Handbook of Pharmaceutical Granulation Technology

© 2005 by Taylor & Francis Group, LLC
DRUGS AND THE PHARMACEUTICAL SCIENCES
Executive Editor
James Swarbrick
PharmaceuTech, Inc.
Pinehurst, North Carolina
Advisory Board
Larry L. Augsburger
University of Maryland
Baltimore, Maryland
Jennifer B. Dressman
Johann Wolfgang Goethe University
Frankfurt, Germany
Jeffrey A. Hughes
University of Florida College of
Pharmacy
Gainesville, Florida
Trevor M. Jones
The Association of the
British Pharmaceutical Industry
London, United Kingdom
Vincent H. L. Lee
University of Southern California
Los Angeles, California
Jerome P. Skelly
Alexandria, Virginia
Geoffrey T. Tucker
University of Sheffield
Royal Hallamshire Hospital
Sheffield, United Kingdom
Harry G. Brittain
Center for Pharmaceutical Physics
Milford, New Jersey
Anthony J. Hickey
University of North Carolina School of
Pharmacy
Chapel Hill, North Carolina
Ajaz Hussain
U.S. Food and Drug Administration
Frederick, Maryland
Hans E. Junginger
Leiden/Amsterdam Center
for Drug Research
Leiden, The Netherlands
Stephen G. Schulman
University of Florida
Gainesville, Florida
Elizabeth M. Topp
University of Kansas School of Pharmacy
Lawrence, Kansas
Peter York
University of Bradford School of Pharmacy
Bradford, United Kingdom
© 2005 by Taylor & Francis Group, LLC
DRUGS AND THE PHARMACEUTICAL SCIENCES
A Series of Textbooks and Monographs
1. Pharmacokinetics, Milo Gibaldi and Donald Perrier
2. Good Manufacturing Practices for Pharmaceuticals: A Plan for Total
Quality Control, Sidney H. Willig, Murray M. Tuckerman,
and William S. Hitchings IV
3. Microencapsulation, edited by J. R. Nixon
4. Drug Metabolism: Chemical and Biochemical Aspects, Bernard Testa
and Peter Jenner
5. New Drugs: Discovery and Development, edited by Alan A. Rubin
6. Sustained and Controlled Release Drug Delivery Systems,
edited by Joseph R. Robinson
7. Modern Pharmaceutics, edited by Gilbert S. Banker
and Christopher T. Rhodes
8. Prescription Drugs in Short Supply: Case Histories, Michael A. Schwartz
9. Activated Charcoal: Antidotal and Other Medical Uses, David O. Cooney
10. Concepts in Drug Metabolism (in two parts), edited by Peter Jenner
and Bernard Testa
11. Pharmaceutical Analysis: Modern Methods (in two parts),
edited by James W. Munson
12. Techniques of Solubilization of Drugs, edited by Samuel H.Yalkowsky
13. Orphan Drugs, edited by Fred E. Karch
14. Novel Drug Delivery Systems: Fundamentals, Developmental Concepts,
Biomedical Assessments, Yie W. Chien
15. Pharmacokinetics: Second Edition, Revised and Expanded, Milo Gibaldi
and Donald Perrier
16. Good Manufacturing Practices for Pharmaceuticals: A Plan for Total
Quality Control, Second Edition, Revised and Expanded, Sidney H. Willig,
Murray M. Tuckerman, and William S. Hitchings IV
17. Formulation of Veterinary Dosage Forms, edited by Jack Blodinger
18. Dermatological Formulations: Percutaneous Absorption, Brian W. Barry
19. The Clinical Research Process in the Pharmaceutical Industry,
edited by Gary M. Matoren
20. Microencapsulation and Related Drug Processes, Patrick B. Deasy
21. Drugs and Nutrients: The Interactive Effects, edited by Daphne A. Roe
and T. Colin Campbell
22. Biotechnology of Industrial Antibiotics, Erick J. Vandamme
23. Pharmaceutical Process Validation, edited by Bernard T. Loftus
and Robert A. Nash
© 2005 by Taylor & Francis Group, LLC
24. Anticancer and Interferon Agents: Synthesis and Properties,
edited by Raphael M. Ottenbrite and George B. Butler
25. Pharmaceutical Statistics: Practical and Clinical Applications,
Sanford Bolton
26. Drug Dynamics for Analytical, Clinical, and Biological Chemists,
Benjamin J. Gudzinowicz, Burrows T.Younkin, Jr.,
and Michael J. Gudzinowicz
27. Modern Analysis of Antibiotics, edited by Adjoran Aszalos
28. Solubility and Related Properties, Kenneth C. James
29. Controlled Drug Delivery: Fundamentals and Applications,
Second Edition, Revised and Expanded, edited by Joseph R. Robinson
and Vincent H. Lee
30. New Drug Approval Process: Clinical and Regulatory Management,
edited by Richard A. Guarino
31. Transdermal Controlled Systemic Medications, edited by Yie W. Chien
32. Drug Delivery Devices: Fundamentals and Applications,
edited by Praveen Tyle
33. Pharmacokinetics: Regulatory • Industrial • Academic Perspectives,
edited by Peter G.Welling and Francis L. S. Tse
34. Clinical Drug Trials and Tribulations, edited by Allen E. Cato
35. Transdermal Drug Delivery: Developmental Issues and Research
Initiatives, edited by Jonathan Hadgraft and Richard H. Guy
36. Aqueous Polymeric Coatings for Pharmaceutical Dosage Forms,
edited by James W. McGinity
37. Pharmaceutical Pelletization Technology, edited by
Isaac Ghebre-Sellassie
38. Good Laboratory Practice Regulations, edited by Allen F. Hirsch
39. Nasal Systemic Drug Delivery, Yie W. Chien, Kenneth S. E. Su,
and Shyi-Feu Chang
40. Modern Pharmaceutics: Second Edition, Revised and Expanded,
edited by Gilbert S. Banker and Christopher T. Rhodes
41. Specialized Drug Delivery Systems: Manufacturing and Production
Technology, edited by Praveen Tyle
42. Topical Drug Delivery Formulations, edited by David W. Osborne
and Anton H. Amann
43. Drug Stability: Principles and Practices, Jens T. Carstensen
44. Pharmaceutical Statistics: Practical and Clinical Applications,
Second Edition, Revised and Expanded, Sanford Bolton
45. Biodegradable Polymers as Drug Delivery Systems,
edited by Mark Chasin and Robert Langer
46. Preclinical Drug Disposition: A Laboratory Handbook, Francis L. S. Tse
and James J. Jaffe
© 2005 by Taylor & Francis Group, LLC
47. HPLC in the Pharmaceutical Industry, edited by Godwin W. Fong
and Stanley K. Lam
48. Pharmaceutical Bioequivalence, edited by Peter G.Welling,
Francis L. S. Tse, and Shrikant V. Dinghe
49. Pharmaceutical Dissolution Testing, Umesh V. Banakar
50. Novel Drug Delivery Systems: Second Edition, Revised and Expanded,
Yie W. Chien
51. Managing the Clinical Drug Development Process, David M. Cocchetto
and Ronald V. Nardi
52. Good Manufacturing Practices for Pharmaceuticals: A Plan for Total
Quality Control, Third Edition, edited by Sidney H. Willig and James R.
Stoker
53. Prodrugs: Topical and Ocular Drug Delivery, edited by Kenneth B. Sloan
54. Pharmaceutical Inhalation Aerosol Technology, edited by
Anthony J. Hickey
55. Radiopharmaceuticals: Chemistry and Pharmacology, edited by
Adrian D. Nunn
56. New Drug Approval Process: Second Edition, Revised and Expanded,
edited by Richard A. Guarino
57. Pharmaceutical Process Validation: Second Edition, Revised
and Expanded, edited by Ira R. Berry and Robert A. Nash
58. Ophthalmic Drug Delivery Systems, edited by Ashim K. Mitra
59. Pharmaceutical Skin Penetration Enhancement, edited by
Kenneth A.Walters and Jonathan Hadgraft
60. Colonic Drug Absorption and Metabolism, edited by Peter R. Bieck
61. Pharmaceutical Particulate Carriers: Therapeutic Applications,
edited by Alain Rolland
62. Drug Permeation Enhancement: Theory and Applications,
edited by Dean S. Hsieh
63. Glycopeptide Antibiotics, edited by Ramakrishnan Nagarajan
64. Achieving Sterility in Medical and Pharmaceutical Products, Nigel A. Halls
65. Multiparticulate Oral Drug Delivery, edited by Isaac Ghebre-Sellassie
66. Colloidal Drug Delivery Systems, edited by Jorg Kreuter
67. Pharmacokinetics: Regulatory • Industrial • Academic Perspectives,
Second Edition, edited by Peter G.Welling and Francis L. S. Tse
68. Drug Stability: Principles and Practices, Second Edition, Revised
and Expanded, Jens T. Carstensen
69. Good Laboratory Practice Regulations: Second Edition, Revised
and Expanded, edited by Sandy Weinberg
70. Physical Characterization of Pharmaceutical Solids, edited by
Harry G. Brittain
© 2005 by Taylor & Francis Group, LLC
71. Pharmaceutical Powder Compaction Technology, edited by
Goran Alderborn and Christer Nystrom
72. Modern Pharmaceutics: Third Edition, Revised and Expanded,
edited by Gilbert S. Banker and Christopher T. Rhodes
73. Microencapsulation: Methods and Industrial Applications,
edited by Simon Benita
74. Oral Mucosal Drug Delivery, edited by Michael J. Rathbone
75. Clinical Research in Pharmaceutical Development, edited by Barry Bleidt
and Michael Montagne
76. The Drug Development Process: Increasing Efficiency and Cost
Effectiveness, edited by Peter G.Welling, Louis Lasagna,
and Umesh V. Banakar
77. Microparticulate Systems for the Delivery of Proteins and Vaccines,
edited by Smadar Cohen and Howard Bernstein
78. Good Manufacturing Practices for Pharmaceuticals: A Plan for Total
Quality Control, Fourth Edition, Revised and Expanded, Sidney H. Willig
and James R. Stoker
79. Aqueous Polymeric Coatings for Pharmaceutical Dosage Forms:
Second Edition, Revised and Expanded, edited by James W. McGinity
80. Pharmaceutical Statistics: Practical and Clinical Applications,
Third Edition, Sanford Bolton
81. Handbook of Pharmaceutical Granulation Technology, edited by
Dilip M. Parikh
82. Biotechnology of Antibiotics: Second Edition, Revised and Expanded,
edited by William R. Strohl
83. Mechanisms of Transdermal Drug Delivery, edited by Russell O. Potts
and Richard H. Guy
84. Pharmaceutical Enzymes, edited by Albert Lauwers and Simon Scharpe
85. Development of Biopharmaceutical Parenteral Dosage Forms,
edited by John A. Bontempo
86. Pharmaceutical Project Management, edited by Tony Kennedy
87. Drug Products for Clinical Trials: An International Guide to Formulation •
Production • Quality Control, edited by Donald C. Monkhouse
and Christopher T. Rhodes
88. Development and Formulation of Veterinary Dosage Forms: Second
Edition, Revised and Expanded, edited by Gregory E. Hardee
and J. Desmond Baggot
89. Receptor-Based Drug Design, edited by Paul Leff
90. Automation and Validation of Information in Pharmaceutical Processing,
edited by Joseph F. deSpautz
91. Dermal Absorption and Toxicity Assessment, edited by Michael S. Roberts
and Kenneth A.Walters
© 2005 by Taylor & Francis Group, LLC
92. Pharmaceutical Experimental Design, Gareth A. Lewis, Didier Mathieu,
and Roger Phan-Tan-Luu
93. Preparing for FDA Pre-Approval Inspections, edited by Martin D. Hynes III
94. Pharmaceutical Excipients: Characterization by IR, Raman, and NMR
Spectroscopy, David E. Bugay and W. Paul Findlay
95. Polymorphism in Pharmaceutical Solids, edited by Harry G. Brittain
96. Freeze-Drying/Lyophilization of Pharmaceutical and Biological Products,
edited by Louis Rey and Joan C. May
97. Percutaneous Absorption: Drugs–Cosmetics–Mechanisms–Methodology,
Third Edition, Revised and Expanded, edited by Robert L. Bronaugh
and Howard I. Maibach
98. Bioadhesive Drug Delivery Systems: Fundamentals, Novel Approaches,
and Development, edited by Edith Mathiowitz, Donald E. Chickering III,
and Claus-Michael Lehr
99. Protein Formulation and Delivery, edited by Eugene J. McNally
100. New Drug Approval Process: Third Edition, The Global Challenge,
edited by Richard A. Guarino
101. Peptide and Protein Drug Analysis, edited by Ronald E. Reid
102. Transport Processes in Pharmaceutical Systems, edited by
Gordon L. Amidon, Ping I. Lee, and Elizabeth M. Topp
103. Excipient Toxicity and Safety, edited by Myra L.Weiner
and Lois A. Kotkoskie
104. The Clinical Audit in Pharmaceutical Development, edited by
Michael R. Hamrell
105. Pharmaceutical Emulsions and Suspensions, edited by
Francoise Nielloud and Gilberte Marti-Mestres
106. Oral Drug Absorption: Prediction and Assessment, edited by
Jennifer B. Dressman and Hans Lennernas
107. Drug Stability: Principles and Practices, Third Edition, Revised
and Expanded, edited by Jens T. Carstensen and C. T. Rhodes
108. Containment in the Pharmaceutical Industry, edited by James P.Wood
109. Good Manufacturing Practices for Pharmaceuticals: A Plan for Total
Quality Control from Manufacturer to Consumer, Fifth Edition, Revised
and Expanded, Sidney H. Willig
110. Advanced Pharmaceutical Solids, Jens T. Carstensen
111. Endotoxins: Pyrogens, LAL Testing, and Depyrogenation, Second Edition,
Revised and Expanded, Kevin L. Williams
112. Pharmaceutical Process Engineering, Anthony J. Hickey
and David Ganderton
113. Pharmacogenomics, edited by Werner Kalow, Urs A. Meyer,
and Rachel F. Tyndale
114. Handbook of Drug Screening, edited by Ramakrishna Seethala
and Prabhavathi B. Fernandes
© 2005 by Taylor & Francis Group, LLC
115. Drug Targeting Technology: Physical • Chemical • Biological Methods,
edited by Hans Schreier
116. Drug–Drug Interactions, edited by A. David Rodrigues
117. Handbook of Pharmaceutical Analysis, edited by Lena Ohannesian
and Anthony J. Streeter
118. Pharmaceutical Process Scale-Up, edited by Michael Levin
119. Dermatological and Transdermal Formulations, edited by
Kenneth A.Walters
120. Clinical Drug Trials and Tribulations: Second Edition, Revised and
Expanded, edited by Allen Cato, Lynda Sutton, and Allen Cato III
121. Modern Pharmaceutics: Fourth Edition, Revised and Expanded,
edited by Gilbert S. Banker and Christopher T. Rhodes
122. Surfactants and Polymers in Drug Delivery, Martin Malmsten
123. Transdermal Drug Delivery: Second Edition, Revised and Expanded,
edited by Richard H. Guy and Jonathan Hadgraft
124. Good Laboratory Practice Regulations: Second Edition,
Revised and Expanded, edited by Sandy Weinberg
125. Parenteral Quality Control: Sterility, Pyrogen, Particulate, and Package
Integrity Testing: Third Edition, Revised and Expanded, Michael J. Akers,
Daniel S. Larrimore, and Dana Morton Guazzo
126. Modified-Release Drug Delivery Technology, edited by
Michael J. Rathbone, Jonathan Hadgraft, and Michael S. Roberts
127. Simulation for Designing Clinical Trials: A Pharmacokinetic-
Pharmacodynamic Modeling Perspective, edited by Hui C. Kimko
and Stephen B. Duffull
128. Affinity Capillary Electrophoresis in Pharmaceutics and Biopharmaceutics,
edited by Reinhard H. H. Neubert and Hans-Hermann Ruttinger
129. Pharmaceutical Process Validation: An International Third Edition,
Revised and Expanded, edited by Robert A. Nash and Alfred H.Wachter
130. Ophthalmic Drug Delivery Systems: Second Edition, Revised
and Expanded, edited by Ashim K. Mitra
131. Pharmaceutical Gene Delivery Systems, edited by Alain Rolland
and Sean M. Sullivan
132. Biomarkers in Clinical Drug Development, edited by John C. Bloom
and Robert A. Dean
133. Pharmaceutical Extrusion Technology, edited by Isaac Ghebre-Sellassie
and Charles Martin
134. Pharmaceutical Inhalation Aerosol Technology: Second Edition,
Revised and Expanded, edited by Anthony J. Hickey
135. Pharmaceutical Statistics: Practical and Clinical Applications,
Fourth Edition, Sanford Bolton and Charles Bon
136. Compliance Handbook for Pharmaceuticals, Medical Devices,
and Biologics, edited by Carmen Medina
© 2005 by Taylor & Francis Group, LLC
137. Freeze-Drying/Lyophilization of Pharmaceutical and Biological Products:
Second Edition, Revised and Expanded, edited by Louis Rey
and Joan C. May
138. Supercritical Fluid Technology for Drug Product Development,
edited by Peter York, Uday B. Kompella, and Boris Y. Shekunov
139. New Drug Approval Process: Fourth Edition, Accelerating Global
Registrations, edited by Richard A. Guarino
140. Microbial Contamination Control in Parenteral Manufacturing,
edited by Kevin L. Williams
141. New Drug Development: Regulatory Paradigms for Clinical Pharmacology
and Biopharmaceutics, edited by Chandrahas G. Sahajwalla
142. Microbial Contamination Control in the Pharmaceutical Industry,
edited by Luis Jimenez
143. Generic Drug Product Development: Solid Oral Dosage Forms ,
edited by Leon Shargel and Izzy Kanfer
144. Introduction to the Pharmaceutical Regulatory Process ,
edited by Ira R. Berry
145. Drug Delivery to the Oral Cavity: Molecules to Market, edited by
Tapash K. Ghosh and William R. Pfister
146. Good Design Practices for GMP Pharmaceutical Facilities ,
edited by Andrew Signore and Terry Jacobs
147. Drug Products for Clinical Trials, Second Edition , edited by Donald
Monkhouse, Charles Carney, and Jim Clark
148. Polymeric Drug Delivery Systems, edited by Glen S. Kwon
149. Injectable Dispersed Systems: Formulation, Processing, and Performance,
edited by Diane J. Burgess
150. Laboratory Auditing for Quality and Regulatory Compliance,
Donald Singer, Raluca-Ioana Stefan, and Jacobus van Staden
151. Active Pharmaceutical Ingredients: Development, Manufacturing, and
Regulation, edited by Stanley H. Nusim
152. Preclinical Drug Development, edited by Mark C. Rogge and David Taft
153. Pharmaceutical Stress Testing: Predicting Drug Degradation, edited by
Steven W. Baertschi
154. Handbook of Pharmaceutical Granulation Technology: Second Edition,
edited by Dilip M. Parikh
© 2005 by Taylor & Francis Group, LLC
Handbook of Pharmaceutical
Granulation Technology
Second Edition
edited by
Dilip M. Parikh
Synthon Pharmaceuticals Inc.
Research Triangle Park, North Carolina, U.S.A.
Boca Raton London New York Singapore
© 2005 by Taylor & Francis Group, LLC
Published in 2005 by
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2005 by Taylor & Francis Group, LLC
No claim to original U.S. Government works
Printed in the United States of America on acid-free paper
10 9 8 7 6 5 4 3 2 1
International Standard Book Number-10: 0-8247-2647-2 (Hardcover)
International Standard Book Number-13: 978-0-8247-2647-8 (Hardcover)
This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with
permission, and sources are indicated. A wide variety of references are listed. Reasonable efforts have been made to publish
reliable data and information, but the author and the publisher cannot assume responsibility for the validity of all materials
or for the consequences of their use.
No part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or
other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information
storage or retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.com
01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and registration for a variety of users. For
organizations that have been granted a photocopy license by the CCC, a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for
identification and explanation without intent to infringe.



Contents
Contributors . . . . xv
1. Introduction . . . . . . 1
Dilip M. Parikh
References . . . . 6
2. Theory of Granulation: An Engineering Perspective . . . . . . . . . . . . . 7
Bryan J. Ennis
1. Introduction . . . . 7
2. Wetting . . . . 19
3. Granule Growth and Consolidation . . . . 34
4. Granule Strength and Breakage . . . . 55
5. Controlling Granulation Processes . . . . 60
References . . . . 77
3. Drug Substance and Excipient Characterization . . . . . . . . . . . . . . 79
L. W. Chan and P. W. S. Heng
1. Introduction . . . . 79
2. Particle Shape, Size, and Surface Area . . . . 80
3. Solubility . . . . 89
4. Crystal Properties and Polymorphism . . . . 95
5. Other Physical Properties . . . . 100
6. Commonly Used Excipients in Granulation . . . . 102
7. Compatibility of Drug and Excipient . . . . 103
8. Conclusion . . . . 105
References . . . . 106
4. Binders and Solvents . . . . . . . . . . . . . . . 109
Ehab Hamed, Derek Moe, Raj Khankari, and John Hontz
1. Introduction . . . . 109
2. Types of Binders . . . . 109
3. Factors Influencing Binder Efficiency . . . . 115
ix
Preface
© 2005 by Taylor & Francis Group, LLC
4. Processing Parameters for Commonly Used Binders . . . . 119
References . . . . 125
5. Spray Drying and Pharmaceutical Applications . . . . . . . . . . . . . . 129
Metin C? elik and Susan C. Wendel
1. Introduction . . . . 129
2. Spray Drying Process Stages . . . . 130
3. Process Layouts . . . . 138
4. Theory of Spray Drying Fundamentals . . . . 139
5. Spray Drying Applications . . . . 146
6. Conclusion . . . . 154
References . . . . 155
6. Roller Compaction Technology . . . . . . . . 159
Ronald W. Miller
1. Introduction . . . . 159
2. Powder Granulation and Compaction . . . . 159
3. Background . . . . 160
4. Benefits of Roller Compaction . . . . 161
5. Compaction Theory . . . . 162
6. Design Features of Roller Compactors . . . . 166
7. Roll Configuration . . . . 167
8. Feed Screw Design . . . . 170
9. Future Trends in Granulation Technology . . . . 172
10. New Findings . . . . 173
11. Deaeration Theory . . . . 178
12. Roller Compaction and Near-Infrared
Spectroscopy . . . . 182
13. Roller Compaction and PAT . . . . 185
References . . . . 188
7. High-Shear Granulation . . . . . . . . . . . . . 191
Rajeev Gokhale, Yichun Sun, and Atul J. Shukla
1. Introduction . . . . 191
2. High-Shear Granulators . . . . 192
3. High-Shear Granulation Process . . . . 195
4. Mechanism of High Shear Wet Granulation . . . . 200
5. Factors Affecting the Granulation Process
and Granule Properties . . . . 204
6. Granulation End-Point Determination . . . . 212
7. Formulation Development (Optimization) . . . . 217
8. Process Scale-Up . . . . 219
9. Conclusion . . . . 224
References . . . . 224
x Contents
© 2005 by Taylor & Francis Group, LLC
8. Low-Shear Granulation . . . . . . . . . . . . . 229
Tom Chirkot and Cecil Propst
1. Introduction . . . . 229
2. Mechanical Agitator Granulators . . . . 231
3. Rotating Shape Granulators . . . . 235
4. Scale-Up . . . . 240
5. End-Point Determination and Control . . . . 243
6. Conclusions . . . . 244
References . . . . 244
9. Batch Fluid Bed Granulation . . . . . . . . . 247
Dilip M. Parikh and Martin Mogavero
1. Introduction . . . . 247
2. Fluidization Theory . . . . 248
3. System Description . . . . 251
4. Particle Agglomeration and Granule Growth . . . . 261
5. Fluid Bed Drying . . . . 265
6. Process and Variables in Granulation . . . . 268
7. Process Controls and Automation . . . . 276
8. Process Scale-Up . . . . 282
9. Safety in Fluid Bed . . . . 286
10. Material Handling Options . . . . 291
11. Fluid Bed Technology Progress . . . . 294
References . . . . 304
10. Single-Pot Processing . . . . . . . . . . . . . . 311
Harald Stahl and Griet Van Vaerenbergh
1. Introduction . . . . 311
2. Typical Single-Pot Process . . . . 313
3. Drying Methods for Single-Pot Processors . . . . 315
4. Other Processes and Applications . . . . 321
5. Scale-Up of Drying Processes . . . . 324
6. Cleaning . . . . 326
7. Product Stability . . . . 326
8. Regulatory Considerations . . . . 326
9. Validation of Single-Pot Processors . . . . 327
10. Control Systems and Data Acquisition Systems . . . . 328
11. Safety . . . . 328
12. Conclusion . . . . 330
References . . . . 330
11. Extrusion/Spheronization as a Granulation Technique . . . . . . . . . 333
Ketan A. Mehta, Gurvinder Singh Rekhi, and Dilip M. Parikh
1. Introduction . . . . 333
2. Applications . . . . 335
3. General Process Description . . . . 336
Contents xi
© 2005 by Taylor & Francis Group, LLC
4. Equipment Description and Process Parameters . . . . 337
5. Formulation Variables . . . . 352
6. Compression of Pellets . . . . 357
7. Conclusions . . . . 359
References . . . . 360
12. Effervescent Granulation . . . . . . . . . . . . 365
Guia Bertuzzi
1. Introduction . . . . 365
2. The Effervescent Reaction . . . . 366
3. Formulation . . . . 366
4. Raw Materials . . . . 367
5. Manufacturing of Effervescent Forms . . . . 371
References . . . . 382
13. Melt Granulation and Pelletization . . . . . 385
T. W. Wong, W. S. Cheong, and P. W. S. Heng
1. Introduction . . . . 385
2. Mechanism of Melt Agglomeration . . . . 389
3. Factors Affecting Melt Agglomeration . . . . 393
4. Control of Melt Agglomeration . . . . 395
5. Conclusions . . . . 400
References . . . . 401
14. Rapid Release Granulation . . . . . . . . . . . 407
P. W. S. Heng, Anthony Yolande, and Lee Chin Chiat
1. Introduction . . . . 407
2. Formulation-Related Factors . . . . 409
3. Granulation-Related Factors . . . . 417
4. Solid Dispersion . . . . 419
5. Conclusions . . . . 423
References . . . . 424
15. Continuous Granulation Technologies . . . . 431
Rudolf Schroeder and Klaus-Ju? rgen Steffens
1. Introduction . . . . 431
2. Comparison of Different Modes of Processing . . . . 431
3. Fluid Bed Systems . . . . 433
4. Mechanical Wet Granulation Systems . . . . 437
5. Roller Extrusion System . . . . 441
References . . . . 456
16. Scale-Up Considerations in Granulation . . 459
Y. He, L. X. Liu, and J. D. Litster
1. Introduction . . . . 459
2. General Considerations in Process Scale-Up: Dimensional Analysis
and the Principle of Similarity . . . . 460
xii Contents
© 2005 by Taylor & Francis Group, LLC
3. Analysis of Granulation Rate Processes
and Implications for Scale-Up . . . . 462
4. Scale-Down, Formulation Characterization, and Formulation Design
in Pharmaceutical Granulation . . . . 473
5. Scale-Up of Fluidized Bed Granulators . . . . 475
6. Scale-Up of High-Shear Mixer Granulators . . . . 480
7. Concluding Remarks . . . . 487
Nomenclature . . . . 487
References . . . . 488
17. Sizing of Granulation . . . . . . . . . . . . . . . 491
Gurvinder Singh Rekhi and Richard Sidwell
1. Introduction . . . . 491
2. Theory of Comminution or Size Reduction . . . . 492
3. Properties of Feed Materials Affecting
the Sizing Process . . . . 493
4. Criteria for Selection of a Mill . . . . 494
5. Classification of Mills . . . . 495
6. Wet Milling . . . . 499
7. Variables Affecting the Sizing Process . . . . 502
8. Scale-Up . . . . 507
9. Case Studies . . . . 509
10. List of Equipment Suppliers . . . . 510
References . . . . 511
18. Granulation Characterization . . . . . . . . . 513
Raj Birudaraj, Sanjay Goskonda, and Poonam G. Pande
1. Introduction . . . . 513
2. Physical and Chemical Characterization
of Granules . . . . 513
3. Conclusion . . . . 531
References . . . . 531
19. Bioavailability and Granule Properties . . . 535
Sunil S. Jambhekar
1. Introduction . . . . 535
2. Bioavailability Parameters . . . . 536
3. Conclusion . . . . 541
References . . . . 542
Recommended Reading . . . . 543
20. Process Analytical Technology . . . . . . . . 545
D. Christopher Watts and Ajaz S. Hussain
1. Introduction . . . . 545
2. Background . . . . 546
3. PAT and Process Understanding . . . . 548
Contents xiii
© 2005 by Taylor & Francis Group, LLC
4. PAT Tools and Their Application . . . . 549
5. Conclusion . . . . 552
References . . . . 552
21. Granulation Process Modeling . . . . . . . . 555
I. T. Cameron and F. Y. Wang
1. Modeling of Granulation Systems . . . . 555
2. Key Factors in Granulation Modeling . . . . 560
3. Representing Granulation Processes Through
Population Balances . . . . 562
4. Solving and Using Population Balances . . . . 570
5. Application of Modeling Techniques . . . . 577
6. Conclusion . . . . 590
References . . . . 591
22. Regulatory Issues in Granulation . . . . . . . 595
Prasad Kanneganti
1. Introduction . . . . 595
2. Pharmaceutical Quality Management . . . . 595
3. Postapproval Change Considerations . . . . 597
4. Validation of Granulation Processes . . . . 609
5. Conclusion . . . . 613
References . . . . 613
xiv Contents




Introduction
Dilip M. Parikh
Synthon Pharmaceuticals Inc., Research Triangle Park, North Carolina, U.S.A.
Perry’s Chemical Engineer’s Handbook (1) defines the granulation process as ‘‘any
process whereby small particles are gathered into larger, permanent masses in which
the original particles can still be identified.’’ This definition is of course particularly
appropriate to a pharmaceutical granulation where the rapid breakdown of agglomerates
is important to maximize the available surface area and aid in solution of the
active drug. The granulation process of size enlargement used within the pharmaceutical
industry has its roots in ancient times. The practice of delivering medicinal powder
by hand rolling into a pill by using honey or sugar has been used for centuries.
It is still the practice to deliver the botanical and herbal extract in homeopathic and
ayurvedic branches of medicine, which are still practiced in India along with allopathic
medicine. The term ‘‘granulated’’ material is derived from the Latin word
‘‘granulatum,’’ meaning grained. The granulated material can be obtained by direct
size enlargement of primary particles, or size reduction from dry compacted material.
In modern times, granulation technology has been widely used by a wide range
of industries, such as coal, mining, and agrochemical. These industries employ
agglomeration techniques to reduce dust, provide ease of handling, and enhance
the material’s ultimate utility.
The development of pharmaceutical granulation was driven by the invention of
the tablet press by W. Brockedon in 1843. Subsequent improvements in the tablet
machinery were patented in the United States by J. A. McFerran (1874), T. J. Young
(1874), and J. Dunton (1876). The demands on the granulation properties were further
enhanced in the 1970s as high-speed tablet and capsule filling machines with automated
controls were introduced. The continuous refinements in the regulatory requirements
such as low-dose products requiring blend uniformity/content uniformity
necessitated knowledge and technology to produce the required granule characteristics.
The high-speed compression and capsule filling machines require a uniform flow
of material to the dies or filling stations that produce pharmaceutical dosage form.
Granulation is an example of particle design. The desired attributes of the
granule are controlled by a combination of the formulation and the process.
Granulation methods can be divided into two major types: wet methods which
utilize some form of liquid to bind the primary particles, and dry methods which do
not utilize any liquid (Fig. 1).

A new approach in the 1990s was to use supercritical fluid technology to produce
uniform particles to replace crystallization. Even though super critical fluids
were discovered over 100 years ago, and the commercial plant was built over 20 years
ago in the United States, it is only now that the technology is used for a number of
pharmaceutical applications (2–5), so as to produce aspirin, caffeine, ibuprofen,
acetaminophen, etc. One of the major areas on which the research and development
of supercritical fluids is focused is particle design. There are different concepts such
as ‘‘rapid expansion of supercritical solution,’’ ‘‘gas antisolvent recrystallization,’’
and ‘‘supercritical antisolvent’’ to generate particles, microspheres, microcapsules,
liposomes, or other dispersed materials.
When the supercritical fluid and drug solution make contact, a volume expansion
occurs leading to a reduction in solvent capacity, increase in solute saturation,
and then supersaturation with associated nucleation and particle formation. A number
of advantages are claimed by using this platform technology (6), such as particle
formation from nanometers to tens of micrometers, low residual solvent levels in
products, preparation of polymorphic forms of drug, etc.
The classical granulation process using either wet or dry methods is employed
in the process industries. Pharmaceutical granulation process is used for tablet and
sometimes capsule dosage forms; however, in some applications the process is used
to produce spherical granules for the modified release indications or to prepare granules
as sprinkles to be used by pediatric patients. In some countries like Japan, having
granulated product in a ‘‘sachet’’ is acceptable where a large dose of the drug
product is not suitable for swallowing. The reasons for granulating a pharmaceutical
compound are listed as follows:
1. To increase the uniformity of drug distribution in the product
2. To densify the material
3. To enhance the flow rates and rate uniformity
4. To facilitate metering or volumetric dispensing
Figure 1 Various granulation techniques.
2 Parikh
© 2005 by Taylor & Francis Group, LLC
5. To reduce dust
6. To improve the appearance of the product.
Five primary methods exist to form an agglomerated granule. They are formation
of solid bridges, sintering, chemical reaction, crystallization, or deposition of
colloidal particles. Binding can also be achieved through adhesion and cohesion
forces in highly viscous binders.
Successful processing for the agglomeration of primary particles depends on
proper control of the adhesional forces between particles, which encourage agglomerate
formation and growth and provide adequate mechanical strength in the
product. Furthermore, the rheology of the particulate system can be critical to
the rearrangement of particles necessary to permit densification of the agglomerate
and the development of an agglomerate structure appropriate for the end-use
requirements. If the particles are close enough then the surface forces such as van
der Waals forces (short-range) and electrostatic forces can interact to bond particles.
Decreasing particle size increases surface–mass ratio and favors the bonding. van der
Waals forces are sevenfold stronger than electrostatic forces and increase substantially
when the distance between them is reduced, which can be achieved by applying
pressure as in dry granulation method.
The cohesive forces that operate during the moist agglomerates are mainly due
to the liquid bridges that develop between the solid particles. Electrostatic forces
keep particles in contact long enough for another mechanism to govern the agglomeration
process.
The processing of drug substance with the excipients can be achieved without
going through the granulation steps. By simply mixing in a blender, a directly compressible
formulation can be processed and compressed in tablets or filled in the hard
gelatin capsules. In the 1970s, microcrystalline cellulose as a directly compressible
vehicle was introduced. The compressible formulation containing microcrystalline
cellulose is suitable for a number of products. This has several obvious advantages,
such as lower equipment cost, faster process time, and efficient operation involving
only two process steps. Sometimes excipient costs may have to be compared against
the savings in the processing steps and equipment by using alternate methods.
There are, however, a number of products that require low dose of drug
substance, where the blend uniformity and the content uniformity in the drug
product are critical. Traditionally, the assessment of the blend uniformity is done
after the blending process is complete. This required considerable delays in obtaining
results, and the sampling techniques and product discharge from the blender
required consistency to obtain satisfactory results. However, with the current interest
in process analytical technology (PAT) on-line measurement of ingredients is possible.
The U.S. Food and Drug Administration (FDA) has recently released guidance
for industry detailing the current thinking on PAT (7).
Other than content uniformity of a low-dose drug substance there are a number
of reasons why direct compression may not be suitable for a wide array of products.
These include the required flow properties; the amount of drug substance in a dosage
form may require it to be densified to reduce the size of the drug product, obtain the
required hardness, friability, disintegration/dissolution, and other attributes.
Another approach which is becoming popular is to use traditional spray-drying
process to produce drum to hopper granulation by-passing the conventional granulation
process. This process may be suitable for large-volume products such as overthe-
counter tablets or capsules.
Introduction 3
© 2005 by Taylor & Francis Group, LLC
Dry compaction technique like roller compaction is experiencing renewed
interest in the industry. There are a number of drug substances which are moisture
sensitive and cannot be directly compressed. The roller compaction provides suitable
alternative technology for processing these products.
Early stages of wet granulation technology development, employed low-shear
mixers or the mixers/blenders normally used for dry blending such as ribbon mixers.
There are a number of products currently manufactured using these low-shear granulators.
The process control and efficiency has increased over the years; however, the
industry has embraced high-shear granulators for wet granulation because of its
efficient and reproducible process and modern process control capabilities. The
high-shear mixers have also facilitated new technologies, such as one-pot processing,
that use the mixer to granulate and then dry using vacuum, gas stripping/vacuum, or
microwave assist in the same vessel.
Fluid-bed processors have been used in the pharmaceutical industry for the last
35 years, initially only as a dryer, and now as a multiprocessor to granulate, dry,
pelletize, and coat particles. The most preferred method of granulation is to use
the high-shear mixer to granulate and use the fluid bed as a dryer in an integrated
equipment setup. This provides the best of both technologies: efficient controllable
dense wet granules and a fast drying cycle using fluid-bed dryer. Here again, the
choice of this approach will be dependent on the product being processed, and its
desired properties at the end of the granulation process. Extrusion/spheronization
is used to produce granulation for the tableting or pelletizing, which involves mixing,
extruding, spheronizing, and drying unit operations. These pellets can be produced
as matrix pellets with the appropriate polymer or are coated in fluid-bed unit to produce
modified release dosage forms. Table 1 illustrates various options to granulate a
pharmaceutical compound.
Many researchers studied the influence of material properties of the granulating
powder and process conditions on the granulation process in a rather empirical
way. In the 1990s a fundamental approach to research was started on various topics
in the granulation research, looking into more detailed aspects of particle wetting,
Table 1 Frequently Used Granulation Techniques and Subsequent Processing
Process Drying technique
Wet granulation Low-shear mixer Tray or fluid-bed dryer
High-shear mixer Tray or fluid-bed dryer
High-shear mixer Vacuum/gas stripping/microwave
assist—one-pot
Fluid-bed granulator/dryer Fluid-bed granulator/dryer
Spray dryer Spray dryer
Extrusion/spheronization Tray or fluid-bed dryer
Continuous mixer granulator
(mechanical)
Fluid bed—continuous or batch
Continuous fluid-bed granulator Fluid bed (continuous)
Dry granulation Process Further processing
Direct compression Blend and process further
Slugging Mill slugged tablets/blend/
recompress/process further
Roller compactor Compacts milled/blend/process
further
4 Parikh
© 2005 by Taylor & Francis Group, LLC
mechanism of granulation, material properties, and influence of mixing apparatus on
the product. The overall hypothesis suggested that the granulation can be predicted
from the raw material properties and the processing conditions of the granulation
process. One of the major difficulties encountered in granulation technology is the
incomplete description of the behavior of powders in general. The ongoing
fundamental research on mixing, segregation mechanisms of powder, surface chemistry,
and material science are necessary to develop the theoretical framework of
granulation technology. An excellent review of the wet granulation process was
presented by Iveson and coauthors (8). The authors have advanced the understanding
of the granulation process by stating that there are three fundamental sets of rate
processes which are important in determining wet granulation behavior. These are
wetting and nucleation, consolidation and growth; and breakage and attrition. Once
these processes are sufficiently understood, then it will be possible to predict the
effect of formulation properties, equipment type, and operating conditions of granulation
behavior, provided these can be adequately characterized according to the
reviewers.
Efficient and cost-effective manufacturing of pharmaceutical products is being
evaluated by the scientists, engineers, and operational managers of pharmaceutical
companies worldwide. In the United States, where 49% of the world pharmaceutical
market is, pharmaceutical companies are under tremendous pressure from the managed
care organizations, politicians, and consumers. The pharmaceutical industry,
worldwide in general and in the United States in particular, faces a unique paradox—
drive future innovation through substantial R&D investments and return competitive
margin to shareholders, while providing access to pharmaceutical products at low or
no cost. The industry has reached a critical juncture in its 100-year history. The industry
is impacted simultaneously by growing competition, declining market performance,
increasing regulation, escalating pricing pressures, and rapidly evolving innovations
for improving people’s health and quality of life. A new report (9) into pharmaceutical
R&D has identified an emerging trend favoring outsourcing of discovery, research,
clinical trials, and manufacturing of dosage forms, providing relief fromthe consistent,
high-growth financial return faced by the majority of pharmaceutical companies. Outsourcing
allows these companies to pursue potential new revenue streams outside of
their core focus areas, and to benefit from improved productivity, emerging technologies,
inlicensing opportunities, and increased growth. Consumers and local governments
in the United States are pressuring the FDA authorities and politicians to
allow importation of the drugs from other countries like Mexico and Canada where
costs are generally lower than in the United States. Demands for price control also
extend to Europe; government-backed pharmaceutical payment plans in Germany
and Italy, for example, have cut back reimbursements. Other European countries have
controls on the drug prices. As a result of these pricing pressures and to enhance the
drugs in the pipeline, mergers and acquisitions have accelerated. Acquisitions remain
the preferred route to quickly enhance a product portfolio. This trend of merging
of equals or takeover of the significant technological companies will continue. This
has created emergence of small niche technology companies as well. Major pharmaceutical
companies are witnessing the end of traditional research and development.
Drug delivery companies are becoming potential targets for mergers or strategic alliances.
During all of the upheaval that the industry is going through, it is becoming
obvious that the cost of development and production, and cost of goods, must be
controlled. Recently released draft guidance by the U.S. FDA for quality systems
approach for the current good manufacturing practices may help to streamline the
Introduction 5
© 2005 by Taylor & Francis Group, LLC
compliance programs in the industry (10). The efficiencies in the research, development,
and manufacturing, which were not necessarily sought after, are becoming
the first priority of the pharmaceutical companies however small they may be in comparison
to the final cost of the product to consumer. The manufacturing of solid
dosage product is no exception.
The significant advances that have taken place in the pharmaceutical granulation
technology are presented in this book to provide the readers with choices that
are available. There is no substitute for good science. The characterization of the
drug substance along with the knowledge of granulation theory, process modeling
capability, in-line or on-line (PAT) tools, process scale-up approaches, and a good
definition of the end product required will prepare the reader to explore the various
options presented in this book. Each drug substance poses a unique challenge that
must be taken into consideration at the process selection stage by the scientists.
The various techniques presented in this book will further help the scientists in their
understanding and selection of the granulation process most appropriate for the
drug substance. For production engineering, validation, and quality professionals
in the industry, this book is intended to provide the fundamental understanding of
the technique of granulation, and the rationale behind the selection of each particular
technique. This will further enhance the ability to design the production plant, carry
out the technology transfer, scale-up, troubleshoot, and maintain the pharmaceutical
granulation operation, in accordance with regulatory compliance.
REFERENCES
1. Ennis BJ, Litster JD. Particle enlargement. Perry RH, Greens D, eds. Perry’s Chemical
Engineer’s Handbook. 7th ed. New York: McGraw Hill, 1997:20-56–20-89.
2. Charoenchaitrakool M, Dehghani F, Foster NR. Micronization by RESS to enhance the
dissolution rates of poorly water soluble pharmaceuticals. Proceedings of the 5th International
Symposium on Supercritical Fluids, Atlanta, GA, April 8–12, 2000.
3. Matson DW, Fulton JL, Petersen RC, Smith RD. Rapid expansion of supercritical fluid
solutions: solute formation of powders, thin films, and fibers. Ind Eng Chem Res 1987;
26:2298–2306.
4. Subra P, Boissinot P, Benzaghou S. Precipitation of pure and mixed caffeine and anthracene
by rapid expansion of supercritical solutions. Proceedings of the 5th Meeting on
Supercritical Fluids, Tome I, Nice, France, March 23–25, 1998.
5. Gilbert DJ, Palakodaty S, Sloan R, York V. Particle engineering for pharmaceutical
applications—a process scale up. Proceedings of the 5th International Symposium on
Supercritical Fluids, Atlanta, GA, April 8–12, 2000.
6. York P, et al. Supercritical fluids ease drug delivery. Manuf Chemist 2000; 26–29.
7. Food and Drug Administration. Guidance for Industry PAT—A Framework for
Innovative Pharmaceutical Development, Manufacturing and Quality Assurance.
FDA, September 2004.
8. Iveson SM, Litster JD, Hopgood K, Ennis B. Nucleation, growth, and breakage phenomenon
in agitated wet granulation process: a review. Powder Technol 2001; 117:3–39.
9. Cambridge Healthcare Advisors (CHA) Report. Report identifies increasing outsourcing
by pharma–Inpharma.com, September 29, 2004.
10. Guidance for Industry. Quality Systems Approach to Pharmaceutical Current Good
Manufacturing Practice Regulations, Draft Guidance, September 2004.
6 Parikh
© 2005 by Taylor & Francis Group, LLC
2
Theory of Granulation: An Engineering
Perspective
Bryan J. Ennis
E&G Associates, Inc., Nashville, Tennessee, U.S.A.
1. INTRODUCTION
1.1. Overview
Wet granulation is a subset of size enlargement (1–5), which involves any process
whereby small particles are agglomerated, compacted, or otherwise brought together
into larger, relatively permanent structures in which the original particles can still be
distinguished. Granulation technology and size-enlargement processes have been
used by a wide range of industries, from the pharmaceutical industry to fertilizer
or detergent production to the mineral processing industries. Size enlargement generally
encompasses a variety of unit operations or processing techniques dedicated to
particle agglomeration. These processes can be loosely broken down into agitation
and compression methods.
Although terminology is industry specific, agglomeration by agitation will be
to a process vessel and is agglomerated, either batchwise or continuously, to form
a granulated product. Agitative processes include fluid bed, pan (or disk), drum,
and mixer granulators. Such processes are also used as coating operations for
controlled release, taste masking, and cases where solid cores may act as a carrier
for a drug coating. The feed typically consists of a mixture of solid ingredients,
referred to as a formulation, which includes an active or key ingredient, binders,
diluents, flow aids, surfactants, wetting agents, lubricants, fillers, or end-use aids
(e.g., sintering aids, colors or dyes, taste modifiers). A closely related process of spray
drying is also included here, but is discussed more fully elsewhere [Refs. 6,7 and
carrier cores, or spray dried product consisting of agglomerated solidified drops.
An alternative approach to size enlargement is agglomeration by compression,
or compaction, where the mixture of particulate matter is fed to a compression device
which promotes agglomeration due to pressure, as depicted
continuous sheets of solid material or solid forms such as briquettes or tablets are
produced. Compaction processes range from confined compression devices such as
7
referred to as granulation. As depicted in Figure 1, a particulate feed is introduced
in Figure 2. Either
Chapter 5]. Product forms generally include agglomerated or layered granules, coated
tableting to continuous devices such as roll presses (Chapter 6), briqueting machines,
© 2005 by Taylor & Francis Group, LLC
such as ram extrusion. Capsule filling operations would be considered as low-pressure
compaction processes.
At the level of a manufacturing plant, the size-enlargement process involves
several peripheral, unit operations such as milling, blending, drying, or cooling,
addition, more than one agglomeration step may be present as in the case of a pharmaceutical
process. In the case of pharmaceutical granulation, granulated material is
almost exclusively an intermediate product form, which is then followed by tableting.
In the context of granulation, therefore, it is important to understand compaction
processes to establish desirable granule properties for tableting performance.
Numerous benefits result from size-enlargement processes as summarized in
tion in the case of pharmaceutical processing is to create free-flowing, nonsegregating
blends of ingredients of controlled strength, which may be reproducibly metered in
Figure 2 The unit operation of compressive agglomeration or compaction. (From Refs. 1–5.)
Figure 1
8 Ennis
and classification, referred to generically as an agglomeration circuit (Fig. 3). In
and extrusion (Chapter 11). Some processes operate in a semicontinuous fashion
Table 1. A wide variety of size-enlargement methods are available; a classification of
these is given in Table 2 with key process attributes.Aprimary purpose of wet granula-
The unit operation of agitative agglomeration or granulation. (From Refs. 1–5.)
© 2005 by Taylor & Francis Group, LLC
subsequent tableting or for vial or capsule filling operations. The wet granulation
process must generally achieve the desired granule properties within some prescribed
range. These attributes clearly depend on the application at hand. However, common
to most processes is a specific granule size distribution and granule voidage. Size
distribution affects flow and segregation properties, as well as compaction behavior.
Granule voidage controls strength, and impacts capsule and tablet dissolution
behavior, as well as compaction behavior and tablet hardness.
Control of granule size and voidage will be discussed in detail throughout this
chapter. The approach taken here relies heavily on attempting to understand interactions
at a particle level, and scaling this understanding to bulk effects. Developing
an understanding of these microlevel processes of agglomeration allows a rational
approach to the design, scale-up, and control of agglomeration processes. Although
Figure 3 A typical agglomeration circuit utilized in the processing of pharmaceuticals
Table 1 Objectives of Size Enlargement
Production of a useful structural form
Provision of a defined quantity for dispensing, with improved flow properties for metering
Improved product appearance
Reduced propensity to caking
Increased bulk density for storage
Creation of nonsegregating blends of powder ingredients
Control of solubility
Control of porosity, hardness, and surface-to-volume ratio and particle size
Source: From Refs. 1–5.
Theory of Granulation 9
involving both granulation and compression techniques. (From Refs. 1–5.)
© 2005 by Taylor & Francis Group, LLC
Table 2 Size-Enlargement Methods and Application
Method
Product size
(mm) Granule density
Scale of
operation Additional comments Typical applications
Tumbling granulators:
Drums
Disks
0.5–20 Moderate 0.5–800
tons/hr
Very spherical granules Fertilizers, iron ore,
nonferrous ore,
agricultural chemicals
Mixer granulators
Handles very cohesive
materials well, both batch
and continuous
Continuous high shear
(e.g., Shugi mixer)
0.1–2 Low Up to 50
ton/hr
Chemicals, detergents, clays,
carbon black
Batch high shear
(e.g., paddle mixer)
0.1–2 High Up to 500 kg,
batch
Pharmaceuticals, ceramics
Fluidized granulators:
Fluidized beds
Spouted beds
Wurster coaters
0.1–2 Low (agglomerated),
moderate (layered)
100–900 kg, batch;
50 ton/hr,
continuous
Flexible, relatively easy to
scale, difficult for cohesive
powders, good for coating
applications
Continuous: fertilizers,
inorganic salts, detergents;
batch: pharmaceuticals,
agricultural chemicals,
nuclear wastes
10 Ennis
© 2005 by Taylor & Francis Group, LLC
Centrifugal granulators 0.3–3 Moderate to high Up to 200 kg,
batch
Powder layering and coating
applications
Pharmaceuticals, agricultural
chemicals
Spray methods
Spray drying 0.05–0.5 Low Morphology of spray dried
powders can vary widely
Instant foods, dyes,
detergents, ceramics,
pharmaceuticals
Prilling 0.7–2 Moderate Urea, ammonium nitrate0
Pressure compaction
Very narrow size
distributions, very sensitive
to powder flow and
mechanical properties
Pharmaceuticals, catalysts,
inorganic chemicals,
organic chemicals, plastic
preforms, metal parts,
ceramics, clays, minerals,
animal feeds
Extrusion >0.5 High to very high Up to 5 ton/hr
Roll press >1 Up to 50 ton/hr
Tablet press 10 Up to 1 ton/hr
Molding press
Pellet mill
Theory of Granulation 11
Source: From Refs. 1–5.
© 2005 by Taylor & Francis Group, LLC
the approach is difficult, qualitative trends are uncovered along the way, which aid in
formulation development and process optimization, and which emphasize powder
characterization as an integral part of product development and process design
work.
1.2. Granulation Mechanisms
Four key mechanisms or rate processes contribute to granulation, as originally outlined
by Ennis (3,4), and later developed further by Ennis and Litster (1,5). These
include wetting and nucleation, coalescence or growth, consolidation, and attrition
or breakage (Fig. 4). Initial wetting of the feed powder and existing granules by
the binding fluid is strongly influenced by spray rate or fluid distribution as well
as feed formulation properties, in comparison with mechanical mixing. Wetting promotes
nucleation of fine powders, or coating in the case of feed particle size in excess
of drop size. Often wetting agents such as surfactants are carefully chosen to enhance
poorly wetting feeds. In the coalescence or growth stage, partially wetted primary
particles and larger nuclei coalesce to form granules composed of several particles.
The term nucleation is typically applied to the initial coalescence of primary particles
in the immediate vicinity of the larger wetting drop whereas the more general term of
Figure 4 The rate processes of agitative agglomeration, or granulation, which include
powder wetting, granule growth, granule consolidation, and granule breakage. These processes
combine to control granule size and porosity, and they may be influenced by formulation
or process design changes.
12 Ennis
© 2005 by Taylor & Francis Group, LLC
coalescence refers to the successful collision of two granules to form a new, larger
granule. In addition, the term of layering is applied to the coalescence of granules
with primary feed powder. Nucleation is promoted from some initial distribution
of moisture, such as a drop, or from the homogenization of a fluid feed to the
bed, as with high-shear mixing. The nucleation process is strongly linked with the
wetting stage. As granules grow, they are consolidated by compaction forces due
to bed agitation. This consolidation stage controls internal granule voidage or granule
porosity, and therefore end-use properties such as granule strength, hardness, or
dissolution. Formed granules may be particularly susceptible to attrition if they are
inherently weak or if flaws develop during drying.
These mechanisms or rate processes can occur simultaneously in all granulation
units, ranging from spray drying to fluidized beds to high-shear mixers. However,
certain mechanisms may dominate in a particular manufacturing process. For example,
fluidized bed granulators are strongly influenced by the wetting process, whereas
mechanical redispersion of binding fluid by impellers and particularly high-intensity
choppers diminishes the wetting contributions to granule size in high-shear mixing.
On the other hand, granule consolidation is far more pronounced in high-shear
mixing than in fluidized-bed granulation. These simultaneous rate processes taken
as a whole—and sometimes competing against one another—determine the final granule
size distribution and granule structure and voidage resulting from a process, and
therefore the final end-use or product quality attributes of the granulated product.
1.3. Compaction Mechanisms
Compaction is a forming process controlled by mechanical properties of the feed in
relationship to applied stresses and strains. Microlevel processes are controlled by
particle properties such as friction, hardness, size, shape, surface energy, and elastic
modulus.
Key steps in any compaction process include (1) powder filling, (2) stress application
and removal, and (3) compact ejection. Powder filling and compact weight
variability is strongly influenced by bulk density and powder flowability (1,2), as well
as any contributing segregation tendencies of the feed. The steps of stress application
Powders do not transmit stress uniformly. Wall friction impedes the applied load,
causing a drop in stress as one moves away from the point of the applied load
(e.g., a punch face in tableting or roll surface in roll pressing.) Therefore, the applied
load and resulting density are not uniform throughout the compact, and powder frictional
properties control the stress transmission and distribution in the compact (8).
The general area of study relating compaction and stress transmission is referred to
as powder mechanics (1,2,8,9). For a local level of applied stress, particles deform at
their point contacts, including plastic deformation for forces in excess of the particle
surface hardness. This allows intimate contact at surface point contacts, allowing
cohesion/adhesion to develop between particles, and therefore interfacial bonding,
which is a function of their interfacial surface energy. During the short time scale
of the applied load, any entrapped air must escape, which is a function of feed
permeability, and a portion of the elastic strain energy is converted into permanent
plastic deformation. Upon stress removal, the compact expands due to the remaining
elastic recovery of the matrix, which is a function of elastic modulus, as well as any
expansion of the remaining entrapped air. This can result in loss of particle bonding,
and flaw development, and this is exacerbated for cases of wide distributions in
Theory of Granulation 13
and removal consist of several competing mechanisms, as depicted in Figure 5.
© 2005 by Taylor & Francis Group, LLC
compact stress due to poor stress transmission. The final step of stress removal
involves compact ejection, where any remaining radial elastic stresses are removed.
If recovery is substantial, it can lead to capping or delamination of the compact.
These microlevel processes of compaction control the final flaw and density
distribution throughout the compact, whether it is a roll pressed, extruded, or tableted
product, and as such, control compact strength, hardness, and dissolution
behavior. Compaction processes will not be discussed further here, with the remainder
of the chapter focusing on wet granulation, agitative processes. For further
1.4. Formulation vs. Process Design
The end-use properties of granulated material are controlled by granule size and
internal granule voidage or porosity. Internal granule voidage egranule and bed
voidage ebed, or voidage between granules, are related by
rbulk ? rgranule?1 
 ebed? ? rs?1 
 ebed??1 
 egranule? ?1:1?
where rbulk, rgranule, and rs are bulk, granule, and skeletal primary particle density,
respectively. Here, granule voidage and granule porosity will be used interchangeably.
Granule structure may also influence properties. To achieve the desired product
quality as defined by metrics of end-use properties, granule size and voidage may be
manipulated by changes in either process operating variables or product material
further by Ennis and Litster (1,2,5). The first approach is the realm of traditional
process engineering, whereas the second is product engineering. Both approaches
are critical and must be integrated to achieve the desired end point in product
quality. Operating variables are defined by the chosen granulation technique and
Figure 5 The microlevel processes of compressive agglomeration, or compaction. These
processes combined control compact strength, hardness, and porosity.
14 Ennis
variables (Figs. 4 and 5) as initially outlined by Ennis (3,4), and later developed
discussion regarding compaction, see Chapter 6 and Refs. 1, 2, and 10.
© 2005 by Taylor & Francis Group, LLC
peripheral processing equipment, as illustrated for a fluidized-bed and mixer granulator
in Figure 6. In addition, the choice of agglomeration technique dictates the
mixing pattern of the vessel. Material variables include parameters such as binder
viscosity, surface tension, feed particle size distribution, powder friction, and the
adhesive properties of the solidified binder. Material variables are specified by the
choice of ingredients, or product formulation. Both operating and material variables
together define the kinetic mechanisms and rate constants of wetting, growth,
consolidation, and attrition. Overcoming a given size-enlargement problem often
requires changes in both processing conditions and product formulation.
The importance of granule voidage to final product quality is illustrated in
granule attrition to increase, and dissolution rate to increase with an increase in
granule voidage. Bulk density is clearly a function of both granule size distribution,
which controls bed voidage or porosity between granules, and the voidage within the
granule itself. The data of Figure 7 are normalized with respect to its zero intercept,
or its effective bulk density at zero granule voidage. The granule attrition results of
fines passing a fine mesh size following attrition in a tumbling apparatus. Granules
weaken with increased voidage. The dissolution results of Figure 9 measure the
length required for granule dissolution in a long tube, or disintegration length (also
based on the CIPAC test method). Increased granule voidage results in increased
dissolution rate and shorter disintegration length. All industries have their own
specific quality and in-process evaluation tests. However, what they have in common
are the important contributing effects of granule size and granule voidage.
An example of the importance of distinguishing the effects of process and
formulation changes can be illustrated with the help of Figures 8 and 9. Let us
assume that the particular formulation and current process conditions produce a
granulated material with a given attrition resistance and dissolution behavior
(indicted as ‘‘current product’’). If one desires instead to reach a given ‘‘target,’’
either the formulation or the process variable may be changed. Changes to the
process, or operating variables, generally readily alter granule voidage. Examples
to decrease voidage might include increased bed height, increased processing time,
Figure 6
Theory of Granulation 15
Figures 7–9 for a variety of formulations. Here, bulk density is observed to decrease,
Figure 8 are based on a CIPAC test method, which is effectively the percentage of
Typical operating variable for pharmaceutical granulation processes. (From Ref. 4.)
© 2005 by Taylor & Francis Group, LLC
or increased peak bed moisture. However, only a range of such changes in voidage,
are due instead to changes in formulation properties. Therefore, it may not be
Figure 7 Impact of granule density on bulk density: Normalized bulk density as a function
Figure 8 Impact of granule density on strength and attrition: Illustration of process changes
vs. formulation changes. Kc is the fracture toughness. (From Ref. 4.)
16 Ennis
and therefore attrition resistance and dissolution, is possible and is illustrated by
moving along the given formulation curve. The various curves in Figures 8 and 9
of granule voidage. (From Ref. 4.)
© 2005 by Taylor & Francis Group, LLC
possible to reach a target change in dissolution without changes in formulation or
material variables. Examples of key material variables affecting voidage would
include feed primary particle size, inherent formulation bond strength, and binder
solution viscosity, as discussed in detail in the following sections. Understanding this
crucial interaction between operating and material variables is critical for successful
formulation, and requires substantial collaboration between processing and formulation
groups, and a clear knowledge of the effect of scale-up on this interaction.
1.5. Key Historical Investigations
A range of historical investigations has been undertaken involving the impact of
cal variables have included the effects of bed hydrodynamics and agitation intensity,
pan angle and speed, fluid-bed excess gas velocity, mixer impeller and chopper
speeds, drum rotation speed, spray method, drop size, nozzle location, and binder
and solvent feed rates. While such studies are important, their general application
and utility to studies beyond the cited formulations and process conditions can be
severely limited. Often the state of mixing, moisture distribution and rates, and material
properties such as formulation size distribution, powder frictional properties,
and solution viscosity are insufficiently defined. As such, these results should be used
judiciously and with care. Often even the directions of the impact of operating variables
on granule properties are altered by formulation changes.
Two key pieces of historical investigation require mention, as the approach
developed here stems heavily from this work. The first involves growth and breakage
mechanisms, which control the evolution of the granule size distribution (14), as
Figure 9 Impact of granule density on disintegration: Gc is the strain energy release rate.
Theory of Granulation 17
operating variables on granulation behavior. As a review, see Refs. 3–5,11–13. Typi-
(From Ref. 4.)
© 2005 by Taylor & Francis Group, LLC
illustrated in Fig. 10. These include the nucleation of fine powder to form initial
primary granules, the coalescence of existing granules, and the layering of raw material
onto previously formed nuclei or granules. Granules may simultaneously be
compacted by consolidation and reduced in size by breakage. There are strong interactions
between these rate processes. In addition, these mechanisms in various forms
have been incorporated into population-balance modeling to predict granule size in
the work of Kapur and Sastry (14–19) (Chapter 23 for details). Given the progress
made in connecting rate constants to formulation properties, the utility of population-
balance modeling has increased substantially.
The second important area of contribution involves the work of Rumpf (20,21)
and Ausburger and Vuppala (22), which studied the impact of interparticle force H
on granule static tensile strength, or
sT ?
9
8
1 
 e
e 
 H
a2 ? A
1 
 e
e 
 g cos y
a
with
A ? 9=4 for pendular state
A ? 6 for capillary state 
?1:2?
Forces of a variety of forms were studied, including viscous, semisolid, solid,
electrostatic, and van der Waals forces. Of particular importance was the contribution
of pendular bridge force arising from surface tension to granule tensile strength.
Capillary pressure deficiency due to the curvature of the pendular bridge in addition
(here, interparticle velocity U?0). This force summed over the granule area results
in a granule static strength, which is a function of pore saturation S as experimentally
plotted. The states of pore filling have been defined as pendular (single bridges),
Figure 10
18 Ennis
to a contact line force results in an interparticle force, as highlighted in Figure 11
Growth and breakage mechanisms in granulation processes. (From Ref. 14.)
© 2005 by Taylor & Francis Group, LLC
funicular (partially complete filling and single bridges), capillary (nearly complete
filling S80–100%) followed by drop formation, and loss of static strength. This
approach will be extended in subsequent sections to include viscous forces and
dynamic strength behavior (U 6? 0).
The approach taken, in this chapter, follows this same vein of research as originally
established by Rumpf and Kapur, namely, relating granule and particle level
interactions to bulk behavior through the development of the rate processes of
wetting and nucleation, granule growth and consolidation, and granule breakage
and attrition. Each of these will now be dealt with in turn in the following sections.
2. WETTING
2.1. Overview
The initial distribution of binding fluid can have a pronounced influence on the size
distribution of seed granules, or nuclei, which are formed from fine powder. Both the
final extent of and the rate at which the fluid wets the particulate phase are
important. Poor wetting results in drop coalescence, and in fewer, larger nuclei with
ungranulated powder and overwetted masses, leading to broad nuclei distributions.
Granulation can retain a memory, with nuclei size distribution impacting final granule
size distribution. Therefore, initial wetting can be critical to uniform nuclei formation
and often a narrow, uniform product. Wide nuclei distributions can lead to a
wide granule size distribution. When the size of a particulate feed material is larger
than drop size, wetting dynamics controls the distribution of coating material which
has a strong influence on the later stages of growth. Wetting phenomena also influence
redistribution of individual ingredients within a granule, drying processes, and
redispersion of granules in a fluid phase. Other granule properties such as voidage,
Figure 11
Theory of Granulation 19
nomenclature.) (From Refs. 20–22.)
Static yield strength of wet agglomerates vs. pore saturation. (See Section 3.5 for
© 2005 by Taylor & Francis Group, LLC
strength, and attrition resistance may be influenced as well. Preferential wetting of
certain formulation ingredients can cause component segregation with granule size.
An extensive review of wetting research may be found in Litster and Ennis (5), Parfitt
2.2. Mechanics of the Wetting Rate Process
As outlined previously, wetting is the first stage in wet granulation involving liquid
binder distribution to the feed powder. There are two extremes: (1) liquid drop size is
large compared to unit or primary particle size of the feed and (2) particle size is
large compared to the drop size.
the wetting process consists of several important steps. First, droplets are formed
related to spray distribution, or spray flux defined as the wetting area of the bed
per unit time. Important operating variables include nozzle position, spray area,
spray rate, and drop size. Second, droplets impact and coalesce on the powder
bed surface if mixing or wet-in time is slow. Third, droplets spread and penetrate into
the moving powder bed to form loose nuclei, again coalescing if wet-in is slow. In the
case of high-shear processes, shear forces break down overwet clumps, also producing
nuclei.
For the second case of small drop size compared to the primary particle size,
collisions between the drop and the particle followed by spreading of the liquid over
the particle surface. If the particle is porous, then liquid will also be sucked into the
pores by capillary action. The wetting dynamics control the distribution of coating
material, which has a strong influence on the later stages of growth as well as coating
quality.
2.3. Methods of Measurement
Methods of characterizing the rate process of wetting include four approaches as
powder. This approach involves the measurement of a contact angle of a drop on
a powder compact, given by the Young–Dupre? equation, or
gsv 
 gsl ? glv cos y ?2:1?
where gsv, gsl, glv are the solid–vapor, solid–liquid, and liquid–vapor interfacial energies,
respectively, and y is measured through the liquid. In the limit of gsv
gslglv,
the contact angle equals 0 and the fluid spreads on the solid. The extent of wetting is
controlled by the group glv cosy, which is referred to as adhesion tension. Sessile drop
studies of contact angle can be performed on powder compacts in the same way as
on planar surfaces. Methods involve the (1) direct measurement of the contact angle
from the tangent to the air–binder interface, (2) solution of the Laplace–Young
equation involving the contact angle as a boundary condition, or (3) indirect calculations
of the contact angle from measurements of, e.g., drop height. The compact can
either be saturated with the fluid for static measurements, or dynamic measurements
granulation processes, the dynamics of wetting are often crucial, requiring that
powders be compared on the basis of a short time scale, dynamic contact angle.
20 Ennis
As depicted in Figure 12 for the first case of fine feeds compared to drop size,
the liquid will coat the particles as depicted in Figure 13. Coating is produced by
may be made through a computer imaging goniometer, as depicted in Figure 14. For
(23), and Hapgood (24), with a summary of references given in Table 3.
illustrated in Table 4. The first considers the ability of a drop to spread across the
© 2005 by Taylor & Francis Group, LLC
Table 3 Summary of Wetting Research Contributions
Workers (Year) Materials Equipment Investigated
Schaefer and Worts
(1978)
Lactose and maize
starch powder with
gelatine, PVP, and
MCC binders
Fluid bed Drop size
Schaafsma et al.
(1999)
Lactose powder with
water and 3–8%
PVP solution
binders
Packed bed
in
petri dish
Drop size
Waldie (1991) Lactose and ballotini
powders with water
and 5% PVP
solution
Fluid bed Drop size
Watano et al. (1994) Lactose and
cornstarch with
water
Hybrid fluid
bed
with
agitator
Drop size, binder
distribution
Holm et al. (1983,
1984)
Lactose and calcium
hydrogen phosphate
with water and 10–
15% PVP solution
High–shear
mixer
Drop size, binder
distribution, and
shear forces
Aulton and Banks
(1979)
Lactose and salicylic
acid powder with
water
Fluid bed Contact angle
Gluba et al. (1990) Talcum, chalk, and
kaolin powder
combinations with
water
Drum Contact angle
Jaiyeoba and Spring
(1980)
Lactose, boric acid,
kaolin,
suphanilamide, and
salicylic acid
powders with water.
Fluid bed Contact angle
Krycer and Pope
(1983)
Paracetamol powder
with HPMC, PVP,
and water
Fluid bed Spreading coefficients
Zajic and Buckton
(1990)
Methycellulose
powder with
HPMC, PVP, and
water
Mixer
granulator
Spreading coefficients
Rankell et al. (1964) Aluminium hydroxide
and sucrose powder
with water
Fluid bed Binder addition rate,
powder flux, spray
zone
Heim and Antkowiak
(1989)
Talc powder and
water
Drum Binder addition rate
Crooks and Schade
(1978)
5% Phenylbutazone in
lactose powder mix
with 10% solution
Fluid bed Binder addition rate
Kristensen and
Schaefer (1994)
Lactose powder with
PEG300 and
PEG6000
High shear Viscosity, binder
dispersion, shear
forces
(Continued)
Theory of Granulation 21
© 2005 by Taylor & Francis Group, LLC
Important factors are the physical nature of the powder surface (particle size, pore
size, porosity, environment, roughness, pretreatment). The dynamic wetting process
is therefore influenced by the rates of ingredient dissolution and surfactant adsorption
and desorption kinetics (25).
The second approach to characterize wetting considers the ability of the fluid
of the extent and rate of fluid rise by capillary suction into a column of powder, better
known as the Washburn test. Considering the powder to consist of capillaries of
radius R, the equilibrium height of rise he is determined by equating capillary and
gravimetric pressures, or
he ?
2glv cos y
DrgR ?2:2?
Figure 12 Stages of wetting for fine powder compared to drop size. (From Ref. 5.)
Table 3 Summary of Wetting Research Contributions (Continued )
Workers (Year) Materials Equipment Investigated
Schaefer et al. (1992) Lactose powder with
PEG300 and
PEG6000
High shear Viscosity, binder
dispersion, shear
forces, powder flux
Cartensen et al. (1976) Lactose, sucrose, and
starch powder
mixed with water
High shear Binder dispersion
Kokubo et al. (1996) Lactose–cornstarch–
MCC with HPMC
3cp
High–speed
mixer
Powder flux
Davies and Gloor
(1971)
Lactose, comstarch,
magnesium stearate,
gelatin, and benzene
powder with gelatin
solution
Fluid bed Spray zone
22 Ennis
to penetrate a powder bed, as illustrated in Figure 15. It involves the measurement
Source: From Ref. 5.
© 2005 by Taylor & Francis Group, LLC
where is Dr the fluid density with respect to air, g is gravity, and glv cos y is the adhesion
tension as before.
In addition to the equilibrium height of rise, the dynamics of penetration are
particularly important. Ignoring gravity and equating viscous losses with the capillary
pressure, the rate (dh/dt) and dynamic height of rise h are given by
dh
dt ?
Rglv cos y
4mh
or h ? ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Rglv cos y
2m  t s ?2:3?
where t is time and m is binder fluid viscosity (23). The grouping of terms in brackets
involve the material properties which control the dynamics of fluid penetration,
namely, average pore radius, or tortuosity R (related to particle size and void distribution
of the powder), adhesion tension, and binder viscosity. The rate of capillary
fluid rise, or the rate of binding fluid penetration in wet granulation, increases with
increasing pore radius (generally coarser powders with larger surface–volume average
particle size), increasing adhesion tension (increased surface tension and
decreased contact angle), and decreased binder viscosity.
The contact angle of a binder–particle system is not itself a primary thermodynamic
quantity, but rather a reflection of individual interfacial energies (Eq. 2.1),
which are a function of the molecular interactions of each phase with respect to
one another. An interfacial energy may be broken down into its dispersion and polar
components. These components reflect the chemical character of the interface, with
the polar component due to hydrogen bonding and other polar interactions and the
dispersion component due to van der Waals interactions. These components may be
determined by the wetting tests described here, where a variety of solvents are chosen
as the wetting fluids to probe specific molecular interactions (26). Interfacial energy
is strongly influenced by trace impurities which arise in crystallization of the active
ingredient, or other forms of processing such as grinding. It may be modified by judicious
selection of surfactants (27). Charges may also exist at interfaces, as characterized
by electrokinetic studies (28). The total solid–fluid interfacial energy (i.e., both
dispersion and polar components) is also referred to as the critical solid surface
energy of the particulate phase. It is equal to the surface tension of a fluid which just
wets the solid with zero contact angle. This property of the particle feed may be
determined by a third approach to characterize wetting, involving the penetration
of particles into a series of fluids of varying surface tension (27,29), or by the variation
of sediment height (30).
The last approach to characterizing wetting involves chemical probing of properties
which control surface energy. As an example, inverse gas chromatography
(IGC) uses the same principles and equipment as standard gas chromatography.
In IGC, however, the mobile phase is composed of probe gas molecules that move
Figure 13
Theory of Granulation 23
© 2005 by Taylor & Francis Group, LLC
Stages of wetting for coarse powder compared to drop size. (From Ref. 5.)
Table 4 Methods of Characterizing Wetting Dynamics of Particulate Systems
Mechanism of wetting Characterization method
Spreading drops on powder surface Contact angle goniometer:
Contact angle
Drop height or volume
Spreading velocity
References:
Kossen, Heertjes. Chem Eng Sci 1965;
20:593
Pan et al. Dynamic Properties of Interfaces
and Association Structure. New York:
American Oil Chem Soc Press, 1995
Penetration of drops into powder bed Washburn test:
Rate of penetration by height or volume
Bartell cell:
Capillary pressure difference
References:
Parfitt, ed. Dispersion of Powders in
Liquids. Washburn: Elsevier Applied
Science Publishers Ltd., 1986, and Phys Rev
1921; 17:273
Bartell, Osterhof. Ind Eng Chem 1927;
19:1277
Penetration of particles into fluid Flotation tests:
Penetration time
Sediment height
Critical solid surface energy distribution
References:
Ayala R. Ph.D.Thesis, Chem. Eng.,
Carnegie Mellon University, 1985
Fuerstaneau et al. Coll Surf 1991; 60:127
Vargha-Butler. In: Botsaris, Glazman, eds.
Interfacial Phenomena in Coal Technology,
Chap. 2, 1994
Chemical probing of powder Inverse gas chromatography:
Preferential adsorption with probe gases
Electrokinetics:
Zeta potential and charge
Surfactant adsorption:
Preferential adsorption with probe
surfactants
References:
Lloyd et al., eds. ACS Symposium Series.
Vol. 391. Washington, DC: ACS, 1989
Aveyard, Haydon. An Introduction to the
Principles of Surface Chemistry. Cambridge
University Press, 1973
Shaw. Introduction to Colloid & Surface
Chemistry. London: Butterworths & Co.
Ltd., 1983
24 Ennis
Source: From Refs. 1,2,5.
© 2005 by Taylor & Francis Group, LLC
through a column packed with the powder of interest, which is the stationary phase.
As the probe molecules travel through the packed column, they adsorb to and desorb
from the powder. The rate and degree of this interaction is determined by the surface
chemistry of the powder and the probe molecules. Since the surface chemistry of the
probe molecules is known, this allows calculation of the surface chemistry (surface
Figure 15
Figure 14
Theory of Granulation 25
Characterizing wetting by dynamic contact angle goniometry. (From Refs. 4,25.)
Characterizing wetting by Washburn test and capillary rise. (From Refs. 4,23.)
© 2005 by Taylor & Francis Group, LLC
energies) of the powder with the help of a series of plots of both alkane and various
polar probes. A distinct advantage of IGC is the reproducible measurements of physical
chemical surface properties which control adhesion tension.
The strength of the solid/liquid interactions determines the average retention
time of a probe. Retention time data for each probe is converted into net retention
volumes, VN. The free energy of desorption is given by
DG ? RT ln VN ? c ? 2Na ffiffiffiffiffiffiffiffiffiffiffi gDS
gDL
q ? ffiffiffiffiffiffiffiffiffi gPS
gP
L q 
 ? c ?2:4?
where R is the universal gas constant, T is the column temperature, c is a system constant,
N is Avagadro’s number, and a is the surface area of a probe molecule. As illustrated
in Figure 16, a plot of RT ln VN vs. a ffiffiffiffiffi gDL
p should give a straight line for a
series of alkanes, the slope of which allows determination of the solid’s dispersive
surface energy gDS
. Plotting RT ln VN vs. a ffiffiffiffiffi gDL
p for the polar probes will give a point
that is generally somewhere above the alkane reference line. The polar solid energy
gPS
is then found from a plot of these deviations.
2.4. Granulation Examples of the Impact of Wetting
Wetting dynamics has a pronounced influence on initial nuclei distribution formed
from fine powder. As an example, the influence of contact angle on the average granangle
was varied by changing the percentages of hydrophilic lactose and hydrophobic
salicylic acid (31). Note that granule size in this study is actually nuclei size, since
little growth has taken place in the process. Nuclei size is seen to improve with contact
angle. In addition, the x-coordinate would more appropriately be replaced with
adhesion tension. Aulton et al. (32) also demonstrated the influence of surfactant
concentration on shifting nuclei size due to changes in adhesion tension.
drop profiles are imaged as a drop wets in to a formulation tablet. Note that the time
Figure 16
26 Ennis
ule size formed in fluid-bed granulation is illustrated in Figure 17. Water contact
Figure 18(A) illustrates an example of dynamic wetting, where a time series of
IGC retention times for glass powder. (From Refs. 4,5.)
© 2005 by Taylor & Francis Group, LLC
scale of wetting in this case is 2 sec, with nearly complete wet-in occurring in 1 sec.
This particular formulation was granulated on a continuous pan system in excess
a second lot—referred to as problem technical—experiences significantly degraded
granule strength and requires production rates to be substantially reduced. This is
associated with nearly twice the initial contact angle (120) and slower spreading
velocity when compared with the good technical. Poor wetting in practice can translate
into reduced production rates to compensate for increased time for drops to
work into the powder bed surface. Weaker granules are also often observed, since
poor wet phase interfacial behavior translates in part to poor solid bond strength
and high granule voidage. Note that differences in the lots are only observed over
the first 0.25–0.5 sec, illustrating the importance of comparing dynamic behavior
of formulations, after which time surfactant adsorption/desorption reduces the
contact angle.
As an example of Washburn approaches as
of fluid penetration rate and the extent of penetration on granule size distribution for
drum granulation was shown by Gluba et al. (Powder Hand. & Proc 1990; 2:323).
Increasing penetration rate, as reflected by Eq. 2.3 increased granule size and
decreased asymmetry of the granule size distribution.
2.5. Regimes of Nucleation and Wetting
ders consists of several important steps. First, droplets are formed related to spray
distribution, or spray flux defined as the wetted area of the bed per unit time.
Figure 17 The influence of contact angle on nuclei size formed in fluid-bed granulation of
lactose/salicylic acid mixtures. Powder contact angle determined by goniometry and percent
Theory of Granulation 27
of 2 tons/hr. Figure 18(B) compares differences in lots of the formulation. Note that
illustrated in Figure 19, the effect
As previously depicted in Figure 12, the wetting and nucleation process for fine powlactose
of each formulation are given in parentheses. (From Ref. 31.)
© 2005 by Taylor & Francis Group, LLC
Figure 18 Dynamic imaging of drop wetting, and its impact on continuous pan granulation:
(A) Dynamic image of a drop wetting into a formulation, with good active ingredient.
(B) Comparison of surface spreading velocity and dynamic contact angle vs. time for good
and bad active ingredients or technical. Bad technical required reduced production rates.
28 Ennis
(From Ref. 4.)
© 2005 by Taylor & Francis Group, LLC
Second, droplets impact and coalesce on the powder bed surface if mixing or wet-in
time is slow. Third, droplets spread and penetrate into the moving powder bed to
form loose nuclei, again coalescing, if wet-in is slow. In the case of high-shear processes,
shear forces break down overwet clumps, also producing nuclei.
Two key features control this wetting and nucleation process. One is the time
required for a drop to wet-in to the moving powder bed, in comparison to some circulation
time of the process. As discussed previously, this wet-in time is strongly
influenced by formulation properties (e.g., Eq. 2.3). The second is the actual spray
rate or spray flux, in comparison to solids flux, or mixing rates. Spray flux is strongly
influenced by manufacturing and process design.
One can envision that drop penetration time and spray flux define regimes of
nucleation and wetting. If the wet-in is rapid and spray fluxes are low, individual
drops will form discrete nuclei somewhat larger than the drop size, defining a droplet
controlled regime. At the other extreme, if drop penetration is slow and spray flux is
large, drop coalescence and pooling of binder material will occur throughout the
powder bed, which must be broken down by mechanical dispersion. In the mechanical
dispersion regime of nucleation, shear forces control the breakdown of wetting
clumps, independent of drop distribution.
Falling between these two extreme regimes, drop overlap and coalescence
occur to varying extent defining an intermediate regime of nucleation, being a function
of penetration time and spray flux. To better define wetting, particularly in the
sense of process engineering and scale-up, we will consider drop penetration or wet-in
time and spray flux in greater detail.
Beginning with penetration time, Eq. 2.3 defines key formulation properties
controlling capillary rise in powder beds. From considering a distribution of macrodetermined
a total drop penetration time tp of
tp ? 1:35
V2=3
d
e2
eff
m
Reff g cos y   ?2:5?
As shown previously, drop wet-in time decreases with increasing pore radius
Reff, decreasing binder viscosity and increasing adhesion tension. In addition, drop
penetration time decreases with decreasing drop size Vd and increasing bed porosity
eeff. Effective pore radius Reff is related to the surface-volume average particle size
d32, particle shape, and effective porosity of packing eeff by
Reff ?
jd32
3
eeff
1 
 eeff 
  ?2:6?
To remain with a droplet controlled regime of nucleation, the penetration time
given by Eq. 2.5 should be less than some characteristic circulation time tc of the
granulator in question. Circulation time is a function of mixing and bed weight,
and can change with scale-up.
powder bed of width B moves past a flat spray of spray rate (dV/dt) at a solids velocity
of w. For a given spray rate, the number of drops is determined by drop volume,
which in turn defines the drop area a per unit time which will be covered by the
spray, giving a spray flux of:
Theory of Granulation 29
and micropores in the moving powder bed, as shown in Figure 20, Hapgood (24)
Now turning our attention to spray flux, consider Figure 21 where an idealized
© 2005 by Taylor & Francis Group, LLC
Figure 19 Influence of capillary penetration on drum granule size: Increasing penetration
rate, as reflected by Eq. 2.3 increases granule size and decreases asymmetry of the granule size
distribution.
Figure 20
30 Ennis
Drop penetration in a moving powder bed. (From Ref. 5.)
© 2005 by Taylor & Francis Group, LLC
da
dt ?
dV=dt
Vd
pd2
d
4 
 ?
3
2
dV=dt ? ?
dd ?2:7?
As droplets contact the powder bed at a certain rate, the powder moves past
the spray zone at its own velocity, or at solids flux given for this simple example by
dA
dt ? Bw ?2:8?
The ratio of the droplet spray flux to the solids flux defines a dimensionless
spray flux given by
ca ?
da=dt ? ?
dA=dt ? ??
3
2
dV=dt ? ?
dd dA=dt ? ? ?2:9?
The dimensionless spray flux is the ratio of the rate at which wetted area is covered
by droplets to the area of flux of powder through the spray zone, and is a measure
of the density of drops falling on the powder surface. As with drop penetration
time, it plays a role in defining the regimes of nucleation as illustrated in Figure 22
(5,23). For small spray flux (ca  1), drops will not overlap on contact and will form
separate discrete nuclei for fast penetration time. For large spray flux (ca1), however,
significant drop overlap occurs, forming nuclei much larger than drop size, and
in the limit, independent of drop size.
Figure 22 Monte-Carlo simulations of drop coverage. (From Ref. 5.)
Figure 21
Theory of Granulation 31
Idealized flat spray zone in a spinning riffle granulator. (From Ref. 5.)
© 2005 by Taylor & Francis Group, LLC
For the case of random drop deposition as described by a Poisson distribution,
Hapgood (23) showed the fraction of surface covered by spray was given by
fsingle ? 1 
 exp?
ca? ?2:10?
In addition, the fraction of single drops forming individual nuclei (assuming
rapid drop penetration) vs. the number of agglomerates formed was given by
fsingle ? exp?
4ca? ?2:11?
fagglom ? 1 
 exp?
4ca? ?2:12?
Examples of the above as applied to nucleation are depicted in Figures 23 and
tion of drop size and spray flux. Lactose was sprayed with a flat spray in a spinning
spray flux of , ca?0.22, a clear relationship is seen between nuclei size and spray
distribution, with nuclei formed somewhat larger than drop size as shown in Figure
23. However, as the speed of the riffles is slowed down (i.e., solids velocity and solids
flux are decreased, and dimensionless spray flux increased), the nuclei distribution
The spray flux captures the impact of equipment operating variables on nucleation,
and as such is very useful for scale-up if nucleation rates and nuclei sizes are to
be maintained constant. The overall impact of dimensionless spray flux on nucleation
and agglomerate formation is illustrated in with agglomerates
increasing with increased spray flux as clearly governed by Eq. 2.12 for the case of
rapid drop penetration.
Figure 23 Effect of spray drop distribution (low spray flux) on nuclei distribution: Lactose
32 Ennis
25,
feed powder in spinning granulator. (From Ref. 33.)
24 and described by Litster et al. (33). Here, nuclei distribution was studied as a funcwidens
with the formation of agglomerates as depicted in Figure 24.
riffle granulator, mimicking the geometry of Figure 21. For a moderate, intermediate
Figure
© 2005 by Taylor & Francis Group, LLC
the help of dimensionless drop penetration time tp and spray flux ca, or
tp ?
tp
tc ?
penetration time
circulation time
and ca ?
da=dt ? ?
dA=dt ? ??
spray flux
solids flux ?2:13?
A droplet controlled nucleation regime occurs when there is both low spray
flux—relatively few drops overlap—and fast droplet penetration—drops wet into
Figure 24 Effect of powder velocity (variable spray flux—water) on nuclei distribution:
Figure 25 Agglomerate formation vs. spray flux: Lactose powder with water and HPLC
Theory of Granulation 33
The regimes of nucleation may now be defined, as depicted in Figure 26 with
Lactose feed powder in spinning granulator. (From Ref. 33.)
solutions. (From Ref. 5.)
© 2005 by Taylor & Francis Group, LLC
the bed completely before bed mixing allows further drop contact. Nuclei will be
formed of the order of drop size.
A mechanical dispersion regime occurs at the other extreme of high spray flux—
giving large drop overlap and coalescence, and large drop penetration times, promoted
by poor wet-in rates and slow circulation times and poor mixing. In this
regime, nucleation and binder dispersion occur by mechanical agitation. Viscous,
poorly wetting binders are slow to flow through pores in the powder bed in the case
of poor penetration time. Drop coalescence on the powder surface occurs (also
known as ‘‘pooling’’) creating very broad nuclei size distributions. Binder solution
delivery method (drop size, nozzle height) typically has minimal effect on the nuclei
size distribution, though interfacial properties may affect nuclei and final granule
strength.
An intermediate regime exists for moderate drop penetration times and moderate
spray flux, with the resulting nuclei regime narrowing with decreases in both.
There are several implications with regard to the nucleation regime map of
Figure 26 with regard to trouble shooting of wetting and nucleation problems. If
drop penetration times are large, making adjustments to spray may not be suffi-
ciently narrow granule size distributions if remaining in the mechanical regime. Significant
changes to wetting and nucleation only occur if changes take the system
across a regime boundary. This can occur in an undesirable way if processes are
not scaled, with due attention to remaining in the drop controlled regime.
3. GRANULE GROWTH AND CONSOLIDATION
3.1. Overview
bution of a particulate feed in a granulation process is controlled by several mechanisms,
including wetting, nucleation, and coating; granule coalescence and
consolidation; and granule breakage. Nucleation of fine powders and coating of
existing granules by the fluid phase have been discussed in the previous section.
Figure 26 A possible regime map of nucleation, relating spray flux, solids mixing (solids flux
34 Ennis
As previously outlined in Figures 4 and 10, the evolution of the granule size distriand
circulation time), and formulation properties. (From Refs. 5,24.)
© 2005 by Taylor & Francis Group, LLC
Breakage mechanisms will be treated in the following section. Here, we focus particularly
on growth and consolidation mechanisms.
Granule growth includes the coalescence of existing granules as well as the
layering of fine powder onto previously formed nuclei or granules. The breakdown
of wet clumps into a stable nuclei distribution can also be included among coalescence
mechanisms. As granules grow by coalescence, they are simultaneously compacted
by consolidation mechanisms, which reduce internal granule voidage or
porosity.
There are strong interactions between these rate processes, as illustrated in
Figure 27 for the case of drum granulation of fine feed. Here, granule size is illustrated
to progress through three stages of growth, including rapid, exponential
growth in the initial nucleation stage, followed by linear growth in the transition
stage, finishing with very slow growth in a final balling stage. Simultaneously with
growth, granule porosity or voidage is seen to decrease with time as the granules
are compacted. Granule growth and consolidation are intimately connected;
increases in granule size are shown here to be associated with a decrease in granule
porosity. This is a dominant theme in wet granulation.
As originally outlined in Ennis (3), these growth patterns are common throughout
fluidized-bed, drum, pan, and high-shear mixer processes for a variety of formulations.
Specific mechanisms of growth may dominate for a process—sometimes to
the exclusion of others (4,5,11,12). However, what all processes have in common
is that the prevailing mechanisms are dictated by a balance of critical particle level
properties, which control formulation deformability, and operating variables which
control the local level of shear, or bed agitation intensity.
3.2. Mechanics of the Growth Rate Process
In order for two colliding granules to coalesce rather than break up, the collisional
Figure 27 Granule porosity and mean (pellet) size: Typical regimes of granule growth and
Theory of Granulation 35
kinetic energy must first be dissipated to prevent rebound as illustrated in Figure 28.
consolidation. (Adapted from Refs. 15–19.)
© 2005 by Taylor & Francis Group, LLC
In addition, the strength of the bond must resist any subsequent breakup forces in
the process. The ability of the granules to deform during processing may be referred
to as the formulation’s deformability, and deformability has a large effect on growth
rate. Increases in deformability increase the bonding or contact area thereby dissipating
and resisting breakup forces. From a balance of binding and separating forces
and torque acting within the area of granule contact, Ouchiyama and Tanaka (34)
derived a critical limit of size above which coalescence becomes impossible, or a maximum
growth limit given by
Dc ? ?AQ3B=2K3=2sT?1=?4
?3=2?Z? ?3:1?
Here, K is deformability, a proportionality constant relating the maximum
compressive force Q to the deformed contact area, A is a constant with units of
[L3/F], which relates granule volume to impact compression force, and sT is the tensile
strength of the granule bond. Granules are compacted as they collide. This expels
pore fluid to the granule surface thereby increasing local liquid saturation in the contact
area of colliding granules. This surface fluid (1) increases the tensile strength of
the liquid bond sT and (2) increases surface plasticity and deformability K.
Dc represents the largest granule that may be grown in a granulation process,
and it is a harmonic average granule size. Therefore, it is possible for the collision of
two large granules to be unsuccessful, their average being beyond this critical size,
whereas the collision of a large and small granule leads to successful coalescence.
The growth limit Dc is seen to increase with increased formulation deformability
(which will be shown to be a strong function of moisture and primary particle size
distribution), increased compressive forces (which are related to local shear levels
in the process), and increased tensile forces (which are related to the interparticle
forces previously discussed.) The parameters z and Z depend on the deformation
mechanism within the contact area. For plastic deformation, z?1, Z?0, and K /
1/H where H is hardness. For elastic, Hertzian deformation, z?2/3, Z?2/3, and
K / (1/E)2/3 where E is the reduced elastic modulus. Granule deformation is initially
dominated by inelastic behavior of contacts during collision (35).
Figure 28 Mechanisms of granule coalescence for low- and high-deformability systems:
Rebound occurs for average granule sizes greater than the critical granule size Dc.
36 Ennis
K?deformability. (From Refs. 1,2.)
© 2005 by Taylor & Francis Group, LLC
3.3. Types of Granule Growth
The importance of deformability to the growth process depends on bed agitation
intensity. If little deformation takes place during granule collisions, the system is
referred to as a low deformability or low agitation intensity process. This generally
includes fluid bed, drum, and pan granulators. Growth is largely controlled by the
extent of any surface fluid layer and surface deformability, with surface fluid playing
a large role in dissipating collisional kinetic energy. Growth generally occurs at a faster
time scale than overall granule deformation and consolidation. This is depicted in
Figure 29, where smaller granules can still be distinguished as part of a larger granule
structure, or a ‘‘pop-corn’’-type appearance as often occurs in fluid-bed granulation.
(Note that such a structure may not be observed if layering and nucleation alone
dominate with little coalescence of large granules.) In this case, granule coalescence
and consolidation have less interaction than they do with high deformability systems,
making low deformability–low agitation systems easier to scale and model.
For high shear rates or bed agitation intensity, large granule deformation
occurs during granule collisions, and granule growth and consolidation occur on
the same time scale. Such a system is referred to as a deformable or high agitation
intensity process, and it generally includes continuous pin and plow shear-type mixers,
as well as batch high-shear pharmaceutical mixers. In these cases, substantial
collisional kinetic energy is dissipated with deformation of the wet mass composing
the granule. Rather than a ‘‘sticking’’ together process as often occurs in the low
deformability process of fluid beds, granules are ‘‘smashed’’ together as with a highshear
mixer, and smaller granules are not distinguishable with the granule structure,
as depicted in Figure 29. High-agitation, high-deformable processes generally produce
denser granules than those with low deformablity and low agitation intensity. In addition,
the combined and competing effects of granule coalescence and consolidation
make high-agitation processes difficult to model and scale. Both coalescence and
consolidation increase with increase in shear and deformability, whereas as granules
densify, they become less deformable, which works to lower coalescence in the later
stages of growth.
Figure 29 Granule structures resulting from (A) low- and (B) high-deformability systems,
Theory of Granulation 37
typical for fluid-bed and high-shear mixer granulators, respectively. (From Ref. 4.)
© 2005 by Taylor & Francis Group, LLC
Bedagitation intensity is controlled bymechanical variables of the process such as
fluid-bed excess gas velocity, or mixer impeller and chopper speed. Agitation intensity
controls the relative collisional and shear velocities of granules within the process and
therefore growth, breakage, consolidation, and final product density. Figure 30 summarizes
typical characteristic velocities, agitation intensities,andcompaction pressures,
and product relative densities achieved for a variety of size-enlargement processes.
Last, it should be noted that the process or formulation itself cannot define
whether it falls into a low- or high-agitation intensity process. As discussed more
fully in the following sections, it is a function of both the level of shear as well as
the formulation deformability. A very stiff formulation with low deformability may
behave as a low deformability system in a high-shear mixer, or a very pliable formulation
may act as a high-deformable system in a fluid-bed granulator.
3.4. Granule Deformability
Granule deformability—and hence Dc—is a strong function of moisture, as illu-
K is related to both the yield strength of the material sy, i.e., the ability of the material
to resist stresses, and the ability of the surface to be strained without degradation
or rupture of the granule, with this maximum allowable critical deformation strain
c
of agglomerates during compression as a function of liquid saturation, with strain
denoted by DL/L. In general, high deformability K requires low yield strength sy
and high critical strain (DL/L)c. For this formulation, increasing moisture increases
deformability by lowering interparticle frictional resistance, which also increases
mean granule size (Fig. 31).
In most cases, granule deformability increases with increasing moisture,
decreasing binder viscosity, decreasing surface tension, decreasing interparticle
Figure 30 Classification of agglomeration processes by agitation intensity and compaction
38 Ennis
strated in Figure 31 by the marked increase in average granule size. Deformability
denoted by (DL/L) . Figure 32 illustrates the low shear rate stress–strain behavior
pressure. (From Refs. 1,2,5.)
© 2005 by Taylor & Francis Group, LLC
friction, and increasing average particle size (specifically dsv, or the surface–volume
mean size), as well as increasing bed agitation intensity. The important contributions
of binder viscosity and friction to granule deformability and growth are illustrated
Note that 60% of the energy is dissipated through viscous losses, with the majority of
the remainder through interparticle friction. Very little is loss due to capillary forces
controlled by surface tension. Therefore, modern approaches to granule coalescence
rest in understanding the impact of granule deformability on growth, rather than the
original capillary framework alone, as put forth by Rumpf (20,21) regarding pendular
and funicular forces due to interparticle liquid bridges.
3.5. Interparticle Forces Due to Pendular Bridges
Interstitial fluid and the resulting pendular bridges play a large role in both granule
growth and granule deformability. Pendular bridges between particles of which a
granule is composed, give rise to capillary and viscous interparticle forces. These
forces develop interparticle pressure, which also allows friction to act between point
deformability, warrant further attention. Note that capillary forces for small contact
Figure 31 Effect of granule saturation on mean granule diameter, indicating the marked
increase in granule deformability. Mean granule diameter is a measure of the critical limit
of size Dc. Granulation of calcium hydrogen phosphate–aqueous binder solutions. Fielder
Theory of Granulation 39
by fractions of energy dissipated during a granule collision as depicted in Figure 33.
contacts (Fig. 34). Interparticle forces due to pendular bridges, and their impact on
PMAT 25 VG mixer. (From Ref. 36.)
© 2005 by Taylor & Francis Group, LLC
angle attract particles (but repel for y>90), whereas viscous and frictional forces
act to resist the direction of motion.
separated by a gap distance 2ho approaching one another at a velocity U. The two
particles may represent two primary particles within the granule, in which case we
are concerned about the contribution of interparticle forces to granule strength
and deformability. Or they may represent two colliding granules, in which case we
are concerned with the ability of the pendular bridge to dissipate granule kinetic
energy and resist breakup forces in the granulation process. The two spheres are
bound by a pendular bridge of viscosity m, density r, and surface tension g. The
pendular bridge consists of the binding fluid in the process, which includes the added
solvent and any solubilized components. In some cases, it may also be desirable to
include very fine solid components within the definition of the binding fluid, and,
therefore, consider instead a suspension viscosity and surface tension. These material
parameters vary on a local level throughout the process, and are also time dependent
and a function of drying conditions.
For the case of a static liquid bridge (i.e., U?0) with perfect wetting, surface
tension induces an attractive capillary force between the two particles due to a threephase
contact line force and a pressure deficiency arising from interfacial curvature.
The impact of this static pendular bridge force on static granule strength has been
studied and reported extensively (3,20–22). It is important to recognize that in most
processes, however, the particles are moving relative to one another and, therefore,
the bridge liquid is in motion. This gives rise to viscous resistance forces which can
contribute significantly to the total bridge strength. The strength of both Newtonian
Figure 32 The influence of sample saturation S on granule deformability: Deformation
strain (DL/L) is measured as a function of applied stress, with the peak stress and strain
denoted by tensile strength sy and critical strain (DL/L)c of the material. Dicalcium phosphate
with 15 wt% binding solution of PVP/PVA Kollidon
 VA64. Fifty percent compact porosity.
40 Ennis
As shown in Figures 11 and 34, consider two spherical particles of radius a
(From Ref. 37.)
© 2005 by Taylor & Francis Group, LLC
and non-Newtonian pendular bridges have been studied extensively (3,39,40). For
Newtonian fluids (39), the dynamic strength was shown to be given by
F
pga ? Fcap ? Fvis ? Fo ? 3Ca=e where
Fcap ? ?2 
 2Ho? sin2 j
Fvis ? 3Ca=e
Ca ? mU=g
8<:
?3:2?
Here, all forces have been made dimensionless with respect to a measure of the
capillary force, or pga. Note that the bridge strength consists of two components.
Fcap is the strength of a static capillary bridge, and is a function of curvature of
Figure 33 Distribution of energy dissipation during agglomerate collisions, with granular
simulations of wall impact for 128 ms duration for invisicid and viscous binder agglomerates.
Figure 34 Interparticle forces and granule deformability: Interparticle forces include
Theory of Granulation 41
(From Ref. 11.)
capillary forces, viscous lubrication forces, and frictional forces. (From Ref. 4.)
© 2005 by Taylor & Francis Group, LLC
the interface Ho and the filling angle. In dimensional form, it is given by
F cap ? pga?2 
 2Ho? sin2 j ?3:3?
Fvis is the strength of a viscous, dynamic bridge, and is equivalent to the force
between two spheres approaching one another in an infinite fluid. This force is a
function of binder viscosity m, and the collision velocity U. Here, e?2ho/a is gap distance,
and not granule voidage. In dimensional form, the viscous force is given by
F vis ? 6pmUa=e ?3:4?
From Eq. 3.2, one finds that the dynamic bridge force begins with the static
bridge strength—which is a constant independent of velocity (or Ca), and then
increases linearly with Ca, which is a capillary number representing the ratio of viscous
(pmUa) to capillary (pga) forces, and is proportional to velocity. This is con-
firmed experimentally as illustrated in Figure 35 for the case of two spheres
approaching axially. Extensions of the theory have also been conducted for
non-Newtonian fluids (shear thinning), shearing motions, particle roughness, wettability,
and time-dependent drying binders. The reader is referred to Ennis (3,39)
for additional details. Curves for non-Newtonian fluids are also included in
Figure 35.
For small velocities, small binder viscosity, and large gap distances, the
strength of the bridge will approximate a static pendular bridge, or Fcap, which is
proportional to and increases with increases in surface tension. This force is equivalent
to the static pendular force H previously given in Eq. 1.2 as studied by Rumpf
(20,21) and Augsburger and Vuppala (22). On the other hand, for large binder viscosities
and velocities, or small gap distances, the bridge strength will approximately
be equal to Fvis, which is proportional to and increases with increases in binder viscosity
and velocity. This viscous force is singular in the gap distance and increases
dramatically for small separation of the particles. It is important to note that as
Figure 35 Maximum strength of a liquid bridge between two axial moving particles as
42 Ennis
a function of Ca for Newtonian and shear thinning fluids. (From Ref. 3.)
© 2005 by Taylor & Francis Group, LLC
granules are consolidated, resulting in decreases in effective interparticle gap distance,
and as binders dry, resulting in large increases in binder viscosity, the dynamic
bridge strength can exceed the static strength by orders of magnitude.
3.6. Dynamic Wet Mass Rheology and Deformability Revisited
the case of slow yielding. However, the dependence of interparticle forces on shear
rate clearly impacts wet mass rheology and therefore deformability. Figure 36 illustrates
the stress–strain response of compacts, demonstrating that the peak flow or
yield stress increases proportionally with compression velocity (41). In fact, in a similar
fashion to dynamic liquid bridge forces, the peak flow stress of wet unsaturated
compacts (initially pendular state) can be seen to also increase with Ca as follows
sPeak
y
g=a ? so ? ACa
B
where
so ? 5:0 
 5:3
A ? 280 
 320; B ? 0:58 
 0:64
Ca ? m_ea=g
8<:
?3:5?
There are several important issues worth noting with regard to these results.
First is the similarity between the strength of the assembly or compact given by
Eq. 3.5 and the strength of the individual dynamic pendular bridge given by Eq.
3.2; both curves are similar in shape with a capillary number dependency. As with
the pendular bridge, two regions may be defined. In region 1, for a bulk capillary
number of Ca < 10
4, the strength or yield stress of the compact depends on the static
pendular bridge, and therefore on surface tension, particle size, and liquid load-
Figure 36 Typical compact stress response for fast compression vs. crosshead compression
velocity for glass ballotini (d32?35 mm) and compact diameter 20 mm, length 25 mm. (From
Theory of Granulation 43
The yield stress of powder compacts has previously been reported in Figure 32 for
(Fig. 37):
Ref. 41.)
© 2005 by Taylor & Francis Group, LLC
ing. In region 2, for Ca > 10
4, the strength depends on the dynamic pendular
bridge, and therefore on binder viscosity and strain rate, in addition to particle size.
Second, the results of Figure 37 do not clearly depict the role of saturation and
compact porosity, though these properties are known to affect strength. Decreases
in compact porosity generally increases compact strength through increases in inter-
(36,37,42). Hence, the curve of Figure 37 should be expected to shift with these variables,
particularly since the viscous force for axial approach is singular in the interparticle
gap distance (Eqs. 3.4).
Third, the static granule or assembly strength (Eq. 1.2), as originally developed
by Rumpf (20,21) and Augsburger and Vuppala (22), is captured by the constant,
low Ca value of the yield stress or
sPeak
y  so g=a ? ?
9
8
1 
 e
e 
 H
a2 
9
4
1 
 e
e 
 g cos y
a
for Ca < 10
4 ?3:6?
Fourth, the mechanism of compact failure also depends on strain rate.
function of capillary number. At low Ca, compacts fail by brittle fracture with
macroscopic crack propagation, whereas at high Ca, compacts fail by plastic flow.
Within the context of granulation, small yield stresses at low Ca may result in
unsuccessful growth when these stresses are compared with large breakup forces.
With increased yield stress comes stronger granules but also decreased deformability.
Therefore, high strength might imply a low-deformability growth mechanism for
low-shear processes such as a fluid bed. On the other hand, it might imply smaller
growth rates for high-shear processes which are able to overcome this yield stress
and bring about kneading action and plastic flow in the process. Therefore, it is
important to bear in mind that increased liquid saturation may initially lower yield
stress, allowing more plastic deformation during granule collisions. However, as
granules grow and consolidate and decrease in voidage, they also strengthen and rise
Figure 37
44 Ennis
Dimensionless peak flow stress of Fig. 36 vs. bulk capillary number, for various
particle friction, whereas increases in saturation lower strength (e.g., Figure 32)
Figure 38 illustrates schematically the crack behavior observed in compacts as a
binder solutions. (From Ref. 41.)
© 2005 by Taylor & Francis Group, LLC
in yield stress, becoming less deformable with time and withstanding shear forces in
the granulator. Hence, the desired granule strength and deformability is linked in
a complex way to granulator shear forces and consolidation behavior.
3.7. Low-Agitation Intensity–Low-Deformability Growth
For those low-agitation processes or formulations which allow little granule deformation
during granule collisions, consolidation of the granules occurs at a much
slower rate than growth, and granule deformation can be ignored to a first approximation.
The growth process can be modeled by the collision of two nearly stiff granules
each coated by a liquid layer of thickness h. A pictorial depiction of a granule
collision is illustrated in Figure 38. For the case of zero plastic deformation, the
probability of successful coalescence as developed by Ennis and coworkers (3,43)
is governed by a dimensionless energy, or Stokes number St given by
St ?
4ru0a
9m ?3:7?
where u0 is the relative collisional velocity of the granules, r is granule density, a is
the harmonic average of granule diameter, and m is the solution phase binder viscosity.
The Stokes number represents the ratio of initial collisional kinetic energy to the
energy dissipated by viscous lubrication forces, and it is one measure of normalized
bed agitation energy. Successful growth by coalescence or layering requires that
St < St where St ? 1 ?
1
er 
 ln?h=ha? ?3:8?
where St is a critical Stokes number representing the energy required for rebound.
The binder layer thickness h is related to liquid loading, er is the coefficient of restitution
of the granules, and ha is a measure of surface roughness or asperities. The
critical condition given by Eq. 3.8 controls the growth of low deformability systems.
This criteria has also been extended to capillary coalescence (3) and for cases
of plastic deformation (44).
Binder viscosity is a function of local temperature, collisional strain rate (for
non-Newtonian binders), and binder concentration dictated by drying rate and local
Figure 38 Collisions between surface wet granules, beginning with approach, and ending
Stokes model.
Theory of Granulation 45
with separation. (From Refs. 3,5,43,44.) Note that no deformation takes place in the original
© 2005 by Taylor & Francis Group, LLC
mass transfer and local bed moisture. It can be controlled through judicious selection
of binding and surfactant agents and measured by standard rheological techniques
(45). The collisional velocity is a function of process design and operating variables,
and is related to bed agitation intensity and mixing. Possible choices of u0 are sum-
Note that u0 is an interparticle collisional velocity, which is not necessarily the local
average granular flow velocity.
Three regimes of granule growth may be identified for low-agitation intensity–
low-deformability processes (3,43), as depicted for fluid-bed granulation in Figure 39,
and outlined as follows:
Noninertial regime: For small granules or high binder viscosity lying within a
noninertial regime of granulation, all values of St will lie below the critical
value St and therefore all granule collisions result in successful
growth provided binder is present. Growth rate is independent of granule
kinetic energy, particle size, and binder viscosity (provided other rate
processes are constant). Distribution of binding fluid and degree of mixing
then control growth and, and this is strongly coupled with the rate
process of wetting (see previous section). As shown in Figure 39, both
binders have the same initial growth rate for similar spray rates, independent
of binder viscosity. Increases in bed moisture (e.g., spray rate, drop
rate) and increases in granule collisions in the presence of binder will
increase the overall rate of growth. It must be borne in mind, however,
that there is a 100% success of these collisions, since dissipation of energy
far exceeds collisional kinetic energy.
Inertial regime: As granules grow in size, their collisional momentum increases,
leading to localized regions in the process where St exceeds the critical
value St. In this inertial regime of granulation, granule size, binder
Figure 39 Median granule diameter for fluid-bed granulation of ballotini with binders of dif-
46 Ennis
marized in Figure 30, and discussed more fully regarding scale-up in Chapter 18.
ferent viscosity indicating regimes of growth. (From Refs. 3,43.)
© 2005 by Taylor & Francis Group, LLC
viscosity, and collision velocity determine the proportion of the bed in
which granule rebound is possible. Increases in binder viscosity and
decreases in agitation intensity increase the extent of granule growth—
i.e., the largest granule that may be grown [e.g., Dc of Eq. 3.1]. This is
confirmed in with the CMC binder continuing to grow,
whereas the PVP system with lower viscosity slows down in growth. Note
that the rate of growth, however, is controlled by binder distribution and
mixing, and not binder viscosity. For example, increasing binder viscosity
will not affect growth rate, or initial granule size, but it will result in
an increased growth limit.
Coating regime: When the spatial average of St exceeds St, growth is balanced
by granule disruption or breakup, leading to the coating regime of granulation.
Growth continues by coating of granules by binding fluid alone.
The PVP system with lower viscosity is seen to reach its growth limit and
therefore coating regime in Figure 39.
Transitions between granulation regimes depend on bed hydrodynamics. As
demonstrated by Ennis et al. (3,4,43), granulation of an initially fine powder may
exhibit characteristics of all three granulation regimes as time progresses, since St
increases with increasing granule size. Implications and additional examples regarding
the regime analysis are highlighted by Ennis (3,4,43). In particular, increases in
fluid-bed excess gas velocity exhibit a similar but opposite effect on growth rate
to binder viscosity; namely, it is observed to not affect growth rate in the initial
Noninertial regime of growth, but instead lowers the growth limit in the inertial
regime.
Figure 40 (A) Exponential growth in drum granulation reaching a growth limit dmax, or
maximum extent of growth (kt)max, which are functions of moisture saturation (15–19). (B)
Maximum extent of noninertial growth (kt)max as a linear function of saturation of the powder
feed and binder viscosity. (C) Maximum extent normalized for differences in binder viscosity,
Theory of Granulation 47
Figure 39
drum speed, granule density by Stokes number. (From Ref. 46.)
© 2005 by Taylor & Francis Group, LLC
3.8. Example: Extent of Noninertial Growth
Growth by coalescence in granulation processes may be modeled by population
nel which describe growth. For fine powders within the noninertial regime of growth,
all collisions result in successful coalescence provided binder is present. Coalescence
occurs via a random, size independent kernel which is only a function of liquid loading
y, as well as mixing, or
b?u; v? ? k ? kf ?y? ?3:9?
For random growth and in the presence of sufficient binding fluid, it may be
rigorously proven that the average granule size increases exponentially with time, or
d ? d0 ekt ?3:10?
This exponential increase in size with time is confirmed experimentally in
granule saturation S is connected to liquid loading y and porosity.) Based on the
regime analysis work above, growth will continue in a process while the conditions
of Eq. 3.8 are met, i.e., dissipation exceeds collisional kinetic energy. Examples of
these growth limits are seen in the drum granulation work of Kapur (15–19) in
the maximum extent of granulation (kt)max occurring within the noninertial regime is
given by
Figure 41 Granule diameter as a function of time for high-shear mixer granulation, illustrating
the influence of deformability on growth behavior. Directions of increasing viscosity and
impeller speed indicated by arrows. (A) Ten-liter melt granulation of lactose with 15 wt% binder
and impeller speed of 1400 rpm for two different viscosity grades of polyethylene gylcol
binders. (B) Ten-liter vertical high-shear mixer granulation of dicalcium phosphate with
15 wt% binder solution of PVP/PVA Kollidon
 VA64, liquid loading of 16.8 wt%, and chop-
48 Ennis
Figure 40, where increases in liquid loading f (y) increase growth rate. (Note that
Figure 40, as well as fluid beds (Fig. 39) and mixers (Fig. 41). It may be shown that
balances (Chapter 21). It is necessary to determine both the mechanism and the kerper
speed of 1000 rpm for varying impeller speed. (From Refs. 47,48.)
© 2005 by Taylor & Francis Group, LLC
?kt?max ? 6 ln?St=St0? f ?y? / ln
m
ru0d0 
  ?3:11?
St0 is the Stokes number based on initial nuclear diameter d0 (46). Extent
(kt) max is taken as the logarithm of the growth limit in the first random stage of
growth, or dmax. The growth limits dmax
in Figure 40(B). Here, (kt)max is observed to depend linearly on liquid loading y.
Therefore, the maximum granule size depends exponentially on liquid loading, as
From Eq. 3.11, it is possible to scale or normalize a variety of drum granulation
data to a common drum speed and binder viscosity. Maximum granule size dmax
and extent (kt)max depend linearly and logarithmically, respectively, on binder
viscosity and the inverse of agitation velocity. This is illustrated in Figure 40(B),
where the slope of each formulation line depends linearly on binder viscosity.
Figure 40 illustrates the normalization of extent (kt)max for the drum granulation
of limestone and fertilizers, correcting for differences in binder viscosity, granule
density, and drum rotation speed, with the data collapsing onto a common line
following normalization.
3.9. High-Agitation Intensity–Deformable Growth
For high-agitation processes involving high-shear mixing or for readily deformable
formulations, granule deformability, plastic deformation, and granule consolidation
can no longer be neglected as they occur at the same rate as granule growth. Typical
growth are evident, which reveal the possible effects of binder viscosity and impeller
speed, as shown for data replotted vs. impeller speed in Figure 42. The initial, nonequilibrium
stage of growth is controlled by granule deformability, and is of most
Figure 42 Granule diameter as a function of impeller speed for both initial nonequilibrium
Theory of Granulation 49
of Figure 40(A) are replotted as extents
observed experimentally (Fig. 31).
growth profiles for high-shear mixers are illustrated in Figure 41. Two stages of
and final equilibrium growth limit for high-shear mixer granulation (From Ref. 10,48.)
© 2005 by Taylor & Francis Group, LLC
practical significance in manufacturing for high-shear mixers. Increases in St due to
lower viscosity or higher impeller speed increase the rate of growth, since the system
becomes more deformable and easier to knead into larger granule structures (e.g.,
based on rigid, low-deformability granules, where high viscosity and low velocity
increase the growth limit.
Growth continues until disruptive and growth forces are balanced in the process.
This last equilibrium stage of growth represents a balance between dissipation
and collisional kinetic energy, and so increases in St decrease the final granule size, as
expected from the Stokes analysis. Note that the equilibrium granule diameter
decreases with the inverse square root of the impeller speed, as it should based on
St?St, with u0?a(du/dx)?oa.
The Stokes analysis is used to determine the effect of operating variables and
binder viscosity on equilibrium growth, where disruptive and growth forces are
balanced. In the early stages of growth for high-shear mixers, the Stokes analysis
in its present form is inapplicable. Freshly formed, uncompacted granules are easily
deformed, and as growth proceeds and consolidation of granules occur, they will
surface harden and become more resistant to deformation. This increases the importance
of the elasticity of the granule assembly. Therefore, in later stages of growth,
older granules approach the ideal Stokes model of rigid collisions. In addition, the
Stokes number controls in part the degree of deformation occurring during a collision
since it represents the importance of collision kinetic energy in relation to viscous
dissipation, although the exact dependence of deformation on St is presently
unknown.
The Stokes coalescence criteria of Eq. 3.8 developed by Ennis must be generalized
to account for substantial plastic deformation in order to treat the initial nonequilibrium
stages of growth in high-agitation systems such as high-shear mixers. In
this case, granule growth and deformation are controlled by a generalization of St,
or a deformation Stokes number Stdef, as originally defined by Tardos et al. (49,50):
Stdef ?
ru2
o
sy ?
r?duo.dx?2d2
sy ?3:12?
Viscosity has been replaced by a generalized form of plastic deformation controlled
by the yield stress sy, which may be determined by compression experiments.
As shown previously, yield stress is related to deformability of the wet mass, and is a
function of shear rate, binder viscosity, and surface tension, primary particle size and
Critical conditions required for granule coalescence may be defined in terms of
the viscous and deformation Stokes number, or St and Stdef, respectively. These
represent a complex generalization of the critical Stokes number given by Eq. 3.9,
and are discussed in detail elsewhere (5,44).
An overall view of the impact of deformability on growth behavior may be
in a regime map, and yield stress has been measured by compression experiments
(51). Growth mechanism depends on the competing effects of high-shear promoting
growth by deformation on the one hand and the breakup of granules giving a growth
limit on the other. For high velocities, growth is not possible by deformation due to
high shear, and the material remains in a crumb state. For low pore saturation,
50 Ennis
Fig. 29). These effects are contrary to what is predicted from the Stokes analysis
friction, and saturation and voidage as previously given by Eq. 3.5 and Figures 32
gained from Figure 43, where types of granule growth are plotted vs. deformability
© 2005 by Taylor & Francis Group, LLC
and 37.
growth is only possible by initial wetting and nucleation, with the surrounding
powder remaining ungranulated. At intermediate levels of moisture, growth occurs
at a steady rate for moderate deformability, but has a delay in growth for low
deformability. This delay, or induction time, is related to the time required to
work moisture to the surface to promote growth. For high moisture, very rapid
and potentially unstable growth occurs.
The currrent regime map as presented requires considerable development.
Overall growth depends on the mechanics of local growth, as well as the overall mixing
pattern and local/overall moisture distribution. Levels of shear are poorly understood
in high-shear processes. In addition, growth by both deformation and the rigid
growth model is possible. Lastly, deformability is intimately linked to both voidage
and moisture. They are not a constant for a formulation, but depend on time and the
growth process itself through the interplay of growth and consolidation. Nevertheless,
the map provides a starting point, and is discussed in additional detail in
3.10. Determination of St
The extent of growth is controlled by some limit of granule size, either reflected by
the critical Stokes number St or by the critical limit of granule size Dc. There are
several possible methods to determine this critical limit. The first involves measuring
the critical rotation speed for the survival of a series of liquid binder drops during
drum granulation (3). A second refined version involves measuring the survival of
granules in a couette-fluidized shear device (49,50). Both the onset of granule deformation
and the complete granule rupture are determined from the dependence of
granule shape and the number of surviving granules, respectively, on shear rate.
The critical shear rate describing complete granule rupture defines St, whereas
the onset of deformation and the beginning of granule breakdown defines an
Figure 43 Regime map of growth mechanisms, based on moisture level and deformabilty of
Theory of Granulation 51
Chapter 18.
formulations. (From Ref. 51.)
© 2005 by Taylor & Francis Group, LLC
additional critical value (Fig. 44). The third approach is to measure the deviation in
the growth rate curve from random exponential growth (52). The deviation from
random growth indicates a value of w, or the critical granule diameter at which
noninertial growth ends. This value is related toDc
Figure 44 Determination of the onset of granule deformation and complete granule breakdown
with the fluidized couette constant shear device. Y is a yield number, Fsur is the fraction
of surviving granules, andDdef
Figure 45 Determination of critical granule diameter, or growth limit, from the evolution of
52 Ennis
(Fig. 45). (See Chapter 21 regarding
is the average degree of granule deformation. (From Refs. 49,50.)
the granule size distribution. (From Ref. 51.)
© 2005 by Taylor & Francis Group, LLC
modeling for further discussion.) The last approach is through the direct measurement
of the yield stress through compression experiments.
3.11. Example: High-Shear Mixer Growth
An important case study for high-deformability growth was conducted by Holm et al.
(42) for high-shear mixer granulation. Lactose, dicalcium phosphate, and dicalcium
phosphate/starch mixtures (15% and 45% starch) were granulated in a Fielder
PMAT 25 VG laboratory scale mixer. Granule size, porosity, power level, temperature
rise, and fines disappearance were monitored during liquid addition and wet
massing phases. Impeller and chopper speeds were kept constant at 250 and 3500
rpm, respectively, with 7.0–7.5 kg starting material. Liquid flow rates and amount
of binder added were varied according to the formulation.
Figure 46 illustrates typical power profiles during granulation, whereas
wet massing time (as opposed to total process time) is defined as the amount of time
following the end of liquid addition, and the beginning of massing time is indicated
in Figure 47.
Clear connections may be drawn between granule growth, consolidation,
power consumption, and granule deformability. From comparing Figures 46 and
47 for the case of lactose, one may note that there is no further rise in power following
the end of water addition (beginning of wet massing), and this corresponds to no
further changes in granule size and porosity. In contrast, dicalcium phosphate continues
to grow through the wet massing stage, with corresponding continual
increases in granule size and porosity. Last, the starch formulations are noted to
have power increase for approximately 2 min into the wet massing stage, corresponding
to 2 min of growth; however, growth ceases when power consumption levels off.
Therefore, power clearly tracks growth and consolidation behavior.
Further results connecting power and growth to compact deformability are
provided in Holm (42). The deformability of lactose compacts, as a function of
saturation and porosity, is shown to increase with moisture in a stable fashion.
Figure 46 Power consumption for lactose, dicalcium phosphate, and dicalcium phosphate/
Theory of Granulation 53
Figure 47 illustrates the resulting granule size and voidage (or porosity). Note that
starch mixtures (15% and 45% starch) granulated in a Fielder PMAT 25 VG. (From Ref. 42.)
© 2005 by Taylor & Francis Group, LLC
Therefore, growth rates and power rise do not lag behind spray addition, and growth
ceases with the end of spraying. Dicalcium phosphate compacts, on the other hand,
remain undeformable until a critical moisture is reached, after which they become
extremely deformable and plastic. This unstable behavior is reflected by an inductive
lag in growth and power after the end of spray addition, ending with unstable growth
and bowl sticking as moisture is finally worked to the surface.
In closing, a comment should be made with regard to using power for control
and scale-up. While it is true the power is reflective of the growth process, it is a
dependent variable in many respects. Different lots of an identical formulation,
e.g., may have different yield properties and deformability, and a different dependence
on moisture. This may be due to minute particle property changes controlling
the rate processes. Therefore, there is no unique relationship between power and
growth. However, power measurements might be useful to indicate a shift in formation
properties.
Lastly, specific power is required for scale-up, where power is normalized by
the active portion of the powder bed. The impact of scale-up on mixing and distribution
of power in a wet mass, however, is only partly understood at this point.
3.12. Granule Consolidation
Consolidation of granules determines granule porosity or voidage, and hence granule
density. Granules may consolidate over extended times and achieve high densities if
there is no simultaneous drying to stop the consolidation process. The extent and
rate of consolidation are determined by the balance between the collision energy
and the granule resistance to deformation. The voidage e may be shown to depend
on time as follows:
e 
 emin
e0 
 emin ? exp?
bt? where b ? fn?y; St; Stdef? ?3:13?
Here, y is liquid loading, e0 and emin are the beginning and final minimum porosity,
respectively (53). The effect of binder viscosity and liquid content are complex
and interrelated. For low-viscosity binders, consolidation increases with liquid
Figure 47 Granule size and porosity vs. wet massing time for lactose, dicalcium phosphate,
and dicalcium phosphate/starch mixtures (15% and 45% starch) granulated in a Fielder
54 Ennis
PMAT 25 VG. (From Ref. 42.)
© 2005 by Taylor & Francis Group, LLC
content as shown in Figure 48 (53). This is the predominant effect for the majority of
granulation systems, with liquid content related to peak bed moisture on average.
Increased drop size and spray flux are also known to increase consolidation. Drying
effects peak bed moisture and consolidation as well through varying both moisture
level as well as binder viscosity. For very viscous binders, consolidation decreases
with increasing liquid content. As a second important effect, decreasing feed particle
size decreases the rate of consolidation due to the high specific surface area and low
permeability of fine powders, thereby decreasing granule voidage. Last, increasing
agitation intensity and process residence time increases the degree of consolidation
by increasing the energy of collision and compaction. The exact combined effect
of formulation properties is determined by the balance between viscous dissipation
and particle frictional losses, and therefore the rate is expected to depend on the
viscous and deformation Stokes numbers (53).
4. GRANULE STRENGTH AND BREAKAGE
4.1. Overview
Dry granule strength impacts three key areas of pharmaceutical processing. These
include the physical attrition or breakage of granules during the granulation and
drying processes, the breakage of granules in subsequent material handling steps
such as conveying or feeding, and last, the deformation and breakdown of granules
in compaction processes such as tableting. Modern approaches to granule strength
rely on fracture mechanics (54). In this context, a granule is viewed as a nonuniform
physical composite possessing certain macroscopic mechanical properties, such as a
generally anisotropic yield stress, as well as an inherent flaw distribution. Hard materials
may fail in tension, with the breaking strength being much less than the inherent
tensile strength of bonds because of the existence of flaws. Flaws act to concentrate
or notches have been added to the tablets, which were subsequently broken under
three-point bend loading (see later). In all cases, the tablets break at the razor
score—which acts as a sharp flaw to concentrate stress—rather than at the tableted
original indentation notch.
Figure 48 Effect of binder liquid content and primary feed particle size on granule porosity
for the drum granulation of glass ballotini: Decreasing granule porosity corresponds to
Theory of Granulation 55
stress, as depicted in Figure 49 for commercial metamucil tablets. Here, razor scores
increasing extent of granule consolidation. (From Ref. 53.)
© 2005 by Taylor & Francis Group, LLC
Bulk breakage tests of granule strength measure both inherent bond strength
trates granule attrition results for a variety of formulations. Granule attrition clearly
increases with increasing voidage; note that this voidage is a function of granule consolidation
discussed previously. Different formulations fall on different curves, due
to inherently differing interparticle bond strengths. It is often important to separate
the impact of bond strength vs. voidage on attrition and granule strength. Processing
influences flaw distribution and granule voidage, whereas inherent bond strength is
controlled by formulation properties.
The mechanism of granule breakage is a strong function of material properties
of the granule itself as well as the type of loading imposed by the test conditions (55).
Ranking of product breakage resistance by ad hoc tests may be test specific, and in
the worst case differ from actual process conditions. Instead, material properties
should be measured by standardized mechanical property tests which minimize the
effect of flaws and loading conditions under well-defined geometries of internal
stress, as described in the following section.
4.2. Mechanics of the Breakage Rate Process
Fracture toughness Kc defines the stress distribution in the body just before fracture
and is given by
Kc ? Ysf ffiffiffiffiffi pc p ?4:1?
where sf is the applied fracture stress, c is the length of the crack in the body, and Y
The elastic stress is increased dramatically as the crack tip is approached. In practice,
however, the elastic stress cannot exceed the yield stress of the material, implying a
region of local yielding at the crack tip. Irwin (57) proposed that this process zone
size rp be treated as an effective increase in crack length dc. Fracture toughness is
then given by
Kc ? Ysf ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi p?c ? dc? p with d c  rp ?4:2?
The process zone is a measure of the yield stress or plasticity of the material in
comparison to its brittleness. Yielding within the process zone may take place either
plastically or by diffuse microcracking, depending on the brittleness of the material.
Figure 49 Breakage of metamucil tablets by under three-point loading with razor scoring.
(A) Upper left: unscored; upper right: notched scored; bottom: scored 90 to notch. (B) Break-
56 Ennis
and granule flaw distribution and voidage (3,4). Figure 8 presented previously illusis
a calibration factor introduced to account for different body geometries (Fig. 50).
age results. (From Ref. 4.)
© 2005 by Taylor & Francis Group, LLC
For plastic yielding, rp is also referred to as the plastic zone size. The critical strain
energy release rate Gc is the energy equivalent to fracture toughness, first proposed
by Griffith (58). They are related by
Gc ? K2
c =E ?4:3?
4.3. Fracture Measurements
In order to ascertain fracture properties in any reproducible fashion, very specific
test geometry must be used since it is necessary to know the stress distribution at
predefined, induced cracks of known length. Three traditional methods are: (1) the
three-point bend test, (2) indentation fracture testing, and (3) Hertzian contact
cal geometry and force response for the case of a three-point bend test. By breaking a
series of dried formulation bars under three-point bend loading of varying crack
Figure 50
Figure 51 Three-point bend and indentation testing for fracture properties. (From Ref. 56.)
Theory of Granulation 57
compression between two spheres of the material. Figures 51 and 52 illustrate a typi-
Fracture of a brittle material by crack propagation. (From Ref. 56.)
© 2005 by Taylor & Francis Group, LLC
length, fracture toughness is determined from the variance of fracture stress on crack
In the case of indentation fracture, one determines hardness H from the area of
the residual plastic impression, and fracture toughness from the lengths of cracks
propagating from the indent as a function of indentation load F (59). Hardness is
a measure of the yield strength of the material. Toughness and hardness in the case
of indentation are given by ture toughness Kc is seen to range from 0.01 to 0.06MPa m1/2, less than that typical
for polymers and ceramics, presumably due to the high agglomerate voidage. Critical
strain energy release rates Gc are from 1 to 200 J/m2, typical for ceramics but less
than that for polymers. Process zone sizes dc are seen to be large and of the order
of 0.1–1 mm, values typical for polymers. Ceramics on the other hand typically have
process zone sizes less than 1 m m. Critical displacements required for fracture may
be estimated by the ratio Gc/E, which is an indication of the brittleness of the material.
This value was of the order of 10
7–10
8 mm for polymer–glass agglomerates,
similar to polymers, and of the order of 10
9 mm for herbicide bars, similar to ceramics.
In summary, granulated materials behave not only similar to brittle ceramics,
which have small critical displacements and yield strains, but also similar to ductile
polymers, which have large process or plastic zone sizes.
4.4. Mechanisms of Breakage
The process zone plays a large role in determining the mechanism of granule
breakage (56), with mechanisms such as those previously presented in
Figure 52 Typical force–displacement curve for three-point bend semistable failure. (From
58 Ennis
Table 5 compares typical fracture properties of agglomerated materials. Frac-
Table 1.
Ref. 56.)
length, as given by Eq. 4.2. (For details, see Ref. 56.)
© 2005 by Taylor & Francis Group, LLC

Agglomerates with process zones small in comparison to granule size break by a brittle
fracture mechanism into smaller fragments, called fragmentation or fracture. On
the other hand, for agglomerates with process zones of the order of their size, there is
insufficient volume of agglomerate to concentrate enough elastic energy to propagate
gross fracture during a collision. The mechanism of breakage for these materials is
one of wear, erosion, or attrition brought about by diffuse microcracking. In the
limit of very weak bonds, agglomerates may also shatter into small fragments or primary
particles.
Each mechanism of breakage implies a different functional dependence of
breakage rate on material properties. Granules generally have been found to have
large process zones, which suggests granule wear as a dominant mechanism of breakage
or attrition. For the case of abrasive wear of ceramics due to surface scratching
by loaded indentors, Evans and Wilshaw (60) determined a volumetric wear rate V of
V ?
d1=2
i
A1=4K3=4
c H1=2
P5=4l ?4:5?
where di is indentor diameter, P is applied load, l is wear displacement of the indentor,
and A is apparent area of contact of the indentor with the surface. Therefore,
wear rate depends inversely on fracture toughness. For the case of fragmentation,
Yuregir et al. (61) have shown that the fragmentation rate of organic and inorganic
crystals is given by
V 
H
K2
c
ru2a ?4:6?
where a is crystal length, r is crystal density, and u is impact velocity. Note that hardness
plays an opposite role for fragmentation than for wear, since it acts to concentrate
stress for fracture. Fragmentation rate is a stronger function of toughness as well.
Drawing on analogies with this work, the breakage rates by wear Bw and fragmentation
Bf for the case of fluid-bed granulation and drying processes should be of
the forms:
Bw ?
d1=2
0
K3=4
c H1=2
h5=4
b ?U 
 Umf? ?4:7?
Bf 
H
K2
c
r?U 
 Umf ?2a ?4:8?
where d is granule diameter, d0 is primary particle diameter, (U
Umf) is fluid-bed
excess gas velocity, and hb
sion rate on material properties for bars and granules undergoing a wear mechanism
of breakage, as governed by Eqs. 4.5 and 4.7.
5. CONTROLLING GRANULATION PROCESSES
5.1. An Engineering Approach to Granulation Processes
Future advances in our understanding of granulation phenomena rest heavily in
engineering process design. A change in granule size or voidage is akin to a change
60 Ennis
is bed height. Figure 53 illustrates the dependence of ero-
© 2005 by Taylor & Francis Group, LLC
in chemical species, and so analogies exist between granulation growth kinetics and
chemical kinetics and the unit operations of size enlargement and chemical reaction.
considered for successful process design. Let us begin by considering a small-volume
either the molecular or the primary particle/single-granule scale. On the granule scale
of scrutiny, the design of chemical reactors and granulation processes differs conceptually
in that the former deals with chemical transformations, whereas the latter
deals primarily with physical transformations controlled by mechanical processing
 Here, the rate processes of granulation are controlled by a set
of key physicochemical interactions. These rate processes have been defined in the
proceeding sections, including wetting and nucleation, granule growth and consolidation,
and granule attrition and breakage.
We now consider a granule volume scale of scrutiny, returning to our smallvolume
element of material A of Figure 54. Within this small volume for the case
of chemical kinetics, we generally are concerned with the rate at which one or more
chemical species is converted into a product. This is generally dictated by a reaction
rate constant or kinetic constant, which is in turn a local function of temperature,
pressure, and the concentration of feed species, as was established from previous
physicochemical considerations. These local variables are in turn a function of overall
heat, mass, and momentum transfer of the vessel controlled by mixing and
heating/cooling. The chemical conversion occurring within a local volume element
may be integrated over the entire vessel to determine the chemical yield or extent
of conversion for the reactor vessel; the impact of mixing and heat transfer is generally
considered in this step at the process volume scale of scrutiny. In the case of a
granulation process, an identical mechanistic approach exists for design, where
 This approach was pioneered by Hans Rumpf (62) and others in the early 1960s at the Universita
?t Karlsruhe, leading to development of mechanical process technology within chemical
engineering in Germany, or Mechanische Verfahrenstechnik. This was key to the founding of
powder technology, an area in which the United States has traditionally lagged behind Europe
and Japan in the areas of chemical engineering and pharmaceutical processing (61–63).
Figure 53 Bar wear rate and fluid-bed erosion rate as a function of granule material
properties. Kc
Theory of Granulation 61
These analogies are highlighted in Figure 54, where several scales of analysis must be
element of material A within a mixing process as shown in Fig. 55, and consider
is fracture toughness and H is hardness. (From Ref. 56.)
[Refs. 62–64] .
© 2005 by Taylor & Francis Group, LLC
chemical kinetics is replaced by granulation kinetics. The performance of a granulator
may be described by the extent of granulation of a species. Let (x1, x2, . . . , xn) represent
a list of attributes such as average granule size, porosity, strength, and any other
generic quality metric and associated variances. Alternatively, (x1, x2, . . . , xn) might
represent the concentrations or numbers of certain granule size or density classes, just
as in the case of chemical reactors. The proper design of a chemical reactor or a granulator
then relies on understanding and controlling the evolution (both time and spatial)
of the feed vector X to the desired product vector Y. Inevitably, the reactor or
granulator is contained within a larger plant scale process chain, or manufacturing circuit,
with overall plant performance being dictated by the interaction between individual
unit operations. At the plant scale of scrutiny, understanding interactions
between unit operations can be critical to plant performance and product quality.
These interactions are far more substantial with solids processing, than with liquid–
gas processing. Ignoring these interactions often leads processing personnel to misdiagnose
sources of poor plant performance. Tableting is often affected by segregation or
poor mixing. Segregation becomes vital for preferential wetting and drug assay
Figure 54 Changes in state as applied to granulator kinetics and design.
62 Ennis
© 2005 by Taylor & Francis Group, LLC
variation per size class. This is often effected by trace impurites in the production of
drug or excipients (1,2,64).
There are several important points worth noting with regard to this approach.
First, the engineering approach to the design of chemical reactors is well developed
and an integral part of traditional chemical engineering education [e.g. (65)]. At
present, only the most rudimentary elements of reaction kinetics have been applied
to granulator design (1,2,5). Much more is expected to be gleaned from this approach
over the coming decade.
Examples might include staged addition of ingredients, micronization of processes,
and tailored process designs based on specific formulation properties. Second,
an appreciation of this engineering approach is absolutely vital to properly scale-up
granulation processes for difficult formulations. Last, this perspective provides a
logical framework to approach and unravel complex processing problems, which
often involve several competing phenomena. Significant progress has been made
with this approach in crystallization (66) and grinding (67).
Figure 55 Granulation within a local volume element, as a subvolume of a process
granulator volume, which controls local size distribution.
Theory of Granulation 63
© 2005 by Taylor & Francis Group, LLC
Many complexities arise when applying the results of the previous sections
detailing granulation mechanisms to granulation processing. The purpose of this section
is to summarize approaches to control these rate processes by placing them
within the context of actual granulation systems and granulator design. Additional
5.2. Scale: Granule Size and Primary Feed Particles
When considering a scale of scrutiny of the order of granules, we ask what controls
the rate processes, as presented in detail in the previous sections. This key step links
formulation on material variables to the process operating variables, and successful
granulator design hinges on this understanding. Two key local variables of the
volume element A include the local bed moisture and the local level of shear (both
shear rate and shear forces). These variables play an analogous role of species concentration
and temperature in controlling kinetics in chemical reaction. In the case
of chemical reaction, increased temperature or concentration of a feed species generally
increases reaction rate. For the case of granulation considered here, increases
in shear rate and moisture result in increased granule/powder collisions in the presence
of binding fluid, resulting in an increased frequency of successful growth events
and increases in granule growth rate. Increases in shear forces also increase the granule
consolidation rate, and aid growth for deformable formulations. In the limit of
very high shear (e.g., due to choppers), they promote wet and dry granule breakage,
or limit the growth at the least. Last, in the case of simultaneous drying, bed and gas
phase moisture and temperature control heat and mass transfer and the resulting
drying kinetics.
5.3. Scale: Granule Volume Element
Next, consider a scale of scrutiny of the level of a small bulk volume of granules, or
granule size distribution controlled by the local granulation rate processes as shown
with fine powder to form initial nuclei, either directly or through mechanical breakdown
of pooled overwetted regions. It is generally useful to consider the initial powder
phase and drop phases as independent feed phases to the granule phase. In
addition, the granule phase can be broken down into separate species, each species
corresponding to a particular granule mesh size cut. Nucleation therefore results
in a loss of powder and drop phases, and the birth of granules. Granules and initial
nuclei collide within this volume element with each other and with the surrounding
powder phase, resulting in both granule growth, and consolidation due to compaction
forces. Granule growth by coalescence results in the discrete birth of granules to
a new granule size class or species, as well as loss or death of granules from the
originating size classes. On the other hand, granule growth by layering and granule
consolidation result in a slow differential increase and decrease in granule size,
respectively. Granule breakage by fracture and attrition (or wear) act in a similar,
but opposite fashion to granule coalescence and layering, increasing the powder
phase and species of smaller granules. Last, this volume element of granules interacts
with the surrounding material of the bed, as granulated, powder, and drop
phases flow to and from the surrounding volume elements. The rate processes of
granulation and the flows or exchanges with the surrounding elements combine to
64 Ennis
volume element of material A in Figure 55. This volume element has a particular
pictorially in Figure 56. In the wetting and nucleation rate process, droplets interact
details of modeling and granulator design can be found in Chapters 16 and 21.
© 2005 by Taylor & Francis Group, LLC
control the local granule size distribution and growth rate within this small-volume
element.
As illustrated in Figure 56, conducting an inventory of all granules entering
and leaving a given size class n(x) by all possible granulation mechanisms leads to
a microlevel population balance over the volume element given by
@na
@t ?
@
@xi ?naui? ? Ga ? Ba 
 Da ?5:1?
where n(x,t) is the instantaneous granule size distribution, which varies with time and
position. G, B, and D are growth, birth, and death rates due to granule coalescence
and granule fracture. The second left-hand side term reflects contributions to the
distribution from layering and wear, as well as interchanges of granules from the
surrounding volume elements. Nucleation rate would be considered a boundary
condition of Eq. 5.1, providing a source of initial granules. Eq. 5.1 governs the local
granule growth of volume element A.
Although actual processes would require specific examination, some general comments
are warranted. Beginning with nucleation, in the case of fast drop penetration
into fine powders, and for small spray flux, new granules will be formed of the order
of the drop size distribution, and contribute to those particular size cuts or granule
species. If spray is stopped at low moisture levels, one will obtain a bimodel distribution
of nuclei size superimposed on the original feed distribution. Very little growth
may occur for these low moisture levels. This should not be confused with induction
type growth, which is a result of low overall formulation deformability. In fact, the
moisture level of the nuclei themselves will be found to be high and nearly saturated.
Moisture, however, is locked up within these nuclei, surrounded by large amounts of
Figure 56 The population balance over a sieve class, over specific granule size class.
Theory of Granulation 65
Solutions to this population balance are described in greater detail in Chapter
21, as well as (5,46,52). Analytical solutions are only possible in the simplest of cases.
© 2005 by Taylor & Francis Group, LLC
fine powder. Therefore, it is important not to confuse granule moisture, local moisture,
and the overall average peak bed moisture of the process; they are very much
not the same and are influenced by proper vessel design and operation. As moisture
levels increase, and the concentration of the ungranulated powder phase decreases,
the portion of the granule phase increases. As granules begin to interact more fully
due to decreased surrounding powder and greater chances of achieved wet granule
interaction, granule coalescence begins to occur. This in turn results in a decrease
in granule number, and a rapid often exponential increase in granule size, as previously
demonstrated. Coalescence generally leads to an initial widening of the granule
size distribution until the granule growth limit is reached. As larger granules
begin to exceed this growth limit, they can no longer coalesce with granules of similar
size. Their growth rate drops substantially as they can only continue to grow by coalescence
with fine granules, or by layering with any remaining fine powder. At this
point, the granule size distribution generally narrows with time. Note that this provides
a local description of growth, whereas the overall growth rate of the process
depends greatly on mixing described next, as controlled by process design.
5.4. Scale: Granulator Vessel
The local variables of moisture and shear level vary with volume element, or position
in the granulator, which leads to the kinetics of nucleation, growth, consolidation,
and breakage being dependent on position in the vessel, leading to a scale of scrutiny
and nucleation will be high at position D. Significant growth will occur at position B
due to increased shear forces and granule deformation, as well as increased contacting.
Significant breakage can occur at position C in the vicinity of choppers. Each of
these positions or volume elements will have theirown specific granule size distribution.
Solids mixing (2,68) impacts overall granulation in several ways. First, it
controls the local shear. Local shear rates and forces are functions of shear stress
transfer through the powder bed, which is in turn a function of mixer design and
bed bulk density, granule size distribution, and frictional properties. Local shear
rates determine granule collisional velocities. This first area is possibly one of the
least understood areas of powder processing, and requires additional research to
establish the connection between operating variables and local shear rates and
Second, solids mixing controls the interchange of moisture, powder phase, and
droplet phases among the local volume elements. Third, it controls the interchange
of the granulated phase. Within the context of reaction kinetics (65), one generally
considers extremes of mixing between well-mixed continuous and plug flow continuous,
or well-mixed batch processes. The impact of mixing on reaction kinetics is
well understood, and similar implications exist for the impact of mixing on granulation
growth kinetics. In particular, well-mixed continuous processes would be expected to
provide the widest granule size distribution (deep continuous fluidized beds are an
example), whereas plug flow or well-mixed batch processes should result in narrower
distributions.y In addition, it is possible to narrow the distribution further by
y All else being equal, plug flow continuous and batch well mixed processes should produce
identical size distributions. It is very difficult to achieve uniform mixing in practice, with
properly operating fluidized beds possibly coming the closest.
66 Ennis
of the vessel size. As shown in Figure 55, moisture levels, drop phase concentration,
forces. It is also a very important scale-up consideration, as discussed in Chapter 18.
© 2005 by Taylor & Francis Group, LLC
purposely segregating the bed by granule size,z or staging the addition of ingredients,
though this is a less explored area of granulator design. Last, it should be possible to
predict the effects of dispersion, backmixing, and dead/stagnant zones on granule size
distribution, based on previous work regarding chemical reaction kinetics.
Equation 5.1 reflects the evolution of granule size distribution for a particular
volume element. When integrating this equation over the entire vessel, one is able to
predict the granule size distribution vs. time and position within the granulator. Last,
it is important to understand the complexities of scaling rate processes on a local
level to overall growth rate of the granulator. If such considerations are not made,
misleading conclusions with regard to granulation behavior may be drawn. Wide distributions
in moisture and shear level, as well as granule size, and how this interacts
with scale-up must be kept in mind when applying the detailed description of rate
processes discussed in the previous sections. With this phenomenological description
of granulation in place, we will now discuss controlling wetting, growth and consolidation,
and breakage in practice, as well as the implications for two of the more
common pharmaceutical granulation processes, namely fluid-bed and high-shear
mixer granulation.
5.5. Controlling Processing in Practice
mixer granulation. From a processing perspective, we begin with the uniformity of
the process in terms of solids mixing. Approaching a uniform state of mixing as previously
described will ensure equal moisture and shear levels and, therefore, uniform
granulation kinetics throughout the bed; on the other hand, poor mixing will lead to
differences in local kinetics. If not accounted for in design, these local differences will
lead to a wider distribution in granule size distribution and properties than is necessary,
and often in an unpredictable fashion—particularly with scale-up.
Increasing fluid-bed excess gas velocity (U
Umf) will increase solids flux and
decrease circulation time. This can potentially narrow nuclei distribution for intermediate
drop penetration times. Growth rates will be minimally affected due to
increased contacting; however, the growth limit will decrease. There will be some
increase in granule consolidation, and potentially a large increase in attrition. Lastly,
initial drying kinetics will increase. Impeller speed in mixers will play a similar role in
increasing solids flux. However, initial growth rates and granule consolidation are
likely to increase substantially with an increase in impeller speed. The growth limit
will decrease, partly controlled by chopper speed.
Fluidized beds can be one of the most uniform processes in terms of mixing
and temperature. Powder frictional forces are overcome as drag forces of the fluidizing
gas support bed weight, and gas bubbles promote rapid and intensive mixing. In
the case of mixers, impeller speed in comparison to bed mass promotes mixing, with
choppers eliminating any gross maldistribution of moisture and over growth.
With regard to bed weight, forces in fluid beds, and therefore consolidation and
granule density, generally scale with bed height. As a gross rule of thumb, ideally, the
power input per unit mass should be maintained with mixer scale-up, related in part
z Pan granulation is a specific process promoting segregation by granule size. Since large granules
interact less with smaller granule size classes, layering can be promoted at the expense of
coalescence, thereby narrowing the granule size distribution.
Theory of Granulation 67


In summary, for the case of fluid-bed granulation, growth rate is largely controlled
by spray rate and distribution and consolidation rate by bed height and peak bedmoisture.
For the case of mixers, growth and consolidation are controlled by impeller and
chopper speed. From a formulation perspective, we now turn to each rate process.
5.6. Controlling Wetting in Practice
summarizes typical changes in material and operating variables which
improve wetting uniformity. Also listed are appropriate routes to achieve these
changes in a given variable through changes in either the formulation or the processing.
Improved wetting uniformity generally implies a tighter granule size distribution
and improved product quality. Eqs. 2.5, 2.9, and 2.13 provide basic trends of
the impact of material variables on wetting dynamics and extent, as described by
the dimensionless spray flux and drop penetration time.
Since drying occurs simultaneously with wetting, the effect of drying can substantially
modify the expected impact of a given process variable and this should not
be overlooked. In addition, simultaneous drying often implies that the dynamics of
wetting are far more important than the extent.
Adhesion tension should be maximized to increase the rate and extent of both
binder spreading and binder penetration. Maximizing adhesion tension is achieved
by minimizing contact angle and maximizing surface tension of the binding solution.
These two aspects work against one another as surfactant is added to a binding fluid,
and in general, there is an optimum surfactant concentration for the formulation (27).
Surfactant type influences adsorption and desorption kinetics at the three-phase contact
line. Inappropriate surfactants can lead to Marangoni interfacial stresses which
slow the dynamics of wetting (25). Additional variables which influence adhesion tension
include (1) impurity profile and particle habit/morphology typically controlled in
the particle formation stage such as crystallization, (2) temperature of granulation,
and (3) technique of grinding, which is an additional source of impurity as well.
Decreases in binder viscosity enhance the rate of both binder spreading and binder
penetration. The prime control over the viscosity of the binding solution is through
binderconcentration.Therefore, liquidloadinganddryingconditionsstrongly influence
binder viscosity. For processes without simultaneous drying, binder viscosity generally
decreases with increasing temperature. For processes with simultaneous drying, however,
the dominant observed effect is that lowering temperature lowers binder viscosity
and enhances wetting due to decreased rates of drying and increased liquid loading.
Changes in particle size distribution affect the pore distribution of the powder.
Large pores between particles enhance the rate of binder penetration, whereas they
decrease the final extent. In addition, the particle size distribution affects the ability
of the particles to pack within the drop as well as the final degree of saturation (69).
The drop distribution and spray rate of binder fluid have a major influence on
wetting. Generally, finer drops will enhance wetting as well as the distribution of binding
fluid. The more important question, however, is how large may the drops be or how
high a spray rate is possible. The answer depends on the wetting propensity of the feed.
If the liquid loading for a given spray rate exceeds the ability of the fluid to penetrate
and spread on the powder, maldistribution in binding fluid will develop in the bed. This
maldistribution increases with increasing spray rate, increasing drop size, and decreasing
spray area (due to, e.g., bringing the nozzle closer to the bed or switching to fewer
nozzles). Themaldistribution will lead to large granules on the one hand and fine ungranulated
powder on the other. In general, the width of the granule size distribution will
70 Ennis
Table 7
© 2005 by Taylor & Francis Group, LLC
Table 7 Controlling Wetting in Granulation Processes
Typical changes in material or operating
variables which improve wetting uniformity
Appropriate routes to alter variable
through formulation changes
Appropriate routes to alter variable
through process changes
Increase adhesion tension Alter surfactant concentration or type to maximize
adhesion tension and minimize Marangoni effects
Control impurity levels in particle formation
Maximize surface tension Alter crystal habit in particle formation
Minimize contact angle Precoat powder with wettable monolayers,
e.g., coatings or steam
Minimize surface roughness in milling
Decrease binder viscosity Lower binder concentration Raise temperature for processes without
simultaneous drying Change binder
Decrease any diluents and polymers which act
as thickeners
Lower temperature for processes with
simultaneous drying since binder
concentration will decrease due to
increased liquid loading
Increase pore size to increase rate of
fluid penetration
Modify particle size distribution of feed ingredients Alter milling, classification, or formation
conditions of feed if appropriate to modify
particle size distribution Decrease pore size to increase extent of
fluid penetration
Improve spray distribution (Related to
dimensionless spray flux, given by ratio
of spray to solid fluxes)
Improve atomization by lowering binder fluid viscosity Increase wetted area of the bed per unit mass
per unit time by increasing the number of
spray nozzles, lowering spray rate,
increasing air pressure or flow rate of two
fluid nozzles
Increase solids mixing (Related to
dimensionless spray flux)
Improve powder flowability of feed Increase agitation intensity (e.g., impeller
speed, fluidization gas velocity, or rotation
speed)
Minimize moisture buildup and losses Avoid formulations which exhibit adhesive characteristics
with respect to process walls
Maintain spray nozzles to avoid caking and
nozzle drip; avoid spray entrainment in
process air streams, and spraying process
walls
Theory of Granulation 71
Source: Refs. 1,2,5.
© 2005 by Taylor & Francis Group, LLC
Table 8 Controlling Growth and Consolidation in Granulation Processes
Typical changes in material or operating
variables which maximize growth and
consolidation
Appropriate routes to alter variable
through formulation changes
Approprziate routes to alter variable
through process changes
Rate of growth (low deformability):
Increase rate of nuclei formation Improve wetting properties (see Wetting subsection);
Increase binder distribution
Increase spray rate and number of drops
Increase collision frequency Increase mixer impeller or drum rotation
speed or fluid-bed gas velocity
Increase residence time Increase batch time or lower feed rate
Rate of growth (high deformability):
Decrease binder viscosity Decrease binder concentration or change binder; decrease
any diluents and polymers which act as thickeners
Decrease operating temperature for systems
with simultaneous drying; otherwise
increase temperature
Increase agitation intensity Increase mixer impeller or drum rotation
speed or fluid-bed gas velocity
Increase particle density
Increase rate of nuclei formation,
collision frequency and residence
time, as above for lowdeformability
systems
72 Ennis
© 2005 by Taylor & Francis Group, LLC
Extent of growth:
Increase binder viscosity Increase binder concentration, change binder, or add
diluents and polymers as thickeners
Increase operating temperature for systems
with simultaneous drying; otherwise
decrease temperature
Decrease agitation intensity Decrease mixer impeller or drum rotation
speed or fluid-bed gas velocity Decrease particle density
Increase liquid loading Extent observed to increase linearly with
moisture
Rate of consolidation:
Decrease binder viscosity As above for high deformability systems As above for high deformability systems; in
addition, increase compaction forces by
increasing bed weight, or altering mixer
impeller or fluid-bed distributor plate
design
Size is controlled in milling and particle
formation
Increase agitation intensity
Increase particle density
Increase particle size
Particle size and friction strongly interact with binder
viscosity to control consolidation; feed particle size
may be increased and fine tail of distribution
removed
Source: From Refs. 1,2,5.
Theory of Granulation 73
© 2005 by Taylor & Francis Group, LLC
Table 9 Controlling Breakage in Granulation Processes
Typical changes in material or operating
variables which minimize breakage
Appropriate routes to alter variable through
formulation changes
Appropriate routes to alter variable
through process changes
Increase fracture toughness Increase binder concentration or change binder; bond
strength strongly influenced by formulation and
compatibility of binder with primary particles
Decrease binder viscosity to increase
agglomerate consolidation by altering
process temperatures (usually decrease for
systems with simultaneous drying)
Maximize overall bond strength
Minimize agglomerate voidage
Increase bed agitation intensity (e.g.,
increase impeller speed, increase bed
height) to increase agglomerate
consolidation
Increase granulation residence time to
increase agglomerate consolidation, but
minimize drying time
Increase hardness to reduce wear Increase binder concentration or change binder; binder
plasticity strongly influenced by binder type
See above effects which decrease
agglomerate voidage Minimize binder plasticity
Minimize agglomerate voidage
Decrease hardness to reduce fragmentation Change binder; binder plasticity strongly influenced by
binder type
Reverse the above effects to increase
agglomerate voidage Maximize binder plasticity
Maximize agglomerate voidage Apply coating to alter surface hardness
74 Ennis
© 2005 by Taylor & Francis Group, LLC
Decrease load to reduce wear Lower formulation density Decrease bed agitation and compaction
forces (e.g., mixer impeller speed, fluid-bed
height, bed weight, fluid-bed excess gas
velocity)
Decrease contact displacement to reduce
wear
Decrease contacting by lowering mixing and
collision frequency (e.g., mixer impeller
speed, fluid-bed excess gas velocity, drum
rotation speed)
Decrease impact velocity to reduce
fragmentation
Lower formulation density Decrease bed agitation intensity (e.g., mixer
impeller speed, fluid-bed excess gas
velocity, drum rotation speed)
Also strongly influenced by distributor plate
design in fluid beds, or impeller and
chopper design in mixers
Source: From Refs. 1,2,5.
Theory of Granulation 75
© 2005 by Taylor & Francis Group, LLC
increase and generally the average size will decrease. Improved spray distribution can be
aided by increases in agitation intensity (e.g., mixer impeller or chopper speed, drum
rotation rate, or fluidization gas velocity) and by minimizing moisture losses due to
spray entrainment, dripping nozzles, or powder caking on process walls.
5.7. Controlling Growth and Consolidation in Practice
imize granule growth and consolidation. Also listed are appropriate routes to
achieve these changes in a given variable through changes in either the formulation
or the processing. Growth and consolidation of granules are strongly influenced by
rigid (especially fluid beds) and deformability (especially mixers) Stokes numbers.
Increasing St increases energy with respect to dissipation during deformation of
granules. Therefore, the rate of growth for deformable systems (e.g., deformable formulation
or high-shear mixing) and the rate of consolidation of granules generally
increases with increasing St. St may be increased by decreasing binder viscosity or
increasing agitation intensity. Changes in binder viscosity may be accomplished by
formulation changes (e.g., the type or concentration of binder) or by operating temperature
changes. In addition, simultaneous drying strongly influences the effective
binder concentration and viscosity. The maximum extent of growth increases with
decreasing St and increased liquid loading, as reflected by Eqs. 3.11. Increasing
particle size also increases the rate of consolidation, and this can be modified by
upstream milling or crystallization conditions.
5.8. Controlling Breakage in Practice
minimize breakage. Also listed are appropriate routes to achieve these changes in a
given variable through changes in either the formulation or the processing. Both
fracture toughness and hardness are strongly influenced by the compatibility of the
binder with the primary particles, as well as the elastic/plastic properties of the binder.
In addition, hardness and toughness increase with decreasing voidage and are influenced
by previous consolidation of the granules. While the direct effect of increasing
gas velocity and bed height is to increase breakage of dried granules, increases in these
variables may also act to increase consolidation of wet granules, lower voidage, and
therefore lower the final breakage rate. Granule structure also influences breakage rate,
e.g., a layered structure is less prone to breakage than a raspberry-shaped agglomerate.
However, it may be impossible to compensate for extremely low toughness by changes
in structure. Measurements of fracture properties help define expected breakage rates
for a product and aid product development of formulations.
ACKNOWLEDGMENTS
This chapter is the result of many collaborative efforts. Support for initial granulation
research was provided by the International Fine Particle Research
Institute, G. Tardos and R. Pfeffer of The City College of the City University of
New York, and E.I Du Pont de Nemours & Company. Based on earlier versions
with A. Maraglou of Du Pont, the material was developed into a training course
by E&G Associates, and later refined in collaboration with J. Litster of the Univer-
76 Ennis
Table 8 summarizes typical changes in material and operating variables which max-
Table 9 summarizes typical changes in material and operating variables necessary to
© 2005 by Taylor & Francis Group, LLC
sity of Queensland. All of these collaborations, as well as discussions with P. Mort,
A. Adetayo, J. Seville, S. Pratsinis, S. Iveson, K. Hapgood, J. Green, P. C. Kapur, T.
Schaefer, and H. Kristensen are acknowledged with great appreciation. Last, to the
countless course participants over the last decade goes a special thank you.
REFERENCES
1. Ennis BJ, Litster J. Size enlargement and size reduction. In: Green D, Perry R, eds.
Perry’s Chemical Engineers’ Handbook. 7th ed. New York: McGraw-Hill, 1994:
Section 21.
2. Ennis BJ, ed. Solids-solids processing. In: Green D, Perry R, eds. Perry’s Chemical Engineers’
Handbook. 8th ed. New York: McGraw Hill, 2005:Section 19.
3. Ennis BJ. On the Mechanics of Granulation. Ph.D. thesis, The City College of the City
University of New York, University Microfilms International, 1990.
4. Ennis BJ. Design & Optimization of Granulation Processes for Enhanced Product
Performance. Nashville, TN: E&G Associates, 1990–2004.
5. Litster J, Ennis BJ. The Science & Engineering of Granulation Processes. Dordrecht, The
Netherlands: Kluwer Academic, 2004.
6. Masters K. Spray Drying Handbook 3rd ed. New York: Wiley & Sons, 2002.
7. Masters K. Spray Drying in Practice. ApS, Denmark: SprayDryConsult International,
1979.
8. Ennis BJ, Mehos G, Wu S, Patel P. Characterizing the impact of flow aids on flowability
of pharmaceutical excipients by automated shear cell. AAPS Annual Meeting.
Baltimore: MD, 2004.
9. Brown RL, Richards. Principles of Powder Mechanics. Oxford: Pergamon Press, 1970.
10. Stanley-Wood N. Enlargement and Compaction of Particulate Solids. London: Butterworth
& Co. Ltd, 1983.
11. Ennis BJ. Powder Technol 1996; 88:203.
12. Turton R, Tardos G, Ennis BJ. Fluidized Bed Coating and Granulation. In: Yang WC,
Fluidization, Solids Handling & Processing, New Jersey: Noyes Publications, Westword,
1999.
13. Pietsch W. Size Enlargement by Agglomeration. Chichester: John Wiley & Sons Ltd.,
1992.
14. Sastry K, Fuerstenau D. In: Sastry, ed. Agglomeration ’77. New York: AIME, 1977:381.
15. Kapur PC. Adv Chem Eng 1978; 10:55.
16. Kapur PC. Chem Eng Sci 1971; 26:1093.
17. Kapur PC, Fuerstenau DW. Ind Eng Chem Eng (Proc Des Dec) 1966; 5:5.
18. Kapur PC. Chem Eng Sci 1971; 26:1863.
19. Kapur PC. Chem Eng Sci 1971; 26:1093.
20. Rumpf H. The strength of granules and agglomerates. In: Knepper WA, ed. Agglomeration.
New York: Interscience, 1962: 379–414.
21. Rumpf H. Particle adhesion. In: Sastry KVS, ed. Agglomeration ‘77, New York: AIME,
1997: 97–129.
22. Augsburger L, Vuppala M. Chap. 2, In: Parikh D, Handbook of Pharmaceutical Granulation
Technology. 1st ed., New York: Marcel-Dekker, 1999.
23. Parfitt G, ed. Dispersion of Powders in Liquids. Amsterdam: Elsevier Applied Science
Publishers Ltd., 1986.
24. Hapgood K. Nucleation and Binder Dispersion in Wet Granulation, Ph.D. Thesis,
University of Queensland, 2000.
25. Pan R, Green J, Ennis BJ, Maderelli C. Dynamic Properties of Interfaces & Association
Structure. New York: American Oil Chemical Society Press, 1995.
26. Zisman WA. Contact angle, wettability, and adhesion. Adv Chem Ser 1964; 43:1.
27. Ayala R. Ph.D. Thesis, Chem. Eng., Carnegie Mellon University, 1985.
Theory of Granulation 77
© 2005 by Taylor & Francis Group, LLC
28. Shaw R. Introduction to Colloid and Surface Chemistry. London: Butterworths & Co.
Ltd, 1983.
29. Fuerstaneau DW, Diao J, Williams MC. Coll Surf 1991; 60:127.
30. Vargha-Butler EI. In: Botsaris GD, Glazman YM, eds. Interfacial Phenomena in Coal
Technology. Chap. 2. New York: Marcel-Dekker, 1989.
31. Aulton ME, Banks M. Proceedings of Powder Technology in Pharmacy Conference:
Powder Advisory Centre, Basel, Switzerland, 1979.
32. Aulton ME et al. J Pharm Pharmacol 1977; 29:59.
33. Litster JD, Hapgood KP, Kamineni SK, Hsu T, Sims A, Roberts M, Michaels J. Powder
Technol 2001; 124:272.
34. Ouchiyama N, Tanaka T. I and EC. Proc Des Dev 1982; 21:29.
35. Johnson KL. Contact Mechanics. Cambridge University Press, 1985.
36. Ritala M, Holm P, Schaefer T, Kristensen HG. Drug Dev Ind Pharm 1988; 14(8):1041.
37. Holm P, Schaefer T, Kristensen HG. Powder Technol 1985; 43:213.
38. Adams M, Thornton C, Lian G. Agglomeration & size enlargement. In: Ennis BJ, ed.
Proceedings of the First International Particle Technology Forum. Vol. 1. Denver,
CO: AIChE, 1994:155–286.
39. Ennis BJ, Li J, Tardos G, Pfeffer R. Chem Eng Sci 1990; 45(10):3071.
40. Mazzone D, Tardos G, Pfeffer R. J Coll Interface Sci 1986; 113:544.
41. Iveson SM, Breathe JA, Page NW. Powder Technol 2002; 127:149.
42. Holm P, Schaefer T, Kristensen HG. Part V and VI. Powder Techol 1985; 43:213–233.
43. Ennis B, Tardos G, Pfeffer R. Powder Technol 1991; 65:257.
44. Liu LX, Litster JD, Iveson SM, Ennis BJ. AIChE J 2000; 46(3):529.
45. Bird RB, Armstrong R, Hassager O. Dynamics of Polymeric Liquids. Vol. 1. New York:
John Wiley & Sons, Inc., 1977.
46. Adetayo AA, Lister JD, Pratsinis SE, Ennis BJ. Powder Technol 1995; 82:37.
47. Schaefer T, Holm P, Kristensen HG. Drug Dev Ind Pharm 1990; 16(8):1249.
48. Schaefer T, Holm P, Kristensen HG. Pharm Ind 1990; 52(9):1147.
49. Tardos G, Khan MI. AIChE Annual Meeting, Miami, 1995.
50. Tardos GI, Khan MI, Mort PR. Powder Technol 1997; 94:245.
51. Iveson SM, Wauters PAL, Forrest S, Litster JD, Meester GMH, Scarlett B, Ennis B.
Powder Technol 2001; 117:83.
52. Adetayo AA, Ennis B. AIChE J 1997; 43(4):927.
53. Iveson SM, Litster JD, Ennis BJ. Powder Techol 1996; 88:15.
54. Lawn B. Fracture of Brittle Solids. 2nd ed. Cambridge University Press, 1975.
55. Benbow JJ, Bridgwater J. Powder Technol 1987; 49:97.
56. Ennis BJ, Sunshine G. Tribology International 1993; 26:319.
57. Irwin J Appl Mech 1957; 24:361.
58. Griffith AA. Philos Trans T Soc 1920; A221:163.
59. Johnsson L, Ennis BJ. In: Ennis BJ, ed. Proceedings of the First International Particle
Technology Forum. Vol. 2. Denver, CO: AIChE, 1994:178.
60. Evans AG, Wilshaw TR. Acta Metall 1976; 24:939.
61. Yuregir KR, Ghadiri M, Clift R. Chem Eng Sci 1987; 42:843.
62. Rumpf H. Translated FA Bulletin. Particle Technology. New York: Chapman & Hall, 1990.
63. Ennis BJ, Green J, Davies R. The US Legacy of Neglect. Chem Eng Prog 1994; 90(4):32.
64. Ennis BJ, Green J. Visualizing Pharmaceutical Manufacturing as an Integrated Series of
Particle Processes. New York: Interphex, 1995.
65. Levenspiel O. Chemical Reaction Engineering. 2nd ed. New York: Wiley, 1972.
66. Randolph AD, Larson MA. Theory of Particulate Processes. 2nd ed. San Diego: Academic
Press Inc., 1991.
67. Prasher CL. Crushing and Grinding Process Handbook. New York: Wiley, 1987.
68. Weineko? tter R, Gericke H. Mixing of Solids. Dordrecht, The Netherlands: Kluwer
Academic, 2000.
69. Waldie B. Chem Eng Sci 1991; 46:2781.
78 Ennis
© 2005 by Taylor & Francis Group, LLC
3
Drug Substance and Excipient
Characterization
L. W. Chan and P. W. S. Heng
National University of Singapore, Singapore
1. INTRODUCTION
Characterization of drug substances and excipients is a very important step at the
preformulation phase of product development. Although testing will involve additional
time and cost, failure to carry out the appropriate characterization tests can
be even more costly to manufacturers if the product made is not within specifi-
cations. Preformulation characterization of raw materials creates a body of information
which is very useful in the development of products. The lack of such
information leaves the formulator with little leeway for remediation action when
a problem arises from the production process or from the quality of the finished
product. It is important to eliminate the possible influences of the raw material characteristics
before venturing into investigation of processing variables. The knowledge
derived from the characterization of raw materials can also serve to enable better
specifications to be drawn up for procuring materials with the aim of either reducing
cost or improving product quality. In addition, a review of material characterization
results can provide an excellent database for the assessment of suppliers who can
provide materials of consistent quality.
Materials from reputable companies may be supplied with detailed speci-
fications and their methods of determination may be obtained, if requested. The
information on specifications such as purity or content is very often available.
Nevertheless, it is prudent to confirm such information. The information provided
by different suppliers may vary. The type of tests carried out or the techniques used
for the characterization of a particular physical property, for example, the particle
size distribution, may be different. Comparison of materials from different suppliers
can therefore be difficult. Sometimes, the analytical result supplied by the manufacturer
is given as falling within a certain range and this gives virtually no information
about batch to batch variation of the material.
It is therefore important to have a system of in-house characterization of raw
materials alongside the stability and functional tests for the finished product. Whenever
possible, tests carried out should yield quantitative results rather than a pass/
fail or present/absent assessment. Retrospective studies of the finished product test
79
© 2005 by Taylor & Francis Group, LLC
results together with well-documented production process validation and characterization
tests of raw materials can provide directions for refinement of the production
process and improve the specifications for raw materials.
The method of material characterization varies considerably as it depends on
the nature and form of material used as well as the process involved in the conversion
of the raw materials to the finished products. The desirable characterization test
for each material depends on the material itself and the likely use or influence of
a particular material property on the process or product. For instance, detailed
information of the particle size distribution of a drug material may be less important
when the end product is a solution, more important when the material is to be granulated
and very critical when preparing an inhalant. As unnecessary testing can be
translated into additional cost, careful consideration of the type and number of tests
to be carried out on the raw materials should be weighed against the usefulness of
tests to give information on the identity and quality of raw materials, their effects
during processing and manufacturing and the functionality and esthetics of the
finished products.
The task of building up a body of information on materials is indeed difficult
given the wide spectrum of drugs and excipients used in pharmaceutical granulation.
Many compendial tests are concerned with the chemical aspects of testing and rarely
address the physical characterization of excipient materials. The physical aspects of
raw materials are more likely than their chemical properties to exert a greater
influence on the granulation process as well as the quality and functionality of the
finished products. The ability to define excipients using the correct functionality tests
would undoubtedly benefit the formulator greatly as better-defined excipients could
help to eliminate many processing problems. It is obvious that some material
characterization tests such as determination of identity and purity are important
but discussion on them is not included as they are better dealt separately.
2. PARTICLE SHAPE, SIZE, AND SURFACE AREA
2.1. Particle Shape
Particle shape is an important parameter which can have a significant effect on the
bulk properties of a powder. It is well known that spherical particles flow better,
pack better, and have a lower surface to volume ratio. Despite the well-recognized
importance of particle shape, the method of shape determination has not been
clearly defined owing to the complexity and variability of the three-dimensional
particles. In general, shape measurement methods are only able to define accurately
the shape if the shape of constituent particles can be correctly predicted based on a
two-dimensional model.
Shape of particles may be assessed descriptively by terms such as spherical,
elongated, acicular, angular, or a host of other terms. Although these are descriptive
terms, if accurately used, they can convey a general idea of the particle shape. However,
they reveal little about the degree to which the particles take upon a particular
shape. Without a comparable quantitative measure, it may be difficult to assess the
effects of particle shape on a process or product.
From the linear dimensional description of breadth, length, and height, some
shape data can be derived. Breadth is usually defined as the minimum distance
between two parallel lines bracketing the particle while length is the maximum
distance between two parallel lines enclosing the particle and is perpendicular to
80 Chan and Heng
© 2005 by Taylor & Francis Group, LLC
the breadth. The height is the thickness of the particle resting in its most stable
orientation. The measurement of height is generally difficult for small particles as
they are usually viewed through a microscope.
Direct microscopic measurement of particle dimensions is very tedious. The
particle dimension is obtained using a linear eyepiece graticule. A camera lucida
attachment may be used to trace out the outline of particles onto a paper. The perimeter
of particle tracing may be obtained by using either a string or a planimeter.
Projected area can be obtained by using paper with grids or by cutting out tracing
of particle and weighing. By determining the area per unit weight of the paper,
weight of tracing may be converted into an area measurement. The popularity of
this laborious manual mode of dimensional measurement has declined significantly
with the introduction of image analyzers. When a video camera is attached to a
microscope, the image obtained on a high-resolution monitor may be digitized
and analyzed using a computer as shown in Figure 1. Information on the breadth,
length, perimeter, and area of particles can be determined very rapidly. The degree
of accuracy of such measurements depends on the clarity of image available and
the separation of particles from one another.
There are a large number of transformation models which can be used to
analyze the image dimensional parameters in the determination of particle shape.
Some methods require additional information of particle surface area, volume or
thickness to give better estimates of shape but for most purposes, a simple approach
is often preferred. The common treatment of data, namely, breadth, length, perimeter,
and area to reflect the particle shape (1,2) is given as follows:
Aspect or elongation ratio ?
Length
Breadth ?2:1?
Figure 1 Image analysis system.
Drug Substance and Excipient Characterization 81
© 2005 by Taylor & Francis Group, LLC
Bulkiness factor ?
Area
?Length ? Breadth? ?2:2?
Formfactor ?
4p?Area?
?Perimeter?2 ?2:3?
The elongation ratio is very useful for assessing deviation from a spherical
shape to an elongated form. The formfactor, on the hand, gives a measure of
sphericity and a perfect sphere has a formfactor value of unity. The bulkiness factor
gives an indication of solidity and large indentations on the particles give rise to low
values (3).
For many users of image analyzers, a wider array of mathematical models
is usually available on the software. The problem that confronts many users is
not the mathematical treatment of image analysis data but rather the selection of an
appropriate transformation model. Procedures for Fourier shape analysis and fractal
measurement (4–7) are also available to allow further treatment of image analysis
data. Indirect methods using two or more measurement techniques like sedimentation
and sieving have also been used for particle shape determination (8). Vendors
for image analysis systems include Data Translations, Malboro, MA; Edax,
Mahwah, NJ; and Synoptics, Cambridge, U.K.
2.2. Particle Size
There is much literature on the measurement of particle size and size distribution.
With the abundance of information on particle sizing and sizing methods, a less
experienced operator can find it daunting as to which method is the best. Upon
completion of a measurement, the task of data interpretation and validation of
the accuracy and reliability of measurement can be difficult. Often, standards are
used for comparison and the information obtained from these standards are used
for calibration of the size measuring system. Standards used are very often ideal
particles with a high degree of sphericity and a narrow size distribution. The real
samples for size analysis are often not spherical and can have a considerably wide
size distribution. It is therefore important to regard all results with some suspicion
and in a comparative perspective, which depends on the standard calibrator used
and the method of measurement. Proper and stringent development of the method
for size measurement should be carried out to ensure reliability, reproducibility,
and sensitivity of measurement method.
The present discussion on the particle sizing methods does not attempt to
establish yet another theory and practice guide on particle sizing but to review the
popular methods of particle sizing, their problems and usefulness in providing
information valuable to a formulator. Little attempt will be made to explain the theories
of various sizing methods as they are dealt with in many other comprehensive
publications on this subject (8–10).
2.2.1. Microscopy
Microscopy is a very old technique capable of sizing fine powders accurately. It is
nowadays regarded as an unpopular method to use due mainly to the tedium in measurement
and the availability of more sophisticated and easy-to-use particle sizers.
82 Chan and Heng
© 2005 by Taylor & Francis Group, LLC
Nevertheless, microscopy still presents as the most direct method of particle size
analysis. However, as a particle sizer, a minimum of at least 625 particles must be
measured for computation of reasonable size statistics (11). In a particle sizing
laboratory, the value of the microscope is usually not as a primary equipment for
particle sizing but rather for the preliminary examination of the size distribution
of a powder sample prior to the use of a less direct particle size analyzer. The estimation
of particle size characteristics obtainable from the microscope would certainly
provide a rough guide for the setting of an automated particle sizer to the correct
sizing range or checking if the size analysis data obtained are within an expected
range.
Unlike other particle sizers, the microscope gives the operator a clear visual
outline of each individual particle being measured. In addition, the material may
be presented in its normal form, dry or in a liquid medium. Particles that are aggregated
can be identified and will not be measured as a single entity. Problems associated
with microscopic sizing lie not only in the tedious nature of the technique
but also in the slide sample preparation, optical resolution, and operator bias.
Sample preparation for microscopy must ensure a representative distribution
of particles on the slide to be used for sizing. When powder material is to be made
as a suspension, there should be neither loss of smaller particles through dissolution
nor loss of large particles due to sedimentation prior to transferring onto a slide. It is
best to wet a sample on the slide itself. Presentation of the area of slide specimen for
sizing must be carried out in a systematic and orderly manner by moving the slide to
view different areas of the sample. There is an intrinsic orientation problem as particles
when presented always find their most stable orientation. Microscopy involves
measuring a particle from only the top view and measurement is two dimensional.
Thus, flakes or discoids will tend to be oversized. Decision on the dimension(s) to
measure needs to be made should a linear eyepiece graticule be used. An area graticule
may also be employed for projected area determination. Orientation problems
may be overcome by dispersing powders onto adhesive before viewing or setting particles
in plastic or wax, which is then sectioned for viewing. These techniques will
definitely make an already tedious technique even more so.
Clarity is always a problem as sizes of particles approach the lower limits of
light microscope optics, which is about 1 mm. The quality of lenses may also play
a part. There is often a tendency to oversize slightly due to fringe effects around
particles. Various microscope accessories and lighting techniques can improve the
viewed image resolution to a varying extent. In some cases, the use of dyes can help
improve the contrast between particle and background.
An operator with good technique is necessary in order to obtain reliable results
with minimal systematic errors. Operator technique can influence accuracy of microscopic
determination of particle size. There is a natural tendency for an operator
to pay greater attention to large particles as he is less likely to miss them. Not only
do small particles tend to be missed, they also pose problems by being hidden by
the larger particles or appearing as clumps. However, in volumetric terms, small
particles have a lesser influence.
Automation of the microscopy technique began with an image shearing
technique (12) and progressed to image projection and measurement and, finally,
computer image digitization and image analysis (13). The preceding discussion on
particle shape has covered the use of image analysis. Information for particle size
and size distribution is easily obtained from the set of data for deriving shape factor
of the particles. An extension of light microscopy is scanning electron microscopy.
Drug Substance and Excipient Characterization 83
© 2005 by Taylor & Francis Group, LLC
Measurements may be made directly from the video image or via photomicrographs.
Attached image analysis systems may also be utilized for sizing. There have also been
improved methods to incorporate image processing systems to microscopic methods.
By taking images at varying focal distances, a composite image with highly
sharpened edges can be digitally produced.
2.2.2. Sedimentation
Sedimentation technique for particle sizing and classification has been in use for
a long time. In recent years, however, the success of light scattering particle sizers
has effectively eroded away the market for sedimentation-based particle sizers.
Sedimentation technique for sizing is based on the settling of particles under gravity
described by Stokes’ law. For a particle of diameter, d, and density, r1, under the
force of gravity, g, in a fluid of viscosity, Z, and density, r2, at its terminal velocity,
n, the accelerating force due to gravity is balanced by the viscous drag and
n ?
d2g?r1 
 r2?
18Z ?2:4?
The Andreasen pipette introduced in the 1920s is perhaps the most popular
manual apparatus for sampling from a sedimenting suspension. Determination of
the change in density of the sampled particle suspension with time enables the calculation
of size distribution of the particles. As Stokes’ law applies only to spherical
particles, the nonspherical particles give a mean diameter referred to as Stokes’
equivalent diameter. The size range measurable by this method is from 2 to
60 mm (8). The upper limit depends on the viscosity of liquid used while the lower
limit is due to the failure of very small particles to settle as these particles are kept
suspended by Brownian motion.
An improvement of the Andreasen pipette method is to use a pan attached to a
sensitive balance which records the changes in weight of the pan as an increasing
amount of suspending particles settle on it. Later, sedimentation techniques using
light extinction by changes in turbidity of the suspension and x-ray were introduced
for more sensitive and rapid measurements.
The introduction of centrifugal sedimentation makes the technique capable of
determining the distribution of particles below 5 mm. The lower limit depends on the
centrifugal velocity. The time of analysis is reduced drastically and multiple samples
in cells can be analyzed simultaneously.
2.2.3. Sieving
Sieving is probably the oldest method of sizing, used initially for particle classification
rather than size analysis. The introduction of high-quality standardized woven-wire
sieves in a ffiffiffi 2 p progression, starting from 75 mm has helped to establish sieving as a
widely used particle sizing method, especially for the larger particles. Conceptually,
particle sizing by sieving is easily understood as the different meshes classify the particles
to different weight-based size fractions giving rise to the frequency distribution.
The process of sieving involves a nest of sieves, usually five to eight, arranged
from the largest apertures to the finest followed by a receiving pan. The sieves with a
lid and receiving pan are then placed on a sieve shaker, which may gyrate, oscillate,
or vibrate the sieves. Most commonly used shakers are vibrators. It is important to
determine the time required for completion of sieving. This is commonly taken as the
time when there is no further change in the weight of material retained on each sieve
84 Chan and Heng
© 2005 by Taylor & Francis Group, LLC
after additional sieving time is given. The load for sieve analysis should be sufficient
in order that the amount present after sieving on each mesh can be accurately
weighed.
In size analysis using sieves, it must be borne in mind that the aperture size of a
given sieve is not an absolute cutoff value for the size of particles permitted to pass
through. The permitted aperture tolerance is much wider and following the maximum
tolerance allowed by the British Standard 410 (1976), particles from 55 to
95 mm may go through or be arrested by a 75 mm aperture size sieve. Particles tend
to pass through based on their narrower cross-sectional area. An elongated particle,
such as a rod-shaped particle, will pass through an aperture just larger than its minimum
cross-sectional profile. Woven sieves can be made down to about 30 mm. Finer
sieves of much higher precision from 100 mm down to a few micrometers can be made
using etching methods and these sieves are often referred to as electroformed micromesh
sieves. Sieves with bases that have accurately drilled or punched circular holes
instead of woven mesh are available for sizing larger particles, generally about
500 mm and larger.
Size analysis by sieving is a relatively slow process and there may be problems
with dust pollution. For many drugs, the safety of the operator needs to be considered.
In addition, wire mesh stretches with repeated use. Apertures may be blanked
by improper or inadequate washing. For fine powders, aggregation of powders due
to cohesive or electrostatic charges can give inaccurate results. Inadequate sieving
time will also produce unreliable data. It is also important to ensure that the sieving
process itself does not bring about size reduction.
As sieving fine powder in a dry state may be a problem due to the cohesive nature
of such a powder and a long sieving time is required, wet sieving by suspending
the powder in a suitable liquid can improve sieving efficiency. However, the procedure
becomes more tedious. It now requires the additional drying of size fractions.
It is recommended that sieving starts with the finest mesh to remove the fines with
a volume of liquid, followed by classifying the powder retained with the largest to
the smallest mesh.
Air-jet sieving is a much more popular method for sizing fine powders below
75 mm than wet sieving. It involves the use of a vacuum pump to remove air from
the underside of a sieve. Air current is also supplied from the underside of the sieve
through a rotating arm of jets, which helps to unclog the mesh. A collecting cyclone
may be attached in the vacuum line to collect the fines. In-line filters may also be used
to collect the fines. Air-jet sieving is usually used as a one-mesh sieving. For information
of size distribution, composite size distribution may be obtained from separate
air-jet sieving operations using different meshes for samples of the same powder.
A common point of discontent with size analysis using sieves is that the process
requires quite a bit of preparatory work, weighing, and subsequent washing. Yet,
a typical analysis would yield only seven to eight points on the size distribution
and this may not be sufficiently discriminating. Nevertheless, sieving is a straightforward
and robust technique suitable for a wide variety of fine to very coarse powders.
Material properties such as density, optical property, water solubility, or conductivity
are not required for computation of the particle size.
2.2.4. Electrical Sensing
The electrical sensing zone principle, which is more commonly known as the Coulter
principle, is based on a simple electrical property that the electrical resistance
Drug Substance and Excipient Characterization 85
© 2005 by Taylor & Francis Group, LLC
between two compartments containing an electrolyte and connected by an aperture
is proportional to the electrical conducting area of the aperture. Figure 2 illustrates
the basic Coulter principle. By drawing electrolyte from one compartment to the
other, particles streaming through will decrease the conducting area of the aperture.
By fast time-based tracing of the resistance, resistance pulses coinciding with the passage
of particles through the aperture will be obtained. The amplitude of the pulse is
proportional to the volume of the particle. With a pulse analyzer, the Coulter counter
can analyze a large number of particles within a short time.
Particle size range detectable depends on the aperture tube used. Each aperture
tube is effective over a size range of about 2–40% of its nominal diameter. Apertures
of sizes from 15 to 4000 mm are available. Before use, it is necessary to calibrate the
equipment with a standard latex containing monosize spherical particles of mean size
within 5–20% of the aperture diameter.
When working with small apertures, aperture blockage may be a problem.
With large apertures for sizing large particles, settling of the large particles may give
rise to sizing errors. It is important that the material for sizing is nonconductive and
nonporous. For porous particles, sizing values obtained may be much smaller than
those derived by visual inspection. Prior to addition of the powder for sizing, it is
necessary to ensure a low background count. When dispersed in the electrolyte,
the powder particles must not be flocculated and do not dissolve in the electrolyte
solution. Care must be taken to ensure proper dispersion of powder. The concentration
of particles must be within the range acceptable. Higher concentration will result
in higher errors due to coincidence while low concentration will necessitate a longer
counting time.
2.2.5. Light Scattering
There has been a tremendous growth in the application of light scattering technique
for particle sizing in recent years and light scattering particle sizers have taken a
Figure 2 The Coulter principle.
86 Chan and Heng
© 2005 by Taylor & Francis Group, LLC
lion share of the market for particle sizers. The large number of manufacturers of
instruments using light scattering technique has also a hand in convincing many
laboratories of their need for such a sizer. It is however unfair to ascribe the recent
popularity of light scattering sizing to just its marketing. Light scattering particle
sizers for both wet and dry samples are indeed much easier to use and are highly
efficient. The measurement time is short and the method is able to produce detailed
and reproducible particle size information.
Without dwelling on the theories of light scattering, particle sizers using this
principle can be roughly divided into three groups, two for determining particle size
down to about 1 mm using light obscuration and laser diffraction and the third, for
submicron particle size using photon correlation spectroscopy. A schematic diagram
of the various techniques of light scattering is shown in Figure 3.
Light obscuration or blockage technique involves measuring particles singly.
The passage of a particle across the light beam produces a reduction in the amount
of transmitted light which is detected by a sensor directly opposite the incident light.
The pulses are then classified, giving the frequency distribution. The degree of light
diffraction, opacity, and orientation of the particle as it passes the light beam can
affect the extent of light blockage and may affect measurement accuracy.
As a small particle passes through a beam of light in a laser diffraction sizer, it
will scatter light, which will be directed onto a diode array detector directly opposite
the incident light. The detector has a series of photodiodes arranged outward from
a central photodiode detector. Since the intensity of the light scattered decreases as
the scatter angle increases, photodiode elements are generally larger as they are
further from the center. Calculations for particle size and size distribution involve
rather complex mathematics. Simply put, sizing of a particle is based on the angle
of diffracted light, with small particles diffracting at wider angles than larger
particles. Thus, from the light scattering pattern, information on the size distribution
of the particles can be obtained through a series of complex calculations.
Sample presentation for light obscuration sizer involves the dispersion of the
particles in a liquid medium like in the case of the electrical sensing sizer. The main
difference is that light obscuration sizer may operate in the absence of an electrolyte.
Figure 3 Schematic diagram for light-scattering particle sizing.
Drug Substance and Excipient Characterization 87
© 2005 by Taylor & Francis Group, LLC
For laser diffraction, sizing may be carried out on both particles dispersed in liquid
as well as dry powders using dry powder feeders. Laser diffraction is also very useful
for sizing aerosol particles and sprays. Increasingly, there has been an attempt to use
information from diffraction to elucidate shape parameters.
The possibility of measuring the particle size and size distribution of powders
in their dry state with an acceptable level of accuracy and in a very short timespan
has attracted many laboratories. For many users, the laser diffraction sizer is an effi-
cient equipment for producing detailed and accurate size distribution information of
the powder particles. However, the sizing of some powders using a dry powder feeder
with the laser diffraction sizer can be fraught with problems. Possible causes of nonreproducible
results are the poor control of ambient humidity, cohesive nature of the
powder, powder particles that fragmentate easily, large size span of the particles,
variable rate of introduction of particles during measurement period, possible segregation
of powders during introduction, and stray powder particles depositing on the
lens. The sizing of a powder composed of very large and very small particles can be
problematic as a portion of the small particles will adhere onto their larger counterparts.
The complete dislodging of the smaller particles can be an extremely difficult
task and with different feed rates of particles for sizing, different results may be
obtained. In addition, humidity of the atomizing air can also affect sizing results.
2.2.6. Photon Correlation Spectroscopy
The measurement of submicrometer particles had been difficult until the introduction
of photon correlation spectroscopy for particle sizing. This technique enables
particles from a few nanometers to a few micrometers to be measured. The measurement
principle involves the determination of fluctuations in the scattered light intensuspended
particles in the liquid. Large particles will diffuse more slowly than smaller
ones and therefore, the rate of decay in intensity of scattered light at a particular
measuring point will depend on the size of the particle. The particle size distribution
is computed using complex calculations and approximations from the intensities of
scattered light (normally at 90 to the incident beam) and their rate of decay. Multiple
angle measurements are sometimes applied to improve the quality of the size
parameters obtainable.
Although the photon correlation sizer represents a very interesting instrument
for the sizing technologist, powders for granulation do not generally fall into the
submicrometer range. The few possible uses of this instrument are in the evaluation
of polymeric materials employed for binding and coating. Suppliers for both laser
diffraction and scattering particle sizers include Beckman Coulter, Fullerton, CA,
USA; Hach Ultra Analytics, Grant Pass, OR, USA; Malvern Instruments, Worcs,
U.K.; and Horiba, Kyoto, Japan.
2.3. Particle Surface Area
Surface area measurement is usually carried out by either gas permeability or
adsorption. The technique of gas permeability depends on measuring the resistance
to gas flow through a packed bed of particles. It is important that packing of the bed
is uniform and from the volumetric flow rate of the gas through and pressure drop
across the bed, the specific surface area of the powder can be calculated. The measurement
of specific surface area by gas permeability does not take into account
88 Chan and Heng
sity at an angle (Fig. 3). These fluctuations are the result of Brownian motion of
© 2005 by Taylor & Francis Group, LLC
the very small pores or fissures since the flow of gas is not hindered as it passes over
them. More accurate measurements can be made by measuring gas flow under
reduced pressure but still, the accuracy cannot match that obtainable by gas adsorption
if the total area to be determined includes those of the fine pores. Although gas
permeability gives a lower specific area for a powder compared with gas adsorption,
the value obtained is sometimes more useful in explaining factors like lubricity and
flow, which would not involve the micropores present in the particles.
Gas adsorption is carried out by placing a powder sample in a chamber and
evacuating the air within. The latter process is commonly referred to as degassing.
Upon achieving a very high vacuum, known volumes of an adsorbing gas are introduced.
From the knowledge of pressures and temperatures before and after introduction
of the adsorbing gas, usually nitrogen, calculations of total sample surface area
can be made. The surface area determination by gas adsorption is based on a simple
principle. From Avogadro’s number, a known volume of air at a certain temperature
and pressure contains a determinable number of molecules. When varying volumes
of gas are introduced to a degassed sample, the small pressure changes in the chamber
are recorded and using a calculation technique known as the BET method, the
initial amount of gas molecules which are adsorbed onto the surface forming
a monolayer can be calculated. Thus, the surface area covered by the gas molecules
can be determined by multiplying the number of molecules needed with the surface
area occupied per molecule. Samples are usually cooled to a low temperature using
liquid nitrogen. There are variations in the technique for gas adsorption by different
instrument manufacturers (14).
In addition to determining the specific surface area, pores below 50nm
may also be characterized by gas adsorption. Distribution of larger pores, 0.003–
0.004 mm, can be determined by mercury intrusion porosimetry technique, where
the volume of mercury intruded under pressure represents the volume of pores whose
entrant diameter can be calculated from the applied pressure (15). The main suppliers
of equipment for surface and porosimetry include Quantachrome, Boynton, FL,
USA, and Micromeritics, Norcross, GA, USA.
3. SOLUBILITY
The solubilities of drugs and excipients are an important physicochemical property
as they affect the bioavailability of the drug, the rate of drug release into the dissolution
medium, and consequently, the therapeutic efficacy of the pharmaceutical product.
It must be borne in mind that a drug must first be in solution in order to be
absorbed into the blood circulation. If the solubility of the drug is less than desirable,
steps must be taken to improve its solubility or to use another more soluble drug
form. Excipients which are poorly soluble in water might retard the release of drug
into the dissolution medium. Hence, the determination of drug and excipient solubilities
constitutes an important aspect of formulation study.
The solubility of a material is usually determined by the equilibrium solubility
method, which employs a saturated solution of the material. The saturated solution
is obtained by stirring an excess of the material in the solvent for a prolonged period
of time at a constant temperature until equilibrium is attained. As a guide, stirring
the mixture overnight is usually adequate for achieving equilibrium. The saturated
solution can also be obtained by warming the solvent with an excess of the material
and allowing the mixture to cool to the required temperature. This, however, may
Drug Substance and Excipient Characterization 89
© 2005 by Taylor & Francis Group, LLC
produce a supersaturated solution for some materials and therefore, heating should
be applied with caution. A portion of the saturated solution obtained by either
method is then removed with the aid of a syringe through a membrane filter at different
time intervals. The determination is completed only if at least two successive
samples have the same results, indicating that equilibrium is attained. The final value
thus obtained is the solubility of the material. The material present in the sample of
saturated solution may be assayed by a variety of methods such as UV spectrophotometry,
electrical conductivity measurement, gravimetric or volumetric analysis,
and chromatographic methods.
The following precautions should be observed in order to obtain accurate and
reproducible solubility values:
 The material and solvent must be pure
 The temperature must be properly controlled
 It is essential that some undissolved material is present in the solution to
ensure that the solution obtained is saturated
 The saturated solution used for assay must be free from undissolved
material
 The method of assay must be reliable
A good understanding of the factors affecting solubilities of materials is pertinent
in explaining the changes in solubility under different conditions. These factors
can therefore be employed to improve the solubility and bioavailability of various
drugs.
3.1. Nature of Solvent
Some materials dissolve very readily in a solvent while others dissolve sparingly. The
solubility of the material in a given solvent depends on the ability of the solvent to
overcome the forces that bind the atoms or molecules of the material. Studies have
shown a definite correlation between the solubility and the molecular structures of
the material and solvent. It is noted that the greater the similarity in molecular structure,
the higher would be the solubility of the material in the solvent. As a rule, polar
materials dissolve readily in polar solvents and nonpolar materials in nonpolar
solvents. Materials with both polar and nonpolar groups in their molecules may
dissolve in polar solvents but their solubilities tend to decrease as the proportion
of nonpolar groups in the molecule increases.
A large number of materials have poor solubilities in water and often pose
a problem in the formulation of pharmaceutical products. Addition of another
solvent, in which these materials are more readily soluble, will increase the concentration
of the materials in the solution. This additional solvent is known as a
cosolvent and common examples include glycerin, sorbitol, propylene glycol, and
polyethylene glycols. The proportion of cosolvent required varies from system to
system.
3.2. Temperature
The solubilities of most materials increase with rising temperature due to their
endothermic dissolution process. Similarly, the solubilities of materials which exhibit
exothermic dissolution decrease with rising temperature. The relationship between
90 Chan and Heng
© 2005 by Taylor & Francis Group, LLC
solubility and temperature is expressed by solubility curves. Three typical solubility
curves are shown in Figure 4.
A material whose solubility increases with rising temperature exhibits a continuous
curve with a positive slope (Curve A) while that whose solubility decreases with
rising temperature exhibits a negative slope (Curve B). Some solubility curves show
an abrupt change in slope at certain temperatures (Curve C). This phenomenon is
attributed to the change in nature of the material at the temperature where the slope
changes direction. For example, the solubility curve C is derived from a material
which can exist in two forms. The curve shows that Form I is converted to
Form II at temperature t. The dissolution of Form I in water is endothermic, which
explains the increasing solubility of the material with rising temperature until t.
Above this temperature, Form I is converted to Form II, which exhibits exothermic
dissolution. The slope of the solubility curve therefore changes from positive to
negative as the temperature exceeds t.
3.3. Crystal Characteristics
Materials may exist as amorphous or crystalline substances. Some materials such as
cortisone, tetracycline, sulfathiazole, and chloramphenicol palmitate can exist in
more than one crystalline form and this property is described as polymorphism.
The different crystalline forms which are known as polymorphs exhibit different
degrees of stability. The lattice structure of the crystalline substances may be altered
by the incorporation of molecules of the solvent from which crystallization occurs.
The resultant crystals obtained are called solvates. If the solvent is water, the crystals
are said to be hydrated.
The different forms of a material have varying solubilities. The amorphous
substance is more soluble than the crystalline counterpart. Among the crystalline
forms, the metastable polymorphs are generally more soluble than the stable polymorphs.
Hydrated crystals tend to exhibit a lower solubility in water compared with
their anhydrous form. On the contrary, the aqueous solubilities of the nonaqueous
solvates are often greater than those of the unsolvated forms.
Figure 4 Typical solubility curves.
Drug Substance and Excipient Characterization 91
© 2005 by Taylor & Francis Group, LLC
3.4. Particle Size
It is important to distinguish equilibrium solubility from intrinsic solubility of
a material. Unlike intrinsic solubility, equilibrium solubility is not affected by the
particle size of the material. The solubilities of materials reported in the literature
generally refer to the equilibrium solubilities. The method of determining equilibrium
solubility is given in an earlier section.
The intrinsic solubility of a material is dependent on the particle size of the
material. Smaller particles have a higher intrinsic solubility compared with their larger
counterparts. This is aptly explained by the existence of a higher interfacial free
energy on smaller particles resulting in a thermodynamically unstable system which
is corrected by greater dissolution of the particles and production of a supersaturated
solution. The increase in intrinsic solubility with decrease in particle size, however,
ceases when the particles have a very small radius and any further decrease in particle
size causes a decrease in intrinsic solubility.
3.5. pH
The solubility of a material will be affected by the pH of the liquid medium if the
material is acidic or basic. For example, a weakly acidic drug is more soluble in
an alkaline solution while a weakly basic drug is more soluble in an acidic solution.
This phenomenon is due to the formation of more soluble salts as a result of acid–
base reaction. Conversely, the weakly acidic drug will precipitate from the solution
if the pH is lowered by the addition of an acid while the weakly basic drug will
precipitate from the solution if the pH is raised by the addition of an alkali. The
precipitation is a result of the conversion of the drug in solution to the less soluble
unionized form.
The relationship between pH and solubility of a material is given by Eqs. 3.1
and 3.2. These equations which are modified from the Henderson–Hasselbalch
equation are useful in the estimation of the solubility of materials under different
pH conditions.
For acidic materials,
pH ? pKa ? log
S 
 S0
S0 ?3:1?
For basic materials,
pH ? pKa ? log
S0
S 
 S0 ?3:2?
where S is the overall solubility of the drug and S0 is the solubility of its unionized
form.
3.6. Additives
Additives refer to other substances incorporated into the solvent in which the drug
is dissolved. The drug in the solution may exist as ionized or nonionized forms.
Drugs that dissociate in the solvent to form ions are described as ionizable while
those that do not dissociate are nonionizable. Like the drugs, some additives ionize
readily in the solvent while others do not. Among the additives that ionize, some
produce a similar ion to the drug. The effect of additives on the solubility of a
92 Chan and Heng
© 2005 by Taylor & Francis Group, LLC
drug depends on the nature of the drug as well as the additives. For ease of understanding,
the additives will be classified as common ions, indifferent electrolytes, and
nonelectrolytes.
3.6.1. Effect of Common Ions on the Solubility of Ionizable Drugs
The solubility of a sparingly soluble drug is decreased by an additive that possesses a
similar ion to the drug. This phenomenon which is a result of common ion effect can
be explained by the Law of Mass Action.
The equilibrium of a saturated solution in the presence of undissolved drug is
represented as follows:
AB?s? , AB , A? ? B

Undissolved Undissociated Ions
drug molecule
?3:3?
If the material is sparingly soluble, the concentration of dissolved drug is sufficiently
small to assume complete dissociation into ions. The overall equilibrium may therefore
be expressed as:
AB?s? , A? ? B
 ?3:4?
From the Law of Mass Action, the equilibrium constant (K) for this reversible reaction
is given by the following equation:
K ? ?A??B

?AB?s? ?3:5?
where the square brackets indicate the concentrations of the respective components.
Since the concentration of a solid may be regarded as being constant, Eq. 3.5 can be
rewritten as:
K0s ? ?A??B
 ?3:6?
where K0s
is a constant which is known as the solubility product of the drug AB.
Some drugs have molecules which contain more than one ion of each type. The equilibrium
and solubility product of these drugs are similarly expressed as:
AxBy?s? , xA? ? yB
 ?3:7?
K0s ? ?A?x?B
y ?3:8?
If the additive dissociates to produce either A? or B
, the concentrations of these
ions in the solvent will increase. As a result, the product [A?][B
] or [A?]x[B
]y will
also increase. It should be recalled that K0s is a constant. Therefore, if K0s
is exceeded
by the product of the concentrations of ions, the equilibrium will move toward the
left in order to restore the equilibrium and the drug will be precipitated. This
explains the decrease in the solubility of a drug by common ion effect.
3.6.2. Effect of Indifferent Electrolytes on the Solubility of Ionizable Drugs
Additives which dissociate to form ions different from those of the drug are known
as indifferent electrolytes. Unlike common ions, indifferent electrolytes may increase
the solubility of a sparingly soluble drug.
Drug Substance and Excipient Characterization 93
© 2005 by Taylor & Francis Group, LLC
The solubility product defined by Eq. 3.6 is only an approximation from the
more exact thermodynamic relationship expressed by the following equation:
Ks ? aA?  aB
 ?3:9?
where Ks is the solubility product of drug AB and aA? and aB
 are the activities of the
respective ions. The activity of an ion is defined as the effective concentration of the
ion in solution. It generally has a lower value than the actual concentration because
some of the ions are ‘‘taken out of play’’ by strong association with oppositely
charged ions.
At infinite dilution, the wide separation of ions prevents interionic association
and the actual concentration and activity of the ion are equal. This situation is
applicable to a sparingly soluble drug where the concentrations of ions produced
are so small that the ions are completely unassociated. Therefore, the solubility product
can be expressed by Eq. 3.6. However, if the concentration of ions increases, the
effects of interionic association are no longer negligible and the activity becomes less
than the actual concentration. The activity coefficient, which is the ratio of activity
to actual concentration (Eq. 3.10), indicates the extent of interionic association. An
activity coefficient of unity shows that the ions are completely unassociated while
smaller values show greater interionic association:
aA?
?A? ? fA? or aA? ? fA?  ?A? ?3:10?
where fA? is the activity coefficient of ion A.
Eq. 3.9 can thus be expressed as:
Ks ? fA?  ?A?  fB
  ?B
 ? fA?  fB
  ?A?  ?B
 ?3:11?
According to Eq. 3.6, the product of the concentrations is the constant K0s. The
product of the activity coefficients of the respective ions may be equated to f 2
A?B
 ,
where fA?B
 is the mean activity coefficient of the drug. Hence,
Ks ? K0sf 2
A?B
 ?3:12?
The value of the activity coefficient decreases from unity as the overall concentration
of ions of the solution increases. Since Ks is a constant, it follows that K0s
will increase
with the ionic strength and become larger than Ks. The increase in the solubility of a
drug by the addition of indifferent electrolytes is attributed to the increase in the
ionic strength of the solution due to the indifferent electrolytes.
3.6.3. Effect of Nonelectrolytes on the Solubility of Ionizable Drugs
The solubility of an ionizable drug depends on the dissociation of the drug into ions.
The degree of dissociation is affected by the dielectric constant of the solvent. Solvents
with a high dielectric constant, being polar in nature, are able to reduce the
forces that attract the oppositely charged ions produced by dissociation of the drug.
Addition of a nonelectrolyte such as alcohol will lower the dielectric constant of the
solvent. This will decrease the dissociation and, subsequently, solubility of the drug.
3.6.4. Effect of Electrolytes on the Solubility of Nonionizable Drugs
Nonionizable drugs do not dissociate into ions in solution. They exist as single
molecules and their solubility in the solvent depends on the formation of weak
94 Chan and Heng
© 2005 by Taylor & Francis Group, LLC
intermolecular bonds with the molecules of the solvent. For example, the solubility
of the drug in water is dependent on the formation of hydrogen bonds between
the molecules of the drug and water. When an electrolyte is added, it will dissociate
to form ions which have a high affinity for water. These ions will compete with the
molecules of the drug for water and reduce the solubility of the drug in water.
3.6.5. Effect of Surfactants
Surfactants are solutes that cause a marked decrease in the surface tension of the solvent.
These substances are commonly employed as solubilizing, wetting, and emulsifying
agents. They are composed of a lipophilic group that has little affinity for water
and a hydrophilic group that has strong affinity for water. At a specific concentration
known as the critical micellar concentration, the surfactant molecules exist as
large aggregates called micelles. In an aqueous system, the hydrophilic group is
orientated on the exterior while the lipophilic groups are on the interior of the
micelles. Drugs which are poorly soluble in water may be taken into the interior
of these micelles, resulting in more of the drug being able to go into solution. The
enhanced solubility obtained as a result of the solubilization phenomenon is known
as the apparent solubility of the drug.
3.6.6. Effect of Complex Formation
The amount of drug that can go into solution may be altered by the addition of a
substance which interacts with the drug to form a complex. The solubility of the
complex will determine the apparent solubility of the drug. If the complex is more
soluble than the drug, a larger amount of the drug will dissolve to form the complex.
Thus, the drug will show a higher solubility in the solvent. Similarly, if the complex is
less soluble, some of the drug will be precipitated in the form of the complex. The
drug will therefore show a lower solubility in the solvent. It should be borne in mind
that the modified solubility obtained is not the solubility but the apparent solubility
of the drug.
4. CRYSTAL PROPERTIES AND POLYMORPHISM
Materials may occur as amorphous substances without any definite structure or as
crystalline particles with a definite structure and shape. Some materials may exist
in more than one crystalline form (polymorph) and are described as exhibiting polymorphism.
The type of crystal formed depends on the conditions, such as temperature
and type of solvent, under which crystallization is induced. At a specific
temperature or pressure, more than one polymorph can exist but only one will be
thermodynamically stable. The less stable or metastable forms will be converted to
the stable form with time. Studies show that it may take from minutes to years to
revert to the stable lattice structure.
The different crystalline forms of a material generally differ in many physical
characteristics, such as solubility, melting point, optical and electrical properties,
density, hardness, and stability. The use of metastable polymorphs frequently results
in higher solubility and dissolution rates while the stable polymorphs are often more
resistant to chemical degradation. It is obvious that any change in the crystalline
form will affect the therapeutic efficacy of a pharmaceutical product. Therefore,
Drug Substance and Excipient Characterization 95
© 2005 by Taylor & Francis Group, LLC
knowledge of the crystalline form of a drug in the formulation of pharmaceutical
products is very important and steps should be taken to ensure that the crystals
do not convert from one form to another during production and storage of the product.
Some problems which may arise if the crystal properties of the drug are not
properly characterized are precipitation, low stability, and poor bioavailability of
drug. The significance of polymorphism in pharmacy is discussed in detail by
Haleblian and McCrone (17).
A number of techniques may be used to identify the crystalline form of a material.
It is advisable to employ more than one method in the analysis as the use of only
one method has been found to be unreliable at times.
4.1. Dissolution Study
An amount of the material in excess of its solubility is added to the dissolution
medium and aliquot samples are removed and assayed at appropriate time intervals.
The concentration of the material in solution as a function of time is then plotted.
The crystalline form which constitutes the material is reflected by the shape of the
dissolution curve. The typical dissolution profile of metastable polymorph which
readily reverts to the stable form is shown in Figure 5A.
The concentration of the metastable polymorph is noted to increase much
more rapidly at the initial period of the dissolution study and then drop to that
of the stable polymorph. For the stable polymorph, the dissolution profile just
increases gradually to a plateau. The solubility of the metastable form is indicated
by the peak of its dissolution curve. In some cases, the metastable polymorph
does not revert readily to the stable form. The dissolution curve of such a metastable
form lies above that of the stable form, indicating that the former is more
soluble (Fig. 5B). The plateau of each curve indicates the solubility of the respective
polymorph.
Figure 5 Typical dissolution profiles.
96 Chan and Heng
© 2005 by Taylor & Francis Group, LLC
4.2. X-Ray Diffraction
Crystalline materials in powder form exhibit characteristic X-ray diffraction patterns
with peaks of varying heights and in different positions. These diffraction patterns
are also known as X-ray diffractograms. A typical X-ray diffractogram is illustrated
in Figure 6.
The polymorphs of a material have different crystal packing arrangements and
thus produce differences in their diffractograms from which the crystalline form of
the material is identified. This method of analysis is nondestructive and requires a
very small sample of the material which can be examined without further processing.
X-ray diffraction studies are especially useful for investigations on the changes of
crystal form during processing. In some cases, the extent of the conversion of a crystalline
drug to the amorphous form can be determined.
Conventional X-ray diffraction method cannot distinguish polymorphs which
have relatively small crystallites because of the low resolution of diffractometers.
Synchrotron sources have been employed to obtain high-resolution electron diffraction
patterns which enable differentiation of such polymorphs (18). The development
of very sensitive charge-coupled detectors (CCD) has allowed electron diffraction
patterns to be recorded in a few seconds using very low electron currents.
4.3. Infrared Analysis
As mentioned earlier, the polymorphs of a material show varying crystal packing
arrangements and produce different X-ray diffractograms. The crystal packing
arrangement also affects the energy of molecular bonds and results in different IR
spectra for the polymorphs of a material. Identification of the crystalline form of
a material is based on the spectrum derived. Infrared (IR) analysis can be used for
both qualitative and quantitative identification. It is important to use only materials
in the solid form as the polymorphs of a material in solution have identical IR
spectra.
In the area of substance identification, near IR is increasingly being used
for confirming the identity of a chemical substance. Chemometric methods using
Figure 6 Typical X-ray diffractogram.
Drug Substance and Excipient Characterization 97
© 2005 by Taylor & Francis Group, LLC
near IR have also been developed for source identification of pharmaceutical
products.
4.4. Thermal Analysis
In this method, the polymorphs are identified by their thermal behaviors. The change
in energy of the polymorph as it undergoes transformation when it is heated is
recorded as a thermogram. An example of a thermogram is given in Figure 7. The
thermogram consists of characteristic peaks, including melting point (Tm) and glass
transition temperature (Tg). The peaks pointing downward indicate endothermic
changes such as melting, sublimation, and desolvation. The different polymorphs
of a material will exhibit different thermograms which allow them to be identified.
Differential scanning calorimetry (DSC) and differential thermal analysis
(DTA) are two methods of thermal analysis commonly used for these studies. In
DSC, the energy resulting from the crystalline transformation is recorded as a function
of temperature. In DTA, the energy is expressed by differential temperature
(sample vs. inert substance).
Conventional DSC has a major limitation: if a Tg occurs in the same temperature
range as another transition, for example, water or solvent loss, the two events
cannot be separated. This limitation may be overcome by employing modulated temperature
DSC (MTDSC), where the measurements are conducted using sine wave
temperature programs defined by underlying heating rate, amplitude, and period.
The heat capacity change associated with the Tg can be separated from the heat flow
changes caused by melting, drying, and solvent loss. By use of the phase angle curve
produced from the MTDSC data analysis, very small changes in specific heat can be
detected, thereby increasing the sensitivity of the method. Based on thermal behavior,
MTDSC is able to differentiate the amorphous and polymorphic forms
of a material with much greater clarity (19). One of the disadvantages of this
Figure 7 Example of a thermogram.
98 Chan and Heng
© 2005 by Taylor & Francis Group, LLC
method is that the data analysis and interpretation are more difficult than for DSC.
In addition, the experiment process can be long as much lower heating rates
are used.
4.5. Hot Stage Microscopy
The polarizing microscope fitted with a hot stage is very useful for identifying
the crystalline forms of a material. In this method, the polymorph is heated to a
temperature at which it undergoes a change in birefringence and/or appearance
which is characteristic of the polymorph.
4.6. Raman Spectroscopy
Raman spectroscopy provides molecular information about the crystalline, as well as
the amorphous forms of a material. In this method, the material is subjected to
a laser beam and a spectrum of the scattered light obtained. The spectrum shows
vibrational bands of the material at their characteristic frequencies. The amorphous
and polymorphic forms of a material can be distinguished by their characteristic
spectra.
Raman spectroscopy and IR spectroscopy complement each other. The former
measures a change in polarization whereas the latter measures a change in dipole
moment. IR-inactive vibrations can be strong in Raman spectra and vice versa.
For example, vibrations in the region of 10–400cm
1 are more easily studied by
Raman than by IR spectroscopy.
One advantage of the Raman spectroscopy method is that no sample preparation
is required, thus the likelihood of inducing phase changes through processing is
reduced. However, representative sampling is critical for quantitative analysis (20).
The results are affected by the particle size of the material. The use of Fourier transform
Raman spectrometers with a longer wavelength laser of 1064nm eliminates the
problem of any fluorescent background. With the utilization of fiber optics, real-time
crystallization can be monitored (21). Thus, this method is useful for on-line analysis
of pharmaceutical processes.
4.7. Solid-State Nuclear Magnetic Resonance Spectroscopy
Solid-state nuclear magnetic resonance (SSNMR) spectroscopy is a more advanced
method for differentiating the polymorphs of a material. The substance is placed in
a strong magnetic field and subjected to radiofrequency radiation. The individual
nuclei experience different magnetic environments and thus show different changes
in resonant frequency characterized by chemical shift. SSNMR spectra show sharp
resonance at chemical shifts characteristic of the molecular and crystal structure.
The polymorphs are differentiated by their characteristic spectra.
This method is suitable for the study of molecular motions in a wide range of
time scales (102–10
10 sec) and thus complements the X-ray diffraction method,
which is limited to motions that are relatively fast (10
18 sec). It has a few
advantages. It can be used for characterization of solid-state forms that cannot be
crystallized and studied by the X-ray diffraction method. It is also useful for
quantifying components of heterogenous mixtures. In contrast to IR and Raman
spectroscopy, the results are less affected by the particle size of the test material.
Drug Substance and Excipient Characterization 99
© 2005 by Taylor & Francis Group, LLC
5. OTHER PHYSICAL PROPERTIES
It is undoubted that the type of physical characterization tests for a drug or excipient
depends very much on the material concerned as well as the processing involved.
Material testing can be broadly divided into two types, namely physical testing
and functionality testing. Physical testing, which is used to determine properties such
as size, solubility, and crystal form is generally more direct and the procedures are
better established. Functionality testing which evaluates, for example, lubricity, flow,
creep, and tack is less well established but, if carried out, may yield useful information
about the raw materials and their potential effects on the processing.
With the introduction of automated helium pycnometer, true densities of raw
materials can be determined with ease and at a high degree of accuracy. The true
density can serve to assure the formulator the identity of the material and sometimes
reveal the state of raw materials, like partial changes of a powder from anhydrous to
hydrated form.
Packing studies of the powder can reveal its rheological properties. These
studies are carried out by filling a volumetrically calibrated cylinder with powder
and tapping it. From the weight of powder filled and the changes in volume of the
powder on tapping, changes in apparent densities can give information about the
powder flowability (22). A small change in apparent density before and after tapping
indicates good flow properties. The Hausner ratio, the ratio of tapped density to
poured density has been found to be extremely sensitive to particle shape (23).
The poured density is the undisturbed packing density of a powder in a calibrated
cylinder after filling, usually done by passing powder through a sieve. The tapped
density is the packing density after tapping a bed of powder until no change in
the packing density is seen. An alternative expression used to predict powder flowability
is the compressibility index or Carr index, which is the ratio of the difference
between the tapped and poured densities to the tapped density, expressed as
a percentage.
Flowability of a powder may be determined directly using a flow cup, which is
often vibrated, or indirectly by the angle of repose. The angle of repose measurement
usually involves forming a conical powder heap and measuring the maximum
inclination of heap. The powders that can flow well give small angles of repose. A
variation of the heap forming measurement is to use a loosely packed powder bed
or transparent cylinder partially filled with powder and slowly rotating the powder
bed away from the initial horizontal plane of the powder surface. The angle of
inclination when sliding of powder just occurs is used to characterize the powder
flowability. Determination of powder avalanches is a simple and efficient way to
evaluate the flow properties of pharmaceutical powders. Powder flow properties
are assessed by comparing their flow indices using a powder flowability analyzer
(TSI Incorporated, USA). The flow analyzer consisted of a transparent acrylic drum
The drum could be programmed to rotate at user-defined rates. A metal mesh
collar was fitted onto the inner circumference of the drum to prevent the sliding of
powder particles down the circumferential wall instead of avalanching. Test powders
should be sieved before being poured into the sample drum.
Analysis should be performed at a range of drum speeds in order to determine
the flow index, as calculated by the equation described as follows:
100 Chan and Heng
fitted on a vertical bayonet-type mount (Fig. 8).
© 2005 by Taylor & Francis Group, LLC
Flow index ?
1
nXn
i?1
Si ?5:1?
where n is the total number of speeds tested and Si is the standard deviation of the
tested speed.
Strange attractor diagrams are generated by plotting the time difference
between each avalanche and the next formed throughout the test duration. These
plots provide a visual comparison of the flow properties of powders. Free-flowing
powders will generate strange attractor diagrams that are dense and closer to the
origin. In addition, their calculated flow indices will be correspondingly small.
Shear studies of powder bed can also be carried out to evaluate the cohesiveness
of powders. There are many variations of shear cells for evaluation of powder
cohesiveness.
The change in the packing properties of powders under pressure is used to
study the compactability of powders. This study can be made using compression
punches attached to pressure transducers (24). The force exerted and the displacement
can be utilized to calculate useful parameters for assessing the compactability
of a powder or powder mix.
For polymers, mechanical testing such as creep testing of films formed can provide
information on the suitability of the polymer or the additives added for their
film forming function (25). When polymers are used as a binder, adhesive properties
in addition to polymer viscosity may be determined. The adhesive property can be
evaluated by the measurement of tack or stickiness. This measurement involves
determining the force required to detach two platens held together by the polymer
solution. Other tests that may be carried out include measurements for surface
Figure 8 Avalanche powder flowability tester.
Drug Substance and Excipient Characterization 101
© 2005 by Taylor & Francis Group, LLC
activity, glass transition temperature, cloud point, and adhesion strength of dried
polymer.
6. COMMONLY USED EXCIPIENTS IN GRANULATION
Excipients for granulation can be largely divided into two categories, as bulking
agents and as functional additives. It is true that bulking agents or fillers also serve
a function in that they form the core or structure of a dosage form. Nevertheless,
bulking agents generally differ from the functional additives in that they are usually
inert materials that are relatively inexpensive. Functional additives include binders,
disintegrants, lubricants, colorants, and stabilizing agents. Besides the pharmacopoeia,
several recently published handbooks can provide a compilation of commonly
used pharmaceutical excipients (26,27).
The choice of excipients depends on a number of factors, namely, the drug
used, the process involved, the formulator, and the cost of excipient. Differences will
be seen in the choice of excipients by innovator companies and generic companies
as they have different cost considerations. Some granulation processes like fluid
bed granulation would require tighter control of drug and excipient specifications
compared with wet granulation using a paddle mixer. In fluid bed, besides the
control of particle size distribution, particle density of excipient should not be too
greatly different from that of the drug. Extrusion–spheronization would generally
require microcrystalline cellulose.
A very common filler is lactose, although other sugars, dicalcium phosphate,
starch, pregelatinized starch, and microcrystalline cellulose are also used. The
starches and microcrystalline cellulose are also disintegrants in tablets. For tablets,
other commonly used disintegrants include sodium starch glycolate, croscarmellose,
crospovidone, and low-substituted hydroxypropylcellulose. Lubricants are usually
not added till prior to filling or tableting of the granules. The most commonly used
lubricant is magnesium stearate. Other lubricants used include calcium stearate,
stearic acid, wax, hydrogenated vegetable oil, talc, and starch. Much work has been
done on lubricants and it is well established that the physical characterization of the
lubricant is very important to ensure consistency in functionality, especially when
a change is made in the supplier (28).
Binders used in granulation consist of a wide variety of sugars and polymers—
natural, semisynthetic, and synthetic. Sugars used include sucrose, glucose, and sorbitol.
Examples of natural polymers are acacia, alginic acid, sodium alginate, gelatin,
and starch. There is an inherent variability in the natural polymers of different
batches and this sometimes gives rise to problems in production. It is essential to
characterize binders for their viscous properties, at the least, to minimize potential
processing problems. The semisynthetic binders include ethylcellulose, sodium carboxymethylcellulose,
methylcellulose, hydroxypropylcellulose, and hydroxypropylmethylcellulose.
By comparison with the natural binders, variability between
batches of the semisynthetic binders from the same supplier can be expected to be
much less. For each polymer type, a number of viscosity grades are available. The
two types of synthetic binders in use are polyvinylpyrrolidone and polyethylene glycol.
Polyvinylpyrrolidone is widely used in both wet massing and fluid bed granulation.
The main application of polyethylene glycol as a binder is in melt granulation.
Other excipients employed in pharmaceutical products include colorants,
coating aids, stabilizers, pH modifiers, and release rate modifiers. They play very
102 Chan and Heng
© 2005 by Taylor & Francis Group, LLC
important roles and would require specific characterization tests. In the making of a
product, the functional excipients to be added must be carefully selected to ensure
that the final product is bioavailable and esthetically pleasing. As all raw materials
have to undergo a processing schedule for conversion of the materials to the finished
product, it is likely that the processing can be very unpredictable if standardizing of
raw materials has not been carried out.
7. COMPATIBILITY OF DRUG AND EXCIPIENT
In the formulation of any pharmaceutical product, it is imperative to ensure that
the ingredients used are compatible with one another. Incompatibilities can occur
between drug and excipient as well as between the excipients themselves. Incompatibilities
may be manifested through many modes, such as acid–base interaction and
complex formation, resulting in lower potency and/or stability and eventually poor
therapeutic efficacy of the product. It is therefore essential to avoid incompatibilities
and this is achieved by carrying out studies to detect potential interactions between
the components used in the formulation.
7.1. Stability Study
This is the traditional method of detecting incompatibilities. Mixtures of the drug
and excipient are prepared and stored under exaggerated conditions of heat, light,
and humidity. A detailed discussion on the ‘‘realistic’’ proportions of drug and excipient
to be used in the investigation was reported by Akers (29). The mixtures are
examined for any physical change and aliquot samples are withdrawn for assay of
the intact drug at varying time intervals. Incompatibility is reflected by various signs
such as appearance of precipitate and decrease in the concentration of intact drug.
7.2. Thermal Analysis
A relatively simple approach for the investigation of potential interaction between
a drug and an excipient can be carried out using differential scanning calorimetry.
The drug, individual excipients, and binary mixtures of the drug and excipient are
separately scanned at a standard rate over a temperature range that encompasses
all the thermal features of the drug and excipients. Each mixture consists of 50%
drug and excipient in order to maximize the likelihood of an interaction. The thermograms
of the mixtures and the appropriate individual components are compared.
Interaction is deduced by changes in the thermal features such as elimination or
Changes in shape, onset, maximum temperature, and relative height of the peaks
may also indicate interaction. However, it should be cautioned that these changes
could also arise from physical mixing of the components.
A big advantage of differential scanning calorimetry over the traditional
stability test is the speed of determination. However, like all methods, differential
scanning calorimetry has its own limitations. It is not applicable if the test materials
exhibit properties that make data interpretation difficult, such as eutectic formation,
coincident melting, and dissolution of one component in the melt of the other. It is
not advisable to rely on differential scanning calorimetry alone to determine incompatibility.
Chrzanowski et al. (30) reported that differential scanning calorimetry
Drug Substance and Excipient Characterization 103
appearance of a peak in the thermogram of the mixture. This is illustrated in Figure 9.
© 2005 by Taylor & Francis Group, LLC
indicated no incompatibilities in mixtures of fenretinide–excipient and mefenidil–
excipient, whereas the traditional stability study showed some incompatibilities.
Hence, differential scanning calorimetry should only be used to supplement the
stability test by eliminating the incompatible excipients and reducing the number
of test samples.
7.3. Chromatographic Methods
Chromatography was first used for the separation of colored leaf pigments. The
operation of chromatography is based on the distribution of a material between a
stationary phase and a mobile phase. The stationary phase can be a solid or a liquid
supported on a solid while the mobile phase can be a gas or a liquid which flows continuously
around the stationary phase. The different components in a mixture can be
separated and identified as a result of differences in their affinity for the stationary
phase.
In addition to its application in the separation and identification of materials,
chromatography is also employed to detect potential interactions between materials.
Both thin-layer chromatography and liquid chromatography are commonly employed
in this area of study. In thin-layer chromatography, the stationary phase
consists of a powder adhered onto a glass, plastic, or metal plate. The powders commonly
used are silica, alumina, polyamides, cellulose, and ion-exchange resins. Solutions
of the drug, excipient, and drug–excipient mixture are prepared and spotted on
the same baseline at one end of the plate. The plate is then placed upright in a closed
chamber containing the solvent, which constitutes the mobile phase. As the solvent
moves up the plate, it carries with it the materials. Those materials which have
a stronger affinity for the stationary phase will move at a slower rate. The material
is identified by its Rf value, which is defined as the ratio of the distance which the
material has moved to the distance the solvent front has moved. The position of
the material on the plate is indicated by spraying the plate with certain reagents or
Figure 9 Thermograms indicating drug–excipient interaction.
104 Chan and Heng
© 2005 by Taylor & Francis Group, LLC
exposing the plate to ultraviolet radiation. If there is no interaction between the
drug and excipient, the mixture will produce two spots whose Rf values are identical
with those of the individual drug and excipient. If there is interaction, the complex
formed will produce a spot whose Rf value is different from those of the individual
components.
In liquid chromatography, the distribution of the material between the solid
stationary phase in a column and the liquid mobile phase is determined. The material
is identified by the time taken for it to elute from the column. Solutions of the drug,
excipient, and drug–excipient mixture are prepared and injected into the column.
The materials will elute from the column at different speeds, depending on their affi-
nity for the column. The concentration of the material that elutes from the column is
detected and plotted against time to give to a chromatogram. If there is interaction
between the drug and excipient, the complex formed will exhibit an elution time different
from those of the individual components. This is illustrated in Figure 10,
which shows the chromatograms of the drug, excipient, and drug–excipient mixture
plotted together. Similarly, gas chromatography may be used.
8. CONCLUSION
The greatest difficulty for any process technologist is to decide on the type and extent
of material characterization to be undertaken such that it is cost-effective in the long
run. Often, it is the problem from the production run that necessitates further material
characterization to be carried out either for the purpose of resolving the problem
or to prevent future occurrences. This chapter serves to identify the more common
material characterization methods that can be carried out and the potentially useful
information that can be inferred from the tests. It is hoped that the discussion of the
many methods of material characterization could help in the choice of characterization
methods for material testing.
Figure 10 Chromatograms illustrating drug–excipient interaction.
Drug Substance and Excipient Characterization 105
© 2005 by Taylor & Francis Group, LLC
REFERENCES
1. Heywood H. The evaluation of powders. J Pharm Pharmac 1963 (suppl 15):56T.
2. Wadell H. Volume, shape and roundness of rock particles. J Geol 1932; 40:443.
3. Stanley-Wood NG. Particle characterisation by size, shape and surface for individual
particles. Stanley-Wood NG, ed. Enlargement and Compaction of Particulate Solids.
London: Butterworths, 1983:18.
4. Kaye BH. Multifractal description of ragged fine particle profile. Particle Char 1984;
1:14.
5. Kaye BH. Specification on the ruggedness and/or texture of a fine powder profile by its
fractal dimension. Powder Technol 1978; 21:1.
6. Flook AG. The use of dilation logic on the Quantimet to achieve dimension characterisation
of textured and structured profiles. Powder Technol 1978; 21:295.
7. Meakin P. Simulations of aggregation processes. Avnir D, ed. The Fractal Approach to
Heterogeneous Chemistry. Chichester: Wiley, 1990:131–160.
8. Allen T. Particle Size Measurement. 4th ed. London: Chapman and Hall, 1990.
9. Svarosky L. Characterization of powders. Rhodes MJ, ed. Principles of Powder
Technology. Chichester: John Wiley, 1990:35.
10. Washington C. Particle Size Analysis in Pharmaceutics and Other Industries. U.K: Ellis
Horwood, 1992.
11. British Standard 3406, Part 4, 1963.
12. Timbrell V. A method of measuring and grading microscopic spherical particles. Nature
1952; 170:318.
13. Russ JC. Computer-Assisted Microscopy. New York: Plenum Press, 1990.
14. Sing KSW. Adsorption methods for surface area determinations. Stanley-Wood NG,
Lines RW, eds. Particle Size Analysis. Cambridge, U.K: The Royal Society of Chemistry,
1992:13.
15. Lowell S, Shields JE. Particle Surface Area and Porosity. 2nd ed. London: Chapman and
Hall, 1983:205.
16. Buckley HE. Crystal Growth. New York: Wiley, 1951:29.
17. Haleblian J, McCrone W. Pharmaceutical applications of polymorphism. J Pharm Sci
1969; 58:911.
18. Li ZG, Harlow RL, Foris CM, Li H, Ma P, Vickery RD, Maurin MB, Toby BH. Polymorph
determination for the GP IIb/IIIa antagonist, roxifiban, using a combination of
electron diffraction and synchrotron x-ray powder diffraction techniques. J Pharm Sci
1999; 88(3):297.
19. Bottom R. The role of modulated temperature differential scanning calorimetry in the
characterization of a drug molecule exhibiting polymorphic and glass forming tendencies.
Int J Pharm 1999; 192:47.
20. Taylor LS, Zografi G. The quantitative analysis of crystallinity using FT-Raman spectroscopy.
Pharm Res 1988; 15(5):755.
21. Findlay WP, Bugay DE. Utilization of Fourier transform-Raman spectroscopy for the
study of pharmaceutical crystal forms. J Pharm Biomed Anal 1998; 16:921.
22. Yamashiro M, Yuasa Y, Kawakita K. An experimental study on the relationships
between compressibility, fluidity and cohesion of powder solids at small tapping numbers.
Powder Technol 1983; 34:225.
23. Kostelnik MC, Beddow JK. New techniques for tap density. In: Hausner HH, ed. Modern
Developments in Powder Metallurgy IV, Processes. Proceedings of the 1970 International
Powder Metallurgy Conference. New York: Plenum Press, 1970.
24. Watt PR. Tablet Machine Instrumentation in Pharmaceutics: Principles and Practice.
Chichester: Ellis Horwood, 1988.
25. Okhamafe AO, York P. Interaction phenomena in pharmaceutical film coatings and
testing methods. Int J Pharm 1987; 39:1.
106 Chan and Heng
© 2005 by Taylor & Francis Group, LLC
26. Handbook of Pharmaceutical Excipients. Washington, DC: American Pharmaceutical
Association, London: Pharmaceutical Society of Great Britain, 1986.
27. Brittain HG, ed. Analytical Profiles of Drug Substances and Excipients. San Diego: Academic
Press, 1993.
28. Chowhan ZT. Excipients and their functionality in drug product development. Pharm
Tech 1993; 17:72.
29. Akers MJ. Preformulation testing of solid oral dosage form drugs. Can J Pharm 1976;
11:1.
30. Chrzanowski FA, Ulissi LA, Fegely BJ, Newman AC. Preformulation excipient compatibility
testing. Application of differential scanning calorimetric method versus a wet
granulation simulating isothermal stress method. Drug Dev Ind Pharm 1986; 12:703.
Drug Substance and Excipient Characterization 107
© 2005 by Taylor & Francis Group, LLC
4
Binders and Solvents
Ehab Hamed, Derek Moe, and Raj Khankari
CIMA Labs Inc., Eden Prairie, Minnesota, U.S.A.
John Hontz
Biovail, Chantilly, Virginia, U.S.A.
1. INTRODUCTION
Binders are adhesives that are added to solid dosage formulations. The primary role
of binders is to provide the cohesiveness essential for the bonding of the solid particles
under compaction to form a tablet. In a wet-granulation process, binders promote
size enlargement to produce granules and thus improve flowability of the blend during
the manufacturing process. Binders may also improve the hardness of the tablets
by enhancing intragranular as well as intergranular forces. In a direct compression
process, binders often act as fillers and impart compressibility to the powder blend.
The cohesive properties of binders may reduce friability of the tablets and thus aid
in their durability and elegance. Although the purpose of using binders in a tablet formulation
is not to influence its disintegration and dissolution rate, these properties
may be modified due to the altered wettability of the formulation.
2. TYPES OF BINDERS
Binders are usually natural polymers, synthetic polymers, or sugars. The selection of
the type of binder for a particular system is quite often empirical and dependent on
the previous experience of the formulator, in conjunction with examination of excipient
compatibility studies. Selection of the quantity of binder required in a particular
system can be determined by optimization studies, using parameters such as
granule friability, tablet friability, hardness, disintegration time, and the drug dissolution
rate. Some commonly used binders in wet granulation with their usual concentration
range, along with the granulating system are listed in Basic
properties of some widely used binders with their method of incorporation will be
discussed in this section (1–4).
109
Table 1.
© 2005 by Taylor & Francis Group, LLC
2.1. Natural Polymers
2.1.1. Starch
Starch is a polymeric carbohydrate obtained from various plant sources, such as
potato, wheat, maize, rice, and tapioca. It is a GRAS-listed material and traditionally
was one of the most widely used tablet binders. It is insoluble in cold water and in
alcohol but it gelatinizes in hot water to form a paste. Starch paste can be prepared
by dispersing starch in 1–1.5 parts of cold water for initial wetting followed by addition
of two to four times as much boiling water with continuous stirring until a translucent
paste is obtained. This paste is further diluted by cold water to the desired
concentration. Alternatively, starch paste can also be prepared by heating the cold
water suspension of starch to boiling in a steam-jacketed kettle with constant stirring.
Freshly prepared starch paste is used at a concentration of 5–25% (w/w) in a
tablet granulation. Relatively soft and friable granules are produced when starch
paste is used as a binder. Consequently, it yields tablets that disintegrate readily.
During the wet-massing process, high viscosity of the starch paste can sometime
make it difficult to evenly distribute the binder in the powder blend.
Table 1 Commonly Used Granulating Systems
Binder
Method of
incorporation
Percentage
used in
formula Solvent
Percentage used
in granulating
system
Natural polymers
Starch Wet mixing 2–5 Water 5–25
Pregelatinized starch Wet mixing 2–5 Water 10–15
Dry mixing 5–10 Water
Gelatin Wet mixing 1–3 Water 5–10
Acacia Wet mixing 3–5 Water 10–15
Alginic acid Dry mixing 1–5 Water
Sodium alginate Wet mixing 1–3 Water 3–5
Synthetic polymers
PVP Wet mixing 0.5–5 Water or
hydroalcoholic
solution
5–10
Dry mixing 5–10
Methyl cellulose Wet mixing 1–5 Water 2–15
Dry mixing 5–10
HPMC Wet mixing 2–5 Water or
hydroalcoholic
solution
5–10
Dry mixing 5–10
Na-CMC Wet mixing 1–5 Water 5–15
Dry mixing 5–10
Ethyl cellulose Wet mixing 1–5 Ethanol 2–10
Dry mixing 5–10
Sugars
Glucose Wet mixing 2–25 Water 25–50
Sucrose Wet mixing 2–25 Water 50–67
Sorbitol Wet mixing 2–10 Water 2–25
110 Hamed et al.
Source: Modified from Ref. 3.
© 2005 by Taylor & Francis Group, LLC
2.1.2. Pregelatinized Starch
Pregelatinized starch is a modified starch used in tablet formulations as a binder,
diluent, and disintegrant. It is obtained by chemically and mechanically processing
the starch to rupture all or part of the starch granules. This process renders starch
flowable, directly compressible, and soluble in warm water without boiling. As a
binder in a wet-granulation process, pregelatinized starch can be used either as a solution
reconstituted in water or as a dry blending followed by wetting with water. The
latter process requires two to four times more binder to achieve the same binding effect.
Pregelatinized starch is available in fully or partially pregelatinized forms. The
degree of pregelatinization determines its solubility in cold water. Cold-water soluble
fraction for a partially pregelatinized starch is 10–20%. Starch 1500 is partially pregelatinized
starch containing 20% maximum cold-water soluble fraction, which makes
it useful for wet granulation. The water-soluble fraction acts as a binder, while the
remaining fraction facilitates the tablet disintegration process.
2.1.3. Gelatin
Gelatin is a mixture of purified protein fractions obtained by partial acid hydrolysis
(Type A gelatin) or alkali hydrolysis (Type B gelatin) of animal collagens. It is insoluble
in cold water and in alcohol, but is soluble in hot water. In hot water, gelatin forms
a gel on cooling to 35–40
C. At temperatures >40
C, the system exists as a solution.
Therefore, the gelatin solutions must be used when warm to avoid gel formation.
During the preparation of gelatin solution, gelatin must be wetted in cold water
and then heated with gentle agitation to ensure dissolution. The agitation intensity must
be kept low to prevent air entrapment in the viscous solution. Use of gelatin as a binder
is limited in general-purpose tablets because it produces tablets characterized by high
hardness and slow disintegration. However, these properties of gelatin along with its
smooth mouth feel can prove to be advantageous in a lozenge formulation.
Gelatin reacts with aldehydes, aldehydic sugars, anionic and cationic polymers,
electrolytes, metal ions, plasticizers, preservatives, and surfactants. In a wet-granulation
process of a formulation containing color, the migration of dyes toward the upper
surface of the static bed during the drying operation is often amplified by the presence
of gelatin due to its high affinity for dyes. The gelatin solutions are susceptible to microbial
contamination upon storage, therefore freshly prepared solutions should be used.
2.1.4. Acacia
Acacia is a natural gum obtained from the acacia trees. It is a complex, loose
aggregate of sugars and hemicelluloses. It is commercially available in a powdered
form, a granular form, or as a spray-dried product. As a tablet binder, it is used
in an aqueous solution or added in dried form prior to moistening with water. Acacia
forms very hard tablets, which disintegrate slowly. Aqueous solutions are susceptible
to bacterial and enzymatic degradation. It is incompatible with amidopyrine, cresol,
phenol, ethanol, ferric salts, and a number of other substances. Acacia, which was
widely used in the past as a tablet binder, is rarely used today, in favor of one of
the many synthetic polymers.
2.1.5. Tragacanth
Tragacanth is a naturally occurring dried gum. It poses similar problems as those of
acacia. Dry addition to the blend followed by addition of water works better than
addition of the solution, because it is difficult to prepare and use the mucilage.
Binders and Solvents 111
© 2005 by Taylor & Francis Group, LLC
2.1.6. Alginic Acid
Alginic acid is a polymannuronic acid extracted from seaweed. It is used as a binder
and disintegrating agent at concentrations between 1% and 5%. It slowly hydrolyzes
at room temperature and is insoluble in water. Therefore, it is best incorporated in a
dry state. It is incompatible with strong oxidizing agents. It forms insoluble alginates
with the alkaline earth metals and group III metals, with the exception of magnesium.
These alginates may delay disintegration of the tablets due to their gelling
properties.
2.1.7. Sodium Alginate
Sodium alginate slowly dissolves in water to form a viscous solution. A 3–5% solution
typically is used in wet-granulation processes. It has also been used in sustained
release formulations because it delays the dissolution of a drug from tablets. It is
hygroscopic and its aqueous solution is susceptible to microbial contamination.
2.2. Synthetic Polymers
2.2.1. Polyvinylpyrrolidone (Povidone)
Polyvinylpyrrolidone (PVP) is versatile and one of the most commonly used binders.
It is readily soluble in water and freely soluble in alcohol and many other organic
solvents. It is available in a variety of grades of different molecular weights. PVP
is generally used in the form of a solution; however, it can be added to the blends
in the dry form and then granulated in situ. It is frequently used as a binder in effervescent
and chewable tablets because the tablets manufactured using PVP as a binder
generally harden with age. Aqueous or hydroalcoholic solutions of PVP are used to
granulate insoluble materials and alcoholic solutions are used for granulating soluble
materials. It is used as a binder at concentrations between 0.5% and 5%. Low- to
medium-viscosity grades are preferred since the high-viscosity grades of PVP have
been known to cause dissolution problems. It is highly hygroscopic and picks up significant
amounts of moisture at low relative humidities and can deliquesce at high
relative humidities.
2.2.2. Methyl Cellulose
Methyl cellulose (MC) is a long-chain substituted cellulose in which 27–32% of the
hydroxyl groups are in the form of methyl ether. It is available in a variety of grades
of different degrees of substitution and average molecular weight. Therefore, it offers
considerable latitude in binding strength. Efficiency of MC as a binder improves with
increasing molecular weight. Low- or medium-viscosity grades are preferred when
used as a binder. It may be added as a dry powder or as a solution. Although an
aqueous solution of 1–5% can be used to granulate soluble or insoluble excipients,
it is a better binder for soluble excipients such as lactose and mannitol.MC produces
granulations that compress easily. Granulation produced using 5% MC solution is
equivalent in hardness to 10% starch paste. It produces robust tablets with moderate
hardness, which does not increase with age.
MC is practically insoluble in hot water, ethanol, chloroform, ether, and saturated
salt solutions. In cold water, it swells and disperses slowly to form a clear to
opalescent, viscous dispersion. To produce an aqueous solution, the appropriate
quantity of MC powder is suspended in 25% of the required amount of water at
112 Hamed et al.
© 2005 by Taylor & Francis Group, LLC
80
C. The remaining amount of water is added cold, or ice water is added to the hot
slurry with vigorous stirring in order to cool it down to 20
C. A clear aqueous solution
of MC is obtained. MC can also be mixed as a dry powder with another powder
prior to mixing with cold water or may be moistened with an organic solvent such as
95% ethanol prior to the addition of water.
2.2.3. Hydroxypropyl Methyl Cellulose (Hypromellose)
Hydroxypropyl methyl cellulose (HPMC) is a propyleneglycol ether of MC. HPMC
is available in a variety of viscosity grades. Its binding properties are comparable to
those of MC. Concentrations of 2–5% (w/w) may be used as a binder in either wetor
dry-granulation processes. HPMC is soluble in cold water and forms a viscous
colloidal solution. To prepare an aqueous solution, HPMC is first hydrated in
20–30% of required water at 80–90
C with vigorous stirring. Cold water is added
to produce the required volume. Hydroalcoholic solutions or mixtures of water
and other water-miscible solvents such as glycol can also be used to dissolve HPMC.
HPMC is first dispersed in the organic solvent, at a ratio of five to eight parts of
solvent to one part of HPMC. Cold water is then added to produce the required
volume. HPMC is incompatible with some oxidizing agents.
2.2.4. Sodium Carboxymethyl Cellulose
Sodium carboxymethyl cellulose (Na-CMC) is a sodium salt of carboxymethyl ethers
of cellulose. It is available in a variety of molecular weights which influence the
viscosity of the solution and its swelling properties. It easily disperses in water at
all temperatures to form a clear colloidal solution. Its aqueous solubility varies with
its degree of substitution, which is the average number of hydroxyl groups substituted
per anhydroglucose unit. A 5–15% solution may be used for the granulation
of soluble as well as insoluble powders. The granulations produced using Na-CMC
as a binder are softer, but have good compressibility. It forms tough tablets of
moderate hardness. Na-CMC is a highly hygroscopic material. It can adsorb a large
quantity (>50%) of water at high relative humidities. Therefore, the tablets using
Na-CMC as a binder have a tendency to harden with age. Na-CMC is incompatible
with strongly acidic solutions and with metal salts of iron, aluminum, zinc, etc.
2.2.5. Ethyl Cellulose
Ethyl cellulose, an ethyl ether of cellulose, is available in a variety of grades, which
differ in their viscosity. Low-viscosity grades are used as binders in concentrations of
2–10% in ethanol. Ethyl cellulose produces softer granules, which compress into
tablets that easily disintegrate. However, the dissolution of the active ingredient from
these tablets may be slower because ethyl cellulose is insoluble in water. Ethyl cellulose
may be used in a dry form or in an ethanolic solution for wet granulation. Ethyl
cellulose is a good nonaqueous binder for water sensitive formulations.
2.2.6. Polyethylene Glycol
Polyethylene glycols (PEGs), by themselves, have limited binding action; however,
they can enhance the effectiveness of tablet binders and impart plasticity to granules.
They can also be used in thermoplastic granulations. In this process, a powder blend
containing 10–15% (w/w) of PEG-6000 is heated to 70–75
C to obtain a paste-like
Binders and Solvents 113
© 2005 by Taylor & Francis Group, LLC
mass, which forms granules if stirred while cooling. This technique is used in lozenge
formulations.
2.2.7. Polymethacrylates (Eudragit)
Polymethacrylates (Eudragit NE 30D and Eudragit RS 30D) can be used as binders
in aqueous or nonaqueous wet-granulation processes. They are supplied as 30% aqueous
dispersions. Dilution with water prior to use is recommended. Eudragit RS 30D
is incompatible with magnesium stearate.
2.2.8. Polyvinyl Alcohol
Polyvinyl alcohols (PVAs) are available in a variety of viscosity grades. Viscosity
ranges from 10 to 100 cps lend themselves for tablet granulations. PVAs are
water-soluble polymers. They form softer granulations, which yield tablets that do
not harden with age.
2.3. Sugars
2.3.1. Glucose (Dextrose)
Glucose, when applied as a syrup in concentrations >50% in wet-granulation processes,
exhibits good bonding properties. It produces moderately strong granules
and tablets, which are hard and brittle. Glucose is also used as a direct compression
tablet diluent and binder, primarily in chewable tablets. Anhydrous dextrose adsorbs
significant amounts of moisture at 25
C and 85% relative humidity to form a monohydrate.
The monohydrate also absorbs moisture at 85% relative humidity. Dextrose
is a reducing sugar and in its aldehyde form, can react with amines, amides, amino
acids, etc. Brown coloration may occur in tablets containing dextrose and strong
alkali or amines.
2.3.2. Sucrose
Sucrose is commercially available in several forms such as granular, fine granular,
fine, superfine, and confectioners sugar. The confectioners sugar is most commonly
used in wet-granulation formulations. Sucrose syrup, containing 50–67% (w/w)
sucrose, is used as a binder in wet-granulation processes. It can also be used as a
dry binder where it is granulated with water or hydroalcoholic solutions. Like glucose,
sucrose produces strong, but hard and brittle tablets. The amount of binder
determines the tablet hardness. Softer granules can be obtained by using hydroalcoholic
mixtures as granulating solutions. Finely divided sugar is hygroscopic. Tablets
containing large amounts of sucrose may harden with age, which may result in
slower disintegration. In systems where quick overwetting occurs, the amount and
the rate of addition of the sucrose syrup must be carefully monitored. Use of sucrose
with starch paste may improve the tablet quality. Sucrose is incompatible with aluminum.
It hydrolyses in the presence of acids. Contamination of powdered sucrose
with heavy metals may lead to incompatibility with substances such as ascorbic acid.
2.3.3. Sorbitol
Sorbitol is an optical isomer of mannitol. It is highly hygroscopic at relative humidities
of 65% and above. Therefore, it used as a humectant in pharmaceutical formulations.
114 Hamed et al.
© 2005 by Taylor & Francis Group, LLC
Whenever this property of sorbitol as a moisture control agent is desirable, it can be
used as a binder. Up to 2–20% of sorbitol can be added as 10–25% aqueous solution
in wet-granulation formulations.
2.4. New Binders
Recently, several new natural and synthetic binders have been investigated for potential
applications in pharmaceutical formulations. Nearly all the natural binders
investigated were gums obtained from different plant origins, including Khaya
gum (5), Leucaena leucocephala seed gum (6), Anacardium occidentale gum (7), Gellan
gum (8), and a combination of detarium gum and veegum (9). While some of
these materials have been approved by the U.S. FDA for use as food additives,
the issues of compendial requirements, lack of literature support, and potential
reproducibility problems associated with materials obtained from plant origin can
limit their applicability in the pharmaceutical industry. New synthetic binders that
have been investigated include maltrodextrins (10) and chitosan derivatives (11).
3. FACTORS INFLUENCING BINDER EFFICIENCY
The function of binders in a tablet formulation is to impart strength and to reduce
the friability of granules and tablets. Several factors influence the effectiveness of
a binder in a formulation. Some of these factors are related to the drug and other
excipients in the formulation and others are related to the binder and solvent
system used. Storage conditions as well as granulation process parameters can also
significantly affect the binder efficiency.
3.1. Properties of the Drug and Other Excipients in the Formulation
3.1.1. Particle Size
The primary particle size of the drug or excipients can affect granule strength, porosity,
and consolidation rate during high-shear granulation. Smaller particles with
higher surface areas have more contact points to allow the formation of stronger
granules than larger particles when granulated with enough binder (12,13). However,
because of the greater surface area, the solvent and binder requirements increase as
the primary particle size decreases (14–16). Therefore, to better anticipate the effect
of reducing the excipients’ primary particle size on granule properties, the degree of
liquid saturation and binder level must be considered. At the same binder and solvent
level, excipients with larger particle sizes consolidate easier and produce less
porous granules as compared to finer particles (17,18).
Another factor that can be related to the excipients’ primary particle size is the
distribution of drug through different size granules. When the drug particles were
smaller than the filler, larger granule size fractions exhibited the highest drug content
and when the drug particles were larger, the highest drug concentration was found in
the smallest granule size fraction (12). The results can be explained in terms of the
ability of small drug particles to form stronger granules that resist breakdown during
high-shear granulation and grow to a further extent than those formed primarily by
larger filler particles.
Binders and Solvents 115
© 2005 by Taylor & Francis Group, LLC
3.1.2. Solubility
The solubility of drugs and other excipients in the granulating solvent can signifi-
cantly affect the granules’ properties. Increasing the excipients’ solubilities in the
granulating solvent decreased the solvent requirement and led to the formation of
granules of tighter particle size distribution and reduced friability (19,20). Accordingly,
lactose-based granules are formed at a lower degree of water saturation compared
to insoluble fillers such as dicalcium phosphate (21,22) and calcium hydrogen
phosphate (23). The findings can be attributed to the increased plasticity of the moist
lactose agglomerates owing to its solubility in water, which increased the propensity
for particle coalescence and growth during high-shear granulation processes. Holm
et al. (24) described lactose granulation as less sensitive to moisture content and
kneading intensity than calcium hydrogen phosphate.
Changing the formulation proportions of water-soluble excipients can dramatically
alter the granules’ properties. The addition of microcrystalline cellulose to a
lactose-based formulation increased the solvent requirement and produced larger
granules for both PVP and gelatin-based granulations (25). Increasing the starch
content in a lactose:starch-based formulation leads to the formation of smaller granules
with a wider particle size distribution (26). The results were explained in terms of
water absorption by starch particles and the formation of weaker granules that could
not grow in size to the same extent as when starch was absent (27).
Drug solubility in the granulating solvent can affect its distribution in different
granule size fractions. Drugs with high solubility in the granulating solvent have a
higher tendency to migrate during drying creating a drug-rich crust. Abrasion during
subsequent milling leads to the formation of highly concentrated fines relative to
larger granules (13).
3.2. Properties Related to the Binder and Solvent System
3.2.1. Mechanical Properties of the Binder
The mechanical and film-forming properties of a binder determine the strength and
deformation behavior of a binder matrix. These properties of the binder matrix primarily
determine the effectiveness of the binder. The tensile strengths of the films of
acacia, gelatin, methylhydroxyethyl cellulose, PVP, and starch prepared with varying
moisture levels by equilibration at different relative humidities were reported (28).
mentioned earlier. The results show that acacia and PVP form weak films while gelatin
films possess the highest tensile strength. PVP was also reported to exhibit low
values of Young’s modulus showing that it is the most deformable binder (28). This
high deformability of PVP aids in consolidation during compaction. Consequently,
PVP is considered the most preferred wet binder by many (29).
3.2.2. Binder–Substrate Interactions
As mentioned earlier, significant determinants of the granule and tablet strength
are wettability of the substrate by the binder solution, binder cohesion, and
binder–substrate adhesion (30). Rowe (31–33), in a series of publications, presented
theoretical approaches to predict the binder–substrate interactions. According to
Rowe, dense nonfriable granules are expected when the spreading coefficient of binder
over substrate is positive while negative spreading coefficient leads to the formation
of porous granules (31). For a low-polarity substrate, such as griseofulvin, either
116 Hamed et al.
Figure 1 shows the effect of moisture content on the tensile strength of the binders
© 2005 by Taylor & Francis Group, LLC
PVP or starch would be the optimal binder while for high-polarity substrates, such as
theophylline, acacia or HPMC would be the optimal binder (31). PVP and HPMC
were reported to have positive spreading coefficient over lactose, acyclovir, and pentoxyfilline
with the latter having the highest spreading coefficient and acyclovir having
the lowest spreading coefficient (34). When granulated in fluid bed with both
binders, pentoxyfilline produced the least friable granules followed by lactose and
then Acyclovir. The results also showed that PVP is more efficient than HPMC
due to the lower work of cohesion and adhesion of the latter (34). This approach
is based solely on the hypothesis that optimum spreading of the binder is the main
criterion for successful formulation. It does not take into account other equally
important factors, such as disintegration, dissolution, and flow properties. However,
the studies show the potential of using the theoretical approach in binder selection
and formulation optimization.
PVP-based granulation of detergent-cleaned hydrophilic glass beads produced
larger granules with more crushing resistance and lower friability than those of
dimethylsilane surface treated hydrophobic glass beads (35). The results can be
attributed to the better adhesion of PVP to hydrophilic surfaces.
3.2.3. Binder Solution Viscosity and Surface Tension
To better understand the importance of binder solution viscosity and surface tension,
the mechanism of granule formation must be considered. When two particles
collide during granulation, they either coalesce or spring back from each other. It
is the balance of the kinetic energy of the colliding particles against the other forces
acting to bring the particles together that decides the outcome of the collision process
Figure 1 The effect of moisture content on the tensile strength of binder films. &, gelatin;
G, methylhydroxyethylcellulose; , starch; &, acacia; 
, PVP. The vertical bars show the
Binders and Solvents 117
limits of error of the means at P ? 0.95. (From Ref. 28.)
© 2005 by Taylor & Francis Group, LLC
(36,37). The forces that help dissipate the collision energy and keep the colliding particles
together include the static pendular bridge (surface tension effect), viscous and
interparticle friction forces (17).
Increasing binder solution viscosity increased the granule size and decreased
the amount of binder required to initiate granule growth in both high-shear (37)
and fluid-bed (36) granulation processes. However, when the binder solution viscosity
is very high, problems with binder spreading and distribution arise. The highest
binder solution viscosity that can be used depends on the excipients and drug properties
and the processing parameters, such as the ability of existing pump and/or spray
systems to handle viscous liquids. Experiments in high-shear granulators with initial
binder solution viscosities exceeding 100 mPa.s have been reported (38).
When different molecular weights of PVP and HPMC were dissolved in water,
reductions in surface tension were observed (39). The decrease in surface tension was
higher for HPMC but independent of the concentrations of both binders and inversely
dependent on the molecular weight of PVP (39). Decreasing binder solution surface
tension decreased the capillary suction pressure and decreased the friction
resistance to consolidation and, therefore, increased the granule consolidation rate,
but also increased the minimal attainable granule porosity, i.e., granules consolidated
fast but not as far (18). Decreasing the binder solution surface tension was also
reported to decrease the liquid requirement to attain overwetting (40).
3.2.4. Solvent Properties
Granule properties can be significantly affected by changing the granulating solvent.
The effect stems from the change in excipient solubility and wettability, as well as the
mechanism of granule consolidation. In the pharmaceutical industry, only aqueous,
hydroethanolic, and ethanolic solvent systems are widely used. Lactose granulations
prepared using ethanolic PVP solution had higher porosity and friability compared
to those prepared using water as a granulating solvent (41). For hydroethanolic solvent
systems, increasing the ethanol content has been shown to increase tablet
strength (41). The results were attributed to the increased fragmentation tendency
of granules when ethanol contents were increased (42). Changing the solvent system
can affect the formulation excipients’ wettability and influence binder distribution.
For example, PVP distribution in low-shear granulation process improved when
water was replaced with a hydroalcoholic solution (43).
3.3. Storage Conditions
Storage conditions can severely alter the tablet properties through their effect on the
binder’s physical characteristics. For example, the hardness and disintegration time
of Ranitidine tablets prepared using the PVP wet granulation method decreased
when the tablets were stored under elevated moisture levels (44). PVP can absorb
a significant quantity of water when exposed to elevated humidity, which decreases
the polymer glass transition temperature (45,46). PVP is proposed to stay in a glassy
stage at room temperature at a relative humidity 55% and completely convert
to the rubbery state at room temperature at a relative humidity >75% (45). Such
changes in the physical state of a binder can dramatically change a tablet’s hardness,
friability, disintegration, and drug release. Moreover, the presence of other additives
in the formulation can augment the plasticizing effect of moisture and cause the granule
and tablet properties to change at a lower moisture level than those used in accelerated
stability protocols.
118 Hamed et al.
© 2005 by Taylor & Francis Group, LLC
HPC was reported not to have a glass transition temperature, but rather melt
at a temperature of 190–195
C (46). When compared to PVP, tablets prepared using
HPC maintained its dissolution profile after storage for 24 weeks under different
moisture/temperature conditions while those prepared using PVP showed a reduction
in pyrydazine release when stored at 40
C/75% relative humidity (46).
4. PROCESSING PARAMETERS FOR COMMONLY USED BINDERS
4.1. Polyvinylpyrrolidone
4.1.1. High-Shear Granulation
The granulation process in high-shear equipment is influenced by a score of variables
including impeller speed, amount of binder and granulating solvent, liquid addition
rate, wet massing time, method of binder incorporation (wet vs. dry), method of
solution addition (dripping vs. atomization), equipment design, and chopper speed.
These parameters influence the binder performance through their combined effects
on the binder distribution/dissolution, the degree of saturation, and the extent to
which granules are allowed to grow and densify before they are broken down by
the high-speed chopper.
4.1.1.1. Impeller Speed. Impeller speed is the most significant parameter
affecting the binder performance in high-shear granulation. Increasing the impeller
speed in PVP and PVP–PVA copolymer-based formulations increases the granule
size (21,22,47–49) and the bulk and tapped density (21,49), but decreases the percent
fines (47,49) and the granule porosity (21,49). However, the effect of impeller speed
on binder performance is dependent on the degree of solvent saturation, the binder
amount, the stage at which the impeller speed is varied (liquid addition vs. wet massing),
and the properties of the formulation materials. At a low degree of solvent
saturation or with an insufficient amount of binder, the effect of increasing impeller
speed is in favor of breaking up the weak nuclei/agglomerates and hindering any
further granule growth (26,50).
4.1.1.2. PVP Concentration.
dicalcium phosphate granules vs. binder concentration for PVP in comparison to
other binders (51).
As the binder concentration increases, the crushing strength of the granules
increases. The plot also shows that starch and gelatin can produce much stronger
granules with lower concentrations compared to acacia, PVP, or PEG 4000 (51).
As expected, the particle size of granules also increases as the binder concentration
concentration (25).
The intragranular bonds formed during granule drying that contribute to the
tablet strength include cohesion of binder film and binder substrate adhesion. Jarosz
and Parrott (52) showed that the radial and axial tensile strengths of dicalcium
phosphate tablets increased with increasing concentrations of PVP in the formula-
High molecular weight PVP produces larger calcium hydrogen phosphate granules
than hydrolyzed gelatin and low-viscosity HPMC (38). When compared to PVP,
gelatin-based granules start growing at a lower degree of liquid saturation and the
dependence of granule size on binder concentration was more prominent (38). Cutt
et al.(53) compared glass ballotini granules prepared using PVP, hydrolyzed gelatin,
and HPMC. PVP produced the least friable granules at lower concentrations and
Binders and Solvents 119
Figure 2 shows a plot of crushing strength of
increases. Figure 3 shows a plot of mean particle size of lactose granules vs. binder
tion (Fig. 4).
© 2005 by Taylor & Francis Group, LLC
HPMC produced the most friable and smallest granules owing to its lower work of
adhesion and cohesion (53). Gelatin produced the strongest granules that needed
more work to compress due to their higher resistance to deformation (53).
4.1.1.3. Liquid Addition Rate and Wet Massing. Liquid addition rate can significantly
influence binder distribution and granular uniformity. Lower water addition
rate improves PVP–PVA copolymer distribution and reduces the liquid
requirement to obtain an acceptable granulation (26,54). Hydroalcoholic and alcoholic
solutions can behave differently due to heat generation and faster evaporation
in high-shear mixers. The effect of liquid addition rate is dependent on the solubility
of the formulation excipients in the solvent used. Slower PVP solution addition rate
enhances granule growth and decreases porosity of the insoluble dicalcium phosphate
granulations (21); lactose-based formulations, on the other hand, are less sensitive
to liquid addition rate (22,49). Increasing the wet massing time of PVP-based
formulations increases granules size and density, while decreasing the percentage of
fines and granule porosity (16,21,49,54).
4.1.1.4. Method of Binder Incorporation. PVP can be incorporated in the
granulation either by dry or by wet addition methods. The method of binder incorporation
affects the granule’s properties through its effect on binder dissolution and
Figure 2 Crushing strength of wet granulated dicalcium phosphate. Binders: , gelatin;
120 Hamed et al.
& , potato starch mucilage; , acacia; D, povidone; and 
, PEG 4000. (From Ref. 51.)
© 2005 by Taylor & Francis Group, LLC
distribution. However, given the good hydration and high solubility in combination
with the use of high impeller and chopper speed, PVP is usually uniformly distributed
when incorporated by either method using water, hydroalcoholic, or alcoholic
solvents. Nevertheless, a higher PVP level may be needed with the dry addition
method to obtain similar granulation results.
4.1.2. Fluid-Bed Granulation
4.1.2.1. PVP Concentration and Molecular Weight. Binder solution concentration
is one of the most crucial parameters controlling granule properties in fluid-bed
granulations. Increasing PVP concentration increases the granule size, narrows the
granule size distribution, and improves granule flowability and strength (55,56).
The increase in granule size can be related to the increase in binder solution viscosity
with a subsequent increase in droplet size together with an increase in the
force binding the particles at the higher binder concentration.
However, granule size enlargement with increasing PVP level reaches a plateau
at a critical PVP concentration (25,34). Increasing the PVP level beyond this critical
concentration does not lead to further growth of granules but rather widens the size
distribution and decreases the friability of the granules (25).
4.1.3. Roller Compaction
4.1.3.1. PVP Concentration. The effect of binder concentration on the
properties of granules and tablets prepared by roller compaction is dependent on
other processing parameters including feed rate, roll pressure, and speed. Increasing
the binder level increases tablet hardness and decreases their friability due to the
Figure 3 Plot of the mean granule size as a function of binder concentration for the granu-
Binders and Solvents 121
lation of lactose. (From Ref. 25.)
© 2005 by Taylor & Francis Group, LLC
imparted plasticity of the formed granules (57). However, at high binder level,
increasing the roll pressure increases the tablet friability and decreases the tablet’s
hardness due to the formation of less friable granules with higher fragmentation
resistance during compression (25). When the PVP level was reduced to 1%,
the granules properties were dependent on the drug properties (58). Increasing
the roll pressure in this case had no effect on tablet hardness or friability.
However, increasing the roll and feed rate can lead to a slight decrease in tablet
friability (58).
4.1.4. Extrusion/Spheronization
Several processing factors can affect the binder performance during extrusion and
spheronization processes including hydration level, extruder speed, shape, hole size,
spheronizer speed, and residence time. While the use of microcrystalline cellulose
alone or in combination with lactose is widely adopted, the addition of binders can
aid in the formation of a plastic mass than can be easily extruded and spheronized.
Increasing the PVP or starch/gelatin level increases the pellet size and decreases their
friability (59).
Figure 4 The influence of povidone on the tensile strength of tablets of dibasic calcium phosphate
dihydrate compressed at 2268 kg (______) and at 4536 kg (- - - - -). D, axial; 
, radial.
122 Hamed et al.
(From Ref. 52.)
© 2005 by Taylor & Francis Group, LLC
When microcrystalline cellulose was spray dried with PVP, HPC, HPMC, Na-
CMC, and pregelatinized starch, the properties of the produced pellets were
improved in terms of yield, sphericity, and surface smoothness (60). PVP and
HPC were described as superior to other binders in terms of water requirements,
process sensitivity, and yield (60).
4.2. Cellulosic Polymers
4.2.1. High-Shear Granulation
4.2.1.1. Polymer Concentration. The effect of cellulosic polymer concentration
on granule properties is dependent on their method of addition. In the wet addition
method, increasing the concentration of HPC, different viscosity grades of
HPMC and MC in lactose–cornstarch–microcrystalline cellulose systems increases
the percentage of binder-rich oversize granules (61). The effect can be attributed
to the increase in binder solution viscosity and heterogenous binder distribution.
In the dry addition method, the effect of increasing polymer concentration on a
granule’s properties depends on the nature of the polymer and its molecular weight/
viscosity. Increasing the concentration of HPC and low-viscosity HPMC (3 cP)
increases the median granule size (61). Increasing the concentration of mediumand
relatively high-viscosity HPMC (6 and 15 cP) and MC is accompanied by a
slight decrease in granule median size and increase in the percentage of oversized
granules (61). The different behavior can be related to the extent and rate of binder
dissolution as well as binder adhesiveness at different degrees of hydration.
4.2.2. Fluid-Bed Granulation
4.2.2.1. Polymer Concentration. Similar to a high-shear granulation process,
the effect of increasing HPMC and HPC concentrations on granule properties is
dependent on their method of addition in fluid-bed granulation. The polymers are
more uniformly distributed when added by the wet addition method and the granule
size and strength increase with increasing the polymer concentrations (62,63). In the
dry addition method, HPMC is more concentrated in the larger granules and
the granule size is independent of the concentration of HPMC (62,64). Agitating
the fluid bed, however, showed no difference in HPMC and HPC distributions
between the wet and dry addition methods (65). In general, HPC produced stronger
granules than HPMC when added in the dry state since it maintains its flexibility and
adhesiveness at low moisture content, which allows more binding during drying (65).
Increasing the level of MC, on the other hand, is not accompanied by an increase in
granule size or strength when it is included by both dry and wet addition methods
(65). MC has the lowest thermal gelation temperature and thus it loses its adhesiveness
at the low moisture level applied in fluid-bed granulation.
4.3. Gelatin
Gelatin was one of the most commonly used binders during the 1960s and 1970s.
Recently, the use of gelatin in pharmaceutical formulations has significantly reduced.
Many factors contributed to the lower popularity of gelatin including its high
Binders and Solvents 123
© 2005 by Taylor & Francis Group, LLC
binding ability and the resultant slow tablet disintegration, and perhaps most importantly,
the introduction of new semisynthetic binders whose methods of preparation
are easier and exhibited better reproducibility and reliability than gelatin.
In gelatin-based formulations, granule properties are highly sensitive to the
level of gelatin used (54,66). Gelatin produces stronger granules than CMC, PEG
(67,68), MC (69), acacia, and PVP (25) due to the high tensile strength of the gelatin
film compared to other binders. In fluid-bed granulation processes, increasing the
gelatin concentration increases the granular size up to a certain critical concentration
beyond which the granule size becomes independent of gelatin concentration (25).
The granulation method can affect the properties of granules and tablets prepared
using gelatin through their effect on gelatin distribution within the granules.
Acetaminophen granules and tablets prepared by wet massing, roller compaction,
and spray drying of the substrate-binder slurry were compared (70–72). Wet massing
produced a sponge-like binder matrix embedding substrate particles while the roller
compaction produced a distribution of discrete binder particles in the substrate
particles. Spray drying coated the granules with an outer shell of the binder and
produced superior tablets when compared to wet massing and roller compaction
(70–72).
4.4. Starch
Despite the reproducibility problems associated with a variety of starch sources and
the method of paste formation, starch paste is still a relatively popular binding agent
in the pharmaceutical industry owing to its superior compatibility with many drugs.
The granule properties are greatly affected by the method of starch paste preparation,
in particular the gelatinization temperature. Starches obtained from different
sources start gelatinization at different temperatures (corn starch starts at 75
C
and potato starch starts at 60
C) and yield pastes with different concentration gradient
viscosities (73). Increasing the temperature at which starch is gelatinized
increases the degree of gelatinization with a subsequent increase in tablet hardness
(73). Variability in ascorbic acid tablet properties decreases when starch paste was
prepared and pumped under precise temperature control (74).
Starch paste concentration can also affect granule properties. When compared
with diluted paste (5%, w/w) with good fluidity, a concentrated paste (15%, w/w)
with high viscosity produced lactose granules with a sharper particle size distribution
when using a high-shear mixer (75). Nevertheless, high impeller speed and longer
processing time was needed for deaggregation and a better distribution of the
concentrated starch paste. However, diluted starch paste produced stronger granules,
better compressibility, and longer disintegration time (76).
Starch can also be used as a disintegrant owing to its wicking and swelling
properties. Starch paste by itself is not sufficient to introduce effective disintegration
and a disintegrant, starch or other, is usually recommended to enhance the effect of
the starch introduced as a paste. When the quantity of disintegrant starch exceeds
that of paste starch, the disintegration time decreases as the compression force
increases. When paste starch is used in higher quantity than dry, the disintegration
time increases with compression force. The effect can be related to different disintegration
mechanisms introduced by disintegrant and starch pastes (77).
When starch is dextrinized by the addition of a-amylase, the resultant paste
produces stronger granules with better flowability that produce tablets with shorter
disintegration time, lower friability, and less variability in weight and hardness.
124 Hamed et al.
© 2005 by Taylor & Francis Group, LLC
However, excessive dextrinization leads to the formation of paste with lower adhesive
and tackiness properties and poor binding capabilities (78). Starch can be also
modified through pregelatinization or cross-linking. Cross-linking does not improve
the granulation ability of conventional starch. When compared to conventional
and cross-linked starches, pregelatinized starch yields coarser granules with lower
friability (79).
REFERENCES
1. Wade A, Walker PJ, eds. Handbook of Pharmaceutical Excipients. 2nd ed. London: The
Pharmaceutical Press, 1994.
2. Lieberman HA, Lachman L, Schwartz JB, eds. Pharmaceutical Dosage Forms: Tablets.
Vol. 1. 2nd ed. New York: Marcel Dekker, Inc., 1989.
3. Kristensen HG. Binders. Encycl Pharm Technol 1993; 451–465.
4. Mendes RW, Roy SB. Tableting excipients, part II. Pharm Tech 1978; 3:61–74.
5. Odeku OA, Itiola OA. Evaluation of Khaya gum as a binder in a paracetamol tablet
formulation. Pharm Pharmacol Commun 1998; 4:183–188.
6. Deodhar UP, Paradkar AR, Purohit, AP. Preliminary evaluation of leucaena leucocephala
seed gum as a tablet binder. Drug Dev Ind Pharm 1998; 24(6):577–582.
7. Chukwunweike Onukwo G, Okoye J. Evaluation of anacardium occidentale gum as
a binder in lactose based tablet formulations. Bole Chim Farm 1997; 136:569–574.
8. Antony PJ, Sanghavi NM. A new binder for pharmaceutical dosage forms. Drug Dev
Ind Pharm 1997; 23(4):417–418.
9. Attama AA, Adikwu MU, Esimone CO. Binary combination of detarium gum and veegum
as a binder in sodium salicylate tablets. Bole Chim Farm 1999; 138:199–202.
10. Soyeux P, Delacourte A, Delie B, Lefevre P, Boniface B. Influence of optimization of
operating parameters with a new binder in wet granulation. I: use in powder form.
Eur J Pharm Biopharm 1998; 46:95–103.
11. Aiedeh KM, Taha MO. Synthesis of chitosan triethylene glycol phthalate and its evaluation
as a binder in wet granulation procedures. Die Pharmazie 1999; 54:614–619.
12. Vromans H, Poels-Janssen HGM, Egermann H. Effect of high shear granulation on
granulate homogeneity. Pharm Dev Technol 1999; 4:297–303.
13. Van den Dries K, Vromans H. Relationship between inhomogeneity phenomenon and
granule growth mechanisms in a high-shear mixer. Int J Pharm 2002; 247:167–177.
14. Paris L, Stamm A. Optimal massing liquid volume determination by energy consumption
measurement: study of the influence of some physical properties of solvent and products
used. Drug Dev Ind Pharm 1985; 11(2,3):361–386.
15. Tapper GI, Lindberg NO. Granulation of some lactose qualities with different particle
size distributions in a domestic type mixer. Acta Pharm Suec 1986; 23:47–56.
16. Schaefer T, Holm P, Kristensen HG. Wet granuation in laboratory scale high shear
mixer. Pharm Ind 1990; 52(9):1147–1153.
17. Iveson SM, Lister JD, Ennis BJ. Fundamental studies of granule consolidation part 2:
effects of binder content and binder viscosity. Powder Technol 1996; 88:15–20.
18. Iveson SM, Lister JD. Fundamental studies of granule consolidation part 2: quantifying
the effects of particles and binder properties. Powder Technol 1998; 99:243–250.
19. Wells JI, Walker CV. The influence of granulating fluids upon granule and tablet properties:
the role of secondary binding. Int J Pharm 1983; 15:97.
20. Dias VH, Pinto JF. Identification of the most relevant factors that affect and reflect the
quality of granules by application of canonical analysis. J Pharm Sci 2002; 91(1):273–281.
21. Schaefer T, Holm P, Kristensen HG. Comparison between granule growth in a horizontal
and a vertical high speed mixer. I. Granulation of dicalcium phosphate.. Arch Pharm
Chem Sci Ed 1984; 14:1–16.
Binders and Solvents 125
© 2005 by Taylor & Francis Group, LLC
22. Schaefer T, Holm P, Kristensen HG. Comparison between granule growth in a horizontal
and a vertical high speed mixer. II. Granulation of lactose. Arch Pharm Chem Sci Ed
1986; 14:17–29.
23. Kristensen HG, Holm P, Jaegerskou A, Schaefer T. Granulation in high speed mixers.
Part 4: effect of liquid saturation on the agglomeration. Pharm Ind 1984; 46(7):763–767.
24. Holm P, Jaegerskou A, Schaefer T, Kristensen HG. Granulation in high speed mixers.
Part 1: effect of process variables during kneading. Pharm Ind 1983; 45(8):806–811.
25. Rohera BD, Zahir A. Granulations in a fluidized-bed: effect of binders and their concentrations
on granule growth and modeling the relationship between granules size
and binder concentration. Drug Dev Ind Pharm 1993; 19(7):773–792.
26. Holm P, Schaefer T, Larsen C. End point detection in a wet granulation process. Pharm
Dev Technol 2001; 6(2):181–192.
27. Holm P. Effect of impeller and chopper design on granulation in a high speed mixer.
Drug Dev Ind Pharm 1987; 13(9–11):1675–1701.
28. Healey JNC, Rubinstein MH, Walter V. The mechanical properties of some binders used
in tableting. J Pharm Pharmac 1974; (suppl 26):41.
29. Shangraw RF, Demarest DA Jr. A survey of current industrial practices in the formulation
and manufacture of tablets and capsules. Pharm Tech 1993; 17:32.
30. Readind SJ, Spring MS. The effects of binder film characteristics on granule and tablet
properties. J Pharm Pharmac 1984; 36:421–424.
31. Rowe RC. Binder–substrate interactions in granulation: a theoretical approach based on
surface free energy and polarity. Int J Pharm 1989; 52:149–154.
32. Rowe RC. Surface free energy and polarity effects in the granulation of a model system.
Int J Pharm 1989; 53:75–81.
33. Rowe RC. Polar/non-polar interactions in the granulation of organic substrates with
polymer binding agents. Int J Pharm 1989; 56:117–122.
34. Planinsek O, Pisek R, Trojak A, Srcic S. The utilization of surface free-energy parameters
for the selection of a suitable binder in fluidized bed granulation. Int J Pharm 2000;
207:77–88.
35. Cutt T, Fell JT, Rue PJ, Spring MS. Granulation and compaction of a model system. I.
Granule properties. Int J Pharm 1986; 33:81–87.
36. Ennis BJ, Tardos G, Pfeffer R. A microlevel-based characterization of granulation phenomena.
Powder Technol 1991; 65:257–272.
37. Hoornaert F, Wauters PAL, Meesters GMH, Pratsinis SE, Scarlett B. Agglomeration
behavior of powders in a lodige mixer granulator. Powder Technol 1998; 96:116–128.
38. Ritala M, Jungersen O, Holm P, Schaefer T, Kristensen HG. A comparison between binder
in the wet phase of granulation in a high shear mixer. Drug Dev Ind Pharm 1986;
12(11–13):1685–1700.
39. Parker MD, York P, Row RC. Binder–substrate interaction in wet granulation. 2: the
effect of binder molecular weight. Int J Pharm 1991; 72:243–249.
40. Pepin X, Blanchon S, Couarraze G. Power consumption profiles in high-shear granulation.
I: liquid distribution in relation to powder and binder properties. J Pharm Sci 2001;
90(3):322–331.
41. Wikberg M, Alderborn G. Compression characteristics of granule materials. VII. The
effect of intragranular binder distribution on the compactibilty of some lactose granulations.
Pharm Res 1993; 10(1):88–94.
42. Alderborn G. Granule properties of importance to tableting. Acta Pharm Suec 1988;
25:229–238.
43. Shah NH, Railkar AS, Phuapradit W, Zeng FW, Chen A, Infeld MH, Malick W. Effect
of processing techniques in controlling the release rate and mechanical strength of hydroxypropyl
methylcellulose based hydrogel matrices. Eur J Pharm Biopharm 1996; 42(3):
183–187.
126 Hamed et al.
© 2005 by Taylor & Francis Group, LLC
44. Uzunarslan K, Akbuga J. The effect of moisture on the physical characteristics of ranitidine
hydrochloride tablets prepared by different binders and techniques. Drug Dev Ind
Pharm 1991; 17(8):1067–1081.
45. Kiekens F, Zelko R, Remon JP. Effect of the storage conditions on the tensile strength of
tablets in relation to the enthalpy relaxation of the binder. Pharm Res 2000; 17(4):
490–493.
46. Fitzpatrick S, McCabe JF, Petts CR, Booth SW. Effect of moisture on polyvinylpyrrolidone
in accelerated stability testing. Int J Pharm 2002; 246:143–151.
47. Lindberg NO, Jo?nsson C. The granulation of lactose and starch in a recording high-speed
mixer, Diosna P25. Drug Dev Ind Pharm 1985; 11(2,3):387–403.
48. Vojnovic D, Chicco D, El Zenary H. Doehlert experimental design applied to optimization
and quality control of a granulation process in a high shear mixer. Int J Pharm 1996;
145:203–213.
49. Badawy SIF, Menning MM, Gorko MA, Gilbert DL. Effect of process parameters on
compressibility of granulation manufactured in high-shear mixer. Int J Pharm 2000;
198:51–61.
50. Vojnovic D, Selenati P, Rubessa F, Moneghini M. Wet granulation in a small scale high
shear mixer. Drug Dev Ind Pharm 1992; 18(9):961–972.
51. Armstrong NA, March GA. J Pharm Sci 1976; 65:200.
52. Jarosz PJ, Parrott EL. Factors influencing axial and radial strength of tablets. J Pharm
Sci 1982; 71:607–610.
53. Cutt T, Fell JT, Rue PJ, Spring MS. Granulation and compaction of a model system. III.
Compaction properties of the granules. Int J Pharm 1989; 49:157–161.
54. Lindberg NO, Jonsson C. Granulation of lactose in a recording high-speed mixer, Diosna
P25. Drug Dev Ind Pharm 1983; 9(6):959–970.
55. Aulton M, Banks M. The factors affecting fluidized bed granulation. Manuf Chem Aeros
News December 1978.
56. Wan LSC, Heng PWS, Ling BL. Fluidized bed granulation with PVP K90 and PVP
K120. Drug Dev Ind Pharm 1995; 21(7):857–862.
57. Sheskey PJ, Dasbach TP. Evaluation of various polymers as dry binders in the preparation
of an immediate-release tablet formulation by roller compaction. Pharm Tech
1995:98–112.
58. Murray M, Laohavichien A, Habib W, Sakr A. Effect of process variables on rollercompacted
ibuprofen tablets. Pharm Ind 1998; 60(3):257–262.
59. Varshosaz J, Kennedy RA, Gipps EM. Effect of binder level and granulating liquid on
phenylbutazone pellets prepared by extrusion–spheronization. Drug Dev Ind Pharm
1997; 23(6):611–618.
60. Law MFL, Deasy PB. Use of hydrophilic polymers with microcrystalline cellulose to
improve extrusion–spheronization. Eur J Pharm Biopharm 1998; 45:57–65.
61. Kokubo H, Nakamura S, Sunada H. Effect of several cellulosic binders on particle size
distribution of granules prepared by a high-speed mixer. Chem Pharm Bull 1993;
41(12):2151–2155.
62. Kokubo H, Nakamura S, Sunada H. Effect of several cellulosic binders on particle
size distribution in fluidized bed granulation. Chem Pharm Bull 1995; 43(8):
1402–1406.
63. Xu, G, Sunada H, Zhang R. Indomethacin controlled release matrix tablet prepared by
wet granulation procedure. Chem Pharm Bull 1997; 45(1):214–216.
64. Dasbach T, Sheskey P, Caswell O. Evaluation of the Use of Methocel in Fluid Bed Granulation
as a Dry Binder. Dow Product Information Publications, 2002.
65. Kokubo H, Nakamura S, Sunada H. Effect of several cellulosic binders and method of
their addition on the properties and binder distribution of granules prepared in an agitating
fluidized bed. Chem Pharm Bull 1998; 46(3):488–493.
66. Nagykaldi VA, Gyarmati L. Untersuchungen zur feuchtgranulation im diosna-mscher V
25. Pharm Ind 1980; 42(5):526–531.
Binders and Solvents 127
© 2005 by Taylor & Francis Group, LLC
67. Solvang S, Finholt P. Effect of tablet processing and formulation factors on dissolution
rate of the active ingredient in human gastric juice. J Pharm Sci 1970; 59(1):49–52.
68. El-Gindy NA, Smaha MW, El-Marandy HA. Evaluation of binder activities on the physical
properties and compression characteristics of granules prepared by two different
modes. Drug Dev Ind Pharm 1988; 14(7):977–1005.
69. Schaefer T, Worts O. Control of fluidized bed granulation. IV. Effect of binder solution
and atomization on granule size and size distribution. Arch Pharm Chem Sci Ed 1978;
6:14–25.
70. Seager H, Burt I, Ryder J, Rue PJ, Murray S, Beal N, Warrack JK. Int J Pharm Tech
Prod Manuf 1998; 1:36.
71. Seager H, Rue PJ, Burt I, Ryder J, Warrack JK. Int J Pharm Tech Prod Manuf 1981;
2:41.
72. Rue PJ, Seager H, Ryder J, Burt I. Int J Pharm Tech Prod Manuf 1980; 1:2.
73. Makino T, Kitamori N. Importance of gelatinization degree of starch paste binder in
hardness and disintegration of tablets. Chem Pharm Bull 1995; 43(3):514–516.
74. Makino T, Wada N, Kitamori N. Preparation temperature in fluidized-bed granulation
on resultant tablet properties. Chem Pharm Bull 1995; 43(3):473–475.
75. Makino T, Kitamori N. Effect of starch paste concentration on particle size distribution
of fine granules produced by agitation granulation. Chem Pharm Bull 1995; 43(7):
1231–1233.
76. Hill PM. Effect of compression force and corn starch on tablet disintegration time.
J Pharm Sci 1975; 65(11):1694–1696.
77. Hill PM. Starch paste granulation: binder dilution effects on granulations and tablets.
J Pharm Sci 1976; 65(2):313–314.
78. Bandyopadhyay AK, Chaudhuri B, Bhattacharjee PK. Starch paste granulation I: effect
of dextrinized corn starch mucilage as binding agent on granulation and tablet quality.
Aust J Pharm Sci 1980; 9(3):85–89.
79. Visavarungroj N, Herman J, Remon JP. Crosslinked starch as binding agent. I. Conventional
wet granulations. Int J Pharm 1990; 59:73–78.
128 Hamed et al.
© 2005 by Taylor & Francis Group, LLC
5
Spray Drying and Pharmaceutical
Applications
Metin C? elik
Pharmaceutical Technologies International, Inc., Belle Mead, New Jersey, U.S.A.
Susan C. Wendel
Elan NanoSystems, King of Prussia, Pennsylvania, U.S.A.
1. INTRODUCTION
Spray drying is one of the oldest forms of drying and one of the few technologies
available for the conversion of a liquid, slurry, or low-viscosity paste to a dry solid
ing process schematically. The simplicity and flexibility of the spray-drying process
make it ideal for handling a wide variety of pharmaceutical products.
1.1. Background
The first detailed description of the drying of products in spray form was mentioned
in a patent of 1872 entitled ‘‘Improvement of Drying and Concentration of Liquid
Substances by Atomizing’’ (2). However, this process found its first significant applications
in the milk and detergent industries in the 1920’s (3). In current times, spray
drying is utilized extensively in many aspects of our daily life from food products,
cosmetics, and pharmaceuticals to chemicals, fabrics, and electronics. Typical
pharmaceutical examples include spray-dried enzymes (such as amylase, protease,
lipase, and trypsin), antibiotics (such as sulfathiazole, streptomycin, penicillin, and
tetracycline) and many other active pharmaceutical ingredients, vitamins (such as
ascorbic acid and vitamin B12), and excipients for direct compression (such as lactose,
mannitol, and microcrystalline cellulose).
1.2. Advantages and Limitations
There are several reasons why the technology of spray drying has found many applications
in numerous industries. It is a continuous process. As long as liquid feed can
continue to be supplied to the drying system the spray-dried product will continue to
be produced. In some instances, this process has been operated for months without
interruption. The physical properties of the resulting product (such as particle size
129
(free-flowing powder) in one unit operation (1). Figure 1 shows a general spray-dry-
© 2005 by Taylor & Francis Group, LLC
and shape, moisture content, and flow properties) can be controlled through the
selection of equipment choices and the manipulation of process variables. The actual
spray-drying process is almost instantaneous as the major portion of the evaporation
takes place in as short a time as milliseconds or a few seconds at most, depending on
the design of the equipment and process conditions. This makes spray drying well
suited for heat-sensitive products. In addition, corrosive and abrasive materials
can be readily accommodated because the contact between the mechanical parts
and materials is minimal as compared to other granulation processes. Also, spray
dryers have few moving parts. In fact, careful selection of various components can
result in a system having no moving parts in direct contact with the product. Operation
requirements of small and large dryers are the same. This makes spray drying a labor
cost-effective process, especially for high-volume products. Last, the spray-drying
process can be fully automated. Commercial-scale spray dryers are controlled by programmable
logic controllers or solid-state controllers. These control systems monitor
exhaust air temperature or humidity and provide an input signal that, by way of a
setpoint, modulates the energy supplied to the process (4).
Like all other granulation processes, spray drying also has some limitations.
For example, it is not typically well suited for producing granules with mean particle
temperatures and the exhaust air stream contains heat, which often requires sophisticated
heat exchange equipment for removal.
2. SPRAY DRYING PROCESS STAGES
The spray-drying process is carried out in three fundamental stages as shown
Figure 1 A schematic representation of a general spray-drying process with primary and secondary
product separation.
130 C? elik and Wendel
size >200 mm as shown in Figure 2. It also has poor thermal efficiency at lower inlet
schematically in Figure 3. The first stage is atomization of a liquid feed into fine
© 2005 by Taylor & Francis Group, LLC
droplets. In the second stage, spray droplets mix with a heated gas stream and the
dried particles are produced by the evaporation of the liquid from the droplets.
The final stage involves the separation of the dried powder from the gas stream
and collection of these powders in a chamber. The second stage (i.e., the mixing
and drying step) has also been considered as separate steps (5). The following sections
detail each of these stages, their process parameters, and related equipment details.
2.1. Atomization
Atomization is the process by which a liquid is disintegrated into many fine droplets.
The formation of a spray with high surface/mass ratio is highly critical for optimum
liquid evaporation conditions and, consequently, the desired properties of the resulting
product. Although ideally the sizes of all droplets should be the same, in practical
terms formation of droplets with a narrow size distribution would be satisfactory.
2.1.1. Atomizer Types and Designs
Formation of the atomized spray requires application of a force. The commercially
available systems employ one of the following in order to create an atomized spray:
centrifugal energy, pressure energy, kinetic energy or sonic energy, and vibrations.
2.1.1.1. Centrifugal Atomizers. Centrifugal atomizers utilize either a rotating
disk or a wheel to disintegrate the liquid stream into droplets (6). Examples of rotary
wide variety of spray characteristics can be obtained for a given product through
combinations of feed rate, atomizer speed, and atomizer design. The droplet size distribution
is fairly narrow for a given method and process conditions but the mean
droplet size can be varied from as small as 15 mm to as large as 250 mm, depending
on the amount of energy transmitted to the liquid. Larger mean sizes require larger
drying chamber diameters. Wheels are well suited for producing sprays in the fine to
medium-coarse size range while disks are used to produce coarse sprays.
Rotary atomizers normally operate in the range of 5000–25,000 rpm with wheel
diameter of 5–50 cm. The mean size of the droplet produced is inversely proportional
Figure 2 Range of mean particle sizes achievable by control of atomizer operation at low to
medium feed rates.
Spray Drying and Pharmaceutical Applications 131
atomizers are shown in Figure 4. These devices form a low-pressure system and a
© 2005 by Taylor & Francis Group, LLC
to the wheel speed and directly proportional to the feed rate and its viscosity. Solid
content and surface tension are the other factors having minor effects on the droplet
size. For example, an increase in feed rate may slightly increase the particle size, but
the use of a variable-speed drive on the centrifugal atomizer facilitates correction to
the specified size.
Centrifugal atomizer designs include wheels with vanes or bushings and vaneless
disks. Vaned atomizer wheels produce sprays of high homogeneity and are the
most commonly used as compared to other designs. In this type of atomizer, liquid
fed onto a wheel moves across the surface until contained by the rotating vane.
Figure 3 Schematic of spray-drying process shown in stages: stage I: atomization; stage II:
spray–air contact and evaporation; stage III: product separation. (From Ref. 5.)
132 C? elik and Wendel
© 2005 by Taylor & Francis Group, LLC
The liquid flows outward under the influence of centrifugal force and spreads over
the vane, wetting the vane surface as a thin film. At very low liquid vane loadings,
the thin film can split into streams. No liquid slippage occurs on a wheel once liquid
has contacted the vanes. Whether radial or curved, the vanes prevent transverse flow
of liquid over the surface. Abrasive materials are best handled using atomizer wheels
with bushings. Since the feed material is in direct contact with rotating parts,
the bushings feature wear-resistant surfaces and require additional maintenance.
Vaneless (disk) designs are often applied when coarse powders are required at high
production rates.
Bulk pharmaceutical excipients and fine chemicals, such as antacids, are often
produced using centrifugal atomizers. The particles produced by this technique are
generally free flowing and, unless intentionally produced with very fine atomization,
dust free. The porous structure of the particles provides increased solubility and the
relatively low density and friability of these particles result in generally good compaction
properties. Also, the batch-to-batch reproducibility and dryer-to-dryer transferability
of this technique are excellent. As mentioned earlier, if larger spray-dried
particles are desired, larger production drying chambers must be employed.
2.1.1.2. Kinetic Energy Nozzles. Kinetic energy is applied in the form of two-
fluid or pneumatic atomization. This is the most commonly used atomization technique
within the pharmaceutical industry. Here, atomization is accomplished by the
interaction of the liquid with a second fluid, usually compressed air. High air velocities
are generated within the nozzle for effective feed contact, which breaks up the feed
into a spray of fine droplets. Neither the liquid nor the air requires very high pressure,
with 200–350 kPa being typical. A typical two-fluid nozzle is shown in Figure 5.
Particle size is controlled by varying the ratio of the compressed airflow to that
of the liquid. The main advantage of this type of atomization is that the liquid has a
relatively low velocity as it exits the nozzle; therefore, the droplets require a shorter
Figure 4 Common rotary atomizers. (Courtesy of Niro Pharma Systems.)
Figure 5 Schematic presentation of a typical two-fluid nozzle. (Courtesy of Niro Pharma
Systems.)
Spray Drying and Pharmaceutical Applications 133
© 2005 by Taylor & Francis Group, LLC
flight path for drying. Because many pharmaceutical applications use relatively small
spray dryers, pneumatic nozzles are often used. Another advantage is the simple
design that lends itself to easy cleaning, sterile operation, and minimal contamination.
Pneumatic nozzles can be designed to meet the most stringent requirements
for sterile or aseptic applications. Special consideration must be given to supplying
a sterile source of compressed air for atomization.
Another type of kinetic energy nozzle is the three-fluid nozzle. The spray
characteristics obtained by two- and three-fluid nozzles are similar when atomizing
low-viscosity feeds at up to intermediate feed rates. Use of the second air stream with
three-fluid nozzles causes a waste of energy, except for high feed rates of low-viscosity
feeds.
2.1.1.3. Pressure Nozzles. The second most common form of atomization
for pharmaceutical applications is hydraulic pressure nozzle atomization. Here,
the feed liquid is pressurized by a pump and forced through a nozzle orifice as a
high-speed film that readily disintegrates into fine droplets. The feed is made to
rotate within the nozzle, resulting in a cone-shaped spray pattern emerging from
the nozzle orifice. Rotary motion within the nozzle can be achieved by use of swirl
inserts or spiral grooved inserts (Fig. 6). The swirl inserts have comparatively large
flow passages and enable such nozzles to handle high solids feed without causing any
wear or clogging.
Because the liquid spray exits the nozzle with a relatively high velocity, a spraydrying
chamber at least 2.5m in diameter and 3.0m in cylinder height is usually
required to operate with pressure nozzles.
The differential pressure across the orifice determines the mean droplet diameter.
The distribution about the mean is similar to, but in most cases is narrower
Figure 6 Schematic presentation of pressure nozzles. (Courtesy of Niro Pharma Systems.)
134 C? elik and Wendel
© 2005 by Taylor & Francis Group, LLC
than, two-fluid atomization. In contrast, sprays from pressure nozzles handling high
feed rates are generally wider in distribution and coarser than sprays from vaned
wheels. At low feed rates, spray characteristics from nozzles and wheels are comparable.
Mean size of spray is directly proportional to the feed rate and inversely
proportional to the pressure.
Pressure nozzles are generally used to form coarse spray-dried particles (120–
300 mm mean particle size) with good flow properties. Antibiotics are a typical application
for such a dryer.
2.1.1.4. Sonic Energy Atomizers. The use of sonic energy and vibrations
for atomization in spray drying has found growing interest in the last two decades.
However, this type of atomizer has not yet found significant commercial applications.
The advantages of sonic nozzles operating at low pressure and having wide
flow channels suggest that they may be suitable for abrasive and corrosive materials,
but it is most likely that sonic nozzles will continue to be developed as atomizers for
special applications, such as very fine sprays of mean size 20 mm, where the nature of
the spray angle and cone minimizes droplet coalescence (7).
2.1.2. Atomizer Selection
The function of any atomizer is to produce as homogenous a spray as possible. The
nature of the feed, the characteristics of the spray, and the desired properties of the
resulting dried product play very important roles in the selection of the atomizer
type. With proper design and operation, nozzles and rotary atomizers can produce
sprays having similar droplet size distribution. In all atomizer types, the size of droplets
can be altered by either increasing or decreasing the atomization energy (e.g.,
increased atomization energy results in smaller droplet size). For a given amount
of energy, the viscosity and surface tension values of the feed influence the size of
the droplet (e.g., higher values result in larger spray droplets).
In general, rotary atomizers are utilized to produce a fine to medium-coarse
product with a mean size of 20–150 mm though larger spray-dried particles can also
be obtained if a very large drying chamber is used. Nozzle atomizers are used to produce
spray-dried product with a coarse mean particle size of 150–300 mm (8).
For a given spray-drying application, the selection between rotary and nozzle
atomizers involves the following considerations (9):
1. The feed capacity range of the atomizer for which complete atomization is
attained
2. Atomization efficiency
3. The droplet-size distribution at identical feed rates
4. Spray homogeneity
5. Operational flexibility
6. The suitability of dryer chamber design for atomizer operation
7. Feed properties
8. The atomizer experience available for the product in question.
2.2. Spray–Air Contact and Evaporation
Once the liquid is atomized, it must be brought into intimate contact with the heated
gas for evaporation to take place equally from the surface of all droplets. This contact
step takes place within a vessel called the drying chamber. The heated gas is
Spray Drying and Pharmaceutical Applications 135
© 2005 by Taylor & Francis Group, LLC
introduced into the chamber by an air dispenser, which ensures that the gas flows
equally to all parts of the chamber.
2.2.1. Spray–Air Contact
The way in which spray contacts the drying air is a critical factor in spray-drying
operations. Spray–air contact is determined by the position of the atomizer in relation
to the air inlet.
Inlet air is introduced to the drying chamber via an air disperser, which uses perforated
plates, or vaned channels through which the gas is equalized in all directions.
It is critical that the air entering the disperser is well mixed and has no temperature
gradient across the duct leading into it; otherwise, the drying will not be even within
the chamber. The air disperser is normally built into the roof of the drying chamber
and the atomization device is placed in or adjacent to the air disperser. Thus, instant
and complete mixing of the heated drying gas with the atomized clouds of droplets
can be achieved.
Spray droplet movement is classified according to the dryer chamber layout
and can be designated as cocurrent, countercurrent, or mixed flow although these
designations are not complete representations of the actual conditions.
1. Cocurrent flow is the configuration in which the spray and drying air pass
through the dryer in the same direction. This arrangement is widely used
and is ideal for heat-sensitive products. Spray evaporation is rapid, the drying
air cools accordingly, and overall evaporation times are short. The particles
are not subject to heat degradation. In fact, low-temperature
conditions are achieved throughout the entire chamber in spite of very
hot air entering the chamber.
2. Countercurrent flow is the configuration in which the spray and the air
enter at the opposite ends of the dryer. This arrangement has excellent heat
utilization. Countercurrent flow is used with nozzle atomization and is well
suited for meeting the final spray-dried properties of non-heat-sensitive
materials.
3. Mixed flow is the configuration in which both cocurrent and countercurrent
flows are incorporated. The advantage of this type of arrangement
is that coarse free-flowing products can be produced in relatively small drying
chambers. In mixed flow systems the powder is subjected to higher particle
temperature. A mixed flow system can be integrated with a fluid-bed
drying chamber when lower particle temperatures are necessary.
The spray–air contact design can be selected according to the required particle
size and the temperature to which the dried particle can be subjected. For example, if
a low product temperature must be maintained at all times, a cocurrent rotary atomizer
is selected for producing fine particles while a countercurrent pressure atomizer
is preferred for obtaining coarser particles. If coarse particles with predetermined
porosity and bulk density properties are desired, a countercurrent pressure nozzle
atomizer is well suited as high product temperature can be maintained for obtaining
the desired porosity and bulk density of the resulting product. For obtaining coarse
spray-dried particles of heat-sensitive materials, a mixed flow nozzle system can be
selected. Integration with a fluid bed is recommended for agglomerated or granulated
powders.
136 C? elik and Wendel
© 2005 by Taylor & Francis Group, LLC
2.2.2. Drying
The largest and most obvious part of a spray-drying system is the drying chamber.
This vessel can be tall and slender or have a large diameter with a short cylinder
height. Selecting these dimensions is based on two process criteria that must be
met. First, the vessel must be of adequate volume to provide enough contact time
between the atomized cloud and the heated gas. This volume is calculated by determining
the mass of air required for evaporation and multiplying by the gas residence
time, which testing or experience dictates.
The second criterion is that all droplets must be sufficiently dried before they
contact a surface. This is where the vessel shape comes into play. Centrifugal atomizers
require larger diameters and shorter cylinder heights. In contrast, nozzle atomizer
systems must have narrower and taller drying chambers. Most spray dryer
manufacturers can estimate, for a given powder’s mean particle size, the dimensions
that are needed to prevent wet deposits on the drying chamber walls.
2.2.3. Drying Gas
In pharmaceutical applications of spray drying, the feedstock can be prepared by
suspending or dissolving the product to be spray dried in water. However, the utilization
of a wide variety of organic solvents in feedstock preparations is also common.
Alcohols, such as ethanol, methanol, and isopropanol are preferred organic solvents
in spray drying of pharmaceuticals although other organic solvents such as ketones
are also used in other industries; often the synthesis process upstream from the drying
step determines the solvent selection. The drying characteristics of the solvents
are also important. For example, a solvent with a low boiling point may be the only
choice for heat-sensitive materials.
Although evaporating organic solvents by a spray drying process is very effi-
cient due to the resulting shorter residence time, as compared to the evaporation
of water, the risk of explosion makes the use of these solvents very hazardous. Therefore,
an inert gas, usually nitrogen instead of air, must be used as drying gas for the
evaporation of the solvents. Use of inert gas requires the use of a closed-cycle system
for spray drying in order to recover the solvent and to limit the gas usage. However,
for small drying tests and laboratory work, the nitrogen can be used without recirculation,
using a carbon bed on the exhaust gas to collect the solvent.
2.3. Dried Powder Separation
Powder separation from the drying air follows the drying stage. In almost every case,
spray-drying chambers have cone bottoms to facilitate the collection of the dried
powder.
Two systems are utilized to collect the dried product. In the first type of system,
when coarse powders are to be collected, they are usually discharged directly from
the bottom of the cone through a suitable airlock, such as a rotary valve. The gas
stream, now cool and containing all of the evaporated moisture, is drawn from
the center of the cone above the cone bottom and discharged through a side outlet.
In effect, the chamber bottom acts as a cyclone separator. Because of the relatively
low efficiency of collection, some fines are always carried with the gas stream. These
must be separated in a high-efficiency cyclone followed by a wet scrubber or in a
fabric filter (bag collector). Fines collected in the dry state (bag collector) are often
added to the larger powder stream or recycled. When very fine powders are being
Spray Drying and Pharmaceutical Applications 137
© 2005 by Taylor & Francis Group, LLC
produced, the side outlet is often eliminated and the dried product together with the
exhaust gas is transported from the chamber through a gooseneck at the bottom of
the cone. The higher loading of entrained powder affects cyclone design but has little
or no effect on the bag collector size.
In the second type of system, total recovery of dried products takes place in the
separation equipment. This type of system does not need a product-conveying system;
therefore, the separation efficiency of the equipment becomes very critical.
Separation of dried product from the air influences powder properties by virtue of
the mechanical handling involved during the separation stage. Excessive mechanical
handling can produce powders with a high percentage of fines.
3. PROCESS LAYOUTS
The most widely used spray-drying process layout is the open-cycle layout in which
the air is drawn from the atmosphere, passed through the drying chamber, and
exhausted back to the atmosphere. This layout is used for aqueous feedstock and
employs air as the drying gas. There are numerous variations of open-cycle layout
systems, two of which are common in pharmaceutical applications (Fig. 7) (10).
The most common and cost-effective layout utilizes a high-efficiency cyclone and
scrubber (Fig. 7A). In this layout, the loss of very fine particles to the atmosphere
cannot be prevented. If the desired particle size of the spray-dried product is too
small to be recovered by cyclone and scrubber systems then the use of a layout
employing a bag filter is recommended (Fig. 7B).
Closed-cycle layouts are mainly used for nonaqueous (i.e., organic solvents)
feedstock and generally require the use of inert gas as the drying medium. They
are also employed when flammable, explosive, or toxic products are used in the spray
flow diagram for a closed-cycle layout schematically (11). These systems require
a good control of the scrubbing–condensing stage at precise temperatures.
Figure 7 Typical layout of the open-cycle spray dryer system: (A) cyclone/scrubber and
(B) bag filter. a, air; f, feed; p, spray-dried product. 1, spray dryer chamber; 2, cyclone;
3, wet scrubber; 4, bag filter/collector. (Adapted from Ref. 10.)
138 C? elik and Wendel
drying process or atmospheric pollution is not permitted. Figure 8 illustrates the
© 2005 by Taylor & Francis Group, LLC
In addition to open-cycle and closed-cycle systems, there are semiclosed-cycle
layouts which are not strict in terms of type of drying medium and are operated
under slight vacuum conditions.
4. THEORY OF SPRAY DRYING FUNDAMENTALS
4.1. Droplet Formation
4.1.1. Rotary Atomizer
During rotary atomization, bulk liquid feed is accelerated to a high centrifugal velocity.
During this acceleration, the liquid feed forms a thin film over the rotating surface.
For smooth disk atomizers, the film or liquid feed disintegrates into droplets at
the edge of the wheel by one of the three mechanisms: (a) direct droplet formation,
type of droplet formation mechanism that occurs during processing is a function of
the surface tension and viscosity of the feed as well as the wheel speed and feed rate
(12). Direct droplet formation occurs at low wheel speeds when surface tension and
viscosity dominate the atomization mechanism. The other variables that could
potentially affect direct droplet formation are inertia and air friction. However,
due to liquid slippage on the surface of the wheel, inertia is limited and the low
release velocities minimize air friction effects so that the effect of these variables
is minimized at low wheel speeds. As wheel speeds and feed rates increase, the
amount of feed in each vane increases giving rise to ligaments instead of droplets
on the periphery of the wheel. These ligaments disintegrate into droplets with larger
droplets forming from feeds with higher viscosity and higher surface tension.
While the first two atomization mechanisms are partially controlled by the physical
properties of the feed, sheet formation is a result of inertial forces becoming predominant
over these properties. At high wheel speeds and feed rates, the ligaments join
to form a liquid sheet that extends beyond the edge of the wheel. The liquid sheet
Figure 8 Typical layout of the closed-cycle spray dryer system: c, coolant (diluent); f, feed; l,
solvent recovery; p, spray-dried product. 1, dried powder; 2, cyclone; 3, liquid-phase indirect
heater; 4, heat exchanger; 5, scrubber–condenser. (Adapted from Ref. 11.)
Spray Drying and Pharmaceutical Applications 139
(b) ligament formation, or (c) sheet formation as shown in Figure 9, respectively. The
© 2005 by Taylor & Francis Group, LLC
disintegrates into a broad droplet distribution as it extends from this edge. In order
to produce a narrow droplet distribution from this mechanism, high wheel speeds are
combined with low wheel loading, which is often achieved with a decreased feed rate.
In contrast, a vaned wheel directs the flow of the liquid feed across the surface
of an inner liquid distributor in which liquid slippage over the surface of the distributor
occurs until there is contact with the vane or channel. The feed then flows outward
due to centrifugal force and forms a thin film across the surface of the vane. As
the liquid film leaves the edge of the vane, droplet formation occurs as a result of the
radial and tangential velocities experienced. Atomizer wheel characteristics that
influence droplet size include speed of rotation, wheel diameter, and wheel design,
e.g., the number and geometry of the vanes.
4.1.2. Two-Fluid Nozzle
Using the two-fluid nozzle, also referred to as a pneumatic nozzle, atomization is
achieved by impacting the liquid feed with high-velocity air, which results in high
frictional forces that cause the feed to disintegrate into droplets. In order to achieve
optimal frictional conditions, this high relative velocity between liquid and air can be
accomplished by either expanding the air to sonic velocities or destabilizing the thin
liquid film by rotating it within the nozzle prior to spray–air contact.
There are several two-fluid nozzle designs available to produce the conditions
necessary for liquid–air contact. A common design is one in which the liquid and
air come into contact outside the nozzle. This nozzle is often referred to as an external
mixing nozzle and its main advantage is the greater control available over the
Figure 9 Smooth disk atomizer droplet formation mechanisms: (A) direct droplet formation,
(B) ligament formation, and (C) sheet formation. (From Ref. 12.)
140 C? elik and Wendel
© 2005 by Taylor & Francis Group, LLC
atomization through the independent control of both the liquid and the air streams.
Other two-fluid nozzle designs include: (a) an internal mixing design with the air and
liquid contacting within the nozzle head, (b) a combined internal–external mixing
design created by using two airflows in the nozzle head (also called three-fluid),
and (c) a pneumatic cup design with liquid–air contact occurring at the rim of
a rotating nozzle head.
In general, two-fluid nozzles are capable of producing small droplet sizes over
a wide range of feed rates. These droplets are then carried away from the nozzle by
the momentum of the spray and the expanding atomizing air. The most important
variable involved in the control of droplet size is the mass ratio of airflow to feed
rate, which is also known as the air-to-feed ratio. An increase in this ratio causes
a decrease in droplet size. This ratio generally ranges from 0.1 to 10. At ratios
approaching 0.1, atomization is difficult even for low-viscosity feeds while a ratio
of 10 approaches the limit above which atomization occurs using excess energy without
an appreciable decrease in particle size (13).
Sprays formed by two-fluid nozzles are symmetrical with respect to the nozzle
axis and have a cone-shaped pattern. The angle of this cone is called the spray angle
and, for two-fluid nozzles, it is narrow and cannot be varied greatly by adjusting the
air-to-feed ratio. The maximum spray angle available is 70–80
, which can be
obtained by employing the maximum feed rate and airflow in a high-throughput
nozzle. In general, an increase in air pressure will increase the spray angle if the feed
rate is maintained at a constant level as long as the maximum angle has not been
obtained. Spray angles are maintained if an increase in airflow is accompanied by
an increase in feed rate resulting in a similar air-to-feed ratio.
4.2. Droplet Drying Mechanisms
Evaporation of water from a spray is often characterized using a curve that describes
the change in drying rate as a function of time. This drying rate curve or evaporation
history is a function of temperature, humidity, and the transport properties of the
droplet formulation as well as the air surrounding the droplet. However, many characteristics
of droplet evaporation can be characterized using a general drying rate
The general drying rate curve has three main phases: an initial drying phase,
a phase in which the rate of drying is mainly constant, and a final phase during which
the rate of drying decreases (falling rate phase) (15). During the second phase, the
removal of moisture from the droplet is at a near-constant rate representing the highest
rate achieved during the evaporation history. This constant evaporation rate
results in a near-constant droplet surface temperature with the wet bulb temperature
representing the droplet temperature. During this phase, the majority of the droplet
moisture is removed. The droplet surface is maintained at saturation by moisture
migration from within the droplet to the surface. In contrast, during the falling rate
phase, the rate of moisture migration is rate limiting to the drying rate causing
a decrease in the overall rate of drying. The surface moisture content is no longer
maintained and the droplet temperature rises.
This general drying rate curve is directly applicable to the spray-drying process.
The initial drying phase begins during the spray–air contact phase immediately upon
contact of the droplet with the drying air. During this initial phase, as the drying rate
increases toward equilibrium, a slight increase in droplet surface temperature occurs.
Spray Drying and Pharmaceutical Applications 141
curve (Fig. 10).
© 2005 by Taylor & Francis Group, LLC
The drying rate continues to increase until equilibrium across the droplet–air interface
is established and the drying rate becomes constant. In the later phase, the solid
layer of the spray-dried particle becomes rate limiting to mass transfer and the drying
rate decreases. The evaporation rate continues to decrease until the droplet reaches
equilibrium moisture content with the surrounding air stream unless the product is
removed from the spray dryer before equilibrium moisture content is reached.
In addition, all evaporation histories, regardless of material type or spray dryer
configuration, have two main points in common: the majority of the evaporation
is completed in an extremely short time interval, usually <1.5 sec, and the temperature
of the drying air decreases rapidly during evaporation.
While the general evaporation history is representative of the processes occurring
during spray drying, the actual rate of moisture migration is affected by several
factors including the temperature of the surrounding air. If the inlet temperature is
so high that the evaporation rate is higher than the moisture migration rate needed
to maintain surface wetness, the constant rate drying phase is very short. This is
because a dried layer forms instantaneously at the droplet surface that acts as a barrier
to additional moisture transfer and retains moisture within the droplet causing
the surface temperature to be much higher. In contrast, lower inlet temperatures
actually yield a lower initial drying rate with a surface temperature equal to wet bulb
temperature for a longer period of time.
It is important to note that the drying curve is only representative. In reality,
there are no defined points during an evaporation history. Some phases may not
even occur or will be very short depending on the process conditions. One example
of this is a spray-drying process for a heat-sensitive material where the inlet temperature
is low. In this case, the initial phase may extend until a critical point where
moisture migration becomes rate limiting, effectively eliminating the constant rate
drying period. In reality, the actual evaporation rate is dependent on several factors
including the droplet shape, composition, physical structure, and solids concentration.
The actual drying time is a sum of the constant rate period and the falling rate
period until the desired moisture content is achieved.
Figure 10 General drying rate curve. (From Ref. 14.)
142 C? elik and Wendel
© 2005 by Taylor & Francis Group, LLC
4.3. Effect of Formulation on Droplet-Drying Mechanisms
Droplet composition also plays a significant role in droplet evaporation history.
Typically, sprays are differentiated into three main types: pure liquids, feeds containing
undissolved solids, and feeds containing dissolved solids (16).
4.3.1. Pure Liquid Sprays
For sprays composed of pure liquids, the droplet evaporates away completely. While
this type of spray is not useful for pharmaceutical formulations, its behavior is representative
of very dilute feed materials. The evaporation of pure liquids is dependent
on the dryer configuration. For low-velocity sprays in a low-velocity air stream
(countercurrent dryer) or for low-velocity sprays in a high-velocity air stream (cocurrent
dryer), the evaporation of the pure liquid spray causes the air temperature and
the evaporation rate to decrease. Pure liquid sprays having a wide droplet distribution
evaporate more quickly than narrow distributions having the same mean droplet
size due to the smaller droplets in the wide distribution. In addition, the size
distribution of the droplet changes during evaporation. If the initial spray is a homogenous
or very narrow distribution, the mean droplet diameter decreases during evaporation.
In contrast, if the initial spray is nonhomogenous or a wide distribution,
then the mean droplet diameter initially increases prior to decreasing. In general, a
distribution is the best representation of a spray since the mean of this distribution
may not adequately describe all characteristics of the distribution. For dryer configurations
of high relative velocities such as coarse atomization in cocurrent or fountain-
type dryers, the droplets travel farther before a given fraction is evaporated.
The relative velocity between droplet and drying air affects evaporation rates more
significantly at higher velocities and higher drying temperatures.
4.3.2. Feeds Containing Insoluble Solids
For droplets containing insoluble solids, the droplet temperature is equal to the wet
bulb temperature of the pure liquid droplet during the constant rate phase since insoluble
solids have negligible vapor pressure lowering effects. The total drying is the
sum of the two drying periods. The drying time for the first period is short compared
to the falling rate period. The falling rate period depends on the nature of the solid
phase and can be estimated given the specific gravity of the feed slurry, the density of
the dried product, and the thermal conductivity of the gaseous film around the droplet,
where gaseous film temperature is the average between the exhaust temperature
and droplet surface temperature. The droplet surface temperature is equal to the
adiabatic saturation temperature of the suspension spray.
4.3.3. Feeds Containing Dissolved Solids
Droplets containing dissolved solids have lower evaporation rates than pure liquid
droplets of equal size. The dissolved solids decrease the vapor pressure of the liquid,
thus reducing the driving forces for mass transfer. Drying results in the formation of
a solid crust at the droplet surface, which does not occur from pure liquid droplets.
Vapor pressure lowering causes droplet temperature to increase over wet bulb temperature
as in the previous two examples. Formation of dried solid during evaporation
has a significant effect on the subsequent evaporation history. During evaporation,
spray–air contact and constant rate period occur but may be shorter. The main effect
of dissolved solids is seen when the first period of drying ends and droplet moisture
Spray Drying and Pharmaceutical Applications 143
© 2005 by Taylor & Francis Group, LLC
content falls to critical value representing the formation of the solid phase at the surface.
During the falling rate period, the migration of moisture decreases due to resistance
to mass transfer caused by solid phase increasing. Last, the heat transfer is
greater than the mass transfer and the droplet temperature increases. Vaporization
of the moisture within the droplet during this phase may occur if the transfer is suffi-
ciently high.
The relationship between mass transfer and heat transfer for droplets containing
dissolved solids can lead to the formation of many different particle morphologies
depending on process conditions and material characteristics. Charlesworth and
Marshall have defined these morphologies as falling into two groups dependent on
the temperature of the drying air relative to the boiling point of the droplet solution
during the major part of the evaporation period (Fig. 11) (17).
If the air temperature exceeds the boiling point of the droplet solution, then a
vapor will be formed. As the solid crust forms around each droplet, vapor pressure
within the droplet is formed and the resultant effect of this pressure is dependent on
the nature of the crust. A porous crust will release the vapor, but a nonporous crust
may rupture resulting in fractured particles or fines from disintegrated particles.
Alternately, the droplet temperature may not reach boiling point levels due to
cocurrent airflow or because the residence time of droplets in the hottest regions of
the dryer is often very short. In this case, moisture migration occurs through diffusion
and capillary mechanisms.
In both cases, the porosity of the solid crust is often evident in the characteristics
of the falling rate period of the drying curve. If the film is highly nonporous, the
rate will fall sharply and the evaporation time will be prolonged. However, if a highly
Figure 11 Potential spray-dried particle morphologies in relation to process conditions and
material characteristics. (From Ref. 17.)
144 C? elik and Wendel
© 2005 by Taylor & Francis Group, LLC
porous film exists, then vapor is easily removed from the droplet–air interface and
the drying rate is similar to that found during the first period of drying.
These drying mechanisms result in a range of particle shapes including solid,
hollow, shriveled, and disintegrated, examples of which are shown in Figure 12.
However, it is important to note that particle morphology is also dependent on several
material characteristics including solubility, temperature of crystallization, melting
point, and thermal conductivity since they will also impact the rate of crust
formation, the porosity of the crust, and the subsequent drying rate.
It is also possible to influence particle density and size distribution through the
modification of process parameter settings such as atomizer settings, temperature
levels, and feed rates (18). For example, an increased feed rate while maintaining
a constant inlet temperature results in particles that have higher moisture content
and a resultant increased bulk density. By increasing the temperature of the feedstock,
the ability of the feed to be atomized is often improved due to the reduction
in ligament formation causing an increase in the bulk density of the dried particles.
Also, an increase in the concentration of the feed solids often increases the bulk density
of the dried particle, as does the use of a rotary atomizer, since many wheel
designs reduce air entrapment. Alternatively, bulk density may be decreased through
feed aeration or an increase in inlet temperature. Also, cocurrent spray-air contact is
often effective for reduction in bulk density because the wettest droplets encounter
the hottest air facilitating rapid evaporation and air entrapment. It is important to
note that the outlined process modifications are generally applicable, but that exceptions
to each can be found based on material characteristics.
In a similar manner, it is often possible to influence spray-dried particle
size distribution by changing process parameter settings. As mentioned earlier, the
size of the droplets formed during atomization is affected by process parameters
such as atomization type, atomizer settings, feed solids concentration, feed physical
Figure 12 Various particle forms of skim milk powder. (Courtesy of Niro Pharma Systems.)
Spray Drying and Pharmaceutical Applications 145
© 2005 by Taylor & Francis Group, LLC
properties, and drying temperatures. The size of the resultant particles following evaporation
is a function of the initial droplet size as well as the material characteristics
such as solid state and film formation mechanisms.
5. SPRAY DRYING APPLICATIONS
5.1. Feasibility Assessments
Before any spray-drying application work begins, it may be advantageous to conduct
the following simple, qualitative tests at the laboratory bench using very little material
to determine the feasibility of the application (19). A rheological profile of the
solution or suspension should be evaluated or, alternatively, a small sample can be
tested to see if droplets from a stirring rod can be readily formed. In the latter test,
if the liquid strings from the surface or forms peaks, then high viscosity is indicated
and the product may not be a candidate for spray drying without formulation
changes. The behavior of non-Newtonian fluids (pseudoplastic, thixotropic, dilatant,
etc.) has been found to influence atomization and the resultant droplet size (20).
However, while it is expected that Newtonian and non-Newtonian fluids atomize differently,
this difference was not found to be as important as the more significant
effect of the wheel speed on droplet size. It is also important to note that highly
viscous materials cannot be atomized by pressure nozzles.
Once the effect of viscosity has been evaluated, it may be advisable to dry a few
drops of product on a glass slide using a heated air gun. During this bench drying
test, the air temperature is recorded and the material is observed for the presence
of stickiness, color changes, or other physical changes. If the dried powder is found
to be suitable at the air temperature applied, it can be placed on a variable temperature
hot bench to determine the temperature at which the powder becomes tacky.
For spray drying to be successful, this temperature must be higher than the outlet
temperature of the dryer.
If the initial feasibility evaluation is successful, it is reasonable to commit additional
materials for a spray-drying trial. A laboratory dryer at least 500mm in diameter
is recommended for such tests. Bench-scale spray dryers are available but are
limited in their ability to provide adequate atomization or sufficient process air flow
for the successful production of dried particles. The laboratory unit, however, combined
with very fine atomization (two-fluid or rotary) will often produce acceptable
product for further testing. A series of tests can be performed at different inlet–outlet
temperature combinations using small quantities of material and these samples can
be tested for chemical stability to evaluate thermal effects from process air contact.
The relationship between outlet temperature and final product moisture can also be
established for this scale. While samples produced in a laboratory dryer are suitable
for evaluating the effect of spray drying on the product, they are not suitable for use in
downstream processing because the fine particle distribution produced as a result of
the small drying chamber dimensions may not be representative of the final spraydried
product.
Production of coarser particles requires a larger, pilot-scale dryer, which in turn
requires larger feed volumes. This pilot-scale work is often conducted at a spraysdrying
development center since many companies have laboratory dryers but few
have the sizes and variety of process types needed to fully develop a spray-dried product
from pilot scale through 1/10th of the commercial scale and into final production.
These facilities are usually found at spray-drying manufacturers or custom
146 C? elik and Wendel
© 2005 by Taylor & Francis Group, LLC
processing companies. In addition to having the equipment, manufacturers and
custom processors often have the expertise to more quickly optimize product
characteristics.
5.2. Spray Drying to Produce a Specific Type of Particle
Because of its inherent costs, spray drying is not always considered as a processing
option for many conventional formulations. However, when a specialized particle
type is required by the active ingredient or dosage form, spray drying can become
a feasible alternative to more conventional manufacturing processes. Such particle
types include microcapsules, controlled release particles, nanoparticles, and liposomes.
The application of spray drying to pharmaceuticals has been extensively
discussed in review articles (21,22).
5.2.1. Granulation
Spray drying is a unique process in several ways as compared to other granulation
methods. The feedstock is a homogenous liquid, which results in the uniform distribution
of all components of the spray-dried product in the same ratio in each individual
particle and eliminates the concerns of uniformly granulating dry components
with a liquid. While granule characteristics may exhibit batch-to-batch variation,
which in turn may influence the compaction behavior of the formulation or the
postcompaction properties of the tablet, granules produced using spray drying are
extremely consistent in terms of particle size, bulk density, and compaction behavior.
These features make spray drying a suitable process for the production of directly
compressible excipients such as lactose, microcrystalline cellulose, and mannitol.
Spray-dried lactose is by far the most commonly encountered spray-dried excipient
(21).
Many granulation methods utilize mechanical energy to transform very fine
particles into granules. Although shear forces are employed in nozzle and centrifugal
atomizers to create sprays, this form of energy will not destroy microencapsulated
material as can happen in high-shear granulation. In spray drying, some trial and
error is encountered in establishing the nozzle combinations and liquid pressures
to obtain equivalent particle size distribution during scale-up; however, the resultant
powder will have similar physical properties such as bulk density and compaction.
Also, it is important to note that, within the spray dryer, the product is never in contact
with moving parts, which facilitates the proper cleaning process greatly.
If the granulation size is a critical criterion for a given formulation, then the
selection of the granulation process may be determined based on the desired particle
limitations of numerous granulation techniques.
As seen in this figure, the spray-drying process results in smaller-size particles
as compared to some other granulation methods such as fluid-bed granulation or
high-shear (high-intensity) granulation. One option for producing larger agglomerates
using spray-drying technology is to employ fluidized spray drying, which combines
the features of the spray-drying process with fluid-bed granulation. The result
of this process is particles similar to those obtained from a fluid-bed granulation
operation, and yet the process is a continuous type in contrast to the batch operation
of a fluid-bed granulator. In a fluidized spray-drying system, the bottom cone of a
conventional spray dryer has been modified to include an integral fluid bed
Spray Drying and Pharmaceutical Applications 147
size and feasible operating temperatures. Figure 13 compares the general particle size
© 2005 by Taylor & Francis Group, LLC
(Fig. 14). When this process is implemented, atomization and spray–air contact
occur as they do in a conventional spray dryer. However, when the partially dried
particles reach the lower portion of the dryer, instead of undergoing product separation,
the particles are fluidized by the second stream of drying gas. Controlled temperature
and humidity of this fluid bed gas stream ensure that the particles retain
enough moisture to be suitable for agglomeration. At the end of the process, each
granule is an agglomerate of spray-dried droplets. Droplets that dry completely
before agglomerating and granules that experience attrition during fluid-bed drying
create fines, which are entrained in the gas stream and carried upward to the drying
gas exhaust. As a result, these fines pass through the atomized spray, providing an
additional opportunity for agglomeration. Fines that are carried through the exhaust
Figure 14 Schematic of fluidized spray drying system (FSDTM). (Courtesy of Niro Pharma
Systems.)
Figure 13 Particle size range of the methods utilized in particle growth. (From Ref. 21.)
148 C? elik and Wendel
© 2005 by Taylor & Francis Group, LLC
are collected in cyclones or a bag collector and can be pneumatically recycled into
the dryer.
A variation of this system is the integrated fluid-bed dryer (Fig. 15). This system
includes integrated filter bags, which are suspended from the chamber roof. This
roof is perforated and serves a dual purpose as a gas disperser. The chamber above
the roof contains clean gas that supplies the inlet process drying air and also has an
exhaust point for clean gas, since any remaining fines are entrapped in the filters.
Products produced using fluidized spray drying have a broader particle size distribution
and lower bulk density than the particles produced by conventional spray
dryers with a typical mean size particle size range of 150–400 mm. This process is not
meant to replace conventional spray drying processes but instead is a feasible alternative
for spray drying applications that require larger mean particle sizes.
5.2.2. Modification of Solid-State Properties
Characterization and modification of solid-state properties of drug substances are
profoundly important in developing pharmaceutical products with the desired drug
release properties. The importance of the process understanding and improvement of
the dissolution rate for poorly water-soluble drugs has been known for decades (24).
Particle size reduction methods (such as grinding, micronization, and ball milling),
precipitation, melt quenching, freeze drying, and spray drying have been used
extensively for improving the solubility and dissolution rate of poorly water-soluble
materials (25–30). These processes generally impart a polymorphic change in the
drug substance by transforming a low-energy crystalline form to a higher-energy
crystalline form or amorphous form.
Spray drying has many advantages over the other methods in achieving such
improvements. First, the spray-dried particles are generally free-flowing and spherical;
thus, no additional processing (such as dry granulation) is needed before compaction
of these particles. The hollow structure of the spray-dried particles
increases the solubility and subsequent dissolution rate of the drugs by several folds.
For example, the dissolution rate of poorly water-soluble salicylic acid was found to
be almost instantaneous and 60 times faster when spray dried as compared to that of
the original powder (25).
Figure 15 Schematic of an integrated fluid-bed dryer (IFDTM). (Courtesy of Niro Pharma
Systems.)
Spray Drying and Pharmaceutical Applications 149
© 2005 by Taylor & Francis Group, LLC
In addition, the energy of the amorphous state depends, to some extent, on the
method of preparation (26). The rapid nature of the spray-drying process (i.e., the
short residence time of the droplets during drying) improves the stability of otherwise
unstable amorphous forms.
Last, the spray-drying process is suitable for integration of stability agents
(such as PVP and PEG) into spray-dried particles. Spray drying of the poorly soluble
drug with 50% PVP resulted in enhanced dissolution when compared to a physical
mixture of micronized drug with PVP (27). A physically stable amorphous form
of ibuprofen, which has a low melting point, was obtained when spray dried in
the presence of 50–75% PVP (28). Recently, in a study on the effect of spray drying
varying lactose/PEG compositions, it was found that the most amorphous particles
were obtained when PEG was present at 10% (w/w) concentration. Conversion to
more crystalline materials occurred over time and the crystallization of lactose
appeared to be retarded at low PEG concentrations (29). In other work, the
increased amount and molecular weight of PVP was found to have the potential
to increase the physical stability of amorphous lactose (30).
5.2.3. Microencapsulation
The preparation of microcapsules involves the coating of particles or liquid droplets
with a biodegradable polymer. Applications for microspheres in the pharmaceutical
industry include controlled release, particle coating, flavor stabilization, taste masking,
and physical or chemical stabilization. Microencapsulation can be achieved through a
number of processes, but, in general, an active pharmaceutical ingredient (API) is
trapped within a reservoir or matrix. This process often begins with the preparation
of a three-phase, immiscible system containing a liquid vehicle, a core particle, and
a coating material or polymer. Several manufacturing techniques can be employed
to deposit the polymer around the particle and cause this coating to become rigid.
These methods include spray drying, Wurster fluid-bed coating, pan coating, coacervation,
and emulsion evaporation. In the spray-drying process, the encapsulation process
is achieved in one step in which desolvation and thermal cross-linking occur
concurrently and the particle is coated. A review of the main factors involved in the
application of spray drying for achieving microencapsulation references many works
which detail pharmaceutical applications, especially drug delivery systems (31).
Microencapsulation is a process that is often used for the purpose of providing
controlled release of a protein or drug. Several authors have studied microencapsulation
formulations manufactured from a spray-drying process as a means to achieve
controlled release. In one case, the effect of polymer hydrophilicity on API release
was evaluated and the most hydrophilic polymer was found to gel faster and retard
drug release the most (32). The size and cohesiveness of the resultant spray-dried
particles were found to be a function of the polymer and also affected drug release
with the smaller, more cohesive particles tending to agglomerate and delay drug
release. In another case, release of a model drug was controlled using a spray-dried,
water-activated, pH-controlled microsphere (33). Water influx into the microcapsule
caused buffer to dissolve and adjusted the inner pH causing the fraction of unionized
drug to increase resulting in the increased release of the drug.
One specific polymer type that has been employed in the spray drying of microspheres
to modify release is acrylic resin. A commercial blend of neutral methacrylic
acid esters was used for the preparation of spray-dried controlled release microcapsules
containing model drugs (34). Dissolution results of tablets compressed from the
150 C? elik and Wendel
© 2005 by Taylor & Francis Group, LLC
microspheres showed successful controlled release with advantages over a matrix
system. In a similar study, sustained release and enteric tablets were prepared by
directly compressing spray-dried microspheres produced using different types of
acrylic resins (35). Complete enteric properties were observed for tablets made from
pH-dependent, anionic acrylic polymers, while a sustained release profile was
observed for tablets made from microspheres containing pH-dependent, cationic
acrylic polymers.
Two common biodegradable polymers used in microencapsulation are polylactide
(PLA) and polylactide-co-glycolide (PLGA). The efficacy of spray drying as a
method for PLA and PLGA microsphere preparation was investigated using a model
lipophilic drug (36). The spray-drying process was tailored to each polymer and the
microspheres obtained were evaluated for shape, size, drug content, and polymer
influence on these characteristics. Polymer type, polymer molecular weight, and
polymer concentration were shown to be the greatest contributing factors to these
characteristics. In-vitro dissolution testing revealed different release profiles depending
on polymer type and microsphere morphology.
5.2.4. Inhalation Dosage Forms
In order for inhalation dosage forms to be clinically effective, the drug should
deposit in the lower airways. In general, the site of drug deposition in the lungs
depends on the particle size and size distribution of the drug particles or droplets,
the inhaler device and formulation, the patient’s breathing patterns, and airway geometry
(37). Generally, the aerosol particles or droplets must be <5 mm aerodynamic
diameter to be deposited into the lower respiratory tract.
A formulation of mucoadhesive microspheres for nasal administration was
examined through the preparation of microspheres containing active and one of
two polymer types using a spray-drying procedure (38). The mean diameter of the
spray-dried particles was 3–5 mm and surface morphology was dependent on polymer
type. Microspheres containing active and either polymer were more mucoadhesive
than any of the starting materials alone and the dissolution rate decreased with
increasing polymer content.
The ability to control the particle size and density of particles for inhalation
was investigated using lactose solutions atomized with a two-fluid nozzle and dried
in a laboratory scale dryer. It was found that droplet size during atomization was
affected by nozzle orifice diameter and atomization airflow but not by feed concentration.
However, dried particle size was influenced by feed concentration and it was
suggested that the shell thickness of the hollow particles increased with increasing
feed concentration (39).
An alternative method of atomization for the formation of respirable particles
is the airblast atomizer. This type of two-fluid nozzle introduces a liquid feed
pumped at a slow rate into a high-velocity gas stream via single or multiple jets. This
atomizer type was utilized at laboratory scale to evaluate the effect of grounded vs.
electrostatically charged tower configurations on the median particle size of the
spray-dried product (40). This study found significant differences between the two
configurations with the latter producing small particles but compromising collection
efficiency.
Spray Drying and Pharmaceutical Applications 151
© 2005 by Taylor & Francis Group, LLC
5.2.5. Nanoparticles
Attempts have been made to manufacture particles on the nanometer scale for
applications such as controlled release and intravenous delivery systems. A comparison
evaluating the processability and solid dosage performance of spray-dried
nanoparticles and microparticles was conducted (41). In this study, nanoparticle suspensions
were prepared by wet comminution in the presence of stabilizers, converted
into dried particles using a spray-drying process and subsequently compressed.
Compacts prepared from microparticles and nanoparticles were found to differ in
their internal structure and micromechanical deformations.
In another study, solid, lipid nanoparticles were produced using high-pressure
homogenization and loaded with drug using hot or cold methods for lipophilic or
hydrophilic drugs, respectively (42). Surfactant addition was investigated and stability
and entrapment efficiency were evaluated. Long-term sterile storage of these dispersions
was difficult and spray drying was investigated as a potential, feasible
technique.
The feasibility of developing nanoparticles for aerosol delivery has also been
investigated (43). The spray-dried nanoparticles produced using one carrier type
were found to be hollow while others had a continuous matrix. Particle size was measured
before spray drying and after the spray-dried powder was re-dissolved. Both
carrier types resulted in an increase in particle size after spray drying, although both
were found to remain in the nanometer range after drying and were suitable for effi-
cient lung delivery.
5.2.6. Liposomes
Another particle type capable of being produced by spray drying is liposomes. Traditional
preparation of liposomes begins with the preparation of a solution containing
the lipids to be used in a volatile organic solvent mixture. Following filtration of the
solution, the solvent mixture is removed under conditions which ensure that phase
separation does not occur. The dry lipid mixture is then hydrated by an aqueous
mixture containing the drug to be entrapped. Last, this mixture is dried. Spray drying
is one method available for accomplishing one or both of these drying steps. For
example, lipid vesicles were produced using a spray-drying process instead of the first
step of the traditional process (44). Vesicles containing phosphatidyl choline
(soybean lecithin) were produced by extruding the phospholipid through a 0.2 mm
polycarbonate membrane followed by spray drying with 10% lactose. The particle
size, vesicle size distribution, and stability of the multilamellar vesicles were measured.
The mean particle diameter after spray drying with a rotary atomizer was
7 mm and the dry particles could be reconstituted in water to liposomes without
any major change to the vesicle size distribution. In addition, the chemical stability
of the liposomes was not significantly affected by the spray-drying process. In subsequent
work, the same authors utilized spray drying for the hydration step of the
traditional process (45).
5.2.7. Peptides and Proteins
Recent advances in biotechnology have made it possible to use macromolecules such
as peptides and proteins as therapeutic agents. Spray drying has been used for decades
for processing antibiotics, vaccines, and, for the last few decades, macromolecular
drugs.
152 C? elik and Wendel
© 2005 by Taylor & Francis Group, LLC
The effect of spray drying process parameter settings on the activity of peptides
and proteins is often difficult to study. Consequently, enzymes are frequently used as
model protein drugs due to the ease with which their activity can be determined.
Investigations of the application of spray drying for the production of some enzymes
and proteins and the effects that processing parameters have on enzyme activity have
been discussed in a review article (21).
Many proteins and peptides are susceptible to degradation upon spray drying
due to relatively high temperatures. In a recent study, the effects of inlet and outlet
temperatures on some spray-dried peptides and proteins were reported (46). In
another study, enzyme activity was found to be susceptible to spray-drying temperature
and only half of its activity remained after spray drying without additives at outlet
temperatures <50
C (47). In this study, it was found that the activity of a
formulation consisting of enzyme and mannitol was maintained at outlet temperatures
<50
C and compromised at temperatures >50
C. Replacing mannitol with trehalose
stabilized the spray-dried enzyme and its activity was maintained at 100% at
an outlet temperature of 100
C.
5.2.8. Dry Elixirs and Emulsions
A dry elixir is a novel dosage form developed by spray-drying actives and excipients
dissolved or suspended in ethanol/water mixtures. One example is a dry elixir in
which the feed solution contained actives, dextrin, and sodium lauryl sulfate in a
mixture of ethanol/water (48). The spray-dried product was spherical in shape with
a smooth surface and a mean diameter of 13 mm. A comparison with the active in
powder form revealed a major decrease in dissolution time from >60 to 2 min.
A similar dosage form to the dry elixir is the dry emulsion. In this case, the
emulsified drug or oily drug solution with additives is spray dried to produce dry
emulsion particles. A dry emulsion of a water-insoluble nutrient was studied and
release from the spray-dried particle was found to be dependent on the type and
amount of oily carrier and surfactant used (49). Differences in release among the different
formulations were attributed to the differences in the physical state of the drug
and surfactant in the dried particle.
5.2.9. Effervescent Products
Spray-dried particles have also been incorporated into effervescent products. In one
study, spray drying was used to protect a degradation-sensitive active by coating fine
particles of the drug with a sugar alcohol solution (50). In vivo results of tablets
made using the spray-dried particles combined with coated citric acid and sodium
bicarbonate revealed that the active was rapidly absorbed from the tablet.
5.2.10. Other Process Variations
Two variations of the spray-drying process have been developed in response to product
requirements. The first variation is spray congealing. In this process, solids such
as wax or monoglycerides are melted. Other ingredients such as drugs, flavors, or fragrances
are dissolved or suspended in the molten material. This molten feed is
sprayed using the same basic spray-drying equipment except that no heat source is
required. Depending on the freezing point of the feed, ambient or chilled air may
be used during the drying process. This process has been described in more detail
and a comparison between particles produced by both the spray drying and the
Spray Drying and Pharmaceutical Applications 153
© 2005 by Taylor & Francis Group, LLC
spray congealing techniques has also been drawn (51). One study compared microcapsules
produced using both methods and found that the solvent used, the lipid
type, and the chain length were variables that influenced the surface properties of
both particle types (52,53).
A second variation of the spray drying process is spray freeze drying. In this
process, the feed is sprayed into freezing air causing the droplets to freeze. The frozen
droplets are subsequently sublimed under vacuum conditions producing a dry product.
One study investigated this method further by eliminating the use of vacuum
conditions for sublimation (54). In this study, the feasibility of spraying pharmaceutical
solutions at atmospheric pressure was investigated using very low air temperatures
and desiccated air for the removal of the water from the frozen particles. The
process resulted in fine, free-flowing powder with a high surface area, good wetting,
and good solubility characteristics.
Another type of atomization employed for pharmaceuticals is supercritical
fluid nebulization. The process uses carbon dioxide as an aerosolization aid, which
permits drying at lower temperatures than is usually needed in conventional spray
drying (55). Within the atomization system, supercritical carbon dioxide is intimately
mixed with aqueous solutions containing API, often proteins or peptides. The outcome
is the formation of microbubbles, which are rapidly dried in <5 sec, resulting
in dried particles predominately <3 mm in diameter (56,57). This method is generally
applied for the production of materials for pulmonary use or to achieve increased
bioavailability (58).
6. CONCLUSION
Spray drying has found many applications in numerous industries. The scale-up of
the spray drying process is less troublesome as the operation requirements of small
and large dryers are the same when compared to other conventional granulation
processes such as high-shear granulation method.
Spray drying, being a continuous process, is well suited for the production of
bulk drug substances and excipients. Using this process, the physical properties of
the resulting product (such as particle size and shape, moisture content, and flow
properties) can be controlled through the selection of equipment choices and manipulation
of process variables; thus, the final spray-dried particulate matter may not
need further processing (wet or dry granulation) before compaction. In addition,
the spray-drying process matches the directives outlined in the Process Analytical
Technology Initiative which is currently being guided and championed by the
FDA (59).
Spray-drying processes offer several advantages when solid-state properties of
drug substances need to be modified. Using this process, solubility and dissolution
rates of properties of poorly soluble materials can be increased several fold and
the stability of the amorphous form of the materials can be improved significantly.
Because of its initial inherent costs, spray drying is not always considered as a
processing option for many conventional formulations, especially for small batch
size operations. However, when a specialized particle type is required by the active
ingredient or dosage form, spray drying can become a feasible alternative to more
conventional manufacturing processes. Such particle types include microcapsules,
controlled release particles, nanoparticles, and liposomes.
154 C? elik and Wendel
© 2005 by Taylor & Francis Group, LLC
ACKNOWLEDGMENTS
We gratefully acknowledge the contributions of Niro Pharma Systems, specifically
Bob Turock and Jim Schell, for providing the content for several of the figures and
Fred Shaw, whose chapter on spray drying in the first edition of this book provided
guidance and content. We also acknowledge the significant contribution of Keith
Masters whose handbook was referenced extensively as a leading text in the field.
REFERENCES
1.
2. Percy SR. Improvement in drying and concentrating of liquid substances by atomizing.
US Patent 125,406, April 9, 1872.
3. Masters K. Introduction. Spray Drying Handbook. 5th ed. Essex, U.K: Longman Scientific
Technical, 1991:1–20.
4.
5. Masters K. Spray drying fundamentals: process stages and layouts. Spray Drying Handbook.
5th ed.. Essex, U.K: Longman Scientific Technical, 1991:24.
6. Masters K. Spray drying fundamentals: process stages and layouts. Spray Drying Handbook.
5th ed. Essex, U.K: Longman Scientific Technical, 1991:26.
7. Masters K. The process stages of spray drying. Spray Drying Handbook. 5th ed. Essex,
U.K: Longman Scientific Technical, 1991:268.
8. Masters K. Spray drying fundamentals: process stages and layouts. Spray Drying Handbook.
5th ed.. Essex, U.K: Longman Scientific Technical, 1991:31.
9. Masters K. The process stages of spray drying. Spray Drying Handbook. 5th ed. Essex,
U.K: Longman Scientific Technical, 1991:271.
10. Masters K. Spray drying fundamentals: process stages and layouts. Spray Drying Handbook.
5th ed. Essex, U.K: Longman Scientific Technical, 1991:43.
11. Masters K. Spray drying fundamentals: process stages and layouts. Spray Drying Handbook.
5th ed. Essex, U.K: Longman Scientific Technical, 1991:47.
12. Masters K. The process stages of spray drying. Spray Drying Handbook. 5th ed. Essex,
U.K: Longman Scientific Technical, 1991:199.
13. Masters K. The process stages of spray drying. Spray Drying Handbook. 5th ed. Essex,
U.K: Longman Scientific Technical, 1991:255.
14.
15. Masters K. Drying of droplets/sprays. Spray Drying Handbook. 5th ed. Essex, U.K:
Longman Scientific Technical, 1991:309.
16. Masters K. Drying of droplets/sprays. Spray Drying Handbook. 5th ed. Essex, U.K:
Longman Scientific Technical, 1991:326.
17. Charlesworth DH, Marshall WR. Evaporation for drops containing dissolved solids.
AIChE J 1960; 6(1):9–23.
18. Masters K. Drying of droplets/sprays. Spray Drying Handbook. 5th ed. Essex, U.K:
Longman Scientific Technical, 1991:345.
19. Shaw F. Spray drying as a granulation technique. Parikh DM, ed. Handbook of Pharmaceutical
Granulation Technology. New York: Marcel Dekker, 1997:93–96.
20. Filkova I, Weberschinke J. Apparent viscosity of non-Newtonian droplet on the outlet of
wheel atomizers. Mujumdar AS, ed. Drying ’82. New York: McGraw Hill, 1982:
165–170.
Spray Drying and Pharmaceutical Applications 155
Traub DA. Spray Dryers, Part 1, Process Heating, August 2001. hhttp://www.processheating.
com/CDA/ArticleInformation/Drying_Files_Item/0,3274,61137,00.htmli.
Traub DA. Spray Dryers, Part 2, Process Heating, September 2001. hhttp://www.processheating.
com/CDA/ArticleInformation/Drying_Files_Item/0, 3274, 63175, 00.htmli.
Traub DA. The Drying Curve, Part 2, Process Heating, October 2002. hhttp://www.process-
heating.com/CDA/ArticleInformation/Drying_Files_Item/0, 3274, 84744, 00.
htmli.
© 2005 by Taylor & Francis Group, LLC
21. Broadhead SK, Rouan E, Rhoads CT. The spray drying of pharmaceuticals. Drug Dev
Ind Pharm 1992; 18(11,12):1169–1206.
22. Wendel SC, C? elik M. An overview of spray-drying applications. Pharm Technol 1997;
21(10):124–156.
23. Kadam KL. Granulation Technology for Bioproducts. Boca Raton, FL: CRC Press,
1990.
24. Fincher JH. Particle size of drugs and its relationship to absorption and activity. J Pharm
Sci 1968; 57(11):1825–1835.
25. Kawashima Y, Satio M, Takenaka H. Improvement of solubility and dissolution rate of
poorly water-soluble salicylic acid by a spray-drying technique. J Pharm Pharmacol 1975;
27:1–5.
26. Pikal MJ, Lukes AL, Land JE, Gaines K. Quantitative crystallinity determination for
beta-lactam antibiotics by solution calorimetry: correlations with stability. J Pharm Sci
1978; 67:767–773.
27. Junginger K, Wedler M. Acta Pharm Technol 1984; 30(1):68.
28. Corrigan OI, Holohan EM, Reilly MR. Physicochemical properties of indomethacin and
related compounds co-spray dried with polyvinyl-pyrrolidone. Drug Dev Ind Pharm
1985; 11(2,3):677–695.
29. Corrigan DO, Healy AM, Corrigan OI. The effect of spray drying solutions of polyethylene
glycol (PEG) and lactose/PEG on their physicochemical properties. Int J Pharm
2002; 235(1,2):193–205.
30. Berggren J, Alderborn G. Effect of polymer content and molecular weight on the
morphology and heat- and moisture-induced transformations of spray-dried composite particles
of amorphous lactose and poly(vinylpyrrolidone). Pharm Res 2003; 20:1039–1046.
31. Re MI. Microencapsulation by spray drying. Drying Technol 1998; 16(6):1195–1236.
32. Wan LS, Heng PW, Chia CG. Spray drying as a process for microencapsulation and the
effect of different coating polymers. Drug Dev Ind Pharm 1992; 18(9):997–1011.
33. Sutinen R, Laasanen V, Paronen P, Urtti A. pH-controlled silicone microspheres for controlled
drug delivery. J Controlled Release 1995; 33:163–171.
34. Palmieri GF, Wehrle P, Stamm A. Evaluation of spray-drying as a method to prepare
microparticles for controlled drug release. Drug Dev Ind Pharm 1994; 20(18):2859–2879.
35. Takeuchi H, Handa T, Kawashima Y. Controlled release theophylline with acrylic polymers
prepared by spray drying technique. Drug Dev Ind Pharm 1989; 15(12):1999–2016.
36. Pavanetto F, Genta I, Giunchedi P, Conti B. Evaluation of spray drying as a method for
polylactide and polylactide-co-glycolide microsphere preparation. J Microencap 1993;
10(4):487–497.
37. Gonda I. Targeting by deposition. Hickey AJ, ed. Pharmaceutical Inhalation Aerosol
Technology. New York: Marcel Dekker, Inc., 1992:61–82.
38. Vidgren P, Vidgren M, Aronen P, Vainio P, Nuutinen J. Nasal distribution of radioactive
drug administered using two dosage forms. Eur J Drug Metab Pharmacokinet 1991;
3:426–432.
39. Elversson J, Millqvist-Fureby A, Alderborn G, Elofsson U. Droplet and particle size
relationship and shell thickness of inhalable lactose particles during spray drying.
J Pharm Sci 2003; 92:900–910.
40. Dunbar CA, Concessio NM, Hickey AJ. Evaluation of atomizer performance in production
of respirable spray-dried particles. Pharm Dev Technol 1998; 3(4):433–441.
41. Lee J. Drug nano-and microparticles processed into solid dosage forms: physical properties.
J Pharm Sci 2003; 92(10):2057–2068.
42. Muller-Mehnert RH, Lucks JS, Schwarz C, Ruhl D, et al. Solid lipid nanoparticles (sln)-
alternative colloidal carrier system for controlled drug delivery. Eur J Pharm Biopharm
1995; 41(1):62–69.
43. Sham JO, Zhang Y, Finlay WH, Roa WH, Libenberg R. Formulation and characterization
of spray-dried powders containing nanoparticles for aerosol delivery to the lung. Int
J Pharm 2004; 269:457–467.
156 C? elik and Wendel
© 2005 by Taylor & Francis Group, LLC
44. Goldbach P, Brochart H, Stamm A. Spray-drying of liposomes for a pulmonary administration.
Part 1. Chemical stability of phospholipids. Drug Dev Ind Pharm 1993;
19(19):2611–2622.
45. Goldbach P, Brochart H, Stamm A. Spray-drying of liposomes for a pulmonary administration.
Part 2. Retention of encapsulated materials. Drug Dev Ind Pharm 1993;
19(19):2623–2636.
46. Costantino HR, Andya JD, Nguyen PA, Dasovich N, Sweeney TD, Shire SJ, Hsu CC,
Maa YF. Effect of mannitol crystallization on the stability and aerosol performance of
a spray-dried pharmaceutical protein, recombinant humanized anti-ige monoclonal antibody.
J Pharm Sci 1998; 87:1406–1411.
47. Broadhead J, Ruan SK, Rhodes CT. The effect of process and formulation properties on
the properties of spray dried b-galactosidase. J Pharm Pharmacol 1994; 46:458–467.
48. Kim CK, Soon YS. Development of digoxin dry elixir as a novel dosage form using
a spray-drying technique. J Microencap 1995; 12(5):547–566.
49. Takeuchi H, Sasaki H, Niwa T, Hino T, Ozawa H, et al. Design of redispersible dry
emulsion as an advanced dosage form of oily drug (vitamin E nicotinate) by spray-drying
technique. Drug Dev Ind Pharm 1992; 18(9):919–937.
50. Timmington H. Improved aspirin. Chemist Druggist 1973; 199:482–483.
51. Killen MJ. Process of spray drying and spray congealing. Pharm Eng 1993; 13(Jul–Aug),
56:58–62.
52. Eldem T, Speiser P, Altorfer H. Polymorphic behavior of sprayed lipid micropellets and
its evaluation by differential scanning calorimetry and scanning electron microscopy.
Pharm Res 1991; 8(Feb):178–184.
53. Eldem T, Speiser P, Hincal A. Optimization of spray-dried and -congealed lipid micropellets
and characterization of the surface morphology by scanning electronmicroscopy.
Pharm Res 1991; 8(1):47–54.
54. Mumenthaler M, Leuenberger H. Atmospheric spray-freeze drying: suitable alternative
in freeze dry technology. Int J Pharm 1991; 72(27):97–110.
55. Sievers RE, Huang ETS, Villa JA, Kawamoto JK, Evans MM, BrauerPure PR. Lowtemperature
manufacturing of fine pharmaceutical powders with supercritical fluid aerosolization
in a bubble dryer. Appl Chem 2001; 73(8):1299–1303.
56. Sievers RE, Karst U. Methods for fine particle formation. U.S. Patent 5,639,441, June
1997.
57. Sievers RE, Karst U. Methods and apparatus for fine particle formation. U.S. Patent
6,095,134, August 2000.
58. Sievers RE, Karst U, Milewski PD, Sellers SP, Miles BA, Schaefer JD, Stoldt CR, Xu
CY. Formation of aqueous small droplet aerosols assisted by supercritical carbon dioxide.
Aerosol Sci Tech 1999; 30:3–15.
59. U.S. Department of Health and Human Services, Food and Drug Administration, Center
for Drug Evaluation and Research (CDER), Center for Veterinary Medicine (CVM),
Office of Regulatory Affairs (ORA). Guidance for Industry: PAT—A Framework
for Innovative Pharmaceutical Development, Manufacturing, and Quality Assurance,
September 2004.
Spray Drying and Pharmaceutical Applications 157
© 2005 by Taylor & Francis Group, LLC
6
Roller Compaction Technology
Ronald W. Miller
Bristol-Myers Squibb Company, New Brunswick, New Jersey, U.S.A.
1. INTRODUCTION
This chapter is different from the original; new works, findings, and best practices
from various investigators since 1997 have been added. The interested reader will
have to wait for a comprehensive book (in preparation now) to link numerous journal
articles, chapters, and investigative works for technology completeness.
2. POWDER GRANULATION AND COMPACTION
Powder granulation is a process of powder size enlargement that incorporates small
particles into larger ones. The definition of granulation comprises a range of different
size enlargement methods that can be classified as either dry or wet. In wet methods, a
suitable liquid is used to agglomerate the small powder particles into a mass. The wet
mass is subsequently dried and sized for further downstream processing needs. Wet
granulation methods have been the most widely used powder granulation technology
in the production of pharmaceutical products, particularly in modern pharmaceutical
manufacturing.
The chief reasons to granulate powders for the manufacture of pharmaceutical
dosage forms are described by Kristensen and Schaefer (1):

 To improve powder flow properties for dosage filling and compression
processes

 To eliminate wet granulation induced degradants and to improve product
stability

 To prevent active product ingredient from segregating

 To reduce bulk volume, thereby minimizing storage and enhancing transport

 To reduce potential environmental and safety hazards.
Kristensen and Schaefer provide ample literature references in their chapter in
the Encyclopedia of Pharmaceutical Technology about granulation size enlargement
methods (1): Capes (2,3), Pietsch (4,5), Sherrington and Oliver (6), Kapur (7). Others
referenced about wet granulation technologies are Kristensen and Schaefer (8),
Lindberg (9), Fonner et al. (10), Anderson et al. (11), and Ghebre-Sellassie (12).
159
© 2005 by Taylor & Francis Group, LLC
3. BACKGROUND
Granulation methods are used in the pharmaceutical industry to enlarge and densify
small powder particles into larger ones typically to improve powder flow so that the
material can be processed effectively and efficiently further into solid dosage forms.
There are two methods of granulation, wet and dry. Wet granulation methods are
widely described in the literature: Encyclopedia of Pharmaceutical Technology (Granulations)
is solely devoted to this technology (1). Dry granulation operations do not
use moisture or heat to process powders into densified granules. The pharmaceutical
industry employs two methods of dry granulation: slugging and roller compaction.
Little has been written about pharmaceutical dry granulation technology. Its
contemporary use in the industry is 50–60 years, beginning in the late 1940s. However,
its popularity has risen in the last 15 years in parallel with the increased
research on new efficacious active pharmaceutical ingredients (API) in the pharmaceutical
industry. A number of these new API cannot be processed so easily using
wet granulation and drying processing steps, because of their chemical fragility
and sensitivity. Therefore, this pushes the need for the use of dry granulation processing
techniques to advance new API in the 21st century.
Briefly, in dry pharmaceutical granulation processing, the powder particles are
aggregated under high pressure, typically a pressure of 30–70 bar. Particulate matter
can be aggregated when compressed at high pressure because of bonding forces developed
by the direct contact between the solid surfaces. The high pressure serves to
improve the contact area between the surfaces and thus the overall bonding strength.
Sometimes a binding agent is needed to provide additional bonding strength.
In the pharmaceutical industry, dry granulation processing in the 1950s–1970s
favored a process called slugging. This process design consisted of feeding powder
into a large compression machine, such as a Stokes D3 type compression machine,
where the powder was compressed into large tablets or slugs, typically in the
order of 1 in. diameter with a tablet gauge of about 0.25 in. The tablet slugs were
subsequently milled by a separate sizing machine to an appropriate particle size
distribution, and further processed into pharmaceutical capsules, powder for oral
suspensions, sachets, or tablet dosage forms. The slugging process is still used today
by only a few manufacturing firms that have old pharmaceutical formulation processes.
Today, modern pharmaceutical formulation processes introduced into the
Americas, Western Europe, Australia, and parts of Asia do not use this kind of
dry granulation equipment in newly developed formulations. The slugging process
is a relic of the past in modern pharmaceutical technology; roller compaction is
the key technology to future dry granulation processing.
Some characteristics are described briefly about the slugging process to complete
the technology information for the reader. The slugging process is externally
influenced by raw material feed properties such as powder cohesiveness, density, flow
characteristics, and powder particle size distribution. The slugging machine’s design
characteristics such as machine type, feed hopper, feed frame, die diameter, tooling
features, compression speed, and slugging pressure also influence the slugging process
and the final product properties. In general, the key processing operational
aspect of slugging is to maintain a uniform powder fill weight into the dies during
the dynamics of the slugging process. This assures the best chance to manufacture
uniform powder slugs and, ultimately, uniform densified granules. The compression–
slugging setup is a key essential to maximizing the slugging throughput and
minimizing the hopper feed-frame and die powder flow problems associated with
160 Miller
© 2005 by Taylor & Francis Group, LLC
the process. Slugging compression is normally performed at 4–6 tons hydraulic pressure,
at a rate of 10–30 turret revolutions per minute. The specific machine tonnage,
turret speed, and roll dwell time required for the process are dependent on the
powder blend’s physical properties, the tooling configuration, machine parts, and
ultimately the slug specifications. Typical slugging machine output ranges from 30
to 50 kg/hr and the machines are not instrumented with modern devices to control
their performance. There are many disadvantages with the slugging technology in the
pharmaceutical industry (Table 1).
4. BENEFITS OF ROLLER COMPACTION
This chapter briefly identifies key aspects of roller compaction technology. Unlike
the slugging process technology, roller compaction technology is well suited for
dry granulation agglomeration in the era of modern development of active pharmaceutical
ingredients in pharmaceutical plants.
The increasing scale of manufacturing pharmaceutical products worldwide, the
need for high processing rates, together with increased levels of good manufacturing
practices, necessitate controlled dry granulation processes with as few processing
steps as possible. This has been accomplished by instrumenting roller compactors
to automate and control the mechanical process. Roller compaction technology
plays a very important role in providing competitive cost control, safety, and quality
products in the pharmaceutical industry. Key roller compaction benefits are identi-
fied in Table 2.
Table 1 Disadvantages of Slugging
Single batch processing Excessive air and sound pollution
Frequent maintenance changeover Increased use of storage containers
Poor process control Increased needs of manufacturing space
Poor economies of scale Increase of logistics
Low manufacturing throughput per hour More energy and time required to produce
1 kg of slugs than 1 kg of roller compact
Source: From Ref. 13.
Table 2 Advantages of Roller Compaction
Simplifies processing Uses less raw materials
Facilitates powder flow Eliminates water-induced degradants
Uses minimal energy to operate Improves process cycle time
Requires less man-hours to operate Prevents particle segregation
Improves drug dosage weight control Facilitates continuous manufacturing
Reproduces consistent particle density Improves content uniformity
Produces good tablet and capsule
disintegration
Does not require explosion proof room/
equipment
Eliminates aqueous and solvent granulating Produces a dry product that is process
scaleable
Source: From Ref. 13.
Roller Compaction Technology 161
© 2005 by Taylor & Francis Group, LLC
5. COMPACTION THEORY
The bonding forces in a dry aggregate are important to granulation properties such
as granule integrity, flowability, friability, density, compressibility, and size for
downstream manufacturing process steps (1). Rumpf and coworkers described the
bonding mechanisms occurring during dry granulation as a mixture of van der
Waals forces, mechanical interlocking, and a recombination of bonds established
between freshly created surfaces, and solid bridges, created because of partial melting
and solidification during compression (14). A general theory describes particle bonding
related to roller compaction in the Handbook of Pharmaceutical Granulation
Technology (13). The process of dry granulation relies on interparticulate bond formation.
Granule bond formation is characterized in different stages, which usually
occur in the following order:
1. Particle rearrangement
2. Particle deformation
3. Particle fragmentation
4. Particle bonding.
Particle rearrangement occurs initially as powder particles begin filling void
spaces. Air begins to leave the powder blend’s interstitial spaces, and particles begin
to move closer together. This action increases the powder blend’s density. Particle
shape and size are key factors in the rearrangement process. Spherical particles will
tend to move less than particles of other shapes because of their close initial packing
to one another. Particle deformation occurs as compression forces are increased.
This deformation increases the points of contact between particles where bonding
occurs and is described as plastic deformation (13).
Particle fragmentation follows as the next bonding stage. This occurs at
increased compression force levels. At this stage, particle fracturing creates multiple
new surface sites, additional contact points, and potential bonding sites. Particle
bonding occurs when plastic deformation and fragmentation occur. It is generally
accepted that bonding takes place at the molecular level, and that this is due to
the effect of van der Waals forces (13).
When powder granules undergo an applied force or stress, a stress force is
released from the granules. The granules attempt to return to their original shape
or form; this is described as elastic deformation. A deformation that does not totally
recover after the stress is released is a plastic deformation. Elastic and plastic deformations
can occur simultaneously, but one effect usually predominates.
Parrott identified three theories of compression bonding: mechanical, intermolecular,
and liquid-surface film. Mechanical bonding purports that individual particles
undergo elastic, plastic, and brittle deformation. Bonding of this nature occurs
because particle surfaces intertwine, forming mechanical bonds. Intermolecular theory
identifies that there are some unsatisfied surface ions that have a potential need to
bond to one another. Under pressure, intermolecular forces become pushed together
close enough so that van der Waals forces can act to consolidate particles. The liquidsurface
film theory identifies that bonding occurs because of the existence of a thin
liquid film. The thin liquid film is generated from pressure induced by the energy
of compression. This mechanism acts as a bonding agent promoting mechanical
strength and an enlarged particle (15). Very little information is written about this last
theory.
162 Miller
© 2005 by Taylor & Francis Group, LLC
Dehont et al. provided a simplified approach to roller compaction theory (16).
They described that powder granules move through stages in the feed area. The
material is drawn into the gap by rubbing against the roll surfaces. The densification
that occurs in this area is particle rearrangement. At this stage, the speed of the
powder is slower than the peripheral speed of the rollers. Figure 1 represents compactor
rolls in the horizontal plain; powder is pushed vertically downward into
the compaction area.
Note in Figure 1, a ? nip angle, b ? material in volume space. The material is
located in the compaction area between a and the horizontal axis (Fig. 1). At this
stage, the material undergoes additional compaction forces. The particles undergo
plastic deformation and are bonded. Dehont’s team noted that nip angle varies
according to the material characteristics of particle size and density and the angle
is about 12 (16). They defined the neutral angle, g, which corresponds to the point
where the pressure applied by the rollers is the greatest on the material. They also
defined elastic deformation, d, and that occurs after the compact begins leaving
the compression roll area. Compacted flakes may increase in size due to material
elastic deformation and actually may have a larger thickness than the roll gap, e (16).
Dehont et al. developed Eq. 5.1 for the linear variation of flake thickness at a
specific roll diameter (16):
e1 ? D?d0=d1  d0??1  cos a? ?5:1?
where e1 is flake thickness, D is roll diameter, d0 is material density at angle a, and d1
is flake density. Dehont et al. assumed that the material in the compaction
area remains horizontal and moves at the peripheral speed of the rollers. They
also considered that the angle a is independent of the roller diameter size and noted
that the flake thickness e1 depends on the roller speed, the roller surface, and the
compaction pressure. All these parameters influence the density of the flake d1.
Dehont et al. concluded that if the same flake thickness were obtained with different
roller diameters, the flake density would be greater with larger-diameter rollers (16).
This is due to the greater nip angle formed with the larger rolls allowing more material
to be compacted.
R. W. Heckel considered the compaction of powders analogous to that of a
first-order chemical reaction. The pores were the reactant and the densification of
the material the product. The proportionality between the change of the density with
the pressure and the pore fraction was the process kinetics (17). Heckel explained
Figure 1 Front view of compactor rolls in horizontal plane. (From Ref. 16.)
Roller Compaction Technology 163
© 2005 by Taylor & Francis Group, LLC
mathematical constants that described the compaction behavior of a given powder
and developed a mathematical relationship, Eq. 5.2 (17):
ln?1=1  D? ? KP ? ln?1=1  D0? ?5:2?
where D is the relative powder density, D0 is the relative loose powder density at zero
pressure, P is the pressure applied, 1  D is the pore fraction, and K is the proportionality
constant.
The expression of density–pressure relationship permitted the determination of
density values in the range of the pressures investigated. Heckel described mathematically
that the curved region when plotting ln(1/1  D) vs. P is associated with powder
densification (17). This occurred by a mechanism of individual particle movement
in the absence of interparticle bonding. Heckel concluded that the densification represented
by the linear region of the plot, ln (1/1  D) vs. P, occurred by plastic
deformation of the compact after an appreciable amount of interparticle bonding
had taken place. For quantitative reasons, Heckel, redefined the mathematical
expression denoting a constant, A ? ln(1/1  D0), which is quantitatively valid
except at low pressures. He postulated that the constant A, which is somewhat larger
than ln(1/1  D0), represents the degree of packing achieved at low pressures as
a result of rearrangement processes before appreciable amounts of interparticle
bonding take place (17). Heckel postulated that K, the slope of the linear region,
is a measure of compact densification by plastic deformation.
Heckel concluded that density–pressure data indicate that the rate of the
change of density with pressure, any pressure, is proportional to the pore fraction
in the compact at that pressure. Additionally, density–pressure curves may be
described by two parameters. He theorized that one was related to low-pressure densification
by interparticle motion. The second measured the ability of the compact to
densify by plastic deformation after appreciable interparticle bonding.
J. R. Johanson identified, through very comprehensive mathematical models
and relationships, material properties, press dimensions, and operating conditions
for roll compactors. For more information, the interested person should read the
entire reference (18). In part, he explained that roller compaction involves a continuous
shear deformation of the granules into a solid mass.
To satisfy the theory’s assumption, it was postulated that the material be isothat
undergoes shear deformation into a solid mass (18). In Figure 2, Po ? horizontal
pressure between rolls, y ? angular position of roll bite, a ? nip angle, 2d ? roll
diameter D, h ? height above the roll centerline at which feed pressure Po is applied,
Pm ? horizontal pressure at y ? 0, and S ? roll gap.
Johanson pointed out that no roller compactor theories at that time determined
the angle of the nip and the bulk density at y ? a, except by actually rolling
the granular solid in a roll press. He also provided a method to calculate the nip
angle and the pressure distribution between the rolls. His calculations determined
the pressure distribution above and in the nip area (18).
He provided the technical rationale to calculate the nip pressures in the nip
region. He showed that material trapped in a volume Va between arc-length segments
4L, must be compressed to volume Vy between the same arc-length segments.
The relationship requires that the bulk densities ga, gy in volumes Va, Vy be related
164 Miller
tropic, frictional, cohesive, and compressible. Figure 2 depicts material in a press
by Eq. 5.3 (Fig. 3) (18):
© 2005 by Taylor & Francis Group, LLC
ga=gy ? VyVa ?5:3?
In Figure 3, P0 ? horizontal pressure between rolls, y ? angular position of roll
bite, a ? nip angle, 2d ? roll diameter D, h ? height above the roll center line at
which feed pressure P0 is applied, Pm ? horizontal pressure at y ? 0, S?roll gap,
DL?arc-length segments, Va?material trapped in volume space described by
arc-lengths, Vy?compressed volume space described by arc-lengths, ga and gy?
respective powder bulk densities in volume spaces Va and Vy, and K?a material
property constant for a given moisture content, temperature and time of compaction.
Johanson stated that the pressure sy at any y < a can be determined as a
Figure 3 Front view of compactor rolls, depicting nip angle. (From Ref. 18.)
Figure 2 Front view of compactor rolls in horizontal plane, depicting powder regions at
different compaction forces. (From Ref. 18.)
Roller Compaction Technology 165
© 2005 by Taylor & Francis Group, LLC
function of the pressure sa, at y?a, by the pressure–density relationship. It was
understood that, for increasing pressures, log density was a linear function of log
pressure (19).
Johanson found that the nip angle does not depend on the magnitude of the
roll force or the roll diameter. He demonstrated that the nip angle was affected very
little by the geometry of the press or the cut grooves on the roll surface. It was mostly
influenced by the nature of the materials that were compressed. The very compressible
materials, with small K-values, had very large nip angles. On the other hand,
incompressible materials, with large K-values, had very small nip angles. Ultimately,
Johanson’s results showed that material properties determine the maximum pressure
that a roller press applies to material.
Parrott evaluated compacting of several pharmaceutical powders into compact
sheets that were sized by an oscillating granulator into granules. He studied flake
thickness, bulk and tap densities, angle of repose, and flow rate after sizing. His work
demonstrated that one could improve most granule densities but not necessarily
improve granule flow properties (20). Nearly all particulate matter can be aggregated
when compressed at high pressure. In some cases, pressure alone cannot achieve suf-
ficient bonding strength; therefore, a binding agent is required to be added to the
powder blend (1). Polymeric binders added to mixtures at certain percentages form
highly viscous bridges between particle-to-particle powders. When the polymeric
mixtures are compressed, they increase the overall compact strength and enhance
particle-to-particle bonding and can improve powder flowability.
Pietsch described a compaction capacity throughput equation for a roller compactor
using flat rolls. The equipment capacity equation is theoretical in nature; it
assumes that all compact is 100% usable for downstream processing needs. Compact
capacity throughput, Cc, can be determined by Eq. 5.4 (4):
Cc ? pDlhAn60g ?5:4?
where Cc is the roller compactor throughput (kg/hr), D is the roller diameter (cm),
l is the roller length, working width (cm), hA is the gap width between the rollers, sheet
thickness (cm), n is the number of revolutions per minute (1/min), and g is the apparent
sheet density (kg/cm3).
6. DESIGN FEATURES OF ROLLER COMPACTORS
Certainly, a key enhancement that highlights today’s pharmaceutical industry stateof-
the-art roller compactors is programmable logic controllers (PLCs). They are
used to control and monitor mechanical parts that regulate screw feed rate, roll
speed, pressure and gap, vacuum deaeration, and mill speed. This section will not
discuss PLC designs.
Briefly, a key machine innovation, vacuum deaeration, was a new important
feature design added by some roller compactor vendors in the early mid-1990s.
The design feature has been shown to help premodify raw material density prior
to compacting and increase throughput (21). Other equipment features such as multiple
horizontal or angled feed screws have assisted in manufacturing a uniform raw
material feed across the rolls (21,22). Newly designed roll machine blocks, featuring
cantilever roll systems, offer more efficient ways to clean, handle, and facilitate
product and equipment changeovers. New storage hoppers and various screw feeder
designs have improved delivery of poor flow powder to the rolls.
166 Miller
© 2005 by Taylor & Francis Group, LLC
A history of hopper and feed screw designs showed that each design evolved to
facilitate and improve powder flow to the compactor feed screw conveyance system.
Feed hopper designs shown in Roller Compaction Technology for the Pharmaceutical
Industry depict designs used in older compactor models (23). Dehont et al. described
the powder compaction feeding systems that were in use up to 1989 (24).
Specific hopper designs were used when material had good rheological characteristics.
Materials that had good but unique flows, and needed to be evenly distributed
across the roll-surface, such as granular salts, used hoppers that had guiding
flaps. When the powder material had poor rheological characteristics, a vertical flap
distribution box was fitted under the hopper. It primarily was used when compacting
ores. Force feeder designs are more commonly used in the pharmaceutical industry.
They are employed with various configured designs and shapes to supply poorly
flowing powder to the nip roll region (23). Pietsch described modern feed screw system
designs; depicting a vertical force-feeding screw with a slightly tapered end, an
inclined feed screw, a vertical, tapered, and blade-angled feed screw, and a horizontal
(single or dual) straight feeder screw(s). These designs are described and pictured
(25). Innovatively designed feed screw systems are used commercially in roller compactors
worldwide. An example is shown in Figure 4.
Sizing devices are now trimly fitted to the compactor body and controlled by
variable speed drives. Most compacters no longer require a separate milling machine
in tandem to size compacts as required by slugging technology. Roller compactors
have clean-in-place systems that offer environmental and safety features. These systems
minimize human exposure to chemicals. See Roller Compaction Technology for
7. ROLL CONFIGURATION
Compactor design features have evolved over the years. By the mid-1970s, research
revealed a number of roll design improvements that increased compacting efficiency.
Three key conditions were identified, at that time, which optimized the roll compact
throughput and minimized leakage of noncompacted powder (26):

 Adequate powder supply must enter the gripping zone,

 Powder must be conveyed fully into the narrowest part of the roller gap,

 Compaction pressure must be distributed as uniformly as possible across
the whole roller-gripped powder mass.
Figure 4 Twin horizontal feed screw system with vacuum deaeration system. (Courtesy of
Alexanderwek Inc., USA.)
Roller Compaction Technology 167
the Pharmaceutical Industry for additional information (23).
© 2005 by Taylor & Francis Group, LLC
Equipment engineers and researchers worked on improving feeding equipment
systems and roll designs to satisfy and maximize the above conditions. Some of the
key advances are identified in Ref. 21 and are re-emphasized in this chapter.
Because of powder feed variability at the nip and in the roll gap regions, powder
leakage is produced during the compaction process. This situation produces
excessive fines and possible undesirable processed material. Usually, this problem
is caused by uneven powder flow and compact formed when the powder is fed
toward the middle of the roll width. Granules produced under these conditions
are typically not optimal for further pharmaceutical processing.
Trials performed by Funakoshi et al. using rectangular aperture chutes, fitted
at the end of a feed screw, aided in preventing uneven powder flow across the roll
width. Funakoshi’s team demonstrated the positive effects of having a concavoconvex
roller pair fitted with inner ring walls. This feature counteracted side seal
effects (fractured or incomplete compacted edges). They also designed and tested
other rolls (concavo-convex with rim), which allowed the powder to distribute more
uniformly across the roll width. This design reportedly minimized powder leakage
during compaction (26). Funakoshi’s team also showed that when the roll rim inner
angle is zero the powder is not adequately and uniformly delivered to the gripping
and compacting zones. This occurred because the stationary side seals acted as resistors
to the powder flow (26). Their work demonstrated that the formed compact and
the compaction pressure (using standard rolls) produced an uneven compact across
the roller ends because of side seal friction. When they employed the newly designed
concavo-convex rimmed rolls, it protected the compact from the adverse side seal
fracturing effect. Funakoshi et al. also determined a proper selection of rimmed rollers,
which delivered adequate powder to the compaction zone and also conveyed
powder fully across the roller gap region. They concluded that the height and slope
of the inner walls’ rims optimally influenced the side seal effect. A 65 inner wall
slope produced 2.5–3.0% fines’ leakage during the compaction of lactose. When
the rolls had no inner wall rims, the fines’ leakage was 15%. In summary, the team
optimized the compact and pressure distribution along the rolls by using the concavo-
convex rimmed shaped rolls and found the best roll design was still dependent
on raw material properties (26).
Parrott further substantiated the usefulness of the rimmed concavo-convex
roller pair to increase the density of several pharmaceutical powders. His work
resulted in optimizing a process with an uncompacted leakage rate of 5%. The optimization
depended on the physical properties of the powder and the machine operating
conditions such as the roller gap, feed screw speed, and roll speed (27). Some of
their key design feature findings are summarized:

 Installing a concavo-convex roll surface rather than a flat roller pair

 Installing rectangular feed chute and flaps

 Designing cylindrical, conical/cylindrical, or tapered variable speed auger
feed screws

 Optimizing the roll rim angle to 65

 Installing digital and analog variable feed screw controllers

 Developing horizontal and variable screw feed systems.
Jerome et al. in the 1980s studied the effects of compactor adjustments on powder
blend properties. His team’s research showed that pressure applied by the movable
roller is not a predominant factor. They found that the most important variable was
the compactor feed screw in relation to the roll speed. More will be said about this
168 Miller
© 2005 by Taylor & Francis Group, LLC
point later in the chapter. Jerome correlated improved tablet compression hardness to
the feed screw speed in relation to the speed of the rolls (28). Roll configurations are
well documented (23). Most compactors now have one floating roll and one fixed roll
each on separate bearing blocks and this is illustrated in Figure 5. Currently, one pharmaceutical
manufacturer of roller compactor equipment has a fixed roll-pair system.
This roll design system relies on powder gravity feed and a specially designed compression
feed screw to continuously deliver powder to the fixed roll pair (23).
Pietsch described two pairs of rollers with different diameters D1 and D2, with
identical roll gaps hA. He theorized that if the peripheral speed of both pairs of rollers
is the same, roller compaction takes place more gradually in the case of the larger
roll pair. At the same time, the larger-diameter roll pair pulls a larger powder volume
into the nip region, resulting in a higher-density compacted product. The largerdiameter
roll pair also minimizes air entrainment more efficiently than the smallerdiameter
roll pair when both are operating at the same peripheral speed (25). Pietsch
noted that the peripheral roll speed and particulate powder speed are not equivalent
in the entire compaction zone. Throughput does not increase proportionally with
roll speed.
There are two effects that hinder throughput: starved conditions in the feed
zone and too much squeezed air from the particle mass that flows upward and
against the powder flow, reducing the supply of material to the nip area (25).
What is the optimum roll speed for a compaction process?What factors does the
formulating scientist or process engineer need to consider to maximize compact quality
and compaction throughput? Johanson (29) attempted to answer these questions
by predicting roll-limiting speeds for briqueting presses. He developed mathematical
expressions considering even the gas and liquid effects as they can theoretically be
Figure 5 Fixed and floating roll pair. (From Ref. 13.)
Roller Compaction Technology 169
© 2005 by Taylor & Francis Group, LLC
squeezed from a solid mass. Solid properties, press dimensions, and operating conditions
were evaluated to predict optimumroll speeds. The results necessary for a quality
briquette are most critical for low-density fine particles. Johanson’s work showed the
relation between feed pressure and roll speed to essentially be proportional to the
material’s permeability (porosity). For example, when the initial powder was compacted,
sized, and recompacted, the bulk powder permeability increased; however,
when the powder was compressed, the compression force decreased, and compactor
feed pressure force requirement decreased significantly. Johanson (29) also demonstrated
that if the compactor feed pressure is kept constant and the press speed is
increased, the maximum pressure applied to the briquette decreases as the roll speed
increases. As a result, the briquette density and strength decrease likewise. For
additional information, the interested reader should peruse the reference list.
Sheskey and Hendren in 1999 studied the effect of roll surface configuration on
the drug release and physical properties of a hydroxylpropylmethyl cellulose
(HPMC) matrix controlled-release dosage form (30). Smooth and axial-grooved
roll surface designs were studied using a Vector Model TF-Mini roller compactor
(Vector Corporation, Marion, IA). Their hypothesis was that the greater the depth
of the concavity on the roll surface, the pressure exerted on the powder being compacted
would not be as evenly displaced as with a smooth roll surface. Therefore, as
with tablet tooling design, the top of the crown area of the compacted ribbon would
theoretically be softer than the rest of the ribbon. However, the results of a particle
size distribution test performed on milled ribbons generated using both smooth and
axial-grooved roll pairs showed similarity between tested samples. In addition,
results showed little difference in tablet crushing strength values between samples
manufactured using either type (smooth, axial-grooved) of roll surface design. Drug
release profiles of tablets prepared from roller compacted granulations using both
roll surface configurations were also similar. It was noted that the authors did not
measure product throughput rates to address the possibility of improved efficiency
when using an axial-grooved roll pair vs. a smooth roll pair.
8. FEED SCREW DESIGN
The consistency and evenness of the powder feed into a roll pair determines to
a large extent, how complete a compact is made and ultimately the success of a compaction
process. Most roller compacting systems suffer the disadvantage of leakage,
i.e., 20–30% powder particles (depending on the formulation) are not compacted.
This primarily occurs because of uneven powder feed and powder slippage between
individual loose particles and the roll surfaces. Under these conditions, it is usually
necessary to recycle the uncompacted powder or fines. Recycling a compaction process
is a significant drawback because of additional capital expenditure, labor costs,
and increased throughput time.
The chief objective of roller compaction is to consistently make an agglomerate
of sufficient strength that meets the required density, granulometry, and powder flow
specifications. The key operational goal in compacting is to maintain a pressure
range on the feedstock, independent of the fluctuating powder granulometry and
flow into the rolls, to maintain a consistent compact. The compaction process is
managed by controlling the input material, the quantity per unit of time, the roll
speed, and the roll gap. Allowing the roll gap to float unchecked could influence
the production rate and the compact quality. Therefore, it is important to control
170 Miller
© 2005 by Taylor & Francis Group, LLC
the compaction process by setting a constant powder feed rate during the compaction
operation.
Design innovations on the powder feed input side of the roller compactor are
complex. The complexity is double-edged: powder materials, which flow and move
well, are more easily handled on the feed side of the compactor. These types of materials
generally do not necessarily need significant densification to marginally improve
their flow handling characteristics on a large scale. On the other hand, poor flowing
and low-density powders require special equipment and considerations to feed a
compactor. Additionally, it is necessary to maintain a constant powder flow and
quantity of material to the rolls during the compaction process. Therefore, the delivery
feed system plays a very important role in delivering poor flowing low-density
bulk powder materials to the roll compactor.
Johanson, in 1995, described an arch-breaking hopper design that effectively
delivered poor flowing powder from the hopper to a horizontal feed screw. The
unique hopper design eliminated the ‘‘rat-holing’’ effect of poor flowing powders
during hopper voiding and is described in Figure 6 (31).
Miller noted that the total compaction power requirement is the sum of the
power required for the feed screw(s) and the roll drive (13). Feed screws not only
convey the powder material from the compactor storage hopper but they also help
deaerate the powder in the process. The deaeration of the powder acts as a minicompactor
by precompacting the material just prior to roll compaction. Optimum compaction
pressures and feed screw designs vary widely for different powder material
properties. Changes in bulk powder density and feed screw speeds will affect the roll
gap, the compaction pressure, the throughput, and the quality of the compact (13).
Weggel indicated that feed screw torque varies directly with the precompaction
pressure (32). He suggested that by maintaining a constant feed screw pressure, the
compactor operator can control the compact quality. Variations in precompaction
pressure and in the compacting pressures are directly related to feed screw amperages
and the roll drive motor. The compaction pressures are dependent on a continuous
flow of powder into the feed screw area. If the powder feed flow is intermittent or
Figure 6 Arch-breaking hopper design. (From Ref. 31.)
Roller Compaction Technology 171
© 2005 by Taylor & Francis Group, LLC
interrupted, this will affect the feed screw amperage readings and eventually the roll
drive readings. Weggel noted that forcefeeding the material deaerated the feedstock
and reduced the roll pressure loads. He found that it was more efficient to achieve
maximum densification during several stages of the roller compaction process (32).
Miller noted, in 1996, that the design of vertical cylindrical feed screws appears
to feed powder uniformly to the circumference of the feed screw (13). However, he
concluded that the powder feeding to the rolls did not coincide with the general rectangular
shape of the compactor throat. The observed drawback was that the powder
did not get delivered evenly across the rolls; the middle of the rolls received more
powder than the rolls’ edge. This potentially created a poor quality compact;
a strong middle compact that is weak at the ends because of frayed edges. Frayed
compact edges give rise to uncompacted material and excess fines (13). An example
of the phenomena was depicted when using a light yellow-green color feedstock,
which after roller compaction using a twin feed screw system produced a homogenous
colored compact. The same feedstock passed through a compactor designed with
a single vertical feed screw, exhibited a sinusoidal colored compact (Fig. 7). Miller indicated
that a multiple horizontal feed screw design system provided a more uniform
powder distribution across the rolls in area and in volume than a single vertical feed
screw system. This applied to either vertical or horizontal single feed screw systems
(13).
9. FUTURE TRENDS IN GRANULATION TECHNOLOGY
A number of these attributes, best technology practices, and features were rated for
their industrial and pharmaceutical importance and reported by Miller and Sheskey
in 2001 (33). Survey findings evaluated future industry usage of roller compaction
Worldwide pharmaceutical manufacturing technical operations predict signifi-
cantly increased roller compaction usage in the future. Research and development
scientists indicated worldwide that they are currently involved with roller compac-
Figure 7 Roller compaction: compacts. (Courtesy of R.W. Miller.)
172 Miller
technology in technical operations and in research and development (Fig. 8).
© 2005 by Taylor & Francis Group, LLC
tion technology to develop new formulations. Some key reasons why this is the case
is that most API discovered now are very sensitive to water and heat and have low
bulk densities. The authors concluded that these combined factors push roller
compaction technology to the forefront as the preferred processing technology for
new pharmaceutical chemicals for solid-dose formulations in the pharmaceutical
industry (33). It also appears that pharmaceutical companies are chiefly content to
have the right technology in place to deliver the granulation form needed in the
easiest, quickest, and most efficacious manner.
A significant survey response indicated the importance of both design and service
features to advance roller compaction technology (33). There seems to be an
assorted blend of needs that are due in part to vendor past performances, such as
reliability, technical support programs, and service history. There is also a component
of innovation, new technology advances, such as cantilevered rolls and operator
ceutical Review (33)].
10. NEW FINDINGS
P. Guigon and O. Simon, Compiegne Universite de Technologie, Compiegne Cedex,
France, recently presented experimental results using a laboratory Komarek

out vacuum deaeration system), to feed a roll press, the stress exerted on the compact
was neither homogenous on the roll width nor constant with time.
The heterogeneity was characterized by measuring light transmission through
a sodium chloride compact. The team showed that the powder packing that took
place in the last flight of the feed screw caused heterogenous feeding pressure. The
heterogenous feeding pressure caused the heterogenous compact properties (34).
The team developed a mathematical model to link feed screw geometry to the stress
distribution in the compact. The author notes that this finding is contrary to
Figure 8 (A) Roller compaction current usage in company technical operations. (From
Ref. 33.) (B) Roller compaction current usage in R&D projects. (From Ref. 33.)
Roller Compaction Technology 173
B100QC press (Fig. 9). They demonstrated that when using a single feed screw (withtouch
screens with parametric feedback [Table 3, courtesy of the American Pharma-
© 2005 by Taylor & Francis Group, LLC
the Johanson model noted earlier in this chapter and quite revealing. The laboratory
press was specially instrumented with two flush mounted piezoelectric transducers
fitted on the smooth upper roll (130mm diameter by 50mm width). The transducers
measured stress (0–200 MPa) exerted on the roll surface at 15mm from both sides of
the roll. The position of the piezoelectric transducers was defined once per revolution;
the roll speed, screw feeder speed, and hydraulic pressure were also measured
The team measured the compact density by the distribution of light transmitted
through the sodium chloride compact. Guigon indicated that the milled sodium
chloride crystals were oriented by the applied stress and therefore the incident light
was not diffused similarly in all directions (34). The stress applied on the compact is
Table 3 Preferences for Compactor Design and Service Features
Compactor design and
service features Number of responses Average rating
Machine performance and reliability 56 4.5
Vendor spare parts availability 55 4.2
Variety of model sizes 61 4.1
Vendor IQ and OQ program 51 3.9
Vendor service history 48 3.9
Cantilevered vs. double-bearing rolls 49 3.8
Vendor delivery time 54 3.8
Ease of cleaning 54 3.7
Variety of roll patterns 51 3.7
Vendor training support 54 3.5
Touch screens with parameter feedback 60 3.2
Other 11 3.2
Note: 5 ? high, 1 ? low preferences.
Source: From Ref. 33.
Figure 9 Side and front view of Komarek B100QC press used to determine geometry feed
system and stress distribution and applied on compact. (From Ref. 34.)
174 Miller
and recorded (Fig. 10).
© 2005 by Taylor & Francis Group, LLC
neither homogenous on the roll width nor constant with time. Guigon points out
that the stress maximum for the same sensor varies from one rotation to the next
and comparing the left and right sensors, one has a low signal while the other has
a high signal during each turn. He noted that the maximum stress measured for each
roll revolution was a cycle of the time equal to the time period of the feed screw
stress, illustrating that the maximum stresses are not applied to the center of the roll
but more toward the roll edges. The scientists showed that the applied stress on the
compact in the roller gap is due to the feed pressure distribution from the feed screw.
The nature of the feed pressure was due to the pushing of the powder in the last
spiral of the feed screw. The team took many stress profile readings to measure
the heterogeneity of the applied stress, as it was not constant or homogenous along
the roller width (Fig. 12).
Graphically, the light transmitted through the salt compact and the maximum
measured stresses (from the two locations on the roll) enabled the stress measurement
throughout the compact in the gap. The compact zones that had endured less
stress transmitted less light and therefore appeared darker (34).
indicated that the feed screw exit pressure was a function of the geometry and surface
Figure 10 (A) Applied stress on the compact during one rotation period of feed screw. (B)
Corresponding gray level on the numerical image of light transmitted through the compact.
Sodium chloride d50 ? 74 mm, Vr ? 10 rpm, Vs ? 25.3 rpm. (From Ref. 34.)
Roller Compaction Technology 175
(Fig. 11).
Figure 12 shows a good representation of the heterogeneity of the applied
Guigon sketched the feeding zone just prior to roll compaction (Fig. 13). He
© 2005 by Taylor & Francis Group, LLC
properties of the feed screw and of the compressibility of the powder. Summarizing
his findings, the pressure at a location (y, z planes, perpendicular to the x-direction of
motion) of the feeding plane will vary due to the motion of the last flight of the
screw. When the wall of the screw flight is far from the feeding plane (top part of
and the localized pressure at the feeding plane is low. When the flight at this location
moves forward, the powder first compacts without moving significantly, increasing
progressively its density and local pressure at the feeding plane.
The team proved the observation with two dynamic motion picture experiments.
The first study showed the motion of particles at the check plate wall and
confirmed by video analysis that the particles do not move continuously in the feed
zone (34). Guigon noted that the powder particles are almost stationary. Then, the
powder moves forward before stopping again. The velocity fluctuations have
the same period as the screw. The second experiment consisted of using sodium
chloride and pushing the material to the rolls by means of a piston-feeding device.
the densest part of the compact is through the center, where most of the light is
transmitted.
The work by Guigon and Simon provided new meaning to the importance of
the feed screw pushing powder into the rolls and the feed screw design’s influence
on compact formation. The author of this chapter raises some questions on how a
multiple feed screw system under vacuum deaeration environment would affect compact
formation. No information has been published on investigating these aspects.
Figure 11 Stress profiles recorded simultaneously on the (A) left and (B) right transducer vs.
number of roller rotations during consecutive turns. (Screw speed, Vs; roller speed, Vr; sodium
chloride (Vr ? 8.4 rpm, Vs ? 22.3 rpm). (From Ref. 34.)
176 Miller
Fig. 13), the powder between the flight and the feeding plane is not compacted
The compact is shown in Figure 14 with light passing through it. Most notably,
© 2005 by Taylor & Francis Group, LLC
Figure 12 Distribution of stresses applied to a compact during the dynamic state of roller
compaction. (From Ref. 34.)
Figure 13 Sketch of the feeding zone. (From Ref. 34.)
Roller Compaction Technology 177
© 2005 by Taylor & Francis Group, LLC
11. DEAERATION THEORY
A key factor limiting compaction throughput and quality is air entrapment in powder
materials. During compression, air-occupying voids between particles are compressed
and squeezed. The gas pushes through the powder causing powder
fluidization and a nonuniform level of powder at the roll gap. It is best described
in Figure 15. This situation limits compact throughput and creates a nonuniform
compact density. Air entrapment creates excess fines prior to sizing because of ‘‘spidering’’
compact edges.
The spidering condition occurs when gases rush across the inside of a compact
to thinly and weakly formed flaked edges. The flake edges break apart perpendicular
to the compaction direction. The compact edge breakage appears ‘‘saw tooth’’ in
Figure 14 Distribution of stresses applied to a compact, by piston pushed material into rolls,
not feed-screw pushed. (From Ref. 34.)
Figure 15 Pattern of gas escape from nip region. (From Ref. 36.)
178 Miller
© 2005 by Taylor & Francis Group, LLC
structure and varies in length depending on the nature of the powder binding properties,
the amount of air entrainment, and the roll dwell time.
Johanson predicted theoretical compactor operating conditions to handle air
entrapment effects in materials. In general, he concluded that the critical compacting
pressure level is dependent on a number of factors: roll diameter and speed, powder
permeability, compressibility, and compact strength. Johanson indicated, when
applying these principles in commercial application, that a compactor operator
would have to operate the press at slightly less than maximum pressures to allow
for material inconsistencies and variable in-feed flow rates (35).
Johanson and Pietsch described the forces at work in a roll press when powders
are compacted (25,35,36). Johanson illustrated the three typical work phases during
compaction Figure 16. The first region is the initial solids contact pressure with the
feed screw. Pressure at this point is nil, P0. Later, as the solids begin to get pushed
and gripped by the rolls, pressure continues to build. Ultimately, the solids are
moved into the narrowest point of the nip angle where the maximum pressure occurs
at pmax. The second region describes the solid’s density increase through compaction
and densification occurs rapidly, in less than a second. Some materials have properties
that after discharge from the compactor undergo elastic deformation, which can
reduce the compact’s density, expressed by the dotted curve. The third region of
force at work is air pressure. Models such as the one described take into account that
it is dependent on the porosity of the particulate solids compacted. If the porosity of
the compacted material remains high enough during the compaction process, air
pressure can escape and vent during and after compaction. On the other hand, if
the material has low porosity, air pressure builds up to high levels because it cannot
escape easily. This can be seen in several ways, one of which is through expansion of
the compact bursting. Bursting can be associated with a popping sound when operating
the compactor. The second way occurs when the compacted sheet breaks into
slivers or ‘‘spidering’’ at the compact edges.
Both Johanson and Pietsch reported that expanding gas in a compact is detrimental
to the compaction process: by reducing the compaction throughput and
increasing the amount of fine particles. Johanson illustrates the effects of roll speed
and powder porosity on air pressure in a compacted sheet (43). He shows a relative
large roll speed operating range when compacting a permeable (porous) powder. Air
entrainment does not limit roller speed for coarse granular powders. On the other
Figure 16 Compaction work phases. (From Ref. 43.)
Roller Compaction Technology 179
© 2005 by Taylor & Francis Group, LLC
hand, when compacting very fine powders, the operating roll speed range is signifi-
cantly reduced because of air entrainment.
Miller indicated that the evenness of the powder feed into the rolls determines,
to a large extent, the success of compaction. Roller compactor systems suffer from
two disadvantages: as the powder feed bulk density approaches 0.3 g/cm3 or less,
the compaction throughput efficiency decreases. Secondly and concurrently, the
uncompacted powder leakage generally increases around the rolls (21). Miller, in
1994, described a new machine design improvement that used vacuum deaeration
to remove air entrainment from the powder just prior to the nip angle during roller
compaction. The multiple benefits of such action are significant when compacting
low-density raw materials (21):

 More uniform powder feed to the rollers

 Less voltage and amperage variability for the roll pair

 More uniform and strong compact

 Less powder leakage

 Greater yield

 Less powder adhering to the compact prior to sizing

 Higher compact throughput

 Less airborne particles.
The newly designed equipment involved a compactor fitted with two horizontal
feed screws, which featured vacuum deaeration. Specifically, the roller compactor
was equipped with a conical storage hopper containing a variable speed agitator.
Bulk powder was fed directly from the top of the hopper to the top of twin horizontal
auger feed screws, which directly transported the powder to the nip roll area (13).
Top and side views of the design features are shown in Figure 17 .
A novel stainless steel encasing that leads to the compactor rolls encloses the
variable-speed auger screws. Just before the nip area, a pair of sintered stainless steel
segments are assembled within the horizontal auger feed system, which can operate
under partial vacuum. A small, self-contained vacuum pump draws negative pressure
through a dry filter and a stainless steel line connected to the sintered assembly
plates. The partial vacuum is adjustable from 0.1 to 0.8 bars. The compaction rolls
Figure 17 Compactor side view of auger feed screw system and sintered plate segment.
(From Ref. 13.)
180 Miller
© 2005 by Taylor & Francis Group, LLC
operate at different speeds and are supported on heavy-duty bearings in such a way
that the lower roll is fixed and the upper roll is slightly movable in the vertical plane.
The deaerated material passes through the roll pair, which is under infinitely variable
hydraulic pressure. The deaeration, auger feed screws’ design and speed, roll speed,
and hydraulic roll pressures are the main factors in producing a compact with
specified properties.
In several experiments,Miller studied the effects of using horizontal twin auger feed
screws under partial vacuum (Alexanderwerk Inc. Horsham, PA, USA, model 50/75
compactor). The feed powder was vacuum deaerated just before roller compaction.
The experimental design showed that the compactor’s deaeration feed system signifi-
cantly increased compaction output and minimized powder leakage when compacting
very low-density blends (<0.35 g/cm3); remarkable results were observed (22).
To evaluate the effectiveness of the compactor’s deaeration system, a test was
designed to process a low-density active drug blend with and without the activated
deaeration system. The test showed how much material was compacted with the
deaeration system engaged and how much material was compacted when it was
not engaged. The test also determined how much material was not compacted (powder
roll leakage) in each case. Details of the experimental plan are found in Ref. 38.
Results of this experiment showed that under the influence of vacuum deaeration,
the compactor produced throughputs of 100 kg/hr and the resultant noncompacted
material leakage rate was <2%. The vacuum deaeration was so effective that there
was no need to recirculate the noncompacted powder back to the rolls to meet processing
specifications. Under a second set of compacting conditions, vacuum deaeration
was not activated and the process typically produced 70–80 kg/hr of densified
compact. The powder leakage rate increased to 20–30%. During this set of conditions,
the powder flow to the roll pair was uneven and the ribbon compact was
not uniform. Processing this formulation for a long period of time under these conditions
would have required recirculation equipment to return the uncompacted
meters (trials 1 and 2).
In several other trials, Miller also studied the effects of using two different comviously
described, were carried out using differently designed compactors. The
objective of these trials was to increase the active bulk density of the feedstock from
0.22 to 0.61 g/cm3 (13). Miller reported that the compactor, machine 1, containing
3 density at 80 kg/hr yield. When the
vacuum deaeration was not activated the desired granule density specification could
drug granules during the first compaction run when employing vacuum deaeration.
After a second compaction pass with vacuum deaeration, trial 2, the powder density
was increased slightly to 0.58 g/cm3. It appeared that the level of vacuum being
applied to the active drug substance, in machine 2, was insufficient, ineffective, or
Miller’s experiments evaluated the effects of powder density; screw feed speed,
roll speed, roll pressure, vacuum deaeration pressure, compaction rate, and the compaction
leakage rate. Test results demonstrated that the first compactor’s deaeration
and feed system designs significantly increased compaction output. The new equip-
Roller Compaction Technology 181
the horizontal twin screws and the vacuum deaeration design as described in Figure.
17, produced granules with 0.63–0.64 g/cm
A second compactor, machine design 2 (Fig. 18), did not densify the active
powder to the rolls for further processing (21,37,38). Table 4 for compactor parapactor’s
vacuum deaeration systems (Table 5). The tests, similar to the ones prenot
be achieved. Table 6 (trials 1 and 2) for operating parameters and results.
not optimally positioned, even at its maximum vacuum deaeration level (13). Table
7 (trials 1–3) for machine 2 operating parameters and results.
© 2005 by Taylor & Francis Group, LLC
ment design and process provided high compact yields and virtually eliminated powder
leakage, obviating the need for expensive powder recirculation equipment. The
of approximately 0.2 g/cm3. In summary, a new critical condition, vacuum deaeration,
had been identified in optimizing roller compacting effectiveness and efficiency
(13). Miller concluded that four key processing conditions must exist to optimize
roller compaction throughput and minimize powder leakage around the rolls (13):

 Adequate powder supply must enter the gripping zone.

 Powder must be fully conveyed into the narrowest part of the roller gap.

 Compaction pressure must be distributed as uniformly as possible over the
whole of the roller-gripped powder mass.

 Sufficient vacuum deaeration must be effectively distributed prior to the nip
roll region, particularly for low bulk density powder feed stock.
12. ROLLER COMPACTION AND NEAR-INFRARED SPECTROSCOPY
Work by this author in the mid-1990s investigated roller compaction and near-infrared
spectroscopy (NIRS) technology, by examining three different active ingredient
Table 4 Compactor Operating Conditions and Yields for Poor
Following Low-Density Active Drug Blend, Formulation 1, Vacuum
Deaeration and Non Vacuum Deaeration Trials
Formulation 1
Conditions Trial 1 Trial 2
Powder density (g/cm3) 0.25–0.35 0.25–0.35
Screw feed (rpm) 52 52
Roll speed (rpm) 8 8
Vacuum (bar) (0.78–0.78) 0
Roll pressure (bar) 60–65 60–65
Compact rate (kg/hr) 100 70–80
Compact leakage rate (kg/hr) 2 15–20
Source: From Ref. 13.
Table 5 Compactor Operating Conditions and Yields for Poor
Flowing Low-Density Active Drug Blend, Formulation 2, Vacuum
Deaeration and Non-vacuum Deaeration Trials
Formulation 2
Conditions Trial 1 Trial 2
Powder density (g/cm3) 0.25–0.35 0.25–0.35
Screw feed (rpm) 52 52
Roll speed (rpm) 8 8
Vacuum (bar) (0.78–0.78) 0
Roll pressure (bar) 60–65 60–65
Compact rate (kg/hr) 150 100–110
Compact leakage rate (kg/hr) 1.3 20–30
Source: From Ref. 13.
182 Miller
first compactor’s vacuum deaeration design (Fig. 17) proved to be superior to that
of the second machine (Fig. 18) when compacting an active bulk drug with a density
© 2005 by Taylor & Francis Group, LLC
blends subjected to different compaction conditions and mapped specific in-process
roller compaction processing steps and relationships. Miller illustrated the use of
NIRS to map, in the static mode, roller compaction unit operations. This was the
first published work about using any spectroscopy to map roller compaction processing
steps (39).
Blends were compacted using an Alexanderwerk roller compactor, model
WP50/75N (Alexanderwerk Inc., Horsham, PA) fitted with vacuum deaeration.
All compacts were sized through a continuous series double rotary granulator fitted
with 3.15 and 1.25mm mesh screens, respectively. Blends were finally mixed for
2 min and compressed into tablets using a Kilian E 150 press (Kilian & Co Inc.,
Horsham, PA), which was equipped with standard concave 11mm diameter tooling.
NIR spectral analysis was conducted using a Rapid Content Analyzer, model
6500 NIR spectrophotometer (Foss NIRSystems, Silver Spring, MD). NIRS
libraries were developed for each active drug concentration and the specific
Table 6 Operating Parameters and Results of CompactorMachine
Design 1When Compacting Poor Flowing Low-Density Active Drug
Substance Blends
Conditions Trial 1 Trial 2
Initial density (g/cm3) 0.22 0.22
Powder density (g/cm3) 0.63–64 0.45
Screw feed (rpm) 52 52
Roll speed (rpm) 8 8
Vacuum (bar) 0.65 0
Roll pressure (bar) 50 50
Compact rate (kg/hr) 80 42
Source: From Ref. 13.
Figure 18 Compactor front view of vertical feed screw system with vacuum deaeration.
(From Ref. 13.)
Roller Compaction Technology 183
© 2005 by Taylor & Francis Group, LLC
compactor’s process critical control point’s (PCCP’s) operating conditions, details of
the experimental plan are referenced (39).
The NIRS absorbance shows (optical density) differences between 10 and
20 bar roll pressure compacts and also differences between compacts manufactured
with and without vacuum deaeration (Fig. 19). The associated NIR ibuprofen/
starch compact spectra also show an absorbance (optical density) hierarchy. The
higher the roll pressure with vacuum deaeration the greater the NIR optical density
(10 bar, no vacuum<10 bar, vacuum<20 bar, no vacuum<20 bar, vacuum).
tion. Figure 20 compares the two levels of roll pressure with and without vacuum
deaeration to a third parameter, compaction roll power consumption (CRPC).
CRPC is a power force measured from the roll drive that is continuously monitored
throughout the compaction operating process. An ibuprofen/starch compact trend
analysis shows that the greater the roll pressure with vacuum deaeration the greater
the CRPC (10 bar, no vacuum<10 bar, vacuum<20 bar, no vacuum<20 bar,
vacuum). The analysis demonstrates that employing vacuum deaeration increased
the CRPC at both low and high roll pressure levels. As the vacuum deaeration
increased, the width and gauge of the ibuprofen/starch powder compacts also
increased. Additional characterizations for aspirin/starch and acetaminophen/
starch compacts were completed, Ref. 39.
Table 7 Operating Parameters and Results of Compactor Machine Design 2 When
Compacting a Poor Flowing Low–Density Active Drug
Conditions Trial 1 Trial 2 Trial 3
Initial density (g/cm3) 0.22 0.45 0.22
Powder density (g/cm3) 0.45 0.58 0.46
Screw feed (horizontal) (rpm) 30 30 30
Screw feed (vertical) (rpm) 260 260 260
Roll speed (rpm) 6 6 6
Vacuum (bar) 0.34 0.34 0
Roll pressure (bar) 50 50 50
Compact rate (kg/hr) 40 48 36
Source: From Ref. 13.
Figure 19 NIR absorbance processing 50/50 Ibuprofen/Starch compacts. (From Ref. 39.)
184 Miller
Figure 20 further illustrates the NIRS ibuprofen/starch 50/50-compaction evalua-
© 2005 by Taylor & Francis Group, LLC
Figure 21 depicts granule NIRS sensitivity to vacuum deaeration and roll pressure
sized compacts. In general, granules produced from sized compacts manufactured
without vacuum deaeration and at lower roll pressure levels exhibited higher
made from the same corresponding conditions. The results demonstrated that successive
ibuprofen/starch processing steps, compacting, and sizing produced uniquely
different NIR absorbance amplitudes.
13. ROLLER COMPACTION AND PAT
A new FDA guideline initiative (2003–2004), Process Analytical Technologies (PAT)
encourages manufacturers to use real time in-line, on-line, or at-line nondestructive
sensory technologies to monitor process critical control points. The PAT framework
addresses every aspect from incoming raw materials to optimization to continuous
improvement. It starts with processability of the incoming raw materials; their attributes
would be used to adjust the process parameters. The incoming material attributes
could be used to predict or adjust optimal processing parameters. The link
between PAT and the new 21st Century Good Manufacturing Practices initiative,
from the FDA’s perspective, is ‘‘quality depends on knowledge and PAT brings
Figure 20 50/50 Ibuprofen/Starch compacts CRPC compactor and NIR corresponding
values. (From Ref. 39.)
Figure 21 90/10 Aspirin/Starch granules NIR absorbance with compactor settings. (From
Ref. 39.)
Roller Compaction Technology 185
optical density. NIR spectra of 90/10 ibuprofen-processing effects are shown in Figure
22: final blend, compact 20 bar roll pressure (vacuum deaeration) and granules
© 2005 by Taylor & Francis Group, LLC
more knowledge and understanding of processes. This is the point where science and
risk-based decisions can be made in terms of manufacturing per Dr. A. Hussain’’
(40).
In 2002–2003 Gupta, Morris, and Peck from Purdue University Industrial
Physical Pharmacy School began a series of noninvasive real-time investigations.
Evaluating roller compaction by NIR in the dynamic mode, the team monitored
the variation of compacted ribbon strength that could adversely affect particle size
and distribution of granules obtained after milling a compact (41). Their thinking
was that both the ribbon strength and the particle-size distribution can be determined
off-line but it is time consuming. The ability to use NIR spectra to quantify
the compact strength, the particle size, and to estimate postmilling particle-size
distribution could be done in real time. The most obvious NIR spectral change that
occurs in compacts or tablets prepared under increasing pressure is an upward shift
in the NIR spectral baseline (39). Kirsch and Drennen used the slope of the best-fit
line through spectra to quantify this upward shift and found a linear correlation
between this slope and the tablet hardness determined by the diametral compression
test (42). Gupta et al. applied this treatment to correlate the slope of the best-fit line
through the NIR spectra with the strength of compacts was determined using the
three-point beam bending technique (41). Additionally, slope values were correlated
Figure 22 NIR absorbance 90/10 Ibuprofen/Starch processing effects. (From Ref. 39.)
Figure 23 MCC compacts: effect of roll speed. (From Ref. 41.)
186 Miller
© 2005 by Taylor & Francis Group, LLC
to the particle-size distribution of granules produced from milled compacts. Realtime
on-line techniques were used to monitor the roller compaction unit operation
for microcrystalline cellulose and 10% active blend. The spectra collected on the
different compact segments prepared from the same material under the same roller
compactor settings showed little variation, suggesting good reproducibility in the
data collection by the NIR sensor. A significant shift in the NIR spectra was
observed for compacts prepared at different roll speeds.
At constant feed screw speeds, the spectra shifted toward lower absorbance
prepared from both the placebo and the 10% tolmetin blends. The shift in the
NIR spectra may be quantified, using the slope of the spectral baseline or using
the slope of the best-fit line, by regression analysis through the spectrum (41). Use
of the slope of the best-fit line was advantageous over the slope of the baseline since
the best-fit line uses information from the entire spectrum. The normalized force
values for the compacts showed strong linear correlation with the slopes of the spectral
best-fit lines (Fig. 24). It was observed that the particle-size distribution of the
milled compacts exhibits a steady decrease in the d90, d50, and d10 values with increasing
roller-compactor roll speed at constant rate (Fig. 25).
Figure 24 Relationship between slope of NIR spectrum and compact strength. (From
Ref. 41.)
Figure 25 Relationship of milled 100% compacts at various roll speeds. (From Ref. 41.)
Roller Compaction Technology 187
values with increasing roll speed (Fig. 23). This shift was observed with compacts
© 2005 by Taylor & Francis Group, LLC
The values for the particle-size distribution for the milled 10% tolmetin compacts
correlated well with the slopes of the spectral best-fit line (Fig. 26). This clearly
shows the potential of the technique in predicting the particle-size distribution of
milled compacts under a given set of milling conditions. Multivariate analysis of
the spectral data using PLS regression analysis confirmed the ability of the slope
of the spectral best fit to estimate the compact strength and the postmilling particle-
size distribution. For greater detail in this expanding new technology area the
interested reader should examine the references.
ACKNOWLEDGMENTS
The author gratefully acknowledges and thanks Carol G. Miller for her contributions
and assistance in editing and data manuscript control. The author thanks
and appreciates Dr. Pierre Guigon, Professor and Director of Chemistry at the University
of Technology of Compiegne, France, for his collaborative contributions to
the text. The author thanks and appreciates Mr. Abhay Gupta, Ph. D. former graduate
student at the Industrial Physical Pharmacy School of Purdue University, West
Lafayette, IN, USA. The author acknowledges and thanks Mr. Leonard Minervini,
Alexanderwerk, Inc., Hatboro, PA, USA, for electronic files of photographs and
pictures.
REFERENCES
1. Kristensen HG, Schaefer T. Granulations. Swarbrick J, Boyland J, eds. Encyclopedia of
Pharmaceutical Technology. Vol. 7. New York: Marcel Dekker, 1993:121–126.
2. Capes CE. Particle size enlargement. Williams J, Allen T, eds. Handbook of Powder
Technology. Vol. 1. Amsterdam: Elsevier, 1980.
3. Capes CE. Particle size enlargement. Grayson M, Eckroth D, eds. Kirk-Othmer Encyclopedia
of Chemical Technology. Vol. 21. 3rd ed. New York: Wiley, 1978:77–105.
4. Pietsch WB. Size enlargement by agglomeration. In: Fayed M, Otten L, eds. Handbook of
Powder Science and Technology. New York: van Nostrand Reinhold Co, 1984:Chap. 7.
5. Pietsch WB. In: Roth D, ed. Agglomeration Techniques for the Manufacturing of
Granular Materials with Specific Product Characteristics. The Institute for Briquetting
and Agglomeration Proceedings. Vol. 25. Erie, PA: Gannon University, 1997:49–164.
6. Sherrington PJ, Oliver R. In: Granulation. London: Heyden & Son, 1981.
Figure 26 Milled tolmetin 10% compacts. (From Ref. 41.)
188 Miller
© 2005 by Taylor & Francis Group, LLC
7. Kapur PC. Adv Chem Eng 1978; 10:55.
8. Kristensen HG, Schafter T. Drug Dev Ind Pharm 1987; 13:803.
9. Lindberg N. Industrial wet granulation. Acta Pharm Suec 1988; 25(special issue):
185–280.
10. Fonner DE, Anderson NR, Banker GS. Granulation and tablet characteristics. In:
Lieberman HA, Lachman L, eds. Pharmaceutical Dosage Forms. Tablets. Vol. 2. New
York: Marcel Dekker, Inc., 1982:Chap. 5.
11. Anderson NR, Banker GS, Peck GE. Principles of improved tablet production system
design. In: Lieberman HA, Lachman L, eds. Pharmaceutical Dosage Forms. Tablets.
Vol. 3. New York: Marcel Dekker, Inc., 1982:Chap.1.
12. Ghebre-Sellassie I. Pharmaceutical Pelletization Technology. New York: Marcel Dekker
Inc, 1989.
13. Miller RW. Roller Compaction Technology. Parikh DM, ed. Handbook of Pharmaceutical
Granulation Technology. Vol. 81. New York: Marcel Dekker Inc, 1997:99–150.
14. Pietsch WB. Roll Pressing. London: Heyden, 1976.
15. Parrot EL. Pharmaceutical Dosage Forms. Vol. 2. New York: Marcel Dekker Inc,
1990:203–204.
16. Dehont FR, Hervieu PM, Jerome E, Delacourte A, Guyot JC. Briquetting and granulation
by compaction: a new granulator-compactor. Wells J, Rubinstein M, eds.
Pharmaceutical Technology, Tableting Technology. Vol. 2. London: Ellis Horwood,
1993:1–11.
17. Heckel RW. Density-pressure relationships in powder compaction. Trans Metall Soc
AIME 1961; 221:671–675.
18. Johanson JR. Rolling Theory for Granular Solids. Trans Am Soc Mech Eng 1965:
842–848.
19. Jenike AW, Shield RT. Plastic flow of Coulomb solids beyond original failure. J Appl
Mech 1959; 26:599–602.
20. Parrott EL. Densification of powders by concavo-convex roller compactor. J Pharm Sci
1981; 70(30):288–291.
21. Miller RW. Advances in pharmaceutical roller compactor feed system designs. Pharm.
Technol 1994:154–162.
22. Shileout G, Lammens RL, Kleinebudde P. Dry granulation with a roller compactor Part
1: the function units and operational modes. Pharm Technol Eur 2000:24–35.
23. Miller RW. Roller Compaction Technology for the Pharmaceutical Industry. Encyclopedia
of Pharmaceutical Technology. New York: Marcel Dekker, Inc., 2003 (on-line).
24. Dehont FR, Hervieu PM. Briquetting and granulation by compaction new granulator
compactor for the industry. Drug Dev Ind Pharm 1989; 15(14–16):2245–2263.
25. Pietsch W. Size enlargement by agglomeration. Fayed M, Otten L, eds. Handbook of
Powder Science, and Technology. 2nd ed. New York: Chapman and Hall, 1997:
347–364.
26. Funakoshi Y, Asogawa T, Satake E. Use of a novel roller compactor with a concavoconvex
roller pair to obtain uniform compacting pressure. Drug Dev Ind Pharm 1977;
3(6):555–573.
27. Parrott EL. Densification of powders by concavo-convex roller compactor. J Pharm Sci
1981; 70(3):288–289.
28. Jerome E. Measurement of resulting forces on a roller compactor. Drug Dev Ind Pharm
1991; 17(12):1571–1591.
29. Johanson JR. Predicting limiting roll speed for briquetting presses. Proceedings of the
13th Institute for Briquetting and Agglomeration. Vol. 13. 1975:89–99.
30. Sheskey PJ, Hendren J. The effects of roll compaction equipment variables, granulation
technique, and HPMC polymer level on a controlled-release matrix model drug formulation.
Pharm Technol 1999; 23(3):90–106.
31. Johanson JR. Roll press feed systems. Proceedings of the 24th Institute for Briquetting
and Agglomeration, October. Vol. 24. 1995:149–163.
Roller Compaction Technology 189
© 2005 by Taylor & Francis Group, LLC
32. Weggel RW. The basics of force-feed roll briquetting. Manuf Eng 1982:107–109.
33. Miller RW, Sheskey PJ. Survey of current industrial practices and preferences of roller
compaction technology and excipients year 2000. Am Pharm Rev Spring 2001;
4(1):24–35.
34. Guigon P, Simon O. Correlation between the geometry of feeding system and the stress
distribution applied on the compact. In: Roth D, ed. Proceedings of the 27th Institute for
Briquetting and Agglomeration, Providence, RI, Nov 2001. Vol. 27. Erie, PA: Gannon
University, 2001:31–41.
35. Johanson JR, Cox BD. Fluid entrainment effects in roll press compaction. Proceedings of
the 20th Institute for Briquetting and Agglomeration. Vol. 20. 1987:251–263.
36. Dec RT. Problems with processing of fine powders in roll press. Proceedings of the 24th
Institute for Briquetting and Agglomeration. 1995; 24:199–210.
37. Miller RW. Using vacuum–deaeration feed system to minimize powder leakage during
roll compaction. Powder Bulk Eng 1997; 11(2):71–75.
38. Miller RW. Vacuum deaeration advances in pharmaceutical roller compaction technology.
Proceedings of the 24th Institute for Briquetting and Agglomeration 1995;
Vol. 24:165–173.
39. Miller RW. Roller compaction optimization—NIR in-process mapping. J Pharm Technol
1999; yearbook supplement:30–39.
40. Miller RW. Process analytical technologies PAT part 2. Am Pharm 2003; 6(1):52–61.
41. Gupta A, Peck GE, Morris KR. Near infrared monitoring of roller compaction. In:
Roth, ed. Proceedings of the Institute of Briquetting and Agglomeration. Vol. 28. Erie,
PA: Gannon University, 2003:20–34.
42. Kirsch JD, Drennen JK. Nondestructive tablet hardness testing by near-infrared spectroscopy:
a new and robust spectral best-fit algorithm. J Pharm Biomed Anal 1999; 19(3–4):
351–362.
43. Johanson JR. Reducing Air Entrainment Problems in Your Roll Press. Powder and Bulk
Engineering 1989; 3(2):43–46.
190 Miller
© 2005 by Taylor & Francis Group, LLC
7
High-Shear Granulation
Rajeev Gokhale*
Incyte Corporation, Wilmington, Delaware, U.S.A.
Yichun Sun and Atul J. Shukla
College of Pharmacy, University of Tennessee, Memphis, Tennessee, U.S.A.
1. INTRODUCTION
Granulation is the process of agglomeration of a powder mixture, which results in the
enlargement of the particles. This is often necessary for manufacturing of solid dosage
forms such as tablets. The materials, which are compressed into tablets, must possess adequate
flowability, density, and compressibility. This is because the requisite amount of
powder mixture required to compress each tablet is filled into the die cavity by volume
and not by weight. This requirement of adequate flowability, density, and compressibility
is particularly important during a high-speed tablet production where the dwell time is
often short. For example, active pharmaceutical ingredients such as ibuprofen and acetaminophen,
which have inadequate flow and compression properties, and a relative high
dose, are often granulated prior to compression into tablets. Thus, the overall purpose of
granulation is to improve the flowability and compressibility of the powder mixture.
Besides improving the flowability and compressibility, the granulation process can also

 Densify the powder mixture and reduce dust

 Narrow the particle size distribution of the powder mixture

 Ensure uniform distribution of the drug in the powder mixture

 Improve the dissolution characteristics of the finished tablets.
The three commonly used granulation methods include wet granulation, dry
granulation, and hot-melt granulation. These methods are categorized based on
the type of binder and the process employed during granulation. The equipment that
is used during the granulation processes is classified into the following three major
categories, based on the shearing strength it generates on the powder bed:
1. Low-shear granulators—twin shell (Peterson Kelly, PK) with an agitator
bar, dough mixer or planetary mixer, ribbon blenders, and fluid bed
granulator without the rotogranulator
* Present Address: Merck Research Laboratories, West Point, Pennsylvania, U.S.A.
191
© 2005 by Taylor & Francis Group, LLC
2. Medium-shear granulators—fluid bed granulators with a rotogranulator
attachment
3. High-shear granulators.
The granulation techniques can affect the physical properties of the resulting
granules, which then subsequently affect the tableting process, the quality, and the
performance of the finished tablets, as illustrated in the following examples.
An extended release of matrix tablet formulation for metoprolol tartrate (100mg)
was prepared with three different granulation processes: direct compression, fluid-bed,
or high-shear granulation (1). Different grades of hydroxypropylmethylcellulose
(HPMC) (Methocel K4M, K15M, K100M, and K100LV) were used as fillers and
binders. Direct compression formulations exhibited poor flow, picking and sticking
problems during tableting. High-shear granulation produced hard granules, which were
difficult to mill. However, they yielded good tablets. Granules produced by the fluid-bed
granulation process appeared to have satisfactory flow and tableting performance.
A hydrophilic matrix tablet formulation containing 60–70% drug, and lowor
medium-viscosity grades of HPMC for extended oral delivery of zileuton, was
prepared using the wet-granulation techniques (2). The granules were prepared with
planetary (low shear) and high-shear granulators. In vitro drug release was evaluated
using USP apparatus 1. Slower drug release was achieved from the granules prepared
with the high-shear granulation process, because of the slower water penetration into
the denser and less porous granules.
A high-shear mixer was utilized for the evaluation of pregelatinized,
cross-linked, waxy corn starches as binding agents in the wet-granulation process (3).
Lactose granules prepared in a high-shear mixer showed a larger average size than
granules prepared in a low-shear mixer (planetary mixer), using the same amount of
binding solution. The granules produced with high-shear mixer had lower friability
than those prepared with the planetary mixer.
High-shear granulation has been one of the most commonly used methods
to produce granules since the early 1980s (4). Hence, this chapter discusses in detail,
the equipment, process variables, formulation requirements, granulation end-point
determination, and scale-up considerations of the high-shear granulation process.
2. HIGH-SHEAR GRANULATORS
Most of the high-shear granulators consist of a mixing bowl, a three-bladed impeller,
and an auxiliary chopper. The shape of the mixing bowl could be cylindrical or conical.
The mixing bowl can be jacketed for heating or cooling the contents in the bowl,
by circulating hot or cool liquid or steam through the jacket. An impeller is
employed to mix the dry powder and spread the granulating fluid. The impeller of
the high-shear mixer granulator normally rotates at a speed ranging from 100 to
500 rpm. The function of the chopper is to break down the wet mass to produce
granules. The rotation speed of the chopper ranges from 1000 to 3000 rpm. The
high-shear granulator could be termed as either vertical or horizontal, based on
the orientation and the position of the impeller. The vertical high shear granulator
view of a top-driven vertical high-shear granulator, ULTIMAGRALTM/ULTIMAPRO
TM
vertical high-shear granulators: ULTIMATM, GMXTM, and GMATM
shows photographs of top-driven, vertical, high-shear granulators: UltimaGral 150,
192 Gokhale et al.
could be either a top-driven or a bottom-driven unit. Figure 1 shows the schematic
© 2005 by Taylor & Francis Group, LLC
. Figure 2(a–c) shows photographs of choppers and impellers of top-driven
. Figure 3(a–d)
Figure 1 Schematic view of top-driven vertical high shear granulator. (ULTIMAGRALTM/
ULTIMAPROTM courtesy of Niro Pharma Systems.)
Figure 2 (A) Photograph of an ULTIMATM top-driven impeller and chopper. (Courtesy
of Niro Pharma Systems.) (B) Photograph of a GMXTM top-driven impeller and chopper
(Courtesy of Vector Corporation.) (C) Photograph of GMATM top-driven impeller and
chopper (Courtesy of L. B. Bohle.)
High-Shear Granulation 193
© 2005 by Taylor & Francis Group, LLC
GMX 10, GMA 1200, and Glatt. shows the schematic view of a
bottom-driven vertical high-shear granulator, with a horizontal chopper shaft
Pharma MatrixTM
photograph of a plow-shaped mixing blade and a chopper inside a horizontal
Figure 3 (A) Photograph of a top-driven vertical high shear granulator (UltimaGral 150).
(Courtesy of Niro Pharma Systems.) (B) Photograph of a top-driven vertical high shear granulator.
(GMX 10 courtesy of Vector Corporation.) (C) Photograph of a top-driven vertical
high shear granulator. (GMA 1200 courtesy of L. B. Bohle.) (D) Photograph of a top-driven
vertical high shear granulator. (Courtesy of Glatt Air Techniques.)
194 Gokhale et al.
Figure 4
(PMA through-the-wall design). Figure 5(A) and 5(B) shows photographs of top
views of bottom-driven impellers with horizontal choppers (Diosna and Glatt). Figure
PMA600 Podium, Diosna P 150, and Glatt. Figure 7 shows the
schematic view of a horizontal high-shear granulator (Loedige). Figure 8 shows a
© 2005 by Taylor & Francis Group, LLC
6(A–C) shows the photographs of bottom-driven vertical high-shear granulators:
high-shear mixer granulator (FKM-1200–Loedige).
The size of the high-shear granulators varies considerably from the small
laboratory-scale models to large production-scale models, depending on the type of
the model and end use. High-shear granulators are often equipped with granulation
end-point control devices, which are used to detect the end point of the granulation
process. Some of the recently introduced features in high-shear mixer granulators are
clean in place (CIP) or wash in place (WIP), and one-pot processing. High-shear
granulators with CIP or WIP features can be cleaned easily, without removing the
bowl, the impeller, and the chopper. The advantage of the one-pot processing system
is that the wet granules can be dried in the same bowl without additional equipment
for drying. Some of the suppliers of the commercially available high-shear granula-
3. HIGH-SHEAR GRANULATION PROCESS
3.1. Wet Granulation
The composition of a powder mixture for granulation generally consists of an active
pharmaceutical ingredient (API), a filler, a disintegrant (usually not included in controlled
release tablets), and a binder.
A high-shear wet-granulation process includes the following steps:
1. Loading all the ingredients into the mixing bowl, which can be accomplished
by either of the following methods: gravity feeding with manual
or pneumatic valve, and vacuum feeding.
2. Mixing of dry ingredients such as API, filler, and disintegrant, at high
impeller and chopper speeds for a short period of time (2–5 min).
3. Addition of a liquid binder (either binder solution or solvent) into the powder
mixture, while both the impeller and the chopper are running at a low
speed.
Figure 4 Schematic view of a bottom-driven vertical high shear granulator with horizontal
chopper shaft. (PMA through-the-wall courtesy of Niro Pharma Systems.)
High-Shear Granulation 195
high-shear mixer granulator (Loedige). Figure 9 shows a photograph of a horizontal
tors are listed in Table 1.
© 2005 by Taylor & Francis Group, LLC
4. Wet massing with both the impeller and the chopper running at a high
speed.
5. Removal of the resulting wet granules from the granulator bowl, and drying
them using an appropriate drying technique such as fluid-bed or tray
drying.
6. Sieving the dried granules.
The high-shear wet-granulation process offers several advantages over the
other granulation processes. These include:

 Short processing time

 Use of less binder solution

 Granulation of highly cohesive materials containing hydrophilic polymers,
which is not achievable with low-shear granulation processes

 Greater densification and production of less friable granules

 Production of reproducible granules with a uniformparticle size distribution
Figure 5 (A) The top view of a bottom-driven impeller with a horizontal chopper. (Courtesy
of Diosna.) (B) The top view of a bottom-driven impeller with a horizontal chopper inside a
conical shaped mixing bowl. (Courtesy of Glatt Air Techniques.)
196 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC

 Reduction of process dust, thus minimizing exposure to workers

 Obtaining predictable granulation end-point determination.
Despite the above advantages, the process is not immune to challenges such as:

 Production of less compressible granules, compared to low-shear granulation
processes

 Narrow range of operating conditions
A commonly employed high-shear granulation process requires multiple unit
operations, such as drying and sieving after wet granulation. The preferred drying
method is fluid-bed drying, because of the distinct advantages over tray drying.
However, a primary problem inherent in the two-step process of high-shear mixer
granulation and fluid-bed drying is the possibility of exposure of the workers and
Figure 6 (A) Photograph of a bottom driven Pharma MatrixTM PMA600 Podium/
Through-The-Wall. (Courtesy of Niro Pharma Systems.) (B) Photograph of a bottom driven
Mixer–Granulator. (P 150 courtesy of Diosna Dierks & Soehne GmbH.) (C) Photograph of a
bottom driven high shear granulator. (Courtesy of Glatt Air Techniques.)
High-Shear Granulation 197
© 2005 by Taylor & Francis Group, LLC
the environment to potentially toxic materials during the transfer of the wet granules
from the high-shear granulator to the fluid bed. In order to remedy the contamination
problem, the processing suites need to be operated under a negative pressure using
powerful air filtration systems. Moreover, dehumidifying and heating of a large volume
of air is also necessary during the fluid-bed drying process. Regulating the airflow and
filtering the exhaust air during the drying process could also be challenging.
However, the above problems can be remedied by a one-pot approach in which
the high-shear wet-granulation and drying processes are combined into a single step.
This approach saves time, since cleaning and validating a single bowl is inherently
easier than tackling two or three units. For example, a Mi-Mi-Pro high-shear granulator,
combined with microwave, was employed to study the effect of process variables
on the properties of the granules obtained, for the mixture of alpha-lactose
monohydrate and microcrystalline cellulose (MCC) (5). The one-pot high-shear
granulators are commercially available from several manufacturers. Options such
Figure 7 Schematic view of a horizontal high shear mixer granulator. (Courtesy of Loedige.)
Figure 8 Photograph of a plough-shaped mixing blade and a chopper inside a horizontal
high shear mixer granulator. (Courtesy of Loedige.)
198 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC
as oscillating the bowl during drying, vacuum drying, gas-assisted vacuum drying,
single-pot processor ULTIMAPROTM, based on the ULTIMAGRALTM with a
cessor at a swinging position (Niro Inc., Pharma Systems Division).
Another approach is to modify the wet-granulation process. This process is
called moisture-activated dry-granulation process. The moisture-activated drygranulation
process consists of two steps, wet agglomeration of the powder mixture
followed by moisture absorption stages. A small amount of water (1–4%) is added
first to agglomerate the mixture of the API, a binder, and excipients. Moisture
absorbing material such as MCC and potato starch is then added to absorb any
excessive moisture. After mixing with a lubricant, the resulting mixture can then
be compressed directly into tablets. Hence, this process offers the advantages of
wet granulation, but eliminates the need for a drying step.
The applicability of a 25 L high-shear mixer for moisture-activated dry granulation
of phenobarbital was investigated by Christensen et al. (6). MCC, potato starch,
or a mixture of 50% of each was used as moisture absorbing material. The results of
the study showed that the physical properties of the tablets were primarily affected by
the water content, the moisture absorbing material, and the compression force.
3.2. Hot-Melt Granulation
Hot-melt granulation in a high-shear granulator is a versatile approach. Hot-melt
granulation utilizes a binder, which is a solid or semisolid at room temperature
and melts at a temperature below the melting point of API. Generally, the melting
point of such binders is between 30C and 100C. The binder, when heated near
its melting point, liquifies or becomes tacky. This tacky and liquified form of binder
agglomerates the powder mixture, which upon cooling forms a solid granulated
mass. The energy to melt the binder may be derived from heat dissipated from the
circulating hot liquid, such as steam, water, or oil through jacketed bowl, or heat
generated by the friction of high-shear mixing. A solvent such as water or an organic
Figure 9 Photograph of a horizontal high shear mixer granulator. (FKM-1200 courtesy of
Loedige).
High-Shear Granulation 199
and microwave drying are also available. Figure 10 shows the schematic view of a
vacuum drying system. Figure 11 shows the photo of UltimaPro 600 single-pot pro-
© 2005 by Taylor & Francis Group, LLC
compound is not necessary to initiate particle binding in this method. The resultant
granules are compressed into tablets.
A series of studies have been carried out by Schaefer and his colleagues to
investigate the effects of formulation and process variables on the hot-melt granulation
process, and the properties of the obtained granules (7–13).
In spite of numerous granulation options available to a formulator, high-shear
wet granulation remains the method of choice.
4. MECHANISM OF HIGH SHEAR WET GRANULATION
The high-shear wet-granulation process can be divided into five stages as shown
below, namely, mixing, adding binder solution, wetting and nucleation, consolidation
and growth, and granule attrition and breakage.
The binder solution or the granulating liquid is distributed through the powder
bed by mechanical agitation created by the impellers. At this stage, the powder mixture
becomes wetted and initiates agglomeration via nucleation. Nucleation of particles
occurs by the formation of liquid bridges between primary particles, which
adhere together to form agglomerates. At this stage, the concentration of liquid
phase in the powder mixture is relatively low, but high enough to establish liquid
bridges. The addition and distribution of binder solution can have a major impact
on the particle size distribution and quality of the granules. Poor liquid distribution
produces granules with a wide particle size distribution. These granules could include
large, overwetted particles, as well as small dry particles.
Granule growth is dominated by one of the two mechanisms: coalescence
or layering (14). Coalescence is agglomeration which occurs by collision and consolidation
of deformable nuclei/granules, provided the agglomerates could withstand
the shear forces applied by the impellers. Layering occurs when fine particles stick
to larger particles or granules coated with binder. Free liquid on the surface of
agglomerates, which renders the necessary bonding strength and plasticity to the
agglomerate, is required for coalescence or layering. The free liquid on the surface
of the agglomerates can be from the addition of more binder solution or from
200 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC
the expulsion of liquid inside the agglomerates due to the consolidation of the
agglomerates.
Granule attrition and breakage reduce the granule size. Granule breakage
is determined by the dynamic granule strength and the shear forces within the
Table 1 List of High-Shear Granulator Suppliers
Supplier Granulator Contact information
DIOSNA Dierks &
Soehne GmbH
Diosna P
high-shear
granulators
270 Route 46 Rockaway, NJ 07866,
Tel: (973) 586–3708,
Fax: (973) 586–3731,
e-mail: marc_kaufman@servo-lift.com
Glatt Air Techniques VG high-shear
granulators
20 Spear Rd,
Ramsey, NJ 07446,
Tel: (201) 825–8700,
Fax: (201) 825–0389,
e-mail: info@glattair.com
Key International,
Inc.
KG-5
high-shear
granulators
480 Route 9,
Englishtown, NJ 07726,
Tel: (732) 536–9700,
Fax: (732) 972–2630,
e-mail: info@keyinternational.com
L.B. Bohle Inc. GMA, VMA, and
BMG high-shear
granulators
L.B. Bohle, LLC,
700 Veterans Circle,
Suite 100,
Warminster, PA 18974,
Tel: (215) 957–1240,
Fax: (215) 957–1237,
e-mail: info@lbbohle.com
Littleford Lodige KFM and MGT
high-shear
granulators
Lodige Process Technology, Inc.,
203 Forrester Creek Way,
Greenville, SC 29607,
Tel: (864) 254–9194,
www.loedige.de
Niro Inc. Pharma
Systems Division
ULTIMAGRALTM,
ULTIMAPROTM,
and PMA highshear
granulators
9165 Rumsey Road,
Columbia, MD 21045,
Tel: (410) 997–7010,
Fax: (410) 997–5021,
e-mail: info@niroinc.com
Vector Corporation GMX
high-shear
granulators
Vector Corporation,
675 44th St.
Marion, IA 52302–3800,
Tel: (319) 377–8263,
Fax: (319) 377–5574,
www.vectorcorporation.com
Zanchetta
(A division of
Romaco)
ROTO
high-shear
granulators
Romaco USA,
242 West Parkway,
Pompton Plains, NJ 07444,
Tel: (973) 616–0440,
Fax: (973) 616–9985,
e-mail: usa@romaco.com
High-Shear Granulation 201
© 2005 by Taylor & Francis Group, LLC
granulator. If the impact forces are larger than the granule strength, continuous
breakage and immediate coalescence of the granules takes place (15). However, when
the granule strength exceeds the impact forces, granules will not break. In that case,
granule growth is more static, i.e., the exchange of primary particles between the
Figure 10 Schematic view of a single-pot processor ULTIMAPROTM based on the ULTIMAGRAL
TM with a vacuum drying system. (Courtesy of Niro Pharma Systems.)
Figure 11 Photograph of an UltimaPro 600 single pot processor (granulator is in a swinging
position with an operator). (Courtesy of Niro Pharma Systems.)
202 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC
granules is minimal. Thus, there is a balance between agglomerate growth and degradation
of the granules.
The liquid level in the powder bed in a high-shear granulator plays a key role in
the growth of granules. Granule growth and consolidation (densification) occur at
the same time in the high-shear granulation process. Since deposition of free liquid
on the surface of agglomerates is required for coalescence or layering to occur, maintenance
of the free liquid on the surface of the agglomerates is critical for the growth
of granules in a high-shear wet-granulation process (16). The agglomeration of fine
particles is controlled mainly by the liquid saturation of the moist agglomerates,
which is the percentage of intragranular voids filled with the liquid. Increasing the
amount of granulating liquid reduces the yield stress necessary to deform the
agglomerates in collision, thus allowing growth by coalescence with other granules.
This deformation also makes it possible to consolidate the granules, which further
increases the liquid saturation level and squeezes additional fluid to the granule surface
for continued growth.
The moist agglomerates can exist in the following three states, based on the
amount of the liquid they contain: the pendular state, the funicular state, and the
capillary state. The three states can be distinguished by the relative liquid saturation
of the pores S, which is the ratio of pore volume occupied by the liquid to the total
volume of the pores available in the agglomerate.
Agglomerates are in the pendular state when the liquid saturation of the pores
S<25%; in the funicular state when S is between 25% and 80%; and in the capillary
state when S>80%. The liquid saturation S is determined by the amount of binder
solution and the intragranular porosity, which can be calculated based on the
following equation (17):
S ?
H?1  e?
e
r ?4:1?
where H is the ratio of the mass of liquid binder to the mass of solid particles, e is the
intragranular porosity, and r is the particle density.
Therefore, the formulation, process, and granulator factors that affect the
degree of liquid saturation would play a critical role in defining the robustness of
the granulation process and the properties of the granules obtained. The effects of
the starting material properties, the type of granulator, and its operation on the
granule growth can be evaluated by assessing the changes in the intragranular
porosity, which occur during the process, since intragranular porosity affects the
degree of liquid saturation as indicated by Eq. (4.1). The properties of the
granules, such as particle size distribution, porosity, and content uniformity would
affect the properties and performance of the dosage forms prepared from these
granules.
4.1. Liquid Requirement
The optimal amount of liquid binder used for high-shear granulation is within a narrow
range. If the optimal amount of binder solution is not used, there could be
neither growth nor lump formation during the wet-granulation process. In theory,
the amount of liquid used during the wet-granulation process should be equal to
or just exceed the liquid content corresponding to 100% liquid saturation. However,
some grades of lactose and dicalcium phosphate (DCP) could be granulated by coalescence
at liquid saturation far below 100% (18). It would be desirable to predict the
High-Shear Granulation 203
© 2005 by Taylor & Francis Group, LLC
required amount of liquid binder based on the formulation composition. However,
this is a challenging task, since the liquid binder required for high-shear granulation
depends on a number of formulation and process factors, such as physicochemical
properties of the starting materials (API and excipients), type of the binder used,
impeller speed, wet-massing time, etc.
5. FACTORS AFFECTING THE GRANULATION PROCESS
AND GRANULE PROPERTIES
Performance and physical properties of the compressed tablets are driven by properties
of the granules. Yajima et al. (19) examined the relationship between particle size
distribution and physical properties of the obtained granules (e.g., angle of repose,
angle of spatula, loose bulk density, tapped bulk density, compressibility) and the
resulting tablets (hardness and weight variation of tablets). The particle size distribution
of the granules used for tablet compression was expressed as a function of median
particle size, and standard deviation with a logarithmic normal distribution.
Physical properties of granules and the compressed tablets were significantly affected
by this factor. Tablet hardness increased as the median particle size decreased, when
the standard deviation was <1.0.
In another example, Wikberg et al. (20) produced 11 granulations of a common
filler (lactose) and three granulations of a high-dose drug (dipentum) by wet
granulation with polyvinylpyrrolidone (PVP) as binder, in a high-shear mixer granulator.
The agglomeration process was varied to produce granules with varying
granule porosity. The granule fragmentation during compression was evaluated by
measurements of air permeability of the tablets. The results showed that the degree
of granule fragmentation during compression was related to the granule porosity
before compaction. A granulation with a higher porosity had a higher fragmentation
propensity during compression. The tablet strength correlated well with the degree of
fragmentation, i.e., a granulation with a higher degree of fragmentation yielded
tablets of a higher hardness. Variations in compactibility can be explained by variations
in granule porosity for the same formulation that was wet granulated under
different process conditions. Therefore, the formulation variables and granulation
process should be controlled to produce granules with the desired properties. Since
granule formation in high-shear wet granulation is dependent on the degree of liquid
saturation, the factors that influence the degree of liquid saturation should be optimized
to obtain an optimal granulation formulation. The factors that affect the
degree of liquid saturation (which depends on the amount of liquid added as well
as on the densification of the granules) are formulation, process, and apparatus variables.
Hence, the granulation process and the properties of the granules are affected
by the aforementioned variables.
5.1. Formulation Variables
Besides the API, excipients such as a filler or fillers, a disintegrant (not applicable to
controlled release), and a binder are also included in the powder mixture of a tablet
formulation. The fillers used in tablet formulations can be classified into two categories,
based on their water solubility: soluble fillers such as lactose, sucrose, mannitol,
etc., and insoluble fillers such as MCC, starch, calcium carbonate, calcium
phosphate, etc. The binders used in the wet granulation process are water-soluble
204 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC
polymers such as gelatin, PVP, HPMC and sugars such as glucose, sucrose, and sorbitol.
Some of the commonly used disintegrants are sodium starch glycolate, crosslinked
PVP, and cross-linked sodium carboxymethylcellulose. These excipients are
available in various grades from vendors. Hence, the physical properties such as particle
size distribution, particle shape, surface morphology, surface area, and solubility
of API and excipients in the binder solution could vary considerably. The rate
and the final degree of densification are controlled by the physical properties such
as particle size distribution, particle shape, surface morphology, surface area, solubility
in the binder solution, etc., of the starting materials. Therefore, the physical
characteristics of drug and drug loading, the types and amount of excipients, and
the type and amount of binder used could affect the rate and the final degree of
densification of the resulting granules. Moreover, these aspects of the powder mixture
can also affect the amount of liquid required for granulation and the degree
of liquid saturation in the agglomerates, during the wet-granulation process, which
in turn affect the physical properties of the obtained granules.
5.2. Effects of Starting Materials
The physical properties of the granules are influenced by the physical properties of
the starting materials. For example, the shape, size, and size distribution of the particles
of the starting materials can affect the packing pattern of a powder mixture,
and ultimately the final degree of densification and the strength of the resulting granules.
The extent of granule densification affects the degree of liquid saturation of the
powder mixture, and the amount of liquid binder required for granulation. The total
surface area of the starting materials, related to the particle size and porosity of the
starting materials, will affect the amount of liquid binder required for granulation.
The liquid adsorption of the starting material could also affect the amount of liquid
binder required for wet granulation. Therefore, the amount of binder solution required
for granulation depends on the formulation composition, and the physical
characteristics of the starting materials. In a study by Schaefer et al. (21), when
the mean particle size of lactose was decreased, the amount of liquid required for
granulation was increased to obtain a similar particle size of granules in high-shear
granulation.
The influence of the primary particle size of DCP on granule growth in a
Lo?dige M5GR high-shear mixer granulator was investigated by Schaefer et al.
(22). Two batches of DCP, one with a narrower particle distribution and smaller
geometric mean diameter, and the other with a wider particle size distribution and
a larger geometric mean diameter were used. The batch with the narrower particle
size distribution and smaller geometric mean diameter was more easily densified than
the one with the wider particle size distribution and a larger geometric mean diameter,
because it was overwetted at lower moisture content. The granules with wider
particle size distribution were produced from the DCP with a wider particle size distribution
and larger geometric mean diameter.
The strength of wet agglomerates can be dominantly affected by the degree of
liquid saturation and the primary particle size of the starting materials. Consequently,
the growth mechanism and drug content uniformity of the resulting granules
can be affected.
The influence of the primary particle size of lactose on the breakage of erythrosine
granules was investigated by van den Dries et al. (4). The results showed that
a decrease in particle size of the starting material, lactose, led to a decrease in
High-Shear Granulation 205
© 2005 by Taylor & Francis Group, LLC
breakage of granules. When granule breakage was absent, the granules remained
intact and the preferential layering of the smallest particles yielded inhomogenous
granules.
The effect of particle size of the lactose on the homogeneity of low-dose steroid
hormone distribution in granules produced in high-speed mixers was studied by
Vromans et al. (23). When a micronized low-dose steroid hormone was granulated
with unmicronized lactose at 250 rpm, inhomogeneity of active was observed. The
coarse particle size fractions of the granules were found to be superpotent, up to
150% of the mean drug content, whereas the fine size fraction showed a corresponding
subpotency of 50%. However, when the particle size of lactose was reduced,
a better drug distribution was seen. This was attributed to an increase in the tensile
strength of the nuclei due to the smaller particle size of lactose.
The effect of particle size of lactose on the homogeneity of estradiol granules
prepared in a 10 L high-shear mixer was studied by van den Dries et al. (15). Three
different particle sizes of lactose were granulated with 0.1% micronized estradiol
(5 mm) using an aqueous solution of hydroxypropyl cellulose (HPC) as a binder.
Granules prepared with the largest lactose particles (141 mm) yielded a homogenous
granulate. However, after a prolonged processing time, demixing was observed.
Contrary to the largest particles, granulation with the smaller lactose particles (23
and 50 mm) led to demixing in the first minute, although to a lesser extent. It was
concluded that granulation with the largest particles resulted in breakage of the granules,
thereby preventing demixing. However, once the granules were strong enough
(smaller particle size and prolonged process time) to survive the shear forces, demixing
was observed. The extent of demixing depended on the particle size difference.
The size distribution of granules prepared from three different tablet excipients,
lactose, glucose, and mannitol, in a high-shear granulator, were studied by both
sieve analysis and laser light diffractometry (24). Both the shapes and the size distribution
of the granules produced from the three excipients differed.
The crystallinity of the starting materials affected the granulation process and
the properties of the granules. The effect of crystallinity of MCC on wet granulation,
in binary mixtures of MCC and corn starch (CS), was investigated by Suzuki et al.
(25). The crystallinity of MCC was modified, producing amorphous material by
milling in a jet mill (J-MCC) and a vibrational rod mill (R-MCC). Crystallinity of
J-MCC was nearly equal to that of intact MCC (I-MCC), but that of R-MCC
was remarkably low due to the mechano-chemical effect. The growth rate of granules
with R-MCC was greater than that with I-MCC and J-MCC. The formulations containing
I-MCC and J-MCC produced granules with a core (MCC and CS) around
which CS was layered. R-MCC provided granules in which CS and abraded MCC
were homogenously mixed. In addition, R-MCC remarkably decreased the difference
in drug content among the size fractions of granules compared with I-MCC.
The decrease in crystallinity of MCC increased the abrasion of the material by the
impeller. This resulted in a greater growth rate of granulation.
The physical properties of the API also affect the wet granulation process and
the properties of the resulting granules. For example, when the particle shape of the
API was changed from spherical to plate like due to a change in the crystallization
solvent, the compressibility of the resulting granules dramatically decreased (26).
The effect of the particle size of DPC 963 (API) on the characteristics of granules
prepared by high-shear wet granulation was evaluated by Badawy et al. (27).
Particle agglomeration was affected by the particle size of the drug substance.
Granule geometric mean diameter and fraction with particle size greater than
206 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC
250 mm was inversely proportional to the particle size of the drug substance. The
granules prepared with the smaller particle size had higher porosity, thus suggesting
lower tendency for granule densification than those manufactured with the larger
particle size. The compressibility of the granules was increased with decreasing particle
size of the drug substance. The effect of particle size on granulation growth is
the result of increased densification propensity from increased drug particle size.
In another study, a high-shear mixer granulator was used for pelletizing a drug.
High-dose pseudoephedrine-HCl pellets were produced with MCC in a high-shear
mixer granulator (28). The drug loading and drug particle size influenced the bulk
and true densities of the resulting pellets.
5.3. Effects of Type and Amount of Binder Solution Used
A binder is normally required for the wet-granulation process. The binder can be
added into the powder mixture as a dry powder followed by the addition of water,
or an appropriate solvent to activate binding. Alternatively, the binder can also be
added into the powder mixture as a binder solution. Because of the higher densification
of the wet agglomerates, the high-shear granulation process normally only needs
two-thirds to three-quarters of the liquid required during the low-shear granulation
process (18). The bonding characteristics of the binders used for wet granulation
could vary due to the differences in their physicochemical properties. Therefore,
the type of binder used for granulation can influence the granulation process, the
amount of binder required for granulation, and the physical properties of the
obtained granules.
For example, the effect of binder solutions on the size of DCP granules was
investigated by Ritala et al. (29). Five different binders, PVP (Kollidone 90), polyvinylpyrrolidone–
polyvinylactate copolymer (PVP–PVA copolymer), hydrolyzed
gelatin and HPMC (Methocel E5 and Methocel E15) were studied. The granules prepared
with PVP and hydrolyzed gelatin possessed a larger mean particle size and
lower intragranular porosity when the same volume of binder solutions was used.
The other three binder solutions behaved similarly. This could be due to the higher
densification of DCP granule caused by PVP and hydrolyzed gelatin. Thus, the granules
had higher liquid saturation with lower volumes of the binder solutions. It was
apparent that increasing the concentration of the binder solution in the powder
mixture increased the mean granule size.
In another study, Ritala et al. (30) investigated the differences between five
binders during wet granulation using a high-shear granulator. The binders used
for the study were PVP K90, PVP K25, PVP–PVA copolymer, Methocel E5, and
Methocel E15. The results demonstrated that the binder solutions with high surface
tension, such as PVP solutions, produced denser granules with larger mean particle
size than those with low surface tension, such as PVP–PVA copolymer,Methocel E5,
and Methocel E15. The difference of granule growth by different binders in highshear
wet granulation could be due to the variation of surface tension of the binder
solutions.
The amount of binder and granulating liquid used during granulation affect
the granulation process and the physical characteristics of the prepared granules.
In general, the granules are finer when the amount of granulating liquid used for
granulation is decreased. For example, the size of the granules prepared from three
tablet excipients, lactose, glucose, and mannitol, increased when the amount of
granulation liquid was increased (24). Similar results were observed for granulating
High-Shear Granulation 207
© 2005 by Taylor & Francis Group, LLC
mixtures of mannitol and MCC (31). When the mixtures were granulated in a highshear
mixer using HPC as the binder, the mean particle size of the granules increased
with increasing amounts of water and concentration of the binder.
Generally, an increase in binder concentration could result in the production of
larger granules. However, this is not always the case with starch paste since it could
behave differently from the other binders. For example, a high-shear mixer was utilized
for evaluation of pregelatinized and pregelatinized, cross-linked, waxy corn
starches as binding agents in the wet-granulation process (3). Increasing the concentration
of the binders decreased the average granule size; however, no difference in
friability of the granules was observed. The use of pregelatinized or pregelatinized,
cross-linked starches as a paste provided larger and less friable granules than when
the binder was added in the dry form.
The effect of starch paste concentrations on the particle size distribution of fine
granules, produced by high-speed mixer, was investigated by Makino (32). When a
low concentration (5%) of starch paste was used, the growth of granules occurred
mainly during the massing time, since a relatively large amount of free water was
available from the paste. Only the mean particle size of the granules became larger
without changing the particle size distribution of the granules during granulation.
However, when a high concentration (15%) of starch paste was used, rapid deaggregation
of granules occurred due to the applied shear force, following the formation
of large granules, immediately after the addition of the starch paste. Deaggregation
of these large agglomerates became predominant during the wet-massing stage, and
hence, the size distribution of the granules became narrower. Longer massing time
was also required when a high concentration (15%) of starch paste was used, since
only a small amount of water exuded from the more concentrated starch paste during
the agitation.
Besides the effect on the particle size of the granules, the amount of binder
solution used in the wet-granulation process can also affect the content uniformity
of the granules. The effects of the amount of binder solution on the content uniformity
of the oily drug D-alpha-tocopheryl acetate (VE) in granules obtained by highshear
wet granulation was investigated by Kato et al. (33). When the amount of
binder solution was below the amount of water required to reach the plastic limit
(minimum amount of water required for the powder mixture to form a large agglomerate
when kneaded), the content of VE was <50% in the fractionated fine granules,
but was >200% in the fractionated large granules. Large variations were seen in the
contents of VE even if the granulation time was extended up to 30 min. When the
amount of binder solution was at or above the amount of water required to reach
the plastic limit, less variation was observed in the content of VE throughout the
granules, and the content of VE was fairly uniform.
In another study, the effect of the amount of water as the granulating liquid
and HPC as a binder on the pharmaceutical properties of granules of two model
drugs, ascorbic acid and ethenzamide, was investigated by Miyamoto et al. (34).
The dependence of drug content uniformity of these two model drugs on the granule
size was also investigated in this study. For a water-soluble drug, ascorbic acid, an
appreciable dependence of drug content on granule size was not observed in model
formulations. However, for a poorly water-soluble drug, ethenzamide, more drug
was found in small-size granules (<75 mm). The drug content of ethenzamide in
small-size granules decreased with increasing amounts of HPC and granulation
liquid. These observations suggested that drug content uniformity is influenced
208 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC
not only by drug solubility in the binder solution, but also by the amounts of granulation
liquid and the type of the binder used.
Granule friability is also influenced by the binder concentration and the amount
of granulating liquid added to the formulation during granulation. For a new API
granulated with lactose and Avicel PH 101 using low-viscosity HPMC (Pharmcoat
603), increasing the binder concentration and the amount of water tended to produce
higher wet-mass consistencies. Granule friability decreased with an increase in the
binder level. An inverse relationship was observed between granule friability and the
amount of water added to the formulation, especially at lower drug concentrations (35).
The binder can influence not only the physical properties of the granules, but
also the performance of the finished tablets. For example, within a certain range of
binder concentrations and granulating liquid, increasing the amount of binder or
granulating liquid actually decreases the compressibility of the granules (26). In
another study, the effect of different types of binders such as PVP K30, HPMC
(Cellulose HP-M 603), maltodextrin (Lycatab DSH), pregelatinized starch (Lycatab
PGS), and low substituted hydroxypropyl cellulose (L-HPC) (type LH 11) on the
hardness of placebo and paracetamol tablets using high-shear granulation technique
was compared by Becker et al. (36). The placebo granules comprised only lactose and
MCC (Avicel PH 102). The binders were added at 2%, 6%, and 10% in the dry form
and water was used as the granulating liquid. The median particle size increased with
increasing binder concentrations from 2% to 6% except when L-HPC was used as the
binder. The median particle size of the granules decreased with increasing binder
concentrations from 2% to 6% when L-HPC was used as the binder. The granule
strength, and the hardness (crushing strength and friability) of the resulting tablets
increased with increasing binder concentrations. In the preparation of model tablets
containing paracetamol, PVP K30 (6%) and Cellulose HP-M 603 (6%) turned out to
be the binders of choice with respect to crushing strength of the finished tablets.
Lycatab PGS, Lycatab DSH, and L-HPCC-LH 11 could not be used to produce
paracetamol tablets that met the requirements.
The amount of granulating water can also affect the physical properties and
compressibility of MCC, which is one of the most commonly used fillers. The
mechanism of forming hard granules with MCC, using a high-shear mixer granulator,
was investigated by Suzuki et al. (37). The hardness of the MCC granules
increased with granulation time and the amount of water added. The specific surface
area decreased during the granulation process. Crystallite size of cellulose decreased
with granulation time and with increasing amounts of water added. MCC granules
were strengthened with longer granulation time and greater amounts of water, thus
resulting in a more intricate network due to the strong shear force of the impeller.
The effect of granulating water level on the physical–mechanical properties of
MCC and silicified microcrystalline cellulose (SMCC) was investigated by Habib
et al. (38). Granules containing either MCC or SMCC were prepared at different
water levels using a high-shear mixer. The resulting granules were tray dried. The
water level ranged from 0% to 100%. Increasing the water level affected the granule
particle size, increased the granular density and flow properties of the granules, and
decreased the porosity and compactibility. The compactibilities of both materials
were similar and acceptable at each granulating water level up to 40%. However,
both the materials showed poor compactibility at higher water levels. The effect of
amount of water used for granulation did not have a statistically significant difference
on the compressibility of MCC and SMCC. SMCC did not offer practical
advantages over MCC, other than better flow in the powder form, which could be
High-Shear Granulation 209
© 2005 by Taylor & Francis Group, LLC
attributed to slightly larger particle size and the presence of silicon dioxide in its
structure.
5.4. Effect of Process Variables
Process variables play a critical role in the granulation process since they influence
how the binder liquid is distributed in the powder bed and the degree of densification
for the powder mixture. The degree of powder densification affects the level of liquid
saturation of the moist agglomerates. Therefore, process variables could influence
properties such as particle size distribution and the drug content uniformity of the
obtained granules. Specifically, the process variables affecting the granulation process
and the physical properties of the obtained granules are:

 Load of the granulator bowl

 Impeller speed

 Granulating solution addition method

 Granulating solution addition rate

 Chopper speed

 Wet-massing time.
The following examples illustrate how varying the aforementioned process
variables influences the properties of the granules obtained using a high-shear wetgranulation
process.
Processing variables such as impeller speed, granulating solution addition rate,
total amount of water added in the granulation step, wet-massing time, etc., were
evaluated using a Plackett–Burman experimental design in a study by Badawy
et al. (39). The results showed that granule compressibility of the lactose-based formulation
is extremely sensitive to the processing conditions. The granule particle size
was increased by increasing the amount of water added, high impeller speed, and
short wet-massing time. The wet-massing time and impeller speed also had a signifi-
cant impact on granulation compressibility. Increasing the impeller speed and wetmassing
time decreased granule porosity and fragmentation propensity, which led
to the decreased hardness of the finished tablets.
The shear effects of the impeller on the properties of granules prepared from
surface treated sericite were investigated by Oulahna et al. (40). Surface treated
sericite was granulated with an alcoholic solution of polyethylene glycol 20,000 using
a high-shear granulator at three different impeller speeds (100, 500, and 1000 rpm).
The properties of the granules produced under different impeller speeds were examined
in terms of porosity, friability, and binder content. The results indicated that
heterogenous granules (in terms of binder concentration) with finer particles and a
wider particle size distribution were obtained at a low impeller speed (100 rpm).
However, at higher impeller speeds, homogenous granules with less fine particles
and a narrower size distribution were observed. Moreover, the porosity of the granules
decreased with increasing impeller speeds. Therefore, mechanical energy
brought to the powder bed by the impeller is as important as the physicochemical
characteristics of the powder–binder pair in affecting the granule properties.
The effect of granulation processing parameters, fill ratios, impeller speed,
chopper speed, and wet-massing time on granule size distribution of placebo formulations
was studied by Bock et al. (41). High fill ratios in the bowl resulted in an
increased proportion of fines in the obtained granules. Increasing the impeller speed
210 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC
and the massing time increased the granule size. The speed of the chopper did not
affect granule size distribution for the formulations tested.
The mode of addition of the liquid binder can affect the characteristics of the
granules (42). When water, used as a binder liquid, was added to the powder mixture
by atomization, granules with a slightly narrower particle size distribution were
obtained.
The effects of massing time on the properties of the granules of hydrophilic
polymer-based controlled-release formulations were studied by Timmins et al.
(43). The formulations consisted of 30% sodium alginate, 10% HPMC, and 50%
diltiazem hydrochloride or verapamil hydrochloride. The increase in massing time
resulted in an increase in the mean granule size of the formulations. This could be
true for all the matrix controlled release formulations, which contain a high concentration
of the hydrophilic polymers.
The effect of reducing drug loading and mixing time on the content uniformity
of a low-dose drug was investigated by Kornchankul et al. (44). Buspirone hydrochloride
was used as a model drug and was mixed with other ingredients in two different
concentrations (0.5% and 5%, w/w) in a T. K. Fielder high-shear mixer at a
high impeller speed (522 rpm) and a high chopper speed (3600 rpm) for up to
32 min. Samples were withdrawn from nine locations in the mixer at specific time
points using a side-sampling thief probe. The optimum time to mix the 0.5% w/w
formulation was 8 min, while it was only 1 min for the 5% w/w formulation.
MCC was granulated with water in a high-shear mixer (37). The hardness of
the MCC granules increased with granulation time and the amount of water added.
The specific surface area was reduced during the process. These findings suggested
that the long-chain structures in MCC were disrupted, resulting in smaller units with
shorter chain lengths due to the strong shear force of the impeller. These smaller
units then formed a network within the granules. Thus, MCC granules were strengthened
with longer granulation time and greater amounts of water, resulting in a
more intricate network. The change in MCC chain length and physical structure
can be experimentally detected using the small-angle x-ray scattering and wide-angle
powder x-ray diffraction methods.
MCC granules were prepared by wet granulation with water in a high-shear
mixer granulator (45). Samples of wet granules were taken at different time points
after addition of water to examine the physical characteristics of the granules using
near-IR spectrometry, thermogravimetry, and isothermal water vapor adsorption.
The results indicated that the degree of interaction between MCC and water increased
with granulation time due to the change of physical structure of MCC during
the granulation process.
5.5. Effect of Granulator Design
A variety of high-shear granulators with varying processing designs are commercially
available. The orientation of the mixing chamber in the granulators is either
vertical or horizontal. The vertical granulators could be bottom or top driven.
Besides the configuration difference, a significant difference exists among the mixers
in terms of the shapes of the mixing bowl and impeller blades. For example, the vertical
top-driven mixer granulators utilize a curved blade design, thus allowing the
tips of the mixing blade to reach the side wall of the mixing bowl, which is also
curved in a shape similar to the blade. On the other hand, the vertical bottom-driven
high-shear granulators (such as Diosna) have a conical mixing bowl with a flat blade
High-Shear Granulation 211
© 2005 by Taylor & Francis Group, LLC
design. The blade is parallel to the lid and the bottom of the bowl, and only a small
portion of the impeller blade directly sweeps off the side wall of the conical mixing
bowl. During mixing and granulation, the conical mixing bowl pushes the powder
toward the center of the bowl for efficient mixing. The differences in impeller blades
and mixing bowl designs of the different commercially available high-shear granulators
are expected to result in different flow patterns and powder flow dynamics in the
bowl.
During the wet-granulation process, the impellers impact the powder mixture
and keep the mixture moving. During this process, binder solution is distributed
and the powder mixture is compacted and densified. Relative swept volume, the
volume swept out per second by impellor and chopper divided by the volume of
the mixer, is the indicator of the work input on the powder material inside the bowl
of the granulator. Changes in the granulator design could affect the relative swept
volume of the granulation process, which in turn could change the degree of densi-
fication of the wet agglomerates. As a result, the optimal amount of granulating
liquid and the physical properties of the resulting granules may be altered. Therefore,
the design of high-shear granulators must be taken into consideration during the wet
granulation process.
Schaefer et al. found that there was a difference between the horizontal (Lo?dige
M5GR) and vertical (Fielder PMAT 25 VG) granulators in terms of granule growth
for DCP (22). The degree of liquid saturation required was higher for the vertical
(Fielder PMAT 25 VG) granulator in order to obtain the same granule size. Agglomeration
of DCP began at lower degrees of liquid saturation in the horizontal (Lo?dige
M5GR) granulator. This could be because the rolling motion of granules in the horizontal
(Lo?dige M5GR) granulator facilitated agglomeration, whereas stronger
mechanical forces in the vertical (Fielder PMAT 25 VG) granulator promoted the
breakage of granules. A similar trend was observed for lactose granulation (46) as
well.
The effects of impeller and chopper design upon granule growth of DCP in
a laboratory high-speed mixer (Fielder PMAT 25) were investigated by Holm (47).
Three different changeable impeller blades with the same surface area were constructed.
The angles of inclination were kept at 30, 40, and 50. The granules with
low porosity were obtained at high impeller speed for the inclination angles at 40
and 50. The effects of the impeller design with respect to the blade inclination
and impeller rotation speed can be explained in terms of the volume of powder mixture
swept out by the impeller. The relative swept volumes for the impellers at three
different inclination angles were 2.2, 2.82, and 3.36 at 400 rpm. A high-swept volume
causes high densification of the agglomerates and narrow granule size distributions.
Chopper size and rotation speed had no effect on the granule size distribution
because the primary function of the chopper is to disturb the uniform flow pattern
of the mass.
6. GRANULATION END-POINT DETERMINATION
The granule properties for a given formulation are a function of the process parameters
such as the impeller speed, amount of granulating fluid, and granulation time
(wet-massing time). Therefore, the time to end the granulation process during a highshear
granulation process becomes critical. The properties of the granules produced
determine, in part, the ultimate quality and performance of the finished dosage
212 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC
forms. Thus, the determination of the granulation end point becomes necessary. The
ultimate goal of end-point determination in a granulation process is to obtain an
indication of formation of granules with the desired physical properties, such as
acceptable mean particle size range and porosity. The measurements of changes during
the granulation process are employed for end-point determination. The advantages
of using an appropriate method to determine the granulation end point are
listed below:

 Process optimization
– Evaluate raw material
– Determine optimal end point

 Batch reproducibility
– Use end point to achieve batch to batch consistency
– Document adherence to batch protocol

 Process trouble shooting
– Detect mechanical problems
– Identify mixing irregularities
Several different approaches have been explored for granulation end-point
determination. These approaches can be classified into two major categories: indirect
measurements and direct measurements. In the indirect measurements, the electrical
and mechanical parameters of the granulator motor are monitored since the changes
of these parameters are related to the changes in the consistency of the powder mixture
in the wet granulation process. Faure’s study (48) has confirmed that there is a
close relationship between the wet-mass consistency/viscosity of samples prepared in
a mixer granulator and physical properties of the dry granules produced from the
wet mass. The physical properties of the dry granules include granule size distribution,
bulk density, friability, and flow characteristics. Variations in the formulation
and process affect the relationships between the wet-mass consistency and drygranule
properties, and the net power consumption of the mixer granulator.
In the direct measurements, the physicochemical properties of the powder mixture
are monitored during the wet-granulation process. These properties could be
mass conductivity and granule size.
6.1. Indirect Measurements
In the indirect measurement, both electrical and mechanical characteristics of the
motor are monitored to control the granulation end point. The electrical characteristics
of the motor are motor current and power consumption. The mechanical characteristics
of the motor are torque and tachometry.
The power consumption of a granulator motor is related to the resistance of
the mass mixture to the granulator blades, which varies with the consistency of
the powder mixture. Power consumption used as granulation end point control
has been related to the level of liquid saturation of the moist agglomerates (49), densification
of wet mass (30,50,51), and granule growth (52). Leuenberger proposed
that the liquid amount required for granulation corresponds to the plateau in the
power consumption record profile (53).
High-Shear Granulation 213
© 2005 by Taylor & Francis Group, LLC
High-shear granulators capable of monitoring the granulation end point using
power consumption are commercially available. Figure 12 shows a typical power
consumption profile obtained from a commercially available high-shear granulator
with an end-point determination (53). It is evident that the profile consists of five
phases during the wet granulation process. In Phase I, the powder mixture is moistened
by adding the granulation liquid. Since the formation of liquid bridges is not
observed between primary particles, power consumption does not increase during
this granulation phase. In Phase II, the liquid bridges begin to form among the primary
particles, thus resulting in granule formation. The power consumption increases
dramatically in this phase. In Phase III, as more granulation liquid is added, the
interparticular voids are filled by the granulation liquid and coarser granules are
formed. The power consumption remains relatively constant in this phase. In Phase
IV, liquid saturation of the powder bed is reached, and more coarse granules are
formed. The power consumption increases in this phase. In Phase V, as more granulation
liquid is added, a suspension is formed. The power consumption starts
declining in this phase. It has been noticed that only the granules formed during
Phase III are usable for further dosage form development. Thus, the power consumption
profile can be used to monitor the granulation process and to determine
the granulation end point.
The drawback of monitoring power consumption is that the signal is affected
by a number of factors, such as product (formulation), equipment, or process variables.
For example, the densification properties of the starting materials, type and
amount of binder used, addition rate of binder solution, impeller speed, etc., influence
the power consumption profile. Wear and tear of the granulator could also
affect the power consumption signal. Thus, the power consumption profile for a
granulation process is formulation and process specific.
The end-point control of granulation by power consumption measurement in a
25 L high-shear mixer was investigated by Holm et al. (42). The effect of impeller
design, impeller speed, liquid addition rate, type of binder, and mixing ratios between
lactose and starch on the correlation between power consumption and granule
Figure 12 A typical power consumption profile obtained from a commercially available high
shear granulator. (From Ref. 53.)
214 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC
growth was investigated in the liquid addition phase of the process using a fractional
factorial experimental design. A linear correlation between power consumption and
mean granule size was observed. The correlation was dependent on the impeller
design, the impeller speed, and the type of binder. However, a granulation end-point
control, based on power consumption was not found to be sensitive to variations in
the lactose:starch ratio. An end-point control based on the peak detection method
was not generally applicable, because a peak in the differentiated power consumption
signal could not be identified in all the experiments.
Monitoring of power consumption was also explored for automating the wetgranulation
process (54). A vertical high-shear mixer (300 L Collette Gral) was
instrumented with a control unit, which controls the continuous addition of granulating
liquid in relation to the consistency of the powder mass by power consumption
measurement. The instrumentation comprised an important approach for automated
process control, and on-line documentation of critical process parameters with
respect to validation. Similarly, a signal analysis system with memory-programmable
automatic process control for determining power consumption in a high-shear mixer
was introduced by Laicher et al. (55). The signal analysis system made it possible to
perform gliding mean value calculations of the measuring signal and to evaluate the
signal fluctuations. The granulation end point was determined by curve analysis.
Using this system, granules with a reproducible particle size distribution were obtained
from active ingredient paracetamol or ibuprofen/excipient mixtures, which
either showed a tendency to form lumps, in the case of overwetting (paracetamol),
or required a short granulation time due to low melting point (ibuprofen).
Since power consumption is related to both current and voltage of the motor, it
would be logical to use motor current for granulation end-point control. However,
the motors used in high-shear granulators are induction motors, which use alternating
current that lags behind the voltage. Therefore, it is inappropriate to determine
granulation end point by monitoring motor current or the simple combined product
of current and voltage (51). The power consumption of induction motors can be
expressed by the following equation:
Power ? ffiffiffi 3 p IV cos y ?6:1?
where I is the current, V is the voltage, and y is the phase difference between current
and voltage.
An alternative to power consumption is torque measurement. A high-shear
mixer was instrumented with a new capacitive sensor, a watt meter, and a straingauge
torque sensor (50). Placebo formulations containing DCP, MCC 101, MCC
102, and lactose were granulated in the high-shear granulator. A similar map (profile
of torque measurement or power consumption) of the granulation process was
obtained for power consumption and torque measurement.
In another study (51), a high-shear vertical mixer/granulator was instrumented
to monitor power consumption, direct torque (signal from impeller shaft), and reactive
torque (signal from the motor). Of the three methods used, direct torque generated
the most descriptive profile for the granulation process. Granules produced for
the test formulations, based on the end points determined using direct torque measurement,
resulted in tablets with acceptable hardness.Measurement of direct torque
could also detail the granulation process of an overwetted formulation.
Torque profiles of granulation of MCC or lactose in a high-shear granulator
with various operating conditions were studied by Kornchankul et al. (56). The
High-Shear Granulation 215
© 2005 by Taylor & Francis Group, LLC
torque measurement for each batch of granules obtained was correlated with the corresponding
tablet characteristics. The scatter plots of tablet hardness indicated that
the hardness of tablets prepared with MCC granules was >7 Kp, as long as the end
torque was <100 lb in., granulation rate was <0.4 lb in./sec, and granulation extent
was <13,000 lb in. sec.
The relevance of end-point determination based on torque measurement in
controlling performance characteristics of compressed tablets such as hardness, friability,
disintegration time, and dissolution time was determined by Achanta et al.
(57). The measurement of torque was used to determine the end point of the granulation
process in a high-shear granulator. With increasing end-point levels, tablets
compressed with the corresponding granules were found to be harder.
6.2. Direct Measurements
During the high-shear wet-granulation process, the binder liquid is dispersed in the
powder mixture. At this point, the wetted powder becomes agglomerated and densi-
fied. Thus, the physical properties (particle size, density, consistency) of the powder
mixture change with time during the granulation process. This change in the properties
of the powder mixture can be used as an indication of the end point of the granulation
process.
6.3. Consistency Measurement Using a Probe
One approach to determine the granulation end point is based on monitoring the
change of the consistency or strength of the wet mass during the granulation process.
This approach was first developed by the Boots Company for Diosna high-shear
mixers (18). The Disona-Boots probe is designed to detect changes in the momentum
of granules, in a constant velocity region of the mass, with the use of strain gauges
mounted on the probe. To avoid bias due to random events (e.g., production of large
lumps due to inhomogenous solution distribution), signal pulse heights, sampling
times, and pulse density are all considered to attain the final signal. The signal is normalized
over a calibrated range of forces from 0% to 100%. The criterion for endpoint
indication is that the signal obtained during the granulation process must be
greater than the preset value. When properly developed and calibrated for a given
high-shear granulator, the probe can be successfully used to determine a repeatable
granulation end point, i.e., granules with a similar particle size and density are
obtained. However, the signal obtained from the probe is granulator and formulation
dependent.
6.4. Acoustic Emission
Acoustic emission monitoring detects and analyzes the sound produced by a process
or system. During the wet-granulation process, particle size, flow properties of the
powder mixture, and degree of powder densification change with time because of
the addition of binder and formation of agglomerates. The change of physical properties
of the powder mixture could affect the acoustic emission signal. Thus, this
technique can be used to map the granulation process and determine its end point.
The technique is noninvasive, sensitive, and relatively inexpensive.
The acoustic emissions during the wet granulation of a model formulation containing
lactose and MCC were monitored using acoustic emission sensors (58). The
216 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC
sensors were attached to the outside of the high-shear mixer bowl. Undergranulated
to overgranulated granules were prepared by varying the quantity of the binder solution.
Average signal level increased with increasing amount of the added binder solution
and wet-massing time. A strong correlation between the acoustic emissions and
the physical properties of the granules at the end of the granulation process was
demonstrated. This technique was capable of monitoring changes in physical properties
such as particle size, flow properties, and compression properties of powder
material during granulation.
6.5. Image Analysis
In the direct measurement techniques, one of the successful methods is monitoring
the granulation processing based on the particle size of the granules.
A novel system to continuously monitor granule growth in a high-shear
granulation has been developed by Watano et al. (59). The system consists of an
image processing system and a particle image probe comprising a CCD camera,
lighting unit, and air purge system. High-shear granulation was conducted using
pharmaceutical powders, and granule size and product’s yield of various size ranges
were continuously measured by the developed system. Sieve analysis of the granules
sampled out during the granulation was simultaneously conducted, and the data
were compared with that of the on-line image processing system. A close relationship
was found between both the data, thus proving that the system could monitor the
granule growth accurately and continuously throughout.
Such an image processing system was further developed to automatically control
the wet-granulation process by Watano et al. (60). Besides the image probe, a
fuzzy control system using a linguistic algorithm employing ‘‘if-then’’ rules with a
process lag element was developed to control granule growth accurately. The system
consisted of an image processing and a fuzzy control system. An image probe continuously
monitored granule images through the sidewall of the vessel. The images
were digitized by the image processing system, and granule median diameter and
shape factor were calculated automatically. The difference between the desired and
measured granule diameters was used as an input for fuzzy reasoning. The result of
fuzzy reasoning was used to control the output power of the liquid feed pump. The
system could control granule growth with high accuracy, regardless of changes in the
operating conditions.
7. FORMULATION DEVELOPMENT (OPTIMIZATION)
The pharmaceutical properties of the granules, which can significantly affect the performance
and properties of the finished dosage forms are controlled by multiple
interrelated variables such as formulation, process, and granulator parameters. In
most cases, the nature and characteristics of the excipients and API used in the formulation
are fixed. Thus, only the amount of granulating liquid and granulation process
variables need to be optimized. ‘‘Design of experiment (DOE)’’ is often used to
optimize the granulation process. In experimental designs, a number of cause factors
which comprise volume of binder solution, impeller speed, massing time, etc., and
response variables (dependent variables), which comprise mean particle size, tablet
hardness, time for 50% of drug released, etc., are selected. The relationship between
High-Shear Granulation 217
© 2005 by Taylor & Francis Group, LLC
cause variables and response variables is established mathematically. The optimal
response variables can then be obtained by selecting the optimal cause variables.
A Box-Behnken design was used to optimize the composition and process factors
for formulations containing mannitol and MCC, which were granulated in a
high-shear mixer and compressed into tablets (31). Simultaneous optimization of
crushing strength, disintegration time, and ejection force of the tablets was carried
out to find optimal regions in the design space for these tablet properties. The composition
of the tablet mixture and the process of tablet manufacturing were optimized
using such statistical techniques.
A central composite experimental design was used to optimize a granulation
process in a high-shear mixer for a hydrophilic matrix tablet formulation (61).
The parameters tested were the amount of water in the hydroalcoholic granulation
liquid, the amount of granulation liquid, and the massing time. The results indicated
that the amount of granulation liquid was the most important parameter, followed
by the amount of water in the granulation liquid. The influence of the massing time
was negligible. A formulation with granule friability below 20% was obtained under
optimal processing conditions.
The wet-granulation process by a high-speed mixer granulator for two model
drugs, ascorbic acid and ethenzamide, was optimized through experimental design
(spherical central composite experimental design) and simultaneous optimization
methodology (34). The amount of water (granulation liquid) and HPC (binder) were
simultaneously optimized with regard to the four pharmaceutical properties, including
yield, drug content uniformity, geometrical mean diameter of granules, and
uniformity of granule size. A simultaneous optimal point incorporating four pharmaceutical
properties was obtained using the generalized distance function. The
experimental values of the four response variables obtained in newly prepared granules
were found to correspond well with the predicted values of granules containing
both ascorbic acid and ethenzamide.
Various experimental designs have been explored by Vojnovic et al. to optimize
wet-granulation process variables. Some of these examples are described below.
Wet granulation of lactose and corn starch in a 10 L high-shear mixer was optimized
using a response surface design (62). The effect of the process factors, impeller
speed, amount of water added, and granulation time on the properties of the granules
were investigated. Moisture level had a major effect on geometric mean diameter
and flow rate of the granules. The impeller speed markedly influenced the geometric
mean diameter, geometric standard deviation, compactibility index, and percentage
of granules smaller than 1250 mm. The granulation time affected the compactibility
index of the granules. Theoretical optimum conditions were obtained for the following
five response variables: geometric mean diameter, geometric standard deviation,
flow rate of the granules, percentage of granules smaller than 1250 mm, and compactibility
index. These results were comparable to the experimental results.
A central composite response surface design was employed to optimize a wetgranulation
process for lactose and corn starch in a 10 L high-shear mixer (63). The
effect of the optimal combination of three independent variables (moisture level,
impeller speed, and granulation time) on four properties (geometric mean diameter,
percentage of particles smaller than 1250 mm, geometric standard deviation, and flow
rate) of the granules was investigated by the simultaneous optimization method. The
optimum zone determined in 10 L high-shear mixer was analyzed for scale-up. The
same product was manufactured at a 50 L scale and optimum theoretical results were
218 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC
obtained for the four response variables. These results were in agreement with the
experimental data.
Optimization of wet granulation of a placebo formulation in a 10 L high-shear
mixer was carried out using an experimental design, which combined the Scheffe
simplex-centroid design and Doehlert experimental matrix (64). Lactose, corn starch,
and MCC were used as excipients. Mixing ratios of these excipients were selected as
formulation factors. The impeller speed and granulation time were chosen as independent
variables. A two-phase experimental strategy was employed. The optimal
composition of the three excipients was determined in the first phase and the optimal
process conditions to obtain a geometric mean diameter >150 mm was determined in
the second phase.
The optimization of wet granulation in a 10 L high-shear mixer was investigated
using a nonclassical mixture experimental design, which combines mixture
experimental design with the ponderation function (65). HPMC, lactose, corn starch,
and MCC were used as excipients, while PVP was used as a binder. The introduction
of such a modification in the experimental design allowed selection of an experimental
matrix, which could provide enough information with a relatively low number of
experiments and at a minimal cost.
8. PROCESS SCALE-UP
The formulation and process for a new API are normally optimized using small-scale
equipment, owing to a limited amount of API. The granulators used during the
development stage are laboratory scale models. The formulation is generally fixed
during the clinical trial studies. However, the production batches are manufactured
with large-scale equipment. For a solid dosage form prepared using the wet-granulation
process in a high-shear granulator, product scale-up could be challenging. This
is because of significant differences between laboratory and production models of
high-shear granulators in terms of the design, shape, size, and geometry of the impellers.
Even though the composition of the formulation is similar, the physical properties
of the obtained granules could be dramatically different because of the
differences between laboratory scale and production scale granulators. These differences
in the physical properties of the granules could significantly affect the properties
and performance of the final tablets.
The ultimate goal of product scale-up is to maintain similar processing conditions
for the production batches as those used at the laboratory scale so that the
properties of the granules produced after scale-up would not change significantly.
Therefore, it is imperative to identify and monitor key process parameters during
the product development stage, so that the tablets produced from the large-scale
granulation batch perform similarly to those produced from the small-scale batch.
The liquid saturation level of the agglomerates during a high-shear wetgranulation
process plays an essential role in terms of granule growth, as discussed
previously. It is desirable to maintain a constant level of liquid saturation during the
scale-up. From Eq. 4.1, liquid saturation level is controlled by both moisture
content and intragranular porosity. The moisture content is determined by the
amount of applied granulation liquid. The intragranular porosity is affected by
the impaction caused by the impeller and wet-massing time. Due to the different geometric
shapes and sizes of the granulator and impeller, and fill ratio in the mixing
bowl, maintaining constant liquid saturation level could be problematic.
High-Shear Granulation 219
© 2005 by Taylor & Francis Group, LLC
8.1. Linear Scale-up
As described earlier, the process and formulation factors that affect the granulation
process and physical properties of the obtained granules are: rate of addition of the
binder solution, impeller speed, wet-massing time, and the amount of binder or granulating
liquid used. Ideally, the amount of binder and granulation liquid will be
scaled up by linearly increasing the amount of water with the batch size. If the time
of binder solution addition is kept constant by increasing the rate of binder addition
during scale-up, then the variables that need to be controlled narrow down to impeller
speed and wet-massing time. Different approaches have been proposed to control
impeller speed. These approaches include maintaining a constant impeller tip speed,
maintaining a constant relative swept volume, or maintaining a constant Froude
number (dimensionless number) during the scale-up process. The wet-massing time
can be determined by the granulation end-point determination such as power consumption.
However, due to the difference in the size and shape of the mixing bowl
and impeller between laboratory and production scale granulators, linear scale-up
may not work.
A constant relative swept volume and a constant impeller tip speed during the
granulation of lactose in three Gral high-shear mixers (Gral 10, 75, and 300 L), did
not result in a comparable process for different sizes of granulators (66). Thus, even
within the same type of high-shear granulators, compensatory changes in the volume
of binder fluid, impeller speed, and wet-massing time to produce equivalent granules
during scale-up may be required.
The ease of scale up could be formulation dependent, and some formulations
are not affected by the batch size. If the formulation is easy to densify, then the speed
of the impeller may not affect the properties of the obtained granules. In one study
(67), alpha-lactose monohydrate was granulated using a PVP K 30 (2.5% on dry
material) binder solution. The wet granulation process was optimized in a Gral
10 L granulator using a two-factor (water content and impeller tip speed), three-level
experimental design. The granulation process was then scaled down from a Collette
Gral 10 L (8 L batch size) to a serial minigranulator (Pro-C-epT Mi-Pro) with different
bowl volumes of 5 L, and 1900, 900, and 250 mL. In all mixer volumes, the impeller
tip speed range used did not influence the granule or tablet properties. In all bowl
volumes, the influence of water concentration on actual yield, particle size distribution,
and granule friability was similar.
In another study, the formulation containing 5% API, 47.35 % of MCC, and
47.35% of pregel starch was granulated using different sizes of high-shear granulator
ranging from 2 to 25 L (68). A 23 factorial design was used to investigate the effects
of scale-up variables including amount of binder solution, wet-massing time, and
impeller tip speed on the physical properties of the obtained granules. It was shown
that the characteristics of the granulations made under different conditions were
highly reproducible. The excipient system of MCC and pregel starch produced a
robust formulation that was resistant to changes during the scaling-up process in
the high-shear mixers.
Unfortunately, most formulations are not scalable. For example, a less homogenous
liquid distribution, a wider granule size distribution, and a higher intragranular
porosity of the obtained granules were found for the scale-up batches when
DCP was granulated using different sizes of high-shear granulator. This could be
due to the decreased swept volume in the larger granulator during scale-up (18).
220 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC
In another study, a model formulation was developed in a small-scale highshear
granulator (Diosna P 1-6) (41). It was easy to scale up to the P 10 granulator
without changing the formulation or the granulation conditions. However, further
scale-up to Diosna P 25 and P 100 granulators, which are large-scale machines,
resulted in granules which were smaller than those prepared in the laboratory scale
equipment.
To achieve linear scale-up for formulations that are sensitive to the changes
during the scale up, the parameters used during the laboratory scale granulation
process have to be similar to those used during the production scale process. Two
processes may be considered similar if there is a geometrical, kinematic, and dynamic
similarity (69). Two systems are considered to be geometrically similar if they have
the same ratio of linear dimensions. Two geometrically similar systems are kinematically
similar if they have the same ratio of velocities between corresponding points.
Two kinematically similar systems are dynamically similar when they have the same
ratio of forces between corresponding points. For any two dynamically similar systems,
all the dimensionless numbers necessary to describe the process are the same
(70). Dimensionless number of the process is scale independent, and thus can be
employed for scale-up of granulation.
Dimensional analytical procedure was first introduced by Lord Rayleigh in
1915 (71). It is a process whereby the variables pertinent to a physical problem are
systematically organized into dimensionless groups/numbers. Assuming the absence
of chemical reaction and heat transfer, the following influencing and independent
variables are pertinent to the wet-granulation process (72):
DP: net impeller power consumption (motor power consumption minus the
dry blending baseline level)
R: impeller radius
N: impeller speed
h: height of granule bed in the bowl
r: granule bulk or specific density
Z: granule dynamic viscosity
The only three fundamental quantities are mass (M), length (L), and time (T).
Buckingham theorem (p-theorem) (73) states that when there are i physical
variables and dimensional constants and j fundamental qualities (such as mass,
length, and time), the number of dimensionless groups/numbers in a complete set
is equal i  j.
Therefore, according to the Buckingham theorem, there are four dimensionless
numbers for a high-shear granulator.
Newton (power) number (Np) relates the drag force acting on a unit area of the
impeller to the inertial stress:
Np ?
DP
rN3R5 ?8:1?
Reynolds number (Re) relates the inertial force to the viscous force:
Re ?
rNR2
Z ?8:2?
Froude number (Fr) is the ratio of the centrifugal acceleration to the gravitation
constant (g). It can be used as a criterion for dynamic similarity of granulators:
High-Shear Granulation 221
© 2005 by Taylor & Francis Group, LLC
Fr ?
RN2
g ?8:3?
Geometric number (ratio of characteristic lengths) h/R is related to fill level in
the bowl.
An ideal scenario is that the granulation processes in laboratory scale and production
scale granulators are geometrically, kinematically, and dynamically similar.
However, the shape and size of granulators currently available vary from manufacturer
to manufacturer. Therefore, it is difficult for the granulation processes to be
geometrically, kinematically, and dynamically similar. For example, Collette Gral
10, 75, and 300 are not geometrically similar (66).
One practical approach during scale-up is to keep the processes dynamically
similar, which is defined by the same dimensionless numbers for the two different
sizes of granulators. Therefore, dimensionless numbers such as Froude number
can be used for scale-up.
Froude number was suggested as a criterion for dynamic similarity and a
scale-up parameter during wet granulation (66). Lactose granules prepared in Gral
high-shear mixers of three different capacities (Gral 10, 75, and 300 L) were
compared. The granulation process regarding temperature increase and particle size
distribution could be scaled up by keeping the Froude number constant. It therefore
seems appropriate to characterize and compare different mixers by the range of the
Froude numbers. A matching range of the Froude numbers would indicate the
possibility of scaling up the batch even for the mixers such as Collette Gral 10,
75, and 300 that are not geometrically similar. Hence, systematic consideration is
needed to select laboratory and high-production scale granulators to develop a solid
dosage form, which requires high-shear granulation in the manufacturing process.
The Froude numbers of each granulator can be calculated based on Eq. 8.3. If
the Froude numbers for the laboratory scale granulator cannot be matched, linear
scale-up is unlikely.
Dimensionless power relationship between power number and the product of
Froude number, Reynolds number, and fill ratio could be another approach for
scale-up (74). A model formulation granulated in different conditions (batch size,
blade speed) with different sizes of high-shear granulators, but presenting similar
wet-mass characteristics (bulk density and consistency) led to dry granules of similar
properties such as granule size distribution, density, friability, and flow. The wet
masses were characterized by the bulk density and consistency, as measured by mixer
torque rheometry. The power consumption of the granulator was monitored with a
power meter. The power number, Reynolds number, Froude number, and bowl fill
ratio were calculated for each running condition. The relationship between the
power number and the product of Reynolds number, Froude number, and bowl fill
ratio can be established by plotting the decimal logarithms of power number vs. the
decimal logarithms of the product of Reynolds number, Froude number, and bowl
fill ratio. The results showed that under certain conditions, a common scale-up master
curve could be drawn from the data gathered for each bowl, and such a curve
could be used for the determination of power consumption of a mixer granulator
at a defined granulation end point for scale-up.
Dimensionless numbers such as Froude number can also be used to design and
select high-shear granulators of different scales. It is therefore necessary to characterize
and compare high-shear granulators of different types and sizes by the range of
222 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC
the Froude numbers. A matching range of the Froude numbers would indicate the
possibility of a scale-up, even for the mixers that are not geometrically similar
(66). Froude number of a granulator is determined by both impeller speed and the
diameter of the impeller as shown in Eq. 8.3. When the diameter of the impeller is
fixed, the Froude number of a granulator can still be varied by adjusting the impeller
speed. Therefore, it would be desirable for the production scale granulator to be
equipped with variable speeds. Hence, the range of Froude numbers for each size
of granulator needs to be considered in the design of the high-shear granulators.
A systematic approach should be followed to select laboratory scale and production
scale granulators.
The matching Froude numbers for different size granulators provide the possibility
of linear scale-up. However, some precautions are still needed for the geometric
dissimilarity of mixing bowl vessels, and the shape of the impellers in the
granulators of different types and scales.
During scale-up, after selecting the granulator with matching Froude number
and similarly shaped impeller blades, the granulation end point can be determined by
any one of the techniques described earlier. A new approach for scale-up based on
the use of power consumption, as the granulation end-point determination, is shown
below.
Integration of the impeller power (watts) vs. time (seconds) profile for highshear
granulator was also explored for scale-up of high-shear granulation (75).
The energy parameter impeller work during the wet-granulation process of a model
MCC formulation in 5 or 10 L was recorded and normalized, based on the weight of
the dry powder mixture. The relationship between work of granulation and cohesion
indexes (slopes of the tablet breaking strength vs. compression force profiles), or
granulation size distributions was evaluated. Independent of granulator make or
model, the impeller work measured during granulation correlated quantitatively with
changes in the granulation bulk and tapped densities, average particle size of the finished
powders, and cohesion index. Granulation end points were accurately predicted
for 25, 65, and 150 L high-shear granulators on the basis of development
work on 5.0 and 10 L equipment using impeller work values normalized for the
weight of dry powders in the granulator (W sec/g).
8.2. Nonlinear Scale-up
If linear scale-up cannot be achieved, the amount of binder solution, the impeller
speed, and the wet-massing time should be optimized. For example, Kristensen
attempted to compensate the less efficient mixing from large-scale granulation by
increasing the relative amount (ratio between granulation liquid and starting materials)
of granulation liquid used (76). Rekhi et al. recommended maintaining a constant
tip speed, scaling up the granulation liquid linearly, and adjusting the
granulation time based on the ratio of impeller speeds between the granulators used
in both the scales (77).
8.3. Utilization of Optimization Techniques
Since scale-up of high-shear granulation is still a ‘‘trial and error’’ process due to the
complexity of the granulation process and lacks standards for granulators of different
types and sizes, the DOE approach can be useful during production. The process
variables for the large-scale production can be optimized by using the DOE
High-Shear Granulation 223
© 2005 by Taylor & Francis Group, LLC
technique with relatively less number of trials. Alternatively, the granulation process
can be optimized in such a way that the process is scale independent.
A spherical central composite experimental design was used to design experiments
(78). Granules were prepared in the two batches (2 and 5 kg batches) using a
high-speed mixer granulator. Except for the rotation speed of the impeller, the
operation conditions such as chopper speed, mixing time, and granulation time were
kept constant during the granulation processes. The concentration of ethanol in the
aqueous binder solution and the total volume of binder solution used were selected
as independent variables. The yield of granules, geometric mean particle size, and
geometric standard deviation were employed as dependent variables. A computerized
optimizing technique based on a response surface methodology was developed.
A universal optimal formulation unaffected by manufacturing scale could be
obtained by minimizing the integrated optimization function. The optimized characteristics
measured at the production scale coincided well with those obtained at
laboratory scale, suggesting that this approach could be useful in minimizing the
scale-up problems.
9. CONCLUSION
High-shear granulation is one of the most important unit operations in the production
of tablets. Various innovative approaches have been explored to simplify and
control the granulation process, and improve the quality of the produced granules.
Some of these novel approaches include using a one-pot granulation system, moisture-
activated dry granulation and hot-melt granulation techniques, and different
granulation end-point techniques. However, both the process and formulation parameters
for each formulation still need to be individually optimized due to the complexity
of the number of variables involved during the granulation process and the
uniqueness of the formulation. Moreover, the relationship between the properties
of the finished tablets and unit operations for tableting, such as wet granulation, drying,
milling, and tablet compression, is inconclusive. Therefore, a systematic evaluation
of all formulation and process variables involved in the manufacturing of tablets
is required for the optimization of production of tablets. Future advancement in the
equipment and granulation techniques could further improve the granulation process,
thus resulting in better quality of the granules.
REFERENCES
1. Nellore RV, Rekhi GS, Hussain AS, Tillman LG, Augsburger LL. Development of
metoprolol tartrate extended-release matrix tablet formulations for regulatory policy
consideration. J Control Release 1998; 50:247–56.
2. Qiu Y, Hui HW, Cheskin H. Formulation development of sustained-release hydrophilic
matrix tablets of zileuton. Pharm Dev Technol 1997; 2:197–204.
3. Visavarungroj N, Remon JP. Crosslinked starch as a binding agent. II. Granulation in
a high shear mixer. Int J Pharm 1990; 65:43–48.
4. van den Dries K, de Vegt OM, Girard V, Vromans H. Granule breakage phenomena in
a high shear mixer; influence of process and formulation variables and consequences on
granule homogeneity. Powder Technol 2003; 133:228–236.
224 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC
5. Kiekens F, Cordoba-Diaz M, Remon JP. Influence of chopper and mixer speeds and
microwave power level during the high-shear granulation process on the final granule
characteristics. Drug Dev Ind Pharm 1999; 25:1289–1293.
6. Christensen LH, Johansen HE, Schaefer T. Moisture-activated dry granulation in a high
shear mixer. Drug Dev Ind Pharm 1994; 20:2195–2213.
7. Schaefer T, Holm P, Kristensen HG. Melt granulation in a laboratory scale high shear
mixer. Drug Dev Ind Pharm 1990; 16:1249–1277.
8. Schaefer T, Holm P, Kristensen HG. Melt pelletization in a high shear mixer. II. Power
consumption and granule growth. Acta Pharm Nordica 1992; 4:141–148.
9. Schaefer T, Holm P, Kristensen HG. Melt pelletization in a high shear mixer. I. Effects of
process variables and binder. Acta Pharm Nordica 1992; 4:133–140.
10. Schaeferand T, Mathiesen C. Melt pelletization in a high shear mixer. IX. Effects of binder
particle size. Int J Pharm 1996; 139:139–148.
11. Schaeferand T, Mathiesen C. Melt pelletization in a high shear mixer. VIII. Effects of
binder viscosity. Int J Pharm 1996; 139:125–138.
12. Schaefer T, Taagegaard B, Thomsen LJ, Kristensen HG. Melt pelletization in a high
shear mixer. V. Effects of apparatus variables. Eur J Pharm Sci 1993; 1:133–141.
13. Schaefer T, Taagegaard B, Thomsen LJ, Kristensen HG. Melt pelletization in a high
shear mixer. IV. Effects of process variables in a laboratory scale mixer. Eur J Pharm
Sci 1993; 1:125–131.
14. Plank R, Diehl B, Grinstead H, Zega J. Quantifying liquid coverage and powder flux in
high-shear granulators. Powder Technol 2003; 134:223–234.
15. van den Dries K, Vromans H. Relationship between inhomogeneity phenomena and
granule growth mechanisms in a high-shear mixer. Int J Pharm 2002; 247:167–177.
16. Kristensen HG. Agglomeration of powders. Acta Pharm Suecica 1988; 25:187–204.
17. Kristensen HG, Holm P, Jaegerskou A, Schaefer T. Granulation in high speed mixers.
Part 4: effect of liquid saturation on the agglomeration. Pharm Ind 1984; 46:763–767.
18. Kristensenand HG, Schaefer T. Granulation. A review on pharmaceutical wet-granulation.
Drug Dev Ind Pharm 1987; 13:803–872.
19. Yajima T, Itai S, Hayashi H, Takayama K, Nagi T. Optimization of size distribution of
granules for tablet compression. Chem Pharm Bull (Tokyo) 1996; 44:1056–1060.
20. Wikbergand M, Alderborn G. Compression characteristics of granulated materials. IV.
The effect of granule porosity on the fragmentation propensity and the compactibility
of some granulations. Int J Pharm 1991; 69:239–253.
21. Schaefer T, Holm P, Kristensen HG. Wet granulation in a laboratory scale high-shear
mixer. Pharm Ind 1990; 52:1147–1153.
22. Schaefer T, Holm P, Kristensen HG. Comparison between granule growth in a horizontal
and a vertical high speed mixer. I. Granulation of dicalcium phosphate. Arch Pharm
Chem, Scientific Edition 1986; 14:1–16.
23. Vromans H, Poels-Janssen HG, Egermann H. Effects of high-shear granulation on granulate
homogeneity. Pharm Dev Technol 1999; 4:297–303.
24. Juppo AM, Yliruusi J, Kervinen L, Strom P. Determination of size distribution of
lactose, glucose, and mannitol granules by sieve analysis and laser diffractometry. Int
J Pharm 1992; 88:141–149.
25. Suzuki T, Watanabe K, Kikkawa S, Nakagami H. Effect of crystallinity of microcrystalline
cellulose on granulation in high-shear mixer. Chem Pharm Bull (Tokyo) 1994;
42:2315–2319.
26. Chowhan ZT. Aspects of granulation scale-up in high shear mixers. Pharm Technol 1998;
12:26–44.
27. Badawy SIF, Lee TJ, Menning MM. Effect of drug substance particle size on the characteristics
of granulation manufactured in a high-shear mixer. AAPS Pharm Sci Tech 2000:1.
28. Vertommen J, Michoel A, Rombaut P, Kinget R. Production of pseudoephedrine hydrochloride
pellets in a high-shear mixer granulator. Eur J Pharm Biopharm 1994; 40:32–35.
High-Shear Granulation 225
© 2005 by Taylor & Francis Group, LLC
29. Ritala M, Jungersen O, Holm P, Schaefer T, Kristensen HG. A comparison between binders
in the wet phase of granulation in a high shear mixer. Drug Dev Ind Pharm 1986;
12:1685–1700.
30. Ritala M, Holm P, Schaefer T, Kristensen HG. Influence of liquid bonding strength on
power consumption during granulation in a high shear mixer. Drug Dev Ind Pharm 1988;
14:1041–1060.
31. Westerhuis JA, de Haan P, Zwinkels J, Jansen WT, Coenegracht PJM, Lerk CF. Optimization
of the composition and production of mannitol/microcrystalline cellulose
tablets. Int J Pharm 1996; 143:151–162.
32. Makinoand T, Kitamori N. Effect of starch paste concentration on particle size distribution
of fine granules produced by agitation granulation. Chem Pharm Bull (Tokyo) 1995;
43:1231–1233.
33. Kato Y, Moroshima K, Hashizume M, Ando H, Furukawa M. Further observation of
content uniformity of d-alpha-tocopheryl acetate as an oily drug in granules obtained
by wet granulation with a high-shear mixer. Drug Dev Ind Pharm 2001; 27:781–787.
34. Miyamoto Y, Ryu A, Sugawara S, Miyajima M, Ogawa S, Matsui M, Takayama K,
Nagai T. Simultaneous optimization of wet granulation process involving factor of drug
content dependency on granule size. Drug Dev Ind Pharm 1998; 24:1055–1065.
35. Hariharanand M, Mehdizadeh E. The use of mixer torque rheometry to study the effect
of formulation variables on the properties of wet granulations. Drug Dev Ind Pharm
2002; 28:253–263.
36. Becker D, Rigassi T, Bauer-Brandl A. Effectiveness of binders in wet granulation: a comparison
using model formulations of different tabletability. Drug Dev Ind Pharm 1997;
23:791–808.
37. Suzuki T, Kikuchi H, Yamamura S, Terada K, Yamamoto K. The change in characteristics
of microcrystalline cellulose during wet granulation using a high-shear mixer.
J Pharm Pharmacol 2001; 53:609–616.
38. Habib YS, Abramowitz R, Jerzewski RL, Jain NB, Agharkar SN. Is silicified wet-granulated
microcrystalline cellulose better than original wet-granulated microcrystalline
cellulose? Pharm Dev Technol 1999; 4:431–437.
39. Badawy SI, Menning MM, Gorko MA, Gilbert DL. Effect of process parameters on
compressibility of granulation manufactured in a high-shear mixer. Int J Pharm 2000;
198:51–61.
40. Oulahna D, Cordier F, Galet L, Dodds JA. Wet granulation: the effect of shear on granule
properties. Powder Technol 2003; 130:238–246.
41. Bockand TK, Kraas U. Experience with the Diosna mini-granulator and assessment of
process scalability. Eur J Pharm Biopharm 2001; 52:297–303.
42. Holm P, Schaefer T, Larsen C. End point detection in a wet granulation process. Pharm
Dev Technol 2001; 6:181–192.
43. Timmins P, Delargy AM, Minchom CM, Howard JR. Influence of some process variables
on product properties for a hydrophilic matrix controlled-release tablet. Eur
J Pharm Biopharm 1992; 38:113–118.
44. Kornchankul W, Hamed E, Parikh NH, Sakr A. Effect of drug proportion and mixing
time on the content uniformity of a low-dose drug in a high shear mixer. Pharmazie
2002; 57:49–53.
45. Suzuki T, Kikuchi H, Yonemochi E, Terada K, Yamamoto K. Interaction of microcrystalline
cellulose and water in granules prepared by a high-shear mixer. Chem Pharm Bull
(Tokyo) 2001; 49:373–378.
46. Schaefer T, Holm P, Kristensen HG. Comparison between granule growth in a horizontal
and a vertical high speed mixer. II. Granulation of lactose. Arch Pharm Chem, Scientific
Edition 1986; 14:17–29.
47. Holm P. Effect of impeller and chopper design on granulation in a high speed mixer.
Drug Dev Ind Pharm 1987; 13:1675–1701.
226 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC
48. Faure A, Grimsey IM, Rowe RC, York P, Cliff MJ. Process control in a high shear
mixer-granulator using wet mass consistency: the effect of formulation variables. J Pharm
Sci 1999; 88:191–195.
49. Holm P, Schaefer T, Kristensen HG. Granulation in high-speed mixers. Part VI. Effects
of process conditions on power consumption and granule growth. Powder Technol 1985;
43:225–233.
50. Corvari V, Fry WC, Seibert WL, Augsburger L. Instrumentation of a high-shear mixer:
evaluation and comparison of a new capacitive sensor, a watt meter, and a strain-gage
torque sensor for wet granulation monitoring. Pharm Res 1992; 9:1525–1533.
51. Kopcha M, Roland E, Bubb G, Vadino WA. Monitoring the granulation process in
a high shear mixer/granulator: an evaluation of three approaches to instrumentation.
Drug Dev Ind Pharm 1992; 18:1945–1968.
52. Holm P, Schaefer T, Kristensen HG. Granulation in high-speed mixers. Part V. Power
consumption and temperature changes during granulation. Powder Technol 1985;
43:213–223.
53. Leuenberger H. Granulation, new techniques. Pharm Acta Helv 1982; 57:72–82.
54. Werani J. Production experience with end point control. Acta Pharm Suecica 1988;
25:247–266.
55. Laicher A, Profitlich T, Schwitzer K, Ahlert D. A modified signal analysis system for
end-point control during granulation. Eur J Pharm Sci 1997; 5:7–14.
56. Kornchankul W, Parikh NH, Sakr A. Correlation between wet granulation kinetic parameters
and tablet characteristics. Drugs Made Germany 2001; 44:78–87.
57. Achanta AS, Adusumilli PS, James KW. Endpoint determination and its relevance to
physicochemical characteristics of solid dosage forms. Drug Dev Ind Pharm 1997; 23:
539–546.
58. Whitaker M, Baker GR, Westrup J, Goulding PA, Rudd DR, Belchamber RM, Collins
MP. Application of acoustic emission to the monitoring and end point determination of
a high shear granulation process. Int J Pharm 2000; 205:79–92.
59. Watano S, Numa E, Miyanami K, Osako Y. On-line monitoring of granule growth in
high shear granulation by an image processing system. Chem Pharm Bull (Tokyo) 2000;
48:1154–1159.
60. Watano S, Numa T, Miyanami K, Osako Y. A fuzzy control system of high shear granulation
using image processing. Powder Technol 2001; 115:124–130.
61. Bouckaert S, Massart DL, Massart B, Remon JP. Optimization of a granulation procedure
for a hydrophilic matrix tablet using experimental design. Drug Dev Ind Pharm
1996; 22:321–327.
62. Vojnovic D, Selenati P, Rubessa F, Moneghini M, Zanchetta A. Wet granulation in a
small scale high shear mixer. Drug Dev Ind Pharm 1992; 18:961–972.
63. Vojnovic D, Moneghini M, Rubessa F, Zanchetta A. Simultaneous optimization of several
response variables in a granulation process. Drug Dev Ind Pharm 1993; 19:1479–
1496.
64. Vojnovic D, Moneghini M, Rubessa F. Optimization of granulates in a high shear mixer
by mixture design. Drug Dev Ind Pharm 1994; 20:1035–1047.
65. Vojnovic D, Moneghini M, Chicco D. Nonclassical experimental design applied in the
optimization of a placebo granulate formulation in high-shear mixer. Drug Dev Ind
Pharm 1996; 22:997–1004.
66. Horsthuis GJB, van Laarhoven JAH, van Rooij RCBM, Vromans H. Studies on upscaling
parameters of the Gral high-shear granulation process. Int J Pharm 1993; 92:143–150.
67. Ameye D, Keleb E, Vervaet C, Remon Jean P, Adams E, Massart Desire L. Scaling-up of
a lactose wet granulation process in Mi-Pro high shear mixers. Eur J Pharm Sci 2002;
17:247–251.
68. Iskandarani B, Shiromani PK, Clair JH. Scale-up feasibility in high-shear mixers: determination
through statistical procedures. Drug Dev Ind Pharm 2001; 27:651–657.
High-Shear Granulation 227
© 2005 by Taylor & Francis Group, LLC
69. Leuenberger H. Scale-up of granulation processes with reference to process monitoring.
Acta Pharm Technol 1983; 29:274–280.
70. Zlokarnik M. Problems in the application of dimensional analysis and scale-up of mixing
operations. Chem Eng Sci 1998; 53:3023–3030.
71. Lord R. The principle of similitude. Nature 1915; 95:66–68.
72. Faure A, Grimsey IM, Rowe RC, York P, Cliff MJ. A methodology for the optimization
of wet granulation in a model planetary mixer. Pharm Dev Technol 1998; 3:413–422.
73. Buckingham E. On physically similar systems; Illustrations of the use of dimensional
equations. Phys Rev NY 1914; 4:345–376.
74. Faure A, Grimsey IM, Rowe RC, York P, Cliff MJ. Applicability of a scale-up methodology
for wet granulation processes in Collette Gral high shear mixer-granulators. Eur
J Pharm Sci 1999; 8:85–93.
75. Siroisand PJ, Craig GD. Scale-up of a high-shear granulation process using a normalized
impeller work parameter. Pharm Dev Technol 2000; 5:365–374.
76. Kristensen HG. Particle agglomeration in high shear mixers. Powder Technol 1996;
88:197–202.
77. Rekhi GS, Caricofe RB, Parikh DM, Augsburger LL. A new approach to scale-up of
high shear granulation process. Pharm Technol 1996; 20:58–67.
78. Ogawa S, Kamijima T, Miyamoto Y, Miyajima M, Sato H, Takayama K, Nagai T. A
new attempt to solve the scale-up problem for granulation using response surface methodology.
J Pharm Sci 1994; 83:439–443.
228 Gokhale et al.
© 2005 by Taylor & Francis Group, LLC
8
Low-Shear Granulation
Tom Chirkot
Patterson-Kelley Co., East Stroudsburg, Pennsylvania, U.S.A.
Cecil Propst
SPI Pharma, Norton Shores, Michigan, U.S.A.
INTRODUCTION
Shearing of the powder bed occurs in many granulators. The low-shear granulators
included in this chapter are granulators that, for reasons of agitator speed, sweep
volume, or bed pressures, generate less shear than granulators discussed in other
chapters such as extruders or the high-shear mechanical granulators. Also considered
to be low shear, but covered in other chapters, are fluid bed granulators.
1.1. Comparison of Granulators
Comparing final granule characteristics in fluid bed, low-shear, and high-shear applications
is problematic because it is often difficult for the same formulation to be successfully
processed in each piece of equipment. However, some broad conclusions
may be drawn. Two of the characteristics most frequently reported are bulk density
and particle size; therefore, a comparison will be shown for these characteristics.
These granulators may accomplish the same unit operation but the final product outcome
may be quite different. The differences are a result of varied process requirements
in each granulator. The process of wet granulation involves several steps
including blending, liquid binder addition, and wet massing or distribution of the
liquid. After charging the powder to the mixer, a blending step is required to achieve
a homogenous blend. The time required to achieve the blend depends on the amount
of movement characteristic of the unit and the size of the unit. Also, the degree of
homogeneity differs from one type of mixer class to the next. In some classes of
mixers, one can easily overblend and segregate components of the mix.
The binder solution addition step follows the blending step. The selection of
the type of binder and quantity depends on the type of mixer selected for the wet
granulation. Nouh (1) studied a sulfadiazine formulation using several different binders.
He worked in both a fluid bed unit and a conventional wet massing–screening
method. When using 5% gelatin as a binder, he noted a 0.968mm average granule
size in wet massing and 0.574mm in the fluid bed. For acacia as binder at the 5%
229
© 2005 by Taylor & Francis Group, LLC
1.
level, the values were 0.90mm for wet massing and 0.605mm for the fluid bed. For
a polyvinylpyrrolidone (PVP) binder at the 5% level, the wet massing produced
0.962mm granule size vs. 0.247mm for the fluid bed. Surprisingly, the difference
between the methods did not transfer to bulk density results. Both the acacia and
the PVP formulations had similar bulk densities in both methods. The gelatin did
show a difference, yielding 0.476 g/mL for the wet massing and 0.294 g/mL for
the fluid bed. Gore et al. (2) made a comparison among a fluid bed granulator, a
planetary mixer, and a high-shear granulator. They found that the granule yield
between 20 and 100 mesh was 97.7% in the fluid bed, 53.3% in the planetary mixer,
and 71.8% in the high-shear granulator. The respective bulk densities were 0.39, 0.66,
and 0.71 g/mL. For comparison, in a low-shear tumble granulator, Scarpone et al.
(3) found 84% yield between 20 and 100 mesh in a 0.05 m3 (2 ft3) vessel and
50.7% yield in a 2.83 m3 (100 ft3) vessel. Density measurements averaged 0.486 g/
mL for the two trials of a cardiovascular drug formulation in the smaller vessel.
Sheskey and Williams (4) studied a niacinamide formulation in both low- and
high-shear granulators and concluded that the resulting apparent densities were
nearly identical in both cases. In general, the bulk density values produced in
low-shear tumbling granulators are intermediate in value between those of a fluid
bed and of a high-shear granulator.
Similar conclusions may be drawn about granule morphology as lower-shear
granulators produce fluffier, more porous granules than do high-shear granulators.
A final note related to morphology: If the low-shear tumbling granulator has a rotating
shell, some rounding of granules may be expected as the material flows through
the angle of repose.
1.2. Binder Issues
Because of improved liquid distribution early in the granulation process, some highshear
granulators require 60–80% of the liquid needed in a low-shear mixer. Liquid
requirement for an antacid granulation made in a high-shear horizontal mixer was
4.5% of the total batch, whereas 7% liquid was required for a low-shear planetary
mixer. Some of the low-shear granulators such as the sigma mixer densify granules.
The liquid requirement for the sigma mixer is slightly reduced at 6% for the same
antacid. Some studies indicate that dissolving a solid binder and adding it to the
dry mix vs. adding the solid binder to the dry mix and then wetting increases dried
granule hardness (5).
The final stage of the wet granulation process is the liquid distribution or the
wet massing. This step can be compared with a kneading step during which the voids
between granules are compressed and the granules are thereby densified. The final
density of the granules, therefore, is dependent on the amount of shearing available
in the unit. Because the shear is introduced in the mixing process by mechanical
means by moving impellers or blades, these low-shear mixers cannot compress the
voids between the granules solely by mechanical means and thus require more binder
solution to form granules of some integrity.
The combination of less mechanical shear and lower bed pressure allows for
a thicker binder deposit, especially when the binder is poured into the granulation
unit instead of sprayed. This thicker film often dries to a more effectively spreadable
film during the compression of these granules to form tablets. This increased plasti-
230 Chirkot and Propst
city is especially true at the lower compression forces (Fig. 1). The tablets also have
© 2005 by Taylor & Francis Group, LLC
much lower friability (Fig. 2). The improved friability can be over a much broader
range of compression forces.
2. MECHANICAL AGITATOR GRANULATORS
The machine classes to be considered under mechanical agitator granulators are;
(1) ribbon or paddle blender; (2) planetary mixers; (3) orbiting screw mixers; and
(4) sigma blade mixers.
Figure 2 Friability vs. compression forces.
Figure 1 Hardness vs. compression forces.
Low-Shear Granulation 231
© 2005 by Taylor & Francis Group, LLC
a
2.1. Ribbon or Paddle Blenders
The ribbon blender-type mixer (Fig. 3) is very popular as a dry mixer. However, if
small amounts of liquid are added or if a dry paste is formulated for the machine,
the ribbon blender can serve as a very reliable granulator (5). Most ribbon blenders
are not made to withstand the shaft torque required for manufacturing granules.
One should ensure that the granulator shaft is strong enough before granulating
in a ribbon blender designed to dry mix.
Zoglio et al. (5) studied several kneading times in a chopper ribbon blender and
discovered an optimum range of kneading times for providing the best granule
mechanical properties. A kneading time that is too long densifies the granules. This
increases moisture exposure and particle size and reduces porosity. The tendency of
the material to stick to the side wall is pronounced in the ribbon blender.
Paddles instead of ribbons decrease sticking problems and the torque required.
Paddle blenders as batch granulators can handle wetter paste. These paddle blenders
are occasionally used as continuous granulators and have both lower torque and
more applications than a continuous ribbon blender. The movement of the paste
helps remove material from the paddles. Formulation of a nontacky paste is aided
by using a substance such as microcrystalline cellulose that helps absorb the excess
moisture from the mix, yet is plastic and nonsticky.
Two very popular ribbon or paddle blenders used as granulators are the topogranulator
and the turbulizer.
either compress or mechanically fluidize the granulation. Compression while slightly
wet increases the overall influence of the liquid on the particle size of the granulation.
At the other extreme, fluidization reduces density if used during the granule growth
phase or speeds drying. The topogranulator is also a vacuum dryer.
The topogranulator is used extensively to make effervescent products by liquid
addition under vacuum or by the Murry fusion method (6). Murry’s method uses
liberated moisture from the acid in the mix (i.e., hydrous citric acid) to start the
acid–base reaction, which generates more water. Thus, granulating of the sodium
bicarbonate–citric acid mixture can be accomplished. The water produced must be
removed quickly to reproducibly stop the reaction. The topogranultor, because of
its ability to compress the particles into the binding moisture, makes a larger, denser
Figure 3 Ribbon blender.
232 Chirkot and Propst
The topogranulator (Fig. 4) is a batch-style ribbon blender with the ability to
© 2005 by Taylor & Francis Group, LLC
granulation with lower moisture content. Also, mechanical fluidization and vacuum
drying removes the water from the reaction more quickly, making the process more
reproducible.
Granulation in vacuum removes entrapped air from the particle surface.
Liquid addition into the vacuum provides immediate wetting and also begins the
drying process. This granulation in vacuum allows the manufacturing of a calcium
carbonate-based effervescence with rapid reactivity (7).
The turbulizer is a continuous paddle granulator. By using continuous powder
feeders and liquid metering pumps, the unit produces large quantities of product per
hour in a very small space. The unit provides adjustable mixing, shear, and impact
action based on the revolutions per minute (rpm) of the shaft and the angle of the
impact blade. The paddle adjustments also vary the retention time. This machine
has a very low silhouette. Paddles are easily accessible for cleaning and inspection
either through a clam shell- or drop door shell-opening side wall. The unit is jacketed
for material temperature control.
Another continuous paddle granulator made by Teledyne Readco has been
studied by Ghali et al. (8). The unit is adjustable to produce either low- or highshearing
action. The amount of energy induced by the rotating shaft can be selected
by using various types of pins. A rounded granule results from the action of the pin
tip speed and some rolling of the granulation against the fixed vessel wall.
2.2. Planetary Mixers
The planetary motion of these granulators is created by rotating the agitator off an
assembly in a direction opposite that of the rotation of the agitator assembly as it
mercial names including Hobart, Kitchen Aide, Pony, and AMF Glen granulators.
All of these mixers have the same basic makeup, which includes: (a) planetary
motion, (b) removable bowl, and (c) top-drive agitators.
These mixers tend to be better at mixing dry powders in a horizontal plane than
the vertical. Lack of vertical mixing may require the materials to be dumped and
returned to the bowl to obtain an acceptable dry mix. Reduced vertical mixing
Figure 4 Topogranulator.
Low-Shear Granulation 233
moves around the bowl. The planetary mixers (Fig. 5) are represented by many com-
© 2005 by Taylor & Francis Group, LLC
is much less a problem in the wet-mix phase, because the materials adhere to the
agitator and move in groups without vertical stratification.
Remon and Schwartz, working with microcrystalline cellulose and lactose mixtures
in a planetary mixer, saw decreased friability of the granule with increasing
massing time (9). This increased massing time improves binder distribution and
mechanical strength. Ghanta et al. demonstrated the mechanism of granule growth
in a Hobart mixer (10).
2.3. Orbiting Screw Granulators
The orbiting screw granulator (Fig. 6) is also used mainly for dry mixing. However,
the unit has been fitted with the nozzle through the center agitator to add liquids to
dry powders (11). Also, a jacket can provide both heating and cooling. A sintered
metal plate can allow entry of compressed air through the skin of the mixer or
can exhaust moisture to obtain drying. All of these added features, along with a
Figure 5 Planetary mixer.
Figure 6 Orbiting screw mixer.
234 Chirkot and Propst
© 2005 by Taylor & Francis Group, LLC
chopper in the side wall, allow this blender to be an effective granulator for powders,
slurries, suspensions, and paste. The unit is a very gentle granulator mixer.
2.4. Sigma Blade Mixers
The sigma blade mixer (Fig. 7) is a compressive granulator. Generally, it develops
material into a paste- or dough-type consistency. Often the sigma blade mixer is preceded
by a dry mixer. The unit generates liquid distribution with pressure during
granulating. This creates a uniform granulation with binder matrix and good binder
distribution.
ROTATING SHAPE GRANULATORS
These vessel shapes are usually some derivation of a cylinder, with a double cone and
counterparts, the machine shells rotate around an axis parallel to the ground. The
rotation speed is moderate, and generally falls into a range of 72.2–106.7 m/min
(250–350 ft/min) peripheral speed. The rpm value changes as vessel size grows.
laboratory model may rotate at 25–30 rpm, whereas a larger production model
may rotate at 4–8 rpm. The peripheral speed remains constant in order to maintain
scale-up relationship among vessel sizes.
Figure 7 Sigma blade mixer.
Low-Shear Granulation 235
V-shape (Fig. 8) being common examples. Unlike their fluid bed and high-shear
© 2005 by Taylor & Francis Group, LLC
3.
A
a
A second rotating device is located on the same axis of rotation as the shell.
This bar or arm may be supported at both ends or may be cantilevered, with support
only on a single end. The bar is used to impart substantially more energy into the
system than that delivered by the rotating shell. The bar is driven by an additional,
larger motor that runs independently of the shell motor. The manufacturer’s descriptive
names for these bars, such as agitator or intensifier bars, confirm their use.
Unlike the gentle rolling action induced by the shell movement, the bar movement
has a high-speed nature. This should not be confused with high shear because the
bar movement produces mainly a convective motion within the material. Indeed,
these machines were originally designed as mixers, with elements of diffusive and
convective mixing. Only later in their evolution did they become granulator-dryers.
The increased peripheral speed of the bar is substantial. As a general rule, they
operate at 10 times the speed of the shell or about 914.4 mpm (3000 fpm). The bars
may also serve as liquid addition devices. The vessels may be jacketed for heating
and cooling if there is a necessity for these options during granulation. They are
vacuum capable, which makes them ideal candidates as a single-pot processor for
mixing, granulating, and drying in the same vessel. Ultimate vessel size for these
granulators is constrained by the weight of the agitator bar. The bars have many
parts and must be disassembled often for cleaning. Large bars become increasingly
unwieldy and difficult to remove from the interior of the vessel.
Ample opportunity exists for fine-tuning a granulation process in these machines.
Apparatus variables that can be adjusted are shell speed, bar speed, bar size, and bar
design. Process changes can be made in mix time, liquid addition time, bed temperature,
internal vessel pressure, and spray droplet size.
Overwhelmingly, these granulator types are used in batch operations. However,
a few manufacturers offer continuous granulating systems that emulate the
combination of forced and rolling agglomeration found in the batch machines.
Besides the issue of validating a continuous process, throughput requirement is
usually the influencing criterion when considering a change from batch to continuous
operation.
Figure 8 Schematic of the twin-shell blender that allows blending and granulating in a single
vessel. (Courtesy of Patterson-Kelley Co., Division of Harsco Corp., East Stroudsburg, PA.)
236 Chirkot and Propst
© 2005 by Taylor & Francis Group, LLC
Much of the granulation done in the pharmaceutical industry is by wet massing
and, more recently, by fluidized bed (12). Consequently, the literature is rich with
articles describing the many variables studied in these machines. On the other hand,
the rotating shape granulators originated as mixing devices and a paucity of research
articles exist on their usage as granulators. This makes them a fruitful area for future
research.
3.1. Bar Speed and Energy Input
The combination of agitator bar speed and its accompanying energy input wields
a wide influence on the physical properties of the granulated material. Watano
et al. (13), using a fluidized bed granulator with an agitator blade, varied the rotation
speed of the agitator and observed the effect it had on product density. They were
able to derive a predictive equation that yielded the density value when the agitator
speed was known. The same group measured shape index—the mean ratio of the
short/long diameters—and concluded that as granulation times increased, the shape
tended more toward sphericity. The rotating shape granulators vary energy input to
the powder bed by changing the agitator speed or extending the running time of the
bar after liquid addition has stopped. A common term for the extended bar period is
postmix time.
A study of a low-shear granulator, with a formulation consisting of lactose,
PVP, and water, that varied the peripheral speed of the bar showed no statistically
significant change in granule density despite the substantial change in bar tip speed
from 636 to 846 to 1056 mpm (14). The tip speed was statistically influential on the
amount of granules found in the yield fraction. The lowest speed produced the best
yield indicating that the higher speeds can degrade the formed granules during
a postmix phase and thereby cause a reduction in yield. The ultimate lower limit
on tip speed is the point where adequate distribution of the binder solution is
compromised.
3.2. Disk Size and Bar Design
Addition of binder liquid with spray heads is available with rotating shape granulators.
The double-cone shape is often used with a dry intensifier bar and spray head
combination for granulating. A wide range of bar designs exists for the more typical
situation of adding liquid through the bar. Blade or knife design is one variable to
consider. The blades extend perpendicularly from the disk circumference and then
bend 90

The blade shape is critical for the liquid addition method. As the blades rotate,
they carve out a toroidal void volume in the flowing powder bed. The liquid exiting
the disk coats the interior of the torus. Proper vessel loading ensures that no liquid
impinges on the wall of the vessel.
The action of the blades is quite vigorous and may cause a problem with very
friable materials. A modification to the bar that removes the blades and recesses all
nuts and bolts can be used to granulate these fragile materials. Further fine-tuning of
the spray pattern can be accomplished with changes in the positioning of liquid evacuation
spacing on the disk circumference. A straight pattern creates droplet flow
orthogonal to the axis of rotation, which allows only a small area of the interior
torus to be coated. An angled pattern creates a substantially wider spray effect, with
more efficient droplet distribution.
Low-Shear Granulation 237
in a shape often called a ‘‘dog-eared’’ design (Fig. 9).
© 2005 by Taylor & Francis Group, LLC
Disk diameter is critical. A large-diameter disk imparts more energy as torque
is a function of diameter. The larger diameter offers more circumference for liquid to
be evacuated and carves out additional void volume in the material. The sweep
volume of the disks is considered a major factor in properly scaling up a granulator
(15).
3.3. Liquid Addition Rate
Particle motion is so complex when granulating that it becomes difficult to determine
the optimum rate of liquid addition. Stamm and Paris (16) studied rates of 5, 10, and
20 mL/min in a fixed shell, helicoidal mixer with vertical bar, and determined that
the best results occurred at the slowest rate.
Cliff (17), working in a high-shear granulator, found that a long binder addition
time was needed to prevent overdensification. Lipps and Sakr (18), using a topspray,
fluidized bed granulator, related geometric mean granule size, specific surface
area, and granule flow properties to the binder flow rate.
In rotating shape granulators, the main energy-imparting function and the
liquid distribution function are encompassed within the agitator bar, yielding
a highly efficient system. Liquid is literally ripped into droplets by the peripheral
speed of the bar. Tumbling action ensures that fresh material replaces the wet material
during each shell revolution (19). A wide range of liquid addition rates is permissible.
The lowest rate is undefined and may involve a slow dropwise addition to the
bar. The highest rate is limited to the point when the liquid rate reduces the droplet
making efficiency of the bar and results in liquid being forced out in a sheet-like
pattern to the powder bed.
When one extends the liquid addition period, the bar continues to abrade the
granules. Favorably, any particles that have been fractured can be rewetted and the
fracture surface can easily bind to another particle. This mechanism tends to hold
down the oversized fraction without the expected increase in the undersized fraction.
Extremely rapid liquid addition overwhelms the blending capability of the system.
Many particles simply do not have enough time to become associated with the
droplets, whereas others become overwet due to the concentrated onrush of liquid.
An additional consequence of high liquid addition rate is that liquid may have diffi-
culty releasing from the bar. A back pressure situation develops that may impede
liquid flow in the feed tube and ultimately cause it to reverse its flow. High enough
back pressure can induce a spring-loaded bar to jump from its moorings and damage
Figure 9 Intensifier bar for the twin-shell blender/granulator showing a ‘‘dog-eared’’ design.
(Courtesy of Patterson-Kelley Co., Division of Harsco Corp., East Stroudsburg, PA.)
238 Chirkot and Propst
© 2005 by Taylor & Francis Group, LLC
the interior of the machine. Because the rotating shape granulators are often heatjacketed,
the potential exists for them to provide a partially or fully coated, hard
granule.
Use of the dissolved solids binder addition method through an agitator bar
could be problematic. Any undissolved solids may clog the liquid evacuation spacing
on the disk. Also, binder solutions with a viscosity much beyond 300–500 cP may not
atomize readily when released from the disk.
3.4. Droplet Size
Droplet size is a very important variable in fluidized bed granulation. Droplet size is
unimportant in high-shear granulators because the energy exists to readily distribute
and redistribute the liquid. Liquid addition in the rotating shape granulators is
unique, for the droplets are released from the fluidizing device or agitator bar. An
analogous situation in a high-shear application would mean that liquid was being
added through the impeller or chopper.
The liquid in a rotating shape granulator is fed through a tube and exits in the
interior of the bar through a distribution slot. A pump can be used for metering the
flow, or gravity flow may be sufficient, if desired. The speed of rotation of the bar
actually induces an area of lower pressure within the bar to draw in the liquid.
The liquid exits the agitator bar through openings on the circumference of the rotating
disk. Openings can be adjusted by placement of spacers that vary in thickness.
A droplet diameter cannot exceed this thickness value, although smaller droplets
may be present. Typical spacer openings are 0.025 cm (0.010 in.) with common adjustment
to 0.0125 cm (0.005 in.) and 0.05 cm (0.020 in.), depending on the viscosity of the
liquid.
Many granulation methods deliver the liquid at some pressure through a spray
head. With liquid addition through the bar, only enough pressure is needed to overcome
line loss. Undue pressure is actually counterproductive. The atomization is
influenced by pressure and the liquid exiting the disks under high pressure tends
to be sheet-like rather than easily distributed droplets.
Droplet size in a fluid bed has been measured by capturing droplets on a slide
covered with viscous oil. The size has been determined to be in the 20–100 mm
range (20). Less delicate techniques in the rotating shape granulators have shown
droplet sizes of about 250 mm. Agland and Iverson provide an experimental study
showing the relation of liquid droplet size to granule characteristics such as size
and penetration of liquid (21).
3.5. Vessel Loading
A negative factor for rotating tumble granulators with high-speed internal bars is the
tight constraint for fill level. The fill level is usually about 50–65% of the total
volume, and the powder level must have some contact with the bar. Overloading
the vessel impedes the mixing action.
Underloading causes material to flow beneath the bar during vessel rotation,
resulting in liquid spraying on the interior wall. This spray-through can be a factor
in material sticking during a drying step.
Even careful loading of the machine with the dry powder may not be sufficient
to preclude spraying through to the walls. If substantial densification occurs during
early liquid addition, the load may drop enough during the final stages of liquid
Low-Shear Granulation 239
© 2005 by Taylor & Francis Group, LLC
addition for liquid to find its way to the shell wall. The effective working volume fill
level can be increased if the tip-to-tip blade diameter is extended.
3.6. Low Shear Single Pot Processing
Their original design as mixers ensures even minor ingredients are well-distributed in
the dry mix phase prior to granulation.
After granulation, the heated shell and vacuum capability are used for gentle
tumble drying and collection of condensed vapors. The bar can then be used to provide
a measure of dry sizing, followed by lubricant addition, and tumble blending.
3.7. Continuous Granulation
Although batch granulation is commonly used, there are some examples of lowshear
continuous granulation. A true adaptation of the batch, rotating shape
sists of a high-speed mixing chamber connected to a series of three V-shaped tumble
blenders. The principles of liquid addition and disk speeds covered earlier are equally
applicable to the Zig Zag.
The typical hold-up volume is half of the total volume of the vessel. The holdup
volume can be adjusted somewhat by raising or lowering the discharge end a few
degrees from the horizontal. Residence times are in the 3–5 min range for granulation
applications. Two types of granulation mechanisms occur in this machine.
When the material is in the high-speed mixing chamber, the liquid droplets contact
the powder in a manner similar to that of batch units. As the wet material travels
into the tumbling section, rolling agglomeration occurs. The tumbling section splits
its load in half on each revolution, allowing half the material to proceed forward
while the other half recycles to the rear. This action tends to smooth out any inconsistencies.
Scale-up to larger Zig-Zag units is based on constant residence times. Therefore,
even in the very largest models, a residence time of 3–5 min can be expected.
The largest models are capable of producing 30,000 kg/hr.
4. SCALE-UP
Often, the most difficult part of implementing a granulation process is the scale-up
step. The many variables involved and the poorly understood relations among the
variables often places scale-up in the realm of trial and error. Hancock et al. (22),
using a fixed bowl mixer with meshing blades, linked a torque arm to a dynamometer
to monitor the process. The amplitude of oscillation (torque range) and the mean
torque increase from baseline were the items recorded. Torque showed a relation
for the force required to shear the wet mass and provided an output for determining
the wet mass strength. Integrating the torque-time curve yields the total energy into
the system for a particular time period. Hancock’s group defined the term, cumulative
energy of mixing, and postulated its use as a means of scaling up.
Vojnovic et al. (23) determined that peripheral speed of the impeller in a vertical
high-shear mixer was an important factor and adjusted this speed as a function
240 Chirkot and Propst
The rotating shape granulators offer fine potential for single-pot processing (Fig. 10).
granulators is the Patterson-Kelley Zig Zag mixer (Figs. 11 and 12). This vessel con-
© 2005 by Taylor & Francis Group, LLC
Figure 10 Schematic diagram of Patterson-Kelley Solids Processor. (Courtesy of Patterson-Kelley Co., Division of Harsco Corp., East Stroudsburg, PA.)
Low-Shear Granulation 241
© 2005 by Taylor & Francis Group, LLC
of the diameter of the larger mixer. They found this to be an adequate scale-up
method despite a geometric dissimilarity in the two mixers studied.
Further work on the rheologic character of the wet mass and the effect on
scale-up was performed by York in a high-shear mixer–granulator (24). It was theorized
that the power number–Reynolds number relationship used in the scale-up of
fluid mixing impeller systems could be applicable to the granulation mechanism in
high-shear mixer. To analyze the relationship, appropriate machine information
needed to be measured, as did the density and viscosity of the wet mass. Mixer
torque rheometry was used to determine wet mass viscosity, and yielded a pseudo-
Reynolds number with the units of torque. It is interesting to note that some unit
operations textbooks show similar logic in developing scale-up equations for dry
powder mixing (25).
Figure 12 Zig-Zag mixer–agglomerator. (Courtesy of Patterson-Kelley Co., Division of
Harsco Corp., East Stroudsburg, PA.)
Figure 11 Schematic of a Zig-Zag, which is an example of a continuous granulating machine.
(Courtesy of Patterson-Kelley Co., Division of Harsco Corp., East Stroudsburg, PA.)
242 Chirkot and Propst
© 2005 by Taylor & Francis Group, LLC
Scaling up the process in a low-shear granulator is a daunting task. Scale-up
equations are usually proprietary to the mixer manufacturers and little is written
about them in the literature. The equations have a sound basis in science but are
ameliorated by constants and factors that take into consideration empirical feedback
and intuition concerning the machine function.
Most commonly, an energy/volume ratio is established in the smaller vessel.
This ratio is emulated in the larger vessel, with appropriate consideration for the
added mass and the larger motor, to yield an estimated scale-up time.
Scarpone et al. (3) studied a change in granulation technology from a highshear
mixing bowl to a V-shaped granulator. Developmental work was done in a
0.056 m3 (2 ft3) vessel and the ultimate production in a 2.83 m3 (100 ft3) vessel.
Scale-up based on the manufacturer’s recommendation was not exact and required
some fine-tuning before the process was established in the larger vessel.
This study provides a detailed look at some of the hazards encountered in scaling
up. The initial trial in the 2.83m3 (100 ft3) production vessel failed to provide an
adequate granulation. The main reason for the failure was that the scale-up trial
lacked the active ingredient and contained only excipient. This highlights the
necessity of maintaining as much constancy as possible, both with formulation
and process variables. Additional tests with the active ingredient were much closer
to the results gained in the smaller vessel. By the third trial, a successful scale-up
had been achieved. Further studies were conducted to evaluate the influence of
changes in the physical characteristics of the active ingredient. For example, density
changes had an influence on the binder liquid needed. The researchers were able to
use their knowledge from the scale-up to achieve another scale-up on a different
product.
M. Uliveri, M. Eli, and R. Bianchini (unpublished data) also attempted a scaleup
in a V-shaped granulator. The laboratory vessel was a 0.056m3 (2 ft3) model and
the scale-up vessel was a 0.283m3 (10 ft3). They found it was necessary to increase the
batch size 1.2 times over the direct 10:2 ratio, that the total liquid was 82% of direct
scale-up, and the granules were slightly larger in the scale-up vessel.
Another scale-up attempt in a V-shaped granulator involved a 2 and 57 L vessel
(15). The author used a cumulative torque to mass ratio as a means of relating the
process in both vessels. A separate, individual method is also provided for scaling-up
the liquid addition time.
Suitably, an article by Ennis et al. (26) highlights the fact that understanding
granulation scale-up is a needed growth area for further research.
5. END-POINT DETERMINATION AND CONTROL
The rapidity by which granulation proceeds makes end-point determination a diffi-
cult problem. Growth behavior is nonlinear, preventing easy solutions to the differential
equations describing the process (27). Watano et al. (13) used an infrared
moisture sensor to continuously monitor moisture through a PID feedback loop.
The current was used to control the liquid addition pump until a suitable size was
achieved. Leuenberger et al. (28) plotted the power consumption curve in a planetary
mixer and observed five distinct phases that were dependent on the amount of liquid
added. Zoglio et al. (5), working in a ribbon blender, postulated a potential
end-point determination using specific pore area. Ghanta et al. (10) installed a slip
ring torque sensor between the Hobart mixer agitator and motor and observed five
Low-Shear Granulation 243
© 2005 by Taylor & Francis Group, LLC
phases in the torque profile. This process was repeatable and could be considered for
granulation end point in this type of mixer. Another method uses effusivity to determine
the amount of moisture in the granule for end-point determination (29).
6. CONCLUSIONS
Low-shear granulators offer a middle ground solution for many of the problems formulators
may encounter. Much denser granules may be produced in these vessels,
compared with a fluid bed device, yet the energy expended is not as great as that
found in a high-shear machine. Many of the low-shear granulators are extremely
adaptable devices capable of mixing the formulation constituents before granulating,
and some even dry the materials after granulation is complete. Ample opportunity is
available in these granulators for adjustments: energy input may be altered with variable
speed drives; droplet size may be changed through judicious selection of spray
heads; and various agitating bar designs may be selected. As with many other granulators,
scale-up and end-point detection remain poorly defined. A trial and error
procedure is often the only method to determine end point and scale-up.
REFERENCES
1. Nouh ATI. The effect of variations in concentration and type of binder on the physical
characteristics of sulfadiazine tablets and granulations prepared by wet and fluidized-bed
granulation method. Pharm Ind 1986; 48(6):670.
2. Gore AY, McFarland DW, Batuyios NH. Fluid-bed granulation: factors affecting the
process in a laboratory development and production scale-up. Pharm Technol
1985:114–122.
3. Scarpone AJ, Dalvi UG, Delorimier AE. Preparing tablet granulations in a 100 cu ft
solids processor. Pharm Technol 1986; 10:44.
4. Sheskey PJ, Williams DM. Comparison of low-shear and high-shear wet granulation
techniques and the influence of percent water addition in the preparation of a controlled-
release matrix tablet containing HPMC and a high-dose, highly water-soluble
drug. Pharm Technol 1996:80–90.
5. Zoglio MA, Huber HE, Koehne G, Chan PL, Carstensen JT. Physical aspects of wet
granulation II: factors involved in prolonged and excessive mixing. J Pharm Sci 1976;
65:1205.
6. Murry R. J Pharm Sci 1968; 57:1776–1779.
7. Gergely et al. Effervescent system for effervescent tablets and effervescent granules. US
Patent 5888544, March 30,1999.
8. Ghali S, Contractor AM, Mankad AD, O’Connor RE, Schwartz JB, Auslander DE,
Grim WM. A high-speed mixer for continuous wet granulation. Pharm Technol 1990;
14:60.
9. Remon JP, Schwartz JB. Effect of raw materials and processing on the quality of granules
prepared from microcrystalline cellulose-lactose mixtures. Drug Dev Ind Pharm
1987; 13:1.
10. Ghanta SR, Srinivas R, Rhodes CT. Use of mixer torque measurements as an aid to optimizing
wet granulation process. Drug Dev Ind Pharm 1984; 10:305–311.
11. Day JH & Co. News bulletin: monitoring particle density, moisture content, particle size,
and process weight. Cincinnati, OH, 1983.
12. Schaefer T. Equipment for wet granulation. Acta Pharm Suecica 1988; 25:205.
244 Chirkot and Propst
© 2005 by Taylor & Francis Group, LLC
13. Watano S, Terashita K, Miyanami K. Determination of end-point with a complex granulation
applying infrared moisture sensor. Chem Pharm Bull 1991; 329:1013.
14. Chirkot TS. Characterization of a pharmaceutical wet granulation process in a V-type
granulator. Pharm Eng 1999; 19(4).
15. Chirkot TS. Scale-up and endpoint issues of pharmaceutical wet granulation in a V-type
low shear granulator. Drug Dev Ind Pharm 2002; 28:871.
16. Stamm A, Paris L. Influence of technological factors on the optimal granulation liquid
requirement measured by powder consumption. Drug Dev Ind 1985; 11:333.
17. Cliff MJ. Granulation end point and automated process control of mixer-granulators:
part I. Pharm Technol 1990; 14:112.
18. Lipps D, Sakr AM. Characterization of wet granulation process parameters using
response surface methodology. 1. Top-spray fluidozed bed. J Pharm Sci 1994; 83:937.
19. Fischer JJ. Liquid-solids blending. Chem Eng 1962; 3(Feb 5).
20. Lindberg NO, Jonsson C. The granulation of lactose and starch in a recording high-speed
mixer, Diosna P25. Drug Dev Ind Pharm 1985; 1:387.
21. Agland S, Iverson S. The impact of liquid droplets on powder bed surfaces. CHEMECA
’99, Newcastle, Australia, Sep 26–29, 1999:25.
22. Hancock C, York P, Rowe RC, Parker MD. Characterization of wet masses using
a mixer torque rheometer: 1. Effect of instrument geometry. Int J Pharm 1991; 76:239.
23. Vojnovic D, Moneghini M, Rubessa F, Zanchetta A. Simultaneous optimization of several
response variables in a granulation process. Drug Dev Ind Pharm 1993; 19:1479.
24. York P. Granulation using mechanical agitation. Proceedings of IFPRI Annual Meeting,
Urbana, IL, 1995.
25. McCabe WL, Smith JC, Harriot P. Unit Operations of Chemical Engineering. 4th ed.
New York: McGraw-Hill, 1985:846.
26. Ennis BJ, Green J, Davies R. The legacy of neglect in the US. Chem Eng Prog 1994:32.
27. Oliver R, Ford LJ. The role of agglomeration in chemical and process technology. Proceedings
of the Institute for Briquetting and Agglomeration, Orlando, FL, 1987:1–32.
28. Leuenberger H, Bier HP, Sucker HB. The theory of the granulating liquid requirement in
the conventional granulation process. Pharm Technol 1986; 11:76.
29. Mathis Instruments. Technical Literature. Fredericton, NB, Canada: Mathis Instruments,
2003.
Low-Shear Granulation 245
© 2005 by Taylor & Francis Group, LLC
9
Batch Fluid Bed Granulation
Dilip M. Parikh
Synthon Pharmaceuticals Inc., Research Triangle Park, North Carolina, U.S.A.
Martin Mogavero
Niro Pharma Systems, Columbia, Maryland, U.S.A.
1. INTRODUCTION
Fluidization is the unit operation by which fine solids are transformed into a fluidlike
state through contact with a gas. At certain gas velocities, the fluid will support
the particles, giving them freedom of mobility without entrainment. Such a fluidized
bed resembles a vigorously boiling fluid, with solid particles undergoing extremely
turbulent motion, which increases with gas velocity. Fluidized bed granulation is a
process by which granules are produced in a single piece of equipment by spraying
a binder solution on to a fluidized powder bed. This process is sometimes classified as
the one-pot system. The fluid bed granulation process has received considerable attention
within the pharmaceutical industry; however, other process industries, such as
food, agro-chemical, dyestuffs, and other chemical industries, have adopted the fluid
bed granulation process to address particle agglomeration, dust containment, and
material handling. The fluidization technique, as it is known today, began in 1942,
with the work of the Standard Oil Company (now known as Exxon, in the United
States) and M.W. Kellogg Company, in an effort to produce the first catalytic cracking
plant on a commercial scale (1).
Fluid bed processing of pharmaceuticals was first reported by Wurster, when
he used the air suspension technique to coat tablets (2,3). In 1960, he reported on
granulating and drying of a pharmaceutical granulation, suitable for the preparation
of compressed tablets, using the air suspension technique. In 1964, Scott et al. (4)
and Rankell et al. (5) reported on the theory and design considerations of the process,
using a fundamental engineering approach and employing mass and thermal
energy balances. They expanded this application to the 30 kg-capacity pilot plant
model designed for both batch and continuous operation. Process variables, such
as airflow rate, process air temperature, and liquid flow rate were studied. Contini
and Atasoy (6) later reported the processing details and advantages of the fluid
bed process in one continuous step.
Wolf (7) discussed the essential construction features of the various fluid bed
components, and Liske and Mobus (8) compared the fluidized bed and traditional
247
© 2005 by Taylor & Francis Group, LLC
granulation process. The overall results indicated that the material processed by the
fluid bed granulator was finer, more free-flowing, and had homogenous granules,
which, after compression, produced stronger and faster disintegration of tablets
than the materials processed by conventional wet granulation. Reviews by Sherrington
and Oliver (9), Pietch (10), and a series published on the topic of ‘‘Fluidization
in the Pharmaceutical Industry’’ (11–17) provide an in-depth background on the
fundamental aspects of the fluidized bed and other granulation technologies. The
fluidized bed was used only for drying the pharmaceutical granulation efficiently
in early days, but now it is employed routinely for drying, agglomeration, pelletization,
and production of modified-release dosage forms using air suspension coating.
Because of this, these units are normally classified as multiprocessor fluid bed
units.
The size enlargement of primary particles has been carried out in the pharmaceutical
industry in a variety of ways. One of the most common unit operations used
in the pharmaceutical industry is fluid bed processing. The batch size increase using
fluid bed granulation requires a good understanding of the equipment functionality,
theoretical aspect of fluidization, excipient interactions, and most of all identifying
the critical variables that affect the process of agglomeration.
This chapter will provide the essential understanding of the fluidization theory,
system description that make up the fluid bed processor, and will discuss the critical
variables associated with the equipment, the product, and the process.
2. FLUIDIZATION THEORY
A fluidized bed is a bed of solid particles with a stream of air or gas passing upward
through the particles at a rate great enough to set them in motion; this velocity,
according to Ku? lling and Simon (18), is higher than the incipient fluidizing velocity,
but lower than the entrainment velocity. When the rate of flow of gas increases, the
pressure drop across the bed also increases until, at a certain rate of flow, the frictional
drag on the particles equals the effective weight of the bed. These conditions,
and the velocity of gas corresponding to it, are termed incipient fluidization and incipient
velocity, respectively. The relationship between the air velocity and the pressure
At low gas velocities the bed of particles is practically a packed bed, and the
pressure drop is proportional to the superficial velocity. As the gas velocity is increased,
a point is reached at which the bed behavior changes from fixed particles to suspended
particles. The superficial velocity required to first suspend the bed particles is known as
minimum fluidization velocity (umf). The minimum fluidization velocity sets the lower
limit of possible operating velocities, and the approximate pressure drop can be used
to approximate pumping energy requirements. For the agglomeration process in the
fluid bed processor, the air velocity required is normally five to six times the minimum
fluidization velocity.
At the incipient point of fluidization, the pressure drop of the bed will be very
close to the weight of the particles divided by the cross-sectional area of the bed
(W/A). For the normal gas fluidized bed, the density of the gas is much less than
the density of the solids and the balance of forces can be shown as
DPmf ? W=A where W ? ?1 
 emf ?rp?g=gc?
248 Parikh and Mogavero
drop is as shown in Figure 1 (19).
© 2005 by Taylor & Francis Group, LLC
where DP?pressure drop, emf?minimum fluidization void fraction, A?crosssectional
area, W?weight of the particles, rp?density of particles, g/gc?ratio of
gravitational acceleration and gravitational conversion factor.
The fundamental phenomenon of fluidization was recently studied by researchers
(20) using a small-scale fluid bed unit. The purpose of the study was to compare
experimental and computational minimum fluidizing velocities (umf) of pharmaceutical
materials using miniaturized fluid bed device. Using various materials, researchers
found that the experimental method was more capable of describing the fluidizing
behavior of pharmaceutical materials than the computational approach. Computational
models of fluidization are based on the behavior of various model particles.
Computational models do not take into account particle size and shape distributions,
cohesion and adhesion of pharmaceutical materials.
As the velocity of the gas is increased further, the bed continues to expand and
its height increases with only a slight increase in the pressure drop. As the velocity of
the gas is further increased, the bed continues to expand and its height increases,
whereas the concentration of particles per unit volume of the bed decreases. At a certain
velocity of the fluidizing medium, known as entrainment velocity, particles are
carried over by the gas. This phenomenon is called entrainment. When the volumetric
concentration of solid particles is uniform throughout the bed all the time,
the fluidization is termed as particular. When concentration of solids is not uniform
throughout the bed, and if the concentration keeps fluctuating with time, the fluidization
is called aggregative fluidization.
A slugging bed is a fluid bed in which the gas bubbles occupy the entire crosssection
of the product container and divide the bed into layers.
A boiling bed is a fluid bed, in which the gas bubbles are approximately of the
same size as the solid particles.
A channeling bed is a fluid bed, in which the gas forms channels in the bed
through which most of the air passes.
A spouting bed is a fluid bed in which the gas forms a single opening through
which some particles flow and fall on the outside.
Figure 1 Relation between the air velocity and pressure drop. (From Ref. 19.)
Batch Fluid Bed Granulation 249
© 2005 by Taylor & Francis Group, LLC
Figure 2 shows various types of fluid beds (21).
The mechanisms by which air affects fluidization have been discussed by various
researchers (13,22–26). When the fluidizing velocity is greater than the incipient
velocity, bubbles of air rise through the bed causing mixing of particles. Mixing does
not generally occur when the bed is fluidized at very low or zero excess gas velocities,
because insufficient bubbles are formed to cause bulk displacement of particles. It is
the gas passing through the bed in the form of bubbles that determines the degree of
mixing. The extent of mixing appears to vary with the particle size. Mixing of particles
having a mean particle size of less than approximately 150 mm decreases as their
mean size approaches zero. Different types of beds, described above, are formed
depending on the movement of bubbles through the bed. The pattern of movement
of the gas phase in and out of bubbles depends on several factors, including minimum
fluidization velocity and particle size. These movements affect heat transfer
between air bubbles and particles. The air distributor at the bottom of the container
has a controlling influence on the uniform distribution of gas, minimization of dead
areas, and maximization of particle movement. The most common reason for mixing
problems, such as segregation in the fluid bed, is the particle density differences. The
main characteristic of the fluid bed is the relative velocity imparted to the particles,
U0, which is a strong function of the size of the particles and the gas velocity in the
bed, and was shown to be given by (27)
U0  aY0 ? 18Uba=Dbd2
Figure 2 Various types of fluid beds. (From Ref. 21.)
250 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
where a is the average particle size, Ub is the bubble velocity, Db is the bubble diameter,
and d is the dimensionless bubble spacing. The first expression on the righthand
side of equation applies to fluidized beds with no rotating parts, where shear
is induced by the motion of bubbles only.
The extent of segregation can be controlled in part by maintaining high fluidizing
velocities and high bowl height–bowl diameter ratio. There are standard air velocities
for various processes that can be used as guidelines. The standard velocities are
based on the cross-sectional area at the bottom of the product container.
This is calculated by using the following formula for calculating the air velocity.
Velocity ?m=sec? ? Airflow ?m3=hr?=Area ?m2?  3600
where airflow in cubic meters per hour (CMH)?airflow (CFM)1.696.
Standard air velocities are based on the application. Low air velocities such as
0.8–1.4 m/sec are required for drying. The velocities are higher during the early
stages of drying because of the wet mass present in the bowl, but will be reduced
when the product loses its moisture. The objective is to have good particle movement,
but to keep the material out of filters. Particle movement and quick drying
are important during the agglomeration process. Airflow velocities are normally
1.0–2.0 m/sec.
An indication of good fluidization is a free downward flow of the granulation
at the sight glass of the drying container. However, improper fluidization can also be
detected by monitoring the outlet air temperature. Every product has a unique constant
rate of drying in which the bed temperature remains relatively constant for a
significant length of time. Therefore, if the outlet temperature rises more rapidly than
anticipated, it will indicate an improper fluidization and the process may have to be
stopped, and manual or mechanical intervention may be required to assist the
fluidization.
3. SYSTEM DESCRIPTION
A fluid bed processor is a system of unit operations involving conditioning of process
air, a system to direct it through the material to be processed, and have the same air
cal fluid bed processor with all the components. These components and their utility
in granulation will be reviewed.
An exhaust blower or fan is situated at the downstream end of the fluid bed
processor to draw the air through the entire unit. This arrangement provides negative
pressure in the fluid bed, which is necessary to facilitate material loading, maintain
safe operation, prevent material escape, and carry out the process under good
manufacturing practice guidelines, all of which will be discussed later in the chapter.
3.1. Air Handling Unit
A typical air preparation system includes sections for prefiltering air, air heating, air
dehumidification, rehumidification, and final high-efficiency particulate air (HEPA)
filtering. Generally, outside air is used as the fluidizing medium in a fluid bed processor.
For the air to be used for pharmaceutical products, it must be free from dust and
contaminants. This is achieved by placing coarse dust filters (30–85%) in the air
Batch Fluid Bed Granulation 251
(usually laden with moisture) exit the unit void of the product. Figure 3 shows a typihandling
unit (AHU). Figure 4 shows a typical AHU.
© 2005 by Taylor & Francis Group, LLC
Figure 3 Fluid bed processor installation with all components.
Figure 4 Typical air handling unit.
252 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
Through years of experience and dealing with various types of materials and
various climatic conditions, it is known that incoming air must be controlled very
closely. As an example, it has been found that the humidity of the incoming air
can greatly affect the quality of spray granulation, drying, or coating. Therefore,
air preparation systems are now designed to better control the conditions of the
incoming air. After the installation of the filters, distinct heating or cooling sections
are installed in the air handler, depending on the geographical location of the plant.
In an extremely cold climate, where cooling coils (needed in summer months for
maintaining uniform dew point) can freeze in winter, a preheating section is placed
ahead of the cooling coils. A typical range for the air after pretreatment that one
should aim at achieving is 15–30C dry bulb and 3–5C wet bulb. If the unit is
located in a tropical or humid climate, the humidity removal section is employed
first. The dehumidification of the air is extremely important where the outside air
moisture varies over a wide range. In summer, when the outside humidity is high,
dehumidification of the process air is required to maintain a specific dew point of
the incoming process air. Rehumidification may be necessary during the winter
months in some regions. A steam injector is used for rehumidifying the dry air. Generally,
lower the process air dew point, higher the affinity to entrain moisture and
shorter the process time. When granulating extremely fine powders, inlet air dew
point of 15C is beneficial to reduce static charges and facilitate uniform fluidization.
In many processes, when preheating is required, a bypass loop can be used for preconditioning
the air. This loop allows the required process temperature and humidity
to be attained within the system ducts before the product is subjected to fluidization.
After the conditioned air leaves the humidification/dehumidification section of the
AHU, it is finally heated to the desired process air temperature and then passed
through an HEPA filter of about 99.90–99.99% capacity. As the process air is treated
and filtered, it is transported by the inlet duct. The air is thus brought into the
process vessel in the lower plenum.
3.2. Product Container and Air Distributor
With the air at the desired humidity and temperature, it is ready to be passed through
The air must be introduced evenly at the bottom of the product container,
through an inlet air plenum. Proper airflow in the inlet air plenum is critical to ensure
that equal airflow velocities occur at every point on the air distributor plate. If the air is
not properly distributed before it reaches the bottom of the container, uneven fluidization
can occur. To facilitate the even flow of powder in the product container, conditioned
air is brought in the plenum at various locations by certain manufacturers.
To properly fluidize and mix the material in the container, correct choice of the
container and air distributor must be made. The container volume should be chosen,
such that the container is filled to at least 35–40% of its total volume and not more
than 90% of its total volume. Correct choice of the air distributor is important. These
distributors are made of stainless steel and are available with a 2–30% open area.
Typically, the distributor should be chosen such that the pressure drop across the
product bed and air distributor is 200–300mm of water column. A fine screen of
60–325 mesh normally covers the air distributor and retains the product in the
container. This type of sandwiched construction has been used for the last 30 years
in the fluid bed processors. The classic air distributor with the fine product retaining
screen is shown in Figure 5.
Batch Fluid Bed Granulation 253
the bed of solids. Figure 5 shows a typical product container with the air distributor.
© 2005 by Taylor & Francis Group, LLC
Keeping the screen and air distributors clean has been challenging. Partially to
address the cleaning problems and partially to provide an efficient processing, a overlap
gill air distributors eliminate the need for a fine screen and perform dual functions as
efficient air distributors and product retainers. Other advantages claimed by the manufacturer
are validatable clean in place (CIP), controlled fluidization, and directional
flow of air to discharge the processed product from the container. Because there is no
fine screen, these types of air distributors sometimes sift very fine particles through
the plate, thus losing part of the batch in the plenum. This sifting of fine powder through
these types of air distributors is of concern when a container containing product is
moved around on the production floor, losing some product due to movement of the
container. Before selecting these types of single-plate distributors, product particle size
and a sifting test should be performed. But these types of air distributors offer advantages
when the airflow can be directed to discharge the granulated product from the
3.3. Spray Nozzle
A spray is a zone of liquid drops in a gas, and spraying is the act of breaking up a
liquid into a multitude of these droplets. The general purpose of spraying is to
increase the surface area of a given mass of liquid, in order to disperse it over the
product area. The primary concern is with the increase of surface area per unit mass
achieved by spraying. The nozzle is an orifice through which liquid is forced,
normally by compressed air. This is done by one of three general methods: (a) liquid
may be sucked up by a pressure drop created over the nozzle cap, after which
compressed air atomizes the liquid stream by disintegrating it with air jets (b) the
compressed air operates a piston arrangement that pushes the liquid through the
orifice and then lets surface tension create droplets or (c) two pressure streams of
liquid impinge upon each other, and so form a highly dispersed, uniform spray.
Figure 5 Product container with air distributor.
254 Parikh and Mogavero
gill plate, shown in Figure 6(A) and (B), was introduced in 1990 (28). These new overlap
© 2005 by Taylor & Francis Group, LLC
container in a contained manner. (Please see the material handling options section.)
The type of spray system is usually characterized by one of four nozzle designs
1. Pressure nozzle: The fluid under pressure is broken up by its inherent instability
and its impact on the atmosphere, on another jet, or on a fixed plate.
2. Rotating nozzle (rotary atomizer): Fluid is fed at a low pressure to the center
of a rapidly rotating disk, and the centrifugal force breaks up the fluid.
These types of nozzles are used mainly in a spray drying application.
3. Airless spray nozzle: The fluid is separated into two streams that are
brought back together at the nozzle orifice, where upon impingement, they
form drops.
4. Gas atomizing nozzle (two-fluid nozzle): The two-fluid (binary) nozzle
where the binder solution (one fluid) is atomized by compressed air (second
C). These nozzles are available as single-port or multiport designs. Generally,
the single-port nozzles are adequate up to the 100 kg batch, but for
the spray undergoes three distinct phases. In the first, the compressed air
Figure 6 (A) Schematic of the overlap gill plate—gill arrangements. (B) Container with the
overlap gill distributor. (Courtesy of Niro Pharma Systems.)
Batch Fluid Bed Granulation 255
(29). Figure 7 shows four types of nozzles.
fluid) is the most commonly used nozzle for fluid bed granulation (Fig. 8A–
larger-size batches multiport nozzles, such as either three-port (Fig. 9) or
six-port (Fig. 10) nozzles are required. When these nozzles are air atomized,
© 2005 by Taylor & Francis Group, LLC
(gas) expands, essentially adiabatically, from the high pressure at the nozzle
to that at the fluid bed chamber. The gas undergoes a Joule–Thomson
effect, and its temperature falls. In the second, the liquid forms into discrete
drops. During this atomization, the liquid’s specific surface area usually
increases 1000 times. In the third, the drops travel after being formed, until
they become completely dry or impinge on the product particles. During
Figure 7 Types of nozzle. (From Ref. 29.)
Figure 8 (A) Schematic nozzle showing different parts. (B) Schematic of two-fluid nozzle.
(C) Typical single-port nozzle. (Courtesy of Niro Pharma Systems.)
256 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
this phase, the solvent evaporates and the diameter of the drops decreases.
The energy required to form a drop is the product of the surface tension and
the new surface area. About 0.1 cal/g energy is needed to subdivide 1 g of
water into 1 mm droplets. The air pressure required to atomize the binder
liquid is set by means of a pressure regulator. The spray pattern and spray
angle are adjusted by adjusting the air cap.
Optimum atomization is achieved by fine adjustment of the air cap and atomization
air pressure measured at the nozzle. The binder solution is delivered to the
nozzle port through a spray lance and tubing. The peristaltic or positive displacement
pump is commonly used to pump the binder solution. The pneumatically controlled
nozzle needle prevents the binder liquid from dripping when the liquid flow is
stopped. Nozzle port openings of between 0.8 and 2.8mm in diameter are most common
and are interchangeable.
The two-fluid nozzle in its simplified model is based on energy transmission as
shown below:
Energy ? Liquid
!Two fluid nozzle 
! Droplets ? Heat
The ratio of energy dissipation by heat and by the droplet-making process is
difficult to measure. Masters (30) suggested that <0.5% of applied energy is utilized
in liquid breakup. Virtually, the whole amount is imparted to the liquid and air as
kinetic energy.
Figure 9 Three-port nozzle. (Courtesy of the Vector Corporation.)
Figure 10 Six-port nozzle. (Courtesy of the Glatt Group.)
Batch Fluid Bed Granulation 257
© 2005 by Taylor & Francis Group, LLC
3.4. Disengagement Area and Process Filters
Once the air leaves the product bed, fine particles need to be separated from the air
stream. Two zones are used in the fluid bed to separate the particles from the air
stream—the disengagement area and the exhaust filter. In the disengagement area,
larger particles lose momentum and fall back into the bed. The velocity of the
process air is the highest at the center of the processor and approaches zero at
the sidewalls. A process-air filter system removes the particles from the exhaust
air. The process air is filtered by using bags or cartridges. The bag filters are widely
used and are available with single-bag or with double-bag configuration, where one
bag is mechanically shaking the particles while the other bag remains functional,
thus facilitating uninterrupted fluidization. This alternate shaking of dual bags
allows the process to be consistent from batch to batch. These filter bags can be constructed
out of nylon, polyester, polypropylene, and polytetrafluroethylene (PTFE)
lined materials (Figs. To dissipate the potential static charges from the
product particles, conductive fabrics are also available and are recommended.
Figure 11 Conventional bag filters. (Courtesy of the Glatt Group.)
Figure 12 Conventional bag filters with hanging arrangements. (Courtesy of the Vector
Corporation.)
258 Parikh and Mogavero
11–13).
© 2005 by Taylor & Francis Group, LLC
Cartridge filters lined with PTFE were introduced to the industry in the 1980s (31).
The standard filtration system normally contains a multiple-cartridge filter system
with an alternating blowback pulse arrangement allowing continuous product fluidization.
A cleanable polyester 2 mm material is utilized for processing water-soluble
and -insoluble materials, which has an electrical conductivity for static-free operation.
Recently, cartridges made of stainless steel suitable for CIP have been introduced
(32). Various suppliers of the process equipment have filter arrangements.
The vertical filter cartridge claimed to provide better cleaning, however, requires
mechanical means to bring the filters down to replace them. Cartridge filters located
at an angle do provide better access to take them out from and place them in the
unit. They are equally effective. and (B) shows the different cartridge-
filter arrangements in the fluid bed processor. The stainless steel cartridge filpossibility
of cleaning, using the automated CIP system. For a potent compound
processing these cartridge filters with a CIP capability is normally recommended.
During the granulation or drying process, cloth filters are mechanically shaken
to dislodge any product adhered, while cartridge filters use low-pressure compressed
air blowback system to do the same. Figure 14(A) and (B) shows various PTFE lined
cartridge filters used in the fluid bed processors.
3.5. Exhaust Blower or Fan
Once the air leaves the exhaust filters, it travels to the fan. The fan is on the outlet
side of the system, which keeps the system at a lower pressure than the surrounding
atmosphere. The airflow is controlled by a valve or damper installed just ahead or
after the fan. Manufacturers of the fluid bed normally make the selection of the fan,
Figure 13 Production size unit with filter bags. (Courtesy of the L.B. Bohle Group.)
Batch Fluid Bed Granulation 259
Figure 14(A)
ters (Fig. 15) are an expensive alternative to the cloth filter bags, but provide the
© 2005 by Taylor & Francis Group, LLC
based on the layout and the complexity of the system. Fan size is determined by calculating
the pressure drop (DP), created by all the components that make up the fluid
bed processor, including the product at the highest design airflow volume.
3.6. Control System
A fluid bed granulation process can be controlled by pneumatic analog control
devices, or state of the art, programmable logic controllers (PLCs) or computers.
The electronic based control system offers not only reproducible batches according
to the recipe but also a complete record and printout of all the process conditions.
Process control technology has changed very rapidly and it will continue to change
as advances in computer technology take place and as the cost of control systems
fall. The CFR Part 11 requirements (33) mandated by the US FDA has created a
number of approaches to ensure that these control systems are complying with the
Figure 14 (A) Cartridge filter during processing mode. (B) Cartridge filters during cleaning
mode. (Courtesy of the Vector Corporation.)
260 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
current regulation. The concept of process analytical technology (PAT) is covered in
another chapter of this book, as it pertains to the process control, and is discussed
further in this chapter, as it pertains to fluid bed processing in particular.
3.7. Solution Delivery System
The liquid delivery systems operate at a low pressure. A peristaltic pump capable of
delivering fluid at a controlled rate is desirable. The liquid is transported from the
solution vessel through the tubing, and atomized using a two-fluid (binary) nozzle
in the fluid bed processor.
3.8. Laboratory Units
During feasibility of process development, the supply of active ingredient is scarce.
A smaller unit with all the functionalities of a larger unit is desirable. The process
parameters developed at this stage may not be fully scalable; however, one can proused
for process feasibility and preliminary development work.
4. PARTICLE AGGLOMERATION AND GRANULE GROWTH
Agglomeration can be defined as the size enlargement process, in which the starting
material is fine particles and the final product is an aggregate in which primary particles
can still be identified. The granules are held together with bonds formed by the
binder used to agglomerate. Various mechanisms of granule formation have been
described in the literature (34–36). The chapter on the theory of granulation in this
book discusses the theory of granule growth. To summarize, the researchers have
suggested three mechanisms for granule formation. These are
1. Bridges due to immobile liquids form adhesional and cohesional bridging
bonds. Thin adsorption layers are immobile and can contribute to the
bonding of fine particles under certain circumstances.
2. Mobile liquids where interfacial and capillary forces are present.
3. Solid bridges formed due to crystallization of dissolved substances during
drying.
Figure 15 Stainless steel filters. (Courtesy of the Glatt Group.)
Batch Fluid Bed Granulation 261
cess 20–100 g of material. Figures 16–18 show various smaller-size units that can be
© 2005 by Taylor & Francis Group, LLC
The type of bonds formed approaches through four transition states, described
by Newitt and Conway-Jones (34) as (1) pendular, (2) funicular, (3) capillary, and (4)
droplet, which normally happens during spray drying.
Tardos et al. (37) investigated comprehensive model of granulation. They
developed a pendular bridge apparatus that can be used to test the bridge-forming
characteristics of the binder, and to determine binder penetration and spreading
rates and the critical time of binder strengthening. Iveson (38) reported a mathema-
Figure 16 Tabletop unit. (Courtesy of Pro-cept.)
Figure 17 Tabletop fluid bed processor. (Courtesy of the Glatt Group.)
262 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
tical model for granule coalescence during granulation. He found that current models
had one of two limitations; either they only consider whether a bond formed on impact
is strong enough to survive subsequent impacts or they fail to consider the possibility
of bond rupture after formation at all. He developed a new model that takes into
account both the effects of bond strengthening with time and the distribution of
impact forces. He suggested that his model be combined with existing models that
predict whether or not two granules stick initially on impact, to be able to predict
the probability of permanent coalescence.
Most of the fluid bed granulated products require much less wetting than the
high-shear granulation or spray dryer processed product. In the fluid bed granulation
process, the particles are suspended in the hot air stream and the atomized liquid
is sprayed on it. The degree of bonding between these primary particles to form an
agglomerated granule depends on the binder used, the physicochemical characteristics
of the primary particles being agglomerated, and the process parameters.
Schaefer et al. (39) and Smith and Nienow (40) have reported a description of
the growth mechanisms in the fluid bed, where the bed particles are wetted by liquid
droplets in the spray zone. Atomized liquid from the nozzle tends to spread over the
particle surface, as long as there is an adequate wettability of the particle by the fluid
(41) Wet particles, on impact, form a liquid bridge and solidify as the agglomerate
circulates throughout the remainder of the bed. Solid bridges then hold particles
together. The strength of the binder determines whether these particles stay as
agglomerates. These binding forces should be larger than the breakup forces and
in turn depend on the size of the solid bridge. The breakup forces arise from the
movement of the randomized particles colliding with each other and are related to
the excess gas velocity and particle size.
If the binding forces are in excess of the breakup forces, either in the wet state or
in the dry state, uncontrolled growth will proceed to an overwetted bed or production
of excessive fines, respectively. If a more reasonable balance of forces is present, con-
Figure 18 Lab unit. (Courtesy of Heinen, Germany.)
Batch Fluid Bed Granulation 263
© 2005 by Taylor & Francis Group, LLC
trolled agglomeration will occur, the growth of which can be controlled. Maroglou
and Nienow presented a granule growth mechanism in the fluid bed by the use of
model materials and scanning electron microscope (42). Figure 19 shows the various
paths a liquid droplet can take and their consequences on the particle growth.
The mechanism of formation of a granule and subsequent growth primarily
progresses through three stages:
1. Nucleation
2. Transition
3. Ball growth
beginning of the spraying stage, primary particles form nuclei and are held together
by liquid bridges in a pendular state. The size of these nuclei depends on the droplet
size of the binder solution. As the liquid addition continues, more and more nuclei
Figure 19 Mechanism of granulation in fluid bed.
264 Parikh and Mogavero
Figure 20 shows the growth of the granule relative to the liquid added. In the
© 2005 by Taylor & Francis Group, LLC
agglomerate and continue the transition from the pendular state to the capillary
state.
The uniqueness of the fluid bed agglomeration process is in, how the liquid
addition and drying (evaporation) steps are concurrently carried out. When the granulation
liquid is sprayed into a fluidized bed, the primary particles are wetted and
form together with the binder relatively loose and very porous agglomerates. Densi-
fication of these agglomerates is brought about solely by the capillary forces present
in the liquid bridges. It is therefore important that the quantity of liquid sprayed into
the bed should be relatively large compared with that used in high-shear granulation.
Drying a wet product in a fluid bed is a separate topic, but during the granulation
process it becomes an integral part of the process; hence, understanding fluid
bed drying is important before we review the agglomeration process.
5. FLUID BED DRYING
Drying is usually understood to be the removal of moisture or solvent. Drying
involves heat transfer and mass transfer. Heat is transferred to the product to evaporate
liquid, and mass is transferred as vapor in the surrounding gas; hence, these
two phenomena are interdependent. The drying rate is determined by the factors
affecting the heat and mass transfer. The transfer of heat in the fluid bed takes place
by convection. Convection is the transfer of heat from one point to another within a
fluid (gas, solid, liquid), by the mixing of one portion of the fluid with another.
The removal of moisture from a product granulated in the fluid bed granulator or
in other equipment essentially removes the added water or solvent. This free moisture
content is the amount of moisture that can be removed from the material by drying at
a specified temperature and humidity. The amount of moisture that remains
associated with the material under the drying conditions specified is called the
equilibrium moisture content (EMC).
Figure 20 States of liquid saturation. Liquid bridging state of agglomerates undergoing
(A) binding liquid addition and (B) densification.
Batch Fluid Bed Granulation 265
© 2005 by Taylor & Francis Group, LLC
The evaporation rate of liquid film surrounding the granule being dried is
related to the rate of heat transfer by the equation:
dw
dt ? h
A
H
dT
where dw/dt is the mass transfer rate (drying rate), h is the heat transfer coefficient,
A is the surface area, H is the latent heat of evaporation, and dT is the temperature
difference between the air and the material surface.
Because fluid bed processing involves drying of a product in suspended hot air,
the heat transfer is extremely rapid. In a properly fluidized processor, the product
temperature and the exhaust air temperatures should reach equilibrium. Improper
air distribution and hence poor heat transfer in a fluidized bed causes numerous problems
such as caking, channeling, or sticking. The capacity of the air (gas) stream to
absorb and carry away moisture determines the drying rate and establishes the
duration of the drying cycle. Controlling this capacity is the key to controlling the
drying process. The two elements essential for this control are inlet air temperature
and airflow. The higher the temperature of the drying air, the greater its vapor
holding capacity. Since the temperature of the wet granules in a hot gas depends
on the rate of evaporation, the key to analyzing the drying process is psychrometry
(44–46).
Psychrometry is defined as the study of the relationships between the material
plify the crucial calculations of how much heat must be added and how much moisture
can be added to the air. The process of drying involves both heat and mass
transfer. For drying to occur, there must be a concentration gradient, which must
exist between the moist granule and the surrounding environment. As in heat transfer,
the maximum rate of mass transfer that occurs during drying is proportional to
the surface area, the turbulence of the drying air, the driving force between the solid
and the air, and the drying rate. Because the heat of vaporization must be supplied to
evaporate the moisture, the driving force for mass transfer is the same driving force
required for heat transfer, which is the temperature difference between the air and
the solid.
Scha?fer and Worts (47) have shown that the higher the temperature difference
between the incoming air and the product, the faster the drying rate. Therefore, product
temperature should be monitored closely to control the fluidized bed drying
process.
During fluid bed drying, the product passes through three distinct temperature
the ambient temperature to approximately the wet-bulb temperature of the air in the
dryer. This temperature is maintained until the granule moisture content is reduced
to the critical level. At this point, the material holds no free surface water, and the
temperature starts to rise further.
The drying capacity of the air depends on the relative humidity (RH) of the
incoming air. At 100% RH, the air is holding the maximum amount of water possible
at a given temperature, but if the temperature of the air is raised, the RH drops, and
the air can hold more moisture. If air is saturated with water vapor at a given temperature,
a drop in the temperature will force the air mass to relinquish some of its
moisture through condensation. The temperature at which moisture condenses is the
dew point temperature. Thus, the drying capacity of the air varies significantly during
processing. By dehumidifying the air to a preset dew point, incoming air can be
266 Parikh and Mogavero
and energy balances of water vapor–air mixture. Psychometric charts (Fig. 21) simphases
(Fig. 22). At the beginning of the drying process, the material heats up from
© 2005 by Taylor & Francis Group, LLC
maintained at a constant drying capacity (dew point) and hence provide reproducible
process times.
Julia Gao et al. (48) studied the importance of inlet air velocity to dry the
product granulated in a high-shear granulator and dried in a fluid bed dryer. The
manufacturing process involved granulating the dry components containing 63%
water-insoluble, low-density drug in a high-shear granulator, milling the wet mass,
and drying in a fluid bed dryer. The granules were dried at an inlet air temperature
of 60C. Two different air velocities were examined for their effect on drying uniformity
of the product. The authors observed that the excessive velocity, indicated by
Figure 21 Psychrometry chart.
Figure 22 Product temperature changes during drying. (From Ref. 21.).
Batch Fluid Bed Granulation 267
© 2005 by Taylor & Francis Group, LLC
the rapid rise in the exhaust air temperature resulted in nonuniform drying of the
product besides resulting in an inefficient process.
6. PROCESS AND VARIABLES IN GRANULATION
6.1. Process
As with any granulating system, in fluid bed granulation processing, the goal is to
form agglomerated particles through the use of binder bridges between the particles.
To achieve a good granulation, the particles must be uniformly mixed, and the liquid
bridges between the particles must be strong and easy to dry. Therefore, this system
is sensitive to the particle movement of the product in the unit, the addition of the
liquid binder, and the drying capacity of the air. The granulation process in the fluid
bed requires a binary nozzle, a solution delivery system, and compressed air to atomize
the liquid binder. Figure 23 shows the equipment setup for granulation using
the fluid bed processor.
Thurn (49), in a 1970 thesis, investigated the details of the mixing, agglomerating,
and the drying operations, which take place in the fluid bed process. Results
indicated that the mixing stage was particularly influenced by the airflow rate and the
air volume. It was suggested that the physical properties of the raw materials, such as
hydrophobicity, might exert a strong influence on the mixing stage. At the granulation
stage, particular attention was paid to the nozzle, and it was concluded that a
Figure 23 A typical fluid bed processor setup for fluid bed granulation.
268 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
binary design (two-fluid) nozzle gave a wide droplet size distribution, yielding a
homogenous granule. The need for strong binders was recommended to aid granule
formation and it was suggested that the wettability of the raw materials required
particular attention. Several research papers have been published, on the influence
of raw material (47–65), binder type (5,8,47,49,59,61,66–76), binder concentration,
and binder quantity (8,52,57,61,63,66,68–70,73–75,77–93). Recently, researchers
granulated various products using foam instead of solution and found that fluid
bed granulation using foams of aqueous solutions of low molecular weight Methocel
hypromellose polymers (E3PLV and E6PLV) or conventional solution did not have
an effect on the physical properties of granules or tablets compressed from these
granules. However, they found that due to foam, the granule formation is achieved
more efficiently. It was further claimed that variables associated with nozzles were
eliminated by using foam, and water requirement was reduced along with shorter
production time (94).
Each phase of the granulation process must be controlled carefully to achieve
process reproducibility. When the binder liquid is sprayed into a fluidized bed, the
primary particles are wetted and form together with the binder, relatively loose and
very porous agglomerates. Densification of these agglomerates is brought about,
almost solely by the capillary forces present in the liquid bridges. It is therefore
important that the liquid binder sprayed into the bed should be relatively large in
quantity, compared with that used in high- or low-shear granulation processes. During
spraying, a portion of the liquid is immediately lost by evaporation, so the system
has little tendency to pass beyond the liquid bridge phase. The particle size of the
resulting granule can be controlled to some extent by adjusting the quantity of binder
liquid and the rate at which it is fed, i.e., the droplet size. The mechanical
strength of the particles depends principally on the composition of the primary product
being granulated and the type of the binder used. Aulton et al. (83) found that
lower fluidizing air temperature, a dilute solution of binder fluid, and a greater spray
rate produced better granulation for tableting.
6.2. Variables
Factors affecting the fluid bed granulation process can be divided into three broad
categories:
1. Formulation related variables
2. Equipment related variables
3. Process related variables
6.2.1. Formulation Related Variables
6.2.1.1. Properties of Primary Material. Ideally, the particle properties
desired in the starting material include a low particle density, a small particle size,
a narrow particle size range, the particle shape approaching spherical, a lack of particle
cohesiveness, and a lack of stickiness during the processing. Properties such as
cohesiveness, static charge, particle size distribution, crystalline or amorphous nature,
and wettability are some of the properties which have an impact on the properties
of granules formed. The cohesiveness and static charges on particles present
fluidization difficulty. The same difficulties were observed when the formulation contained
hydrophobic material or a mixture of hydrophilic and hydrophobic materials.
The influence of hydrophobicity of primary particles has been shown by Aulton and
Batch Fluid Bed Granulation 269
© 2005 by Taylor & Francis Group, LLC
Banks (17) where they demonstrated that the mean particle size of the product was
directly related to wettability of the primary particles expressed as cos y (where y is
the contact angle of the particles). It was also reported that as the hydrophobicity of
the mix is increased, a decrease in granule growth is observed. Aulton, Banks, and
Smith, in a later publication, showed that addition of a surface-active agent, such
as sodium laurel sulfate, improves the fluidized bed granulation (57). In a mixture
containing hydrophobic and hydrophilic primary particles, granule growth of hydrophilic
materials takes place selectively, creating content uniformity problems. Formulating
a controlled release granulation can be accomplished by using fluid bed
granulation. A controlled release matrix formulation of Naproxen was successfully
developed using fluid bed granulation (95).
6.2.1.2. Low-Dose Drug Content. Wan et al. (96) studied various methods of
incorporating a low-dose drug, such as chlorphenirmine maleate, in lactose formulation
with Povidone (PVP) as the granulating binder. They concluded that the randomized
movement of particles in the fluid bed might cause segregation of the drug and
that uniform drug distribution was best achieved by dissolving the drug in the granulating
solution. The mixing efficiency of drug particles with the bulk material was
found to increase, in the proportion of the granulating liquid used to dissolve the
drug. The optimum nozzle atomizing pressure was deemed to be important to avoid
spray drying the drug particles or overwetting, which creates uneven drug distribution.
Higashide et al. (97) studied the fluidized bed granulation using 5-fluorouracil
in a concentration of 0.3% in 1:1 mixture of starch and lactose. Hydroxy propyl cellulose
(HPC) was used as the binder. The ratios of starch and lactose contained in
the granules were measured gravimetrically. The researchers found that a larger
amount of the drug and starch was found in larger granules than in smaller granules.
The results were attributed to the hydrophobicity of the 5-fluorouracil, starch, and
the hydrophilicity of lactose.
6.2.1.3. Binder. A more general discussion on the types of binders used in
the pharmaceutical granulations and their influence on the final granule properties
was presented in a previous chapter of this book. Different binders have different
binding properties, and the concentration of the individual binder may have to be
changed to obtain similar binding of primary particles. Thus, the type of binder, binder
content in the formulation, and concentration of the binder have a major influence
on granule properties. These properties include friability, flow, bulk density,
porosity, and size distribution.
Davies and Gloor (98,99) reported that the types of binder such as povidone,
acacia, gelatin, and HPC, all have different binding properties that affect the final
granule properties mentioned above. Hontz (91) investigated microcrystalline cellulose
concentration, inlet air temperature, binder (PVP) concentration, and binder
solution concentration effects on tablet properties. Binder and microcrystalline cellulose
concentration were found to have a significant effect on tablet properties.
Alkan and Ulusoy (76) studied binder (PVP) addition in solution and as a dry powder
in the powder mix. They found a larger mean granule size when the dry binder
was granulated with ethanol. However, when the binder was in solution, the granules
produced were less friable and more free-flowing. A similar finding was confirmed by
other researchers (92,93). Binder temperature affects the viscosity of the solution,
and in turn affects the droplet size. Increased temperature of the binder solution
reduces the viscosity of the solution, reducing the droplet size, and hence producing
smaller mean granule size. Binder solution viscosity and concentration affect the
droplet size of the binder. Polymers, starches, and high molecular weight PVP cause
270 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
increased viscosity, which in turn creates a larger droplet size and subsequently a larger
mean granule particle size (68).
Diluted binders are preferred because they facilitate finer atomization of the
binder solution, provide control of the particle size, reduce friability, and increase the
bulk density, even though the tackiness or binding strength may suffer (8,69,79,83,99).
6.2.1.4. Binder Solvent. In most instances, water is used as the solvent. The
selection of solvent, such as aqueous or organic, depends on the solubility of the binder
and the compatibility of the product being granulated. Generally, organic solvents,
due to their rapid vaporization from the process, produce smaller granules
than the aqueous solution. Different solvents have different heats of vaporization
as shown in Table 1. Incorporating the binder or a mixture of binders of low melting
point with the drug substance in the dry form can eliminate the requirement of a solvent
for the binder. The temperature of the incoming air is sufficient to melt the binder
and form the granules. Seo et al. (100) studied fluid bed granulation using
meltable polymers such as polyethylene glycol (PEG) 3000, or esters of polyethylene
glycol and glycerol (Gelucire 50/13). They showed that melt agglomeration by atomization
of a melted binder in a fluid bed occurs by initial nucleation followed by
coalescence between nuclei. The nuclei are formed by immersion of the solid particles
in the binder droplets, provided that the droplet size is larger than the size of the
solid particles. The agglomerate growth rate is supposed to be practically independent
of the droplet size, if the binder viscosity is so low that the droplets are able
to spread over the agglomerate surface. If the droplets are unable to spread, because
of high viscosity, the growth rate is supposed to be inversely proportional to the droplet
size. These effects of droplet size are different from those seen in aqueous fluid
bed granulation, probably because the aqueous process is affected by the evaporation
of the binder liquid.
6.2.2. Equipment Related Variables
6.2.2.1. Design. To fluidize, and thus granulate and dry the product, a certain
quantity of process air is required. The volume of the air required will vary
based on the amount of material that needs to be processed. The ratio of drying
capacity of the process air to the quantity of the product needs to be maintained constant
throughout the scaling-up process. However, some suppliers of the equipment
provide higher drying capacity for their laboratory unit, but cannot maintain the
same ratio for the production units. This lack of proportionality reduces the drying
capacity per unit volume of the process air, resulting in a longer process time in the
production units. The current design of the fluid bed is a modular one, where multiple
Table 1 Heats of Vaporization for Commonly Used Solvents
Solvent
Solvent boiling point
(C)
Density
(g/mL)
Heat of vaporization
(kcal/g)
Methylene chloride 40.0 1.327 77
Acetone 56.2 0.790 123.5
Methanol 65.0 0.791 262.8
Ethanol 78.5 0.789 204.3
Isopropanol 82.4 0.786 175.0
Water 100.0 1.000 540.0
Batch Fluid Bed Granulation 271
© 2005 by Taylor & Francis Group, LLC
processes such as drying, granulating, Wurster coating, rotary fluid bed granulating
or coating, etc., can be carried out by changing the container specially designed for
the individual process.
6.2.2.2. Air Distributor Plate. The process of agglomeration and attrition
due to random fluidization requires control of the particle during the granulation
process. This is a complex phenomenon due to the prevailing fluidizing conditions
and a particle size distribution, which undergoes changes during the process. As
the conditioned air is introduced through the lower plenum of the batch fluid bed,
the fluidizing velocity of a given volume of air determines how fluidization will be
achieved.
Perforated air-distributor plates covered with the 60–325-mesh fine stainless
steel screen, described previously, provide an appropriate means of supplying air
to the product. These plates are identified by their percentage of open area. Air distributor
plates that have 4–30% open area are normally available. These interchangeable
plates provide a range of loading capacities, so that batches of various sizes can
be produced efficiently and with uniform quality. To prevent channeling, an operator
can select a plate with optimum lift properties. For example, a product with low bulk
density requires low fluidizing velocity. A distributor plate having a small open area
to give a large enough pressure drop may provide uniform fluidization of such a product
without reaching entraining velocity and impinging the process filters. Alternatively,
a product with higher bulk density can be fluidized and processed using a plate
with a larger open area. The air distributor plate consists of a perforated plate and a
fine mesh screen. This arrangement sometimes causes problems like product leakage,
due to a torn screen, and difficulty in cleaning without separating the perforated plate
and the fine mesh screen. To overcome these deficiencies, various air distributor
design have been recently introduced. These distributor designs were discussed earlier
in the chapter.
6.2.2.3. Pressure Drop. The blower creates flow of air through the fluid bed
processor, or a fan located downstream from the process chamber. This fan imparts
motion and pressure to air using a paddle-wheel action. The moving air acquires a
force or pressure component in its direction of motion, because of its weight and inertia.
This force is called velocity pressure and is measured in inches or millimeters of
water column. In operating duct systems, a second pressure that is independent of
air velocity or movement is always present. Known as static pressure, it acts equally
in all directions. In exhaust systems such as fluid bed processors, a negative static pressure
exists on the inlet side of the fan. The total pressure is thus a combination of static
and velocity pressures. Blower size is determined by calculating the pressure drop (DP)
created by all the components of the fluid bed processing system. Proper selection of
the blower is essential in fluid bed design. A blower with appropriate DP will fluidize
the process material adequately. However, a blower without enough DP will not allow
proper fluidization of the product, resulting in longer process time and improper granulation.
A similar effect can be seen when a product with unusually high bulk density
is processed in place of normal pharmaceutical materials, or an air distributor offers
high resistance due to its construction. This creates a pressure drop that the blower
was not designed to handle. A properly sized blower or fan should develop sufficient
DP so that the exhaust damper can be used in the 30–60% open position. Any additional
components such as scrubbers, exhaust HEPA, police filters, or additional components
in the air handling unit would require a larger blower/static pressure which
can be recommended by the supplier of the fluid bed processor.
272 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
6.2.2.4. Shaker/Blow Back Cycle Mechanism. To retain entrained particles
of a process material, process filters are used. To maintain these filters from building
up layers of fine process material, and causing a higher pressure drop and thus
improper fluidization, these filters are cleaned during the granulation process. When
bag filters are used, mechanical means are used to clean them. This mechanical cleaning
of the bag filters requires a cessation of airflow and thus the fluidization during
the filter cleaning process. In units with a single-bag house, this results in a momentary
dead bed, where no fluidization takes place. This interruption in the process
extends the process time. To avoid process interruptions, a multishaking filter bag
arrangement is desired, where the granulation process is continuous. Using bag filters
with a blowback or using cartridge filters where air under pressure is pulsed
through the filters also achieves the continuous process. Generally, filters should
be cleaned frequently during the granulation step, to incorporate the fines back in
the granulation. This is possible if the cleaning frequency is high and the period
between the filter cleaning is short. Rawley (101) reported the effect of bag-shake/
interval cycle. He discussed the possibility of improving particle size distribution
by optimizing the shake time and the corresponding interval between bag shakes.
The following general guidelines for filter cleaning frequency and duration are
recommended:
Single-bag shaker unit: Frequency, 2–10 min between filter cleaning, 5–10 sec
for shaking. This may vary as the fine powders form granules, and the
frequency between the shakes or duration of shaking interval, can be
extended. In any case, the occurrence of a collapsed bed should be kept
at a minimum in a single-shaker unit.
Multiple bags shaker unit: Since this is a continuous process, frequency of shaking
for each section is approximately 15–30 sec between filter cleanings,
and about 5 sec for shaking the filters. If a low-pressure blow back system
is used for the bags, the frequency of cleaning is about 10–30 sec.
Cartridge filters: These offer continuous processing and require cleaning frequency
of 10–30 sec.
The cleaning frequency and cleaning duration are now offered as an automated
system, where instead of having to base the cleaning frequency on time, the trigger
point for filter cleaning is the buildup of a pressure drop across the filters. This automates
the process and eliminates operator input.
6.2.2.5. Other Miscellaneous Equipment Factors. Granulator bowl geometry
is considered to be a factor that may have an impact on the agglomeration process.
The fluidization velocity must drop from the bottom to the top rim of the bowl by
more than half to prevent smaller, lighter particles from being impinged into the filter,
creating segregation from heavier product components in the bowl. Generally,
the conical shape of the container and expansion chamber is preferred, where the
ratio of cross-sectional diameter of the distributor plate to the top of the vessel is 1:2.
Most of the suppliers of this equipment offer units with a multiprocessor concept
where a single unit can be used for drying, agglomerating, air suspension coating,
or rotary fluid bed processing by changing the processing container, while the rest
of the unit is common. This approach does eliminate the concerns about the geometry
of the processor, because of the way these units are constructed.
Batch Fluid Bed Granulation 273
© 2005 by Taylor & Francis Group, LLC
6.2.3. Process Related Variables
The agglomeration process is a dynamic process where a droplet is created by a two-
fluid nozzle, and deposited on the randomly fluidized particle. The binder solvent
evaporates, leaving behind the binder. Before all of the solvent is evaporated, other
randomized particles form bonds on the wet site. This process is repeated numerous
times to produce the desired agglomerated product. There are a number of process
variables that control the agglomeration. Process variables, most important to consider
are listed as follows:
1. Process inlet air temperature
2. Atomization air pressure
3. Fluidization air velocity and volume
4. Liquid spray rate
5. Nozzle position and number of spray heads
6. Product and exhaust air temperature
7. Filter porosity and cleaning frequency
8. Bowl capacity
These process parameters are interdependent and can produce the desired product
if this interdependency is understood. Inlet process air temperature is determined
by the choice of binder vehicle, whether aqueous or organic, and the heat
sensitivity of the product being agglomerated. Generally, aqueous vehicles will
enable the use of temperatures between 60C and 100C. On the other hand organic
vehicles will require the use of temperatures from 50C to below room temperature.
Higher temperatures will produce rapid evaporation of the binder solution and will
produce smaller, friable granules. On the other hand, lower temperatures will produce
larger, fluffy, and denser granules.
let air humidity during the granulation process.
The process of drying while applying spraying solution is a critical unit operation.
This mass transfer step was previously discussed. The temperature, humidity,
and volume of the process air determine the drying capacity. If the drying capacity
of the air is fixed from one batch to the next, then the spray rate can also be fixed. If
the drying capacity of the air is too high, the binder solution will have a tendency to
spray dry before it can effectively form bridges between the primary particles. If on
the other hand, the drying capacity of the air is too low, the bed moisture level will
become too high and particle growth may become uncontrollable. This will result in
unacceptable movement of the product bed.
As previously discussed, the appropriate process air volume, inlet air temperature,
and binder spray rate are critical for achieving proper and consistent particle
size distribution and granule characteristics. There are many ways to arrive at the
proper operating parameters. The following procedure was found by the authors
to be one of the ways one can set the operating parameters when granulating with
fluid bed processors.
1. Determine the proper volume of air, to achieve adequate mixing and particle
movement in the bowl. Avoid excessive volumetric airflow so as to
entrain the particles into the filters.
2. Choose an inlet air temperature that is high enough to negate weather
effects (outside air humidity or inside room conditions). The air temperature
should not be detrimental to the product being granulated. (To
274 Parikh and Mogavero
Figure 24 shows the relationship of inlet and product air temperature, and out-
© 2005 by Taylor & Francis Group, LLC
achieve consistent process year round, a dehumidification/humidification
system is necessary, which provides the process air with constant dew
point, and hence constant drying capacity).
3. Achieve a binder solution spray rate that will not dry while spraying (spray
drying) and will not overwet the bed. This rate should also allow the nozzle
to atomize the binder solution to the required droplet size.
4. As stated earlier, a typical air velocity used for spray granulation is from
a first guess at determining the proper spray rate for a spray granulation
process in a fluid bed processor.
The presence of variables in the fluid bed granulation process, and their impact
on the final granulation, was summarized by Davies and Gloor (102), where they
state that the physical properties of granulation are dependent on both the individual
formulations and the various operational variables associated with the process. The
solution spray rate increase and subsequent increase in average granule size, resulted
in a less friable granulation, a higher bulk density, and a better flow property for a
lactose/corn starch granulation. Similar results were obtained by an enhanced binder
solution, decreasing nozzle air pressure, or lowering the inlet air temperature
during the granulation cycle. The position of the binary nozzle with respect to the
fluidized powders was also studied. It was concluded that by lowering the nozzle,
binder efficiency is enhanced, resulting in average granule size and a corresponding
decrease in granule friability.
Figure 24 Temperature and humidity changes during the granulation process.
Batch Fluid Bed Granulation 275
1.0 to 2.0 m/sec. Table 2, which is based on the psychometric chart, gives
© 2005 by Taylor & Francis Group, LLC
The significant process parameters and their effect on the granule properties
Maroglou (43) listed various parameters affecting the type and rate of growth
meters and the material parameters on the product.
7. PROCESS CONTROLS AND AUTOMATION
The agglomeration process is a batch process, and accurate repeatable control of all
critical process parameters is necessary for a robust system. Earlier designs of the
fluid bed processor used pneumatic control, which provided safe operation in a
hazardous area but relied heavily on human actions to achieve repeatable product
quality and accurate data acquisition. Current designs use PLCs and personal computers
(PCs) to achieve sophisticated control and data acquisition. The operating
conditions are controlled to satisfy parameters of multiple user-configured recipes,
and critical data are collected at selected time intervals for inclusion in an end-ofbatch
report. Security levels protect access to all user-configured data with passwords
permitting access only to selected functions. With the appropriate security
level, not only are operating conditions configured, but also identification of each
Table 2 Calculation of Fluid Bed Spray Rate
Given process data
Air volume range:
Minimum (1.2 m/sec) _________________ m3/hr
Maximum (1.8 m/sec) _________________ m3/hr
Inlet air temperature and humidity to be used: _______ C _______ %RH
% Solids in sprayed solution: _________________ % solids
From psychrometry chart
Air density at point where air volume is measured: ___________ m3/kg air
Inlet air absolute humidity (H): ___________ g H2O/kg air
Maximum outlet air absolute humidity (H): ___________ g H2O/kg air
(Follow line of constant adiabatic conditions)
Use 100% outlet RH for spray granulator or 30–60% RH (as required for column coating)
Calculations for spray rate
Step 1 Convert air volumetric rate to air mass rate
Minimum ________ m3/hr(60 ________ m3/kg air)? ________ kg air/min
Maximum _________ m3/hr(60 ________ m3/kg air)? ________ kg air/min
Step 2 Subtract inlet air humidity from outlet air humidity
___________________ (g H2O/kg air) Hout
___________________ (g H2O/kg air) Hin
? ____________________________________________________ g H2O removed/kg air
Step 3 Calculate (minimum and maximum) spray rate of solution
This will provide range of generally acceptable spray rates based on the airflow used
in the unit
Step 1 (minimum) _________ step 2 ________ [1
( ________ % solids100)]
? _______________________ spray rate (g/min) at minimum airflow
Step 2 (maximum) ________ step 2 ________ [1
( ________ % solids100)]
? _______________________ spray rate (g/min) at minimum airflow
276 Parikh and Mogavero
are summarized in Table 3.
in batch fluidized granulation (Table 4) and showed the influence of process para-
© 2005 by Taylor & Francis Group, LLC
valid recipe and operator is entered. The identification is verified before any operator
actions are permitted and is included with the end-of-run report. The use of computer
related hardware requires some additional validation, but with coordination
between the control system provider and the end user, the validation of software
(B) shows a PLC-based control panel with a typical operator screen.
The most important sensors for control of the drying process are product, inlet
and exhaust air temperature, and sensor for airflow measurement, located in the air
transport system. Other sensors for the spray agglomeration process are atomization
air pressure and volume, pressure drops (across the inlet filter; the product container
with the product being processed, and outlet process air filter), inlet air humidity or
dew point, process filter cleaning frequency and duration, spray rate for the binder
solution, and total process time.
All of these sensors provide constant feedback information to the computer.
These electronic signals may then be stored in the computer’s memory and then
recalled as a batch report. With this ability to recall data analysis, a greater insight
can be gained into the process.
Table 3 Significant Variables and Their Impact on the Fluid Bed Granulation Process
Process parameter Impact on process References
Inlet air temperature Higher inlet temperature produces finer
granules and lower temperature produces
larger stronger granules
(74,85)
Humidity Increase in air humidity causes larger
granule size, longer drying times
(39)
Fluidizing airflow Proper airflow should fluidize the bed
without clogging the filters. Higher airflow
will cause attrition and rapid evaporation,
generating smaller granules and fines
(17,20,74)
Nozzle and position A binary nozzle produces the finest droplets
and is preferred. The size of the orifice has
an insignificant effect, except when binder
suspensions are to be sprayed. Optimum
nozzle height should cover the bed surface.
Too close to the bed will wet the bed faster
producing larger granules, while too high
a position will spray dry the binder, create
finer granules, and increase granulation time.
(59)
Atomization air
volume and
pressure
Liquid is atomized by the compressed air. This
mass-to-liquid ratio must be kept constant to
control the droplet size and hence the granule size.
Higher liquid flow rate will produce larger droplet
and larger granule and the reverse will produce
smaller granules. At a given pressure an increase
in orifice size will increase droplet size
(39,59,87,94)
Binder spray rate Droplet size is affected by liquid flow rate, and
binder viscosity and atomizing air pressure and
volume. The finer the droplet, the smaller the
resulting average granules
(17,56,73,74,94)
Batch Fluid Bed Granulation 277
can be managed. Figure 25 shows the pneumatic control panel and Fig. 26(A) and
© 2005 by Taylor & Francis Group, LLC
7.1. Advances in Process Control and Automation
The degree of the instrumentation of pharmaceutical unit operations has increased.
This instrumentation provides information on the state of the process and can be
used for both process control and research. A central part of optimizing production
is increasing the level of automation. Besides monitoring the process parameters,
a number of approaches are being developed for measuring the moisture of the
product to determine the end point of the process, and consequently the in-process
particle size analysis. A number of publications discuss the on-line moisture
measurement and process end point determination using near-infrared (NIR).
7.1.1. Near-Infrared
The nondestructive character of vibrational spectroscopic techniques, such as NIR,
makes them novel tools for in-line quality assurance (103). NIR has been widely used
for the measurement of water in various applications (28). NIR can be applied both
for quantitative analysis of water and for determining the state of water in solid
material. This gives a tool for understanding the physicochemical phenomena during
manufacture of pharmaceutical granulation.
Table 4 Influence of Operating and Material Parameters on the Granulated Product
Operating parameters
Droplet size NARa
Atomization air velocity
Rheology
Surface tension
Nozzle position
Nozzle type
Bed moisture content Solution type and feed rate
Bed temperature
Fluidization velocity
Aspect ratio
Nozzle position and atomization velocity
Air distributor design
Jet grinding
Binder solution/suspension Concentration Bridge strength and size
Rheology
Material parameters
Binder solution/suspension Concentration Bridge strength and size
Rheology
Type of binder Molecular length and weight
Wettability Particle-solvent interaction
Surface tension
Viscosity
Material to be granulated Average particle size
Size distributiona
Shape and porosity
Drying characteristics
Density and density differencesb
aNAR is the ratio of air to liquid flow rates through the nozzle of a twin fluid atomizer expressed either in
mass units or in volume units (air at STP).
bEspecially important relative to elutriation and segregation.
278 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
The history of NIR dates back to the studies, by Herschel, in 1800. The modern
NIR analysis was developed in 1950 by the works of a group at United States
Department of Agriculture (USDA), headed by Karl Norris (28).
Other branches of chemical industry have also applied for various applications.
One feature of NIR is that the applications have been ahead of theoretical aspects.
This has hindered the general approval of NIR in the pharmaceutical industry. However,
the pharmacopoeias have defined some characteristics of analysis with NIR
(104,105). The NIR spectrum is just above the visible region of electromagnetic spectrum
(EMS). The NIR region covers the interval between 4000 and 12,500 cm
1
(0.8–2.5 mm). Molecules that absorb NIR energy vibrate in two fundamental modes:
stretching and bending. Stretching is a continuous change in the interatomic distance,
along the axis between two atoms, and it occurs at lower wavelengths than
bending vibration. A bending vibration is a change in the bond angle between diatomic
molecules. A band observed at 1940nm is known to be caused by O–H stretching
and bending vibrations, and is the most used analytically (106,107). It has
recently been reported, from measurements on silica gel layers, that water content
has an effect on NIR absorption at all wavelengths, even where water is absorbed
minimally (108). Usually, for a solid sample, the reflected light is the parameter measured
in NIR spectroscopy, known as diffuse reflectance. The reflected light consists
of two components: specular and diffuse. The specular, or mirror-like, component in
the boundary between two media occurs at the sample surface, and it contains little
information about the chemical composition of the substance. The NIR spectroscopy
is particularly based on the diffused component of the reflected light, and it can be
affected by particle size and shape distribution, bulk density surface characteristics,
and temperature (109,110). This portion of the Electro Magnetic Spectrum (EMS)
has, for the last 30 years, been studied and investigated in great detail as an analytical
tool for the analysis of many natural and man-made materials (109,111–114).
Developing a functional automation system requires new measuring techniques;
new in-line measuring devices are needed (115–119). Solid–water interactions
are one of the fundamental issues in the pharmaceutical technology. The state of
Figure 25 Pneumatic control panel. (Courtesy of the Glatt Group.)
Batch Fluid Bed Granulation 279
© 2005 by Taylor & Francis Group, LLC
water in the solid material may be characterized using x-ray diffraction, microscopic
methods, thermal analysis, vibrational spectroscopy, and nuclear magnetic resonance
spectroscopy (120). Traditionally, the control of fluidized bed granulation has been
based on indirect measurements. These control methods applied utilize the properties
of process air by Schaefer and Worts (47). Frake et al. (121) demonstrated the
use of NIR for in-line analysis of the moisture content in 0.05–0.07mm pellets during
spray granulation in fluid bed processor. Rantanen et al. (122,123) described a
similar approach for moisture content measurement using a rationing of three to
four selected wavelengths. He and his coworkers reported that the critical part
of in-line process was the sight glass for probe positioning, which was continuously
Figure 26 (A) PLC based system. (Courtesy of Niro Pharma Systems.) (B) Production unit
with a PLC control panel. (Courtesy of the L.B. Bohle Group.)
280 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
blown with heated air. They also reported spectra baselines caused by particle size and
refractive properties of the in-line samples; they analyzed several data pretreatments
to eliminate these effects on their fixed wavelength setup. Solvents other than water
have also been evaluated for real-time quantification.
On-line measurement has also been possible enabling the monitoring of film
coating on pharmaceutical pellets, in an industrial manufacturing process. Andersson
et al. (124) conducted measurements on solid coated tablets using fiber-optic
probe positioned in the fluid bed processor. In this case, they secured a representative
sampling during processing, by using a sample collector that was emptied with compressed
air inside the processor. Va?zquez recently provided comprehensive review of
FT-NIR application in measuring fluid bed drying end (125). Rantanen et al. (126)
used NIR to monitor the moisture as well as airflow. Using in-line multichannel NIR
the multivariate process data collected were analyzed using principal component
analysis (PCA). The authors showed that robust process control and measurement
system combined with reliable historical data storage can be used for analyzing
the fluid bed granulation process. PCA modeling proved a promising tool to handle
multidimensional data that were collected, and for reduction of the dimensionality of
process data. FT-NIR spectra gave useful information for understanding the phenomenon
during granulation. Rantanen et al. (127) further studied the application
of NIR for fluid bed process analysis. The authors used NIR to study moisture measurement
combined with temperature and humidity measurements. By controlling
the water during the fluid bed granulation, the granulation process was controlled.
They concluded that the varying behavior of formulations during processing can
be identified in a real-time mode. Thus, they found that NIR spectroscopy offered
unique information of granule moisture content during all phases of granulation.
7.1.2. Other Approaches for Process Control
7.1.2.1. Self-Organizing Maps. On-line process data are usually multidimensional,
and it is difficult to study with traditional trends and scatter plots. Rantanen
et al. (128) has suggested a new tool called ‘‘self-organizing maps’’ (SOM), for
dimension reduction and process state monitoring. As a batch process, granulation
traversed through a number of process states, which was visualized by SOM as a
two-dimensional map. In addition, they demonstrated how the differences between
granulation batches can be studied.
7.1.2.2. At-Line Measurement. Laitinen et al. presented a paper at a recent
conference (129) proposing at-line optical techniques to study particle size. Using
a CCD camera with optics and illumination units, with stabilized collimated light
beams, authors took two images of 36 granule samples by illuminating the samples
alternatively. Two digital images, with matrices of their gray scale values, were
obtained and the differences between the two matrices were calculated. This method
provided very rapid (1 min per sample) measurement of particle size, with a very
small sample size (<0.5 g).
7.1.2.3. Focused Beam Reflectance Measurement. This device uses a focused
beam of laser light that scans, in a circular path, across a particle or particle structure
passing in front of the window. Upon hitting the particle, light is scattered in all
directions. The light scattered back toward the probe is used to measure the chord
length or the length between any two points on a particle. Such devices are supplied
commercially and claim to be useful for monitoring on-line measurement of particle
size in the fluid bed granulation process.
Batch Fluid Bed Granulation 281
© 2005 by Taylor & Francis Group, LLC
7.1.2.4. Artificial Neural Network. Neural networks have been used by scientists
for optimizing formulations as an alternative to statistical analysis because of its
simplicity for use and potential to provide detailed information. The neural network
builds a model of the data space, which can be consulted to ask ‘‘what if’’ kinds of
questions. Recently, there has been interest in using artificial neural network (ANN)
for process control. Similar to the human brain, an ANN predicts events or information
based on learned pattern recognition. ANNs are computer systems developed to
mimic the operations of the human brain, by mathematically modeling its neurophysiological
structure (i.e., its nerve cells and the network of interconnections between
them). In ANN, the nerve cells are replaced by computational units called neurons,
and the strengths of the interconnections are represented by weights (130). This
unique arrangement can simulate some of the neurolocal processing ability of the
brain, such as learning and drawing conclusions from experience (131). Using the
process control system, quality assurance results, or energy usage data, an ANN
develops supervisory set points for the system. When ANN and process control systems
are used together, they form a product control system. Product control occurs
when a system measures defined product attributes in real time, and uses the knowledge
to adjust the control system. While process control system runs the process (i.e.,
fans, motors, heaters, etc.), the ANN controls the product (moisture level, consistency).
The fluidized bed processor’s process control system includes an operator interface,
sensing elements, and final control elements. The inputs in that case are inlet air
temperature, outlet air temperature, airflow rate, and energy consumption. Additional
contributing factors are the fouling coefficient of the dryer bags, the quantity of product
in the processor, and the type of product with its unique characteristics.
Watano et al. (132) described a practical method for moisture control in fluid bed
granulation by means of neural network. Wet granulation of pharmaceutical powder
was conducted using an agitation fluidized bed, and the moisture content was continuously
measured by IR moisture sensor. A neural network system for moisture control
was developed using moisture content and its changing rate as input variables, and the
moisture control characteristics were investigated by the neural network system with a
back propagation learning. Good response and stability without overshoot were
achieved by adopting the developed systems. This system also maintained favorable
stability under various operating conditions. Several researchers have published
papers, detailing the use of ANN for different applications (133–135).
8. PROCESS SCALE-UP
8.1. Regulatory
Scale-up is normally identified with an incremental increase in batch size, until a
desired level of production is obtained. In 1991, American Association of Pharmaceutical
Scientists (AAPS) with US FDA held a workshop on scale-up (136), where
several speakers presented scale-up issues from an industrial and regulatory perspective.
For example, Shangraw divided scale-up problems into two general categories:
those related to raw materials or formulation and those related to processing
equipment. He also indicated that it is essential to ascertain whether or not changes
in raw materials have occurred, before one looks at processing/equipment changes
as a source of any problem. The workshop report, as it pertains to the process
and equipment, is reproduced below:
282 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
It is generally recognized that many NDAs and ANDAs contain provision for
multiple manufacturers of the drug substance(s), and that not all drug substance
suppliers, a priori, produce equivalent material. There is then a need for material
quality control to assure the performance and reproducibility of the finished
product. Particle size and distribution, morphology, and intrinsic dissolution of
the drug substance are important considerations. Polymorphism, hygroscopicity,
surface area, wettability, density (bulk and tapped), compressibility (for dry
blending), and powder flow effects should be controlled.
Additionally, the process should be controlled by employment of a validation
protocol, which defines the critical parameters and also establishes the acceptance
criteria for the granulation or blend; which may include sieve analysis, flow,
density, uniformity, and compressibility, moisture content, etc.
In the milling, blending, granulating and or drying processes, the operating
principles of the equipment employed should be defined, and the variables determined.
The impact and mechanism of measurement on in-process variables should
be defined. Time, temperature, work input of equipment, blend/granulation
volume, and granulating rate should be determined. . . . The parameters selected
should be appropriate for the process, . . . In those cases where the manufacturing
process has been controlled and validated as specified in the foregoing discussion;
batch scale-up, changes in site of manufacture, allowance for equipment change
(where the operating principle is the same), minor formulation changes, etc.,
should be determined on the basis of the comparability of both the blend/granulation
and the final product; as assured by: a) appropriate tests; b) specifications; c)
process validation; and d) comparative accelerated stability.
Regulatory guidelines for scale-up and postapproval changes (SUPAC) IR
were released in 1995 (137) and are discussed in another chapter of this book.
8.2. Scale-Up and Equipment Design
The scale-up from the laboratory equipment to production size units is dependent on
equipment design, which may or may not have been scalable as far as its dimensional
feature or components selection is concerned. The importance of scalability is well
understood and accepted by the manufacturers of fluid bed processors. Various sizes
in their product line are logically designated and manufactured. Airflow in the fluid
bed process is a critical parameter. The design and selection of the processor is very
important for the laboratory and production unit. Because airflow is one of the components
of the drying capacity of a fluid bed system, the ratio of air volume per kilogram
or liter of the product is very critical to achieve a scale-up that is linear. The
other design feature is the cross-sectional area of the product container, and how
it has been designed throughout the various sizes that a manufacturer supplies.
The relationship between various sizes of the process containers can be utilized to
calculate the scale-up of binder spray rate, and if the cross-sectional area is designed
linearly, then the spray rate scale-up can be linear.
8.3. Scale-Up and Process Factors
The fluid bed agglomeration process is a combination of three steps, namely, dry
mixing, spray agglomeration, and drying to a desired moisture level. These process
steps are equally important. But the quality of the granules is really determined
during the spraying stage, when constant building of granules and evaporation of
Batch Fluid Bed Granulation 283
© 2005 by Taylor & Francis Group, LLC
binder solvent are taking place. Granule size is directly proportional to the
bed humidity during granulation (47) and hence, control of this humidity during
scale-up is essential.
Gore et al. (138) studied the factors affecting the fluid bed process during scale
up. The authors found that processing factors that affected granule characteristics
the most, were process air temperature, height of the spray nozzle from the bed, rate
of binder addition, and the degree of atomization of the binder liquid.
The atomizing air pressure and the wetness of the bed are two of the most
important elements of fluid bed granulation. A higher atomizing air pressure yields
a finer droplet of binder solution. Therefore, granule growth (as described earlier in
this section) is affected by the atomizing air pressure. A major factor which must be
considered during the scale up of fluid bed granulation process is maintaining the
same droplet size of the binder for assuring successful scale-up. A recent study
(139) confirmed the influence of spray nozzle setup parameters on the drying capacity
of the air. The study concluded that more attention should be paid to the easily
overlooked nozzle atomizing air pressure and volume. When considering the atomizing
air pressure, attention must be paid to ensure that enough air is delivered to the nozzle
tip. This can be ensured by placing air pressure and volume measurement devices at the
nozzle. The data also show that the drying capacity of the process air influences the
final granulated particle size.
Jones (140) has suggested various process related factors that should be considered
during the scale-up of a fluid bed processing. The suggestions are listed in the
following paragraphs.
Due to the higher degree of attrition in the larger unit compared to the smaller
unit, the bulk density of the granulation from the larger fluid bed is 20% higher
than the smaller unit. He also re-emphasized the importance of keeping the bed
moisture level below the critical moisture level, to prevent the formation of larger
agglomerates. Since the higher airflow along with the temperature (drying capacity)
in a larger unit provide higher evaporation rate, one must maintain the drying capacity
in the larger unit, such that the bed temperature is similar to the smaller-unit bed
temperature. This can be accomplished, either by increased spray rate, increased air
temperature, and increased airflow, or by the combination of these variables to
obtain suitable results. Since the ratio of bed depth to the air distributor increases
with the size of the equipment, the fluidization air velocity is kept constant by
increasing the air volume. In the past, the scale-up was carried out by selecting best
guess process parameters. The recent trend is to employ the factorial and modified
factorial designs and search methods. These statistically designed experimental plans
can generate mathematical relationships between the independent variables, such as
process factors, and dependent variables, such as product properties. This approach
still requires an effective laboratory/pilot scale development program and an understanding
of the variables that affect the product properties.
In summary, when scaling up, the following processing conditions should be
similar to the pilot scale studies:
1. Fluidization velocity of the process air through the system
2. The ratio of granulation spray rate to drying capacity of fluidization air
volume
3. Droplet size of the binder spray liquid.
284 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
Each of these values must be calculated based on the results of the operation of
the pilot size unit. Pilot size equipment studies should also be conducted in a wide
range to determine the allowable operating range for the process.
Another chapter of this book provides fluidized bed scale-up and should be
reviewed.
8.4. Case Study
The following case study illustrates how a product is scaled up from 15 to 150 kg in
the equipment supplied by Aeromatic, when one understands the critical process
parameters used when scaling up.
A spray granulation process was developed for a common pharmaceutical
compound. The granulation process involved spraying of a 5% w/w binder solution
onto the fluidized powder. Table 5 shows the data from the 15 kg run and the resulting
successful 150 kg run condition for a spray agglomeration process.
8.4.1. Airflow Calculations
To maintain the same fluidization velocity, the air volume in a larger unit must be
increased, based on the cross-sectional area of the product bowl. In this case, the
cross-sectional area of the base of the larger container was 0.77m2 and that of the
smaller was 0.06m2. The correct airflow should be calculated as 300(0.77/
0.06)?3850CMH. This number was further modified after considering the increase
in bed depth in a larger unit to 4000CMH.
8.4.2. Spray Rate Calculations
To maintain the same particle size, the triple-headed nozzle could spray three times
the pilot unit spray rate at a 2.5 atomization air pressure. However, this could result
in a longer process time. Another approach to maintain a similar droplet size is to
maintain the mass balance of spray rate and the atomization pressure. Thus, by
increasing the atomization pressure to 5 bar, the spray rate was increased to 800 g/
min, keeping the same droplet size, and hence obtaining granulation with the desired
characteristics.
8.4.3. Temperature Calculations
Finally, the required inlet temperature was recalculated based on the change in the
ratio of the air volume to the spray rate. Because the air volume was increased over
13 times but the spray rate was only increased eight times, the inlet temperature was
reduced to 50C. This adjustment in drying capacity was necessary to avoid spray
drying of the spray solution. (A three-headed nozzle used in this scale-up can be
Table 5 Scale-Up Process Parameters from a 15 to a 150 kg Batch
Process parameters 15 kg 150 kg
Airflow (m3/hr) 300 4000
Inlet air temperature (C) 55 50
Spray rate (g/min) 100 800
Nozzle air pressure (bar) 2.5 5
Container cross-sectional area of the base (m2) 0.06 0.77
Number of nozzles 1 3
Batch Fluid Bed Granulation 285
© 2005 by Taylor & Francis Group, LLC
replaced by a six-headed nozzle. This would have resulted in the ability to increase
the spraying rate 13 times above the pilot size unit to match the airflow. The maintenance
of droplet size and temperature could have been achieved with the
six-headed nozzle. The end result would be reduced process time.) Figure 27 shows
the particle size distribution produced, using 15 and 150 kg units.
Recently, Matharu and Patel (141) presented a scale-up case study, where a
low-dose multiple-strength product (0.5–5% w/w/ active) was spray granulated
and scaled up from a pilot scale fluid bed processor up to production size equipment.
Their approach was based on matching air velocity between the two scales of operation.
The impact of droplet size was determined by varying the independent parameters.
Based on their study, authors have suggested an equation, which takes into
account, material and equipment parameters. Rambali (142) scaled up granulation
process from small (5 kg) to medium (30 kg) to large (120 kg) with an aim to obtain
the target geometric mean granule size of 400 mm. The scaling up was based on the
relative droplet size and the powder bed moisture content at the end of the spraying
cycle. Authors found that the effect of the change in relative droplet size on the granule
size was different for each fluid bed. They applied the experimental design on the
small- and medium-scale unit, and regression models for the granule size were proposed
in order to scale up the granulation process on the small to medium scale.
Using only the relative droplet size, authors were able to scale up the process to
the larger unit.
9. SAFETY IN FLUID BED
For an explosion to occur, three conditions must exist: an ignition source, a fuel, and
oxygen. With an explosion, oxygen reacts with the fuel, releasing heat and gases. If a
dust explosion occurs in free space, a fireball of considerable extent arises. If the dust
explosion occurs in a closed container, there is a sudden pressure rise that is mainly
decided by the following factors: type of dust, size of the dust, dust/oxygen ratio,
turbulence, precompression, temperature, shape of the container, and ignition
source. In a container without precompression and with an organic dust of sufficient
fineness, the pressure inside the container can rise to over 10 bar overpressure.
Figure 27 Scale-up case study and resultant particle size distribution.
286 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
The fluid bed process handles a large amount of air. This air, in the presence of
fine product dust, poses potential for an explosion. This hazard can be enhanced
when using flammable solvents. If sufficient ignition energy (static charge) is introduced,
an explosion within the processor can take place. In order to contain these
dust or flammable solvent induced explosions, fluid bed processors are normally constructed
to withstand an overpressure of 2.0 bar. In this case, the fluid beds are provided
with explosion relief flaps, to release the pressure as soon as it starts to build up
inside the processor. The explosion flaps, mounted either horizontally or vertically
(Fig. 28), are designed to vent the pressure buildup as low as 0.06 bar. The explosion
isolates the overpressure within the processor. The 2 bar vented design shows the
of the building. These panels are gasketed, and sealed so that normal fluid bed operation
is not affected. It was an accepted practice to have the production unit with
2 bar pressure shock integrity; however, the cleaning of the gasket area around the
flaps is always difficult. To avoid having the product exposed to the outside during
such an event, a suppression system is used to contain the possible overpressure front
sors located within the processor. These sensors are designed to trigger a series of fire
extinguishers (containing ammonium phosphate), as soon as a preset level (generally,
0.1 bar) of pressure is set within the processor.
Figure 28 Processor with horizontal and vertical relief.
Batch Fluid Bed Granulation 287
protection valve, shown in Figure 29, acts to cut off the airflow to the blower and
propagation of the overpressure (Fig. 30). The explosion flaps open up to the outside
from leaving the unit (Fig. 31). The suppression system consists of low-pressure sen-
© 2005 by Taylor & Francis Group, LLC
With the introduction of potent and costly drug substances, the 2 bar design is
being replaced with a 10 bar design and higher, based on the specific mixtures of the
product(s) being processed. These units can withstand explosions up to 10 bar. Most
of the pharmaceutical dust explosions studied (143) show the overpressure reaching
Figure 29 Explosion protection valve.
Figure 30 Vented design (2 bar unit). (Courtesy of the Glatt Group.)
288 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
9 bar, with a Kst value (constant of explosion speed) of 200. An explosion in a 10 bar
unit is contained within the unit. A 10 bar designed unit does not require any explosion
relief panels or gaskets. This eliminates the concerns about cleaning of the gaskets
and panels. Another advantage of a 10 bar unit is that, in case of explosion, the
processor containing the potent drug substance is contained inside the unit, and exploand
(B) shows the Ventex-ESI valve for passive control of the explosion. The Ventexsion
force (pressure wave), moving ahead of the flame front, hurls the poppet forward to
the valve seat providing an airtight seal. The poppet, once seated, is locked in by a
mechanical shut-off device, which retains the seal until manually reset. The three basic
versions of the standard mechanical Ventex valve are available with a set pressure of
1.5 psiandamaximumpressure of 150 psi.TheVentex-ESI valve closes bythe explosion
shows the various types of Ventex valves installed in a fluid bed system.
In case of granulation requiring flammable solvents, process air and nozzle atomization
air are replaced by an inert gas such as nitrogen and the system is designed
as a closed cycle with a solvent recovery capability (144). A number of approaches
ods for solvent emission control systems.
?
The inert gas used for fluidization circulates continuously. An adjustable volume of
gas is diverted through the bypassed duct, where solvent vapors are condensed and
solvent collected. The circulating gas passes through the heat exchanger to maintain
the temperature necessary for evaporation of the solvent from the product bed. During
Figure 31 Explosion suppression system.
Batch Fluid Bed Granulation 289
sion does not pose an environmental problem as with the 2.0 bar unit. Figure 32(A)
ESI valve requires less maintenance than the active valve shown in Figure 29.Anexplopressure
wave, without external power for horizontal or vertical operation. Figure 33
Kulling and Simon (18) reported the closed loop system shown in Figure 34.
can be taken to handle solvent from the process. Table 6 summarizes various meth-
© 2005 by Taylor & Francis Group, LLC
the agglomeration and subsequent drying process, the solvent load in the gas stream
does vary. The bypass valve controls the flow of the gas to the heat exchanger and
the condenser. By controlling the gas stream in this manner, the drying action is continued
until the desired level of drying is reached. Even though the cost of fluid bed
processor with the solvent recovery is generally double the cost of a regular single-pass
fluid bed processor, such a system offers effective measures for both explosion hazard
reduction and air pollution control.
In 1994 European parliament issued the ATEXdirective (145) on the approximation
of the laws of the member states, concerning equipment and protective systems
intended for use in potentially explosive atmospheres. The scope, these regulations specifically
require ‘‘protective systems,’’ intended to halt incipient explosions immediately
Figure 32 (A) Schematic of the Ventex valve. (B) Ventex valve. (Courtesy of Rico-Sicherheitechnik
AG, Switzerland.)
290 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
and to limit the effective range of explosion flames and explosion pressures. All manufacturers
of fluid bed processors in Europe must comply with this directive.
10. MATERIAL HANDLING OPTIONS
The transfer of materials to and from the fluid bed processor is an important consideration.
The loading and unloading of the processing bowl can be accomplished
either by manual mode or by automated methods.
Figure 33 Setup of fluid bed with different Ventex valves. (Courtesy of Rico-Sicherheitechnik
AG, Switzerland.)
Table 6 Comparison of Different Solvent Emission Control System
Water scrubbing Catalytic burning Carbon absorption Condensation
Open cycle Open cycle Open cycle Closed cycle with
nitrogen
High capital cost Low capital cost Moderate capital cost Low capital cost
High energy
requirement
Low energy Low energy Low energy
External installation External installation External installation Internal installation
Medium space
requirement
High space
requirement
Moderate space
requirement
Small space
requirement
Medium flexibility Medium flexibility Low flexibility Good flexibility
Waste treatment
required
CO2/H2O emission
waste treatment
Waste treatment Concentrated waste
treatment
Batch Fluid Bed Granulation 291
© 2005 by Taylor & Francis Group, LLC
10.1. Loading
The contemporary method for loading the unit is by removing the product bowl
from the unit, charging the material into the bowl, and then placing the bowl back
into the unit. This loading is simple and cost-effective. Unfortunately, it has the
potential of exposing the operators to the product and contaminating the working
area. To avoid the product from having a dust and cleaning hazard, a dust collection
system should be installed to collect the dust before it spreads. A manual process also
depends on the batch size, the operator’s physical ability to handle the material, and
the container full of product. Furthermore, this can be time consuming, since the
material must be added to the product container, one material at a time.
The loading process can be automated and isolated to avoid worker exposure,
minimize dust generation, and reduce loading time. There are two main types of
loading systems. These systems are similar because both use the fluid bed’s capability
to create a vacuum inside the unit. Here, the product enters the fluid bed through a
product in-feed port, on the side of the unit. This is done by having the fan running
and the inlet air control flap set, so that minimum airflow may pass through the product
container and the outlet flap is almost fully open. Typically, when the highshear
granulated material needs to be charged into the fluid bed, this approach helps
valve is closed and the granulating process started. This transfer method uses some
amount of air to help the material move through the tube. Loading can be done
either vertically from an overhead bin, or from the ground. Less air is required
through the transfer pipe when the material is transferred vertically because gravity
is working to help the process. Vertical transfer methods do require greater available
height in the process area. Loading by this method has the advantages of limited
operator exposure to the product, allows the product to be fluidized as it enters
the processor, and reduces the loading time. The disadvantage of this type of system
is the cleaning required between different products.
Figure 34 Schematic of closed loop solvent recovery system.
292 Parikh and Mogavero
(Fig. 35). Once the material has been charged to the fluid bed, the product in-feed
© 2005 by Taylor & Francis Group, LLC
10.2. Unloading
As with loading, the standard method for unloading is by removing the product
bowl from the unit. Once the bowl is removed, the operator may scoop the material
from the bowl, which is the most time consuming and impractical method, because
of its potential of exposure to the product. Alternatively, the product can be vacuum
transferred to a secondary container or unloaded by placing the product bowl into
This hydraulic device is installed in the processing area. The mobile product
container of the fluid bed processor is pushed under the cone of the bowl dumper
and coupled together by engaging the toggle locks. Subsequently, the container is
lifted hydraulically, pivoted around the lifting column, and rotated 180 for discharging.
Use of the bowl dumping device or vacuum unloading device still requires that
the product bowl be removed from the unit. There are contained and automated
methods for unloading the product, while the product bowl is still in the fluid bed
processor. The product may be unloaded either from the bottom of the product
container or from the side. Until recently, the most common contained method
is to unload the material from the bottom of the unit. This requires a ceiling
height high enough to accommodate it, or the installation becomes multistoried
installation.
There are two types of bottom discharge options, gravity or pneumatic.
tainer, which is located below the lower plenum. If the overall ceiling height limitation
prevents discharge by gravity, the gravity/pneumatic transfer combination can
be considered. The gravity discharge poses cleaning problems, since the process air
and the product discharge follow the same path; assurance of cleanliness is always of
prime concern.
The desire to limit the processing area and the development of the overlap gill
air distributor mentioned earlier in the chapter has prompted the consideration of
the side discharge as an option. The product bowl is fitted with the discharge gate,
Figure 35 Loading of fluid bed from high-shear mixer. (Courtesy of the Glatt Group.)
Batch Fluid Bed Granulation 293
bowl dumping discharge device, as shown in Figure 36(A) and (B).
Gravity discharge (Figs. 37 and 38) allows for collection of the product into conas
shown in (Fig. 39A and B).
© 2005 by Taylor & Francis Group, LLC
Most of the product, being free-flowing granules, flows through the side
discharge into a container. The remainder of the product is then discharged by
manipulation of the airflow through the overlap gill air distributor. The discharged
product can be pneumatically transported to an overhead bin if the dry milling of the
granulation is desired.
The contained system for unloading the product helps to isolate the operator
from the product. The isolation feature also prevents the product from being
contaminated due to exposure to the working environment. Material handling consideration
must be thought of early in the equipment procurement process. Fluid bed
processing, whether used as an integral part of high-shear mixer/fluid bed dryer or as
a granulating equipment option, production efficiency, and eventual automation can
be enhanced by considering these loading and unloading options.
11. FLUID BED TECHNOLOGY PROGRESS
The fluid bed processor was used as an efficient way to dry a product due to suspension
of particles in the hot air stream. However, over the last 35 years, the development
in the pharmaceutical industry and the proliferation of the batch fluid bed
processing technology in other industries as food, polymer, detergent, etc., has
provided the opportunity to use the batch fluid bed processor for granulation,
coating of particles, and pelletization. The advances in the fluid bed can be attributed
to several factors. The needs of formulators, the requirements of the regulators,
and technological innovations from the manufacturers of these equipments are
Figure 36 (A) Product discharge system. (B) Bowl inverter. (Courtesy of Niro Pharma
Systems.)
294 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
responsible for these advances. The result of these changes provided units that are
paint free, modular, safer, in compliance with the cGMPs, and are capable of performing
various processes that were not thought of before.
11.1. Developments in Equipment and Process Applications
Parikh (146) has presented a review of all of the equipment developments. Among
various advances, the development of production units that can withstand more than
10 bar pressure shock resistance is a very significant development. These units do not
require pressure relief duct and associated cleaning problems. Units are now equipped
with the air handler that can provide designated humidity and dew point air, throughout
the year and at any geographical location. The fluid bed CIP became a reality with
the introduction of the overlap gill air distributors and the stainless steel cartridge filters
described earlier in this chapter. The coating of the particles is carried out most
Figure 37 Loading and unloading setup with bottom discharge. (Courtesy of the Vector
Corporation.)
Batch Fluid Bed Granulation 295
frequently using Wurster column (Fig. 40A and B). The Wurster process is the most
© 2005 by Taylor & Francis Group, LLC
popular method for coating particles. The Wurster based coating process does not
contain any fluid bed regions in the traditional sense, as it is a circulating fluid bed
process. Four different regions within the equipment can be identified: the upbed
region, the expansion chamber, the downbed region, and the horizontal transport
region. The coating process consists of three phases: the start-up phase, the coating
phase, and the drying and cooling phase. During the coating phase, several processes
take place simultaneously. They are atomization of the coating solution or suspension,
transport of the atomized droplets of the coating solution to the substrate, and drying
of the film. In 1993, the design was modified as in Wurster HS (147).
However, the technology has certain disadvantages, such as nozzle accessibility,
prolonged process time, minimum volume requirement, and difficulty of loading
and unloading. In 1995, to address these shortcomings of the Wurster system, the
Precision Coater
 incorporating modified air suspension technique was introduced
(148). It was designed to allow removal of the nozzles for cleaning, faster process
time because of the patented particle accelerator, good utilization of thermal and
kinetic energies, and scalability from a single-column to a multicolumn setup.
Recently, Glatt organization introduced the STRATOS2 system for applying sustained
release tablet coating, using bottom spray. The modification of the bottom
spray setup offers better efficiency than Wurster HS for coating tablets (149). The
Wurster setup for granulating was presented by Walter et al. (150). Authors claim
that the quality of granulation is superior to the top spray granulation.
Researchers have discussed the incorporation of microwave in the laboratory
fluid bed processor (151,152). Fluid bed process using organic solvent requires an
inert gas such as nitrogen to replace the air used for fluidization, as discussed earlier
in the chapter. It is accompanied by the solvent recovery system. In 1989, a vacuum
fluid bed system was presented by Luy et al. (153). The main feature was the generation
and sustaining of a fluidized bed under vacuum, thereby eliminating the use of
inert gas. Several advantages are claimed by the authors, such as emission reduction,
increased recovery rate of solvent, and an application for oxygen sensitive materials.
Figure 38 Bottom discharge. (Courtesy of the Glatt Group.)
296 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
11.2. Rotary Fluid Bed
The other advance of significance is the development of a rotary fluid bed for producing
denser granulation. Modules were introduced by various manufacturers, and
the technology is discussed below. The 1972 patent (154) for the rotor technology
was awarded for the equipment and coating of the granular material. The subsequent
patents (155,156) were awarded for the coating of the spherical granules.
An advantage of rotary fluid bed processing to produce granules over the conventional
top-spray granulation technique was reported by Jager and Bauer (155). In
this unit, the conventional air distributor is replaced by the rotating disk. The material
to be granulated is loaded on the rotating disk. The binder solution is added
through the atomization nozzle, located tangentially to the wall of the bowl. The centrifugal
force creates a dense, helical doughnut-shaped pattern. This type of motion
is caused by the three directional forces.
The vertical movement is caused by the gap or slit air around the rotating disk,
the gravitational force folds back the material to the center, and the centrifugal force
caused by the rotating disk pushes the material away from the center. The granulation
produced in the rotary fluid bed processor shows less porosity, compared to the
Figure 39 (A) Side discharge. (Courtesy of the Glatt Group.) (B) Side discharge. (Courtesy
of Niro Pharma Systems.)
Batch Fluid Bed Granulation 297
© 2005 by Taylor & Francis Group, LLC
conventionally agglomerated product in the fluid bed processor. There are essentially
two designs of this rotary fluid bed module which are available: the single-chamber
conventional air distributor is replaced by the rotating disk, which has variable size
slit opening between the bowl wall and the rotor disk. The fluidizing air enters the
mixing zone in the bowl through the slit. The single-chamber design is manufactured
by Glatt (W.Glatt GMBH, Binzen, Germany), as Glatt Rotor Granulator
, and by
Figure 40 (A) Wurster Coater schematic. (B) Wurster Coater. (Courtesy of the Vector
Corporation.)
298 Parikh and Mogavero
design and the double-chamber design (Fig. 41). In the single-chamber design, the
© 2005 by Taylor & Francis Group, LLC
Freund (Freund Industrial Co. Ltd., Tokyo, Japan), as a Spir-a-flow
 granulator.
The double-chamber design was patented and manufactured by Aeromatic-Fielder
AG (Aeromatic-Fielder AG–now Niro Pharma Systems, Bubendorf, Switzerland)
(158). An inner stainless steel wall, encompassing the area called the forming zone
is surrounded by an annular drying zone. The forming zone has a rotating disk.
The gap air around the rotating disk allows the free movement of the disk. The stainless
steel wall separating the forming zone and drying zone can be raised, so that the
rotating product transfers in the drying zone through the gap created during the drying
mode between the stationary and movable walls. The process air fluidizes the product
through the annular drying zone. As the partially dried product reaches the upper
boundary of the forming zone, the particle velocity slows down due to the design of
the unit and the particles fall back in the forming zone. These partially dried particles,
eventually, go through the same path in a cyclic pattern, until the desired moisture content
is reached. The rotary fluid bed, with its different designs listed above, do provide
granules which have less porosity, and higher bulk density compared to the granulation
produced by the typical top spray granulated fluid bed process.
Tu?rkoglu et al. produced Theophylline granulation using a rotary fluid bed
(159). The formulation contained lactose, starch, and microcrystalline cellulose,
Figure 41 Various rotary fluid bed inserts.
Batch Fluid Bed Granulation 299
© 2005 by Taylor & Francis Group, LLC
along with Theophylline. They reported that the granules produced were spherical
and dense. Three different drug level formulations were evaluated. The authors concluded
that the rotary fluid bed, as a wet granulator, has the potential to obtain a
better drug content uniformity for tablets even at low active levels such as 1% in
comparison with conventional fluidized beds. The use of rotary fluid bed to produce
spherical granules for modified release application is reported by a number of
authors. Rotary fluid bed technology was reviewed by Li et al. (160), and its usefulness
was described to produce the pellets. The comparison of the rotary fluid bed
processing with the multiple-step extrusion and spheronization was reported by
Robinson et al. (161). The authors manufactured acceptable immediate release acetaminophen
pellets, using both of these techniques. The quality of the pellet produced
improved as the minimum quantity of product was increased in the rotary fluid bed
processor. The advantage of using a single unit, such as the rotary fluid bed, over the
multiple-unit process involving several pieces of equipment was described.
The rotary fluid bed is used for producing a pellet by layering the active drug
suspension or solution on to nonpareil cores, and subsequently coating them with
polymers to impart modified release properties (162) Hileman et al. (163) reported
the manufacture of immediate spheres of a poorly water-soluble drug in a rotary
fluid bed by layering the active drug suspension on to nonpareil cores. These immediate
release spheres were then overcoated with an ethylcellulose/HPMC hydroalcoholic
solution in the same unit, eliminating the need for additional process and
handling steps. Iyer et al. evaluated layering of aqueous solution of phenylpropanolamine
hydrochloride with different binders (164). The layered beads were coated in
the rotoprocessor and the Wurster Coater to compare the utility of the rotoprocessor
as an equipment, not only to produce pellets, but to coat them as well. Various
equipment manufacturers have promoted powder layering on the pellets, in a rotary
fluid bed. In 1992, Jones et al. received a patent for such a process (165). The process
claims to have advantages of layering a drug substance with a relatively small
amount of liquid, thus making this layering process more efficient. The commercial
application of this process has not been reported in the literature. Korakianiti et al.
(166) studied the preparation of pellets using rotary fluid bed granulator. The
authors concluded that the rotor speed and the amount of water significantly
affected the geometric mean diameter of the pellets, and they proposed an equation
to show that correlation. Pis?ek et al. (167) studied the influence of rotational speed
and surface of rotating disk on pellets produced by using rotary fluid bed. They used
a mixture of pentoxifylline and microcrystalline cellulose to produce pellets, using a
suspension of Eudragit
 NE 30 D as a binder. The results showed that both the surface
and the rotational speed of the disk have an influence on the shape, surface, and
size of the pellets, while there was less effect on the density, humidity content, and
yield. They found that the textured surface of the disk produced pellets with a
rougher surface when the rotational speed was increased compared to the smooth
surface, where increased rotational speed produced more spherical pellets with a
larger diameter. Jan Vertommen (168) has summarized the use of rotary fluid bed
processor to produce pellets.
11.3. Integrated Systems
The fluid bed technology is used for drying, agglomerating, coating, and pelletization.
The trend in the industry is toward integrating various steps currently used
to produce the solid dosage products. In an attempt to minimize the number of steps
300 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
involved, fluid bed granulation and drying is increasingly planned in the following
three ways:
1. Fluid bed processor for granulation and drying
2. Integration of high-shear mixer and a fluid bed dryer
3. Integration of high-shear mixer with fluid bed dryer for containment of
potent compounds.
11.3.1. The Granulation and Drying Carried Out in a Single Unit
To minimize material handling steps, the fluid bed units are loaded by gravity or by
vacuum, as described previously. Discharge of the product from the unit is accomplished
by either side or bottom discharge, by employing pneumatic transport
system. The CIP of such a unit can be accomplished by using stainless steel cartridge
filters and an overlap gill air distributor, if desired.
11.3.2. Integration of a High-Shear Mixer and a Fluid Bed Dryer
for controlling dust and cross-contamination. When these two-unit operations are
integrated as a single unit, a number of points must be considered. The following is a
list of some of the questions that the reader may want to consider:
1. Engineering layout and the footprint, ceiling height requirements.
2. How will the high-shear mixer be loaded—by gravity, vacuum, or
manually?
Figure 42 Integrated system. (Courtesy of the Glatt Group.)
Batch Fluid Bed Granulation 301
Figures 42 and 43 show a typical integrated system, where containment is considered
© 2005 by Taylor & Francis Group, LLC
3. How will the binder solution be prepared and delivered to the mixer?
4. How will the granulation end point be determined and reproduced?
5. How will the discharge from the high-shear mixer be accomplished?
6. Are the process parameters for granulation and fluid bed drying established
and reproducible, indicating a robust process?
7. How will the product, discharged from the fluid bed dryer be handled?
Does it require sizing, blending with the lubricants?
8. Is this system dedicated for a single product or multiple products?
9. How will this system be cleaned?
10. Will the control of a process be done individually for each unit or by an
integrated control system?
For the potent compound processing, requiring high-shear granulation and
fluid bed drying, all the questions mentioned previously must be considered. In
addition, the operator safety, special sampling valves, isolated room air handling
systems, and CIP must be considered. Such an integrated system can be dedicated
for a single product.
materials are dispensed, operating personnel do not have to handle the product during
the granulation drying steps. Such a system is costly and does require an enormous
amount of time for process and cleaning validation.
The fluid bed process, like other granulation techniques requires understanding
of the importance of characterization of the raw materials, especially of a drug substance,
the process equipment, limitations of the selected process, establishment of
in-process control specifications, characterization of the finished product, and clean-
Figure 43 Integrated system. (Courtesy of the Glatt Group.)
302 Parikh and Mogavero
Figure 44 shows a system where a potent compound is processed. After the raw
© 2005 by Taylor & Francis Group, LLC
ing and process validation. It is equally important that the formulation and development
scientists do not lose sight of the fact that the process being developed will be
going on the production floor and should be robust enough.
ACKNOWLEDGMENT
My sincere thanks to Ms. Christine Budny, my assistant at Synthon Pharmaceuticals
Inc., for her help in preparing this chapter.
Figure 44 Integrated system for potent compound. (Courtesy of Niro Pharma Systems.)
Batch Fluid Bed Granulation 303
© 2005 by Taylor & Francis Group, LLC
REFERENCES
1. Kunii D, Lvenspiel O. Fluidization Engineering. New York: John Wiley & Sons, Inc.,
1968.
2. Wurster DE. J Am Pharm Assoc Sci Ed 1960; 49:82.
3. Wurster DE. J Am Pharm Assoc Sci Ed 1959; 48(8):451–454.
4. Scott MW, Liberman HA, Rankell AS, Battista JV. J Pharm Sci 1964; 53(3):314–319.
5. Rankell AS, Scott MW, Liberman HA, Chow FS, Battista JV. J Pharma Sci 1964;
53(3):320.
6. Contini S, Atasoy K. Pharm Ind 1966; 28:144–146.
7. Wolf G. Drugs Made Germany 1968; 11:172–180.
8. Liske T, Mobus W. The manufacture and comparative aspects of fluidized layer spray
granulation. Drugs Made Germany 1968; XI:182–189.
9. Sherrington PJ, Oliver R. Granulation monograph (chapter 3). In: Goldberg AS, ed.
Powder Science and Technology. London: Heyden, 1981.
10. Pietch WB. In: Fayed ME, Otten L, eds. Handbook of Powder Science and Technology.
New York: Van Nostrand Reinhold, 1984.
11. Hersey JA. Fluidized bed technology—an overview. Int J Pharm Tech Prod Manuf
1981; 2(3):3–4.
12. Thiel WJ. The theory of fluidization and application to the industrial processing of
pharmaceutical products. Int J Pharm Tech Prod Manuf 1981; 2(5):5–8.
13. Thiel WJ. Solids mixing in gas fluidized beds. Int J Pharm Tech Prod Manuf 1981;
2(9):9–12.
14. Littman H. An overview of flow in fluidized beds. Pharm Technol 1985; 9(3):48.
15. Whitehead AB. Behavior of fluidized bed systems. Pharm Technol 1981; 2(13).
16. Story MJ. Granulation and film coating in the fluidized bed. Pharm Technol 1981;
2(19):19–23.
17. Aulton MJ, Banks M. Fluidized bed granulation; factors influencing the quality of the
product. Pharm Technol 1981; 2(24):24–27.
18. Kulling W, Simon EJ. Fluid bed technology applied to pharmaceuticals. Pharm Technol
1980; 4(1):79–83.
19. Gomezplata A, KugelmanAM.Processing systems. In: MarchelloJMand Gomezplata A,
eds. Gas–Solid Handling in the Process Industries, Chemical Processing and Engineering
Vol. 8. New York: Marcel Dekker, Inc., 1976:chap 4.
20. E. Ra?sa?nen et al. AAPS PharmSci, 3(3) Abstract, AAPS Annual Meeting 2001.
21. Parikh DM. Airflow in batch fluid-bed processing. Pharm Technol 1991; 15(3):100–110.
22. Row PN, et al. Trans Inst Chem Eng 1965; 32T:271.
23. Davis L, et al. Trans Inst Chem Eng 1966; 44T:293.
24. Gorrodnichev VI, et al. Pharm Chem J (USSR) 1974; 8:298.
25. Thiel WJ. The theory of fluidization and application to the industrial processing of
pharmaceutical products. Int J pharm Tech Prod Manuf 1981; 2(5).
26. Zenz FA. In: Kirk-Othmer, ed. Fluidization: Encyclopedia of Chemical Technology,
Vol. 10, 3rd ed. New York: Wiley-Interscience Publication, 548–581.
27. Ennis BJ, Tardos GI, Pfeffer R. A micro-level–based characterization of granulation
phenomena. Powder Technol 1991; 65:257–272.
28. Osborne BG, Fearn T, Hindle PH. In: Practical NIR Spectroscopy with Applications in
Food and Beverage Industry Analysis. 2nd ed. Harlow, U.K.: Longman, 1993:227.
29. Long GE. Spraying theory and practice. Chem Eng 1978:73–77.
30. Masters K. Spray Drying, an Introduction to Principles, Operational Practice and
Application. 2nd ed. New York: John Wiley & Sons Inc., 1976.
31. Japanese Patent Application 61–259696, 1986.
32. Swiss Patent 0176/93, 1993, U.S. Patent 5,444,892, 1995, European Patent 0572356A1.
33. Guidance for Industry, Part 11, Electronic Records; Electronic Signatures—Scope and
Application, February 2003.
304 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
34. Newitt DM, Conway-Jones JM. A contribution to the theory and practice of granulation.
Int J Pharm Tech Prod Manuf 1958; 36:422.
35. Record PC. A review of pharmaceutical granulation technology. Int J Pharm Tech Prod
Manuf 1980; 1:32.
36. Rumpf H. In: Krepper W, ed. Agglomeration. New York: Interscience, 1962:374.
37. Tardos GI, Khan MI, Mort PR. Critical parameters and limiting conditions in binder
granulation of fine powders. Powder Technol 1997; 94:245–258.
38. Iveson SM. Granule coalescence modeling: including the effects of bond strengthening
and distributed impact separation forces. Chem Eng Sci 2001; 56:2215–2220.
39. Schaefer T, Worts O. Control of fluidized bed granulation I. Effects of spray angle, nozzle
height and starting materials on granule size and size distribution. Arch Pharm
Chem Sci Ed 1977; 5:51–60.
40. Smith PJ, Nienow AW. Chem Eng Sci 1983; 38(8):1223–1231, 1323–1240.
41. Aulton ME, Banks M. Int J Pharm Tech Prod Manuf 1981; 2(4):24–29.
42. Maroglou A, Nienow AW. 4th Symposium on Agglomeration, Toronto, Canada, Iron
& Steel Society Inc., June 1985:465–470.
43. Maroglou A, Nienow AW. Fluidized bed granulation technology and its application to
tungsten carbide. Powder Metall 1986; 29(4):195–291.
44. McCabe WL, Smith JC. In: Lachman L, Liberman HA, Kanig JL, eds. Unit Operations
of Chemical Engineering. New York: McGraw-Hill, 1956:Inc.
45. Green Don W, ed. In: Perry’s Chemical Engineer’s Handbook. New York : McGraw-
Hill, Inc., 1984:section 20.
46. Rankell RS, Liberman HA, Schiffman RF. Drying. In: The Theory and Practice of
Industrial Pharmacy. 3rd ed. Philadelphia, PA: Lea & Feabiger, 1986.
47. Schaefer T, Worts O. Control of fluidized granulation III, effects of the inlet air temperature,
and liquid flow rate on granule size and size distribution. Control of moisture
content of granules in the drying phase. Arch Pharm Chem Sci Ed 1978; 6(1):1–13.
48. Gao JZH, et al. AAPS PharmSci, 1(4) Abstract (Tech Note), AAPS Annual Meeting
2000.
49. Thurn U. Dissertation, Eidgenossischen Technischen, Hochschule, Zurich, 1970.
50. Liske T, Mobus W. Drugs Made Germany 1968; 11(4):182–189.
51. Schaefer T, Worts O. Control of fluidized bed granulation, I. Effects of spray angle,
nozzle height and starting materials on granule size and size distribution. Arch Pharm
Chem Sci Ed 1977; 5:51–60.
52. Schaefer T, Worts O. Control of fluidized bed granulation V, factors affecting granule
growth. Arch Pharm Chem Sci Ed 1978; 6:69–82.
53. Ormos Z, Pataki K. Hung J Ind Chem 1979; 7:89–103.
54. Ormos Z, Pataki K. Hung J Ind Chem 1979; 7:105–117.
55. Banks M. Ph.D. Thesis, C.N.A.A., Leicester Polytechnic, 1981.
56. Galmen MJ, GreerW. Paper 2. Fluid Technology and Pharmaceutical Manufacturing
International Conference, 1982.
57. Aulton ME. Paper 3. Fluid Technology and Pharmaceutical Manufacturing International
Conference, Powder Advisory Center, London, 1982.
58. Veillard M, et al. Int J Pharm Tech Prod Manuf 1982; 3(4):100–107.
59. Georgakopoulos PP, et al. Pharmazie 1983; 38(4):240–243.
60. Jinot JC, et al. STP Pharm 1986; 2(13):126–131.
61. Shinoda A, et al. Yakuzaigaku 1976; 36(2):83–88.
62. Aulton ME, et al. J Pharm Pharmacol 1977; 29(suppl):59.
63. Schepky G. Acta Pharm Technol 1978; 24(3):185–212.
64. Aulton ME, et al. Proceedings of the International Conference on Powder Technology
in Pharmacy, Powder Advisory Center, Basel, Switzerland, June 1979.
65. Kocova El-Arini S, et al. Drugs Made Germany 1983; 26(4):205–211.
66. Davies WL, Gloor WT. J Pharm Sci 1972; 61(4):618–622.
67. Rouiller M, et al. Acta Pharm Technol 1975; 21(2):129–138.
Batch Fluid Bed Granulation 305
© 2005 by Taylor & Francis Group, LLC
68. Schaefer T, Worts O. Control of fluidized bed granulation II, estimation of droplet size
of atomized binder solution. Arch Pharm Chem Sci Ed 1977; 5:178–193.
69. Ormos Z, et al. Hung J Ind Chem 1979; 7:131–140.
70. Ormos Z, et al. Hung J Ind Chem 1979; 7:141–151.
71. Ormos Z, et al. Hung J Ind Chem 1979; 7:153–163.
72. Ormos Z, et al. Hung J Ind Chem 1979; 7:221–235.
73. Kocova El-Arini S. Pharm Ind 1981; 43(7):674–679.
74. Jager KF, Bauer KH. Acta Pharm Technol 1984; 30(1):85–92.
75. Nouh ATI. Pharm Ind 1986; 48(6):670–673.
76. Alkan H, et al. Doga Tu J Med Pharm 1987; 11(1):1–7.
77. Bank A, et al. Proc 2nd Conf Appl Phys Chem 1971; 2:687–692.
78. Davies WL, Gloor WT. J Pharm Sci 1973; 62(1):170–172.
79. Ormos Z, et al. Hung J Ind Chem 1973; 1:307–328.
80. Ormos Z, et al. Hung J Ind Chem 1973; 1:463–474.
81. Johnson MCR, et al. J Pharm Pharmacol 1975; suppl:80.
82. Schaefer T, Worts O. Control of fluidized bed granulation IV. Effects of binder solution
and atomization on granule size and size distribution. Arch PharmChem Sci Ed 1978;
6:14–25.
83. Aulton ME, et al. Mfg Chem Aerosol News 1978; 12:50–56.
84. Gorodnichev VI, et al. Pharm Chem J (USSR) 1980; 14(10):72–77.
85. Ceschel GC, et al. II Farm Ed Prat 1981; 36(6):281–293.
86. Rangnarsson G, et al. Int J Pharm 1982; 12:163–171.
87. Meshali M, El-Banna HM, El-Sabbagh H. Pharmazie 1983; 38(5):323–325.
88. Hajdu R, Ormos Z. Hung J Ind Chem 1983; 12:425–430.
89. Devay A, et al. Acta Pharm Technol 1984; 30(3):239–242.
90. Alkan MH, Yuksel A. Drug Dev Ind Pharm 1986; 12(10):1529–1543.
91. Hontz J. Assessment of Selected Formulation and Processing Variables in Fluid Bed
Granulation, Ph.D. Thesis, University of Maryland, Baltimore, MD, 1987; Dissertation
Abs Int 1987; 6:1655-B.
92. Wan LSC, Lim KS. STP Pharm 1988; 4(7):560–571.
93. Wan LSC, Lim KS. STP Pharm 1989; 5(4):244–250.
94. Sheskey P, Keary C, Inbasekaran P, Deyarmond V, Balwinski K. Foam technology: the
development of a novel technique for the delivery of aqueous binder systems in high
shear and fluid-bed wet-granulation applications. Poster at AAPS Annual Meeting,
Oct 26–30, 2003.
95. Dahl TC, Bormeth AP. Naproxen controlled release matrix tablets: fluid bed granulation
feasibility. Drug Dev Ind Pharm 1990; 16(4):581–590.
96. Wan Lucy SC, Heng Paul WS, Muhuri G. Incorporation and distribution of a low dose
drug in granules. Int J Pharm 1992; 88:159–163.
97. Higashide F, Miki Y, Nozawa Y, Ishibashi K. Dependence of drug content uniformity
on particle sizes in fluidized bed granulation. Pharm Ind 1985; 47(11):1202–1205.
98. Davies WL, Gloor WT Jr. Batch production of pharmaceutical granulationin fluidized
bed II: effects of various binders and their concentrations on granulations and compressed
tablets. J Pharm Sci 1972; 61:618.
99. Davies WL, Gloor WT Jr. Batch production of pharmaceutical granulation in fluidized
bed III: binder dilution effects on granulation. J Pharm Sci 1973; 62:170.
100. Seo A, Holm P, Schaefer T. Effects of droplet size and type of binder on the agglomerate
growth mechanisms by melt agglomeration in a fluidized bed. Eur J Pharm Sci 2002;
16:95–105.
101. Rawley FA. Effects of the bag shaking cycle on the particle size distribution of granulation.
Pharm Technol 1989; 13(9):78–82.
102. Davies WL, Gloor WT Jr. Batch production of pharmaceutical granulation in fluidized
bed I: effects of process variables on physical properties of final granulation. J Pharm
Sci 1971; 60(12):1869–1874.
306 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
103. Workman J Jr. A review of process near infrared spectroscopy: 1980–1994. J. Near
Infrared Spectrosc 1993; 1:221–245.
104. U.S. Patent USP XXVII 2004.
105. Near Infra Red Spectrometry. European Pharmacopoeia June 2004:59.
106. Choppin G, Buijs K. Near infrared studies of structure of water-II. Ionic solutions.
J Chem Phys 1963; 39:2042–2050.
107. Osborne BG, Fearn T, Hindle PH. In: Practical NIR Spectroscopy with Applications in
Food and Beverage Industry Analysis, 2nd ed. Harlow, U.K.: Longman, 1993.
108. Fong A, Hieftje GM. Near infrared measurement of relative and absolute humidity
through detection of water adsorbed on a silica gel layer. Anal Chem 1995; 67:
1139–1146.
109. Siesler HW, Ozaki Y, Kawata S, Heise HM. Near-Infrared Spectroscopy: Principles,
Instruments and Applications. Weinheim, Germany: Wiley, 2002.
110. Wetzel DL. Near infrared reflectance analysis: sleeper among the spectroscopic techniques.
Anal Chem 1983; 55(12):1165A–1175A.
111. McDonald BF, Pebble KA. Some application of near-infrared reflectance analysis in the
pharmaceutical industry. J Pharm Biomed Anal 1993; 11(11–12):1077–1085.
112. Sinsheimer J, Poswalk N. Pharmaceutical applications of the near infrared determination
of water. J Pharm Sci 1968; 57:2007–2010.
113. Wetzel DL. Near infrared reflectance analysis: sleeper among the spectroscopic techniques.
Anal Chem 1983; 55(12):1165A–1175A.
114. Workmann JJ. Review of process and non-invasive near-infrared and infrared spectroscopy
1193–1999. Appl Spectrosc Rev 1999; 34:1–89.
115. Callis J, Illman D, Kowalski B. Process analytical chemistry. Anal Chem 1987;
59:624A–637A.
116. Beebe K, Blaser W, Bredeweg R, Chauvel J, Harner R, LaPack M, Leugers A, Martin D,
Wright L, Yalvac E. Process analytical chemistry. Anal Chem 1993; 65:199R–216R.
117. Blaser W, Bredeweg R, LaPack M, Leugers A, Martin D, Pell R, Workman J, Wright L.
Process analytical chemistry. Anal Chem 1995; 67:47R–70R.
118. Hassel D, Bowman E. Process analytical chemistry for spectroscopists. Appl Spectrosc
1998; 52:18A–29A.
119. Workman J Jr, Veltkamp D, Doherty S, Anderson B, Creasy K, Koch M, Tatera J,
Robinson A, Bond L, Burgess L, Bokerman G, Ullman A, Darsey G, Mozayeni F,
Bamberger J, Greenwood M. Process analytical chemistry. Anal Chem 1999;
71:121R–180R.
120. Brittain H. Methods for the characterization of polymorphs and solvates in polymorphism
in pharmaceutical solids. In: Brittain H ed. 1st ed. New York: Marcel Dekker Inc.,
1999:227–278.
121. Frake PD, Greenhlgh SM, Grierson JM, Hempenstall, Rudd DR. Process control and
end-point determination of fluid bed granulation by application of near-infrared spectroscopy.
Int J Pharm 1997; 151:75–80.
122. Rantanen JO, Antikanen JP, Mannermaa, Yliuusi J. Use of near-infrared reflectance
method for measurement of moisture content during granulation. Pharm Dev Technol
2000; 5(2):209–217.
123. Rantanen J, Lehtola S, Ra?met P, Mannermaa JP, Ylirussi. On-line monitoring of moisture
content in an instrumented fluidized bed granulator with multi-channel NIR moisture
sensor. Powder Technol 1998; 99:163–170.
124. Andersson M, Folestad S, Gottfires J, Johansson MO, Josefson M, Wahlund KG.
Quantitative analysis of film coating in a fluidized-bed process by in-line NIR spectroscopy
and multivariate batch calibration. Anal Chem 2000; 72:2099–2108.
125. Vazquz ER. Optimization of Drying-End-Points Measurements for the Automation of
a Fluidized-Bed Dryer Using FT-NIR Spectroscopy. M.S. Thesis, University of Puerto
Rico, 2004.
Batch Fluid Bed Granulation 307
© 2005 by Taylor & Francis Group, LLC
126. Rantanen J, Ka?nsa?koski M, Tenhunen J, Lehtonen S, Rajalahti T, Mannermaa J,
Yliruusi J. Next generation fluidized bed granulator automation. AAPS PharmSciTech
2000; 1(2):article 10.
127. Rantanen JA, Jorgensen E.Ra?sa?nen,Luukkonen P, Airaksinen S, Raiman J, Ha?nninen K,
Antikainen O, Yliruusi J. Process analysis of fluidized bed granulation. AAPS Pharm
SciTech 2001; 2(4):article 21.
128. Rantanen J, Sampsa JL, Osmo KA, Jukka-Pekka M, Olii ES, Jouko KY. Visualization
of fluid bed granulation with self organizing maps. J Pharm Biomed Anal 2001;
24(3):343–352.
129. Laitinen N, Antikainen O, Airaksinen S, Yliuusi J. At-line particle size analysis with a
novel optical technique during a fluidized-bed granulation process. Poster at AAPS
Annual Meeting, 2002.
130. Jha BK, Tambe SS, Kulkarni BD. Estimating diffusion coefficients of a micellar system
using an ANN. J Coll Interface Sci 1995; 170:392–398.
131. Bourquin J, Schmidi H, Van Hoogevest P, Leunberger H. Basic concepts of ANN modeling
in the application to pharmaceutical development. Pharm Dev Technol 1997;
2(2):95–109, 111–121.
132. Watano S, Takashima H, Miyanami K. Control of moisture content in fluidized bed
granulation by neural network. J Chem Eng Jpn 1997; 30(2):223–229.
133. Leane MM, Cumming I, Corrigan OI. The use of artificial neural networks for the selection
of the most appropriate formulation and processing variables in order to predict
the in vitro dissolution of sustained release minitabs. AAPS PharmSciTech 2003;
4(2):article 26.
134. Vandel DA, Davari, Famouri P. Modeling of fluidized bed neural networks. Proceedings
of 32nd IEEE SSST, Tallahassee, FL, March 2000, FAMU-FSU.
135. Quantrille TE, Liu YA. In: Artificial Intelligence in Chemical Engineering. San Diego:
Academic Press, 1991.
136. AAPS/FDA. Scale-up of Oral Solid Dosage Forms, December 11–13, 1991.
137. Food and Drug Administration. Immediate Release Solid Dosage Forms. Scale-up and
Post Approval Changes: Chemistry, Manufacturing and Control, In-vitro Dissolution
Testing and In-Vivo Bioequivalence Documentation. Rockville, MD: FDA Center for
Drug Evaluation and Research, November 30, 1995.
138. Gore AY, McFarland DW, Batuyios NH. Fluid bed granulation: factors affecting the
process in laboratory development and production scale-up. Pharm Technol 1985;
9(9):114.
139. Bonck JA. Spray granulation. Presented at the AIChE Annual Meeting, November,
1993.
140. Jones DM. Factors to consider in fluid bed processing. Pharm Technol 1985; 9(4):50.
141. Matharu AS, Patel MR. A new scale-up equation for fluid bed processing. Poster at
AAPS Annual Meeting, 2003.
142. Rambali B, Baert L, Massart DL. Scaling up of the fluidized bed granulation process.
Int J Pharm 2003; 252(1–2):197–206.
143. Simon EJ. Fluid bed processing of bulk solids. In: Third International Powder Technology
and Bulk Solids Conference, February 20, 1975. Harrogate, U.K.: Heyden & Sons
Publisher, 1975:63–73.
144. Kulling W. Method and Apparatus for Removing a Vaporized Liquid from Gas for Use
in a Process Based on the Fluidized Bed Principle. U.S. Patent 4,145,818, March 27,
1979.
145. ATEX Directive 94/9/EC of the European Parliament and the Council, March 23,
1994.
146. Parikh DM. Fluid bed processing in the 1990s. In: Tabletting and Granulation Yearbook.
Pharm Technol 1996; suppl:40–47.
147. International Patent WO 93/08923, 1993.
148. International Patent WO 95/20432, 1995.
308 Parikh and Mogavero
© 2005 by Taylor & Francis Group, LLC
149. Glatt Brochure on STRATOS2.
150. Walter KT, et al. Fluidized bed precision granulation with real time process determination.
Poster at AAPS Annual Meeting, 2003.
151. Doelling MK, et al. The development of a microwave fluid bed processor I: construction
and qualification of a prototype laboratory unit. Pharm Res 1992; 9(11):1487–1492.
152. Doelling MK, Nash RA. The development of a microwave fluid bed processor II: drying
performance and physical characteristics of typical pharmaceutical granulations. Pharm
Res 1992; 9(11):1493–1501.
153. Luy B, Hirschfeld P, Leuenberger H. Granulation and drying in vacuum fluid-bed system.
Pharm Ind 1989; 51:89.
154. U.S. Patent 3,671,296, June 20, 1977.
155. U.S. Patent 4,034,126, July 5, 1977.
156. U.S. Patent 4,542,043, Sep 17, 1985.
157. Ja?ger KF, Bauer KH. Effect of material motion on agglomeration in the rotary
fluidized-bed granulator. Drugs Made Germany 1982; 25:61–65.
158. U.S. Patent 4,740,390.
159. Tu?rkoglu M, He M, Sakr A. Evaluation of rotary fluidized-bed as a wet granulation
equipment. Eur J Pharm Biopharm 1995; 41(6):388–394.
160. Li Shun P, Kowarski CR, Feld KW, Grim WM. Recent advances in microencapsulation
technology and equipment. Drug Dev Ind Pharm 1988; 14(2,3):353–376.
161. Robinson RL, Hollenbeck G. Manufacture of spherical acetaminophen pellets: comparison
of rotary processing with multiple-step extrusion and spheronization. Pharm Technol
1991; 15(5):48–56.
162. Parikh DM. Layering in rotary fluid bed a unique process for the production of spherical
pellets for controlled release. Interphex, New York, May 18, 1991.
163. Hileman GA, Sarabia RE. Manufacture of immediate and controlled release spheres in
a single unit using fluid bed rotor insert. Poster PT 6167 at AAPS Annual Meeting,
1992.
164. Iyer RM, Augsburger LL, Parikh DM. Evaluation of drug layering and coating: effect
of process mode and binder level. Drug Dev Ind Pharm 1993; 19(9):9891–9998.
165. U.S. Patent 5,132,142, 1992.
166. Korakianti ES, Rekkas DM, Dallas PP, Choulis NH. Optimization of the pelletization
process in a fluid bed rotor granulator using experimental design. AAPS PharmSciTech
2000;1(4):article 35.
167. Pis?ek R, Planins?ek O, Tus? M, Src?ic? S. Influence of rotational speed and surface of rotating
disc on pellets produced by direct rotor pelletization. Pharm Ind 2000; 62:312–319.
168. Vertommen J. Pelletization in a rotary processor using the wet granulation technique.
In: ACTA Biomedica Lovaniesia Leuven. Belgium Leuven University Press, 1998:166.
Batch Fluid Bed Granulation 309
© 2005 by Taylor & Francis Group, LLC
10
Single-Pot Processing
Harald Stahl
Niro Pharma Systems, Muellheim, Germany
Griet Van Vaerenbergh
Collette N.V., Wommelgren, Belgium
1. INTRODUCTION
Single-pot processing was developed to provide the means for mixing, granulating,
drying, and blending pharmaceutical granulations in a single apparatus. Although
gory of processors consists of a high- or low-shear mixer–granulator (similar to conventional
granulators) and outfitted with a variety of drying options. Initially,
vacuum was combined with a heat-jacketed bowl to provide the means for drying
in the single pot. Today, processors are available that provide vacuum drying with
microwaves or that percolate gas under low pressure into the vacuum chamber
(i.e., processing bowl). Another very interesting improvement is the use of swinging
processing bowls (1).
Single-pot processors for the pharmaceutical industry have been available for
years. They received renewed interest in the mid-1980s when microwaves were
coupled with vacuum to enhance the drying operation. Microwaves applied to drying
pharmaceutical granulations became synonymous with single-pot processing,
and there was anticipation that this technology would eventually become the norm
for granulation processing. Several major pharmaceutical companies purchased production-
scale units following successful trials conducted at vendors’ pilot facilities.
In the ensuing years, when the technology did not become as popular as expected,
there were rumors that microwave systems could not be validated and suffered from
excessive regulatory hurdles. The reality is neither, and single-pot technology has
continued to evolve during the last decade. It has de-emphasised its association with
microwave drying while continuing to demonstrate its appropriate role in granulation
technology (2).
In a production setting, single-pot processing may offer a number of advantages.
By integrating granulating and drying capabilities into a single unit, capital
investment in equipment and good manufacturing practice floor space may be lower
than the other alternatives. The number of material-handling steps is decreased; consequently,
the total processing time may be shorter while maintaining a high yield
311
equipment design varies from manufacturer to manufacturer (Figs. 1–3), this cate-
© 2005 by Taylor & Francis Group, LLC
and keeping production support to a minimum. Environmental variables, such as
humidity, are eliminated from the manufacturing process, which may offer advantages
for processing moisture-sensitive formulations. State of the art is to outfit a single-
pot processor with clean-in-place systems, thereby enhancing operator safety by
minimizing exposure to the product both during manufacturing and during cleaning
Figure 2 L.B. Bohle single-pot processor. (Courtesy of the L.B. Bohle Group, Germany.)
Figure 1 UltimaProTM 600 microwave/vacuum single-pot processor with swinging bowl.
(Courtesy of Collette NV, Belgium.)
312 Stahl and Van Vaerenbergh
© 2005 by Taylor & Francis Group, LLC
(3). Requirements for solvent recovery systems are lower for single-pot processors
compared with fluid-bed driers. Single-pot processors outfitted with vacuum are
attractive for evaporating events that are explosive or for containing drug substances
with low-exposure limits.
The versatility and compactness of small-scale (3–25 L) single-pot processors
also make the technology attractive for development and pilot laboratory facilities.
Within the last few years, equipment manufacturers began offering single-pot processors
that can accommodate the batch sizes required during early development
(0.3 g–10 kg). The processors can be used as mixer–blenders for direct compression
formulations, or as mixer–granulators to prepare wet granulations for fluid-bed drying,
or utilized for their full range of capabilities as a single-processing unit for all the
steps required for granulation preparation. Some vendors offer the option of upgrading
their small-scale processors. For example, a user can initially purchase a singlepot
processor with vacuum-drying capabilities and add a microwave drying system
at a later time. Consequently, single-pot processors should be given strong consideration
when equipping a development laboratory or a pilot plant intended to offer
a variety of processing options to the pharmaceutical formulator.
The following text is intended to expose the reader to the drying methods, the
capabilities, and the applications of single-pot processing to pharmaceutical granulations.
Fluid-bed technology, which can also be used to mix, to granulate, and for
drying alternative wet granulation technologies in a single unit, will not be addressed
in this chapter because it is discussed elsewhere in this book.
2. TYPICAL SINGLE-POT PROCESS
The steps and sequence of manufacturing pharmaceutical granulations using singlepot
processing are the same as those that use alternative technologies, except that
Figure 3 Zanchetta single-pot processor. (Courtesy of Zanchetta, Italy.)
Single-Pot Processing 313
© 2005 by Taylor & Francis Group, LLC
several of the steps are performed in the same product chamber. The majority of production
installations make use of its mixing, granulating, and drying capabilities during
the processing of a single batch. Important to note is the absence of a milling step
in between granulation and drying. The advantage of a reduced number of process
steps is obvious while the drawback is that there is no possibility of breaking lumps
generated during granulation. Single-pot operations therefore require excellent control
of the granulation to prevent the formation of oversized material.
2.1. Dry Mixing
Powders are loaded into the single pot either manually (for development-and pilotscale
units) or by a conveying system (for production-scale units). Vacuum pump(s)
used for the drying operation can also be used to charge the processor. Pneumaticand
vacuum-conveying systems contribute to minimizing operator exposure to the
drug product. The powders are mixed in the dry state until the desired degree of uniformity
is obtained. Depending on the geometry of the processing bowl and the effi-
ciency of its mixing blades, optimal mixing for most processors generally occurs
when the bowl is charged to 50–75% of its capacity. Batch size, impeller speed,
and mixing time are the variables that affect the desired degree of blend homogeneity
before the addition of the binder solution.
2.2. Addition of Binder Solution
Once dry mixing is completed, the binder solution is added through a spray lance
connected to a solvent delivery system (such as a pressure pot or peristaltic pump).
For highly viscous binder solutions it is of advantage to use the vacuum system of
the processor for sucking into the processor. Because the single-pot processor is
operating as a granulator at this point, all variables considered during the manufacture
of wet granulations in conventional high-or low-shear mixer–granulators are
applicable. Those variables include the rate of binder action, droplet size, and spray
pattern (the last two being determined by the selection of the spray nozzle and the
distance between the nozzle tip and granulation bed). The speed of the main impeller
and the granulating tool (e.g., high-intensity chopper bar), as well as the jacket temperature
should also be controlled during binder addition.
2.3. Wet Massing
Following binder addition, additional energy may be imparted to the granulation
until the desired consistency is obtained. The speeds of the main impeller and granulating
tool, wet-massing time, and jacket temperature are variables that can affect
the physical attributes of the granulation. Like the bowl shape the impeller design
will also affect the amount of shear imparted to the granulation. Granulation end
point may be controlled by process time, temperature of the product bed, and the
energy consumption or torque of the main impeller.
2.4. Drying
After granulation, the material is dried using one of three approaches: (a) vacuum
drying, (b) gas-assisted vacuum drying, or (c) microwave vacuum drying. Details
314 Stahl and Van Vaerenbergh
© 2005 by Taylor & Francis Group, LLC
of each drying method are summarized in Drying Methods for Single-Pot
Processors. The product bed is usually stirred at low intensity during the drying process
to facilitate solvent removal and promote uniform drying, as well as to prevent
caking of the granulation on the chamber’s walls. Agitation may be applied by
slowly tilting the bowl or operating the impeller at low speed, either continuously
or intermittently, throughout the drying stage. Caution must be exercised to avoid
granule breakdown during drying, which may result in unfavorable compression
characteristics (4). Variables for vacuum drying include the level of vacuum maintained
in the bowl, the jacket temperature, and the degree of agitation. In addition
to the parameters listed for vacuum drying, gas-assisted vacuum drying must also
consider the type of drying gas used and its rate of delivery. When microwave
vacuum drying is used, all of the variables used for vacuum drying are applicable,
as well as the level of microwave power used to dry the granulation. If yield is of
greater importance than process time, a very interesting process option is to follow
with the wall temperature very closely the product temperature and use as a source
of drying energy only the introduced microwave energy. This mode of operation
minimizes the amount of material sticking to the walls for the price of a prolonged
drying operation resulting in a reduced throughput. This option is of special interest
for the processing of highly expensive materials.
If required, cooling can be conducted at the conclusion of the drying operation.
The heated water or steam in the bowl jacket, which supplied conductive heat during
the drying process, can be replaced with a glycol–water solution to provide a contact
surface as low as 10
C. Another approach to cool the granulation is by purging
a cooling gas into the single pot while agitating the granulation bed.
2.5. Sizing and Lubrication
Once the granulation is dried, it is usually necessary to size it. This may be accomplished
by discharging the material through an in-line mill into a receiving vessel,
where it may be blended with any remaining excipients (e.g., lubricant and flavors).
This process design maintains the containment benefits of the single-pot process.
Alternatively, the remaining excipients may be added to the single-pot processor
and blended with the granulation before discharging and milling. This approach
requires that the lubricant be adequately distributed during milling and material
transfer during the compression operation.
3. DRYING METHODS FOR SINGLE-POT PROCESSORS
3.1. Conductive Drying
The bowls of single-pot processors are generally jacketed for temperature control,
which minimizes condensation of the granulating solvent and assists in solvent evaporation
during drying. As a result, conductive heating provided by the heatjacketed
lid and the walls of the single pot contributes to the drying process. Its
dependence on the transfer of heat through pharmaceutical powders, which are poor
conductors of heat, prevents its use as the sole mode of drying in single-pot processors.
Eq. 1 addresses the conductive drying component for solvent removal.
Single-Pot Processing 315
© 2005 by Taylor & Francis Group, LLC
Heat transfer from the vessel walls to the granulation bed is governed by
Q ? hSDT ?3:1?
where Q ? the energy exchange, h ? the exchange coefficient, S ? the contact surface
of the heated wall, and DT ? the temperature difference between the contact wall
and the granulation.
The rate of drying can be facilitated either by increasing the contact area
between the granulation and the vessel walls (which can be achieved by agitating
the product or utilizing a tilting bowl) or by maximizing the temperature differential
between the vessel walls and product (either through increasing the jacket temperature
or through maintaining the temperature of the product as low as possible during
processing).
Eq. 2 is a simple relation that may be of some value for the scaling-up of processes
using conductive heating (5)
tb=ta ? ?A=V?a?A=V?b ?3:2?
where t ? drying time, A ? heat transfer surface (m2), V ? vessel working volume
(m3), a ? refers to pilot scale, and b ? refers to production scale.
This relation accounts for the ratio of the surface area of the jacketed bowl and
the volume of the product requiring drying.
3.2. Vacuum Drying
Single-pot processors using vacuum drying may be considered if the product must be
dried at low temperature (<40
C), if solvent recovery is required, or if the potential
for explosion is high. A vacuum is maintained within the vessel, thereby lowering
the temperature at which the granulating solvent evaporates. Because vapors are
removed from the processing bowl, vacuum drying provides a convenient means
for solvent recovery.
De Smet (6) has discussed the theory, advantages, and limitations of vacuum
drying. Aqueous granulations require a large amount of energy during drying, which
is generally supplied by the transfer of heat through conduction from the jacketed
bowl to the product. The amount of energy required for water removal is dependent
on the level of vacuum applied to the vessel and the osmotic pressure of dissolved
substances. As additional material dissolves in the water, the osmotic pressure
increases and additional energy is necessary to drive off the water. Therefore, as
the material becomes drier, the amount of energy necessary to evaporate the water
increases, and the rate of evaporation slows down. Processing times in vacuum driers
are often long, owing to the limited contact of the granulation with the heat from the
jacketed walls, and the slow rate of evaporation of the solvent from the interior of
the granules.
The drying rate of the vacuum component is dependent on the following relation:
V ? ksDP ?3:3?
where V ? evaporation rate, k ? rate coefficient, s ? total surface of granules,
and DP ? the vapor pressure difference between the granules and the surrounding
space.
316 Stahl and Van Vaerenbergh
© 2005 by Taylor & Francis Group, LLC
The rate of drying can be facilitated by increasing the level of vacuum (i.e.,
decreasing the pressure within the bowl) to increase the differential between granule
and bowl vapor pressure. Figure 4 is a typical drying curve for a vacuum-drying
process. When moisture content in the granulation is high, the rate of drying is
constant because evaporation of solvent from the product surface readily occurs.
As the level of moisture on the granule surface decreases, water must migrate from
the interior of the granule before evaporation. As a result, the rate of evaporation
progressively decreases.
During vacuum-drying processes, one should be aware of various problems
that could arise. For example, granule damage may occur owing to excessive
attrition as the bed is agitated during drying. Vacuum systems should contain adequate
filtering or blowback to prevent the loss of granulation ‘‘fines’’ through the
vacuum line, which may compromise the drug uniformity within the processed
batch.
A condenser positioned between the processor and vacuum pump, should
always be used, especially for granulations manufactured using organic solvents.
The condensate must be sufficiently cooled to prevent it from being released into
the atmosphere. Also, filters may become blocked owing to condensation forming
on the filter or the entrapment of solid particles. Blockage of the filters reduces
the level of vacuum that can be pulled on the bowl and excessively strains the pump
itself.
3.3. Gas-Assisted Vacuum Drying
Accelerating the drying process in vacuum driers is often limited by characteristics of
the product or equipment. Bowl temperature is generally limited by the physicochemical
stability of the product, which can limit the use of higher temperatures
to expedite drying. Increasing the contact area between the product and the vessel
is difficult without significantly altering the design of the equipment. Excessive agitation
of the product can lead to considerable granule attrition, which can lead to poor
granulation flow and compression properties.
Figure 4 Typical curve for vacuum drying. (Courtesy of Manufacturing Chemist).
Single-Pot Processing 317
© 2005 by Taylor & Francis Group, LLC
Gas-assisted vacuum drying improves the efficiency of single-pot processors
that use vacuum drying by continuously introducing a small stream of gas through
the granulation to facilitate solvent removal. Drying continues to be performed at
lower temperatures (compared with tray and fluid-bed drying), but at shorter-processing
times than vacuum drying alone.
The gas may be introduced into the unit through openings in the bottom of the
vessel or through the mixing blades (Fig. 5).
Compressed air or nitrogen (mixed with or without air) is the commonly used
gas for these units. The rate of gas flow and the level of vacuum applied to the bowl
can be adjusted for a specific product to produce optimal drying conditions.
The introduction of gas into a vacuum chamber facilitates the drying process
through several actions. The constant flow of gas through the product improves
the transport of moisture from the product to the vacuum-solvent recovery system
(7). Introducing gas into the bowl also increases the vapor pressure driving force
(8). The pressure gradient across the vessel is increased, resulting in a reduction
in the rate at which water molecules recombine, producing a net increase in the
rate of evaporation. This causes the product temperature to be reduced, which
increases the temperature differential between the granules and the bowl wall.
The gas also reduces drying time by increasing the heat transfer coefficient from
the bowl to the bed. In addition to improving the heat transport through the
bed, the gas can reduce or eliminate product sticking to the sides of the vessel walls
because it improves flow and dries the particle surfaces more quickly. As the vessel
wall is the only notable source of drying energy this technology is best used in the
case of:
Figure 5 Gas-assisted vacuum-drying principle. (Courtesy of Collette NV, Belgium.)
318 Stahl and Van Vaerenbergh
© 2005 by Taylor & Francis Group, LLC
 Small batch sizes (good surface volume ratio)
 Heat-insensitive materials (allows to operate a higher wall temperature)
 Organic solvents (requires only a fraction of the energy water needs for evaporation).
Additionally, the boiling temperature corresponding to the actual vacuum level
is much lower for organic solvents than for water, generating a larger temperature
difference between product bed and vessel.
3.4. Microwave Vacuum Drying
High-shear granulators with microwave vacuum-drying capabilities provide the fastest
drying rates in the family of single-pot processors. Microwave drying is based on
the absorption of electromagnetic radiation by dielectric materials, the theory of
which has been extensively described (9–11). Microwaves are a form of electromagnetic
energy similar to radio waves, the frequencies of which fall between 300 and
3000 MHz (between radio and optical waves, (Fig. 6). The two frequencies allocated
for domestic, scientific, medical, and industrial purposes are 915 and 2450 MHz.
Pharmaceutical processors generally use 2450 MHz, because this frequency is more
desirable when used in conjunction with vacuum. Single-pot processors incorporating
microwave drying are constructed of stainless steel because metal is a common
reflector of microwave energy and contains the energy within the processing chamber.
Teflon is essentially inert to microwaves, making it a suitable material for components
required in the processing bowl (e.g., spray lance and temperature probe).
Energy absorption of materials exposed to microwaves is described by Eq. 4
(10,11):
P ? 2pfV2E0Er tan d ?3:4?
where P ? the power density of the material (W/m3), f ? frequency (Hz), V ? voltage
gradient (V/m), E0 ? dielectric permissivity of free space (8.85  1012 F/m), Er ? dielectric constant of the material, and tan d ? loss tangent.
Figure 6 The electromagnetic spectrum.
Single-Pot Processing 319
© 2005 by Taylor & Francis Group, LLC
For a constant electric field strength V the term (2pfV2E0) is constant. Therefore,
the power absorbed is proportional to the term (Er tan d), called the loss factor,
which is a relative measure of how easily a material absorbs microwave energy.
Various materials commonly used in pharmaceutical formulations have low
loss factors and only absorb microwave energy at high field strengths. Solvents used
in the granulation process (water, ethanol, isopropanol, and such), however, possess
high loss factors relative to the pharmaceutical powders (10). The dipolar component
of the solvents couples with the high-frequency electromagnetic field producing high
heating rates for the solvent, resulting in its evaporation and subsequent removal
from the processing chamber. Table 1 lists the loss factors for various components
in a typical pharmaceutical granulation (12).
Lucisano and Moss (13) performed a study in which a microwave drying process
was conducted in two different processors, one using fixed-output magnetrons
and the other using a variable-power magnetron. The unit using fixed-output magnetrons
had difficulty obtaining low moisture levels (<0.3%) because the E-field
safety set point was exceeded when the moisture was <1%. This problem did not
occur for the unit using the variable-power magnetron because forward power was
reduced as the E-field increased. During the late stages of drying, the unit was primarily
functioning as a vacuum dryer, as the amount of microwave energy being
introduced into the bowl was minimal. Nowadays, most microwave driers use the
variable output magnetrons.
The use of vacuum during microwave drying lowers the temperature at which
the solvent volatilizes, thereby limiting the temperature to which the material is
exposed. For example, at a vacuum of 45 mbar, water-based granulations will dry
at 31
C. Once most of the water is removed from the process, the temperature
of the material will rise as components in the mixture with lower loss factors start
to absorb the microwaves. If too much vacuum is applied to the system, there is a
potential for granule breakdown owing to the excessive pressure between the core
and the surface of the granules. Microwaves are typically applied in a vacuum range
of 30–100 mbar. Introducing microwave into a vacuum <30 mbar risks ignition of
the surrounding atmosphere, a condition known as ‘‘arcing.’’
Control of the drying process is achieved through the simultaneous measurethe
level of microwave forward power, microwave reflected power, and product temperature
at various times during the drying process. During the initial stage, the product
temperature remains relatively constant as the free solvent is preferentially
evaporated, and the reflected power remains relatively low. The amount of vacuum
applied to the bowl, and to a lesser extent the bowl jacket temperature, will affect the
actual product temperature observed. As drying progresses at a constant rate of forward
power, the amount of absorbed energy decreases as the material dries, thereby
increasing the amount of free energy. As the free energy increases, a corresponding
increase in the reflected power is also observed.
The rise in reflected power is accompanied by an increase in product temperature,
which is simultaneously monitored while the magnetron output power is
reduced. This is necessary because the loss factors for some pharmaceutical components
are so small that very low moisture can be achieved before the temperature
rises. For such materials, the reflected power can rise sharply once most of the solvent
has evaporated, resulting in significant temperature gains.
The rise in temperature and reflected power signifies that the end of the drying
process is approaching. Several factors, such as the loss factors of the formulation
320 Stahl and Van Vaerenbergh
ment of product temperature, forward power, and reflected power. Figure 7 depicts
© 2005 by Taylor & Francis Group, LLC
components, the microwave power, and the solvent retention properties of the solids
influence the point at which the previous relation will occur.
For example, lactose has a low loss factor and shows a sharp rise in reflected
power, followed by a slow temperature rise. Conversely, starch has a high loss factor
and demonstrates a fast temperature rise followed by a slow rise in reflected power.
4. OTHER PROCESSES AND APPLICATIONS
Because of the different technologies incorporated into a single-pot processor, it is
capable of executing many different processes, apart from the standard wet granulation
and drying process, while small modifications or additional options can extend
the flexibility even further.
This chapter discusses some of the possible ‘‘special’’ processes and applications
in a single-pot processor. Although many of these processes are used in the
pharmaceutical industry, scientific literature on them is rare. The main reason for
this is that many of these processes were developed by the pharmaceutical industry
as product-specific solutions. This does not imply, however, that these processes cannot
be used more widely.
4.1. Melt Granulation
Melt granulation is a process in which the binder solution of the standard wet granulation
process is replaced with a meltable binder such as a wax or polyethylene glycol
(PEG), which is generally added in solid form, and melted during the process by
adding the necessary energy.
The most common production technique for melt granulation uses extruders,
but melt granulation in a high-shear mixer has also been extensively described in
the literature.
Figure 7 Relationship between microwave forward power, microwave reflected power, and
product temperature during microwave vacuum drying.
Single-Pot Processing 321
© 2005 by Taylor & Francis Group, LLC
In this process, the necessary energy to melt the binder is provided either
by the mixer arm (mainly in laboratory-scale equipment) or by a heated jacket
(14–16).
If the meltable binder used absorbs microwaves, however (such as PEG for
example), using a single-pot processor equipped with microwave drying can present
major time-savings to the production process.
Providing the melting energy by the impeller or the heated jacket can be a very
time-consuming process, especially in production-scale equipment. Microwaves (as
explained earlier) are an instant source of energy that penetrates into the product
and can provide the energy faster and immediately where needed.
In a comparison between the use of the heated jacket to melt the binder and the
use of the heated jacket supplemented with microwaves, the latter method not only
proved to be more than twice faster in melting the binder used (PEG 3000), but also
the granulation step was reduced threefold (17).
At this moment, there is no literature available about other meltable binders in
combination with microwaves.
However, the results from the previously mentioned study show that it is worthwhile
to consider a single-pot processor for the production of melt granulations.
Another step where a single-pot processor can present a major advantage compared
to the standard production techniques, and especially compared to the process
in a high-shear mixer, is the cooling step. To achieve a stable ‘‘dry’’ granule from a
melt granulation process, the product needs to be cooled down to room temperature.
In a high-shear mixer, the cooling process is done by circulating cold water or a glycol–
water mixture in the bowl jacket. As the contact surface between the product and
the jacket is limited, the cooling process generally takes a long time. If the process is
executed in a single-pot processor equipped with a gas-assisted vacuum-drying system,
this system can be used to pass cold air or even liquid nitrogen through the product
to aid the cooling process and reduce the cooling time considerably. If liquid
nitrogen is used, even a fivefold reduction of the cooling time is achievable (17).
A related chapter in this book should be consulted to gain more understanding of
melt granulation.
4.2. Pellet Production
For the production of spheres or pellets, in most cases an extrusion/spheronization
process is used.
There are, however, many references in scientific literature detailing the production
of pellets using a high-shear mixer, most of which concern melt pelletization
(15–21). Taking into account the explanations given earlier on melt granulation, a
single-pot processor can, of course, also be used for this process for the same
reasons.
Also, for other pelletization processes, not using meltable binders, the use of a
single-pot processor can be advantageous. In scientific literature, there are some
references describing the use of a high-shear mixer for such processes (22,23), but
so far none can be found about single-pot processors. Nevertheless, a standard pellet
formulation often contains microcrystalline cellulose, which needs high water content
to obtain a good granule/pellet quality. Drying the pellets is always a part of
the production process. The advantage of a single-pot processor is that the whole
process of pelletization and drying can be executed in the same equipment, making
322 Stahl and Van Vaerenbergh
© 2005 by Taylor & Francis Group, LLC
product transfers redundant and thereby reducing the risk of product loss and contamination
and enhancing containment and operator safety.
To enhance the pelletization/spheronization process, many vendors of highshear
mixers/single-pot processors offer special mixing tools for producing pelletlike
granules. Depending on the geometry of the equipment, this special mixing tool
has either more (up to six) or less (up to two) mixing blades than a standard mixing
tool, which generally has three mixing blades (Fig. 8). All special pellet-mixing tools
have, however, the same purpose: to simulate the product behavior in a spheronizer
and enhance the spheronization process that occurs in the mixer.
4.3. Effervescent Production
The production of effervescent tablets is first of all a conventional solid dosage form
manufacturing process, which has to be taken into consideration, due to the special
characteristics of the product and some unusual features.
For granulation of effervescent products, many different production techniques
can be used, ranging from dry granulation methods over two-step granulation
(granulating acid and alkali phase separately) to one-step granulation using water or
organic solvents.
For the one-step granulation methods, the use of a single-pot processor offers
many benefits. Apart from the overall benefit of eliminating product transfer between
a granulator and a dryer, a single-pot processor allows easy solvent recovery by condensation
in case organic solvents are used as granulation liquid, compared to the
quite complex system for the exhaust gas treatment required for a fluid-bed dryer.
When water is used as granulation liquid for effervescents, the effervescent
reaction will start and cause a chain reaction. The critical point in such a process
is to stop this reaction at the correct time by evaporating the water created by this
reaction. In a single-pot processor, this can be very easily and accurately achieved
by switching on the vacuum-drying system (possibly supplemented with gas-assisted
drying or microwave drying) (24).
Figure 8 Example of a special pelletizing mixer arm. (Courtesy of Collette NV, Belgium.)
Single-Pot Processing 323
© 2005 by Taylor & Francis Group, LLC
4.4. Crystallization
Although a high-shear mixer/single-pot processor is mainly intended for powder
processing, there is a possibility of executing crystallization or recrystallization processes
in this type of equipment as well. Starting either from a solution or from a
powder, which is dissolved in a suitable solvent, the speed of the drying process
can be controlled to achieve the desired crystallization process. Using vacuum drying
only, the drying process will be slow and gradual. Temperatures during this process
will remain low, creating an environment for slow crystal growth. When microwaves
are used for drying, the process will be considerably faster and the temperature of the
product will most likely be higher then under ‘‘pure’’ vacuum conditions. The crystals
that result from such a crystallization process will have different characteristics
from those from the vacuum-drying process. When choosing the appropriate settings
for the drying/evaporation process, combining vacuum and microwave drying, crystals
with specific properties can thus be obtained.
The main advantage of executing a (re) crystallizing process in a single-pot processor
is that the possibility exists of granulating the product at the same time by
varying the mixer speed during the drying process. The resulting product will be suitable
for tableting without the necessity of executing other processing steps (apart
from lubrication).
5. SCALE-UP OF DRYING PROCESSES
Because the rate of solvent removal during vacuum drying is dependent on a favorable
surface area/volume ratio, the drying time in a vacuum processor often increases
substantially during scale-up as described by Formula (3.2). Microwave vacuum
drying is relatively insensitive to the surface area/volume ratio and does not suffer
the same inefficiency as vacuum drying during transition from pilot to production
scale. Pearlswig et al. reported successfully scaling a microwave vacuum-drying
process for a moisture-sensitive formulation that required a drying end point of
<0.2% (25). The drying time remained within a 30–45 min range throughout the
scale-up from 15 kg (Vactron.75) to 300 kg (Vactron.600), whereas the time for
vacuum drying increased threefold. After additional scale-up to 600 kg in a
Vactron.1200, the drying time rose slightly to a 50–55 min range. Poska also
reported attaining equivalent drying times when scaling up in Spectrum processors
for lactose–starch granulation prepared in single-pot processors using vacuum
drying, gas-assisted vacuum drying, or microwave vacuum drying. Although not
as rapid as microwave vacuum drying, gas-assisted vacuum drying can decrease
the drying time by up to 50% compared with that of vacuum drying alone. As the
wall of the vessel is the only source for drying energy, also in this case the scale is
of major importance for the drying time (27).
When performing feasibility trials on a development- or pilot-scale single-pot
processor, it is important to be aware of the maximum energy input capacity of
for each drying method in a pilot- and production-scale single-pot processor are
shown. While for processors equipped with microwaves, the drying time is relatively
independent of the scale, a significant increase in drying time is observed for vacuumand
gas-assisted vacuum-drying processes.
324 Stahl and Van Vaerenbergh
ranging from 65 to 300 L bowl size (26). Figure 9 compares typical drying curves
the corresponding production-scale processor. In Figures 9 and 10, the drying times
© 2005 by Taylor & Francis Group, LLC
Figure 10 Comparison of drying curves for different modes of drying in a 600 L Ultima-
ProTM. (Courtesy of Collette NV, Belgium.)
Figure 9 Comparison of drying curves for different modes of drying in a 75 L UltimaProTM.
(Courtesy of Collette NV, Belgium.)
Single-Pot Processing 325
© 2005 by Taylor & Francis Group, LLC
6. CLEANING
As a single-pot processor is often used for highly potent products, or more generally
for contained production, it is important that also cleaning of the machine can be
executed in a contained, automated fashion to eliminate the risk of operator exposure
to the active product.
All single-pot processors on the market nowadays are equipped with a more or
less extensive clean-in-place system. Vendors have made a great effort to optimize
their design enabling easy cleaning in place, which can be validated (3). The focus
has been to eliminate any dead spots in the equipment where the cleaning water cannot
reach and including cleaning spray balls in critical product contact areas such as
the product filter and discharge valve.
Drying of the single-pot processor after the cleaning cycle to prepare the equipment
for the next batch can be done using the system’s own vacuum-drying system
and jacketed bowl, making a separate drying unit redundant.
A nice case study of an evaluation of a clean-in-place system on a pilot-scale
single-pot processor is given in Ref (3). In this study it was proved that a complete
changeover from one product to the next can take place in <2 hr.
7. PRODUCT STABILITY
The stability of pharmaceutical granulations dried by microwaves is comparable with
that provided by alternative methods. Microwaves are nonionizing and do not
possess the amount of energy required for the formation of free radicals or the
Since the introduction of microwave drying at the end of the 1980s, numerous
new and supplemental drug applications that include the use of microwave vacuum
drying of wet granulations have been approved by the Food and Drug Administration
(FDA). We are unaware of any instance in which the FDA required additional
stability or analytical testing beyond that normally required for other methods of
manufacture. Mandal (28), Moss (29), and others (26,30) have also published or presented
data showing the comparability of the physicochemical characteristics of
granulation dried in microwave processors vs. tray driers and fluid-bed driers.
When microwaves were first introduced, some authors (31) concluded that
microwave drying could not be generally recommended because of the inability to
control the microwaves after they enter the drying cavity and the risk of unacceptable
thermal damage to active substances with high loss factors (i.e., high dielectric
constants). The routine production of pharmaceutical products by several installations
of single-pot processors using microwave vacuum drying indicates that their
general concerns can be addressed by the proper selection of formulation components
and process parameters.
8. REGULATORY CONSIDERATIONS
Single-pot processors combine established technologies into a single piece of equipment
and, in general, deserve no special regulatory consideration when using them to
develop a new product or to manufacture an approved product. Robin and colleagues
(4) surveyed eight European regulatory agencies in 1992 to determine the
326 Stahl and Van Vaerenbergh
liberation of bound water conditions that foster product instability (Fig. 6).
© 2005 by Taylor & Francis Group, LLC
implications of converting from fluid bed drying to microwave vacuum drying within
a single-pot processor. The majority of the agencies required only process validation
data and three suggested limited stability data (up to 6 months of accelerated data).
These requests were no different from those expected for similar types of manufacturing
changes (i.e., change in process or equipment).
Manufacturers considering converting to a single-pot process for an immediaterelease,
solid oral dosage form (tablets, capsules, or the like) with an approved
manufacturing process should consult their appropriate regulatory agencies
governing the practices they use to manufacture their products. For drug products
sold in the United States, manufacturers should refer to FDA’s SUPAC IR
marketed products. This document describes the levels of change that may be made
in a manufacturing process and equipment. It outlines the chemistry, manufacturing,
and control tests and documentation for each level of change as well as the appropriate
regulatory filing (Annual Report, Prior Approval Supplement, or other).
For example, a tablet formulation is currently granulated using a high-shear
granulator and dried in a tray drier. A drug manufacturer wishes to replace this process
with a single-pot processor that incorporates high-shear granulation and gasassisted
vacuum drying. This conversion would be viewed as a change in equipment
to a different design and different operating principles (defined as a Level 2 Equipment
Change in the SUPAC IR Guidance document). Such a change requires the
manufacturer to submit a Prior Approval Supplement with up to three batches with
3-months accelerated stability data (depending on the duration of commercial
experience with the product). The submission would also require updated batch
records including the new equipment, and the generation of multipoint dissolution
profiles. The requirements for this conversion are, however, the same as those for
a conversion of a tray drier to a fluid-bed drying process.
9. VALIDATION OF SINGLE-POT PROCESSORS
Because single-pot processors combine standard engineering approaches into a single
processor, their validation should pose no special problems. Other sources adequately
describe the validation of granulating and drying processes (33), although
validation of the microwave drying system and the approach to process control of
drying end point deserves special mention.
For operational qualification of microwave components, such as forward and
reflected power, and arc detection, we suggest that customers contact the vendors
because of the specialized nature of microwave systems. The cost associated with
the calibration equipment is difficult to justify, and microwave systems should operate
reliably following proper setup and qualification and require no more periodic
maintenance than other granulation approaches.
When microwave processors were first introduced, there was the expectation
that E-field would be a reliable indicator of the drying end point. With experience,
users found that the E-field tends to be too variable, and now view it primarily as
a safety feature monitoring the microwave field within the drying cavity. Most
microwave driers on the market today do not even include E-field monitoring any
more, because of the difficulties with validation of this system.
Product temperature, time, cumulative forward power, and reflected power are
proving to be more reliable indicators of drying end point with verification by some
Single-Pot Processing 327
© 2005 by Taylor & Francis Group, LLC
Guidance document (32), which addresses scale-up and postapproval changes for
in-process control that directly measures the moisture content of a product sample.
The industrial processes in operation today all use one of these approaches for endpoint
determination.
With the pharmaceutical industry moving toward the use of process analytical
technology (PAT), however, the equipment vendors are also investigating the possibility
of applying this technology to single-pot processors and thereby eliminating
the necessity of taking product samples. The first single-pot processors with PAT will
undoubtedly be introduced into the market shortly.
10. CONTROL SYSTEMS AND DATA ACQUISITION SYSTEMS
Users of single-pot processors may use one or all of its processing features (mixing,
granulating, drying, or other). The accompanying data acquisition system must collect,
display, and record the relevant processing conditions and granulation behavior
for each cycle. The degree of sophistication of the system may depend on the venue
in which the processor is used.
In a development setting, the sequence of cycles is often interrupted to collect
samples for analysis, and the user is interested in capturing as much information as
possible to assist in defining a suitable processor for a particular formulation. In
many cases, the control system for a development environment can be limited to a
manual control system. The data acquisition system, however, needs to be more
sophisticated to allow registration of all relevant processing data.
In production, the manufacturing sequence and process parameters are prede-
fined and validated, and information needs are reduced to monitoring critical parameters.
In these settings, an automatic control system with recipes to reduce operator
interventions is indispensable. The data acquisition system may or may not be as
sophisticated as for the development settings, depending on the requirements for
data in the batch records.
For the purposes of trouble-shooting and trend analysis of production
operations, however, the information requirements of development and production
converge, and data acquisition systems for single-pot processors should seek to
address the needs of both types of users.
All data acquisition systems are considered to generate electronic records. If
these electronic records are used as batch documentation, the acquisition system
needs to be compliant with FDA 21 CFR part 11.
Most vendors have addressed these issues by including password control, audit
11. SAFETY
The primary safety concern during the granulation and drying processes is the prevention
of explosions. Bulk powder, dust clouds, and flammable vapors, all have the
potential to explode. Adequate grounding and ventilation during loading and discharging
of the vessel and controlling the various processing conditions can reduce
the risk of explosion.
In Europe, all equipment used in a potentially explosive atmosphere needs to
be compliant with the ATEX guidelines (34).
Two approaches are generally taken toward explosion protection in single-pot
processors. One consists in removing oxygen from the processing chamber and
328 Stahl and Van Vaerenbergh
trails, and point verification systems (Fig. 11).
© 2005 by Taylor & Francis Group, LLC
replacing it with an inert gas (e.g., nitrogen) before any mixing action can take place.
The removal of oxygen reduces the risk of explosion by eliminating one of the elements
necessary to create an explosion. The other approach is to design the equipment in
order to contain the explosion. Executions of 10 or even up to 16 bar of high-shear
granulators and single-pot processors are now becoming available on the market.
Apart from the measures taken to avoid or contain explosions within the processing
chamber, the electrical, electronic, and mechanical parts of single-pot processors
that will be used in potentially explosive atmospheres also need to be explosion
protected.
The leakage of microwave energy is a concern for single-pot processors that use
this drying approach. Industrial microwave processors are expected to meet the
guidelines for microwave leakage specified by the Center for Devices and Radiological
Health within FDA and by the American National Standards Institute (35,36).
The guideline is 5 mW/cm2 maximum exposure at a frequency of 2450 MHz at a
distance of 5 cm from any surface of the microwave cavity. Survey meters for the
detection of microwave leakage are relatively inexpensive and should be purchased
by users of single-pot processors that incorporate microwave drying. The survey
meters are calibrated before shipment and returned to the supplier for recalibration
at periodic intervals. Their use should be incorporated in standard operating procedures
for the equipment. Operator readings that exceed the guideline limit are often
indicative of deteriorating seals around the lid cavity.
In addition to energy leakage standards, microwave processors are designed
with safety interlocks to prevent accidental exposure. For example, the magnetrons
can be activated only if the microwave cavity (i.e., bowl of the processor) is operating
under vacuum, usually 30–100 mbar. If the vacuum falls outside this range, as in the
unlikely event that an operator inadvertently tries to open the lid during microwave
vacuum drying, the magnetrons are disabled. Vendors also incorporate additional
safeguards to ensure that the microwave power is disabled with access to the bowl.
Because of popular misconceptions about the use of microwave ovens (e.g.,
stainless steel should not be used in a microwave cavity) and electromagnetic radiation
(e.g., all types cause biological effects), a training program should be instituted
Figure 11 Screen of a typical control system for a single-pot processor. (Courtesy of Collette
NV, Belgium.)
Single-Pot Processing 329
© 2005 by Taylor & Francis Group, LLC
in any facility that uses microwave drying. This will demystify any unfounded concerns
about the technology and foster a rational approach to a sound safety and
maintenance program.
12. CONCLUSION
Over the last 30 years since its introduction, single-pot processing has developed into
a mature and generally accepted production technique.
Even if the technique historically was often used because of its specific advantages
for effervescent production, potent compounds, organic solvents, or multiproduct
facilities, practice has shown that single-pot processing is also attractive for
standard pharmaceutical solid dosage production.
REFERENCES
1. Van Vaerenbergh G. The influence of a swinging bowl on granulate properties. Pharm
Technol 2001; 13(3):36–43.
2. Lucisano LJ, Poska RP. Microwave technology-fad or the future. Pharm Technol 1990;
14:38–42.
3. Van Vaerenbergh G. Cleaning validation practices using a one-pot processor. Pharm
Technol. Europe. Feb 2004.
4. Robin P, Lucisano LJ, Pearlswig DM. Rationale for the selection of a single pot manufacturing
process using microwave/vacuum drying. Pharm Technol 1994; 18:28–36.
5. Bellini G, Pellegrini L. Non adiabatic drying. In: Goldberg E, ed. Handbook of Downstream
Processing, Chapman & Hall, 1993.
6. De Smet P. Vacuum drying. Manuf Chem Mar 1989; 37–39.
7. LB Bohle, Inc. Technical Report. Granulation and Drying with SYSTEM-VAGAS.
Bristol, PA: LB Bohle, Inc.
8. Niro-Fielder Ltd. Technical Report. AEROVAC System, Accelerated Vacuum Drying.
Eastleigh, Hampshire, U.K.: Niro-Fielder Ltd.
9. Metaxas AC, Meredith RJ. Industrial Microwave Heating. London: Peter Peregrines,
1983.
10. Doyle C, Cliff MJ. Microwave drying for highly active pharmaceutical granules. Manuf
Chem Feb 1987; 23–32.
11. Waldron MS. Microwave vacuum drying of pharmaceuticals: the development of
a process. Pharm Eng 1988; 8:9–13.
12. Poska RP. Microwave processing: the development experience revisited. Proceedings of
International Society of Pharmaceutical Engineers Congress, September, 1992.
13. Lucisano U, Moss RA. Vacuum drying vs. microwave-vacuum drying in three pilot-scale
single pot processors using sodium acid pyrophosphate as the model granulation. AAPS
Annual Meeting, San Antonio, TX, Nov 15, 1992.
14. Schaefer T. Melt Agglomeration with Polyethylene Glycols in High Shear Mixers. Ph.D.
thesis, The Royal Danish School of Pharmacy, Copenhagen, 1996.
15. Schaefer T, Holm P, Kristensen HG. Melt pelletisation in a high shear mixer. I. Effects of
process variables and binder. Acta Pharm Nordica 1992; 4:133–140.
16. Schaefer T, Mathiesen C. Melt pelletisation in a high shear mixer. IX. Effects of binder
particle size. Int J Pharm 1996; 139:139–148.
17. Van Vaerenbergh G. Melt granulation with polyethylene glycol in a one-pot processor.
Pharma. In press.
330 Stahl and Van Vaerenbergh
© 2005 by Taylor & Francis Group, LLC
18. Heng PW, Wong TW, Chan LW. Influence of production variables on the sphericity of
melt pellets. Chem Pharm Bull (Tokyo) 2000; 48:420–424.
19. Hamdani J, Moe?s AJ, Amighi K. Development and evaluation of prolonged release
pellets obtained by the melt pelletization process. Int J Pharm 2002; 245:167–177.
20. Thies R, Kleinebudde P. Melt pelletisation of a hygroscopic drug in a high shear mixer.
Part 1. Influence of process variables. Int J Pharm 1999; 188(2):131–143.
21. Voinovich D, Moneghini M, Perisutti B, Franceschinis E. Melt pelletization in high shear
mixer using a hydrophobic melt binder: influence of some apparatus and process variables.
Eur J Pharm Biopharm 2001; 52(3):305–313.
22. Vonk P, Guillaume CPF, Ramaker JS, Vromans H, Kossen NWF. Growth mechanisms
of high-shear pelletisation. Int J Pharm 1997; 157(1):93–102.
23. Ramaker JS, Albada Jelgersma M, Vonk P, Kossen NWF. Scale-down of a high-shear
pelletisation process: flow profile and growth kinetics. Int J Pharm 1998; 166(1):89–97.
24. Stahl H. Manufacturing effervescent tablets. Pharm Technol 2003; 15(4):25–28.
25. Pearlswig DM, Robin P, Lucisano LJ. Simulation modeling applied to the development
of a single-pot process using microwave/vacuum drying. Pharm Technol 1994; 18:44–60.
26. Poska R. Integrated mixing granulating and microwave drying: a development experience.
Pharm Eng 1991; 11:9–13.
27. Stahl H. Single pot systems for drying pharmaceutical granules. Pharm Technol Eur
2000; 12(5):23–34.
28. Mandal TK. Evaluation of microwave drying for pharmaceutical granulations. Drug
Dev Ind Pharm 1995; 21:1683–1688.
29. Moss RA. Demonstration-microwave/vacuum drying of pharmaceuticals. AAPS
Annual Meeting, Washington, DC, Nov 17, 1991.
30. Van Scoik K. Microwave vacuum processing in the Vactron 300. AAPS Annual Meeting,
Washington, DC, November 1991.
31. Duschler G, Carius W, Bauer KH. Single-step granulation method with microwaves:
preliminary studies and pilot scale results. Drug Dev Ind Pharm 1995; 21:1599–1610.
32. Food and Drug Administration. Immediate release solid oral dosage forms: scale-up and
postapproval changes: chemistry, manufacturing, and controls; in vitro dissolution testing;
in vivo bioequivalence documentation; guidance. Fed Reg 1995; 60:61638–61643.
33. Berry IR, Nash eds. Pharmaceutical Process Validation. New York: Marcel Dekker,
1993.
34. Directive 94/9/EC of the European Parliament and the Council of 23 March 1994 on the
approximation of the laws of the Member States concerning equipment and protective
systems intended for use in potentially explosive atmospheres.
35. Performance Standards for Microwave and Radio Frequency Emitting Products. 21
Code of Federal Regulations, Part 1030. Washington, DC: General Services Administration,
April 1, 2003.
36. IEEE Standard for Safety Levels with Respect to Human Exposure to Radio Frequency
Electromagnetic Fields, 3 kHz to 300 GH—Supplement 1999 (IEEE C95.1–1999). New
York: Institute of Electrical and Electronics Engineers.
Single-Pot Processing 331
© 2005 by Taylor & Francis Group, LLC
11
Extrusion/Spheronization as a Granulation
Technique
Ketan A. Mehta
RO?HM Pharma, Degussa Corp., Piscataway, New Jersey, U.S.A.
Gurvinder Singh Rekhi
Elan Drug Delivery Inc., Gainesville, Georgia, U.S.A.
Dilip M. Parikh
Synthon Pharmaceuticals Inc., Research Triangle Park, North Carolina, U.S.A.
1. INTRODUCTION
Extrusion/spheronization is a multiple step process capable of making uniformly
sized spherical particles. The process is now being widely utilized in the pharmaceutical
industry. This is the revision of the original chapter written by D. Erkoboni
(1). As a pharmaceutical dosage form, pellets are defined as small, free flowing,
spherical, or semispherical units made up of fine powders or granules of bulk drugs
and excipients by variety of processes, extrusion and spheronization being one. It
is primarily used as a method to produce multiparticulates for immediate and controlled
release applications. The major advantage over other methods of producing
drug-loaded spheres or pellets is the ability to incorporate high levels of actives without
producing an excessively large particle. Mehta and Kislalioglu in their study
demonstrated the incorporation of a poorly soluble drug in a pellet matrix up to
40% loading via extrusion/spheronization for controlled drug delivery (2).
Though the process is more efficient than other techniques for producing
spheres, it is more labor and time intensive than the more common granulation techniques.
Therefore, it should be considered as a granulating technique when the
desired particle properties are essential and cannot be produced using more conventional
techniques. However, more recently, pharmaceutical scientists worldwide have
been able to use this method more easily due to advances in extrusion/spheronization
equipment engineering, thus making it simple to use. Pellets offer scientists a great
deal of development flexibility. Chemically incompatible ingredients, for instance,
can be incorporated into single capsule using two pellet types each containing
one of the incompatible ingredients. Pellets of different release characteristics can
be combined to achieve the desired release pattern of the active ingredients. Pellets
are characterized by a low surface area-to-volume ratio compared with powder or
333
© 2005 by Taylor & Francis Group, LLC
granules, which provides excellent coating substrate. Typically, pellets range in diameter
between 0.25 and 1.5 mm. Pellets are normally filled into hard gelatin capsules,
or eventually compressed into tablets which disintegrate into individual
pellets after oral intake.
Spheronization is a process invented by Nakahara in 1964. The patent
describes a ‘‘Method and Apparatus for Making Spherical Granules’’ from wet powder
mixtures (3). The equipment described in the patent was commercialized by Fuji
Denki Kogyo Co. under the trade name Marumerizer
. The process went widely
unnoticed in the pharmaceutical industry until 1970 when two articles were published
by employees of Eli Lilly and Co. Conine and Hadley described the steps
involved in the process including (a) dry blending, (b) wet granulation, (c) extrusion,
(d) spheronization, (e) drying, and (f) screening (optional) (4). Reynolds went on to
further describe the equipment and the mechanics of the process including the movement
of the particles within the spheronizer (5). Both publications cite desirable
product attributes that can be achieved, including good flow, low dusting, uniform
size distribution, low friability, high hardness, ease of coating, and reproducible
packing. Additionally, the resulting pellets offer not only technological advantages
as mentioned before but also therapeutic advantages such as less irritation of the gastrointestinal
tract and a lowered risk of side effects due to dose dumping and reproducibility
of the drug blood levels (6). From the publication of these articles through
present day the interest in extrusion/spheronization has continued to grow. The process
has recently become established in industry but was primarily driven by academia
in the interim. The increased popularity in recent years is, in part, due to a growing
understanding of the effects of process parameters and material characteristics.
In recent times, hot-melt extrusion (HME) has gained subsequent industry and
academic attention and is in the phase of further process maturity for ultimately
gaining wide spread popularity. HME is somewhat widespread in the plastics industry
for the production tubes, pipes, wires, and films. For pharmaceutical systems,
this method has been used to prepare granules, sustained release tablets, and transdermal
drug delivery systems (7). The advantage of HME is that it does not require
the use of solvents and water and few processing steps are needed making the process
somewhat simpler, efficient, and continuous. The disadvantage of HME is that it
may use complicated know-how and typically employs high temperatures around
and over 100
C as a processing requirement. However, it has been used as a method
to increase the solubility of a poorly soluble drug, to taste mask a bitter drug and
overall for controlled release dosage form purposes. The bioavailability of the drug
substance has been demonstrated to improve when it is dispersed at the molecular
level in hot-melt extruded dosage forms. Several examples of melt-extruded molecular
dispersions were presented by Breitenbach and Ma?gerlein (8).
A recent patent (9) describes melt- extrusion process as being used to produce
improved controlled release characteristics. The authors described the process which
includes mixing together a therapeutically effective agent, a water-insoluble retardant,
and a binder to form a homogeneous mixture, heating the mixture and thereafter
extruding the mixture into strands. The strands are then cooled, reduced to the
desired sizes.
The number of hot-melt extrusion patents issued for pharmaceutical systems
has increased more than sixfold annually since the early 1980s, with the United
States, Germany, and Japan leading the field. The field of melt extrusion technology
is growing very rapidly and its use to produce films for transmucosal and transdermal
drug delivery applications was presented by McGinity and Repka (10).
334 Mehta et al.
© 2005 by Taylor & Francis Group, LLC
2. APPLICATIONS
Potential applications are many including both immediate and controlled release. In
a paper by Sood et al., extrusion/spheronization method was used to develop controlled
release dosage forms for diltiazem hydrochloride (11). Two or more actives
can easily be combined in any ratio in the same unit. These combination products
can contain actives that are incompatible or have varying release profiles. Spheres
can be used as a method to limit drug migration. Physical characteristics of the active
ingredients and excipients can be modified to improve physical properties and downstream
processing. As an example, a low-density, finely divided active can be pelletized
to increase density, improve flow, and limit dusting (12). Mehta et al. showed
that this technique can be used for the development of zero order controlled release
drug delivery of poorly soluble drugs (13). Functional coatings can be applied easily
and effectively. Dense multiparticulates disperse evenly within the gastrointestinal
tract and can be used to prolong gastrointestinal transit times (14,15) or to improve
tolerance of some compounds. Regardless of the application, care must be taken to
achieve the required sphere or granule properties.
Spheres for controlled release coating applications will likely have significantly
different physical requirements than granules for compression. A product to be
coated for controlled release should have a uniform size distribution, good sphericity
and surface characteristics as well as low friability. Once coated, the sphere should
have the desired release characteristics. Additionally, if the coated spheres are to
be compressed into tablets, they will require sufficient strength to withstand the
forces of compression. Upon disintegration of the tablet, the individual spheres must
retain their original release profile. Physical properties such as flow, density, friability,
porosity, and surface area become important for granules intended for compression
into tablets. The granules should have good deformation and bonding
characteristics to form tablets having desirable physical properties. Drug release
from the final dosage form must meet the target specification.
Product produced using extrusion/spheronization can range from barely
shaped, irregular particles with physical properties similar to a conventional granule
to very spherical particles having properties that are drastically different (16). Tableting
characteristics can be modified by altering the composition of the spherical particles
(17), granulating fluid (18), or the process conditions used to produce them
(19). Compaction studies conducted on spheres similar to those used for controlled
release applications show the bonding and densification that occur during extrusion/
spheronization can alter the deformation characteristics of some materials (18).
Microcrystalline cellulose (MCC), which deforms plastically in the dry powder state,
exhibits elastic deformation followed by brittle fracture once spheronized (17). The
deformation characteristics, coupled with the larger size particles result in reduced
bonding sites and the production of weak compacts. A compaction profile of
The point is not to dwell on the properties required for each application, but
rather to reinforce the fact that each application will have very specific requirements.
One must first understand the properties required and then tailor the process to
yield the desired effects. The effects of process and formulation variables will be
discussed next.
A review of the literature shows that most investigators have tried to understand
small components of this process isolated from other effects. They have
focused on particular formulation or process parameters. It is valuable to have
Extrusion/Spheronization as a Granulation Technique 335
MCC and spheres prepared from MCC is shown in Figure 1.
© 2005 by Taylor & Francis Group, LLC
a detailed understanding of the main variables; however, this approach fails to take
into consideration the high degree of interaction that exists between the variables.
The use of statistical experimental design is a valuable tool to understand not
only the main effects but also the interactions that can have a profound effect on
the characteristics of the resulting particles (20–22). Additionally, these techniques
are extremely useful during product/process development to understand the effect
of variables and control them to produce product having desired attributes (23).
After pointing out the benefits of design methodology in this application, it should
be understood that, for simplicity, much of the discussion to follow will address the
various topics individually. In reality, however, they truly cannot be isolated from
one another. This chapter will review and discuss the general process, equipment
types, and the effect of process and formulation variables on the properties of
spherical granules.
3. GENERAL PROCESS DESCRIPTION
Extrusion/spheronization is a process requiring at least five units of operation with
an optional sixth screening step. First, the materials are dry mixed (i) to achieve a
homogeneous powder dispersion and then wet granulated (ii) to produce a suffi-
ciently plastic wet mass. The wet mass is extruded (iii) to form rod-shaped particles
of uniform diameter that are charged into a spheronizer and rounded off (iv) into
spherical particles. The spherical particles are then dried (v) to achieve the desired
moisture content and optionally screened (vi) to achieve a targeted size distribution.
cess steps along with critical variables associated with them (24). The end product
Figure 1 Compaction profiles ofmicrocrystalline cellulose powder and spheres. (FromRef. 17.)
336 Mehta et al.
The process flow diagram, shown in Figure 2, has been used to show each of the profrom
each of the steps is shown in Figure 3.
© 2005 by Taylor & Francis Group, LLC
4. EQUIPMENT DESCRIPTION AND PROCESS PARAMETERS
4.1. Dry Mixing
During the first step, powders are dry mixed to achieve a uniform dispersion prior to
wet granulation. It is generally carried out in the same mixer used for the granulation;
Figure 2 Process flow chart of the extrusion/spheronization process showing the process
variables for each individual step. (From Ref. 24.)
Figure 3 Product produced by the first four extrusion/spheronization process steps.
(A) Powder from dry mixing; (B) granules from granulation; (C) extrudate from extrusion;
and (D) spheres from spheronization.
Extrusion/Spheronization as a Granulation Technique 337
© 2005 by Taylor & Francis Group, LLC
however, if a continuous granulator is used, a separate mixer is required for the
dry mix. This step is typically taken for granted because wet massing follows.
The uniformity of the dry mix, however, can have a significant effect on the quality
of the granulation and, in turn, the spherical particles produced. An uneven distribution
of materials having wide differences in properties such as size and solubility
can result in localized over wetting, at least initially, during the granulation step.
The more soluble and finely divided components can also dissolve and become part
of the granulating fluid. The fluids, rich in soluble compounds, can either remain
as overwet regions or, with continued wet massing, can be redistributed (25).
Sphere uniformity (size and shape) is very much dependent on the uniform distribution
and composition of the granulating fluid which include not only the solvent
but also any dissolved ingredients.
4.2. Granulation
The second step is granulation, during which a wet mass having the requisite plasticity
or deformation characteristics is prepared. With a few exceptions, this step is
similar to conventional granulation techniques used to produce product for compression.
It is typically carried out in a batch type mixer/granulator; however, any
equipment capable of producing a wet mass, including the continuous type, can
be used. Batch type processors include planetary mixers, vertical or horizontal high
shear mixers, and sigma blade mixers. Examples of continuous mixers include the
Nica M6 instant mixer (26) and high shear twin screw mixer/extruders (27). The high
shear twin screw mixer/extruders have mixer/feeders that are capable of shearing
and kneading the feed materials. Dry powders and fluids are fed in through separate
ports and mixed by the action of the extruder blades and screws. The mixer/extruder
is capable of being configured to customize the amount of shear and energy used in
the process by changing the configuration of the mixing blades. This can have an
impact on the properties of the extrudate produced (28). As with the batch processors
it is critical to achieve a uniform level of fluid within the wet mass. The proper
fluid/solids ratio is accomplished by maintaining a steady powder and fluid feed into
the mixer/extruder. Both are critical; however, the powder feed is the most
problematic. Small variations in feed rates can cause significant shifts in the moisture
content of the granulation and, therefore, the quality of the spherical particles produced.
The two major differences in the granulation step, as compared to typical
granulations for compression are the amount of granulating fluid required and the
importance of achieving a uniform dispersion of the fluid. The amount of fluid needed
to achieve spheres of uniform size and sphericity is likely to be greater than that for
a similar granulation intended for tableting. Instruments such as a ram extruder (29)
and a torque rheometer (30) have been used to characterize the flow characteristics
of granulations for use in extrusion/spheronization. They are useful tools in quantifying
the rheological effect of formulation and process variations in the granulation.
The ram extruder has been used to characterize the flow of wet masses through a die,
which has been divided into stages. They are: (a) compression, where the materials
are consolidated under slight pressure, (b) steady-state flow, where the pressure
required to maintain flow is constant, and (c) forced flow, where an increase in force
is required to maintain flow. The three stages are shown in the force vs. displacement
The change from steady state to forced flow is caused by the movement of fluid
under pressure. Extrusion in a ram extruder is continuous, and this phenomenon is
338 Mehta et al.
profile in Figure 4.
© 2005 by Taylor & Francis Group, LLC
less likely to be seen in extruders that are discontinuous such as gravity-feed models
Regardless of the mixer used, one must remember that the downstream process
steps of extrusion and spheronization are very dependent on the level of water contained
in the granulation and the quality of its dispersion. High-energy mixers such
as high shear mixers and high shear twin screw mixer/extruders can cause a signifi-
cant rise in temperature. It may be necessary to use a jacket to guard against heat
build-up. High temperatures can result in a greater than tolerable level of evaporation
(33) or an increase in the solubility of some of the solids. A reduction in fluid
will reduce the plasticity of the granulation, while an increase in solubility will
increase the weight ratio of granulating fluid since the solute is then part of that fluid
(34). The water solubility of the drug in the granulation plays a key role in determining
granulation end point for extrusion/spheronization process. A highly water soluble
drug will dissolve in the granulation whereas a highly insoluble drug will have
wetting problems during the granulation step. Upon extrusion and during spheronization
step a granulation containing a highly insoluble drug at a high dose will have
a higher tendency to release moisture due to which the moisture will migrate to the
surface of the extrudates, which might cause interpellet sticking. In order to avoid
this Mehta et al. in their work demonstrated the use of small quantities of talc to
adsorb the surface moisture which helped in the spheronization step without altering
the drug release from the resulting pellets (2).
4.3. Extrusion
The third step is the extrusion step which forms the wet mass into rod-shaped particles.
The wet mass is forced through dies and shaped into small cylindrical particles
having a uniform diameter. The extrudate particles break at similar lengths under
Figure 4 A force–displacement profile for a microcrystalline cellulose–lactose–water mixture
showing the three stages of extrusion on a ram extruder: Compression, steady-state flow, and
forced flow (ram speed, 4 mm/sec; die diameter, 1.5 mm; L/R ratio, 12). (From Ref. 31.)
Extrusion/Spheronization as a Granulation Technique 339
(32). A diagram of a ram extruder is shown in Figure 5.
© 2005 by Taylor & Francis Group, LLC
their own weight. The extrudate must have enough plasticity to deform but not so
much to adhere to other particles when collected or rolled in the spheronizer.
Extruders come in many varieties but can generally be divided into three classes
based on their feed mechanism. They include those that rely on a screw, gravity or a
piston to feed the wet mass into the extrusion zone (35). Examples of extruders from
each class are shown in Figure 5. Screw feed extruders include the (a) axial or end
plate, (b) dome, and (c) radial type, while gravity-feed extruders include (d) cylinder,
(e) gear, and (f) radial types. The screw and gravity-feed types are used for development
and manufacturing with the radial varieties being the most popular for pharmaceutical
applications. The piston feed or ram extruder is primarily used in
research as an analytical tool.
Figure 5 Schematic diagrams of extruder types used in extrusion/spheronization.
340 Mehta et al.
© 2005 by Taylor & Francis Group, LLC
Screw extruders have either one (single) or two (twin) augers that transport the
wet mass from the feed area to the extrusion zone. During the transport process, the
screws compress the wet mass removing most of the entrapped air. Studies have been
conducted on the ram extruder to understand this compression or consolidation
stage. They have shown the apparent density of the wet mass plug prior to extrusion
is approximately equal to the theoretical apparent particle density, indicating that
nearly all of the voids were eliminated (31). Twin screw extruders generally have a
higher throughput than single screw models, while single screw extruders compress
and increase the density of the extrudate more. Other features that can affect the
density of the extrudate are the spacing of the turnings on the screw and the space
between the end of the screw and the beginning of the die (36). Turnings that are
wide and regularly spaced minimize the amount of compression during material
transport. Screws with closer or progressively closer spacing between the turnings
will result in more compression and produce a denser extrudate. Space between
the screw and the die results in a void into which material is deposited and compressed.
The greater the space, the more compression takes place prior to extrusion.
As material builds up, pressure increases and causes the material to be forced, under
hydraulic pressure, to flow through the die. When space between the screw and the
die is at a minimum, extrusion takes place as material is compressed in the nip,
between the extruder blade and the die.
The primary difference between the various types of screw extruders is in the
extrusion zone. An axial or dome extruder transports and extrudes the wet mass
in the same plane. Axial extruders force the wet mass through a flat, perforated
end plate, typically prepared by drilling holes in a plate. The thickness of the plate
can be more than four times the hole diameter, resulting in high die length to radius
(L/R) ratios. An axial extruder is shown in Figure 6(A).
Dome extruders use a dome or half sphere shaped screen as the die. It is prepared
by stamping holes in metal stock having a similar thickness as the hole diameter.
This results in a die L/R ratio close to 2, however, variations in screen
thickness are possible resulting in a slightly higher or lower ratio. A dome extruder
is shown in Figure 6(B).
Unlike axial and dome extruders, radial extruders extrude the wet mass perpendicular
to the plane of transport. Material is transported to the extrusion zone where
it is wiped against the screen die by an extrusion blade. The mass is forced through
the die by pressure generated at the nip. A screw feed radial extruder is shown in
Figure 6(C).
Figure 6 Types of extruders. (A) Axial; (B) dome and (C) radial. (Courtesy of the LCI
Corporation.)
Extrusion/Spheronization as a Granulation Technique 341
© 2005 by Taylor & Francis Group, LLC
As with dome type extruders the die is a stamped screen. Due to the shorter die
lengths and the increase number of holes or dies, dome and radial extruders have the
advantage of higher throughput as compared to the axial type.
As with almost every step in extrusion/spheronization, heat build-up during
extrusion is a significant concern. This is especially true of the screw fed extruders.
Axial extruders generate heat due to their long die lengths. Radial extruders can have
a significant heat differential over the width of the screen. Materials fed into the
extrusion zone will have the lowest temperature. However, as material moves to
the front of the zone, the temperature increases due to the longer residence time
of material. Of the screw feed extruders, the dome type has the highest rates and
is least likely to generate significant heat over an extended period.
Gravity-feed extruders include a cylinder, gear, and radial type. The cylinder
and gear both belong to a broader class referred to as roll extruders. Both use two
rollers to exert force on the wet mass and form an extrudate. The cylinder extruder
has rollers in the form of cylinders, one solid and one hollow with drilled holes to
form the dies. The wet mass is fed by gravity into the nip area between the two cylinders
and forced through the dies into the hollow of the cylinder. Gear type extruders
have rollers in the form of hollow gears. The dies are holes drilled at the base of each
tooth. Wet mass is forced through the holes and collected in the hollow of the gears
as the teeth and the base areas mesh. The last type of gravity-feed extruder to be discussed
is the radial type. One or more arms rotate to stir the wet mass as it is fed by
gravity. Rotating blades wipe the wet mass against the screen, creating localized
forces sufficient to extrude at the nip. There is no compression prior to extrusion
which is the major difference between the gravity and screw feed radial extruders.
The primary extrusion process variables are the feed rate, die opening, and die
length. The water content of the granulation is also very critical, since the properties
of the extrudate and resulting spheres are very dependent on the plasticity and cohesiveness
of the wet mass. The process variables and water content have been the
focus of many studies. Harrison, Newton, and Rowe studied the flow of the wet mass
as it is forced through a die (29,31,37,38). They determined that steady-state flow
which results in uniformly sized spherical particles having good sphericity and surface
characteristics. Materials and processes that did not result in steady state, a
condition referred to as forced flow, produced extrudate having surface impairments.
In moderate cases, the surface is rough, while in more severe cases, a phenomenon
commonly referred to as shark-skinning occurs. Examples of smooth extrudate
Force–displacement profiles of microcrystalline cellulose (MCC) and water at
various ratios, MCC, lactose, and water at a 5:5:6 ratio, and lactose and water at a
Steady state was possible with the MCC and MCC:lactose samples but not
with lactose alone. As can be seen with the MCC samples, the duration of the compression
stage was water level dependent with no effect seen on the steady state stage.
Additional studies indicated the effect of ram speed (extrusion speed) and die L/R
ratio. An increase in ram speed increased duration of the steady-state stage with
no effect on the compression stage. The L/R ratio had no effect on either compression
or steady state. Wet mass composition, therefore, influenced the ability
to achieve steady state while the water level and ram speed influenced duration.
Higher water levels decreased the force to produce steady-state flow but increased
342 Mehta et al.
A gravity-feed extruder is shown in Figure 7.
(described earlier and shown in Fig. 4) was essential to produce a smooth extrudate,
and shark-skinned extrudate are shown in Figure 8(A) and (B), respectively.
8:2 ratio, developed by Harrison et al., are shown in Figure 9.
© 2005 by Taylor & Francis Group, LLC
Figure 7 (A) Gravity-feed rotary extruder; (B) close-up showing the extrusion zone;
(C) operating principle of gravity-feed extruder. (Courtesy of Niro Pharma Systems.)
Figure 8 Scanning electron micrographs showing an example of (A) smooth extrudate and
(B) extrudate having surface impairment, or shark-skinning.
Extrusion/Spheronization as a Granulation Technique 343
© 2005 by Taylor & Francis Group, LLC
the duration. Faster ram speeds (extrusion rates) increased the duration of steady
state and increased the force. As discussed in the following text, other investigators
have reported the correlation between extrusion force and sphere quality.
Fechner et al. indicated that for an optimum extrusion process, a mixture of
spongelike and gel-like behavior might be desirable which are possible by the use
of MCC and powder cellulose (39).
Harrison et al. also indicated that a uniform lubricating layer at the die wall
interface must occur to eliminate the slip-stick phenomenon responsible for forced
flow. Development of a lubricating layer was dependent on the length of the die
(a minimum length required), wall shear stress and upstream pressure loss. They
represent the frictional forces at the die wall interface and the estimated pressure loss
at zero die length in the barrel of the ram extruder. The method for deriving these
values is described in Ref. 27. These parameters allow for a quantitative comparison
between formulations and process; however, no specific values can be targeted since
they vary with materials.
Pinto et al. also showed that, at slow ram speeds, water moves toward the die
wall interface and acts as a lubricant resulting in reduced extrusion forces. At higher
speeds water is unable to move rapidly through the mass resulting in higher forces
(40). They indicated that the water content and its distribution are critical in determining
the particle size and sphericity of the product. Lower water content and
higher speed will reduce the size and sphericity of the particles. The extrusion speed
and water content should be adjusted to achieve the desired effect. Other researchers
have investigated the effect of die length using gravity-feed radial extruder. Helle?n
et al. indicated the extrudate became smoother and more bound as the L/R ratio
Figure 9 Force–displacement profiles at various moisture contents of mixtures of microcrystalline
cellulose and water: (a–d) microcrystalline cellulose–lactose–water (5:5:6);
(e) lactose–water (8:2); (f) at a ram speed of 4 mm/sec, die diameter of 1.0, and a L/R ratio
of 12. Percentage of moisture content of microcrystalline cellulose–water mixture: a, 59.4;
b, 51.1; d, 45.0. (From Ref. 29.)
344 Mehta et al.
© 2005 by Taylor & Francis Group, LLC
of the die was increased (41). Vervaet et al. reported that a higher L/R ratio enables
the use of lower water levels to achieve a more bound extrudate (42). This also
increased the range (drug loading and water level) over which quality spheres could
be produced. They attributed the increased latitude and capability to increased densification
and resulting well-bound extrudate. The average pore diameter and bulk
density reported for extrudate prepared from various MCC:DCP:water ratios at
two L/R ratios are shown in Table 1.
Baert et al. also indicated a similar increase in latitude when a cylinder extruder
having an L/R ratio of 4 was compared to a twin screw extruder having a L/R ratio
of close to 1.8 (43). Other studies have shown that there is an optimal pressure range
over which extrudate capable of yielding acceptable spheres can be produced. Shah
et al. demonstrated the correlation between screen pressure yield and density (44).
A high yield of spheres within a targeted narrow size distribution was produced as
long as the screen pressure was maintained within a given range. The relationship
While many of the researchers have indicated a need for a more cohesive extrudate,
few have expressed a need to remove all surface impairments. Some researchers
have indicated that spheres having acceptable characteristic can be produced from
extrudate having shark-skinning. O’Connor and Schwartz have found the presence
of shark-skinning to be advantageous in facilitating the breakage of the extrudate
during the spheronization step (45).
Experimental design studies conducted to concurrently investigate the effect of
extrusion as well as other process and formulation variables have indicated the extrusion
variables to be less significant than granulating fluid level or variables of the
spheronization step. Hasznos et al. determined that extruder speed had little effect
on the size distribution of the final product or moisture change during processing
as compared to the spheronization variables (46). Hilemann et al. indicated that,
when water/MCC ratios are held constant, a change in screen size results in a significant
change in the size distribution (47). However, in a study where water level was
included as a variable, Erkoboni et al. have shown that the effect of screen size on
size distribution is small compared to the effect of a change in water level. A change
in water level can shift the mean size and still result in an acceptable distribution (21).
This is in agreement with earlier work by Malinowski and Smith who also showed
the mean particle size is typically smaller than the size of the screen itself due to
shrinking during the drying step (12). Vervaet et al. have presented an excellent
review of extrusion spheronization (48).
Table 1 Average Pore Diameter and Bulk Density of Extrudate Composed of DCP–Avicel
PH-101–Water Mixture, Extrudated Using Screens with Different L/R Ratios
Composition DCP–
Avicel–water (w/w) L/R ratio of screen
Average pore
diameter (mm) Bulk density (g/mL)
150:380:470 4 0.982 1.132
150:400:450 4 0.992 1.211
150:380:470 2 1.249 0.949
150:400:4502 2 1.292 0.947
Source: From Ref. 42.
Extrusion/Spheronization as a Granulation Technique 345
between yield and screen pressure is shown in Figure 10.
© 2005 by Taylor & Francis Group, LLC
4.4. Spheronization
The fourth step in the extrusion/spheronization process is the spheronization step. It
is carried out in a relatively simple piece of equipment. The working parts consist of
a bowl having fixed sidewalls with a rapidly rotating bottom plate or disk. The
rounding of the extrudate into spheres is dependent on frictional forces. The forces
are generated by particle-to-particle and particle-to-equipment interactions. For this
reason the disk is generally machined to have a grooved surface which increases the
forces generated as particles move across its surface. Disks having two geometric
patterns are produced, a cross-hatched pattern with the grooves running at right
angles to one another and a radial pattern with the grooves running radially from
the center. The two varieties are shown graphically in Figure 11.
Some studies have shown the rate of spheronization to be faster with the radial
pattern; however, both plates will result in acceptable product (35).
During the spheronization step, the extrudate is transformed from rod-shaped
pellets into spherical particles. This transition occurs in various stages. Once charged
Figure 10 The effect of extruder screen pressure on the yield of particles within an acceptable
distribution. (From Ref. 44.)
Figure 11 Spheronizer disks having two geometric patterns: (A) a cross-hatched pattern
with the groves running at right angle to one another and (B) a radial pattern with the groves
running radially from the center.
346 Mehta et al.
© 2005 by Taylor & Francis Group, LLC
into the spheronizer, the extrudate is drawn to the walls of the extruder due to centrifugal
forces. From here what happens is very much dependent on the properties of
the extrudate. Under ideal conditions, the extrudate breaks into smaller, more uniform
pieces. Within a short period of time, the length of each piece is approximately
equal to the diameter, due to attrition and rapid movement of the bottom plate or
disk. The differential in particle velocity as they move outward to the walls, begin
to climb the walls and fall back onto the rotating bed, along with the angular motion
of the disk results in a rope-like formation (5). Figure 12 shows the rope-like formation
of the extrudates.
This formation can be a critical indicator of the quality of the granulation or
extrudate. As pointed out by Reynolds (5), the disk rotating without movement of
the product indicates an over wet condition. The condition is caused either from
a granulation that was initially over wet or migration of water or a fluid ingredient
to the surface of the extrudate during extrusion or spheronization.
As mentioned, the transformation from cylinder-shaped extrudate to a sphere
occurs in various stages. Two models have been proposed to describe the mechanism
The model proposed by Rowe in 1985 describes a transition whereby the
cylindrical particles (Fig. 13-2A) are first rounded off into cylindrical particles with
rounded edges (Fig. 13-2B), then form dumbbell-shaped particles (Fig. 13-2C), ellipsoids
(Fig. 13-2D), and finally spheres (Fig. 13-2E) (35). The second model proposed
by Baert et al. in 1993 suggests that the initial cylindrical particles (Fig. 13-1A) are
deformed into a bent rope-shaped particle (Fig. 13-1B), then form a dumbbell with a
twisted middle (Fig. 13-1C). The twisting action eventually causes the dumbbell to
break into two spherical particles with a flat side having a hollow cavity (Fig. 13-
1D). Continued action in the spheronizer causes the particles to round off into
spheres (Fig. 13-1E). When the sphere is fractured a hollow particle is revealed
(49). The exact mechanism is likely to be composition dependent. If the extrudate
is overwet, particle growth will occur resulting in a broad size distribution. Under
wet extrudate will not have enough plasticity to further round off in the spheronizer;
and dumbbells that would not deform further.
Figure 12 A characteristic rope-like formation in a spheronizer bowl during operation.
Extrusion/Spheronization as a Granulation Technique 347
and are shown graphically in Figure 13.
the result is the formation of dumbbells. The scanning electron micrographs in Figure
14 show an example of good spheres produced from a sufficiently plastic mass
© 2005 by Taylor & Francis Group, LLC
Of the two process steps unique to extrusion/spheronization, the first, extrusion,
is a continuous process while the second, spheronization, is a batch process.
To make the process viable for commercial operations, two systems have been developed
to enable the extruder to continuously feed material to the spheronizer(s). The
Figure 14 An example of (A) good spheres produced from a sufficiently plastic mass and
(B) dumbbells that would not deform further produced from underwet extrudate.
Figure 13 A graphic representation of the two models proposed to describe the mechanism
of spheronization. The model proposed by Baert et al. (38) describes a transition from initial
cylindrical particles (1A) into a bent rope (1B), dumbbell (1C), two spherical particles with a
hollow cavity (1D), and spheres (1E).The model proposed by Rowe (25) describes a transition
from cylindrical particles (2A) into cylindrical particles with rounded edges (2B), dumbbells
(2C), ellipsoids (2D), and spheres (2E). (From Refs. 35,49.)
348 Mehta et al.
© 2005 by Taylor & Francis Group, LLC
first system is a semicontinuous shuttle system and the second is a cascade system.
The shuttle system is typically used when uniform particles are required, such as
for controlled release coating applications. The cascade system, however, can be used
for applications where less size and shape uniformity is required, such as granulations
intended for compression.
The shuttle system uses two spheronizers in parallel. It is designed to fill one
spheronizer while the second is in the middle of its cycle, continue to collect extrudate
in a shuttle receptacle while they are both full and operational, and fill the second
after it empties and the first unit is in the middle of its cycle. The shuttle system
operation is shown graphically in Figure 15.
A picture of a spheronizing system having twin spheronizers is also shown in
The cascade operation uses one or more spheronizers that are modified to have
the disks some distance below the discharge chute (36). This results in a spheronization
zone having a fixed volume. The product is continually fed from either the
extruder or a previous spheronizer. As the charge volume grows from incoming
material some product is discharged. The residence time is dictated by the feed rate.
The reduced size and shape distributions are due to the percentage of material that
does not reside in the spheronization zone for the intended period of time. The number
of spheronizers placed in sequence depends on the desired outcome. However, if
only a slight rounding with minimal densification is required, one spheronizer with
a short residence time will be sufficient.
A commercial manufacturing of pellets using the extrusion/spheronization
process can be accomplished by discharging the formed pellets from the spheronizer
in a continuous fluid bed unit. This provides the semiautomatic commercial setup.
Variables in the spheronization step include spheronizer size, charge, disk
speed, and residence time. A number of studies have shown each of the variables
Figure 15 A graphic representation of twin spheronizer shuttle system using two spheronizers
in parallel and shuttle receptacle. (A) When both units are full the shuttle receptacle collects the
extrudate. (B) After one empties, (C) the shuttle box fills it. (D–F) The cycle repeats itself for
the second unit. (From Ref. 36.)
Extrusion/Spheronization as a Granulation Technique 349
Figure 16.
Figure 17 illustrates a typical setup in such an equipment configuration.
© 2005 by Taylor & Francis Group, LLC
has the potential to play a significant role in influencing the physical characteristics
of the resulting product. Hasznos et al. (46) showed that a higher disk speed and
longer residence time increased the coarse fraction and mean diameter and decreased
the fine fraction. The faster speed and longer time also increased the moisture loss
during the process. Since the moisture loss can reduce the plasticity of the particle,
it can have the same effect as an underwet granulation. The particles may not round
off into spheres and stay as deformed cylinders or dumbbells. Higher spheronizer
Figure 16 Twin spheronizer with extruder. (Courtesy of Niro Pharma Systems.)
Figure 17 Typical semicontinuous pellet production setup. (Courtesy of LCI Corporation.)
350 Mehta et al.
© 2005 by Taylor & Francis Group, LLC
charges reduced the moisture loss. They also suggested that an interaction between
spheronizer speed and residence time indicated the total number of revolutions of the
disk was critical. A change in one of the variables could be offset by an opposite
change of the other, as long as the total number of revolutions remained constant
(46). Helle?n et al. showed a similar moisture loss during spheronization. In addition,
they indicated that the major factors influencing the shape of the spheres were the
disk speed and residence time. High speed and long time produced more spherical
particles (26). Wan et al. indicated that a minimum disk speed and residence time
was required to round the cylinder-shaped extrudate. Furthermore, an increase in
speed or time, up to a limit, increased the median diameter of the spheres while
higher speeds and longer times caused a reduction in size. Short residence times at
high disk speeds resulted in small but round particles (50).
A number of investigators have reported the effect of disk speed and residence
time on density. Woodruff and Neussle reported the variables to have no effect on
the density of the spheres as compared to the density of the granulation and extrudate
(16). These results are in conflict with most of the other studies; however, they
are likely due to the use of mineral oil in the formulation. The oil can reduce the frictional
forces at the die wall during extrusion and between particles and equipment
surfaces during spheronization. A number of investigators including Malinowski
and Smith reported that an increase in either disk speed or residence time resulted
in an increase in density (20,21,26,47). Mehta et al. studied the effect of spheronization
time on the pellet hardness and drug release (2). It was concluded that pellet
hardness changed with spheronization time until about 10 min after which the hardness
decreased until the 20 min point. No significant difference in pellet hardness
from 20 to 40 min thereafter was observed. Mehta et al. explained this by the densi-
fication process occurring during the spheronization step. As spheronization time
progresses from time zero to a certain time point ‘‘t’’ the extrudates are cut into uniform
particles and shaped into spheres by the centrifugal and frictional forces present
in the spheronizer during the operation. These forces act on each and every
particle, making them denser and more spherical with time. However, after a critical
period, no further densification occurs with an increase in spheronization time. In
another study, Mehta et al. showed the effect of spheronization time on the porosity
parameters of the pellets (51). It was summarized that a processing period of
2–10 min increased the number of pores and the total pore surface area and
decreased the pore diameter. Beyond this time, for up to 20 min of spheronization
time, the porosity was unchanged.
O’Connor et al. indicated that the friability of placebo spheres decreased with
increasing residence time while the mean particle diameter decreased (24). Erkoboni
et al. showed an increase in extruder screen size resulted in reduced friability (21).
4.5. Drying
Drying is the final step in the process. This can be accomplished in any dryer that can
be used for conventional type granulations, including tray dryers, column type fluid
beds, and deck type vibratory fluid beds. Each of the drying techniques has advantages;
however, the major differences are based on the rate of water removal. Tray
drying is the slowest of the processes. Fluidized bed dryers result in a much more
rapid drying rate because of the higher air volumes and the potential use of higher
inlet temperatures. Column fluid beds are batch dryers, while the deck type dryers
offer the advantage of a continuous process. Both have been used successfully in
Extrusion/Spheronization as a Granulation Technique 351
© 2005 by Taylor & Francis Group, LLC
drying product produced by extrusion spheronization. The drying process must be
chosen based on the desired particle properties. For example, pellets to be dried in
fluid bed equipment will have to withstand fluidization process, resist attrition,
and maintain its integrity.
Tray drying is a slow process in a static bed. Because of this, it can offer the
greatest opportunity for a drug to migrate toward the surface and recrystallize
(52). The more rapid rate in a fluid bed will likely minimize the effects of migration.
This phenomenon can have an effect on a number of particle properties. The increased
active concentration at the surface of the particle can increase the rate of dissolution.
This recrystallization, however, can cause a problem for applications requiring film
coating since the smooth surfaces developed by the spheronization process would be
damaged. Additionally, the crushing strength of tray dried particles will likely be
greater than their fluid bed counterparts. The slow recrystallization in the static
bed allows for crystal bridges to develop as the fluid is removed and the solute
recrystallizes.
5. FORMULATION VARIABLES
The composition of the wet mass is critical in determining the properties of the
particles produced. This is clearly understood if we look at what material behaviors
are required during each of the process steps. During the granulation step, a plastic
mass is produced—a simple enough task if ended there. The materials must form
a plastic mass, deform when extruded and break off to form uniformly sized cylindrical
particles. A minimal amount of granulating fluid should migrate to the surface
during extrusion and the particles should stay discrete during collection. During
spheronization the particles must round off to form uniformly sized spheres. They
must not dry out due to temperature or air volume or grow in size due to agglomeration.
The fact is that a lot is expected from materials used in this process. This
is especially true of formulations containing high percentages of active where low
levels of excipients are used to impart the desired properties to the mass.
The importance of using sphere-forming excipients was noted early on. Conine
and Hadley cited the necessity of using microcrystalline cellulose (4). Reynolds went
on to indicate the need for either adhesive or capillary type binders (5). He cited cellulose
gums, natural gums, and synthetic polymers as adhesives and microcrystalline
cellulose, talc, and kaolin as capillary type binders. Since then much work has been
conducted in an attempt to understand the significance of material properties. Some
of the studies are discussed in the following text.
O’Connor et al. studied the behavior of some common excipients in extrusion/
spheronization. The materials were studied as single components using water as the
granulating fluid in an attempt to understand their application in the process. Of the
materials tested, only MCC or MCC with Na-CMC (Na-caboxymethyl cellulose)
was capable of being processed. Others including dicalcium phosphate, lactose,
starch, and modified starch did not process adequately (24).
In an additional study, they investigated the effect of varying drug, excipient,
and excipient:drug ratios. At low drug levels they found the spheronizing excipient
played the most significant role in determining sphere properties. They found that,
for low dose applications, MCC was the best excipient to use since it formed the
most spherical particles. At moderate drug loading (50%), MCC as well as the
two products consisting of MCC coprocessed with Na-CMC (Avicel
 RC-581 and
352 Mehta et al.
© 2005 by Taylor & Francis Group, LLC
Avicel CL-611) resulted in acceptable spheres. At higher loading levels, however, the
MCC did not yield acceptable spheres and the coprocessed materials did. The
spheres produced using Avicel CL-611 were the most spherical. In addition, they
found dissolution to be dependent on the type of excipient used, the solubility,
and concentration of the active. Spheres containing MCC remained intact and
behaved as inert matrix systems, while those containing the coprocessed products
formed a gel plug in the dissolution basket and were described as water-swellable
hydrogel matrix systems. The release profiles for spheres containing each of the excipients
and theophylline in a 50:50 ratio are shown in Figure 18.
Release profiles for spheres containing different drug loads are shown in
An increase in drug load resulted in an increased release rate. Release profiles
for spheres containing actives having different solubilities, including chlorpheniramine
maleate, quinidine sulfate, theophylline, and hydrochlorothiazide are shown
Mehta and Kislalioglu demonstrated the use of polymethacrylate type polymers
such as Eudragit L 100–55 and Eudragit S 100 via extrusion/spheronization
in the development of controlled release pellets (13,54). They theorized that for
the development of zero-order controlled release pellets of a poorly soluble drug,
MCC would not be a good choice to form a pellet system via extrusion/spheronization.
This would be due to the fact that MCC being insoluble would form a nondisintegrating
matrix from which it would be difficult for an insoluble drug to be
released. In their work they showed that Eudragit L100–55 and Eudragit S 100
can be used as pellet forming and release rate governing polymers for developing
a controlled release drug delivery system with out the use of MCC in the matrix.
Zhou and Vervaet produced matrix pellets by combining microcrystalline
waxes, pregelatinized starches, and hydrolyzed starches with model drugs such as
Ibuprofen, chloroquin phosphate, and others (55). They concluded that the combination
of microcrystalline waxes and pregelatinized starches or maltodextrins is a
flexible system for the production of matrix pellets, even with a high drug concentration.
Additionally, they concluded that the drug release with such a system could be
Figure 18 Dissolution profiles of spheres containing 50% theophylline in different Avicel
MCC types  Avicel PH-101;
G
, Avicel RC-581; &, Avicel CL-611. (From Ref. 53.)
Extrusion/Spheronization as a Granulation Technique 353
Figure 19.
in Figure 20. An increase in drug solubility resulted in an increased release rate (53).
© 2005 by Taylor & Francis Group, LLC
modeled by varying the type and the concentration of the wax and the starch. Tapia
et al. described factors influencing the mechanism of release from sustained release
matrix pellets, produced by the extrusion/spheronization process (56).
Kleinebudde and Jumaa concluded that during the extrusion process, water
content in the extrudate and pellet porosity were increased as the degree of polymerization
of MCC and powder cellulose in the matrix was increased (57).
Millili and Schwartz demonstrated the effect of granulating with water and
ethanol at various ratios. The physical properties of the spheres changed significantly
as the ratio of the two fluids was varied. Spheres could not be formed with absolute
ethanol but were possible with 5:95 water:ethanol. An increase in the water fraction
resulted in a decrease in porosity, friability, dissolution, and compressibility and an
increase in density. The porosity of spheres granulated with 95% ethanol was 54%
Figure 19 Dissolution profiles of spheres containing different concentrations of drug in
Avicel CL-611: 
, 10%; &, 59%;
`, 80%. (From Ref. 53.)
Figure 20 Dissolution profiles of spheres containing 10% drug in Avicel PH 101: , chlorphenarimine
maleate; 
, quinidine sulfate; , theophyline; & hydrochlorothiazide. (From Ref. 53.)
354 Mehta et al.
© 2005 by Taylor & Francis Group, LLC
while the water granulated product had a porosity of 14%. When greater than 30%
water was used, spheres remained intact throughout the dissolution test. As previously
discussed, water granulated spheres were very difficult to compress while
spheres granulated with 95% ethanol were significantly more compressible than those
prepared using water (18). In contrast, Mehta et al. showed that an increase in granulation
water level increased the total number of pores in the pellet matrix without
changing the pore diameters (51). Additionally they concluded that this direct
increase in porosity increased the dissolution contact angle due to which dissolution
of the poorly soluble drug was increased. Jerwanska et al. concluded that the rate of
drug release increased with increased levels of granulation liquid because of a greater
degree of porosity obtained after drying (58). They also correlated these results with
differences in hardness of the pellets. This was similar to the findings by Mehta et al.
Jerwanska et al. proposed that for a continuous extrusion process, adequate water is
required to bridge the particles together until liquid saturation in the granulation is
achieved. This strategy is necessary to deform the granulation to form extrudates
and consequently shape them into spheres by spheronization. If the granulation water
level is below the liquid saturation point, then the spheres obtained will be hard and
less porous, thereby leading to decreased drug release rates. Above the liquid saturation
point, the hardness and porosity of the pellets are not significantly decreased.
A tablet hardness vs. compression forces profile is shown in Figure 21.
In a later study, Millili et al. proposed a bonding mechanism, referred to as
autohesion, to explain the differences in the properties of spheres granulated with
water and ethanol. Autohesion is a term used to describe the strong bonds formed
by the interdiffusion of free polymer chain ends across particle–particle interfaces
(59).
Using a ram extruder, Harrison et al. demonstrated that steady state flow could
not be achieved with lactose. Additionally, they demonstrated the reduced sensitivity
of MCC to small changes in moisture as determined by the force required to induce
plug flow in a cylinder. Comparing MCC to a MCC/lactose blend and 100% lactose,
Figure 21 The effect of varying compression force on the hardness of compacted 16/30-
mesh spheres of 10% theophylline–Avicel PH101:
`
, spheres prepared by water; &, spheres
prepared by 95% ethanol granulation. (From Ref. 18.)
Extrusion/Spheronization as a Granulation Technique 355
© 2005 by Taylor & Francis Group, LLC
they found that, with lactose, small changes in moisture caused large changes in force
while with MCC, larger changes in moisture were required to have similar effects on
the force (29).
Baert et al. used mixtures of microcrystalline cellulose and coexcipients at various
ratios to demonstrate the effect of solubility and the total fluid on extrusion
forces. They showed that if the coexcipient was insoluble, such as dicalcium phosphate,
the force required to extrude increased with increasing levels of coexcipient.
When a soluble excipient such as lactose was used, the force required to extrude
decreased with the addition of the initial amounts of lactose. After a certain level,
however, the reduction in force stopped and began to increase. This was due to
the initial solubilization of lactose and the resulting increase in the total fluid level.
Once the fluid was saturated the remaining lactose was not soluble and the force
began to increase. The increase began at about 10% lactose level for a-lactose and
20% for b-lactose. This was due to the difference in solubility between the two materials
(33). The effects of dicalcium phosphate and various lactose grades on extrusion
force are shown in Figure 22.
Funck et al. showed that low levels of common binders could be used to produce
high drug loaded spheres with microcrystalline cellulose. Materials such as carbomer,
Na-CMC, hydroxypropylcellulose (HPC), hydroxypropylmethylcellulose
(HPMC), povidone (PVP), and pregelatinized starch were used. All materials were
capable of producing spheres of acceptable quality. Dissolution testing showed
spheres containing HPC and HPMC remained intact during testing while spheres
containing starch, PVP, and Na-CMC disintegrated (60).
Lender and Kleinebudde reported that spheres produced with powdered cellulose
had higher porosity and faster dissolution than those made using microcrystalline
cellulose. Spheres could not be produced using only powdered cellulose and
drug; a binder was required. The higher porosity of the spheres prepared from powdered
cellulose may be beneficial for applications requiring compression (61).
Figure 22 Influence of the amount of lactose or dicalcium phosphate dehydrate (% total
weight) on the extrusion forces (N) for mixtures of lactose or dicalcium phosphate dehydrate–
Avicel PH 101–water after granulation with a planetary mixer. Each end point is the
mean of six values. The SD is lower than 3% for each point. Six different types of lactose were
used: &, a-lactose monohydrate 80 mesh; 
, a-lactose monohydrate 200 mesh; , a-lactose
monohydrate 325 mesh; &, spray dried lactose DCL 11; , anhydrous b-lactose DCL 21; ,
anhydrous a-lactose DCL 30. One type of dicalcium phosphate dehydrate was used,
G
. (From
Ref. 33.)
356 Mehta et al.
© 2005 by Taylor & Francis Group, LLC
Feilden et al. showed that increasing the particle size of lactose resulted in forced
flow and high extrusion forces, which resulted in poor quality extrudate and spheres having
awide size distribution. Thiswas attributed to the increased pore diameter of themixture
containing the coarse lactose which allowed greater movement of water (62).
Chien and Nuessle (63) showed the use of a surfactant, such as sodium lauryl
sulfate, reduced the migration of drug to the surface of the sphere during drying by
reducing the surface tension of the granulating fluid. The reduction in surface tension
also made it difficult to produce a cohesive extrudate in some cases.
Some miscellaneous observations include the following. Reynolds reported
that excess extrudate friability can be overcome by incorporating more MCC, binder,
or water in the granulation (5). Erkoboni et al. indicated that sphere hardness
was most affected by the level of MCC in the formulation and the level of granulating
fluid used (21). Hileman et al. showed that MCC had a narrower water range
over which quality spheres could be made than MCC coprocessed Na-CMC (47).
Helle?n et al. showed that the surface characteristics were influenced by the water
level with higher water levels giving smoother surfaces (26). Mehta et al. showed that
when concentrations of pellet forming and release rate governing polymers in the
matrix were changed, it altered the dissolution kinetics of a poorly soluble drug (2).
6. COMPRESSION OF PELLETS
Typically, pellets are produced for administering in a capsule dosage form after manufacturing
with desired modified release properties. These pellets can be compressed
into tablets as well. The tablets are normally intended to disintegrate into discrete
pellets in the gastrointestinal tract and the drug should be subsequently be released
in a controlled manner from the individual pellets. One challenge in the production
of such tablets is maintaining the desired drug release after compaction, as the application
of a compaction pressure can lead to structural changes in the film coating
and consequently, altered drug release. Numerous investigations have been made
into the compaction of pellets, coated or uncoated. It is imperative that the coated
pellets do not lose their release properties during the compression stage. Pellets have
been shown to react differently to compaction and consolidation than powders of the
same material. Wang et al. reported compression of various lactose/microcrystalline
cellulose compositions in powder or pellet form (64). Schwartz et al. (17) demonstrated
the compaction characteristics of MCC processed into spheres are signifi-
cantly different than the original powder. The powder material forms hard
compacts at low compression forces, while the spheres are not compressible and
form soft compacts, even at high forces. They indicated that spheres prepared from
MCC showed a high degree of viscoelasticity over the entire compression range.
Inclusion of coexcipients such as lactose and dicalcium phosphate increase the compactability
by decreasing the viscoelastic resistance or pressure range over which the
spheres behave elastically. A reduction in viscoelastic resistance was seen with
spheres containing both lactose and dicalcium phosphate; however, dicalcium phosphate
had a greater effect. Compaction profiles of spheres containing 10% theophylline
with MCC, MCC/DCP, or MCC/lactose in a 22.5/67.5 ratio are shown in
A similar phenomenon was reported by Maganti and Celik when pellets produced
by rotor granulation were compressed (65). They compared the compaction
behavior of pellet formulations, mainly consisting of MCC, to that of the powders
Extrusion/Spheronization as a Granulation Technique 357
Figure 23.
© 2005 by Taylor & Francis Group, LLC
from which they were formed and also found significant differences. The powders
examined were found to compact by plastic deformation and produced strong compacts,
while the pellets exhibited elastic deformation and brittle fragmentation,
resulting in compacts of lower tensile strength. This can be explained by the fact that
the pellets, which are large and spherical in shape as compared to the small, irregular
powder particles they are composed of, have a low surface-to-volume ratio, which
might result in a decreased area of contact between the particles as they consolidate.
Nicklasson (66) investigated the compression behavior of pellets consisting of MCC,
with or without other excipients such as polyethylene glycol and DCP. Deformation
of the aggregates was found to depend on three deformation characteristics, namely,
the capacity for, the mode of and the resistance to deformation. High surface deformation
refers to the great ability of the pellets to conform to the surface of the surrounding
pellets. In pellets containing the soft component, the primary particles can
reposition within the agglomerate and the ability to fill the intragranular pore space
is increased. For pellets containing hard materials, the compaction stress may give
local failure at pellet surfaces. Thus, the material properties of the primary particles
constituting the pellets are important for the compression behavior of pellets. In
number of studies (67,68) various soft materials have been incorporated in pellets
to modify their deformability and compatibility. Nicklasson and Alderborn (69) studied
the modulation of the tableting behavior of pellets through the incorporation of
polyethylene glycol and found that these soft pellets had an increased propensity to
deform and altered mode of deformation to the relatively hardMCC pellets. Iloan?usi
et al. (67) found MCC-based bead formulations incorporating wax to be more compressible
than those made without wax. Salako et al. (68) found that pellets containing
theophylline and MCC were hard and less brittle than the ones containing
glyceryl monostearate which were soft pellets. The soft pellets were found to fracture
under low compression pressures and were able to form a coherent network of
deformable material in the tablets at higher pressures. The hard pellets were unable
Figure 23 The effect of excipients on the compaction profile of spheres. Compaction profiles
of spheres containing 10% theophylline with either MCC, MCC–DCP, or MCC–lactose in
a 22.5:67.5 ratio using the Leunberger model. (From Ref. 17.)
358 Mehta et al.
© 2005 by Taylor & Francis Group, LLC
to form such a network at high pressures and found to reduce more in volume
without bond formation than soft pellets.
The size of the pellets can also have a bearing on their compression behavior.
Small pellets have been shown to be less affected than larger ones by the compaction
process (70,71). Smaller beads were significantly stronger, relative to their size, than
larger ones. Researchers also found that larger pellets were much more readily
deformed (71). The application of coating to the pellet core can influence their compression
characteristics. Magnati et al. (72) added a water-based ethyl cellulose coating
(Surelease) to the MCC-based pellets previously used (65) and thereby altered
their deformation characteristics, introducing plastoelastic properties whereas previously
they had been brittle and elastic. The overall ability of the pellets to deform,
both plastically and elastically, increased with an increasing coating level.Miller et al.
(73) investigated the mechanical properties of tablets compressed from pellets coated
with Surelease to that of uncoated pellets and found them to be comparable with the
exception of the diametrical strain, which increased on coating. This was attributed
to the flexibility of the plasticized ethyl cellulose, allowing greater deformation of the
compact to occur before failure.
It has been found that coated pellets can be compressed into tablets while
retaining controlled release of the drug, provided that the effect of excipients and
compression force is considered and determined. The protective effect of an excipient
is dependent on the particle size and the compaction characteristics of the material.
In general, materials that deform plastically, such as MCC and PEG, give the best
protective effect (74–77). Yuasa et al. studied the protective effect of 14 different
excipients and were able to correlate the plastic energy percentage to the release rate,
that is, materials that deform plastically were shown to protect the coating best (78).
However, Stubberud et al. found that lactose, a fragmented material, gives better
protection than MCC (79). The compressed induced changes in the structure of a
film coating may depend on formulation factors, such as the type and the thickness
of the coating, the properties and the structure of the substrate pellets, and the
incorporation of excipient particles.
The optimum amount of excipients incorporated in the tablet formulation was
concluded to be 30% (80). Palmieri et al. (81) showed that tablets consisting of maximum
40% coated granules had acceptable release profiles, using MCC as the tablet
excipient. Wagner et al. (82) concluded that a tablet content of 70% (w/w) of pellets,
approximately 1mm diameter, is a critical level resulting in pronounced damage to
the coating. On reducing the pellet proportion to 60% (w/w), tablets fulfilling the
compendial dissolution requirements for enteric-coated products can be prepared.
Tuno?n (83) investigated factors that influence the preparation of modified release
pellets, its compression behavior, influence of inter- and intragranular factors and
release of drug release of pellets. The most frequently encountered explanation for
the loss of modified release upon compaction, in terms of an increased drug release
rate, is the occurrence of cracks in the coating. However, knowledge of why and how
cracks are formed in the coating and techniques to avoid them is highly valuable in
the development of multiple unit tablets containing pellets.
7. CONCLUSIONS
Extrusion/spheronization is a versatile process capable of producing granules or
spheres having unique physical properties. Since it may be more labor and time
Extrusion/Spheronization as a Granulation Technique 359
© 2005 by Taylor & Francis Group, LLC
intensive than the more common granulation techniques, it should be considered
as a granulating technique when the desired properties cannot be produced using
more conventional techniques. Potential applications are many including both
immediate and controlled release. Regardless of the application, care must be taken
to understand the desired properties and the formulation and process variables capable
of achieving them. The use of statistical experimental design for formulation
and process development is strongly recommended due to the high degree of interactions
between the variables. Lastly, new technologies such as hot melt extrusion
and spheronization (HME) are gaining considerable interest in the pharmaceutical
drug delivery arena for solving specific problems such as enhancing taste masking,
improving solubility and drug bioavailability and in general for controlled release
drug delivery. Compression of pellets into modified release multiple unit dosage form
is now possible once the proper understanding of the formulation of the core pellets,
type of coating, and protective excipients to maintain the coating integrity of the
pellets is understood.
REFERENCES
1. Erkoboni D. Extrusion–spheronization for granulation. Parikh DM, ed. Handbook of
Pharmaceutical Granulation Technology. New York: Marcel Dekker, 1997.
2. Mehta KA, Kislalioglu MS, Phuapradit W, Malick AW, Shah NH. Effect of formulation
and process variables on matrix erosion and drug release from a multiunit erosion matrix
of a poorly soluble drug. Pharm Technol 2002; February:26–34.
3. Nakahara, US Patent 3,277,520, October 1966.
4. Conine JW, Hadley HR. Preparation of small solid pharmaceutical spheres. Drug Cosmet
Ind 1970; 106:38–41.
5. Reynolds AD. A new technique for the production of spherical particles. Manuf Chem
Aerosol News 1970; 41:40–43.
6. Mehta KA, Kislalioglu MS, Phuapradit W, Malick AW, Ke J, Shah NH. In vivo release
performance of nifedipine in dogs from a novel Eudragit based multi-unit erosion matrix.
Drug Deliv Technol 2002; 2(2):38–42.
7. Young CR, Koleng JJ, McGinity JW. Production of spherical pellets by a hot-melt extrusion
and spheronization process. Int J Pharm 2002; 242:87–92.
8. Breitenbach J, Ma?gerlein M. Melt extruded molecular dispersions. Ghebre-Sellassie I, ed.
Pharmaceutical Extrusion Technology. New York: Marcel Dekker, 2003.
9. Oshlack B, et al. US Patent 6,706,281, issued March 16, 2004.
10. McGinity J, Repka M. Hot melt extruded films. Drug Deliv Technol 2004; 4(7):40–47.
11. Sood A, Ashokraj Y, Panchagnula R. Use of extrusion spenronization to develop controlled
release dosage forms for diltiazem hydrochloride. Pharm Tech 2004; April:62–85.
12. Jalal IM, Malinowski HJ, Smith WE. Tablet granulations composed of spherical-shaped
particles. J Pharm Sci 1972; 61:1466–1468.
13. Mehta KA, Kislalioglu MS, Phuapradit W, Malick AW, Shah NH. Release performance
of a poorly soluble drug from a novel Eudragit based multi-unit erosion matrix. Int
J Pharm 2001; 213:7–12.
14. Clarke, GM, Newton JM, Short MB. Comparative gastrointestinal transit of pellet systems
of varying density. Int J Pharm 1995; 114:1–11.
15. Devereus JE, Newton JM, Short MB. The influence of density on the gastrointestinal
transit of pellets. J Pharm Pharmacol 1990; 42:500–501.
16. Woodruff CW, Nuessle NO. Effect of processing variables on particles obtained by
extrusion-spheronization. J Pharm Sci 1972; 61:787–790.
360 Mehta et al.
© 2005 by Taylor & Francis Group, LLC
17. Schwartz JP, Nguyen NH, Schnaare RL. Compaction studies on beads: compression and
consolidation parameters. Drug Dev Ind Pharm 1994; 20:3105–3129.
18. Millili GP, Schwartz JB. The strength of microcrystalline cellulose pellets: the effect of
granulating with water/ethanol mixtures. Drug Dev Ind Pharm 1990; 16:1411–1426.
19. Malinowski HJ, Smith WE. Effect of spheronization process variables on selected tablet
properties. J Pharm Sci 1974; 63:285–288.
20. Malinowski HJ, Smith WE. Use of factorial design to evaluate granulations prepared by
spheronization. J Pharm Sci 1975; 64:1688–1692.
21. Erkoboni DF, Fiore SA, Wheatley TA, Davan T. The effect of various process and formulation
variables on the quality of spheres produced by extrusion/spheronization. Poster
presentation, AAPS National Meeting, 1991.
22. Chariot M, France`s J, Lewis GA, Mathieu D, Phan Tan Luu R, Stevens HNE. A factorial
approach to process variables of extrusion-spheronization of wet powder masses.
Drug Dev Ind Pharm 1987; 13:1639–1649.
23. Hileman GA, Goskonda SR, Spalitto AJ, Upadrashta SM. Response surface optimization
of high dose pellets by extrusion and spheronization. Int J Pharm 1993; 100:71–79.
24. O’Connor RE, Holinez J, Schwartz JB. Spheronization I: processing and evaluation of
spheres prepared from commercially available excipients. Am J Pharm 1984; 156:80–87.
25. Ojile JE, Macfarlane CB, Selkirk AB. Drug distribution during massing and its effect on
dose uniformity in granules. Int J Pharm 1982; 10:99–107.
26. Helle?n L, Yliruusi J, Merkku P, Kristoffersson E. Process variables of instant granulator
and spheronizer: I. Physical properties of granules, extrudate and pellets. Int J Pharm
1993; 96:197–204.
27. Kleinebudde P, Lindner H. Experiments with a twin screw extruder using a single-step
granulation/extrusion process. Int J Pharm 1993; 94:49–58.
28. Lindberg NO, Tufvesson C, Holm P, Olbjer L. Extrusion of an effervescent granulation
with a twin screw extruder, Baker Perkins MPF 50 D. Influence on intragranular porosity
and liquid saturation. Drug Dev Ind Pharm 1988; 14:1791–1798.
29. Harrison PJ, Newton JM, Rowe RC. The characterization of wet powder masses suitable
for extrusion/spheronization. J Pharm Pharmacol 1985; 37:686–691.
30. Land??n M, Rowe RC, York P. Characterization of wet powder masses with a mixer torque
rheometer. 3. Nonlinear effects of shaft speed and sample weight. J Pharm Sci 1995;
85:557–560.
31. Baert L, Remon JP, Knight P, Newton JM. A comparison between the extrusion forces
and sphere quality of a gravity feed extruder and a ram extruder. Int J Pharm 1992;
86:187–192.
32. Baert L, Fanara D, De Baets P, Remon JP. Instrumentation of a gravity feed extruder
and the influence of the composition of binary and tertiary mixtures on the extrusion
forces. J Pharm Pharmacol 1991; 43:745–749.
33. Ku CC, Joshi YM, Bergum JS, Jain NB. Bead manufacture by extrusion/spheronization—
a statistical design for process optimization. Drug Dev Ind Pharm 1993;
19:1505–1519.
34. Rowe RC. Spheronization: a noval pill-making process? Pharm Int 1985; 6:119–123.
35. Harrison PJ, Newton JM, Rowe RC. The application of capillary rheometry to the extrusion
of wet masses. J Pharm 1987; 35:235–242.
36. Hicks DC, Freese HL. Extrusion and spheronization equipment. Ghebre-Sellassie I, ed.
Pharmaceutical Pelletization Technology, Chap. 4. New York: Marcel Dekker, 1989:71–
100.
37. Harrison PJ, Newton JM, Rowe RC. Flow defects in wet powder masses. J Pharm Pharmacol
1984; 37:81–83.
38. Harrison PJ, Newton JM, Rowe RC. Congergent flow analysis in the extrusion of wet
powder masses. J Pharm Pharmacol 1984; 37:81–83.
39. Fechner P, Wartewig S, Fueting M, Heilmann A, Neubert R, Kleinebudde P. Properties
of microcrystalline cellulose and powder cellulose after extrusion/spheronization as
Extrusion/Spheronization as a Granulation Technique 361
© 2005 by Taylor & Francis Group, LLC
studied by Fourier transform Raman spectroscopy and environmental scanning electron
microscopy. AAPS PharmSci 2003; 5(4):Article 31.
40. Pinto JF, Buckton G, Newton JM. The influence of four selected processing and formulation
factors on the production of spheres by extrusion and spheronization. Int J Pharm
1992; 83:187–196.
41. Helle?n L, Ritala J, Yliruusi P, Palmroos A, Kristoffersson E. Process variables of the
radial screen extruder: I. Production capacity of the extruder and the properties of the
extrudate. J Pharm Tech Int 1992; 4:50–60.
42. Vervaet C, Baert L, Risha PA, Remon JP. The influence of the extrusion screen on pellet
quality using an instrumented basket extruder. Int J Pharm 1994; 107:29–39.
43. Baert L, Remon JP, Elbers JAC, Van Bommel EMG. Comparison between a gravity feed
extruder and a twin screw extruder. Int J Pharm 1993; 99:7–12.
44. Shah R, Kabadi M, Pope DG, Augsberger LL. Physico-mechanical characterization of
the extrusion/spheronization process. Part I. Instrumentation of the extruder. Pharm
Res 1994; 11:355–360.
45. O’Connor RE, Schwartz JB. Extrusion and spheronization technology. Ghebre-Sellassie
I, ed. Pharmaceutical Pelletization Technology, Chap. 9. New York: Marcel Dekker,
1989:187–216.
46. Hasznos L, Langer I, Gyarmathy M. Some factors influencing pellet characteristics made
by an extrusion/spheronization process. Part I. Effects on size characteristics and moisture
content decrease of pellets. Drug Dev Ind Pharm 1992; 18:409–437.
47. Hileman GA, Goskonda SR, Spalitto AJ, Upadrashta SM. A factorial approach to high
dose product development by an extrusion/spheronization process. Drug Dev Ind Pharm
1993; 19:483–491.
48. Baert L, Remon JP. Influence of amount of granulating liquid on the drug release rate
from pellets made by extrusion spheronization. Int J Pharm 1993; 95:135–141.
49. Wan LSC, Heng PWS, Liew CV. Spheronization conditions on spheroid shape and size.
Int J Pharm 1993; 96:59–65.
50. Mehta KA, Kislalioglu MS, Phuapradit W, Malick AW, Shah NH. Effect of formulation
and process variables on porosity parameters and release rates from a multi unit erosion
matrix of a poorly soluble drug. J Control Release 2000; 63:201–211.
51. Dyer AM, Khan KA, Aulton ME. Effect of the drying method on the mechanical and
drug release properties of pellets prepared by extrusion. Drug Dev Ind Pharm 1994;
20:3045–3068.
52. O’Connor RE, Schwartz JB. Spheronization II: Drug release from drug–diluent mixtures.
Drug Dev Ind Pharm 1985; 11:1837–1857.
53. Mehta KA, Kislalioglu MS, Phuapradit W, Malick AW, Shah NH. Multi-unit controlled
release systems of nifedipine and nifedipine: Pluronic F-68 solid dispersions: characterization
of release mechanisms. Drug Dev Ind Pharm 2002; 28(3):275–285.
54. Zhou F, Vervaet C, Remon JP. Matrix pellets based on the combination of waxes,
starches and maltodextrins. Int J Pharm 1996; 133:155–160.
55. Kleinebudde P, Jumaa M, El Saleh F. Influence of degree of polymerization on behavior
of cellulose during homogenization and extrusion/spheronization. AAPS PharmSci 2000;
2(2):Article 21.
56. Jerwanska E, Alderborn G, Newton JM, Nystrom C. The effect of water content on the
porosity and liquid saturation of extruded cylinders. Int J Pharm 1995; 121:65–71.
57. Millili GP, Wigent RJ, Schwartz JB. Autohesion in pharmaceutical solids. Drug Dev Ind
Pharm 1990; 16:2383–2407.
58. Maganti L, Celik M. Compaction studies on pellets I. Uncoated pellets. Int J Pharm
1993; 95:29–42.
59. Funck JAB, Schwartz JB, Reilly WJ, Ghali ES. Binder effectiveness for beads with high
drug levels. Drug Dev Ind Pharm 1991; 17:1143–1156.
60. Linder H, Kleinebudde P. Use of powdered cellulose for the production of pellets by
extrusion/spheronization. J Pharm Pharmacol 1994; 46:2–7.
362 Mehta et al.
© 2005 by Taylor & Francis Group, LLC
61. Fielden KE, Newton JM, Rowe RC. The influence of lactose particle size on the spheronization
of extrudate processed by a ram extruder. Int J Pharm 1992; 81:205–224.
62. Vervaet C, Baert L, Remon JP. Extrusion–spheronization: a literature review. Int
J Pharm 1995; 116:131–146.
63. Tapia C, Buckton G, Newton JM. Factors influencing the mechanism of release from
sustained release matrix pellets, produced by extrusion/spheronization. Int J Pharm
1993; 92:211–218.
64. Chien TY, Nuessle NO. Factors influencing migration during spheronization. Pharm
Technol 1985; 4:44–48.
65. Wang C, Zhang G, Shah NH, Infeld MH, Malick AW, McGinity JW. Compaction properties
of spheronized binary granular mixtures. Drug Dev Ind Pharm 1995; 21:753–779.
66. Nicklasson F. Compression Mechanics of Pharmaceutical Aggregates—Studies on the
Tabletting of Spheronized Aggregates with Varying Composition and Porosity Ph.D.
Thesis, Uppsala University, Sweden, 2000.
67. Iloan? usi NO, Schwartz JB. The effect of wax on compaction of microcrystalline cellulose
beads made by extrusion and spheronization. Drug Dev Ind Pharm 1998; 24:37–44.
68. Salako M, Podczeck F, Newton JM. Investigations into the deformability and tensile
strength of pellets. Int J Pharm 1998; 168:49–57.
69. Nicklasson F, Alderborn G. Modulation of the tableting behavior of microcrystalline cellulose
pellets by the incorporation of polyethylene glycol. Eur J Pharm Sci 1999; 9:57–65.
70. Haslam JL, Forbes AE, Rork GS, Ripkin TL, Slade DA, Khossravi D. Tableting of controlled
release multiparticulates, the effect of millisphere size and protective overcoating.
Int J Pharm 1998; 173:233–242.
71. Johansson B, Nicklasson F, Alderborn G. Effect of pellet size on degree of deformation
and densification during compression and on compactability of microcrystalline cellulose
pellets. Int J Pharm 1998; 163:35–48.
72. Maganti L, Celik M. Compaction studies on pellets II. Coated pellets. Int J Pharm 1994;
103:55–67.
73. Miller RA, Leung EMK, Oates RJ. The compression of spheres coated with an aqueous
ethylcellulose dispersion. Drug Dev Ind Pharm 1999; 16:1411–1426.
74. Be?chard SR, Leroux JC. Coated palletized dosage form: effect of compaction on drug
release. Drug Dev Ind Pharm 1992; 18:1927–1944.
75. Torrado JJ, Augsburger LL. Effect of different excipients on the tableting of coated particles.
Int J Pharm 1994; 106:149–155.
76. Beckert TE, Lehmann K, Schmidt PC. Compression of enteric-coated pellets to disintegrating
tablets. Int J Pharm 1996; 143:13–23.
77. Haubitz H, Mehnert W, Fro?mming KH. Preparation of theophylline multiple units
tablets. Pharm Ind 1996; 58:83–86.
78. Yuassa H, Takashima Y, Omata T, Kanaya Y. Studies on stress dispersion in tablets III.
Suppression of fracture of coated film by an excipients during the preparation of tablets
containing particles. STP Pharm Sci 2001; 11:221–227.
79. Stubberrud L, Eriksson M, Kordnejad K, Graffner C. Water–solid interactions. IV.
Influence of moisture sorption on the compaction of film-coated particles. Pharm Dev
Technol 1998; 3:141–151.
80. Lehmann K, Petereit HU, Dreher D. Scnellzerfallende Tabletten mit gesteuerter Wirkstoffabgabe.
Pharm Ind 1993; 55:940–947.
81. Palmieri GF, Wehrle? P, Martelli S. Drug release from compressed Eudragit RS 30D
coated beads. STP Pharm Sci 1996; 6:118–121.
82. Wagner KG, Krumme M, Beckert TE, Schmidt PC. Development of disintegrating multiple-
unit tablets on a high-speed rotary tablet press. Eur J Pharm Biopharm 2000;
50:285–291.
83. Tuno?n A? . Preparation of Tablets from Reservoir Pellets with an Emphasis on the Compression
Behavior and Drug Release. Ph.D. Thesis, Uppsala University, Sweden, 2003.
Extrusion/Spheronization as a Granulation Technique 363
© 2005 by Taylor & Francis Group, LLC
12
Effervescent Granulation
Guia Bertuzzi
IMA S.p.A, Solid Dose Division, Bologna, Italy
1. INTRODUCTION
Effervescent granulation is an important step in the production of fizzy dosage forms
that, most of the time, cannot be avoided to achieve the desired characteristics of the
effervescent tablets. It is also very critical because it can affect the stability of the
final dosage forms.
The first effervescent preparations were described over two centuries ago in the
official compendia; they were in powder form for use as cathartic salts. Later, in
1815, a patent described ‘‘a combination of neutral salt or powder which possesses
all the properties of the medicinal spring of Seidlitz in Germany, under the name
of Seidlitz Powders,’’ which contains sodium potassium tartarate, sodium bicarbonate,
and tartaric acid, in the proportions 3:1:1, respectively (1). Effervescent
granules and tablets have become more and more popular as the dosage form
because they are readily soluble and easy to consume just by drinking the glass of
water where they are dissolved.
To state the growing interest in such forms, in the 1980s—when electronic
bibliography searching was not available yet—the results of a literature search about
effervescent forms were published so as to help scientists working on new
developments (2).
According to the 4th edition of the European Pharmacopoeia, the effervescent
forms are defined as those granules or tablets that are to be dissolved in water before
administration to patients. Effervescent tablets or granules are uncoated and generally
contain acidic substances and carbonate or bicarbonate which reacts rapidly to
release carbon dioxide when dissolved in water. Disintegration of the tablets usually
occurs within 2 min or even less, due to the evolution of carbon dioxide.
Effervescent forms have many advantages over conventional pharmaceutical
forms. They substitute liquid forms when the active ingredient is less stable in liquid
form because they can be administered only by first dissolving the tablet in water.
Active ingredients that are not stable in liquid form are often more stable in the
effervescent form. Their administration is easy and is particularly helpful to patients,
for instance, children, who are not able to swallow capsules or tablets. They have
a pleasant taste due to carbonation, which helps to mask the bad taste of certain drugs.
365
© 2005 by Taylor & Francis Group, LLC
This could help to avoid the gastric side effect of certain drugs. In certain cases they
can shorten drug absorption rate in the body when compared to traditional tablets,
with a quicker therapeutic effect (3). They are easy to use and appeal to consumers
more than the traditional dosage forms because of their color and fizzy appearance.
Disadvantages of such solid dosage forms are more related to production technology
even if processing methods and equipment are the same as the conventional
ones. The general product requirements are similar to conventional granules: particle
size distribution and shape, uniformity of active distribution, to produce satisfactory
free-flowing granules, capable of being tableted using high-speed rotary tablet press.
However, it is also necessary to focus attention on some aspects of the procedure,
including compression and packaging because effervescent dosage forms—irrespective
of whether it is a sachet or a tablet—are larger and consequently difficult to produce.
The pharmaceutical industry has faced many problems, especially in the preparation
of effervescent tablets as it is certainly the application for which the choice
of the processing equipment is at least as important as the formulation design.
2. THE EFFERVESCENT REACTION
Effervescence is the evolution of gas bubbles from a liquid, as the result of a chemical
reaction. The most common reaction for pharmaceutical purpose is the acid–base
reaction between sodium bicarbonate and citric acid:
3NaHCO3 ?aq? 252g?3mol?
? H3C6H5O7 ?aq? 192g?1mol?
)3H2O?aq? 54g?3mol?
? 3CO2
132g?3mol? ? 3Na3C6H5O7 ?aq? 258g?1mol?
This reaction starts in presence of water, even with a very small amount as catalyzing
agent, and because water is one of the reaction products, it will accelerate the
rate of reaction, leading to difficulty in stopping the reaction. For this reason the
whole manufacturing and storage of effervescent products has to be planned by
minimizing the contact with water. Looking at the stechiometric ratios in the reaction
it is quite easy to understand the reason why effervescent tablets are so large.
Recently, some effervescent systems have been prepared that act as penetration
enhancers for drug absorption, not only in oral forms but also in some topical forms,
such as skin or vaginal applications. In these cases the reaction takes place directly
after administration, in the mouth due to saliva (4), upon the wounds due to blood
serum (5), or when formulated in a suppository. The effervescence can be provoked
by the moisture of the vaginal mucosa to treat vaginal infections or by simply adjusting
the pH.
There are other forms in which effervescence is based on a reaction different
from carbon dioxide formation. Effervescence is due to reactants that evolve oxygen
or other gases, which are safe for human consumption even if they are not suitable
for oral administration but can be employed in preparations for external use such as
antibacterial for dental plate cleaning.
3. FORMULATION
The criteria to choose the raw materials for effervescent products are not very different
from those for conventional tablets, since in both cases good compressibility and
compactability are the targets to be achieved.
366 Bertuzzi
© 2005 by Taylor & Francis Group, LLC
The intrinsic characteristics of effervescent forms bring some considerations
that limit the choice of the raw material, including the selection of the active ingredient.
Moisture content of the raw material is a very significant aspect, because it
affects compressibility and stability of the tablets. To avoid premature effervescent
reaction during the process or once the tablets are formed, raw materials with very
low moisture content have to be used.
Since an effervescent tablet is required to dissolve within 2 min or less in a glass
of water (about 100 mL), the solubility of raw materials and their rate of solubility
are other significant parameters. The active ingredient must be soluble, water dispersible,
or at least solubilized by salt formation during the dissolution in the glass of
water. The rest of the excipients, such as additives like sweeteners, coloring agents,
flavors, also have to be water soluble.
For all the above considerations the list of the excipients has not changed since
many years. However, the physical properties of these raw materials have recently
improved. Many different grades for each material are now commercially available,
also including also some preformulated grades, which are highly recommended for
direct compression. Formulators must choose the excipient grade on the basis of
the active ingredient, tablet size, and the process technology available. Therefore,
the ultimate use of the effervescent granules or tablets mainly affects the choice
of the raw materials.
To design an effervescent formula it is necessary to consider the stechiometric
ratios in the reaction, and the carbon dioxide solubility in water, which is 90 mg/
100mL of water (in STP conditions). The suggested ratio between acid and alkaline
components is about 0.6 but sometimes it might be required to increase the acid
source to get a pleasant taste. In fact, the alkali–acid ratio controls both the effervescence
capacity and the taste of the solution to be administered.
When the solubility of the active ingredients is not pH dependent, the alkali–acid
ratio can be optionally selected. This ratio can also be determined according to the pH
that is required for dissolving the active ingredient. In fact, when the active solubility
increases at the acid side, the pH of the solution is lowered by adding an excess of
the acidic agent. Conversely, an excess of alkaline source must be added when the active
ingredient is more soluble at higher pH (6). However, another approach that can be
used to increase the active ingredient solubility is to increase the volume of carbon
dioxide to be generated by increasing the alkaline component in the formulation.
As far as other excipients, such as diluents or binders, are concerned there is a
very little freedom for the formulator to experiment, because of the large dimension
of the tablet required for effervescent systems. In addition, compressibility cannot be
enhanced by additional binders for effervescent dosage form, because of the larger
size of the tablet.
In the latest development in effervescent forms, some formulations have been
designed to control the rate of effervescence, so to obtain a rapid, intermediate, or
slow rate. The rate control is strictly related to the acid–alkaline components ratio,
but the chemical properties of the effervescent excipients or their combinations can
have influence on it, especially when a slow rate of effervescence is required.
4. RAW MATERIALS
Because of the nature of the effervescence reaction, additional excipients are sparingly
used as the alkaline and acid ingredients also act as fillers to get a tablet bulk.
Effervescent Granulation 367
© 2005 by Taylor & Francis Group, LLC
They are in a so large an amount that effervescent tablets are much larger than conventional
ones. In case it is necessary to add a filler, sodium bicarbonate is the material
of choice due to its lower cost and because it does not influence final pH of the
solution and it also increases the effervescence effect. Sodium chloride and sodium
sulfate are other possible fillers. They are high-density crystalline powders that are
very compatible with the other ingredients.
Additives are added in a small amount to obtain tablets that are more attractive
to users. Flavors, colors, and sweeteners are used as usual in all the formulations.
4.1. Acid Materials
Necessary acidity for effervescence can be provided by three main sources: food
acids, acid anhydrides, and acid salts. Citric acid, tartaric acid, and ascorbic acid
are the most commonly used food acids because they are odorless, have a nice taste,
are not expensive, and are easy to handle.
4.1.1. Citric Acid
Citric acid is the more often used acidic ingredient because of its good solubility and
pleasant taste. It is commercially available in powder form and is either colorless or
exist as white crystals. The particle-size grades are: coarse, medium, fine, powder
(only anhydrous). It is very soluble in water and soluble in ethanol (7). It can be used
as monohydrate or anhydrate, depending on the selected equipment, technology,
and process conditions. It is very hygroscopic; however, the anhydrous form is less
hygroscopic than the monohydrate (8). However, caking of the anhydrous may
occur upon prolonged storage at humidity >70%. The monohydrate melts at
100
C, and releases the water of hydration at 75
C. For this reason, it can be used
as a binder source in hot melt granulation.
4.1.2. Tartaric Acid
It is very soluble in water and very hygroscopic compared to citric acid. In the effervescence
reaction with sodium bicarbonate it behaves like citric acid in producing an
evident effervescence. It must be used in a higher amount to get the proper stoichiometric
proportions, because it is a diprotic acid, while citric acid is a triprotic one.
In terms of compressibility it is comparable to citric acid (8).
4.1.3. Ascorbic Acid
It is white in crystalline form and light yellow in fine powder. It is not hygroscopic
and this may be helpful in production because it is easier to handle. It is freely soluble
in water (1 g in about 3 mL) and absolute ethanol (9). If exposed to light, it gradually
darkens. Its behavior in the effervescent reaction with sodium bicarbonate is
comparable to the other acids (citric and tartaric) in terms of release rate of carbon
dioxide.
4.1.4. Acid Anhydrides
Anhydrides of food acids are potential acid source as they are precursors of the corresponding
acid on hydrolyzation. The effervescent effect is strong and sustained by
the continuous production of acid in the solution. Water has to be avoided during
368 Bertuzzi
© 2005 by Taylor & Francis Group, LLC
the whole process when anhydrides are part of a formulation. Otherwise they would
be hydrolyzed to the corresponding acid before their use in the product (10).
4.1.5. Acids Salts
Sodium dihydrogen phosphate and disodium dihydrogen pyrophosphate are acid
salts that have been used in effervescent formulation. They are soluble in water, producing
acid solution and react quickly with alkaline sources. They are commercially
available either as granules or powder.
4.1.6. Other Less Frequent Sources of Acid
 Fumaric and nicotinic acids, which are not hygroscopic, but have low water
solubility.
 Malic acid has recently been introduced in effervescent formulations because
of its smooth and light taste. It is highly hygroscopic and soluble but has
less acid strength than tartaric or citric acids.
 Acetylsalicylic acid, though an active ingredient that is very commonly administered
in effervescent preparations, cannot be used as an acid source because
of its low water solubility.
 Adipic acid deserves to be mentioned even if it does not function as an acid
source because of its low water solubility. It has given good results as
a lubricant for effervescent calcium carbonate tablets (11).
4.2. Sources of Carbon Dioxide
Solid carbonates salts are the most popular source for effervescents; bicarbonate
forms are more reactive than carbonates.
4.2.1. Sodium Bicarbonate
Sodium bicarbonate is the major source of carbon dioxide in effervescent forms and
is able to provide a yield of 52% of carbon dioxide. It is commercially available in
five grades according to particle size, from free-flowing uniform granule to fine powder,
which are odorless and slightly alkaline in taste. When heated, the bicarbonate is
converted into anhydrous sodium carbonate. This reaction is time and temperature
dependent. Ninety percent of the conversion is achieved within 75 min at 93
C, but
dehydration starts at 50
C, which must be considered as a critical temperature in
processing (12).
Being a nonelastic material sodium bicarbonate has a very low compressibility
but this issue has been overcome since its production by spray-drying technique.
Direct compressible grades are now available but contain some additives such as
polyvinylpyrrolidone (PVP) or silicone oil.
4.2.2. Sodium Carbonate
Sodium carbonate is commercially available in three different forms, all very soluble
in water: anhydrous, monohydrate, or decahydrate (13). It is more resistant to the
effervescent reaction and in some formulations can be used as a stabilizing agent
in amounts not exceeding the 10% of the batch size. It is used as a stabilizing agent
in certain effervescent formulas because it absorbs moisture preferentially, preventing
the effervescent reaction from starting. Of course the anhydrous form is the
Effervescent Granulation 369
© 2005 by Taylor & Francis Group, LLC
preferred choice for this purpose. Recently, a particular grade of sodium bicarbonate
has been produced as round-shaped particles, coated with a carbonate layer to
increase bicarbonate stability (14).
4.2.3. Potassium Bicarbonate and Potassium Carbonate
They substitute the sodium salts when sodium ion is not required (15). They are less
soluble than the corresponding sodium salts and are more expensive.
4.2.4. Calcium Carbonate
Precipitated calcium carbonate occurs as fine, white, odorless, and tasteless powder
or crystals. Its water solubility is very poor, and it is not soluble in ethanol or isopropanol.
It is a high density powder, and thus not suitable for compression. It is
normally used as a drug in effervescent tablets for patients who suffer from calcium
deficiency. It can also be used as alkaline source because it provides stability to the
effervescent system (16).
4.2.5. Sodium Glycine Carbonate
Sodium glycine carbonate provides a light effervescence reaction, but causes rapid
disintegration of the tablets, so it has been applied in the preparation of fastdissolving
sublingual tablets. It is much more compressible than the other alkaline
compounds and has been found suitable for direct compression (17).
4.3. Binders
The use of a binder in effervescent formulations is limited by the fact that any binder,
even if water soluble, will retard the tablet disintegration. Therefore, the amount of
binder in a given formula will be a compromise between desired granule strength and
desired disintegration time.
As will be described in Section 5.1.2, water itself is an effective binder for effervescent
granules when granulated with all the components together. A small amount
of water, very finely distributed on the powder bed, acts as a binder by partially dissolving
the raw materials so that they can be agglomerated. Other solvents, ethanol
and isopropanol are not binders themselves, but can be used as granulating liquids to
dissolve dry binders.
Binders for dry granulation, such as lactose, mannitol, dextrose, are almost
inappropriate, because they would be effective only in larger amount than that
allowed by an effervescent formulations. The binder choice in wet granulation is also
limited by the method of production and consequently by the amount of granulating
liquid.
In case of granulation of both the alkaline and acid components together with
water, it would not make sense to put a binder in the formulation because the small
water amount will never be able to dissolve the binder.
The most popular binder for effervescent tablets is polyvinylpyrrolidone
(PVP). Types K25 and K30 are preferred for their water solubility and dissolution
rate which is an important parameter for the final purpose of effervescent tablets.
PVP is effective at low percentage in the formula, starting from 2% and over. It is
feasible either for dry or wet granulation. It is soluble in water, alcohols, and
hydroalcoholic liquids (18).
370 Bertuzzi
© 2005 by Taylor & Francis Group, LLC
4.4. Lubricants
Because tableting is a critical step in effervescent production, selecting the lubricant
is one of the most important issues. Lubrication of the effervescent mixture is quite
problematic because of the chemical–physical nature of the lubricants. Most of the
lubricants, because of their low water solubility, inhibit the tablet disintegration,
which, as already said, must be very rapid in case of effervescent tablets. The effervescent
tablets—mainly for marketing reasons—are often required to provide a clear
transparent solution, that is, without any insoluble ‘‘scum’’ forming on the water
surface, or any residue left. In selecting a lubricant, proper attention must be given
to its solubility in water along with its compatibility with the active ingredient.
Many different lubricants have been tested for a long time to establish the most
appropriate one for effervescent tablets (19), including the opportunity to carry out
external lubrication of the granules directly in the dies of the tablet-press.
Lubricant substances which are reported in literature as suitable for effervescent
manufacturing because they are water soluble are: sodium benzoate, sodium
acetate, L-leucine, and Carbowax 4000. A very recent application is a combination
of calcium and potassium sorbates, micronized polyethylene glycol with calcium
ascorbate or trisodium citrate (20). Combination of spray-dried L-leucine and polyethylene
glycol 6000 has been reported as a successful lubricant in the literature
(21). Other less soluble lubricants have been used in formulating effervescent tablets,
however, a balance should be found between compression efficiency and water
solubility. Magnesium stearate is also employed but the most suitable, commercially
available type is its combination with sodium lauryl sulfate, a surface-active agent
that helps in its dispersion (22).
4.5. Additives: Sweeteners, Coloring Agents, and Flavors
Coloring agents can include all the dyes soluble and suitable for food, such as those
the F.D.&C. ones and all the natural coloring substances in amounts varying in the
range of 0.1–3.5% of the total weight of the formulation.
Flavors can be selected from synthetic flavors or natural extracts. Lemon,
orange, and other fruit essences are particularly suitable to obtain organoleptic
requirement in amounts varying from 0.5% to 3.0% of the total weight of the
formulation.
5. MANUFACTURING OF EFFERVESCENT FORMS
Manufacturing conditions are crucially important even with regards to stability of
the products after they have been packed. Almost all the raw materials used for effervescent
tablet manufacturing are hygroscopic, and hence moisture absorption from
the air must be prevented to avoid the effervescent reaction prior to use of the
tablets.
mixing or granulating, lubricating, tableting, and packaging) can be carried out in
completely closed and integrated handling system, consisting of intermediate bulk
containers (IBCs), tumblers for IBCs, docking and dosing stations.
In this case only the packaging area will be ventilated with low moisture content
of air. Otherwise traditional open handling systems can be employed but the
whole facility has to be conditioned with air at minimum level of moisture content
Effervescent Granulation 371
The whole production process as shown in Figure 1 (dosing of the ingredients,
© 2005 by Taylor & Francis Group, LLC
(23). In fact, the suggested conditions throughout the plant are: relative humidity
(RH) below 20% and uniform temperature at 21
C. It is known, however, that
25% RH at controlled room temperature (25
C) is enough to avoid instability caused
by atmospheric moisture (24).
Manufacturing effervescent drugs on a large scale is usually done using a semicontinuous
procedure, by paying attention to synchronize all the process steps, in
order to achieve the largest production throughput. A continuous process flow, with
continuous feeding of raw materials and collection of granules, can be performed by
extrusion of the wet mass and drying in a continuous fluid bed dryer. Granulation of
effervescent mixtures must, most of the times, be executed in batches and is definitely
the most critical step of this particular kind of pharmaceutical manufacturing, as it
strongly influences the characteristics of the final forms, granules or tablets, and consequently
the following steps of production. Lubrication of the granulation,
Figure 1 Integrated production plant.
372 Bertuzzi
© 2005 by Taylor & Francis Group, LLC
compression of tablets, and packaging of effervescent tablets should be carefully
planned to produce a suitable effervescent product.
The critical issues discussed earlier, for the compression of effervescent granules,
are related to the low compressibility of the majority of the raw materials,
large dimensions of the tablets, poor content of binder in the formulation, and diffi-
culties of lubricating the mixture. For all these reasons the tableting phase has to be
carried out carefully, and even the choice of the tablet press is of great importance.
To overcome lubrication difficulties, some suppliers of tablet press have developed
equipment that is capable of carrying out external lubrication of the granules. Antiadherent
materials are sprayed directly into the dies of the tablet press, during the
pause phase of compression, so as to prevent sticking of granules upon dies and
punches. External lubrication is not as good a method as using lubricant substances
to disperse in the granules. It is not considered the best solution for lubrication
because it is considered less compliant to the standard GMP (good manufacturing
practices) rules; In addition, the assembling and disassembling of the tablet press
is complex. An alternative method to obtain better tablets, facilitating the compression,
is to tablet the granules while they are still slightly wet, or not dried yet as we
are still considering very low moisture content of <1%. Tablets are then dried and
stabilized by means of a process step in a static ventilated oven.
Packaging operations for both granules and tablets must be done, as already
mentioned, in a low humidity environment. The critical aspects about packaging
of effervescent drugs are obviously related to stability of the tablets and granules
and the main objective is to protect them, as much as possible, not only during
packaging operations but also after they are packed, so as to impart a reasonable
shelf-life. The oldest packaging method for effervescent preparations consisted of
wrapping the acid and alkaline components separately, to avoid the premature effervescent
reaction until use. All the effervescent drugs can be packed in individual dose
units, in airtight containers made of protective aluminum foil or plastic laminates.
Tablets can also be packed by stacking them one by one in plastic or metal tubes,
which have almost the same diameter of the tablets so as to minimize the air which
remains in contact with the tablets. The tubes must be resealed every time, after taking
out each tablet.
Certain tablets are also wrapped in an aluminum foil before packaging in the
tube and this seems to be the best solution for long-term stability. Patented types of
tubes containing silica-gel in the internal side of the cap are the most recent invention
(25).
For tablets that are packed in strips, it is essential that the packaging machine
has a fine control over the temperature of the welding unit, so that an accurate sealing
of the strips can be obtained and overheating phenomenon that could provoke
release of residual water from the tablets can be avoided (24).
5.1. Granulation Methods
Two main granulation methods have been known for a long time. The 1911 edition
of the British Pharmacopoeia reported a detailed description of the manufacturing
procedure (26).
Effervescent granules are made by mixing citric and tartaric acids with the
medicament, and the sodium bicarbonate with the sugar when present; and then
thoroughly mixing the one with the other, and granulating the resulting mixture
by stirring in a pan heated to between 93
 and 104
, passing through sieves of
Effervescent Granulation 373
© 2005 by Taylor & Francis Group, LLC
a suitable size, and drying at temperature not exceeding 54
. This method for
preparing the granules yields satisfactory results but the following alternative
method has also been suggested: Mix the sodium bicarbonate, the sugar, and
the medicament, pass the mixture through a No 20 to No 30 stainless steel sieve,
subject the mixed acids to the same process, and thoroughly mix the two sifted
powders. Place the mixed powders in layers on suitable dish, pan or glass tray,
heated to 75
 to 85
, if required but not exceeding the higher temperature. When
the mass, after being suitably kneaded and compressed, has assumed a uniform
plastic condition, suitable for granulation, rub it to a No 5 to No 10 stainless
steel sieve, according to the size of granules desired, and dry the granules at
a temperature not exceeding 50
.
Only a few aspects have substantially changed in the modern methods. Two
main methods can be executed with various types of granulation equipment:
a. Single-step method. All the components of an effervescent formula are
granulated together, handled with care, running the process in an contained
manner to maintain the stability of the mixture until the step is
completed. This single-step method is normally carried out using dry
granulation, hot-melt granulation, and certain wet granulation processes.
b. Multistep method. The alkaline and the acid components are granulated
separately, then mixed together, just before the tableting or packaging
step. Usually these are also typical applications of wet granulation technologies.
5.1.1. Dry Granulation
Dry granulation by roller compactor is definitely the most appropriate method for its
simplicity, low costs, and higher product throughput. The number of operations and
space required are less and consequently, air ventilation is reduced. On the other
hand, not all the excipient grades are suitable for such a technology. More sophisticated
grades, and thus expensive, pre-prepared by raw material bulk suppliers are
required.
An alternate process to dry granulation is direct compression of the blend of all
the raw materials, in the attempt to avoid operating and stability problems. It would
be the ideal process for effervescent tablet manufacturing, but its application is limited
to a few cases, for example, when the active ingredient cannot be granulated
(e.g., when it is already included in a complex like with a cyclodextrine), or contains
some water of crystallization.
5.1.2. Wet Granulation
Despite some disadvantages, wet granulation is still the most preferred method for
effervescent granulation. As required for conventional tablets, this method assures
homogeneous granules, suitable for compression, and is able to provide uniform
tablets either in terms of weight or active ingredient content.
For this technique, it is necessary to use a granulating liquid that might interact
with the powders initiating the effervescent reaction. Hence it is essential to handle
the process with great care.
A wet granulation process in two separate steps is mostly recommended and
suitable for conventional equipment like high-shear granulator-dryer and fluid bed
processor. Acidic and alkaline ingredients are granulated separately. The two granules
are then mixed together, just before adding the lubricant for tableting. Water,
374 Bertuzzi
© 2005 by Taylor & Francis Group, LLC
alcohols, or hydroalcoholic solutions can be used as a binding liquid because this
process is a standard wet granulation process. The usual procedure is to granulate
only one of the effervescent sources and add the other one in powder in the final
blending. All these possibilities are illustrated in Figure 2.
In certain cases only the acid components of the formulation are granulated
and then mixed with the sodium bicarbonate, preferably when it is of fine granular
grade. Other additives such as flavors and lubricants can then be added and mixed.
This approach increases productivity and reduces costs since one granulation step
has been eliminated.
The peculiar process that distinguishes effervescent granulation consists in a
single-step granulation of all the components of the formulation, which can be performed
either with nonreactive or reactive liquids with reference to the effervescence
reaction.
5.1.3. Single-Step Granulation
Single-step granulation process provides dry effervescent granules directly by granulating
the acid and the alkaline materials together. It is possible to use only water
as the granulating liquid, thus controlling the effervescent reaction to granulate.
A nonreactive liquid like absolute ethanol or isopropanol can also be used, but in
this case it is necessary to use a binder to agglomerate the raw material particles.
5.1.3.1. Process with Water. A very small amount of water, less than 1% of
the batch size, can be used to initiate the effervescent reaction. Some carbon dioxide
is released, but some water too, which acts as the binding liquid as well. It takes a few
minutes (from 5 to10 min) to obtain wet granules. The effervescent reaction rate
rapidly increases and becomes difficult to stop; it must however be terminated very
quickly otherwise no product will remain. An immediate start of the drying of the
granules will control and stop the effervescent reaction. For single-step technology,
both fluid bed granulator and high-shear granulator-dryer are suitable.
Figure 2 Alternative granulation processes.
Effervescent Granulation 375
© 2005 by Taylor & Francis Group, LLC
In high-shear granulator-dryer technology it is possible to suddenly switch to
drying phase by creating vacuum inside the bowl, just after the granulation phase
when wet granules are well massed. Vacuum is created in a few seconds, which
immediately provokes the decrease of the water boiling point down to about
20
C. At the same time the bowl is heated up to provide more energy for water
evaporation. The water released on the granules surface is removed within a few
seconds and the effervescent reaction stops.
The application of microwave radiation, combined with vacuum inside the
bowl of the high-shear granulator (27), can also be used to stop the effervescent reaction
and to dry the effervescent granules (28).
Drying of effervescent granules is a shorter process than drying conventional
granules granulated with water, because of the very small amount of water involved
in the process. However, drying still remains a critical step since it is very hard to
remove, even the small quantity of water, from hydrophilic or hygroscopic materials.
Typical drying time is about 50 min for 20 kg batch of effervescent granulation under
vacuum. See Figure 3 for comparison of drying of conventional and effervescent
granules by vacuum technology.
Drying rate for effervescent granules is lower because the moisture content to
be removed is in a range below 2% of the batch size. Consequently, the drying time,
passing from pilot scale to industrial scale, does not increase as much as for conven-
The great advantage of this method for effervescent tablet preparation lies in
the installation of equipment that does not require explosion-proof systems. An
example of effervescent aspirin produced in 600 L high-shear granulator-dryer
The formulation consists of:
Anhydrous citric acid 116.6 kg
Sodium bicarbonate 154.2 kg
Sodium carbonate 39.2 kg
Acetylsalicylic acid 50 kg
Figure 3 Drying rate of effervescent granules, compared to conventional granules.
376 Bertuzzi
tional granules (Fig. 4).
equipped with vacuum and tilting bowl (29) is shown in Figure 5.
© 2005 by Taylor & Francis Group, LLC
First, the effervescent system is granulated with very small amounts of water
(2–4 mL/kg) and sprayed in very fine droplets. The acetylsalicylic acid is added
later, in the final blending, after granulation is completed. The results of three
Figure 4 Drying time scale up for high-shear granulator-dryer.
Figure 5 ‘‘Rotocube’’ high-shear granulator-dryer equipped with vacuum and tilting bowl.
Effervescent Granulation 377
batches are reported in Table 1.
© 2005 by Taylor & Francis Group, LLC
Table 1 Granules Produced with Single-Step Technology in a High-Shear Granulator-Dryer
Results of samples
Yield
Batch
code
Batch
size (kg) kg %
Sampling
time (min)
Moisture
content
(0.1%)
Acid
neutralizing power
(185mL
0.1N acid/tablet)
pH
(6.0–6.4)
Batch #1
granulated with
720mL of water
(2 mL/kg)
360 352.2 97.80 30 0.048%
60 <0.01%
90 <0.01% 244.6 6.1
Batch #2
granulated with
1440mL of water
(4 mL/kg)
360 336.4 93.44 30 <0.01%
60 <0.015%
After
discharge
<0.01% 244.6 6.4
Batch #3
granulated with
1440mL of water
(4 mL/kg).
Note: no tilting
of the bowl while
drying
360 323.5 89.86 60 0.075% 229.8
After
discharge
<0.01% 245.5 6.25
378 Bertuzzi
© 2005 by Taylor & Francis Group, LLC
The lower yield of the third batch is due to product sticking to the walls of the
bowl, which could not be discharged. The reason why it happened is related to the
set parameters during drying phase: in batch #3, drying was performed by keeping
the bowl static, instead of tilting it as in the other batches. The batch size in actual
production was then increased to 644 kg in 1400 L equipment.
All the components of an effervescent mixture can be granulated together in
a single-step process in a conventional fluid bed granulator-dryer. Granulation occurs
when water is sprayed on the fluidized bed, initiating the effervescent reaction. The
reaction is stopped when water is not sprayed anymore and drying phase is carried
out with warm dry air. It is quite understandable that this method is difficult to
control and reproduce (30). A subsequent patent application (31) describes an
improvement of the method reproducibility, which can be achieved by controlling
the air humidity (it has to be less than 1 g/m3), using an hydroalcoholic solution
instead of water. Even better results in controlling the effervescent reaction are
achieved when spraying and drying phase are combined together.
It is, however, very difficult to reproduce such a process. Therefore, an alternative
procedure to manufacture effervescent granules has been invented using a rotor
fluid bed spray-granulator (32).
Warm air, which is the only method available for drying in fluid bed technology,
is not capable of stopping the reaction all at once as it happens by applying
vacuum inside the processing bowl. Therefore, the only way to proceed is to
minimize in some way the contact between the two components of the effervescent
system. An intelligent and brilliant hypothesis put forward by Professors P. Gautier
and J.M. Aiache (32) is to alternate the granulation of the acid materials with that of
alkaline materials while still using a single equipment such as a rotary fluid bed system.
In the literature a vitamin C formulation is reported to explain this technique
but other active ingredients can also be used (33).
The process consists of two or three continuous steps to produce effervescent
spheres by layering the acid components over alkaline spheres or vice versa. The
binding liquid is, however, a hydroalcoholic solution in which PVP the binder must
be previously dissolved.
The first step is the granulation of alkaline components in the rotary fluid bed.
In the second step, the granulating solution is sprayed in combination with the acidic
powders, which deposit on the alkaline spheres creating an external acid layer separated
by a neutral layer of the binder. When agglomeration is completed, drying
phase with hot air starts with no interruptions (32).
5.1.3.2. Process with Alcohol or Hydroalcoholic Solution. As reported in the
previous example, it is sometimes preferable to granulate with a hydroalcoholic solution
to initiate a lighter effervescence so as to keep the reaction under better control
during the process. Use of alcohols is indispensable in case a binder like PVP is
included in the formulation. In fact, the amount of water required to dissolve
PVP to obtain the binding action will be too high, and it will not be possible to keep
the effervescent reaction under control.
As already suggested for a conventional granulation process that requires the
use of an inflammable liquid, it is convenient to install fully explosion-proof equipment
with an accessory solvent-recovery utility that will limit the emission of vapors
in the atmosphere. Solvent recovery will be more advantageous while drying under
vacuum than with a fluid bed.
Effervescent Granulation 379
© 2005 by Taylor & Francis Group, LLC
Certain high-shear granulator-dryers are equipped to achieve 99% of solvent
recovery by installing a tank to collect the condensate, and by bringing down the
possible residual exhaust with a shower of water.
5.1.4. Hot Melt Granulation
Hot melt process is an alternative technology to wet granulation (which is discussed
in detail in another chapter of this book). Agglomeration of the particles of a powder
mixer can be achieved by melting hydrated citric acid so as to release the hydration
water which acts as the granulating liquid. Once granules are formed it is necessary
to cool them to achieve the proper hardness and mechanical stability. There are two
different techniques:
 Surface hot melt granulation (SHMG), which consists of mixing all the raw
materials together in a blender and then drying the mix in a tray oven at
90
C. Water is then released from citric acid and other ingredients to form
granules (34). Unfortunately, batch reproducibility of this method is very
low.
 Hot melt granulation is normally carried out in high-shear granulator-dryer
with the capability to heat up the bowl. In certain cases the released water
of hydration of citric acid initiates the effervescent reaction to produce
additional water which acts as the binding liquid. However, this process
for obvious reasons is difficult to control. The same process has been
applied to fluid bed spray-granulator where low melting point polymers,
like polyethylene glycols (PEGs) or polyoxyethylene glycols can also be
used as binders (35).
5.1.4.1. Hot Melt Extrusion. Hot melt extrusion is a recent patented method
to produce effervescent granules, especially dedicated to produce granules having a
controllable rate of effervescence (36). The formulations for this technology must
contain a hot-melt extrudable binder. Preferred binders are the polyethylene glycols
with molecular weight in the range 1000–8000 Da, but some other polymers have
been investigated. Binder percentage varies according to the formulation in a range
of 20–40% of the total weight. There are two main extrusion techniques to carry out
in extruders that must be equipped with a solid conveying zone, multiple separate
temperature controllable heating zones, and an extrusion die:
a. A blend of all the ingredients, including the active ingredient of the formulation,
is hot-melt extruded at high temperature, in order to melt or
soften the binder. The extrudate is then ground or chopped to obtain
effervescent granules.
b. The acidic agent and the hot-melt binder are formulated in the right proportions
to obtain an eutectic mixture that has decreased the melting
point temperature. This binary mixture is separately melted, the alkaline
agent is added as powder in the next step. The melted mixture is then
extruded, chopped, or ground as in the previous method.
To control the effervescence rate of the final dosage form, it is possible to
adjust some parameters like the temperature and the rate of extrusion. The temperature
range selected can be critical because degradation of the active agent may occur
alongside decomposition of the effervescent components. This range is usually from
about 50
C to about 120
C.
380 Bertuzzi
© 2005 by Taylor & Francis Group, LLC
Table 2 Manufacturing Process for Effervescent Tablets
Dry granulation Granulation by heating Wet granulation
Process
Slugging or
laminating
Direct
Compression
Surface hot
melt granulation
Hot melt
granulation
Granulation?drying
(two steps)
Granulation?drying
(one step)
Equipment Blender?tablet
press compactor
Blender Blender?tray
dryer
Single-pot mixergranulator-
dryer
Mixer?fluid
bed dryer
Single-pot
mixer-granulator-dryer
Number
of steps
6 3 6 5 6 4
Estimated
time (h)
23 10 22 14 15 12
Advantages Fast process No product
transfers
No granulating
liquid
No product
transfers
Batch
reproducibility
Batch
reproducibility
No product
transfers
No granulation Totally closed No granulating
liquid
No granulation Short time
Disadvantages Raw materials
at low residual
moisture
Raw materials
for direct compression
at low residual
moisture (more
expensive)
Loss of
effervescence
Difficult to
carry out
Difficult
to clean
Difficult
to clean
Dusty process Long time Difficult
to clean
High manpower
required
Dusty process
High manpower
required
Note Obsolete Rarely applied
Effervescent Granulation 381
© 2005 by Taylor & Francis Group, LLC
The rate of extrusion is related to the time of materials exposure to high temperature
which is usually less than 5 min.
The hot-melt extrusion technology in some cases can be run as a continuous
process, having a higher throughput than batch hot-melt granulation process per
batches.
All the previous sections provided an overview of all the possible technologies
to manufacture effervescent granules but how to choose the most appropriate technique
for a certain formulation. An interesting study to figure out the best produc-
(37).
Three technologies were evaluated: dry granulation, hot melt granulation, and
wet granulation. The choice of the process technology was strictly related to the physical
properties of the raw materials, namely particle size, density, flowability, and
moisture content. Moisture content is definitely the most significant parameter since
the powder mixture has less than 0.2–0.3% of moisture content. It would be stable
but difficult to tablet and pointing such a situation, a wet granulation process is
required.
The use of high-shear granulator-dryer has been found as the more economic
and flexible production method by other researchers (38) as well. This technology
allows us to use a wider range of excipient grades, avoiding problems related to
particle size or moisture content of the raw materials. Even if the granules produced
by this technology appear finer, their flow properties are good despite the large
fraction of fine-sized particles. Tableting properties are always in line within the
specifications.
REFERENCES
1.
2. NCF. December 1984 pp. 19–22; May 1985 pp. 21–23.
3. Moller PL, Norholt et al. J Clin Pharmacol 2000; 40(4):370–378.
4. US Patent 6,200,604, 1999.
5. Rapp M. PCT Int Appl 2000.
6. US Patent 6,077,536, 1998.
7. Handbook of Pharmaceutical Excipients. Washington/London: The American Pharmaceutical
Association/The Pharmaceutical Society of Great Britain, 1986:78–80.
8. Schmidt PC, Bro?gmann B. Dtsch Apoth-Ztg 1987; 127:991–997.
9. Handbook of Pharmaceutical Excipients. Washington/London: The American
Pharmaceutical Association/The Pharmaceutical Society of Great Britain, 1986:6–8.
10. Repta AJ, Higuchi T. J Pharm Sci 1969; 58:1110–1113.
11. Romaco Zanchetta R&D Laboratory archive.
12. Handbook of Pharmaceutical Excipients. Washington/London: The American Pharmaceutical
Association/The Pharmaceutical Society of Great Britain, 1986:263–265.
13. Handbook of Pharmaceutical Excipients. Washington/London: The American
Pharmaceutical Association/The Pharmaceutical Society of Great Britain, 1986:436–438.
14. SPI Pharma Group. Fall 2001.
15. Duvall RN, Gold, G. Miles, Inc., USA, 1990, 6 pp.
16. US Patent 6,242,002, 1999.
17. US Patent 6,284,272, 1999.
18. Handbook of Pharmaceutical Excipients. Washington/London: The American Pharmaceutical
Association/The Pharmaceutical Society of Great Britain, 1986:234–239.
382 Bertuzzi
tion method for effervescent tablets was presented by Laugier and Rona (Table 2)
© 2005 by Taylor & Francis Group, LLC
Homan P. Pharm J 2001; 267:911–936 (http://www.pharmj.com).
19. Strickland WA Jr, Higuchi T, Busse L. J Am. Pharm. Assoc. (Sci. Ed.), 1956; 45:51–55.
20. Daher LJ. Bayer Corporation, USA. US Patent 1999, 6 pp.
21. Rotthaeuser B, Kraus G, Schmidt PC. Pharmazeutisches Institut, Eberhard-Karls-Univ.,
Tuebingen, Germany. Pharmazeutische Industrie 1998; 60(6).
22. Rudnic E, Schwartz JB. Oral Solid Dosage Forms, Chapter 45, Tablets, 861, Remington.
23. Armandou J-P, Mattha AG. Pharm Acta Helv 1982; 57:287–289.
24. Mohrle R. Effervescent tablets. Lieberman HA, Lachman L, eds. Pharmaceutical Solid
Dosage Forms. Vol. 1. New York: Marcel Dekker, 1980:225–258.
25.
26. The British Pharmaceutical Codex. Published by direction of the Council of the Pharmaceutical
Society of Great Britain, 1911
27. Collette News Vol. 2, Issue 1, May 2001, GEA Powder Technology Division.
28. Aiache JM, Cardot JM. Utilisation des micro-ondes dans la fabrication des formes
pharmacutiques. Conf. INSA Roeuen, 1999.
29. Romaco Zanchetta R&D Laboratory Archive.
30. Coletta V, Kennon L. J Pharm Sci 1964; 53:1524–1525.
31. Patent EP 673,644.
32. Gauthier P, Aiache J-M. Pharm Technol Eur 2001; 13(10):32 (see also 34
33. US Patent 6,210,711, 1999.
34. Yanze FM, Duru C, Jacob M. Laboratoire de pharmacie galenique, pharmacotechnie
et de Biopharmacie, Universite Montpellier I, UFR de Sciences Pharmaceutiques,
Montpellier, France, Die pharmazie 12/2000.
35. Yanze FM, Duru C, Jacob M. Drug Dev Ind Pharm 2000; 26(11):1167–1176.
36. US Patent 6,488,961, 1999.
37. Laugier M, Rona R. PMPS Spring 2002:26–28.
38. Pearlswig DM. Simulation Modeling Applied to the Single Pot Processing of
Effervescent Tablets. Master’s Thesis, North Carolina State University, Raleigh, NC.
Effervescent Granulation 383
© 2005 by Taylor & Francis Group, LLC
http://www.desiccantcity.com/CASE_HISTORIES/History11.htm.
(http://www.scocca.org/herbmed/eclectic/
bpc1911/granulae.html).
pp. 36–37).
p. and
13
Melt Granulation and Pelletization
T. W. Wong
Faculty of Pharmacy, University of Technology MARA, Selangor, Malaysia
W. S. Cheong and P. W. S. Heng
National University of Singapore, Singapore
1. INTRODUCTION
Melt granulation and melt pelletization are agglomeration processes that have
gathered increasing interest in the pharmaceutical industry for the concept of utilizing
a molten liquid as a binder. Unlike the conventional use of aqueous or organic
solvents as binders, the binding liquid in melt processes remains as a constituent
of the formulation. However, the basic principles in melt agglomeration processes
are relatively similar to those of wet agglomeration processes with solvents except
that the interpretation of the melt agglomeration processes is not complicated by
an evaporation of the molten binding liquid.
1.1. Overview of Agglomeration
Agglomeration is one of the key pharmaceutical manufacturing processes concerning
the conversion of fine solid particles into larger entities by agitating the fine solid particles
in the presence of a binding liquid using equipment such as tumbling drum,
fluid bed granulator, and high-shear mixer. The agglomeration is considered as
a wet process because a binding liquid is required for the wetting of solid particles
prior to agglomeration. The agglomerates formed have a specific mean size and
shape, improved flowability and mechanical strength, narrow bulk density and porosity
values, as well as, a modified drug release rate (1,2). Agglomerates can be classified
into two types: granules and pellets, based on their physical characteristics.
Granules are irregularly shaped agglomerates which have a rather wide size distribution,
typically within the range of 0.1–2 mm. Pellets are spherical agglomerates with
a narrow size distribution, within the range of 0.5–2 mm. Traditionally, the binding
liquid required for agglomeration of fine solid particles is aqueous or organic solvent.
The solid particles are held together by the binding liquid which provides
temporary binding force by means of liquid bridges, and subsequently by solid
bridges of the residual solute after the removal of solvent by evaporation. The hardness
of solid bridges largely determines the strength of the dried agglomerates (3).
385
© 2005 by Taylor & Francis Group, LLC
The solid bridges can also be formed by the resolidification of molten material. This
is the basis by which meltable materials can be used as a binder in agglomeration.
1.2. Development of Melt Processes
The concept of using molten binding liquid for agglomeration instead of the conventional
binding liquid prepared from aqueous or organic solvent that has to be evaporated
off after agglomeration is completed was reported in some research
findings in the late 1970s and early 1980s (4–6). However, the concept was not widely
studied until more than a decade later and the period thereafter, using high-shear
mixer and fluidizing granulator as the processors for melt agglomeration (7–27).
High-shear mixer has gained greater popularity for use in melt agglomeration than
other equipment such as coating pan (4), drum granulator (28,29) or extruder
(30–34) for it provides intense shear forces from the high-speed rotation of the
impeller. The impeller mixing action facilitates a more homogenous distribution of
molten binding liquid, leading to the formation of agglomerates with narrower size
distribution and uniform drug content. In addition, the meltable binder can be
brought to melting by the frictional heat generated from the high shearing forces
of the impeller, without external heat supply (8). Highly spherical pellets can be produced
by a continuous agglomeration and rounding effect of the high-speed impeller
rotation in a high-shear mixer.
The characteristics of melt agglomerates such as size, size distribution, and
shape change progressively with variations in processing and formulation parameters.
The melt agglomeration process can be broadly classified as melt granulation
or melt pelletization, based on the physical properties of the formed agglomerates
(35). For simplicity of presentation, melt agglomeration will be used as the term
representing both melt pelletization and melt granulation, and melt agglomerates
as the term for their resultant products in this discussion.
1.3. Requirement of Melt Agglomeration
Melt agglomeration is carried out by adding the binder to the fine solid particles
either in the form of a molten liquid or in the form of a solid that melts during
the process. Generally, an amount of 10–30% w/w of binder, with respect to that
of fine solid particles, is used. The melting temperature of solid binder can be
achieved by external heat supply or heat generated from interparticulate friction during
high-speed mixing in the processor. A meltable binder suitable for melt agglomeration
has a melting point typically within the range of 50–100
C. Meltable binders
with melting point lower than 50
C are generally unsuitable as the end products are
liable to melting, softening, or sticking during handling and storage. On the other
hand, binders of high melting temperatures are not desirable due to greater risks
of thermal instability to the drugs used when very high heat is required for the melting
of the binders. The meltable binder can be classified as hydrophilic or hydrophobic.
Examples of hydrophilic binders include polyethylene glycols and poloxamers.
Hydrophobic binders include fatty acids, fatty alcohols, waxes, and glycerides.
Hydrophilic meltable binders are used to prepare immediate-release dosage forms
while the hydrophobic meltable binders are preferred for prolonged-release formulations.
One or more meltable binders may be employed in the formulation of a batch
of melt agglomerates. The binding action of meltable binders can be manifested
through liquid bridges formed by the molten liquid. Alternatively, a softened
386 Wong et al.
© 2005 by Taylor & Francis Group, LLC
semisolid can act as a binder (6,36). The local melting zone on the surface of a
semisolid binder could promote particle aggregation and subsequently agglomerate
growth. Nonetheless, a higher binder content up to 50% w/w was possible as there
was less binding liquid available for agglomerative action for a given weight of meltable
binder (6). Furthermore, the process of melt agglomeration is less sensitive
to the level of binder when softened semisolid is used as a binder. Frequently, the
process of agglomeration which employs softened semisolid binder is termed as
thermoplastic agglomeration.
The fine solid particles can be organic or inorganic in origin and in melt
agglomeration studies, inert substances commonly employed are lactose and dicalcium
phosphate. The melting point of fine solid particles should be at least 20
C
higher than that of the maximum processing temperature. This prevents excessive
softening of the solid particles as they form the support for the molten binding liquid
during the building-up process of agglomerate structure. The fine solid particles may
be constituted by one or more components. The physicochemical characteristics of
both meltable binders and nonmeltable fine solid particles have crucial effects on
the outcome of a melt agglomeration and will be discussed in Section 3.
1.4. Advantages of Melt Agglomeration
The growing interest in melt agglomeration has largely been attributed to various
advantages of melt agglomeration which are not attainable using the conventional
wet agglomeration techniques. With an appropriate selection of meltable binders,
both immediate-and prolonged-release agglomerates can be prepared using a onestep,
one-processor approach (14,37–42). Melt agglomeration is applicable for processing
of water-sensitive materials such as effervescent excipients and hygroscopic
drugs, thus omitting the use of either aqueous or organic solvent (43,44). With melt
methods, organic solvent, flame-proof facilities, and solvent recovery equipment are
not required, giving rise to lower cost of operation, and reduced ecological and toxicological
hazards. In addition, the formative process of melt agglomerates does not
require a drying phase and this shortens the total processing time. Melt agglomeration
is an ideal model for the elucidation of an agglomeration process as the process
of particle binding is not complicated by an evaporation of molten binding liquid
unlike the aqueous and nonaqueous wet agglomeration. Moreover, the physicochemical
properties of the molten binding liquid are potentially modifiable by incorporating
an additive (45).
Despite the various advantages offered by melt agglomeration, the technique is
not suitable for processing of heat-labile materials due to the involvement of elevated
temperatures. Nonetheless, successful agglomeration of volatile substances with
almost complete prevention of material loss by vaporization has been reported (46).
The probable reason is that the agglomerate surfaces was covered by layers of solidified
molten binder and loss of volatile substances was further avoided since the
melt process was in a closed chamber. In addition, Lactobacillus acidophilus bacteria
were found to have a higher survival rate in formulations prepared by continuous
melt technology than in formulations prepared using conventional wet techniques,
in spite of the molten mass’ temperature rising to more than 100
C (47). Clearly,
the detrimental effects of heat were considerably reduced by the absence of moisture
in melt agglomeration.
The growth processes of melt agglomerates are highly sensitive to formulation,
processing, and equipment variables. Uncontrollable melt agglomeration is
Melt Granulation and Pelletization 387
© 2005 by Taylor & Francis Group, LLC
frequently encountered when such variables are not optimally adjusted. In view of
challenges in controlling the melt agglomeration process, melt agglomeration has
remained an interesting field of research with the potential for practical and wider
application in the pharmaceutical industry.
1.5. Characterization of Melt Agglomerates
Over the past decade, many researchers have examined the effects of formulation,
processing, and equipment variables on melt agglomeration process through physicochemical
characterization of the formed agglomerates. This has contributed to
a better understanding of the formation and growth processes of melt agglomerates.
Similar to agglomerates produced by wet techniques, the most common method
employed for the characterization of melt agglomerates is the determination of mean
agglomerate size and size distribution by sieving. The size and size distribution of
melt agglomerates generally fit a log-normal mathematical model. They can be represented
by geometric weight mean diameter (dgw) and geometric standard deviation
(sg), respectively, as described by Sch?fer and Worts (48):
log dgw ? Pwi  log di
Pwi ?1:1?
log Sg ? ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffififfiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pwi?log di?2  ?Pwi  log di?2
Pwi s ?1:2?
where di is the mean diameter and wi is the weight of agglomerates of sieve
fraction i.
Alternatively, the size and size distribution of melt agglomerates can be
expressed as the model independent mass median diameter and span, respectively.
The mass median diameter is defined as agglomerate size corresponding to the
50th weight percentile of the cumulative agglomerate size distribution and span
being the quotient of difference between the agglomerate sizes corresponding to
the 90th and 10th weight percentiles of cumulative size distribution to mass median
diameter. The percentages of fines and lumps, generally described with respect to
agglomerate fractions smaller than 90–250 mm and larger than 2000–4000 mm,
respectively, are also reported as an indirect quantitative input on size distribution
of melt agglomerates.
Apart from size and size distribution of melt agglomerates, shape, surface morphology,
specific surface area, moisture content, density, porosity, tensile strength,
friability, flow, and packing properties have often been determined in elucidation
of the mechanism and kinetics of melt agglomeration. The distribution of binder
content in melt agglomerates has a strong bearing on the homogeneity of melt
agglomerate growth. The actual binder content in melt agglomerates has also been
measured directly through experiments or estimated indirectly via the application
of mathematical modeling. A direct measurement of polyethylene glycol content
was determined using near-infrared spectrophotometry (20). The content of binder
in melt agglomerates was also determined with the aid of the high-performance
liquid chromatography analytical method (49). The polyethylene glycol content
was indirectly measured from the true density values of milled agglomerate fraction
determined by means of a gas displacement pycnometer (50) and the polyethylene
388 Wong et al.
© 2005 by Taylor & Francis Group, LLC
glycol content was calculated using Eq. 1.3 as follows:
?1  x?
rl ?
x
rb ?
1
ra ?1:3?
where x is the fraction of polyethylene glycol, rl, rb, and ra are true density values of
lactose, binder, and melt agglomerates, respectively. The content of polyethylene glycol
can also be determined through gravimetric analysis whereby the weight of binder
is obtained by weight subtraction of nonbinder fraction from that of the melt
agglomerates (51,52). In the case of complexation and crystallite behavior of drug,
binder, and fine solid particles, the melt agglomerates have often been subjected to
evaluation by differential scanning calorimetry and x-ray diffractometry. In vitro
dissolution studies have been conducted to examine the drug release characteristics
of melt agglomerates. El-Shanawany (53) and Voinovich et al. (39) found that the
in vivo drug release attributes of melt agglomerates were in accordance with the
in vitro findings. Nonetheless, a quantitative relationship between the in vivo and in
vitro performances of melt agglomerates is still generally lacking.
Another important feature of melt agglomerates is their performance stability
during storage. It was found that the melt agglomerates could maintain their dissolution
profile after a year of storage at 25
C and 60% relative humidity (40). The
release of sulfamethazine from tablets made of melt granules remained unchanged
after 2 years of storage at 25
C in closed containers (37). Nevertheless, the rate
and extent of drug release were found to increase with agglomerates stored at an
elevated temperature of 40
C and at a relative humidity of 75%, probably owing
to the softening of the melt matrices (41,54). On the basis of the meltable nature
of agglomerates, stability testing is preferred to be conducted at an appropriate
choice of storage temperature which will not bring about marked alterations to
the physicochemical state of the agglomerates.
2. MECHANISM OF MELT AGGLOMERATION
The mechanism of melt agglomeration is similar to that of wet agglomeration,
except that the formation and growth processes of melt agglomerates are not complicated
by binding liquid losses via evaporation during the agglomeration process.
The growth process of melt agglomerates is dependent on the interplay between the
size enlargement and size reduction processes. The likelihood of an agglomerate to
grow in size or experience breakages is a result of the balance between externally
applied mechanical forces and agglomerate strength. The agglomerates will grow
in size if they have sufficient strength to withstand the impact of externally applied
forces or vice versa. The strength of agglomerates is affected by the relative magnitude
of capillary, frictional, and viscous forces. The capillary forces aid agglomerate
consolidation by pulling the solid particles together while the viscous and
frictional forces resist both consolidation and dilation of the solid particle assembly.
The mechanisms involved in the formation and growth of melt agglomerates
have been categorized into different stages and will be discussed in the following
sections.
Melt Granulation and Pelletization 389
© 2005 by Taylor & Francis Group, LLC
2.1. Stages of Agglomerate Growth
The mechanism of wet agglomeration has traditionally been subdivided into nucleation,
coalescence, layering, abrasion transfer, crushing, and other concomitant
events such as snow balling and onion skinning, based mainly on the elementary
growth mechanism suggested by Sastry and Fuerstenau (Fig. 1) (55). An alternative
description of the agglomeration mechanism has also been proposed (2,56). The process
of agglomeration consists of a combination of three phases: wetting and nucleation,
consolidation and growth, together with the steps of attrition and breakage
2.1.1. Wetting and Nucleation
Nucleation is the initial phase of agglomeration in which nuclei or small agglomerates
of loose and porous structure are formed after the primary particles are wetted
by a binding liquid droplet. The primary particles are bound by liquid bridges in the
Figure 1 Elementary agglomerate growth mechanisms. (Adapted from Ref. 55.)
390 Wong et al.
(Fig. 2).
© 2005 by Taylor & Francis Group, LLC
pendular state. The degree of liquid saturation of an agglomerate can be increased
either by continuous addition of liquid or through the consolidation of the agglomerate.
Two nucleation mechanisms, namely immersion and distribution, are proposed
by Sch?fer and Mathiesen (50), based on the process of melt agglomeration
agglomeration is a function of the ratio between the sizes of primary particles and
molten binder droplets (25,50,52).
Nucleation by immersion occurs when the size of the molten binder droplets is
greater than that of the fine solid particles. Immersion proceeds by the deposition of
fine solid particles onto the surfaces of molten binder droplets. The propensity of
nucleation by immersion method is promoted by large binder droplet size, high binder
viscosity, and low shearing forces. High binder viscosity and low shearing forces
reduce the opportunity of molten binder droplets to break down, thus keeping their
size comparatively large and several times the size of the fine solid particles.
In nucleation by distribution method, a molten binding liquid is distributed
onto the surfaces of fine solid particles. The nuclei are formed by the collision
between the wetted particles. The formed nuclei have a loose structure with
entrapped air unlike those produced by the immersion method. Generally, small binder
droplet size, low binder viscosity, and high shearing forces are favorable conditions
for nucleation by the distribution method.
The nucleation phase is characterized by the disappearance of fines (35), as
a consequence of coalescence between the wetted primary particles or the primary
particles with the formed nuclei. The resultant nuclei would undergo consolidation
under the impact of the externally applied mechanical forces and acquire sufficient
Figure 2 A recent approach in describing the agglomeration mechanism. (Adapted from
Ref. 2.)
Melt Granulation and Pelletization 391
(Fig. 3). The dominance of either mechanism in the nucleation process of melt
© 2005 by Taylor & Francis Group, LLC
strength to resist further breakdown by impact forces and will be able to grow into
bigger agglomerates.
2.1.2. Consolidation and Growth
Agglomerate growth refers to progression in the size of the formed nuclei, typically
by binary coalescence and layering. The number of nuclei is progressively reduced
with an increase in size of the resultant agglomerates by coalescence. In contrast,
the number of agglomerates remains largely unchanged when size enlargement proceeds
by layering. The layering growth stage generally takes place after the agglomerates
have attained a certain size and rigidity and is associated with the reduced rate
of coalescence (57).
The probability of successful fusion between the collided nuclei is dependent on
the liquid saturation state of nuclei. During the process of agglomeration, the agglomerates
are consolidated by the agitation forces. The reduction in agglomerate pore size
and number promotes the migration of binding liquid from agglomerate core to surfaces,
thus enhancing the surface plasticity and propensity of agglomerate growth by
coalescence. The rate and extent of consolidation of an agglomerate is governed by
interparticulate frictional, capillary, and viscous forces. The interparticulate frictional
and viscous forces resist the consolidation process of agglomerates. Conversely, the
capillary forces promote the consolidation of agglomerates by pulling the particles
together (58). The net extent of consolidation for an agglomerate is dependent on
the relative magnitude of these forces, which is a function of agglomerate size, viscosity,
and surface tension of binding liquid, as well as the physicochemical properties of fine
solid particles (17,20,45,52,59,60).
The degree of liquid saturation of an agglomerate may be increased through
consolidating the agglomerates by agitation. Alternatively, it is attainable by adding
an additional volume of binding liquid to the existing agglomerates, making the liquid
bridges of nuclei progress from pendular to funicular and followed by the capillary
state where the level of liquid saturation can be as high as 80–100%. The degree of
Figure 3 Modes of nucleation mechanism: (A) distribution and (B) immersion. (Adapted
from Ref. 50.)
392 Wong et al.
© 2005 by Taylor & Francis Group, LLC
liquid saturation reaches and probably exceeds 100% prior to agglomeration when a
cohesive powder, highly viscous binding liquid, or an excessive amount of binding
liquid is employed. At the droplet state when excessively high liquid saturation is
attained, there is considerable risk of agglomerate overwetting and uncontrollable
agglomerate growth may result. In contrast to the conventional wet agglomeration
processes, the level of liquid saturation of melt agglomerates cannot be reduced via
the evaporation of solvent. The effects of excess surface wetness are counteracted
by the application of a very high impeller speed which aids in the dispersion of lumps
and large agglomerates or through the use of porous fine particles which could
accommodate binding liquid in pores and crevices, away from the agglomerate
surfaces.
2.1.3. Attrition and Breakage
Attrition and breakage refer to the phenomenon of agglomerate fragmentation in the
are solidified by tray cooling to ambient temperature without the need for drying by
a tumbling process. Consequently, breakage is known to have a more essential role
in affecting the resultant properties of the melt agglomerates during the agglomerative
phase.
The propensity of agglomerate breakage is largely governed by the strength of
the agglomerates. Rumpf (61) described the strength of moist agglomerates in their
static funicular and capillary states as follows:
st ? SC
1  e
e
g
d
cos y ?2:1?
where st is the tensile strength, S is the liquid saturation, C is a constant, e is the
agglomerate porosity, g is the surface tension of binding liquid, d is the surface mean
particle diameter, and y is the contact angle between the solid particles and the binding
liquid. The strength of moist agglomerates is promoted by an elevation in surface
tension of the binding liquid, liquid saturation, and wetting state of solid particles, as
well as a reduction in agglomerate porosity and mean size of solid particles.
The application of a binding liquid with low viscosity and surface tension
values increases the propensity of agglomerates to breakage (17). This deters the
agglomerates from consolidation. The regular breakdown of agglomerates aids in
the distribution of binding liquid across the solid particle bed ensuring homogenous
agglomerate growth. Nevertheless, it prevents a complete buildup of structure which
is needed for size enlargement process. In the case of spheronization, the regular
breakdown of agglomerates is translated to the absence of a plastically deformable
coalesced structure which is needed for rounding through shaping of agglomerate
surfaces by particle rearrangement.
3. FACTORS AFFECTING MELT AGGLOMERATION
The formation and growth processes of melt agglomerates are governed by formulation,
processing, and equipment variables to a greater extent than those of conventional
wet agglomerates as a result of the involvement of high impeller speeds and
high shearing forces for agglomeration and viscous molten binding liquid. Many
Melt Granulation and Pelletization 393
© 2005 by Taylor & Francis Group, LLC
dry and wet states respectively (2). In melt agglomeration, the formed agglomerates
researchers had studied the influences of these variables, through the characterization
of melt agglomerates as previously illustrated, for a better understanding and
control of the melt agglomeration processes.
The effects of processing variables such as mixing time, mixing speed, mixer
load, jacket temperature, and method of binder addition on melt agglomeration have
been extensively investigated (8,11,13,43,52,62–67). Typically, an increase in mixing
time or mixing speed promotes agglomerate growth through squeezing the molten
binding liquid from agglomerate core to surfaces by means of a densification process,
thus increasing the degree of liquid saturation of agglomerates and their propensity
to grow by binary coalescence following the collision between two or more plastically
deformable surfaces. Under the continuous stirring action of impeller rotation,
the deformable surfaces of these agglomerates can be rounded via intra-agglomerate
rearrangement of particles leading to the formation of spherical pellets.
The influences of equipment variables on melt agglomerate growth are more
marked with high-shear mixers than with low-shear mixers and fluid bed granulators.
One main reason is that the intensity of shearing forces is greater in high-shear
mixers. The high shearing forces promote a more even distribution of molten binding
liquid making the size distribution of melt agglomerates becoming narrower. The
level of shearing forces generated in a high-shear mixer is dependent on the geometry
of impeller blade, design of processing chamber, and relative dimension between the
blade and chamber. The construction of both impeller blade and processing chamber
can have a significant impact on the flow pattern of processing material. The use of a
truncated cone-shaped lid (11) and an inner wall lining made of polytetrafluoroethylene
(68) reduced the adhesion of mass undergoing agglomeration onto the wall of
the processing chamber. Variation in the curvature of impeller blade and the distance
from its base to the floor of the chamber likewise brought about changes in the flow
pattern of the wet mass (12,18). The pattern of material flow is related to the
mechanical force distribution and homogeneity of melt agglomeration. It has
a strong bearing on the disparity in size and shape of the formed agglomerates.
In the case of formulation variables, the effects of size, size distribution, shape,
density, and packing properties of fine solid particles have been reported (10,69,70).
The use of solid particles of mean size smaller than 10 mm is usually problematic in
melt agglomeration because a very high level of liquid saturation is needed to overcome
the high agglomerate strength resulting from the cohesiveness of small particles
in order to provide sufficient agglomerate deformability for growth. This, in turn,
can lead to a potentially uncontrollable melt agglomerative process. Generally, solid
particles with size ranges between 20 and 25 mm are preferable for the production of
melt agglomerates (10). Excessively large solid particles, with a mean size of over
100 mm, produce mechanically weak agglomerates or fines. The excessively large or
small solid particles will tend to produce irregularly shaped melt agglomerates as
such aggregates are susceptible to breakdown under the influence of shear forces
or have rigid particle linkages within the agglomerate which inhibit particle deformation
and rearrangement. The effects of size, size distribution, shape, density, and
packing property of fine solid particles on melt agglomeration are interdependent.
Size, size distribution, shape, and density affect the packing geometry of solid particles,
bound by molten binding liquid, within the agglomerate. The strength of particulate
interlocking, state of liquid distribution, and saturation of agglomerates can
be altered via the modification in these properties of fine solid particles.
The effects of binder volume, binder rheology, binder surface property, and
binder particle size on melt agglomeration have largely been reported in relation
394 Wong et al.
© 2005 by Taylor & Francis Group, LLC
to the influences of other formulation or processing variables (15,17,20,36,45,70). The
growth of melt agglomerates is promoted predominantly by an increase in viscosity,
tack, and specific volume as well as a decrease in surface tension of the molten binding
liquid. The viscosity, tack, specific volume, and surface tension govern the intraagglomerate
mobility of molten binding liquid and state of liquid saturation of melt
agglomerates. The influences of viscosity, tack, and surface tension of molten binding
liquid on melt agglomeration are affected by product temperature, mixing speed, and
physicochemical properties of the fine solid particles. The effect of binder particle size
on melt agglomeration is less marked except when high-viscosity binder is used and at
early agglomerative phases (50). At the early agglomerative phase, the dimension of
droplets of a very viscous molten binding liquid can remain large and unreduced by
the high shearing forces, resulting in larger agglomerates for further growth.
4. CONTROL OF MELT AGGLOMERATION
The growth processes of melt agglomerates are sensitive to formulation, processing,
and equipment variables. A better understanding of the mechanism of melt agglomerate
formation aids in the development of methods to control and monitor the inprocess
changes of product attributes. In conventional wet agglomeration process,
the instrumentation of equipment to monitor and control the operation has been studied
by several investigators. The incorporation of sensors and suitable instrumentation
of the equipment enables changes in agglomerate properties during the
agglomeration process to be monitored such that the end point of the process can
be accurately predicted.
4.1. Image Processing/Infrared Spectroscopy/Bed Height
In fluid bed agglomeration process, the control of agglomerate growth through monitoring
agglomerate moisture content using infrared spectroscopy (71–73) or agglomerate
size and shape using an image processing system (74,75) has been investigated.
The application of bed height control to monitor agglomerate growth in fluid bed
agglomeration process has also been studied (76,77). Of all instrumentation techniques,
the image processing system can potentially be adopted as a tool to control the
melt agglomeration process. Such a system requires a heat-resistant optic device capable
of withstanding the processing temperatures associated with the melt agglomeration
process of up to as high as 120
C. The infrared spectroscopy or equivalent
techniques are less suitable for use in monitoring the melt agglomerate growth process.
Unlike the wet agglomeration process, the meltable binder is incorporated with
the processing material prior to mixing and shearing in the melt processor to bring
the binder to its melting temperature. Practically, there is no need for monitoring the
changes of binding liquid content added. Melt agglomeration in high-shear mixer
proceeds by the active centrifugal massing of the processing materials. In contrast
to the fluid bed agglomerator, the bed height of powder particles in the processing
chamber is less affected by gravitational forces and the upward air flow velocity
required for particle suspension. For a given material load, the bed height is expected
to have minimal effects on the changes in product attributes. It is envisaged that the
measurement of bed height would not be useful in reflecting the in-process product
attributes.
Melt Granulation and Pelletization 395
© 2005 by Taylor & Francis Group, LLC
4.2. Torque/Current/Power Consumption
For wet agglomeration in high- or low-shear mixers, the popular method for instrumentation
involves an indirect measurement based on the changes in rheological
properties of moistened mass which are related to the growth propensity of agglomerates.
The process of extrusion in extrusion–spheronization had been monitored
using the measurement of power consumption of the extruder motor or extrusion
force on the screen (78–84). Low-shear mixers had employed torque (85–90), power
consumption (85,91–93), or current (94) measurements as the tool for end-point
determination. The instrumentation technique for high-shear mixers involved the
measurement of torque on impeller shaft (95,96), power consumption (80,95–110),
current (111,112) or motor slip (113–116). Technically, power and current consumption
are the simpler techniques for monitoring the agglomerate growth in a highshear
mixer. It was easier to equip a high-shear mixer with a power consumption
meter than a sensor for torque or slip measurement (117). Torque generated had
been reported to provide a more descriptive profile of the agglomeration process
than power consumption. Nevertheless, there were only small differences between
the two methods for process monitoring (95,96,117). The measurement of torque
is a more expensive method and technically more difficult due to problems associated
with the transmission of signals from the sensor (118). Holm et al. (98) showed that
the power consumption profile of an agglomeration process was related to plasticity
and liquid saturation of the processing material in addition to that of the interparticulate
friction within the agglomerates. Since plastic deformation is important for
agglomerate growth, power consumption may then be related to the agglomeration
propensity.
A slow and continuous binding liquid addition is a prerequisite for obtaining
a power consumption that is suitable for use in process control of wet agglomeration
the first phase, a wetting of powder mixture takes place without any observable rise
in the power consumption.
The subsequent addition of binding liquid brings about particle agglomeration
and a sharp rise in the magnitude of power consumption. The power consumption of
the agglomeration process levels off as the agglomerates gradually become dense and
saturated with the binding liquid. During the fourth phase, the power consumption
rises due to overwetting and excessive growth of agglomerates by coalescence. The
power consumption falls in the fifth phase and it is associated with the formation
of a suspension upon the addition of an excessive amount of the binding liquid.
The optimum end point of the process lies within the third phase of the power consumption
tracing. In the case of melt agglomeration process, the addition of meltable
binder is almost instantaneous. This lack of the typical power consumption profile
makes power consumption measurement less suitable for monitoring melt agglomeration
end point as in the case of the wet agglomeration process. In laboratory
high-shear mixers, an increase in melt agglomerate size is not reflected in the power
consumption signals except at an excessively high impeller speed (11,69). The
deformability of the agglomerates produced by a molten binding liquid is relatively
low in comparison to that of agglomerates formed when using an aqueous binding
liquid. The sensitivity of power consumption measurement is not sufficiently discriminating
to reflect small changes in the plasticity of melt agglomerates.
During the premelt agglomeration phase of the powder mixture, the level of
396 Wong et al.
(118). Five characteristic phases of power consumption are identifiable (Fig. 4). In
power consumption is low and with minimal fluctuations in its magnitude (Fig. 5).
© 2005 by Taylor & Francis Group, LLC
The magnitude of the power consumption falls to a value slightly below that of the
values obtained during the premelt agglomeration phase when the binder particles
begin to melt upon the influence of frictional heat generated by the impeller rotation
(9). Such observation could be related to the lubricating effect of molten binding
liquid on the solid particles undergoing agglomeration. The power consumption rises
sharply following the complete melting of the binder particles. The molten binding
liquid is distributed rapidly throughout the powder mixture. The level of power consumption
becomes lower upon the complete distribution of the molten binding liquid
and increases as the agglomerate growth takes place.
Generally, a higher level of power consumption is recorded when a higher
impeller speed or a more viscous molten binding liquid is employed for the preparation
of melt agglomerates. The rise in power consumption is delayed when a viscous
meltable binder is used (119). The growth process of melt agglomerates is not easily
interpretable from the record of power consumption when the growth propensity of
melt agglomerates is strongly governed by the viscosity of the molten binding liquid.
This is ascribed to greater difficulties in distributing viscous molten binding liquid
and a slower rate of agglomerate densification. A viscous binder hinders the consolidation
of agglomerates as a viscous liquid resists plastic deformation and its migration
from core to the surface of the agglomerate during consolidation. The power
consumption measurements are not directly comparable between high-shear mixers
of different processing capacities. However, the values of power consumption can be
correlated to the changes in size of the agglomerates produced using a large-scale
high-shear mixer, albeit such relationship is not attainable when using a laboratory
scale high-shear mixer (12).
Figure 4 Power consumption profile of a wet agglomeration process. (Adapted from
Ref. 118.)
Melt Granulation and Pelletization 397
© 2005 by Taylor & Francis Group, LLC
4.3. Energy Consumption
The applicability of both processing power and energy consumption as tools to
monitor agglomerate growth in a melt agglomeration process has been studied
(19,65). Postmelt specific power consumption, postmelt specific energy consumption,
and average postmelt specific power consumption were computed. The postmelt specific
power consumption was calculated by dividing the postmelt power consumption
read at specific postmelt processing time, with the total weight of processing material.
The postmelt specific energy consumption was obtained by integrating the area
of postmelt specific power consumption against postmelt processing time curve. The
average postmelt specific power consumption was represented by the quotient of
postmelt specific energy consumption to that of the corresponding length of postmelt
processing time.
4.3.1. Melt Agglomerate Size
The increase in the melt agglomerate size was more closely correlated to the postmelt
specific energy consumption than that of the average postmelt specific power consumption
or postmelt specific power consumption. This is because specific energy
consumption can be considered as the amount of work done for the formative
process of consolidating the primary particles into melt agglomerates. It is more indicative
of the stage of melt agglomerate growth. The postmelt specific energy
consumption and postmelt processing time were linearly correlated and the formation
of larger melt agglomerates at longer processing time involved a greater level
Figure 5 Profile of power consumption at premelt and postmelt phases. (Adapted from
Ref. 19.)
398 Wong et al.
© 2005 by Taylor & Francis Group, LLC
of energy consumption. The postmelt specific energy consumption and melt agglomerate
size were linearly related and such a relationship was not dependent on the effect
of impeller speed except that of the binder concentration and size of solid particles. In
relating the influences of binder concentration and size of solid particles on the outcome
of agglomeration with respect to postmelt specific energy consumption, an
equation was established. The processes of melt agglomerate formation and growth
were described using a modified macroscopic population model. The model of melt
agglomeration, under the specified processing conditions, was summarized as:
d ? f 1?s??1 ? f 2?c?Emelt ?4:1?
where d refers to mean melt agglomerate size, Emelt is the postmelt specific energy
consumption, f1(s) and f2(c) represent polynomial relationships with size of solid
particles and binder concentration as independent variables, respectively. Using
Eq. 4.1, the theoretical value of melt agglomerate size was found in good agreement
with the experimental data obtained from melt agglomeration runs employing various
binder concentrations and different sizes of primary particles. Nonetheless, it
should be emphasized that the application of postmelt specific energy consumption
for the control of melt agglomeration process would be inappropriate in processes
encountering an excessive level of material adhesion or when large amounts of
lumps are formed. The effective weight of processing material is markedly reduced
in the presence of an excessive level of material adhesion. This will give rise to inaccuracies
in computing the postmelt specific energy consumption level. With the
excessive formation of lumps, the rope-like flow pattern of the agglomerates will
be affected. The movement of agglomerates, bound within the processing chamber,
is expected to be variable. As a result, the predictability of the melt agglomeration
process using postmelt specific energy consumption will be substantially reduced by
the changes in product attributes from one batch to another. Practically, Eq. 4.1 is
useful for predicting the end point of an optimal or near-optimal melt agglomeration
run.
Frequently, the end point of an agglomeration process is governed by the size
of the formed agglomerates. The geometric-weight mean diameter of melt agglomerates
is calculated using Eq. 1.1 described by Sch?fer and Worts (48). In the case of
melt agglomerates with size distribution which cannot be fitted into a log-normal
relationship, the mass median diameter of agglomerates shall be adopted for representing
of the size of melt agglomerates produced.
4.3.2. Melt Agglomerate Shape
The shape and tensile strength of melt agglomerates may be used to characterize the
agglomeration process besides the more generally used size of melt agglomerates.
Similar to the measurement of size, both shape and tensile strength can be determined
using quick, simple, and accurate methods. The sphericity of melt agglomerates
is best evaluated using the image analysis system. The image analyzer consists of
a computer system connected to a video camera mounted on a stereomicroscope.
The average projected area and perimeter of melt agglomerates are determined from
the digitized images of agglomerates. By transformation of these basic parameters,
melt agglomerate sphericity is calculated. The sphericity value gives a measure of
melt agglomerate roundness. A sphericity value, determined by Eq. 4.2, of unity
Melt Granulation and Pelletization 399
© 2005 by Taylor & Francis Group, LLC
describes a perfect circle. The agglomerate sphericity is defined as:
Sphericity ?
4p  Area
?Perimeter?2 ?4:2?
The melt agglomeration process can be divided into two spheronization phases,
a fast initial rate followed by a slow rate (65). The change in spheronization rate was
associated with that of agglomerate size, porosity, and flow pattern of the processing
material. The study established a biexponential mathematical model to relate the
agglomerate sphericity with postmelt specific energy consumption. The relationship
of agglomerate sphericity with postmelt specific energy consumption was found to be
independent of the effect of various production variables, such as postmelt impeller
speed, binder concentration, and size of solid particles. The biexponential mathematical
model has an equation as follows:
1  Sphericity ? ?0:4107 exp?0:0118Emelt? ? ?0:0533 exp?0:0004Emelt?
?4:3?
Overall, the characterization of the melt agglomeration end point using the
biexponential mathematical model is simple and without the need to consider the physicochemical
properties of the processing materials used. As the process of agglomeration
is aimed at making highly spherical pellets, predictive methods based on
sphericity will help to capture the ideal end point together with predictive methods
based on size.
4.3.3. Melt Agglomerate Tensile Strength
The formative process of melt agglomerates is accompanied by changes in the size,
shape, and porosity of melt agglomerates (19,65,120). The formation of a less porous
melt agglomerate is related to an increase in the size and level of sphericity of the
product. This is attributed to an enhanced molten binding liquid migration from
core to surfaces of agglomerates upon densification, which promotes rounding and
binary growth of melt agglomerates as previously described. The application of
agglomerate porosity as an in-process control parameter is not practical as such a
measurement requires a long experimental time and the accuracy of such a measurement
is subjected to the skill of an operator. The tensile strength is an indirect
measurement of the porosity of melt agglomerates. It has an inverse relationship
with the porosity of melt agglomerates. The tensile strength of melt agglomerates
can be determined using a tensile tester. The crushing force is usually applied at a
fixed rate. The load required to crush each melt agglomerate is recorded. The evaluation
of tensile strength of melt agglomerates is relatively simple and it is useful as an
in-process control parameter for melt agglomeration processes.
5. CONCLUSIONS
Over the past decade, the properties of processing materials and their effects on melt
agglomeration have been extensively investigated. Principally, solid particle size, size
distribution, shape, density, packing property, as well as solid binder particle size,
molten binding liquid viscosity, molten binding liquid surface tension, and molten
binding liquid tack have their individual influences on the size, shape, and porosity
of melt agglomerates produced. In the case of equipment and processing variables,
400 Wong et al.
© 2005 by Taylor & Francis Group, LLC
off-bottom clearance, impeller geometry, construction characteristics of the processing
chamber, nature of wall lining, impeller speed, jacket temperature, mixing time,
and mixer load are important variables affecting the kinetics of melt agglomeration
and characteristics of the formed agglomerates. The reproducibility in control of a
melt agglomeration process is strongly dependent on the rigid surveillance of all
critical formulation, processing, and equipment variables. Apart from the tight
control of formulation, processing, and equipment parameters, the application of
in-process monitoring tool and predictive mathematical model are imperative to
avoid any uncontrollable melt agglomeration.
REFERENCES
1. Tardos GI, Khan MI, Mort PR. Critical parameter and limiting conditions in binder
granulation of fine powders. Powder Technol 1997; 94:245–258.
2. Iveson SM, Litster JD, Hapgood K, Ennis BJ. Nucleation, growth and breakage
phenomena in agitated wet granulation processes: a review. Powder Technol 2001; 117:
3–39.
3. Ghebre-Sellassie I. Mechanism of pellet formation. Ghebre-Sellassie I, ed. Pharmaceutical
Pelletization Technology. Vol. 37. New York: Marcel Dekker, Inc., 1989:129.
4. Rubinstein MH. A new granulation method for compressed tablet. J Pharm Pharmacol
1976; 28:67.
5. Rubinstein MH, Musikabhuma P. Formulation and evaluation of paracetamol 500 mg
tablets produced by a new direct granulation method. Drug Dev Ind Pharm 1980;
6(5):451–473.
6. McTaggart CM, Ganley JA, Sickmueller A, Walker SE. The evaluation of formulation
and processing conditions of a melt granulation process. Int J Pharm 1984; 19:139–148.
7. Sch?fer T, Holm P, Kristensen HG. Melt granulation in a laboratory scale high shear
mixer. Drug Dev Ind Pharm 1990; 16(8):1249–1277.
8. Sch?fer T, Holm P, Kristensen HG. Melt pelletization in a high shear mixer. I. Effects
of process variables and binder. Acta Pharm Nordica 1992; 4:133–140.
9. Sch?fer T, Holm P, Kristensen HG. Melt pelletization in a high shear mixer. II. Power
consumption and granule growth. Acta Pharm Nordica 1992; 4:141–148.
10. Sch?fer T, Holm P, Kristensen HG. Melt pelletization in a high shear mixer. III. Effects
of lactose quality. Acta Pharm Nordica 1992; 4:245–252.
11. Sch?fer T, Taagegaard B, Thomsen LJ, Kristensen HG. Melt pelletization in a high
shear mixer. IV. Effects of process variables in a laboratory scale mixer. Eur J Pharm
Sci 1993; 1(3):125–131.
12. Sch?fer T, Taagegaard B, Thomsen LJ, Kristensen HG. Melt pelletization in a
high shear mixer. V. Effects of apparatus variables. Eur J Pharm Sci 1993; 1(3):
133–141.
13. Thomsen LJ, Sch?fer T, Sonnergaard JM, Kristensen HG. Prolonged release matrix
pellets prepared by melt pelletization. Part 1. Process variables. Drug Dev Ind Pharm
1993; 19(15):1867–1887.
14. Thomsen LJ, Sch?fer T, Kristensen HG. Prolonged release matrix pellets prepared by
melt pelletization. Part 2. Hydrophobic substances as meltable binders. Drug Dev Ind
Pharm 1994; 20(7):1179–1197.
15. Eliasen H, Sch?fer T, Kristensen HG. Effects of binder rheology on melt agglomeration
in a high shear mixer. Int J Pharm 1998; 176:73–83.
16. Eliasen H, Kristensen HG, Sch?fer T. Electrostatic during a melt agglomeration
process. Int J Pharm 1999; 184:85–96.
17. Eliasen H, Kristensen HG, Sch?fer T. Growth mechanisms in melt agglomeration with
a low viscosity binder. Int J Pharm 1999; 186:149–159.
Melt Granulation and Pelletization 401
© 2005 by Taylor & Francis Group, LLC
18. Heng PWS, Wan LSC, Wong TW. Effect of off-bottom clearance on properties of
pellets produced by melt pelletization. Pharm Dev Technol 1999; 4(1):27–33.
19. Heng PWS, Wong TW, Shu JJ, Wan LSC. A new method for the control of size of
pellets in the melt pelletization process with a high shear mixer. Chem Pharm Bull
1999; 47:633–638.
20. Wong TW, Wan LSC, Heng PWS. Effects of physical properties of PEG 6000 on pellets
produced by melt pelletization. Pharm Dev Technol 1999; 4(3):449–456.
21. Maejima T, Osawa T, Kobayashi M, Noda K. Factors effecting spherical granulation of
drugs by tumbling granulation method. Chem Pharm Bull 1992; 40(2):488–492.
22. Maejima T, Osawa T, Nakajima K, Kobayashi M. Preparation of spherical beads without
any use of solvents by a novel tumbling melt granulation (TMG) method. Chem
Pharm Bull 1997; 45:518–524.
23. Maejima T, Osawa T, Nakajima K, Kobayashi M. Effects of species of non-meltable
and meltable materials and their physical properties on granulatability in tumbling melt
granulation method. Chem Pharm Bull 1997; 45:1833–1839.
24. Maejima T, Kubo M, Osawa T, Nakajima K, Kobayashi M. Application of tumbling
melt granulation (TMG) method to prepare controlled-release fine granules. 1998;
46:534–536.
25. Abberger T, Seo A, Sch?fer T. The effect of droplet size and powder particle size on the
mechanisms of nucleation and growth in fluid bed melt agglomeration. Int J Pharm
2002; 249:185–197.
26. Kidokoro M, Haramiishi Y, Sagasaki S, Shimizu T, Yamamoto Y. Application of
fluidized hot-melt granulation (FHMG) for the preparation of granules for tableting;
properties of granules and tablets prepared by FHMG. Drug Dev Ind Pharm 2002;
28(1):67–76.
27. Seo A, Holm P, Sch?fer T. Effects of droplet size and type of binder on the agglomerate
growth mechanisms by melt agglomeration in a fluidized bed. Eur J Pharm Sci 2002;
16(3):95–105.
28. Pataki K, Horva?th E, Ormo? s Z. Rolling bed granulation with local melt forming.
Proceedings of 4th Conference on Applied Chemistry, Unit Operations and Processes,
Vol. 3. Veszprem, Hungary. 1983:361–362.
29. Litster JD, Sarwono R. Fundamental studies of fluidized drum granulation. Proceedings
of the 1st International Particle Technology Forum, Denver, CO, Part l, Aug
17–19, 1994; 265–270.
30. Edimo A, Leterme P, Denis J, Traisnel M, Gayot AT. Capacity of lipohilic auxiliary
substances to give spheres by extrusion-spheronization. Drug Dev Ind Pharm 1993;
19(7):827–842.
31. Miyagawa Y, Okabe T, Yamaguchi Y, Miyajima M, Sato H, Sunada H. Controlledrelease
of diclofenac sodium granule. Int J Pharm 1996; 138:215–224.
32. Zhang F, McGinity JW. Properties of sustained release tablets prepared by hot-melt
extrusion. Pharm Dev Technol 1999; 4:241–250.
33. Liu J, Zhang F, KcGinity J. Properties of lipophilic matrix tablets containing phenylpropanolamine
hydrochloride prepared by hot-melt extrusion. Eur J Pharm Biopharm
2001; 52:181–190.
34. Young CR, Koleng JJ, McGinity JW. Production of spherical pellets by a hot-melt
extrusion and spheronization process. Int J Pharm 2002; 242:87–92.
35. Sch?fer T. Growth mechanisms in melt agglomeration in high shear mixers. Powder
Technol 2001; 117:68–82.
36. Evrard B, Amighi K, Beten D, Delattre L, Moe?s AJ. Influence of melting and rheological
properties of fatty binders on the melt granulation process in a high-shear mixer.
Drug Dev Ind Pharm 1999; 25(11):1177–1184.
37. Evrard B, Delattre L. In vitro evaluation of lipid matries for the development of
a sustained-release sulfamethazine bolus for lambs. Drug Dev Ind Pharm 1996;
22(2):111–118.
402 Wong et al.
© 2005 by Taylor & Francis Group, LLC
38. Zhou F, Vervaet C, Remon JP. Matrix pellets based on the combination of waxes,
starches and maltodextrins. Int J Pharm 1996; 133:155–160.
39. Voinovich D, Moneghini M, Perissutti B, Filipovic-Grcic J, Grabnar I. Preparation in
high-shear mixer of sustained-release pellets by melt pelletisation. Int J Pharm 2000;
203:235–244.
40. Passerini N, Albertini B, Gonza?lez-Rodr??guez ML, Cavallari C, Rodriguez L. Preparation
and characterization of ibuprofen-poloxamer 188 granules obtained by melt granulation.
Eur J Pharm Sci 2002; 15:71–78.
41. Hamdani J, Moes AJ, Amighi K. Development and evaluation of prolonged release
pellets obtained by the melt pelletization process. Int J Pharm 2002; 245:161–177.
42. Perissutti B, Rubessa F, Moneghini M, Voinovich D. Formulation design of carbamazepine
fast-release tablets prepared by melt granulation technique. Int J Pharm 2003;
256:53–63.
43. Thies R, Kleinebudde, P. Melt pelletisation of a hygroscopic drug in a high shear mixer.
Part 1. Influence of process variables. Int J Pharm 1999; 188:131–143.
44. Yanze FM, Duru C, Jacob M. A process to produce effervescent tablets: fluidized bed
dryer melt granulation. Drug Dev Ind Pharm 2000; 26(11):1167–1176.
45. Heng PWS, Wong TW, Cheong WS. Investigation of melt agglomeration process with
a hydrophobic binder in combination with sucrose stearate. Eur J Pharm Sci 2003;
19:381–393.
46. Ukita K, Murakami T. Preparation of essential oils loaded granule by melt granulation.
Drug Dev Ind Pharm 1994; 20(6):981–992.
47. Appelgren C, Eskilson C. A novel method for the granulation and coating of pharmacologically
active substances. Drug Dev Ind Pharm 1990; 16(15):2345–2351.
48. Sch?fer T, Worts O. Control of fluidized bed granulation: I: effects of spray angle, nozzle
height and starting materials on granule size and size distribution. Arch Pharm
Chem Sci Ed 1977; 5:51–60.
49. Crowley KJ, Forbes RT, York P, Nyqvist H, Camber O. Drug-fatty acid salt with waxlike
properties employed as binder in melt granulation. Int J Pharm 2000; 211:9–17.
50. Schaafer T, Mathiesen C. Melt pelletization in a high shear mixer. IX. Effects of binder
particle size. Int J Pharm 1996; 139:139–148.
51. Johansen A, Sch?fer T. Effects of physical properties of powder particles on binder
liquid requirement and agglomerate growth mechanisms in a high shear mixer. Eur
J Pharm Sci 2001; 14:135–147.
52. Scott AC, Hounslow MJ, Instone T. Direct evidence of heterogeneity during high-shear
granulation. Powder Technol 2000; 113:205–213.
53. El-Shanawany S. Sustained release of nitrofurantoin from inert wax matrices. J Control
Release 1993; 26:11–19.
54. Brabander CD, Vervaet C, Remon JP. Development and evaluation of sustained release
mini-matrices prepared via hot melt extrusion. J Control Release 2003; 89:235–247.
55. Sastry KVS, Fuerstenau DW. Mechanisms of agglomerate growth in green pelletization.
Powder Technol 1973; 7:97–105.
56. Ennis BJ, Litster JD. Particle size enlargement. Perry R, Green D, eds. Perry’s chemical
Engineers’ Handbook. 7th ed. Chapter 20. New York: McGraw-Hill, 1997:56–73.
57. Ennis BJ, Tardos G, Pfeffer R. A microlevel-based characterization of granulation
phenomena. Powder Technol 1991; 65:257–272.
58. Iveson SM, Litster JD. Fundamental studies of granule consolidation. Part 2: Quantifying
the effects of particle and binder properties. Powder Technol. 1998; 99:243–250.
59. Iveson SM, Litster JD, Ennis BJ. Fundamental studies of granule consolidation. Part 1:
effects of binder content and binder viscosity. Powder Technol 1996; 88:15–20.
60. Thies R, Kleinebudde P. Melt pelletization of a hygroscopic drug in a high shear mixer.
Part 3. Effects of binder variation. Chem Pharm Bull 2001; 49(2):140–146.
61. Rumpf H. The strength of granules and agglomerates. In: Knepper WA, ed. Agglomeration.
New York: Wiley-Interscience, 1962:379–419.
Melt Granulation and Pelletization 403
© 2005 by Taylor & Francis Group, LLC
62. Knight PC. An investigation of the kinetics of granulation using a high shear mixer.
Powder Technol 1993; 77:159–169.
63. Knight PC, Instone T, Pearson JMK, Hounslow MJ. An investigation into the kinetics
of liquid distribution and growth in high shear mixer agglomeration. Powder Technol
1998; 97:246–257.
64. Knight PC, Johansen A, Kristensen HG, Sch?fer T, Seville JPK. An investigation of the
effects on agglomeration of changing the speed of a mechanical mixer. Powder Technol
2000; 110:204–209.
65. Heng PWS, Wong TW, Chan LW. Influence of production variables on the sphericity of
melt pellets. Chem Pharm Bull 2000; 48(3):420–424.
66. Heng PWS, Chan LW, Zhu L. Effects of process variables and their interactions on melt
pelletization in a high shear mixer. STP Pharm Sci 2000; 10(2):165–172.
67. Voinovich D, Moneghini M, Perissutti B, Franceschini E. Melt pelletization in high
shear mixer using a hydrophobic melt binder: influence of some apparatus and process
variables. Eur J Pharm Biopharm 2001; 52(3):305–313.
68. Zhou F, Vervaet C, Rernon JP. Influence of processing on the characteristics of matrix
pellets based on microcrystalline waxes and starch derivatives. Int J Pharm 1997;
147(1):23–30.
69. Sch?fer T. Melt pelletization in a high shear mixer. X. Agglomeration of binary
mixtures. Int J Pharm 1996; 139:149–159.
70. Johansen A, Sch?fer T. Effects of interactions between powder particle size and binder
viscosity on agglomerate growth mechanisms in a high shear mixer. Eur J Pharm Sci
2001; 72:297–309.
71. Watano S, Terashita K, Miyanami K. Determination of end-point with a complex granulation
applying infrared moisture sensor. Chem Pharm Bull 1991; 39(4):1013–1017.
72. Watano S, Yamamoto A, Miyanami K. Effects of operational variables on the properties
of granules prepared by moisture control method in tumbling fluidized bed granulation.
Chem Pharm Bull 1994; 42(1):133–137.
73. Frake P, Greenhalgh D, Grierson SM, Hempenstall JM, Rudd DR. Process control and
end point determination of a fluid bed granulation by application of near infrared
spectroscopy. Int J Pharm 1997; 151(1):75–80.
74. Watano S, Miyanami K. Image processing for on-line monitoring of granule size distribution
and shape in fluidized bed granulation. Powder Technol. 1995; 83(1):55–60.
75. Watano S, Sato Y, Miyanami K. Control of granule growth in fluidized bed granulation
by an image processing system. Chem Pharm Bull 1996; 44(8):1556–1560.
76. Watano S, Fukushima T, Miyanami K, Murakami T, Sato T. Automation of the manufacturing
process of cold remedy granules by a tumbling fluidized bed applying the
fuzzy control method. Chem Pharm Bull 1994; 42(6):1302–1307.
77. Watano S, Fukushima T, Miyanami K. Application of fuzzy logic to bed height control
in agitation-fluidized bed granulation. Powder Technol 1994; 81(2):161–168.
78. Baert L, Fanara D, Baets PD, Remon JP. Instrumentation of a gravity feed extruder
and the influence of the composition of binary and ternary mixtures on the extrusion
forces. J Pharm Pharmacol 1991; 43(11):745–749.
79. Baert L, Fanara D, Remon JP, Massart D. Correlation of extrusion forces, raw materials
and sphere characteristics. J Pharm Pharmacol 1992; 44(8):676–678.
80. Elbers JAC, Bakkenes HW, Fokkens JG. Effect of amount and composition of granulation
liquid on mixing, extrusion and spheronization. Drug Dev Ind Pharm 1992;
18(5):501–517.
81. Baert L, Down GRB. A comparison of two methods of instrumenting a small-scale
basket extruder. Int J Pharm 1994; 107(3):219–222.
82. Kleinebudde P, Solvberg AJ, Lindner H. The power consumption-controlled extruder: a
tool for pellet production. J Pharm Pharmacol 1994; 46(7):542–546.
83. Vervaet C, Baert L, Risha PA, Remon JP. The influence of the extrusion screen on pellet
quality using an instrumented basket extruder. Int J Pharm 1994; 107(1):29–39.
404 Wong et al.
© 2005 by Taylor & Francis Group, LLC
84. Kleinebudde P. Use of power consumption-controlled extruder in the development of
pellet formulations. J Pharm Sci 1995; 84(10):1259–1264.
85. Lindberg N-O, Leander L, Wenngren L, Helgesen H, Reenstierna B. Studies on granulation
in a change can mixer. Acta Pharm Suecia 1974; 11:603–620.
86. Travers DN, Rogerson AC, Jones TM. A torque arm mixer for studying wet massing.
J Pharm Pharmacol 1975; 27(suppl):3.
87. Lindberg N-O. Studies on granulation in a small planetary mixer. II. Granulation of
lactose and an antacid mixture. Acta Pharm Suecica 1977; 14:197–204.
88. Lindberg N-O, Leander L, Nilsson PG, Reenstierna B. Studies on granulation in a small
planetary mixer. I. Instrumentation Acta Pharm Suecica 1977; 14:191–196.
89. Ghanta SR, Srinivas R, Rhodes CT. Use of mixer torque measurements as an aid to
optimizing wet granulation process. Drug Dev Ind Pharm 1984; 10:305–311.
90. Tapper G.L, Lindberg N-O. The granulation of some lactose qualities with different
particle size distributions in a domestic-type mixer. Acta Pharm Suecica 1986; 23:47–56.
91. Bier HP, Leuenberger H, Sucker HB. Determination of the uncritical quantity of
granulating liquid by power measurements of planetary mixers. Pharm Ind 1979;
41:375–380.
92. Leuenberger H. Granulation, new techniques. Pharm Acta Helv 1982; 57:72–82.
93. Paris L, Stamm A. Optimal massing liquid volume determination by energy consumption
measurement: study of the influence of some physical properties of solvents and
products used. Drug Dev Ind Pharm 1985; 11(2,3):361–385.
94. Chirkot T, Propst CW. Low shear granulators. Parikh DM, ed. Handbook of
Pharmaceutical Granulation Technology. New York: Marcel Dekker, Inc., 1997:
205–255.
95. Corvari V, Fry WC, Seibert WL, Augsberger L. Instrumentation of a high shear mixer:
evaluation and comparison of a new capacitance sensor, a watt meter, and a strain-gage
torque sensor for wet granulation monitoring. Pharm Res 1992; 9(12):1525–1533.
96. Kopcha M, Roland E, Bubb G, Vadino WA. Monitoring the granulation process in
a high shear mixer/granulator: An evaluation of three approaches to instrumentation.
Drug Dev Ind Pharm 1992; 18(18):1945–1968.
97. Leuenberger H. Monitoring granulation. Manuf Chem 1984; 55:67,69,71.
98. Holm P, Sch?fer T, Kristensen HG. Granulation in high-speed mixers. Part V. Power
consumption and temperature changes during granulation. Powder Technol 1985;
43:213–223.
99. Holm P, Sch?fer T, Kristensen HG. Granulation in high-speed mixers. Part VI. Effects
of process conditions on power consumption and granule growth. Powder Technol
1985; 43:225–233.
100. Ritala M, Jungersen O, Holm P, Sch?fer T, Kristensen HG. A comparison between binders
in the wet phase of granulation in a high shear mixer. Drug Dev Ind Pharm 1986;
12:1685–1700.
101. Terashita K, Oh’ike A, Kato M, Miyanami K. Granulation process and end-point in
high speed mixer granulator. J Pharm Soc Jpn 1987; 107:377–383.
102. Ritala M, Holm P, Sch?fer T, Kristensen HG. Influence of liquid bonding strength on
power consumption during granulation in a high shear mixer. Drug Dev Ind Pharm
1988; 14:1041–1060.
103. Werani J. Production experience with end point control. Acta Pharm Suecica 1988;
25:247–266.
104. Leuenberger H, Usteri M, Imanidis G, Winzap S. Monitoring the granulation process:
granulate growth, fractal dimensionality and percolation threshold. Boll Chim Pharm
1989; 128:54–61.
105. Watano S, Terashita K, Miyanami K. Frequency analysis of power consumption in
agitation granulation of powder materials with sparingly soluble acetaminophen. Chem
Pharm Bull 1992; 40(1):269–271.
Melt Granulation and Pelletization 405
© 2005 by Taylor & Francis Group, LLC
106. Shiraishi T, Kondo S, Yuasa H, Kanaya Y. Studies on the granulation process of
granules for tableting with a high speed mixer. I. Physical properties of granules for
tableting. Chem Pharm Bull 1994; 42(4):932–936.
107. Vertommen J, Michoel A, Rombaut P, Kinget R. Production of pseudoephedrine HC1
pellets in a high shear mixer-granulators. Eur J Pharm Biopharm 1994; 40(1):32–35.
108. Shiraishi T, Sano A, Kondo S, Yuasa H, Kanaya Y. Studies on the granulation process
of granules for tableting with a high speed mixer. II. Influence of particle size of active
substance on granulation. Chem Pharm Bull 1995; 43(4):654–659.
109. Landin M, York P, Cliff MJ, Rowe RC, Wigmore AJ. The effect of batch size on scaleup
of a pharmaceutical granulation in a fixed bowl mixer granulator. Int J Pharm 1996;
134:243–246.
110. Laicher A, Profitlich T, Schwitzer K, Ahlert D. Modified signal analysis system for end
point control during granulation. Eur J Pharm Sci 1997; 5(1):7–14.
111. D’Alonzo GD, O’Connor RE, Schwartz JB. Effect of binder concentration and method
of addition on granule growth in a high intensity mixer. Drug Dev Ind Pharm 1990;
16(12):1931–1944.
112. Lin K, Peck GE. Development of agglomerated talc. II. Optimization of the processing
parameters for the preparation of granulated talc. Drug Dev Ind Pharm 1995;
21(2):159–173.
113. Lindberg N-O, Jo?nsson C. Granulation of lactose in a recording high speed mixer,
Diosna P25. Drug Dev Ind Pharm 1983; 9(6):959–970.
114. Timko RJ, Johnson JL, Skinner GW, Chen ST, Rosenberg HA. Instrumentation of
a vertical high shear mixer with a motor slip monitoring device. Drug Dev Ind Pharm
1986; 12(10):1375–1393.
115. Cliff MJ. Granulation end point and automated process control of mixer-granulators:
part I. Pharm Technol 1990; 14(4):112–132.
116. Cliff MJ. Granulation end point and automated process control of mixer-granulators:
part II. Pharm Technol Int 1990; 18:20–23.
117. Kristensen HG, Sch?fer T. Granulations. In: Swarbrick J, Boylan JC, eds. Encyclopedia
of Pharmaceutical Technology. Vol. 7. New York: Marcel Dekker, Inc., 1993:
121–160.
118. Holm P. High shear mixer granulators. Parikh DM, ed. Handbook of Pharmaceutical
Granulation Technology. New York: Marcel Dekker, Inc., 1997:151–204.
119. Sch?fer T, Mathiesen C. Melt pelletization in a high shear mixer. VIII. Effects of binder
viscosity. Int J Pharm 1996; 139:125–138.
120. Wong TW, Chan LW, Heng PWS. Study of the melt pelletization process focusing on
the micromeritic property of pellets. Chem Pharm Bull 2000; 48(11):1639–1643.
406 Wong et al.
© 2005 by Taylor & Francis Group, LLC
14
Rapid Release Granulation
P. W. S. Heng
National University of Singapore, Singapore
Anthony Yolande and Lee Chin Chiat
International Specialty Products (ISP) Asia Pacific Pte. Ltd., Singapore
1. INTRODUCTION
Many potential drug candidates which had been screened to show therapeutic
activities could not be formulated into suitable oral dosage forms, which are the
most preferred form of drug administration, due to poor aqueous solubility. These
compounds were abandoned during preformulation stage because they did not exhibit
sufficient solubility in the aqueous environment, implying reduced bioavailability
upon administration and the inability to attain the necessary drug levels in blood for
trating various oral dosage forms and it is obvious that there are many factors
affecting the absorption of drug.
The factors can be broadly classified into patient or dosage-form related. The
focus of the current chapter is to examine some of the dosage-form related factors
which can be controlled from the formulator’s perspective, to enhance the dissolution
of the bioactive compounds in order to increase the bioavailability of poorly
water-soluble compounds through the granulation process, which is making granules
to be filled into capsules or tableted. Most of the dosage-form related factors are
granulation-related factors.
Rapid release granulation is expected to benefit the class of compound where
absorption is highly dependent on the dissolution of the drugs in the gastrointestinal
tract, termed as Class II compounds under the Biopharmaceutics Classification
1.1. Definition
Solution can be defined as a mixture of at least two components forming a single
phase, which is homogenous at the molecular level. The transfer rate of molecules
or ions from solid state into solution is known as dissolution rate and the extent
to which this proceeds under a given set of conditions is known as solubility.
407
therapeutic effect. Figure 1 depicts the possible events that can occur upon adminisoutlined
in Figure 2. This could be further classified into formulation-related or
Scheme (Table 1).
© 2005 by Taylor & Francis Group, LLC
Thus, the rate of solution (dissolution rate) and the amount that can be dissolved
(solubility) are two separate concepts but in practice, high solubility is usually
associated with high dissolution rate (2).
1.2. Factors Affecting Dissolution Rate
Most of the dosage-form related factors highlighted in Figure 2 can be explained by
the terms in the modified Noyes
Whitney equation (3,4), which is as follows:
Figure 2 Dosage-form related factors.
Figure 1 Schematic illustration of the possible events upon administering various oral
dosage forms.
408 Heng et al.
© 2005 by Taylor & Francis Group, LLC
dC
dt ?
AD Cs 
 C ? ?
h ?1:1?
where dC/dt is the dissolution rate, A is the surface area available for dissolution, D
is the diffusion coefficient of the compound, Cs is the solubility of the compound in
the dissolution medium, C is the concentration of drug in the dissolution medium at
time t, and h is the thickness of the diffusion boundary layer adjacent to the surface
2. FORMULATION-RELATED FACTORS
2.1. Drug
2.1.1. Free Acid, Free Base, or Salt Form
The use of the appropriate form of drug prior to granulation is of utmost importance
to ensure the production of rapid release granules. Sodium salts of weak acids or
hydrochloride salts of weak bases can cause marked increase in aqueous solubility
when compared to the corresponding free acids or bases. This is attributed to an
increase in interactions between drug and water, giving rise to a greater degree of
ionic dissociation of the drug when it dissolves in water. An example of this effect
is the aqueous solubilities of salicylic acid and its sodium salt, which are 1:550 and
1:1, respectively (2).
2.1.2. Particle Size
The most common approach for improving the dissolution rate (dC/dt) of drug
from rapid release granules is to increase the surface area (A) available for dissolution
and this is often achieved by employing finely divided particles [Eq. (1.1)]. Size
reduction has also been shown to decrease the diffusion boundary layer (h) of
sparingly soluble drugs (5–7). Hence, the combined effects of A and h improve the
dissolution rate of the drug. Griseofulvin represents a classical example of a drug
where improvement in rate of absorption can be brought about by an increase in dissolution
rate due to the increase in surface area by size reduction. This drug, which
was initially marketed as coarse particles, resulted in many cases of therapeutic
failures due to low bioavailability. Kraml et al. (8) demonstrated that 0.5 g of micronized
griseofulvin produced the same serum level as 1.0 g of the unmicronized form.
Subsequently, the use of micronized griseofulvin enabled a dosage reduction and
contributed not only to lower drug cost to the patient but more importantly,
a decrease in therapeutic failures caused by poor absorption (9). However, size
Table 1 Biopharmaceutics Classification Scheme
Class Solubility Permeability
I High High
II Low High
III High Low
IV Low Low
Source: Amidon et al. (1)
Rapid Release Granulation 409
of the dissolving compound (Table 2).
© 2005 by Taylor & Francis Group, LLC
Table 2 Terms That are Being Affected by the Various Dosage-Form Related Factors
Term in modified
Noyes–Whitney
equation Affected by Comments
A Formulation-related factors
Drug
Particle size A is inversely related to the drug
particle size
Additives
Effervescent agent
Hydrophilic agent
Surface active agent
Superdisintegrant
Poorly water-soluble drug particles
tend to form a coherent mass in the
dissolution medium; the dispersibility
of these particles can be achieved
with the additives listed to enhance
dissolution
Granulation-related factors
Granulating equipment
Liquid binder
Wet massing time
Tableting force
In general, these factors affect the
hardness of the granules; hardness
is usually associated with lower
granule porosity and this will lead
to poorer dissolution as the
granules do not break up easily
and the penetration of dissolution
medium is impeded
Cs Formulation-related factors
Drug
Free acid, free base or
salt form
Polymorphs or solvates
Amorphous state
Sections 2.1.1, 2.1.3, and 2.1.4
Additives
Complexing agent
Surface active agent
Sections 2.2.1 and 2.2.5
H Formulation-related factors
Drug
Particle size The smaller the drug particle size the lower
the diffusion boundary layer (5–7)
Additives
Effervescent agent
Hydrophilic agent
Surface active agent
Superdisintegrant
Poorly water-soluble drug particles
tend to form a coherent mass in
the dissolution medium having a
large resultant diffusion boundary
layer; dispersing these particles into
individual particles decreases this value
D Granulation-related factors
Granulating equipment
Liquid binder
Wet massing time
Tableting force
Drug does not readily diffuse into the
dissolution medium from hard
granules
Blending Section 3.4
Source: Modified from Ref. 2.
410 Heng et al.
© 2005 by Taylor & Francis Group, LLC
reduction has its practical limits as micronized particles tend to aggregate due to the
high surface energy per unit mass. Aggregation reduced the surface area available for
dissolution and lowered the drug dissolution rate (10). Wetting effect is particularly
important under such a situation as it increases the effective surface area (11). This
principle is commonly applied in interactive mixtures (Section 2.1.2.1). In a study
using rapid release granules, micronized DPC 961, which was an antiretroviral drug,
was wet granulated by high shear and the granules were subsequently tableted and
fed to beagle dogs (12). It was found that for the fasted dogs, tablets made from
micronized DPC 961, which displayed higher in vitro dissolution, produced a higher
plasma drug concentration due to faster dissolution and absorption. The effect of
particle size reduction was not obvious for the nonfasted dogs, probably because
the solubilization afforded by the nonfasted state was sufficient for rapid dissolution
of drug irrespective of particle sizes. Hence, particle size of the drug is important if it
is to be consumed before meals.
There are several micronization techniques available and they can be broadly
classified into mechanical- and solution-based micronization (Table 3) (13).
Table 3 illustrates some of the commonly established micronization techniques.
An example of a technique that could be viewed as a solution-based micronization is
solvent deposition, which could be classified as a form of spray drying. The poorly
soluble drug, tolbutamide, was dissolved in aqueous ammonia, and mixed with a
core material, either low-substituted hydroxypropylcellulose or partly pregelatinized
corn starch. The suspension was subsequently spray dried leaving a thin film of fine
drug crystals on the surface of the core materials. The rapid solvent evaporation prevented
the growth of large drug crystals. Using disintegrant as the core material
further improved drug release because the swelling of disintegrant would dislodge
the thin film of fine drug crystals from the core for better dissolution, thus effectively
dispersing the fine drug crystals (14). This study illustrated the possibility of concurrent
granulation with micronization. The concept was probably pioneered by
Monkhouse and Lach (15,16) when they loaded poorly water-soluble drugs onto
fumed silicon dioxide by dissolving the drugs in organic solvent and suspending
the fumed silicon dioxide in the drug solution, simultaneously evaporating the
organic solvent. The drugs loaded onto the fumed silicon dioxide improved the dissolution
profiles. This improvement was attributed to particle size reduction and
changes in the crystalline structures (Section 2.1.3). The crystalline nature of the drug
has to be assessed after undergoing micronization as certain crystal forms may be
more suitable for formulation of rapid release granules (Section 2.1.3). Drug can also
be micronized with additives, usually water-soluble carriers, such that the intimate
interaction between drug and additives enabled rapid drug release when in solution.
This will be discussed under Solid Dispersion (Section 4).
2.1.2.1. Interactive Mixtures. The concept of interactive mixing was first
introduced by Hersey (17) to describe the mixing of fine and cohesive particles. This
form of mixing is different from random mixing because of the adhesion of fines to
coarser particles (18). The adhesion may be due to many factors such as humidity,
adsorption, chemisorption, surface tension, friction, and electrostatic forces, unlike
random mixing where gravitational force predominates over the other types of
adhesion forces. Interactive mixtures could be viewed as ‘‘miniature granules’’ since
the fine drug and coarser particles exist as agglomerates. The forces of adhesion can
be so great that it is able to resist the fluidization force in a fluid bed, thus facilitating
the coating of the ‘‘miniature granules’’ (19).
The basic mechanism in enhancing the dissolution rate of a poorly watersoluble
drug in an interactive mixture is by reduction of drug aggregates upon mixing
with highly soluble and coarse carrier particles, thus improving the wettability of
the micronized drug. By adding carrier particles, fine drug particles will adhere onto
the coarser carrier particles, thereby increasing the exposed surface area for drug
dissolution upon rapid concommitant dissolution of the carrier into the dissolution
medium (20–22). Thus, the solubility of the coarse carrier particles plays an
important role in the dissolution of the adhered drug. The use of carriers having
surface active properties will further improve the dissolution rate of the drugs by
solubilization effect during dissolution.
2.1.3. Polymorphs and Solvates
Drug can exist in one or more crystalline forms. The term crystalline implies that the
structural units, known as unit cells, are repeated regularly and indefinitely in threedimensional
space. Crystalline forms of a drug can be further classified into polymorphs
and solvates, also known as pseudopolymorphs. Crystalline polymorphs
have the same chemical composition but different crystal structures because of the
manner in which molecules arrange themselves in crystal lattices during crystallization.
Solvates are crystalline forms containing solvent molecules in the crystal structures
in stoichiometric or nonstoichiometric proportion. The differences in crystal
packing of the polymorphs and solvates would mean that they differ in lattice energy
and entropy, thereby significantly affecting physicochemical properties, one of which
is the solubility of the drug (23,24). This implied that the dissolution and absorption
of a poorly water-soluble drug may also be affected by the actual polymorphs or
solvates present in the oral dosage forms. Usually, only one stable polymorph exists
for a given set of environmental conditions. In terms of enhancing dissolution, it is
better to employ metastable polymorphs of the drug. However, in processing, it is
a manufacturer’s nightmare as the use of metastable polymorphs may lead to
stability problems on storage unless a way is found to stabilize the polymorph.
Hence, it is always desirable to choose the most soluble and stable form of the drug
during the initial stages of product development (25).
The choice of polymorph before granulation determines the dissolution rate.
Otsuka et al. (26) found that the use of carbamazepine polymorphs I and IV, which
are the respective anhydrate and dihydrate forms, gave rise to a higher dissolution rate
than polymorphs II and III, which are the anhydride forms, when hydroxypropylcellulose
solution was employed as binder. This was because the tablets made from polymorphs
I and IV were not as hard and thus disintegrated faster upon contact with the
412 Heng et al.
© 2005 by Taylor & Francis Group, LLC
dissolutionmedium. During wet granulation, formsI, II, and III were converted to form
IV. Thus, the anhydrate and the anhydrides were converted to the dihydrate form. Only
2.5% of form I was converted as compared to 80.3% of form II and 35.2% of form III.
The granulation was unsuccessful when forms II and III were employed. This was
because water from the liquid binder was used to hydrate these two polymorph forms
and little was available for wet massing. Only upon adding more water could the granule
yield be improved. Upon drying, the transformed dihydrate lost the water molecules to
form polymorph III. Drying resulted in a reduction in particle size as the removal of
solvent from solvates is a size reduction process as well. Thus, the specific surface area
for granules obtained using forms II and III were higher than forms I and IV, therefore
contributing to higher mechanical strength of the tablet and hence poorer dissolution.
The effect of the types of liquid binder on polymorphic conversion is clearly illustrated
by the work carried out by Otsuka et al. (27) when it was found that carbamazepine form
I was readily converted to the dihydrate form, which is form IV, by using 50% aqueous
ethanol solution but relatively unaffected by distilled water or ethanol. Thus, this
implies that if 50% aqueous ethanol solution is to be used, the resultant tablets would
be poorly soluble as the above-mentioned transformations would have occurred.
2.1.4. Amorphous State
Amorphous state can be defined with reference to a crystalline solid. It is similar to
crystalline solid as an amorphous solid has short-range molecular arrangement with
neighboring molecules, but unlike crystalline solid, an amorphous solid has no longrange
order of molecular packing. The major interest in amorphous solids is in their
higher solubilities and dissolution properties as well as better compressibility as compared
to the crystalline forms. However, amorphous solids are generally not as stable
physically and chemically. Currently, many researchers are exploring various
methods to maintain the stability of drugs in their amorphous state.Methods include
stabilization of the amorphous solids by additives and storing the amorphous state
under appropriate storage temperatures (28).
2.2. Additives
2.2.1. Complexing Agent
An additive is able to improve the dissolution of a poorly water-soluble drug by
forming a water-soluble intermolecular complex. The additive is called solubilizing
Table 4 Solubilizing Agents for Poorly Soluble Substances
Poorly water-soluble substance Solubilizing agent
Caffeine Sodium benzoate, sodium salicylate
Calcium theobromine Calcium salicylate
Iodine Potassium iodide
Mephenesin Salicylic acid
Quinine hydrochloride Urethane, carbamide
Riboflavin Nicotinic amide
Theobromine Sodium salicylate
Theophylline Sodium acetate, ethylenediamine
Source: From Ref. 29.
Rapid Release Granulation 413
© 2005 by Taylor & Francis Group, LLC
agent and it must be present at an optimal concentration to maximize the solubility
One commonly used additive is cyclodextrin. Cyclodextrins are bucket-shaped
oligosaccharides produced from starch.
As a result of their molecular structures and shapes, cyclodextrins are able
to entrap poorly water-soluble drugs in the hydrophobic cavities forming watersoluble
inclusion complexes with the poorly water-soluble drugs. Cavallari et al. (30)
showed that complexation and granulation of piroxicam with b-cyclodextrin by steam
granulation in a single process improved in vitro drug release when compared to a
physical mixture of drug and b-cyclodextrin in the same composition. 2-Hydroxypropyl-
b-cyclodextrin was employed by Uekama et al. (31) to form water-soluble
inclusion complex with nifedipine. The drug and additive were dissolved in
ethanol-dichloromethane and subsequently spray dried. Tablets prepared using the
inclusion complexes were found to promote nifedipine in vitro release rate. Upon
feeding the test tablets to dogs, superior oral bioavailability was obtained. Storage
at 60C and 75% relative humidity did not result in nifedipine crystal growth because
the nifedipine molecules were basically entrapped within the 2-hydroxypropylb-
cyclodextrin, minimizing the opportunity for the nifedipine molecules coming
together and recrystallizing. Thus, the stability in dissolution profile could be
achieved throughout the entire period of study.
2.2.2. Effervescent Agent
Effervescence is defined as the evolution of bubbles of gas from a liquid as the result
of a chemical reaction. The generation of gas, which is usually carbon dioxide, is
usually achieved by spontaneous chemical reaction between a soluble acid source
and an alkali metal carbonate in the presence of water. The reaction can be triggered
in the presence of a small amount of water, thus moisture protection of such oral
dosage forms is paramount. This is because effervescence reaction when initiated will
proceed spontaneously because one of the by-products is water.
Several researchers had found that the bioavailability of aspirin from effervescence
tablet was higher than conventional or enteric-coated tablets (32–34). The reasons
for this included the dramatic disintegration rate of the tablets that enabled rapid
release of the drug particles for dissolution and the increase in gastric emptying rate.
2.2.3. Hydrophilic Agent
In the formulation of rapid release granules, it is important to consider the formulation
in its totality. It is always better to employ more hydrophilic agents which may
be water-soluble or water-insoluble as binders or fillers in the formulation so that
water penetration into the granules will not be impeded. Hydrophilic agents
commonly employed include hydroxylpropylcellulose, hydropropylmethylcellulose,
lactose, microcrystalline cellulose, polyvinylpyrrolidone, starch, and many others.
These additives aid in wetting the poorly water-soluble drug. Polyvinylpyrrolidone
K30 was able to improve the wetting of nifedipine complexed with 2-hydroxypropyl-
b-cyclodextrin, thus negating the decrease in nifedipine release brought about
by the poor wettability and good compressibility of the complex (35).
414 Heng et al.
of a poorly water-soluble drug. Table 4 lists some of the common solubilizing agents.
© 2005 by Taylor & Francis Group, LLC
2.2.4. Lubricant
Lubricant is employed in a formulation to facilitate the ejection of tablets from the
die. Excessive sticking of tablets to the punches or wear of punches and dies will be
reduced. Lubricant functions by forming a thin film at the interface of the tablet and
die, reducing the shearing force during the ejection of tablet. An optimum amount of
lubricant must be used for each formulation in order to optimize the drug dissolution.
Excess lubricant interferes with both disintegration and bioavailability
by waterproofing the granules and tablets. Gordon (36) found that doubling the
quantity of magnesium stearate to be mixed with naproxen granules before tableting
resulted in a slower drug dissolution rate.
2.2.5. Surface Active Agent
This additive is capable of forming aggregates called micelles above the critical
micelle concentration. This is the major distinguishing feature between surface active
agent and complexing agent. In an aqueous environment, the cores of these
aggregates resemble a separate organic phase providing a suitable environment for
poorly water-soluble drugs. Solubilization results in the increase in the aqueous solubility
of drugs and aids dissolution. It is also able to improve wetting of the granules.
Surfactant can also potentially disrupt the integrity of membranes to enhance drug
absorption.
Sodium lauryl sulfate was employed in the formulation of ibuprofen tablet to
enhance ibuprofen dissolution. Sodium lauryl sulfate was incorporated during the
wet granulation phase and the resultant dissolution profiles were even better than
commercial products (37). In another study by Gohel and Patel (38), the authors
used a combination of hydrophilic and surface active agents to bring about rapid
release of nimesulide. The hydrophilic agents studied included polyethylene glycol
400, propylene glycol and polyvinylpyrrolidone K30, whereas the surface active
agents investigated were sodium lauryl sulfate and polyoxyethylene sorbitan monooleate.
Aqueous solutions of the hydrophilic agents and surface active agents were
mixed with the drug before the suspension was further mixed with the rest of the
powder mass during granulation. Thus, additives were incorporated as part of the
liquid binder. The granules formed were shown to display enhanced dissolution as
compared to the pure drug and the improvement in drug release property was
attributed to increase in wettability and the solubilizing effects by the additives.
The solubilizing effect of sodium lauryl sulfate on griseofulvin in the griseofulvin–
polyethylene glycol 3000 system was demonstrated by Sjo? kvist et al. (39). At an
optimum concentration of sodium lauryl sulfate, the crystalline griseofulvin was converted
into a solubilized state, which was in the molecular form. The change of phase
could be detected by x-ray diffractometry and differential scanning calorimetry.
Thus, the drug molecules existing in a molecularly dispersed state could be easily
released into the dissolution medium upon contact with the dissolution medium.
Dry granulation via slugging and roller compaction of hydroxypropylmethylcellulose
with naproxen, nifedipine, or carbazepine resulted in improvement of their
dissolution profiles. The mechanism for dissolution enhancement was believed to be
a microenvironment hydroxypropylmethylcellulose surfactant effect facilitated by
placing the hydroxypropylmethylcellulose in close proximity to the drug as simple
mixing did not result in enhanced dissolution (40). Another example was the use
of polyethylene 40 hydrogenated castor oil to improve the in vitro release of hydrochlorothiazide
(41). The release was dependent on the concentration of surface active
Rapid Release Granulation 415
© 2005 by Taylor & Francis Group, LLC
agent and a higher concentration gave rise to better dissolution. Mehta et al. (42)
managed to enhance nifedipine release from granules with copolymer of propylene
oxide and ethylene oxide as the surface active agent.
2.2.6. Superdisintegrant
The addition of disintegrant is to facilitate the breakup of tablet, thus presenting the
micronized drug to the dissolution medium. In general, for a drug having solubility
of 10 mg/mL or less, disintegration rate of the solid dosage form has a profound
effect on the dissolution profile (43). Superdisintegrants represent a subclass of
disintegrants that are associated with dramatic disintegration rates. Some of the
common superdisintegrants are crospovidone, croscamellose sodium, and sodium
starch glycolate. Studies in the 1970s and 1980s usually used the disintegration test
to investigate the effect of incorporating superdisintegrants into formulations. Even
though tablet disintegration is often a necessary precursor for drug dissolution, it
does not ensure that the drug of interest dissolves but in practice, the two are highly
correlated. In 1993, Gordon and coworkers (44) studied the effect of superdisintegrants
employed during the wet granulation process in terms of dissolution and
confirmed the usefulness of superdisintegrants in enhancing dissolution of p-aminobenzoic
acid in tablets containing mainly lactose, dibasic calcium phosphate dehydrate,
or naproxen, in order to vary the water affinity of the tablet manufactured.
Bolhuis et al. (45) also verified the effectiveness of superdisintegrants in improving
the dissolution of methylprednisolone and phenylbutazone. Granules produced from
wet granulation were either filled into capsule or tableted. Both types of oral dosage
forms displayed enhanced dissolution. Superdisintegrants were found to be useful in
melt granulation as well as where the intragranular addition of crospovidone during
the melt granulation of carbamazepine, polyethylene glycol 4000, and lactose
monohydrate was able to reduce the dissolution t90% by half (46).
Figure 3 Possible factors affecting the rapid release of drug in the respective steps under dry
and wet granulation methods.
416 Heng et al.
© 2005 by Taylor & Francis Group, LLC
3. GRANULATION-RELATED FACTORS
some of the factors that may affect the final release of drug from capsules or tablets
made from the granules.
The first step of both methods, which is interactive mixing, plays an important
part in improving the dissolution of poorly soluble drug because the mixing of micronized
drug with the appropriate additives reduces the cohesiveness of the drug particles,
leading to better dissolution of the drug (Section 2.1.2.1). Mitchell et al. (40) processed
poorly water-soluble drugs with hydropropylmethylcellulose by dry granulation and
achieved enhanced drug dissolution probably due to the reduction in cohesiveness of
the drugs and surface active property of hydropropymethylcellulose. Levy et al. (47)
reported that the dissolution rate of salicylic acid tablets increased with a decrease in
the granule size prepared by the slugging method. Gao et al. (48) improved the dissolution
of an experimental drug by carrying out wet granulation in fluid bed and high-shear
mixer and this could be attributed to the combined effects of employing the micronized
form of the drug, a surface active agent, a superdisintegrant, and reduction in the
cohesiveness of the drug by interactive mixing.
3.1. Liquid Binder
Wet granulation offers an opportunity for the transformation of crystal forms and
the choice of the liquid binder plays an important role in determining the final crystal
form of the drug in the granules obtained (Section 2.1.3). Transformation usually
occurs during the addition of liquid binder to the powder mass during wet massing
and drying of the formed granules. Addition of the binder could be viewed as suspending
drug in a mixture of solvent and additives, hence, encouraging transformation
of anhydrates to solvated forms. If sufficient liquid binder is added, this could be
viewed as a solution step and subsequent drying of granules as the recrystallization
step. Thus, close attention has to be paid to polymorph conversion during wet
granulation. An example of such a situation is theophylline. Theophylline readily
converts to the monohydrate form (II) upon exposure to water during wet granulation.
The hydration process could be followed closely using near-infrared spectroscopy
(49) and charge-coupled device Raman spectroscopy (50). This conversion
could not be prevented by having high water absorbing capacity additive such as
the silicified microcrystalline cellulose because the minimal amount of water for
effective wet massing was sufficient to trigger this conversion (51). Theophylline
monohydrate under vacuum dehydration forms the metastable anhydrate (I). This
metastable anhydrate on storage recrystallized to stable anhydrous form (I) and in
the process, formed solid bridges within the tablet (52). A decrease in dissolution
would result if the recrystallization process were not prevented. The use of water
as the liquid binder also affected the dissolution of naproxen sodium because of drug
hydration (53) resulting in a poorer dissolution of the drug. Thus, if wet granulation
has to be carried out for these drugs, and in order to maintain rapid dissolution,
polymorphic conversion may be prevented by using ethanol as the liquid binder.
The solubility of drug in the liquid binder also affects the dissolution of the
resulting capsules or tablets made from these granules. This was illustrated by a
study carried out by Wu et al. (54) who found a correlation between dissolution rate
constants of zindotrine and the solubility of zindotrine in liquid binder containing
varying ethanol concentrations. This was explained by the dissolution of a certain
Rapid Release Granulation 417
Figure 3 refers to some of the common steps in dry and wet granulation methods and
© 2005 by Taylor & Francis Group, LLC
amount of zindotrine, followed by the recrystallization of fine crystals during the
drying phase of the granules. The subsequent enhancement in dissolution was due
to the finer crystals formed due to the recrystallization process. Thus, the drug
had effectively undergone the micronization process during wet granulation.
The quantity of liquid binder also gives rise to different dissolution profiles.Alow
amount used would result in the production of smaller granules and the resultant tablets
formed displayed much faster dissolution as compared to granules formed using a
higher amount of liquid binder (36). Besides forming larger granules, a higher amount
of liquid binder used is also expected to increase the hardness of granules. Hard tablets
will give rise to poorer dissolution because the tablets require more timeto break up (55).
This observation was found to be dependent on the moisture content of the tablets. At
moisture contents of 1.6% and 2.0%, ticlopidine hydrochloride dissolution was highly
dependent on the tablet strength. However, with moisture contents of more than
3.0%, the drug dissolution was independent of tablet strength (56).
3.2. Granulating Equipment
The use of the appropriate granulating equipment also plays a part in the release rate
of the drug. Acetaminophen beads made from extrusion/spheronization were compared
to beads made from pan coating. Beads made from pan coating displayed
higher dissolution rates as compared to those made from the former method. This
was attributed to the disintegration of pan-coated beads and the ones made from
extrusion/spheronization were denser and less friable due to the higher energy input
during wet massing and thus did not disintegrate during dissolution. Thus, the
selection of equipment for wet granulation affects the hardness of the granules
and ultimately influences drug release (57). Chowhan et al. (56,58) also reported that
granules made from high-speed shear mixer were lower in porosity as compared to
those prepared from planetary mixer. Low porosity did not facilitate solvent
penetration and, hence, caused poorer drug dissolution.
3.3. Wet Massing
Wet massing was found to play an important factor in the dissolution rate of dyphylline.
Increasing the time during wet massing resulted in an increase in bulk density of
the granules. The maximum bulk density value coincided with the minimum dissolution
rate indicating that the dissolution of drug required the diffusion of dissolution
medium into granules via pores to dissolve the drug (59). Thus, the duration of wet
massing affects the hardness of granules and ultimately, dissolution of the drug.
3.4. Blending
The dissolution rate is affected by the type of blending equipment employed, the
duration of blending of granules with disintegrant, glidant, and lubricant. It was
found that the type of blender affected the distribution of magnesium stearate and
hence, drug dissolution. High-speed blender was employed to mix interactive mixture
of theophylline with magnesium stearate before tableting. It was found that a 15 min
duration was sufficient to impair theophylline dissolution whereas an impairment
of dissolution was not observed for a lower-speed blender. The impairment to drug
dissolution increased with an increase in the duration of blending (60). This was
418 Heng et al.
© 2005 by Taylor & Francis Group, LLC
attributed to the coating of the granules with a thin film of lubricant, a water-repellant,
and hence, compromising the wetting of the granules.
4. SOLID DISPERSION
The term, solid dispersion, refers to a composite solid of one or more drugs in a watersoluble
carrier or matrix prepared by melt (fusion), solvent, or melt–solvent method
(61). Solid dispersions have been traditionally employed to enhance the dissolution rate
of drug, with a viewto improve bioavailability. The common approach to achieve rapid
drug dissolution is to use inert but water-soluble carriers such as polyethylene glycol or
polyvinylpyrrolidone. Solid dispersion enhances the dissolution of drugs by the formation
of fine drug crystals via the eutectic or monotectic systems. Hence, solid dispersion
could also be viewed as a size reduction technique (Section 2.1.2). Under certain conditions,
the drug maybe entrapped in the water-soluble carrier matrix without undergoing
recrystallization to form an amorphous solid solution (Section 2.1.4), a water-soluble
complex with the carrier (Section 2.2.1), or it may dissolve in the carrier to form a true
solid solution. The drug particles in solid dispersions were released as fine or amorphous
colloidal entities upon dissolution of the matrix, enhancing drug dissolution rate. This
concept is deemed to be utilized for drug enhancement if the drug and water-soluble carriers
have been fused prior to or during granulation via any of the three preparation
solid dispersion and the ones suitable for large-scale manufacturing are further discussed
under Section 4.2. Excellent reviews have been published in this area for further
reading (61,96–98).
4.1. Preparation Methods
4.1.1. Melt/Fusion Method
Early pioneers in the field of solid dispersions were Sekiguchi and Obi (99) who utilized
the melt or fusion method to produce dispersion systems. Essentially, a physical mixture
of a drug and a water-soluble carrier was heated directly until melting occurred. The
molten mixture was subsequently solidified in an ice bath. The cooled mass was milled
and sieved.Administration of the product obtained was found to improve absorption of
the drug. Subsequent workers proposed numerous variations to this method.
4.1.2. Solvent Method
Tachibana and Nakumara (100) introduced the concept of solid dispersion preparation
by a solvent method. The proposed method involved dissolving the drug and
water-soluble carrier in a common solvent and subsequent evaporation of the solvent
under vacuum. Solid dispersions produced by this manner are commonly known as
coprecipitates, a term coined by Bates (101). This term is a misnomer and should be
appropriately renamed as coevaporates. In a strict sense, coprecipitates are formed
by the addition of a second solvent to the original solution, which decreases the solubilities
of the drug and carrier in the resultant solutions, thus coprecipitating them.
4.1.3. Melt-Solvent Method
This is a hybrid of the two methods discussed. The drug is dissolved in a suitable
organic solvent and the solution is incorporated directly into a molten carrier.
Subsequently, the organic solvent is evaporated off (61).
Rapid Release Granulation 419
methods (Section 4.1). Table 5 lists a number of processes that employ the concept of
© 2005 by Taylor & Francis Group, LLC
4.2. Processes
4.2.1. Melt Extrusion
This process originated from the plastic industry and it entails conversion of a raw
material into a product of uniform shape and density. The drug and additives are
usually mixed followed by forcing the powder mass through a die under controlled
conditions. The extruded mixtures always exhibited more rapid release of drug vs. the
corresponding physical mixtures as shown by Perissutti et al. (67) using carbamazepine
as the model drug. Hu? lsmann et al. (65) extruded 30% of 17-b-estradiol hemihydrate
with 30% polyvinylpyrrolidone K30 and 40% saccharose-monopalmitate, which, respectively,
functioned as a water-soluble carrier and a surface active agent, to achieve
a good release of drug and this was unchanged upon tableting as the dissolution profiles
of the exudates and tablets were similar. Ndindayino et al. (68) used this technique to
produce hydrochlorothiazide tablets by directly extruding the melt into a tablet-shaped
cavity. The molded tablet thus formed displayed higher in vitro dissolution and relative
bioavailability in healthy volunteers. The improvement in dissolution rate is drug
specific as demonstrated by Forster et al. (66) after investigating the extrusion of
indomethacin, lacidipine, nifedipine, and tolbutamide with either polyvinylpyrrolidone
Table 5 Summary of the Processes Reported in the Literature for Solid Dispersion
Production
Preparation
method Process Investigators
Melt
method
Compression moulding Broman et al. (62)
Extrusion/spheronization Vervaet et al. (41), Vervaet and Remon (63),
Mehta et al. (42)
Melt extrusion El-Egakey et al. (64), Hu?lsmann et al. (65),
Forster et al. (66), Perissutti et al. (67),
Ndindayino et al. (68)
Hot-spin melting Dittgen et al. (69)
In situ granulation Ford and Rubinstein (70)
Melt granulation Kinget and Kemel (71), Passerini et al. (72),
Perissutti et al. (46), Gupta et al. (73–75)
Roll mixing Nozawa et al. (76,77)
Ultrasound-assisted
compaction
Fini et al. (78)
Ultrasound-assisted spray
congealing
Passerini et al. (79), Fini et al. (78)
Solvent
method
Lyophilization Corveleyn and Remon (80–82)
Solvent deposition Monkhouse and Lach (15,16), Kim and
Jarowski (83), Takeuchi (14)
Solvent extrusion/
spheronization
Deasy and Gouldson (84)
Spray drying Corrigan and Holohan (85), Arias et al. (86)
Spray freezing into liquid Hu et al. (87), Rogers et al. (13,88–91)
Steam granulation Cavallari et al. (30), Rodriguez et al. (92)
Supercritical carbon dioxide Senc?ar-Boz?ic? et al. (93), Moneghini et al. (94),
Corrigan and Crean (95)
420 Heng et al.
© 2005 by Taylor & Francis Group, LLC
K30 or copolymer of polyvinylpyrrolidone and vinyl acetate, both of which were acting
as water-soluble carriers. The stability of the extrudates was speculated to be dependent
on the degree of hydrogen bonding between the drug and polymer. An excellent
review by Breitenbach (102) provides an in-depth view of this process in the rapid
release of drug.
A possible modification of this process is the extrusion/spheronization, where
the extrudates are spheronized to form pellets. Pellets are highly spherical granules
and Vervaet et al. (41) found that the use of polyethylene glycol 400 as the watersoluble
carrier and polyethylene glycol 40 hydrogenated castor oil as the surface
active agent was able to enhance the in vitro release of the hydrochlorothiazide pellets
produced. The pellets were subsequently filled in hard gelatin capsule and fed to
eight Caucasian male volunteers. Comparison of mean serum concentration of
volunteers between a commercial product and the pellets indicated that dissolution
and absorption from the pellets were faster (63). A slightly different method was
employed by Mehta et al. (42) for rapid drug release. They prepared a solid dispersion
of nifedipine and copolymer of propylene oxide and ethylene oxide serving as
the surface active agent, and the resultant powder was mixed with other excipients
before extrusion. The nifedipine drug release was reduced to 12 h from 24 h as compared
to when only micronized nifedipine powder was used.
Deasy and Gouldson (84) employed organic solvents, instead of using nonsolvent
extrusion, for wet massing of the powder. The wetted mass was left aside for 12 h
for in situ formation of enteric coprecipitate of nifedipine with hydroxypropylmethylcellulose
phthalate, which is a water-soluble carrier at pH 6.8. However, the
release of the nifedipine was poor as indicated by in vitro dissolution upon changing
the dissolution medium from pH 1.2 to 6.8, to simulate the passage of the pellets
through the gastrointestinal tract. Enhanced release was only achieved by the addition
of sodium lauryl sulfate and sodium starch glycolate, which, respectively, acted
as the surface active agent and superdisintegrant. The combined effects of wetting,
solubilization, and disintegration exposed the enteric coprecipitates for rapid
dissolution at pH 6.8.
4.2.2. Melt Granulation
Melt granulation is a process by which powders are agglomerated with the aid of
a binder, in either a molten state or solid state that melts during the process. The
apparatus of choice is a high-shear mixer, where the temperature of a powder can
be raised above the melting point of a meltable binder by either a heating jacket
or frictional forces generated by the impeller blades. Liquid binding is possible by
the molten binder, thus melt granulation does not require the use of solvents. The
choice of the meltable binder plays an important role in the process. It has to melt
at a relatively low temperature of 50–80C. The use of hydrophilic binder that melts
at a low temperature will aid in the rapid release of drug. Passerini and coworkers
(72) used a copolymer of ethylene glycol and propylene glycol as the surface active
agent, to achieve fast release ibuprofen granules. The improvement in dissolution
could be correlated to the formation of a eutectic mixture between the drug and
binder, hence utilizing the solid dispersion concept (Section 4). Polyethylene glycol
4000 and lactose monohydrate were employed as the hydrophilic meltable binder
and hydrophilic filler, respectively, by Perissutti et al. (46), in the formulation of
rapid release carbamazepine tablets by the melt granulation process. Dissolution
Rapid Release Granulation 421
© 2005 by Taylor & Francis Group, LLC
profiles of the granules were found to be better than physical mixtures of the same
compositions and the pure drug.
Gupta and coworkers (73) employed a variation of the above method. They
first formed a molten mixture of drug and polyglycolized glycerides, which was
the surface active agent employed, before adding this to a surface adsorbent such
as magnesium alumino silicate. The use of a surface adsorbent is to from granules
with good flow and compressibility properties. The granules produced had enhanced
dissolution properties as compared to the physical mixtures or pure drug. Tableting
further improved the release rate. Storage under 40C and 75% relative humidity
also improved the dissolution rate. This was attributed to an increase in the degree
of hydrogen bonding between drug and adsorbant under elevated temperatures. This
was mediated by the solubility of the drug in the polyglycolized glycerides matrix
(74,75).
4.2.3. Roll Mixing
Roll mixing involves the feeding of the powder mix through two rollers. The powder
mass usually includes the drug with one or more water-soluble carriers. Depending
on appropriate processing conditions such as roller pressure or the use of solvent, the
drug may be converted into an amorphous state. Nozawa et al. (76) managed to
improve phenytoin in vitro dissolution release by roll mixing of the drug and polyvinylpyrrolidone
K30, which served as the water-soluble carrier, as compared to the
physical mixtures and coprecipitates. The use of an organic solvent such as ethanol
would further improve the dissolution of the drug. Nozawa et al. (77) employed this
solvent during the roller mixing of nifedipine with two water-soluble carriers, polyvinylpyrrolidone
K30 and polyethylene glycol 1500. They were able to achieve a 400
times higher dissolution rate as compared to nifedipine alone.
4.2.4. Lyophilization
Lyophilization, also popularly known as freeze drying, is a process whereby water is
sublimed from frozen solutions, frozen suspensions, or frozen emulsions; under
reduced pressure and temperature, leaving a dry porous mass of approximately
the same shape and size as the original frozen mass. Lyophilization essentially
consists of three steps, freezing, primary drying, and secondary drying. The materials
are cooled until they are frozen. Primary drying is accomplished under vacuum and
slight heating to remove most of the water by sublimation. The last stage is to
remove sorbed water, which is carried out under elevated temperature. The lyophilization
technology has been applied to the formation of rapid dispersible tablets.
This type of dosage form caters mainly to the pediatric and geriatric patients who
experience difficulties in swallowing. This also caters to active working patients when
there is limited access to water. The ease of administration will decrease the incidence
of noncompliance. The clinical effects of some drugs are enhanced when formulated
into this dosage form as pregastric absorption from the mouth, pharynx, and
esophagus will increase the bioavailability and decrease the side effects. The firstpass
metabolism could also be bypassed. The most notable lyohilization process in
the pharmaceutical industry was called Zydis Technology, which was codeveloped
by Wyeth Laboratories and R. P. Scherer (103). This technology essentially involved
the filling of blister pocket with a suspension of drug followed by freeze drying to
form a porous tablet of sufficient strength before sealing of the blister pack (104).
Thus, this technology could be viewed as a ‘‘macrogranulation’’, whereby each tablet
422 Heng et al.
© 2005 by Taylor & Francis Group, LLC
is considered a ‘‘macrogranule.’’ Instead of lyophilizing a suspension, Corveleyn and
Remon (81) managed to lyophilize an emulsion of hydrochlorothiazide into a tablet.
Disadvantages of lyophilized tablets include poor tablet strength and tendency to be
disturbed by air currents during sealing. This problem was examined by Thapa et al.
(105) and overcome by lyophilization of sealed liquid via holes pierced through
impermeable membrane. The lyophilization technology provides an opportunity
for the formation of amorphous solid solution and enhanced drug release by solid
dispersion concept. This technique had been successfully employed by Corveleyn
and Remon (80,82) to produce lyophilized hydrochlorothiazide tablets displaying
better in vitro dissolution as well as higher bioavailability in human volunteers.
4.2.5. Steam Granulation
Steam granulation is a derivative of wet granulation technique, which involves the
use of steam instead of traditional liquid binder. Instead of spraying the liquid
binder, steam is emitted into the wet massing chamber. The processing time is short
because less moisture is needed as steam has a higher diffusion rate into the powder
mass, achieving a higher distribution of the binder. Another reason is the more
favorable thermal balance during the drying step. This is because after condensation
of the steam in the powder mass, water forms a hot thin film and this requires only a
small amount of energy for vaporization as compared to evaporating the water from
room temperature. This process was proposed by Rodriguez et al. (92) and they
successfully produced diclofenac-polyethylene glycol 4000 rapid release granules
whereby polyethylene glycol 4000 acted as the water-soluble carrier. The authors
compared the granules with granules manufactured from wet and melt granulation
techniques and found that the granules produced were more spherical and had a
larger surface area. The larger surface area was proposed to enhance drug release.
This group of researchers also used steam to complex piroxicam with b-cyclodextrin,
as the complexing agent, forming steam granulated granules simultaneously, thus
carrying out two processes in a single step to produce granules having dissolution
profiles better than the original drug or a physical mixture of the two (30).
5. CONCLUSIONS
The preparation of rapid release granules for capsule or tablet preparation is a multifactorial
situation whereby it has to be approached concurrently, in a number of
ways in order to achieve the desired end result. This approach is dependent on a
number of factors, which could be broadly divided into formulation-related and
granulation-related, but in reality, they are closely related. This chapter aims to
highlight the key factors that the formulator has to consider in order to optimize
the dissolution properties in the preformulation stage to end product so that appropriate
decisions could be made pertaining to the selection of the form of drug to be
presented, additives to be used, granulation methods employed, and processing
variables to be controlled, and the employment of solid dispersion processes.
ACKNOWLEDGMENT
The authors thank Ms. Loh Zhi Hui for assisting in the literature search.
Rapid Release Granulation 423
© 2005 by Taylor & Francis Group, LLC
REFERENCES
1. Amidon GL, Lennerna?s H, Shah VP, Crison JR. A theoretical basis for a biopharmaceutic
drug classification: the correlation of in vitro drug product dissolution and in vivo
bioavailability. Pharm Res 1995; 12(3):413–420.
2. Aulton M. Dissolution and solubility. Aulton AE, ed. Phamaceutics: The Science of
Dosage Form Design. Edinburgh: Churchill Livingstone, 2000:15–32.
3. Noyes AA, Whitney WR. The rate of solution of solid substances in their own solutions.
J Am Chem Soc 1897; 19:930–934.
4. Nernst W. Theorie der reaktionsgeschwindigkeit in heterogenen systemen. Z Phys Chem
1904; 47:52–55.
5. Anderberg EK, Bisrat M, Nystro?m C. Physicochemical aspects of drug release VII. The
effect of surfactant concentration and drug particle size on solubility and dissolution
rate of felodipine, a sparingly soluble drug. Int J Pharm 1988; 47:67–77.
6. Bisrat M, Nystro?m C. Physicochemical aspects of drug release VIII. The relation
between particle size and surface specific dissolution rate in agitated suspensions. Int
J Pharm 1988; 47:223–231.
7. Bisrat M, Anderberg EK, Barnett MI, Nystro?m C. Physicochemical aspects of drug
release XV. Investigation of diffusional transport in dissolution of suspended, sparingly
soluble drugs. Int J Pharm 1992; 80:191–201.
8. Kraml M, Dubuc J, Gaudry R. Gastrointestinal absorption of griseofulvin: II. Influence
of particle size in man. Antibiot Chemother 1962; 12(4):239–242.
9. Levy G. Effect of particle size on dissolution and gastrointestinal absorption rates of
pharmaceuticals. Am J Pharm 1963; 135:78–92.
10. Lin SL, Menig J, Lachman L. Interdependence of physiological surfactant and drug
particle size on the dissolution behavior of water-insoluble drugs. J Pharm Sci 1968;
57(12):2143–2148.
11. Craig DQM. Polyethylene glycols and drug release. Drug Dev Ind Pharm 1990;
16(17):2501–2526.
12. Aungst BJ, Nguyen NH, Taylor NJ, Bindra DS. Formulation and food effects on the
oral absorption of a poorly water soluble, highly permeable antiretroviral agent.
J Pharm Sci 2002; 91(6):1390–1395.
13. Rogers TL, Johnston KP, Williams RO. Solution-based particle formation of pharmaceutical
powders by supercritical or compressed fluid CO2 and cryogenic spray-freezing
technologies. Drug Dev Ind Pharm 2001; 27(10):1003–1015.
14. Takeuchi H, Handa T, Kawashima Y. Enhancement of the dissolution rate of a poorly
water-soluble drug (tolbutamide) by a spray-drying solvent deposition method and
disintegrants. J Pharm Pharmacol 1987; 39:769–773.
15. Monkhouse DC, Lach JL. Use of adsorbents in enhancement of drug dissolution I.
J Pharm Sci 1972; 61(9):1430–1435.
16. Monkhouse DC, Lach JL. Use of adsorbents in enhancement of drug dissolution II.
J Pharm Sci 1972; 61(9):1435–1441.
17. Hersey JA. Ordered mixing: a new concept in powder mixing practice. Powder Technol
1975; 11:41–44.
18. Hersey JA. Preparation and properties of ordered mixtures. Aust J Pharm Sci 1977;
6(1):29–31.
19. Thiel WJ, Sberna FJ. Fluidized bed film coating of an interactive powder mixture to
produce microencapsulated 2-5 mm particles. J Pharm Pharmacol 1986; 38:166–171.
20. McGinity JW, Ku CT, Bodmeier R, Harris MR. Dissolution and uniformity properties
of ordered mixes of micronized griseofulvin and a directly compressible excipient. Drug
Dev Ind Pharm 1985; 11(4):891–900.
21. Nystro?m C, Westerberg M. The use of ordered mixtures for improving the dissolution
rate of low solubility compounds. J Pharm Pharmacol 1986; 38:161–165.
424 Heng et al.
© 2005 by Taylor & Francis Group, LLC
22. Ibrahim H, Sallam E, Takieddin M, Shamat MA. Dissolution characteristics of
interactive powder mixture. Part one: effect of solubility and particle size of excipients.
Drug Dev Ind Pharm 1988; 14(9):1249–1276.
23. Udupa N. Characterization of different crystal forms of ibuprofen, tinidazole and
lorazepam. Drug Dev Ind Pharm 1990; 16(9):1591–1596.
24. Gul MK, Zhu J. Preparation, characterization, and evaluation of physicochemical
properties of different crystalline forms of ibuprofen. Drug Dev Ind Pharm 1998;
24(5):463–471.
25. Vippagunta SR, Brittain HG, Grant DJW. Crystalline solids. Adv Drug Delivery Rev
2001; 48:3–26.
26. Otsuka M, Hasegawa H, Matsuda Y. Effect of polymorphic forms of bulk powders on
pharmaceutical properties of carbamazepine granules. Chem Pharm Bull 1999;
47(6):852–856.
27. Otsuka M, Hasegawa H, Matsuda Y. Effect of polymorphic transformation during
the extrusion-granulation process on the pharmaceutical properties of carbamazepine
granules. Chem Pharm Bull 1997; 45(5):894–898.
28. Yu L. Amorphous pharmaceutical solids: preparation, characterization and stabilization.
Adv Drug Delivery Rev 2001; 48:27–42.
29. Kawashima Y. Solubility and dissolution rate. Gotoh K, Masuda H, Higashitani K,
eds. Powder Technology Handbook. New York: Marcel Dekker, 1997:217–230.
30. Cavallari C, Abertini B, Gonza?lez-Rodr??guez ML, Rodriguez L. Improved dissolution
behaviour of steam-granulated piroxicam. Eur J Pharm Biopharm 2002; 54:65–73.
31. Uekama K, Ikegami K, Wang Z, Horiuchi Y, Hirayama F. Inhibitory effect of 2-hydroxypropyl-
b-cyclodextrin on crystal-growth of nifedipine during storage: superior
dissolution and oral bioavailability compared with polyvinylpyrrolidone K-30. J Pharm
Pharmacol 1992; 44:73–78.
32. Ekenved G, Elofsson R, Solvell L. Bioavailability studies on a buffered acetylsalicylic
acid preparation. Acta Pharm Suecica 1975; 12:323–332.
33. Mason WD, Winer N. Kinetics of aspirin, salicylic acid, and salicyluric acid following
oral administration of aspirin as a tablet and two buffered solutions. J Pharm Sci 1981;
70(3):262–265.
34. Mason WD, Winer N. Influence of food on aspirin absorption from tablets and buffered
solutions. J Pharm Sci 1983; 72(7):819–821.
35. Wang Z, Hirayama F, Ikegami K, Uekama K. Release characteristics of nifedipine from
2-hydroxypropyl-b-cyclodextrin complex during storage and its modification of
hybridizing polyvinylpyrrolidone K-30. Chem Pharm Bull 1993; 41(10):1822–1826.
36. Gordon MS. Process considerations in reducing tablet friability and their effect on in
vitro dissolution. Drug Dev Ind Pharm 1994; 20(1):11–29.
37. Ghosh LK, Ghosh NC, Chatterjee M, Gupta BK. Product development studies on the
tablet formulation of ibuprofen to improve bioavailability. Drug Dev Ind Pharm 1998;
29(3):299–310.
38. Gohel MC, Patel LD. Processing of nimesulide-PEG 400-PG-PVP solid dispersions:
preparation, characterization, and in vitro dissolution. Drug Dev Ind Pharm 2003;
29(3):299–310.
39. Sjo? kvist E, Nystro?m C, Alde?n M. Physicochemical aspects of drug release XIII. The
effect of sodium dodecyl sulphate additions on the structure and dissolution of a drug
in solid dispersions. Int J Pharm 1991; 69:53–62.
40. Mitchell SA, Reynolds TD, Dasbach TP. A compaction process to enhance dissolution
of poorly water-soluble drugs using hydroxypropyl methylcellulose. Int J Pharm 2003;
250:3–11.
41. Vervaet C, Baert L, Remon JP. Enhancement of in vitro release by using polyethylene
glycol 400 and PEG-40 hydrogenated castor oil in pellets made by extrusion/spheronisation.
Int J Pharm 1994; 108:207–212.
Rapid Release Granulation 425
© 2005 by Taylor & Francis Group, LLC
42. Mehta KA, Kislalioglu MS, Phuapradit W, Malick AW, Shah NH. Multi-unit controlled
release systems of nifedipine and nifedipine:Pluronic


F-68 solid dispersions:
characterization of release mechanisms. Drug Dev Ind Pharm 2002; 28(3):275–285.
43. Storey DE. The role of dissolution testing in the design of immediate release dosage
forms. Drug Inf J 1996; 30:1039–1044.
44. Gordon MS, Rudraraju VS, Rhie JK, Chowhan ZT. The effect of aging on the
dissolution of wet granulated tablets containing super disintegrants. Int J Pharm
1993; 97:119–131.
45. Bolhuis GK, Zuurman K, Wierik GHP. Improvement of dissolution of poorly soluble
drugs by solid deposition on a super disintegrant II. The choice of super disintegrants
and effect of granulation. Eur J Pharm Sci 1997; 5:63–69.
46. Perissutti B, Rubessa F, Moneghini M, Voinovich D. Formulation design of carbamazepine
fast-release tablets prepared by melt granulation technique. Int J Pharm 2003;
256:53–63.
47. Levy G, Antkowiak J, Procknal J, White D. Effect of certain tablet formulation factors
on dissolution rate of the active ingredient II: granule size, starch concentration, and
compression pressure. J Pharm Sci 1963; 52(11):1047–1051.
48. Gao JZH, Jain A, Motheram R, Gray DB, Hussain MA. Fluid bed granulation of
a poorly water soluble, low density, micronized drug: comparison with high shear
granulation. Int J Pharm 2002; 237:1–14.
49. Ra?sa?nen E, Rantanen J, Jorgensen A, Karjalainen M, Paakkari T, Yliruusi J. Novel
identification of pseudopolymorphic changes of theophylline during wet granulation
using near infrared spectroscopy. J Pharm Sci 2001; 90(3):389–396.
50. Jorgensen A, Rantanen J, Karjalainen M, Khriachtchev L, Ra?sa?nen E, Yliruusi J.
Hydrate formation during wet granulation studied by spectroscopic methods and
multivariate analysis. Pharm Res 2002; 19(9):1285–1291.
51. Airaksinen S, Luukkonen P, Jorgensen A, Karjalainen M, Rantanen J, Yliruusi J.
Effects of excipients on hydrate formation in wet masses containing theophylline.
J Pharm Sci 2003; 92(3):516–528.
52. Phadnis N, Suryanarayanan R. Polymorphism in anhydrous theophylline-implications
on the dissolution rate of theophylline tablets. J Pharm Sci 1997; 86(11):1256–1263.
53. Bansal P, Haribhakti K, Subramanian V, Plakogiannis F. Effect of formulation and
process variables on the dissolution profile of naproxen sodium from tablets. Drug
Dev Ind Pharm 1994; 20(3):2151–2156.
54. Wu P, Attarchi F, Anderson N, Carstensen JT. Effect of drug solubility in wet granulation
fluids on the dissolution rates of the resulting tablets. Drug Dev Ind Pharm 1989;
15(1):11–16.
55. Chowhan ZT, Amaro AA. Optimization of tablet friability, maximum attainable crushing
strength, weight variation and in vitro dissolution by establishing in-process variable
controls. Drug Dev Ind Pharm 1988; 14(4):1079–1106.
56. Chowhan ZT, Yang IC, Amaro AA, Chi LH. Effect of moisture and crushing strength
on tablet friability and in vitro dissolution. J Pharm Sci 1982; 71(12):1371–1375.
57. Zhang G, Schwartz JB, Schnaare RL. Effect of spheronization technique on drug
release from uncoated beads. Drug Dev Ind Pharm 1990; 16(7):1171–1184.
58. Chowhan ZT. Moisture, hardness, disintegration and dissolution interrelationships in
compressed tablets prepared by the wet granulation process. Drug Dev Ind Pharm
1979; 5:41–62.
59. Ertel KD, Zoglio MA, Ritschel WA, Carstensen JT. Physical aspects of wet granulation
IV-effect of kneading time on dissolution rate and tablet properties. Drug Dev Ind
Pharm 1990; 16(6):963–981.
60. Otsuka M, Gao J, Matsuda Y. Effects of mixer and mixing time on the pharmaceutical
properties of theophylline tablets containing various kinds of lactose as diluents.
Drug Dev Ind Pharm 1993; 19(3):333–348.
426 Heng et al.
© 2005 by Taylor & Francis Group, LLC
61. Chiou WL, Riegelman S. Pharmaceutical applications of solid dispersion systems.
J Pharm Sci 1971; 60(9):1281–1302.
62. Broman E, Khoo C, Taylor LS. A comparison of alternative polymer excipients and
processing methods for making solid dispersions of a poorly water soluble drug. Int
J Pharm 2002; 222:139–151.
63. Vervaet C, Remon JP. Bioavailability of hydrochlorothiazide from pellets, made by
extrusion/spheronisation, containing polyethylene glycol 400 as a dissolution enhancer.
Pharm Res 1997; 14(11):1644–1646.
64. El-Egakey MA, Soliva M, Speiser P. Hot extruded dosage forms, part I technology and
dissolution kinetics of polymeric matrices P. Pharm Acta Helv 1971; 46:31–52.
65. Hu?lsmann S, Backensfeld T, Keitel S, Bodmeier R. Melt extrusion–an alternative
method for enhancing the dissolution rate of 17b-estradiol hemihydrate. Eur J Pharm
Biopharm 2000; 49:237–242.
66. Forster A, Hempenstall J, Rades T. Characterization of glass solutions of poorly watersoluble
drugs produced by melt extrusion with hydrophilic amorphous polymers.
J Pharm Pharmacol 2001; 53:303–315.
67. Perissutti B, Newton JM, Podczeck F, Rubessa F. Preparation of extruded carbamazepine
and PEG 4000 as potential rapid release dosage form. Eur J Pharm Biopharm
2002; 53:125–132.
68. Ndindayino F, Vervaet C, Van Den Mooter G, Remon JP. Bioavailability of hydrochlorothiazide
from isomalt-based moulded tablets. Int J Pharm 2002; 246:199–202.
69. Dittgen M, Fricke S, Gerecke H, Osterwald H. Hot spin mixing-a new technology to
manufacture solid dispersions, part 1: testosterone. Pharmazie 1995; 50:225–226.
70. Ford JL, Rubinstein MH. Formulation and ageing of tablets prepared from indomethacin-
polyethylene glycol 6000 solid dispersions. Pharm Acta Helv 1980; 55(1):1–7.
71. Kinget R, Kemel R. Preparation and properties of granulates containing solid
dispersions. Acta Pharm Technol 1985; 31(2):57–62.
72. Passerini N, Albertini B, Gonza?lez-Rodr??guez ML, Cavallari C, Rodriguez L.
Preparation and characterization of ibuprofen-poloxamer 188 granules obtained by
melt granulation. Eur J Pharm Sci 2002; 15:71–78.
73. Gupta MK, Goldman D, Bogner RH, Tseng YC. Enhanced drug dissolution and bulk
properties of solid dispersions granulated with a surface adsorbent. Pharm Dev Technol
2001; 6(4):563–572.
74. Gupta MK, Bogner RH, Goldman D, Tseng YC. Mechanism for further enhancement
in drug dissolution from solid-dispersion granules upon storage. Pharm Dev Technol
2002; 7(1):103–112.
75. Gupta MK, Tseng YC, Goldman D, Bogner RH. Hydrogen bonding with adsorbent
during storage governs drug dissolution from solid-dispersion granules. Pharm Res
2002; 19(11):1663–1672.
76. Nozawa Y, Mizumoto T, Higashide F. Improving dissolution rates of practically insoluble
drug phenytoin by roll mixing with polyvinyl pyrrolidone. Pharm Acta Helv 1985;
60(5,6):175–177.
77. Nozawa Y, Mizumoto T, Higashide F. Increases in dissolution rate of nifedipine by roll
mixing with polyvinyl pyrrolidone. Pharm Acta Helv 1986; 61(12):337–341.
78. Fini A, Rodriguez , Cavallari C, Albertini C, Passerini N. Ultrasound-compacted and
spray-congealed indomethacin/polyethyleneglycol systems. Int J Pharm 2002; 247:
11–22.
79. Passerini N, Perissutti B, Moneghini M, Voinovich D, Albertini B, Cavallari C,
Rodriguez L. Characterization of carbamazepine-gelucire 50/13 microparticles
prepared by a spray congealing process using ultrasounds. J Pharm Sci 2002;
91(3):699–707.
80. Corveleyn S, Remon JP. Formulation and production of rapidly disintegrating tablets
by lyophilisation using hydrochlorothiazide as a model drug. Int J Pharm 1997;
152:215–225.
Rapid Release Granulation 427
© 2005 by Taylor & Francis Group, LLC
81. Corveleyn S, Remon JP. Formulation of a lyophilized dry emulsion tablet for the
delivery of poorly soluble drugs. Int J Pharm 1998; 166:65–74.
82. Corveleyn S, Remon JP. Bioavailability of hydrochlorothiazide: conventional versus
freeze-dried tablets. Int J Pharm 1998; 173:149–155.
83. Kim KH, Jarowski CI. Surface tension lowering and dissolution rate of hydrocortisone
from solid solutions of selected n-acyl esters of cholesterol. J Pharm Sci 1977;
66(11):1536–1540.
84. Deasy PB, Gouldson MP. In vitro evaluation of pellets containing enteric coprecipitates
of nifedipine formed by non-aqueous spheronization. Int J Pharm 1996; 132:131–141.
85. Corrigan OI, Holohan EM. Amorphous spray-dried hydroflumethiazide-polyvinylpyrrolidone
systems: physicochemical properties. J Pharm Pharmacol 1984; 36:217–221.
86. Arias MJ, Gine?s JM, Moyano JR, Pe?rez-Mart??nez JI, Rabasco AM. Influence of the
preparation method of solid dispersions on their dissolution rate: study of triamterene-
D-mannitol system. Int J Pharm 1995; 123:25–31.
87. Hu J, Rogers TL, Brown J, Young T, Johnston KP, Williams RO. Improvement of
dissolution rates of poorly water soluble APIs using novel spray freezing into liquid
technology. Pharm Res 2002; 19(9):1278–1284.
88. Rogers TL, Nelsen AC, Hu J, Brown JN, Sarkari M, Young TJ, Johnston KP, Williams
RO. A novel particle engineering technology to enhance dissolution of poorly water
soluble drugs: spray-freezing into liquid. Eur J Pharm Biopharm 2002; 54:271–280.
89. Rogers TL, Hu J, Yu Z, Johnston KP, Williams RO. A novel particle engineering
technology: spray-freezing into liquid. Int J Pharm 2002; 242:93–100.
90. Rogers TL, Johnston KP, Williams RO. Physical stability of micronized powders
produced by spray-freezing into liquid (SFL) to enhance the dissolution of an insoluble
drug. Pharm Dev Technol 2003; 8(2):187–197.
91. Rogers TL, Johnston KP, Williams RO. Enhanced aqueous dissolution of a poorly
water soluble drug by novel particle engineering technology: spray-freezing into liquid
with atmospheric freeze-drying. Pharm Res 2003; 20(3):485–493.
92. Rodriguez L, Cavallari C, Passerini N, Albertini B, Gonza?lez-Rodr??guez ML, Fini A.
Preparation and characterization by morphological analysis of diclofenac/PEG 4000
granules obtained using three ifferent techniques. Int. J Pharm 2002; 242(1,2):285–289.
93. Senc?ar-Boz?ic? P, Src?ic? S, Knez Z? , Kerc? J. Improvement of nifedipine dissolution
characteristics using supercritical CO2. Int J Pharm 1997; 148:123–130.
94. Moneghini M, Kikic I, Voinovich D, Perissutti B, Filipovic?-Grc?ic? J. Processing of carbamazepine-
PEG 4000 solid dispersions with supercritical carbon dioxide: preparation,
characterisation, and in vitro dissolution. Int J Pharm 2001; 222:129–138.
95. Corrigan OI, Crean AM. Comparative physicochemical properties of hydrocortisone-
PVP composites prepared using supercritical carbon dioxide by the GAS anti-solvent
recrystallization process by coprecipitation and by spray drying. Int J Pharm 2002;
245:75–82.
96. Ford JL. The current status of solid dispersions. Pharm Acta Helv 1986; 61(3):69–88.
97. Serajuddin ATM. Solid dispersion of poorly water-soluble drugs: early promises,
subsequent problems, and recent breakthroughs. J Pharm Sci 1999; 88(10):1058–1066.
98. Leuner C, Dressman J. Improving drug solubility for oral delivery using solid
dispersions. Eur J Pharm Biopharm 2000; 50:47–60.
99. Sekiguchi K, Obi N. Studies on absorption of eutectic mixture I. A comparison of the
behavior of eutectic mixture of sulfathiazole and that of ordinary sulfathiazole in man.
Chem Pharm Bull 1961; 9:866–872.
100. Tachibana T, Nakamura A. A method for preparing an aqueous colloidal dispersion of
organic materials by using water-soluble polymers: dispersion of beta-carotene by
polyvinylpyrrolidone. Koll Z Polym 1965; 203:130–133.
101. Bates TR. Dissolution characteristics of reserpine-polyvinylpyrrolidone co-precipitates.
J Pharm Pharmacol 1969; 21:710–712.
428 Heng et al.
© 2005 by Taylor & Francis Group, LLC
102. Breitenbach J. Melt extrusion: from process to drug delivery technology. Eur J Pharm
Biopharm 2002; 54:107–117.
103. Virley P, Yarwood R. Zydis—a novel, fast dissolving dosage form. Manuf Chem 1990;
61(2):36–37.
104. Seager H. Drug-delivery products and the Zydis fast-dissolving dosage form. J Pharm
Pharmacol 1998; 50:375–382.
105. Thapa P, Baillie AJ, Stevens HNE. Lyophilization of unit dose pharmaceutical dosage
forms. Drug Dev Ind Pharm 2003; 29(5):595–602.
Rapid Release Granulation 429
© 2005 by Taylor & Francis Group, LLC
15
Continuous Granulation Technologies
Rudolf Schroeder

L.B. Bohle Maschinen und Verfahren GmbH, Ennigerloh, Germany
Klaus-Ju?rgen Steffens
Department of Pharmaceutical Technology, University of Bonn, Bonn, Germany
1. INTRODUCTION
Granulation of powders is a well-known process in the pharmaceutical industry
where primary particles are made to adhere in order to form larger entities called
granules. Primarily, it is used for an improvement of the flow characteristics as well
as for assuring the homogeneity and mechanical stability of the blend during further
production steps. Especially for the compaction of modern actives, granulation is
often unavoidable due to the hydrophobic characteristics of these substances. By
binder granulation the actives are ‘‘coated’’ with a hydrophilic agent, thus establishing
electrostatic interactions between the granulate particles and enabling water to
penetrate the tablet in order to improve disintegration and dissolution behavior.
Also, the granulation process can be used to influence the dissolution behavior of
a tablet, e.g., to gain a drug with sustained release characteristics. In this case melt
agglomeration by hot melt extrusion is a widespread process where a particular excipient
is able to melt and mechanically stabilizes the agglomerates after cooling (1,2). This
excipient embeds the active components in a so-called ‘‘matrix’’ and alters the dissolution
characteristics from immediate release to a diffusion controlled sustained release.
2. COMPARISON OF DIFFERENT MODES OF PROCESSING
To be able to compare the different modes of operation like continuous, semicontinuous,
and batch oriented processing it is necessary to define them and to delimit from
In the case of batch oriented processing the total product volume M is released
after the process duration T whereas for (semi) continuous processing the product output
occurs gradually. The semicontinuous process arises from the division of the total

 Present Address: Abbott GmbH & Co. KG, Ludwigshafen, Germany.
431
each other. A first distinction can be made by the product output per time (Fig. 1).
© 2005 by Taylor & Francis Group, LLC
batch sizeMinto a defined number n of subbatches. The quotient of subbatch size mS
and the process time for one particular subbatch tS is constant for every subbatch.
For really continuous processes neither the total batch size nor a subbatch size
is given by itself. Consecutively, the batch size Mis a result of the product output mC
per time tC. The quotient of mC and tC is constant and the product output is proportional
to the production time. The mathematical descriptions are displayed in
Table 1. Arising from these circumstances some consequences are given, resulting
in both advantages and disadvantages in terms of process control.
For batch oriented processing always a large amount of raw materials is being
granulated; thus, it is quite stable against small deviations of single components in
terms of weighing. Furthermore, it is possible to equalize these deviations by adapting
the process control or the total process duration. Therefore, an oos result due to
these reasons is very unlikely. In the case of continuous processing it is absolutely
unavoidable to carry out the subprocesses in a continuous way and to ensure that
the speed of the subprocess is proportional to the process speed. Otherwise, deviations
in product properties are inescapable. In most cases a correction and thereby
compensation of these deviations is not possible.
The semicontinuous processing takes an intermediate position. On the one
hand, small deviations can be tolerated and equalized; on the other hand, the homogeneity
from one subbatch to another has to be ensured to guarantee the homogeneity
of the whole batch. From the latter point of view the semicontinuous processes
are better comparable to the batch oriented than to the continuous ones.
For pharmaceutical purposes the product properties can be ensured with the
IPC correlating to the needs of the process. Furthermore, the results of the in-process
control (IPC) can be used to influence the process control if speaking of batch
oriented or semicontinuous processing. In the case of continuous processing one will
take care of a robust process control to avoid alterations to a maximum during the
Table 1 Mathematical Descriptions of the Different Types of Processing
Batch oriented Semicontinuous Continuous
Batch size M M? nmS
?n ? 0; 1; 2; . . .?
M ? T
mC
tC  ? T
dm=dt while dm=dt ? constant
Process
duration
T T? ntS T ? M
dm=dt1
Note: For variables, see text.
Figure 1 Product output for batch oriented, semicontinuous, and continuous processing.
432 Schroeder and Steffens
© 2005 by Taylor & Francis Group, LLC
process. Measurement of the process parameters should be for documentation only,
for instance, to show the homogeneity.
The transfer of a process from one machine to another and the simultaneous
increase of the batch size while keeping the product properties, the so-called
‘‘scaling-up,’’ is a challenge in the pharmaceutical industry where most granulation
processes are batch oriented.
For instance in a high-shear granulator the product properties are influenced
by the alteration in the geometry of the bowl and the speed and shape of the impeller
as well as the not negligible influence of material load. The occurring deviations have
to be compensated with alterations in the process control.
This problem does not occur in continuous processing because the batch size
can be adjusted by the process time without interfering in the process control. For
semicontinuous processes the batch size can be adjusted by the number of subbatches
resulting in identical opportunities as for continuous processes. This
approach is limited due to economic aspects with respect to the duration of the whole
process, also taking into account preceding or following process steps.
The transfer of a batch oriented process to a semicontinuous one is possible by
a repetition of the process and a narrow linking of the subprocesses. The homogeneity
and thereby the quality of the product can be stated if the process is always running
under the same conditions. This also includes the cleanliness of the machine
with regard to material buildup arising from the former subbatches. For really continuous
processes this procedure is not possible in most cases.
One big advantage of continuous processing arises from the demand for a precise
and stable process control. Hereby, the IPC can be reduced and the surveillance of the
process can be reduced.Therefore, continuous processes have a reduced need for control
chapter of this book is also a good example of a continuous granulation technique.
3. FLUID BED SYSTEMS
Today there are a large number of suppliers manufacturing continuously working
fluid bed systems for drying and granulation. But as the origin of these systems does
not lie in the pharmaceutical industry, few companies only designed and manufactured
units corresponding to the cGMP standards.
3.1. Process Steps
Along with the traditional batch oriented fluid bed systems also a continuously
working machine has to perform the following steps to produce satisfying granules:
1. Product feeding
2. Product preheating and mixing
3. Granulation (spraying)
4. Drying
5. Cooling
6. Discharging
But in contrast to the batch oriented processes in this case the subprocesses are
to be carried out in a continuous way conjunct with all advantages and risks menrated
from each other as the following explanations will show.
Continuous Granulation Technologies 433
after reaching steady-state conditions (Fig. 18).Roller compaction described in another
tioned in Chapter 2. The zones are not necessarily mechanically and spacially sepa-
© 2005 by Taylor & Francis Group, LLC
3.2. Single-Cell Systems
The first investigations for realizing continuous fluid bed systems in the pharmaceutical
industry were made by adapting batch oriented fluid bed units. The machines
had to be modified to allow continuous powder feeding and product reception
(Fig. 2). The feeding unit could be carried out using a rotary valve. Instead of a standard
product outlet the continuously working fluid bed is equipped with an air
separator to lead back the fine powders into the granulation area.
One major problem of single-cell fluid beds is the fact that all subprocesses of a
granulation process have to take place in one reaction chamber. This means the
wetted powders for granulation and the dried granules are colocated in the same
physical room and can interfere with each other.
To avoid the simultaneous presence of these conflicting process steps multicell
systems were developed allowing a segregation and separation of the wet process
steps from the dry ones.
3.3. MultiCell Systems
With respect to the individual features of every multicell system the general arrangement
of the common aspects like air handling, product treatment, feeding, and discharging
is quite similar. Therefore, the following paragraphs will give a short
overview on the functional basics. Special properties of some exemplary systems will
be discussed later in this text.
Figure 2 Single-cell fluid bed processor.
434 Schroeder and Steffens
© 2005 by Taylor & Francis Group, LLC
3.3.1. Air Handling and Product Flow
As the course of the inlet air conditions during a batch oriented fluid bed process is
not constant, the air supply of a continuous fluid bed is segmented to allow differentiated
air temperatures and humidity. Usually, the number of air chambers is
equal to or more than the number of process steps of a standard fluid bed process.
The boundary between inlet air supply and product chamber is the air distribution
plate. Often this plate is carried out as a gill-plate (Fig. 3). Because of the structure
of the plate directed air jets give a forward transport of the product. But this
only affects the lowest layer of the powder bed. Above, the air stream is scattered
by the powder particles and directed upwards. Therefore, back-mixing occurs and
the flow of the product is more or less based on statistical distributions. Without
any interior fixtures like baffles or walls the particle flow can be expressed with Bateman’s
function (3). Thus, the forwarding of the material is a result of the incoming
raw materials. The main advantages of the gill structure are to be seen in the emptying
of the machine—as the final residue of product is affected by the jets—and the
improved cleaning facilities compared to a flat plate.
The product flow can be improved by vibration or by a sloping arrangement of
the air distribution plate from the in-feed toward the product discharge. The latter
leads to a voluntary flow of the particles taking advantage of the liquid-like behavior
of fluidized powders.
Figure 3 Niro ContipharmTM. (Courtesy of Niro Pharma Systems, Muellheim, Germany.)
Continuous Granulation Technologies 435
© 2005 by Taylor & Francis Group, LLC
The most important aspect of the product flow is a homogenous transport
throughout the whole process without segregation of powder particles. Furthermore,
it has to be ensured that fine powders can be moved as well as larger agglomerates
(lumps). Consequently, the sloped design is to be seen as less favorable as it can
result in a nonhomogenous product flow.
The upper boundary of the product chamber is given by a set of filter cartridges
whereas the specific design follows up different purposes like particle separation and
usability for CIP (cleaning in place). In the latter case the filter cartridges are usually
made of stainless steel.
3.3.2. Spray Nozzle Arrangement
Like in batch oriented fluid bed processors the spray nozzles can be arranged either
as top-spray or bottom-spray. For blends that can be hardly granulated, typically
bottom-spray arrangements are preferred to take advantage of local overwettings.
3.3.3. Feeding and Discharging
Prior to the steady-state phase of the production usually the bed has to be filled to a
certain level, that is, mainly depending on the design of the equipment and the pressure
drop over the bed. The constancy of the bed height can also be ensured via the
pressure drop. Therefore, this value is important for the control of the powder feed
and product discharge.
3.3.4. Continuous Fluid Bed Processor: Niro ContipharmTM
boundary is fixed to the wall and carried out as a torospherical head. The front
boundary of identical shape can be moved hydraulically to open the product chamber
for inspection and cleaning.
The interior between the dished disks is divided horizontally into three sections.
The first section from bottom to top is the inlet air section consisting of three
subdivisions separated from each other by vertical walls. This allows adjusting the
inlet air conditions for preblending, granulation, and drying corresponding to the
product characteristics and process requirements independently. The upper boundary
of the inlet air chamber is given by the air distributor plate carried out as a
groove-like gill-plate with a flattened bottom to keep the powders in the middle of
the plate. Additionally, it contains a set of two-component nozzles to add the granulation
liquid in a bottom-spray arrangement. On the upper end the product chamber
is bordered by exhaust air filters made of stainless steel. The whole rig is prepared
for CIP.
The product in-feed and discharge are carried out as rotary valves to assure a
constant material supply and bed height.
As for most continuous granulation processes the production is divided into
three major steps. In the first step the machine is filled with 50 kg of powder. Consecutively,
the granulation is started for about 1 hr in a batch-like manner with powder
inlet and discharge closed. After the granulation is done the machine is switched
to continuous mode with a product throughput of about 160–240 kg/hr. When all
materials are processed the production chamber is emptied and automatically
cleaned with the integrated CIP unit (4).
436 Schroeder and Steffens
The product chamber is of a round shape with a diameter of 2m (Fig. 3). The rear
© 2005 by Taylor & Francis Group, LLC
3.3.5. Heinen Continuous Fluid Bed Processes
The basic concept of the Heinen fluidized bed system is based on a sectional structure
with separated inlet zones for the drying air (Fig. 4). This means, the drying or fluidizing
air is added section by section in multiple, separate temperature and air zones
as explained earlier.
Mostly, the air distribution plate is carried out as a rectangular gill-plate to
ensure the forwarding of the product, supported by integrated pneumatic vibrators.
The granulation, drying and cooling zones are physically separated from each
other by vertical walls while the orifice between granulation and drying is permanently
open and the cooling chamber can be closed by controllable valves. This
should give a more efficient heat exchange and reduced back mixing of the dried
granules into the wetting area. Furthermore, it reduces the heat exchange between
the hot inlet air for drying and the cold air for cooling.
Usually, the spray nozzles are arranged in top-spray orientation. With the inte-
Because this results in a compact system it is possible to implement widely differing
process steps and temperature sequences adapted to the specific product . This
provides a variety of options for setting and optimizing process parameters.
4. MECHANICAL WET GRANULATION SYSTEMS
The variety of mechanical granulation systems is very highly justified with the manifold
requirements of the formulations on the market. The various system available on the
Figure 4 Heinen continuous fluid bed processor. (Courtesy of Heinen Trocknungstechnologie
GmbH, Varel, Germany.)
Continuous Granulation Technologies 437
Consecutively, it passes the different zones of the granulation process (Fig. 5).
grated cleaning nozzles the complete rig is prepared for CIP (Fig. 6).
© 2005 by Taylor & Francis Group, LLC
market can be differentiated and classified by means of several parameters as product
holding times, energy input, product loss and product hold-up, respectively.
Product holding time or dwell time and energy input cannot be seen as separated
from each other. Only the combination of both enables a prediction of the
granulation characteristics of the machine. Especially in the case of hardly soluble
substances or raw materials with a need for intensive blending either a high holding
Figure 5 Continuous fluid bed processor during operation. (Courtesy of Heinen Trocknungstechnologie
GmbH, Varel, Germany.)
Figure 6 Exemplary PI-diagram of a granulation process. (Courtesy of Heinen Trocknungstechnologie
GmbH, Varel, Germany.)
438 Schroeder and Steffens
© 2005 by Taylor & Francis Group, LLC
time combined with an average energy input or a low holding time together with
high energy input can lead to satisfying granules.
Product holdup (product staying inside the machine after granulation) is of
importance during the drug development phase as it limits the minimum batch size
possible. The larger the batch size the less important is the product holdup of the
machine as it is a constant value independent of the total batch size. Nevertheless,
a low product holdup helps in reducing waste and thus helps in cleaning.
A lot of mechanical continuous granulation equipment can be seen as derived
from single-or twin-screw extrusion systems, mostly improved with special fixtures
inside like choppers or baffles to fulfill the needs of pharmaceutical wet granulation
processes. Screw extrusion systems are treated elsewhere in this book and will not be
discussed at this point.
Other machines like the Schugi mixer–agglomerator (Hosokawa Schugi B.V.,
Lelystadt, The Netherlands) or the Nica M6 turbine mixer and granulator (Aeromatic-
Fielder Ltd., Eastleigh, U.K.) were already mentioned and discussed in the last
edition of this book.
4.1. Semicontinuous Processing: Glatt MulticellTM
This machine takes an intermediate position between fluid bed systems and mechansize
reduction of the manufacturing equipment combined with the tight conjunction
of the single subprocesses (containing product loads addressed as subunits) while
each subprocess is located in a separate process chamber, in order to reduce scaleup
efforts. From raw material to dried granules the product passes through the following
stages:
1. Weighing system for raw materials
2. Wet granulation by small-scale plough-share blender and simultaneous
injection of granulation liquid
3. Milling of the wet granules via a rotary sieve machine
4. Predrying, final drying, and cooling in a three-chambered fluid bed dryer
5. Final milling in a second rotary sieve machine
All process chambers are linked by tubes for contained product transfer. The
transfer itself is realized pneumatically by airflow (5).
The minimum subunit size depends on the minimumcapacity of the plough-share
blender. The manufacturer specifies 1–25 kg/subunit as acceptable range, covered by
three different sizes of equipment. The throughput depends on the process and properties
of the raw material. But as the system should be able to run unattended in a semicontinuous
way the real throughput is less important if economic aspects are not taken
care of.
The machine is prepared for automated CIP with approx. 40 cleaning nozzles
installed.
4.2. High-Throughput Mixer–Granulator: Lo? dige Ploughshare
 Mixers
The main elements of the plough-share blender are a horizontal drum and granulation
chamber and a horizontally rotating shaft, equipped with different transporting
Continuous Granulation Technologies 439
ical granulation equipment (Fig. 7). The philosophy this machine is based on, is the
and blending elements, the shovels (Fig. 8). Usually, the arrangement of the shovels
© 2005 by Taylor & Francis Group, LLC
gives a three-segmented process. The first part of the axle is equipped with a relatively
low amount of shovels and mainly acts as a feeding section to preblend the
mixture and forward it into the granulation section where the blending elements
are carried out as blending sticks. At this point the granulation liquid is added
and forwarding of the material is less important than a sufficient power intake.
Consecutively, the material is entering the postprocessing section where the granules
are formed to their final shape and moved to the discharging orifice.
The revolution speed of the axle is high enough to form a fluid bed-like behavior.
Due to the shape of the blending elements and the material characteristics, nearly
spherical particles can be obtained. Due to the blending characteristic, a disadvantage
of the plough-share blender is the buildup of large agglomerates. High throughputs
from 4.5 to 900 m3/hr can be achieved (6).
At an even higher revolution speed and slight alterations in the arrangement of
flow characteristics of the powder blend inside the machine is very different from
plough-share blenders. Due to the high speed of the axle circular layers of material
are formed, resulting in higher shearing forces and generally an increased power
input. As a result the granules are not as spherical as in plough-share blenders
and the particle size distribution is narrower. Thus, the flow characteristic of the
Figure 7 Multicell semicontinuous granulation processor. The dosing processor and granulation
unit can be seen on the left side. (Courtesy of Glatt GmbH, Binzen, Germany.)
440 Schroeder and Steffens
the blending elements, high-speed continuous mixers can be achieved (Fig. 9). The
© 2005 by Taylor & Francis Group, LLC
product is worse but the dissolution of the granules is improved even due to the more
homogenous blending.
The throughput is comparable to the plough-share blenders and therefore
mainly suitable for the production of large batches.
5. ROLLER EXTRUSION SYSTEM
As a new system in the pharmaceutical market the function and usability of a roller
extrusion system for pharmaceutical wet granulation will be discussed in more detail.
Figure 8 Continuous plough-share blender. (Courtesy of Gebr. Lo?dige GmbH, Paderborn,
Germany.)
Figure 9 High-speed continuous mixer. (Courtesy of Gebr. Lo?dige GmbH, Paderborn,
Germany.)
Continuous Granulation Technologies 441
© 2005 by Taylor & Francis Group, LLC
Roller extrusion is a special type of extrusion not comparable to the classical
single- or twin-screw extrusion systems. In screw extrusion systems the main part
and functional center of the machine is the combination of the screw itself and the
barrel as casing for the screw. The different aims like degassing, ventilation, blending,
melting, and cooling can be made possible by the screw design—in case of a
twin-screw extruder also the cooperation of both intermeshing screws—and the temperature
control of the barrel. Usually, there is no temperature control for the
screws. To achieve a sufficient forward transport it is necessary to take care that
the friction between material and barrel is higher than the friction between material
and screw (14).
For roller extruders or planetary roller extruders these explanations are not
applicable. The design and shape are very different. The arrangement of the spindles
and the barrel as heart of the machine is comparable to a planetary gear but with a
helical gearing of 45. In the center lies the central spindle attached to the main drive.
Around it are the planetary spindles floating freely in the plastic mass, varying in
length and number while they are covered by the barrel. On rotation of the central
spindle the planetary spindles roll on the central spindle and the internally toothed
barrel (Fig. 10). The plastic material floating in the gaps between the planetary spindles
is rolled on the surface of the central spindle and the barrel in very thin layers.
Thus, it is blended on the whole extrusion distance while the friction heat arising
from the homogenization is transferred away into the coolant flowing inside the central
spindle and the barrel. The forward transport of the material is due to the helical
shape of the gearing. It is imaginable that the contact surface area of a planetary
of a planetary extruder is about six times larger (8).
Arising from these considerations the planetary roller extruder seems to be particularly
suitable for the wet and melt agglomeration of pharmaceutical powder
blends. In the following is described a system for continuous wet granulation based
on a planetary extruder and the usage thereof.
5.1. System Description
To take advantage of the manifold opportunities of an extrusion system the machine
is built up modularly to be more flexible also in terms of process control. Therefore,
the subprocesses known from the classical granulation processes as weighing, blending/
granulation, drying, and milling are realized as insulated procedures of an integrated
process line. In addition to the classical steps a cooling step can be
Figure 10 Appearance of a planetary roller extruder. The arrows on the left side indicate the
movements during extrusion.
442 Schroeder and Steffens
roller extruder is very high (Fig. 11). Compared to a twin-screw extruder the area
© 2005 by Taylor & Francis Group, LLC
implemented to enable the manufacturing of granules by hot melt extrusion. The
control of these subprocesses is on the one hand separated from each other; on the
other hand, all controls are integrated into a centralized control and report system.
The system is available for very different outputs from 2 up to 1,000kg/hr,
realized by four different machine sizes only. Due to the continuous processing very
different batch sizes can be granulated with one machine size. For instance, in the
case of a granulator with a nominal output of 10kg/hr, normally the realizable output
is about 2–20kg/hr, leading to batch sizes from 500g in 15min up to 480kg/24 hrs.
5.1.1. Feeding
It is to distinguish between feeding of the powder components and—if applicable—
the liquid components, e.g., binder solutions.
Powder components are usually added by mass-controlled twin-screw feeders
as it is standard in the pharmaceutical industry. But still there are two possible procedures
for the weighing the powder components. One begins with the weighing of
the whole formulation by hand, followed up by an external blending step and ends
with the feeding of the screw feeder with a preblended mixture. The second and more
convenient way is the usage of several screw feeders, one for each raw material in
order to completely avoid the manual weighing. This can be done for feeding rates
of 200g/hr and more. Due to the limited space and the fact that all feeders have to
feed into a single tube the number of feeders is limited. Usually, no preblending has
to be done due to the excellent blending capabilities of the granulation system.
Depending on the flow properties the raw materials immediately fall into the
feed hopper or they are force-fed inside the feeding tube by a screw or agitator of
a different shape. Subsequently, the raw materials are taken over by the feed screw
of the granulation system and moved into the granulator.
The feeding of the liquid components can be done with various pumps of different
outputs and pressures. As a standard a peristaltic pump is used, controlled by
a flow meter based on the Coriolis principle. With respect to the requirements of the
formulation and process, different types of pumps can be used.
Figure 11 Comparison of different extrusion systems.
Continuous Granulation Technologies 443
© 2005 by Taylor & Francis Group, LLC
5.1.2. Granulation
As mentioned earlier, the granulation is realized through a planetary roller extruder.
By the feed screw the material is taken into the granulation section of the planetary
system (Fig. 12).
In the case of binder agglomeration the powder components are plasticized by
the addition of the binder liquid at the very beginning of the granulation process.
Apart from the physical properties of the raw materials the binder liquid is the main
plasticizer and lubricant for achieving a viscosity low enough to allow extrusion. The
homogeneity and granulation are completed at the end of the first stage of the extruder.
There the material has to pass a narrow ring with the function of an intermediate
die. Afterward, the mass re-expands and is—if required—blended with nitrogen
injected through the middle stop ring in extrusion direction. The gas injection leads
to an increase in product porosity and subsequently to a change of product properties
as tableting and dissolution behavior. At the product outlet of the granulation
unit again the mass is pressed through a die-like ring and brought into shape through
a multihole forming unit. If injection of nitrogen is not needed or not wanted, the
second stage of the extruder can be removed reducing the extrusion length and
energy intake to one-half.
Usually, for melt agglomeration processes the addition of liquids is not necessary
because the mass can be plasticized by melting one or more components. The
other facts concerning the process are identical to binder agglomerations.
5.1.3. Drying
Due to the modular design the drying process can be carried out in different ways.
For example, two different types of dryers are described as they are specially adapted
to the extrusion/granulation system.
pipe as dryer inlet is flanged to the granulator directly. By the rotation of the glass
Figure 12 Overview on granulation process consisting of extruder/granulator, microwave
dryer and mill.
444 Schroeder and Steffens
The first one is based on a microwave tube dryer (Figs. 12 and 13). The glass
© 2005 by Taylor & Francis Group, LLC
pipe the product is cut into cylindrical pieces of 0.5–1in. in length, the diameter
depends on the forming unit of the granulator; usual sizes are 0.08–0.2in. As the
sticks fall into the glass pipe they roll due to the round shape. In the same way they
are blended and moved forward as an effect of the rotation of the pipe and the following
material. The drying is done with low pressure of about 0.14psi absolute
under microwave radiation and the addition of a stripping gas for improvement of
the drying time. To achieve precise drying times barriers can be mounted into the
glass pipe (e.g. spirals). The product outlet is carried out as a double-valve lock
due to the vacuum inside.
If drying by microwave energy is not possible or not wanted a contact dryer
can be used instead (Fig. 14). The design is quite similar to the microwave dryer.
Due to the different drying principles the contact dryer has to be larger. The glass
pipe is replaced by a stainless steel pipe with a spiral mounted inside. Both the pipe
and the spiral can be heated. All additional features for drying improvement like
vacuum and stripping gas are implemented.
Figure 13 Sectional sketch of a microwave dryer.
Figure 14 Sectional sketch of a contact dryer.
Continuous Granulation Technologies 445
© 2005 by Taylor & Francis Group, LLC
In case the desired dryer is already present the extruder/granulator can be
combined with nearly every drying method imaginable.
5.1.4. Cooling
Usually, for melt agglomeration processes no drying is needed. The intermediate
product coming out of the extrusion step is warmed up and shows a soft to pastelike
behavior. Therefore, milling of the warm product is not possible and a cooling
step has to be implemented in between to achieve a brittle consistency that allows
milling.
The standard cooling belt is carried out as a stainless steel conveyor while the
cooling is done by pharma air (Fig. 15). The belt swings at the entry to deposit the
warm product with maximum contact surface. This allows short belts even for high
temperature differences between extrudate entry and product outlet.
5.1.5. Milling
the behavior of the material it is to be decided whether to use a flat screen or a rasping
screen. The flat screen allows milling with high throughput for brittle substances
where a higher amount of fines does not matter. If fines have to be avoided to a maximum
or the material to be sieved is very hard or plastic, a rasping mill inset should
be used.
5.1.6. Raw Material Supply and Product Reception
To enable constant product flow it is very important to have a look at the material
supply and its reception after granulation. The design of the supply chain depends
mostly on the process mode. This means in cases of small batches or batches with
a volume of not more than one bin the realization is quite easy. With respect to
Figure 15 Sketch of a cooling belt.
446 Schroeder and Steffens
In most processes milling can be done with a standard conical mill (Fig. 16). From
© 2005 by Taylor & Francis Group, LLC
the powder components the raw material bins can be docked onto the feeding unit
consisting of the screw feeders themselves and a system for the refill of the feeders.
The latter can be realized by a refill screw, for instance. For product reception only
one bin is needed that can be docked at the beginning of the process. It is imaginable
to connect the further production steps like final blending and tableting immediately
instead of a material bin.
If necessary the granulation liquid can be prepared for the whole batch before
manufacturing starts.
For continuous processes over days or weeks the arrangement becomes more
complex. The bins for the raw materials have to be changed in between creating
the need for an intermediate storage of the powders. For establishing this kind of
manufacturing the feeding unit is refilled from a small bin. On the other hand the
small bins are filled from the raw material bins. This assembly gives a chance to
the operators to change the raw material bins in a certain time leading to a constant
material flow. The same solution can be used for the granulation liquid.
During the production itself the material handling even of highly potent substances
is not critical due to the total encapsulation of the single production steps
and the production line as a whole. Therefore, the only locations to take care of
are the inlet of the raw material and the product outlet. Consecutively, this means
that at every connection designed and qualified for separation during the process
a high containment conjunction has to be implemented. With respect to total containment
further modifications are not necessary.
tainment conditions in a building with three floors.
Figure 16 Sectional sketch of the milling unit.
Continuous Granulation Technologies 447
Figure 17 shows the proposal for processing of smaller batches under high con-
© 2005 by Taylor & Francis Group, LLC
5.2. Process Control and Automation
As in nearly every continuous production the whole process has to be seen as divided
into three phases. The first one or start-up phase begins with the starting of the
machine and the raw material supply and ends with the reaching of a stable state
of the whole production line. At that point the product reception can be started
and the production phase is reached. After the required amount of product is
From these facts it is obvious that the yield increases with the batch size.
Depending on the intended output and the process requirements a certain amount
of raw materials is needed to reach the production phase. In some cases this might
be a few grams, in other cases a few kilograms are needed. But it does not depend on
the batch size. The product loss during the end phase depends on the machine size
because in most cases the yield loss is simply the material remaining inside
the machine. Therefore, the most economic way of manufacturing is the use of
Figure 17 Sketch of a batch process with requirements for total containment.
448 Schroeder and Steffens
received the feeders are stopped and the process enters the end phase (Fig. 18).
© 2005 by Taylor & Francis Group, LLC
a relatively small machine running in a real continuous manner; this means 24 hr a
day and 7 days a week.
Due to the fact that only a small amount of raw materials are processed at the
same time it has to be ensured that the instrumentation of the machine allows detecting
of any inhomogeneities during production. And furthermore, the measurement
has to give reliable values for the product quality.
In terms of the feeding process instrumentation is very easy. As it is standard in
the pharmaceutical industry the feeders for the powdery materials are mass-controlled,
recognizing any derivations from the set-point. Assuming the use of a combination
of a pump and a Coriolis-based flow-meter the same explanations are
applicable for the granulation liquid.
Unfortunately, it is more complicated to establish a reliable instrumentation
system for the extrusion/granulation section. As a prerequisite one has to assume
that the product properties are directly linked to the measured ambient conditions.
This means temperature, pressure, and torque. For establishing a stable process it is
not necessary to know the correct values, rather it is sufficient to know that the measured
values are linked with the interesting ones and do not show major fluctuations.
As there will not be any measured value without fluctuations it is necessary to detect
the range thereof and the influence on the product behavior at the upper and lower
limits, especially in terms of process validation.
Temperature probes are installed inside the middle and end stop ring because
these are the locations where the material has to pass a narrow orifice and therefore
the highest temperatures can be expected. It is an indirect measurement of the mass
temperature by measuring the ring temperature very close to the inner surface. Additionally,
the coolant temperatures are detected. Because the ring temperature at the
outer surface is set by the coolant the mass temperature is proportional to the
detected temperature of the ring.
Pressure probes are also installed at both rings. Due to the size and installation
of the rings indirect measurements are useless at these locations. Therefore, the mass
pressure is being detected by strain gauges separated from the mass by a diaphragm.
Figure 18 Typical course of an extrusion/granulation process with startup, production, and
end phase.
Continuous Granulation Technologies 449
© 2005 by Taylor & Francis Group, LLC
As an average value of the mechanical power intake the torque measurement of
the extruder drive is a very important value. Any fluctuation in powder or liquid
dosage or speed of the extruder leads to a fluctuation in torque. Additionally, the
viscosity of the mass is also influenced by its temperature and pressure. Thus, the
torque measurement is also a backbone for the detection of any deviations occurring
during granulation. It is taken from the frequency converter of the main drive.
Therefore, it does not give absolute values but the possibility of recognizing a process
trend that might lead to products being out of specification.
In case of microwave drying it is very important to have a reliable instrumentation.
With a view on the needs of the product the interesting values are the holding
time of the product inside the dryer, the amount of microwave radiation, the amount
of stripping gas and the pressure inside the dryer. Not all products need a very precise
holding time due to the special behavior of microwaves. Normally, microwaves
affect materials that contain a minimum amount of water or other suitable liquids.
When the product is dried microwaves lose the ability to affect the material. Primarily,
the drying and heating will occur where it is needed and wanted (9,11).
For security reasons and with respect to the lifetime of the magnetron the
microwave reflection is detected. The stripping gas can be controlled by pressure
or with a roots pump allowing determining of the gas amount very precisely.
In contrast to the microwave dryer in a contact dryer it is very important to
assure constant material flow and thus constant holding times. Therefore, the contact
dryer is equipped with a spiral inside the drying pipe to forward the material
with a constant speed. Additionally, the temperature of the pipe is determined at
all interesting spots.
For both dryers the measurement of the product temperature at the product
outlet is a matter of course.
For a comprehensive characterization of the product properties like moisture or
active content this application gives the chance of implementing online detection instruments
like NIR spectroscopy, for instance.With this powerful instrument it is possible to
analyze the complete batch in a nondestructive way. Once calibrated, the measurement
can be reduced to a detection of homogeneity that can be carried out very easily.
For automation purposes and deviation control the dependencies of temperature,
pressure, and revolution speed for the granulation unit have to be understood
very well. The influence of different temperatures can be explained quite easily.
Usually, increasing temperatures lead to decreasing viscosities and thus the pressure
and the torque will decrease. With an increasing speed of the central spindles the filling
ratio inside the extruder will decrease and thereby the pressure and the torque,
not taking care of the alternations by increasing temperature caused by a higher friction
inside. To make it more difficult at the beginning of a speed increase the pressure
increases rapidly and then decreases slowly to a lower level due to the fact that a
change of the filling ratio is a slow process.
The pressure is to be seen as a result of speed and temperature and thus not
directly controllable. Even the product temperature can be influenced by two or
more factors, the temperature of the barrel, the speed of the central spindle, and
the physical properties of the materials, mainly the lubricant characteristics and
thereby the inner friction.
Summarizing, the regulation circuits of planetary extrusion systems have to be
carried out very carefully. When the process reaches the production phase usually regulation
circuits are not required if raw material supply and process control can be
maintained in a stable status as the measured values are a result thereof. If deviations
450 Schroeder and Steffens
© 2005 by Taylor & Francis Group, LLC
occur the regulating steps have to be designed according to the intended process by
defining the regulating ranges of each parameter and the dependency algorithms
describing the influences on each other.
5.3. Typical Granulation Processes
5.3.1. Wet Granulation
As there are countless variations of wet granulation processes it is possible only to
demonstrate a small choice of formulations to show the feasibility and limitations
of different raw materials. During testing of the granulation unit various excipients
of higher importance for solid dosage forms were evaluated to know their behavior
under this particular treatment, concentrating more on the behavior with respect to
the extrusion system (Table 2). For these particular formulations, it was possible to
achieve product temperatures below 50C after extrusion. Compression after final
blending led to tablets of a tensile strength of more than 1.0MPa in the examined
range of 5–15kN compression force. By using nitrogen injection it was possible
to improve the compression characteristics of recipe I due to an increase of the gran-
10min; in some cases, it was adjustable by nitrogen pressure.
The granules of all placebo blends were of a round and compact structure,
leading to excellent flowing properties at an LOD of about 2.5%, which was also
the granule moisture before compression.
A different way of evaluating the machine was the granulation of hardly wettable
substances in higher concentrations as models for modern actives. As a first
granulation, caffeine was granulated with 2.5% of povidone resulting in tablets with
Table 2 Placebo Formulations
I II III IV V VI
Granules
Lactose
monohydrate
82.5 80.0 70.0 82.0
Mannitol 72.0 80.0
Corn starch 14.6 15.0 15.0 5.0 10.0
Cellulose 15.0 10.0 15.0 15.0
Providone 2.9 3.0 3.0
HPMC 5.0
Povidone,
cross-linked
5.0
Final blend
Magnesium
stearate
0.5 0.5 0.5 0.5 1.0 1.0
Silicon dioxide
Providone, crosslinked
5.0 2.0 5.0 5.0 5.0
Carboxymethylcellulose
Sodium, crosslinked
3.0
Note: The components of the final blend are to be seen as percentage parts of the dried granules. All values
in % (w/w).
Continuous Granulation Technologies 451
ule porosity (Fig. 19). The disintegration times of the tablets were mostly below
© 2005 by Taylor & Francis Group, LLC
satisfying compression and disintegration behavior and a dissolution t80 value of
about 5min for a tensile strength of 1.0MPa. Hardening after 2 weeks resulted
in useless tablets with a t80 value of nearly 3hr. This could be avoided by addition
of 5% cellulose resulting in stable tablets over a period of 6 months. The components
of the final blend were 3% povidone cross-linked, 0.5% silicon dioxide, and 0.5%
magnesium stearate.
Ibuprofen and mefenamic acid (12), both granulated with 3% povidone and 5%
cellulose and afterward blended with the identical substances as for caffeine, could
be easily compressed at forces of 4kN for ibuprofen and 6kN for mefenamic acid.
Disintegration time was 1.7 and 1.0 min for ibuprofen and mefenamic acid, respectively.
The dissolution t80 value was 6.7 and 17 min, respectively, stable for the examination
period of 6 months. The LOD of the granules before final blending was 1.4%;
disintegration and dissolution behaviors were examined at a tensile strength of
1.0MPa.
A more detailed examination of the ibuprofen granulations as factorial designs
(13) showed satisfying granules for povidone concentrations of 1–4% and cellulose
lose-rich starch type (Eurylon
 7, Roquette) could be classified as suitable for the
granulation. The satisfying ranges were 1–4% of HPMC and 3–8% of starch whereas
Gas injection resulted in worse disintegrating tablets and prolonged dissolution
due to the displacement of the hydrophilic binder from the hydrophobic active by the
hydrophobic gas resulting in higher contact angles between the granules and water
As the results of the aforementioned granulations the following facts could be
found:
 Main advantages of this process are the excellent blending quality and temperature
control associated with low energy requirements for granulation
and drying and a high efficiency of the microwave dryer.
Figure 19 Compression characteristics of recipe I. The second y-axis shows the influence of the
nitrogen pressure on the porosity. Porosity was calculated by the quotient of upward pressure of
the coated granules in silicon oil as a measure of the apparent density and the true density. The
three different charts display the compression characteristics for three compression forces
(5–15 kN).
452 Schroeder and Steffens
concentrations of 3–13% (Figs. 20 and 21). Additionally, HPMC and a special amylower
concentrations gave better results in disintegration and dissolution (Fig. 22).
(Table 3).
© 2005 by Taylor & Francis Group, LLC
 Lactose and mannitol are both suitable as filler ingredients due to their slow
solubility in water.
 The necessary amount of water for granulation mainly depends on the
lubricating properties of the dry blend. In this process the main purpose
of water is lubrication, which enables a very low amount of water of about
7% (w/w) for some blends.
 Ibuprofen could be granulated with active concentrations of maximum 96%
(granules) and 92% (tablets); mefenamic acid could be granulated with
active concentrations of 92% (granules) and 88% (tablets) as an effect of
the excellent blending qualities.
 Cellulose can be used as disintegrant and/or hardening inhibitor in
amounts of 3–7% (w/w).
Figure 20 Dependency of different concentrations of povidone and cellulose on the disintegration
times of tablets containing ibuprofen.
Figure 21 Dependency of different concentrations of povidone and cellulose on the mean
dissolution times of tablets containing ibuprofen.
Continuous Granulation Technologies 453
© 2005 by Taylor & Francis Group, LLC
 Corn starch was gelatinized to a high degree even at low temperatures
resulting in good dispersion qualities for nitrogen at recipe I. Therefore,
starch can be used as a binder added as native powder to the dry blend.
 The process is mainly suitable for the granulation of hardly wettable substances
at high concentrations and the granulation of poor or slowly soluble
substances.
As an indicator for the excellent blending quality and the high shearing forces
brought into the product an SEM picture of a granulate containing mefenamic acid
tals. In standard granulation procedures the binder avoids contact with the active
resulting in insulated spots of binder on the surface of the active crystals. With this
process a forced contact between binder and active is given (small binder bridges in
the picture) leading to tablets of higher hardness. Due to the hydrophilization caused
by the ‘‘binder coating’’ the disintegration and dissolution are very fast at high active
concentrations.

5.3.2. Melt Granulation
Melt granulation processes are the standard applications for extrusion processes in
the pharmaceutical industry. But as this is not a standard machine for manufacturing
pharmaceutics the behavior of the machine during processing and the behavior
of the products are hardly comparable to granules manufactured with a single- or
twin-screw extruder. The main difference is originated in the working principle.
While screw extruders mainly push and shift the material inside the extruder and
blending can be carried out with specially constructed blending elements the planetary
extruder rolls and blends the material on the whole extrusion length. Therefore,
planetary extruders need more lubrication to forward the material but enable excellent
blended mixtures. Usually, the material pressures inside are lower than in screw
extruders and the mass temperatures are more homogenous. As a result from these
considerations formulations transferred from a twin-screw extruder to a planetary
extruder have to be slightly adapted.
extrusion process developed on a corotating twin-screw extruder.
5.4. Differences Compared to Classical Granulation Processes
The main advantages compared to classical granulation methods as high-shear granulation
or fluid bed granulation arise from the continuous way of manufacturing in
contrast to the batch-wise orientation of the classical ones. Usually, the output of a
continuously working machine is expressed in mass per time allowing adjusting of
the batch size by changing the process time or, to be more precise, the production
time of the process (Fig. 18). Therefore, in a certain range the batch size is independent
of the machine size, which makes scale-up for processes developed on a continuous
machine much easier. Even during production the batch size can be chosen
freely depending on the need for that particular product.
The product quality can be ensured due to the fact that at the same time only a
small amount of raw material stays inside the machine for being processed. Furthermore,
it is obvious that the actual energy intake is lower than in a batch oriented
Figure 23 SEM picture of granulate containing mefenamic acid.
Continuous Granulation Technologies 455
Exemplarily, Figure 18 shows the process parameters of a pharmaceutical melt
© 2005 by Taylor & Francis Group, LLC
granulation process for the equivalent batch size. Resulting from the constant product
flow, inhomogeneities in the means of blending or binder distribution can be
detected and avoided by the instrumentation of the granulation unit and detected
at the end of the process by a nondestructive inline measurement tool like NIR spectroscopy
for instance. Therefore, it is not necessary to rely on a small amount of
samples for release; moreover, the whole batch can be analyzed.
For a stable process after reaching the production cycle there is no alternation
in any of the process parameters in a certain range as all parts of the machine are
running at the same time. A separation in several production steps like in known
granulation machines is not given. This makes it very easy to detect deviations
during production.
ACKNOWLEDGMENTS
The authors thank Dr. Daniel Rytz and Dr. Bernhard Luy of Glatt, Klaus Scho?rnbo
?rner of Heinen, Reiner Lemperle of Lo?dige and Harald Stahl of Niro/Aeromatic-
Fielder Division for their support in the preparation of this chapter.
REFERENCES
1. Dittgen M, Kala H, Moldenauer H, Zessin G, Schneider J. Zur pharmazeutischen Technologie
der Granulierung. Pharmazie 1980; 35(4):237–S249.
2. Kristensen H. Particle Agglomeration. Advances in Pharmaceutical Sciences. London:
Academic Press, 1990:221–272.
3. Paul S. Pharmazeutische Eignung von Verfahren zur kontinuierlichen Granulation. Dissertation,
Erlangen-Nu?rnberg, 1996.
4. Product Information, Aeromatic-Fielder.
5. Product Information, Glatt GmbH.
6. Product Information, Gebr. Lo?dige GmbH.
7. Pu?schner H. Wa?rme durch Mikrowellen. Eindhoven, The Netherlands: Philips Tech.
Bibl., 1964.
8. Lu?chtefeld S. Mischer la?sst Planeten kreisen. Chemie Technik 1998; 27(6):66–68.
9. Parikh D, ed. Handbook of Pharmaceutical Granulation Technology. 1st. ed. Vol. 81.
New York: Marcel Dekker, 1997.
10. Schu? tte A. Untersuchungen zur Feuchtgranulierung hydrophober Arzneistoffe am Beispiel
der Mefenaminsa?ure. Dissertation, Bonn, 2001.
11. Stahl H. Trocknung pharmazeutischer Granulate in Eintopfsystemen. Pharm Ind 1999;
61(7):656–661.
12. Schmidt PC, Christin I. Wirk- und Hilfsstoffe fu? r Rezeptur, Defektur und Gro?herstellung.
1st ed. Stuttgart: Wiss. Verl. Ges., 1999.
13. Davies O. The Design and Analysis of Industrial Experiments. 2nd ed. London: ICI,
1978:Longman.
14. Rauwendaal C. Understanding Extrusion. Munich: Carl Hanser Verlag, 1998.
15. Stahl PH. Feuchtigkeit und Trocknen in der pharmazeutischen Technologie. Darmstadt:
Steinkopff, 1980.
16. Dietrich R, Brausse R. Erste Erfahrungen und Validierungsversuche an einem neu
entwickelten GMP—gerechten und instrumentierten Pharma-Extruder. Pharm Ind 50
1988; 50:1179–1186.
17. Fikentscher H, Herrle K. Polyvinylpyrrolidone. Mod Plast 1945; 23(3):157–
161,212,214,216,218.
456 Schroeder and Steffens
© 2005 by Taylor & Francis Group, LLC
18. Goodhart FW, Draper JR, Niger FC. Design and use of a laboratory extruder for pharmaceutical
granulations. J Pharm Sci 1973; 62:133–136.
19. Keleb EI, Vermeire A, Vervaet C, Remon JP. Cold extrusion as a continuous single-step
granulation and tabletting process. Eur J Pharm Biopharm 2001; 52S:359–368.
20. Kibbe AH, ed. In: Handbook of Pharmaceutical Excipients. 3rd ed. London: Pharmaceutical
Press, 2000.
21. Kleinebudde P, Lindner H. Experiments with an instrumented twin-screw extruder using
a single-step granulation/extrusion process. Int J Pharm 1993; 94S:49–58.
22. Lindner H, Kleinebudde P. Use of powdered cellulose for the production of pellets by
extrusion spheronization. J Pharm Pharmcol 1994; 46:2–7.
23. Scheffler E. Statistische Versuchsplanung und auswertung, Eine Einfu?hrung fu? r Praktiker.
3rd ed. Stuttgart: Dt. Verl. Fu? r Grundstoffindustrie, 1997.
24. Sucker H. Methoden zum Planen und Auswerten von Versuchen: I. Factorial Design,
eine Einfu?hrung, Informationsdienst der APV. 1971; 17:52–68.
Continuous Granulation Technologies 457
© 2005 by Taylor & Francis Group, LLC
16
Scale-Up Considerations in Granulation
Y. He, L. X. Liu, and J. D. Litster
Particle and Systems Design Centre, Division of Chemical Engineering,
School of Engineering, The University of Queensland,
Queensland, Australia
1. INTRODUCTION
Scale-up of any engineering process is a great technical and economic challenge.
Scale-up of granulation processes, in particular, is difficult and often problematic
due to the inherently heterogenous nature of the materials used. However, recent
improved understanding of the rate processes that control granulation improves
our ability to do rational scale-up.
There are two situations where process scale-up is needed: (1) commercialization
of newly developed processes and products and (2) expansion of production
capacities in response to increased market demand.
For pharmaceutical applications, the challenge is almost always associated
with new product development. Scale-up in the pharmaceutical industry is unique
in that experiments at laboratory and pilot scale are also required to produce
product of the desired specification for different stages of clinical trials. This gives
additional constraints and challenges to engineers and technologists during
scale-up.
A change in scale invariably impacts on process conditions and, consequently,
on the product quality. For pharmaceutical industries, the Food and Drug Administration
(FDA) ranks the impacts on the drug product arising from changes of
(1). Level 1 is reserved for changes that are unlikely to have any detectable impact on
the formulation quality and performance (2). For all practical purposes, scale-up
should aim to achieve an impact equivalent to or less than Level 1.
In this chapter, we will first consider general scale-up approaches from a
chemical engineering perspective. We will then look specifically at understanding
pharmaceutical granulation scale-up through considering granulation as a combination
of rate processes. Each rate process is affected by changes in process during
scaling, as well as by formulation decisions. Finally, we will present suggestions for
scaling of fluid-bed and high-shear mixer granulation that follow from this
approach.
459
process conditions including production scales into three levels as shown in Table 1
© 2005 by Taylor & Francis Group, LLC
2. GENERAL CONSIDERATIONS IN PROCESS SCALE-UP:
DIMENSIONAL ANALYSIS AND THE PRINCIPLE OF SIMILARITY
It is important to recognize that designing a commercial scale operation via several
stages of scale-up is, in one sense, an admission of failure. If we have a strong understanding
of our processes, then full-scale design can be performed using appropriate
mathematical models, given feed formulation properties and clear required product
specifications. Mature chemical engineering processes, such as distillation, are
designed this way.
However, most solids processing technology do not have this level of maturity
yet. In this case, scale-up studies reduce uncertainties in the design and operation of
the scaled unit most economically. On this basis, the starting point in scale-up must
really be the commercial unit. In theory, once sufficient information for the commercial
unit is known, scale-up can be done by applying the similarity principles from
data collected on a smaller unit. The similarity principle states (3): Two processes
can be considered similar if they take place in similar geometric space and all dimensionless
groups required to describe the processes have the same numerical values.
To establish the necessary dimensionless groups, a systematic dimensional analysis
needs to be carried out where Buckingham P theorem is used to reduce the number
of dimensionless groups (4). Assuming that a process can be described by k
variables, we can express one variable as a function of the other k 
 1 variables, i.e.,
x1 ? f ?x2; x3;K; xk
1? ?2:1?
To conform to the dimensional homogeneity, the dimensions of the variable on
the left-hand side of the equation must be equal to those on the right-hand side.With
some simple mathematical rearrangements, Eq. 2.1 can be transformed into an

Eq. 2.2 is a relationship among k
r independent dimensionless products,
where r is the minimum number of reference dimensions required to describe the
variables. While the Buckingham P theorem itself is straightforward, development
of a dimensionless expression for a process or a phenomenon requires a systematic
dimensional analysis [Ref. 4 for more details]. For most engineering problems, variables
can be divided into three groups: (1) geometric variables, (2) material property
variables, and (3) process variables. The reference dimensions are normally the basic
dimensions such as mass (M), length (L), and time (T).
It is important to note that a systematic dimensional analysis can only be
applied to processes where a clear understanding of the processes is established.
Omission of any important variables of the process will lead to an erroneous
outcome of the dimensional analysis, inevitably causing major problems in scale-up.
Zlokarnik et al. (3) divided the application of dimensional analysis to five general
cases with different levels of understanding in each case:
1. The science of the basic phenomenon is unknown—dimensional analysis
cannot be applied;
2. Enough is known about the science of the basic phenomenon to compile
a tentative draft list—the resulting P set is unreliable;
3. All the relevant variables for the description of the problems are known—
application of dimensional analysis is straight forward;
4. The problem can be described by a mathematical equation—mathematical
functions are better than P relationships, which may help reducing the
number of dimensionless groups.
5. A mathematical solution of the problem exists—application of dimensional
analysis is unnecessary.
Clearly, the more we understand a process or phenomenon, the better we can
scale it up with confidence.
Full application of the similarity principle requires all the relevant P groups to
be measured at the small scale and kept constant during scale-up. Unfortunately,
most industrial processes are very complex with many physical and chemical
phenomena occurring. This leads to a large set of dimensionless groups required
to fully characterize the process. This is particularly the case with processes involving
particulate materials such as granulation. Maintaining all the dimensionless groups
constant on the two scales is very difficult, if not impossible, due to constraints on the
degrees of freedom in variables that can be changed on scale-up. In this case, scale-up
can only be done on the basis of ‘‘partial similarity.’’ That is, not all dimensionless
numbers can be maintained the same on the two scales.
To scale up on the basis of partial similarity, experiments are carried out on a
succession of equipment at different scales and results extrapolated to the final scale.
That is, the scale-up ratio is kept low. With conflicting requirements on the dimensionless
groups during scale-up, a common approach is to maintain one dimensionless
group constant and check the effect of other dimensionless groups on the
dependent variable by varying these dimensionless groups during experimentation.
Once determined, only the dominant dimensionless number will be kept constant
on scale-up. This partial similarity approach is often applied to granulation.
Scale-Up Considerations in Granulation 461
© 2005 by Taylor & Francis Group, LLC
3. ANALYSIS OF GRANULATION RATE PROCESSES
AND IMPLICATIONS FOR SCALE-UP
Many of the required granule product attributes are directly related to the size, size
distribution, and density of the granule product. These granule properties develop as
a result of three classes of rate process in the granulator (Fig. 1):
1. Wetting and nucleation
2. Growth and consolidation
3. Breakage and attrition.
Each of these processes is analyzed in depth in Litster and Ennis (5). In this
section, we will summarize each rate process in turn, particularly highlighting the
main formulation properties and process variables. Wherever possible, we will define
dimensionless groups that can be used in scale-up.
3.1. Wetting and Nucleation
The first step in granulation is the addition of a liquid binder to the powder to form
nuclei granules. Within the granulator, the key region for wetting and nucleation is
the spray zone where liquid binder droplets contact the moving powder surface. The
1. Droplet formation
2. Droplet overlap and coalescence at the bed surface
Figure 1 A classification of granulation rate processes. (From Ref. 5.)
462 He et al.
nucleation process is considered to consist of four stages (Fig. 2):
© 2005 by Taylor & Francis Group, LLC
3. Drop penetration into the bed by capillary action
4. Mechanical dispersion of large clumps within the powder bed (only
applicable to mixer granulators).
Poor wetting and nucleation lead to broad granule size distributions and poor
distribution of the liquid binder, which increases substantially the chances of poor
drug distribution. Despite the action of other rate processes, the broad size distributions
and poor liquid distribution often persist throughout the granulation.
For ideal nucleation, the granulator should operate in the drop-controlled
regime. Here, each drop which hits the powder bed penetrates into the bed to form
a single nucleus granule. There is (almost) no drop overlap at the bed surface and
mechanical dispersion of large, wet powder clumps is unnecessary.
To predict the required conditions for drop-controlled nucleation we must
understand
1. The thermodynamics and kinetics of drop penetration, largely controlled
by formulation properties
2. The flux of drops onto the bed surface, largely controlled by process parameters.
The drop penetration time tp can be estimated using a model, which considers
the rate at which liquid flows into the pores in the powder surface under capillary
action (6):
tp ? 1:35
V2=3
0
e2
effReff
m
gLVcosy ?3:1?
where V0 is the drop volume, m is the liquid viscosity, and gLV cosy is the adhesive
tension between the liquid and the powder. The effective pore size Reff and porosity
eeff of the powder bed are given by
Reff ?
fd32
3
eeff
?1 
 eeff ? ?3:2?
eeff ? etap?1 
 e ? etap? ?3:3?
Figure 2 Wetting and nuclei formation in the spray zone of a granulator.
Scale-Up Considerations in Granulation 463
© 2005 by Taylor & Francis Group, LLC
where j is the particle sphericity, d32 is the specific surface mean particle size, e is the
loose packed bed porosity, and etap is the tapped bed porosity.
For drop-controlled nucleation, the drop penetration time must be small compared
to the bed circulation time tc before that section of powder passes again
through the spray zone, i.e., the dimensionless penetration time should be small:
tp ?
tp
tc
< 0:1 ?3:4?
To avoid drop overlap on the bed surface and caking of the powder, the dimensionless
spray flux c must also be kept small. ca is the ratio of the rate of production
of drop projected area by the nozzle to the rate at which powder surface area passed
through the spray zone and is defined as
ca ?
3V
2Add ?3:5?
Figure 3 shows how the nuclei granule size distribution broadens as the spray
flux increases. For drop-controlled nucleation, the dimensionless spray flux should
be kept <0.2. For ca >0.7 the surface of the powder bed in the spray zone is effectively
caked.
controlled nucleation is achieved only when both ca and tp
an example of full granulation data from a 25 L fielder mixer granulator on this type
of regime map. The granule size distribution is much narrower when nucleation is
kept in the drop-controlled regime (lower left-hand corner). This illustrates that poor
nucleation usually results in broad final granule size distributions despite the impact
of other processes occurring in the granulator.
3.2. Growth and Consolidation
Granule growth is very complex. The key question in establishing growth behavior
is: Will two granules which collide in a granulator stick together (coalesce) or
Figure 3 Effect of powder velocity on nuclei size distribution for lactose with water at 310
kPa. (From Ref. 6.)
464 He et al.
We can represent the nucleation behavior in a regime map (Fig. 4). Dropare
low. Figure 5 shows
© 2005 by Taylor & Francis Group, LLC
Figure 4 Nucleation regime map. For ideal nucleation in the drop-controlled regime, it must
have (1) low ca and (2) low tp. In the mechanical dispersion regime, one or both of these
conditions are not met, and good binder dispersion requires good mechanical mixing.
Figure 5 Nucleation regime map in 25 L Fielder mixer at 15% liquid content. Merck lactose
with water and HPC as liquid binders. (From Ref. 6.)
Scale-Up Considerations in Granulation 465
© 2005 by Taylor & Francis Group, LLC
rebound? To answer this question it is useful to look at two extreme cases, which
cover most applications:
Deformable porous granules: These granules are typically formed by a nucleation
process described earlier with the drop size of the same order, or
larger, than the powder size. Most of the liquid in the granule is contained
in the pores between particles in the granule and held there by
capillary action. For successful coalescence, this liquid must be made
available at the contact point between colliding granules. This model is
often suitable for drum mixer granulation.
Near-elastic granules: Here, the wetted granule is considered as a nearly elastic
sphere with a liquid layer on the surface. This is a good model for cases
where the drop size is much smaller than the granule size and the granulator
has simultaneous drying. This model is often suitable for fluid-bed
granulation.
The different growth modes for deformable porous granules can be represented
on a regime map (Fig. 6). For growth to occur by coalescence, the liquid content
needs to be large enough to provide 85–105% saturation of the pores in the granule.
Granules that are weak, i.e., form large contact areas on collision, fall into the steady
growth regime. When two granules collide, a large contact area is formed and liquid
is squeezed into the contact area, allowing successful coalescence. In this regime,
granules grow steadily and the growth rate is very sensitive to moisture content
Strong granules do not deform much on collision and granules rebound, rather
than coalesce. However, as granules consolidate slowly, eventually liquid is squeezed
into the granule surface and this liquid layer causes successful coalescence. This is the
Figure 6 Granule growth regime map. (From Ref. 7.)
466 He et al.
(Fig. 7A).
© 2005 by Taylor & Francis Group, LLC
induction growth regime (Fig. 7B). At lower moisture contents, nuclei granules form
and consolidate. Some growth by layering may occur, but there is insufficient liquid
for growth by coalescence. This is the nucleation regime. Very weak granules simply
fall apart and cannot sustain growth. This is the crumb regime.
There are two dimensionless groups that dictate the growth behavior, the
Stokes deformation number Stdef and the maximum pore saturation smax, which
are defined as
Stdef ?
rgU2c
2Y ?3:6?
Smax ?
wrs?1 
 emin?
r1emin ?3:7?
where rg, rs, and rl, are the granule, particle, and liquid densities, respectively; Uc is
the effective granule collision velocity; Y is the granule yield strength; w is the liquid
content (kg liquid/kg dry powder); and emin is the minimum granule porosity after
complete consolidation.
Understanding where your system sits on the growth regime map is important
for troubleshooting and scale-up. Granules that grow in the induction regime are
easy to scale with respect to granule size, provided that the induction time is not
exceeded. However, granule density often changes with scale because consolidation
kinetics are important and these kinetics can change with scale. On the other hand, in
the steady growth region it is difficult to control granule size, but granule density
quickly settles to a minimum value and varies little with process parameters.
To make effective use of the granulation regime map we need reasonable estimates
of the effective collision velocity Uc (controlled by process conditions) and
of the average and maximum collision velocities for different process equipment. In
Figure 7 Coalescence growth modes for deformable granules.
Scale-Up Considerations in Granulation 467
dynamic yield stress Y (a function of formulation properties). Table 2 gives estimates
© 2005 by Taylor & Francis Group, LLC
high-shear mixers, the difference between the average and maximum collision
velocities can be very large.
The dynamic yield stress of the granule matrix is a function of strain rate due
to the contribution of viscous dissipation to the granule strength. Therefore, it is
dangerous to use static strength measurements to predict performance in the granulator.
Iveson et al. (8) show how dynamic yield stress can be estimated from peak
flow stress measurements in a high-speed load frame. They were able to correlate
data for different formulations and strain rates in a single line when plotted as the
dimensionless peak flow stress (Str) vs. the capillary number (Ca) (Fig. 8). This line
can be fitted by a simple empirical equation of the form
Str ? k1 ? k2Can ?3:8?
where Str?spkdp/g cos y is the dimensionless peak flow stress, Ca?m_edp/g cos y is
the ratio of viscous to capillary forces, spk is peak flow stress, _e is the bulk strain rate,
and y is the solid–liquid contact angle. k1 gives the static strength of the pellets. k2
determines the transition between strain-rate independent and strain-rate dependent
behavior. n is an exponent which gives the power law dependency of the flow stress
on viscosity and strain rate. The best-fit value of n was found to be 0.58  0.04 and
the transition between strain-rate independent and dependent flow stress occurred at
Ca  10
4.
Table 2 Estimates of Uc for Different Granulation Processes
Type of granulator Average Uc Maximum Uc
Fluidized beds (6Ubdp)/db (6Ubdp)/dbd2
Tumbling granulators odp oDdrum
Mixer granulators oidp, ocdp oiD, ocDc
Source: From Ref. 5.
Figure 8 Dimensionless flow stress vs. capillary number for widely sized 35 mm glass ballotini
with six different binders.
468 He et al.
© 2005 by Taylor & Francis Group, LLC
The rate of consolidation of granules can also be correlated with Stdef in the
form
kc ? bcexp?a  Stdef? ?3:9?
where bc and a are constant, kc is the consolidation rate constant for a first-order
consolidation equation of the form
e 
 emin
e0 
 emin ? exp?
kct? ?3:10?
For near-elastic granules the conceptual model originally developed by Ennis
et al. (9) considers the collision between two near-elastic granules each coated with
a layer of liquid (Fig. 9). In this case the key dimensionless group is the viscous
Stokes number Stv
Stv ?
4rgUcdp
9m ?3:11?
Stv is the ratio of the kinetic energy of the collision to the viscous dissipation in
the liquid layer. Successful coalescence will occur if Stv exceeds some critical value
St and we can define three growth regimes as follows:
Noninertial growth (Stv,maxSt): The kinetic energy in most or all collisions
exceeds viscous dissipation in the liquid layer. There is no coalescence.
Granule growth will only occur by the successive layering of new material
in the liquid phase (melt, solution, or slurry) onto the granule.
Figure 10 shows an example of granule growth in a fluidized bed where the
growth regime changes as the granules grow. Glass ballotini is grown with two liquid
binders of different viscosity. Initially, both systems grow steadily at the same rate
(noninertial regime). When the granule size reaches 800 mm, the PVP bound granule
growth begins to slow down indicating a transition to the inertial growth regime
(only some collisions are successful). Finally, the PVP granules level off at a maximum
size of 900 mm showing transition to the coating regime, where no granule
collisions are successful. In contrast, the more viscous CMC-M granules grow steadily
throughout the 8 hr experiment, i.e., they remain in the noninertial regime for
the whole experiment.
3.3. Breakage and Attrition
Breakage and attrition really cover two separate phenomena:
1. Breakage of wet granules in the granulator
2. Attrition or fracture of dried granules in the granulator, drier, or in subsequent
handling.
Breakage of wet granules will influence and may control the final granule size
distribution. It is only an important phenomenon for high-shear granulators. Wet
granule breakage is much less studied than nucleation and growth. There is very little
Figure 10 Growth of glass ballotini granules in a fluidized bed with binders of different
viscosity. (From Ref. 9.)
470 He et al.
© 2005 by Taylor & Francis Group, LLC
quantitative theory or modeling available to predict conditions for breakage, or the
effect of formulation properties on wet granule breakage.
Tardos et al. (10) considered that a granule breaks if the applied kinetic energy
during an impact exceeds the energy required for breakage. This analysis leads to a
Stokes deformation number criteria for breakage:
Stdef > Stdef ?3:12?
where Stdef is the critical value of Stokes number that must be exceeded for breakage
schematics of the failure mode of different formulations in dynamic yield strength
measurements. Failure behavior varies widely from semibrittle behavior at low capillary
numbers to plastic failure at high capillary numbers. We expect a purely plastic
granule to smear rather than break when its yield stress is exceeded. At high impeller
speeds such materials will coat the granulator wall or form a paste. Semibrittle granules
will break at high-impact velocity giving a maximum stable granule size or a weak
crumb. Nevertheless, Eq. 3.12 provides a good starting point for quantifying wet
granule breakage.
Dry granule attrition is important where drying and granulation occur simultaneously
(e.g., in fluidized beds) and in subsequent processing and handling of the
granular product. We can consider dry granule breakage as brittle or semibrittle
phenomena. The key granule properties that control the breakage are the granule
fracture toughness Kc and the flaw or crack size c in the granule. Kc is set by formulation
properties, while c is closely related to granule porosity controlled by the
consolidation process in the granulator.
Dry granule breakage usually results in production of fines by wear, erosion, or
attrition brought about by diffuse microcracking. Within a fluid bed there are a large
number of low-velocity collisions between particles as they shear past each other.
This process is analogous to abrasive wear. For abrasive wear of agglomerates,
the volumetric wear rate V is given by (11)
V ?
d1=2
i
A1=4K3=4
c H1=2
P5=4l ?3:13?
where di is indentor diameter, P is applied load, H is the hardness of the particles, l is
wear displacement of the indentor, and A is the apparent area of contact of the
indentor with the surface. The number and relative velocity of the collisions depend
on the number of bubbles in the bed and hence the excess gas velocity (u 
 umf). The
applied pressure in a fluid bed depends on bed depth. Thus, the attrition rate Bw in
a fluidized bed granulator is
Bw ?
d1=2
0
K3=4
c H1=2
L5=4?u 
 umf? ?3:14?
where d0
shows the attrition rates of several formulations in a fluidized bed with a direct
correlation between attrition rate and the material properties grouping in Eq. 3.13
and (3.14).
Note that Eq. 3.13 and (3.14) only hold for breakage via a wear mechanism.
For attrition during impact or compaction, there are different dependencies of the
attrition rate on the materials properties [Ref. 5 for more details].
Scale-Up Considerations in Granulation 471
to occur. However, this model is probably an oversimplification. Figure 8 shows
is the distributor hole orifice size and L is the fluidized bed height. Figure 11
© 2005 by Taylor & Francis Group, LLC
3.4. Implications for Scale-Up
Table 3 summarizes the key controlling dimensionless groups for the rate processes
described earlier and the main process parameters and formulation properties that
impact on these groups.
In addition to these groups, there are dimensionless groups to describe: (1) the
geometry of the equipment and (2) the flow of the powder and granules in the granulator.
Both these classes of controlling groups are very equipment dependent.
For scale-up using full dimensional similarity, all these dimensionless groups
need to be held constant. This is normally impossible due to the small number of
Figure 11 Erosion rates of agglomerate materials during attrition of granules in a fluidized
bed. (From Ref. 12.)
Table 3 Summary of Controlling Groups for Granulation Rate Processes
Rate process Controlling groups
Key formulation
properties
Key process
parameters
Wetting and
nucleation
Dimensionless spray
flux c,
dimensionless
penetration time tp
tp, m, g cos y, dp,
e, etap
V&, A&(influenced by
nozzle design and
position, no. of
nozzles, and powder
flow patterns)
Growth and
consolidation
Stokes deformation
number Stdef,
viscous Stokes
number Stv, liquid
saturation S
Y, rg, m, g cos y,
dp, etap
Uc (influenced by
powder flow
patterns; see
Table 2)
Attrition and
breakage
Stokes deformation
number Stdef
Kc, H, Y, rg, m,
g cos y, dp, etap
Uc, L, u 
 umf
472 He et al.
© 2005 by Taylor & Francis Group, LLC
degrees of freedom and the large number of constraints. In particular, for regulatory
reasons it is usually not possible to change formulation properties during scale-up
except during the very early stages of process development. This leaves only a
relatively small number of process parameters as degrees of freedom.
Therefore, a partial similarity approach for scale-up is recommended. The
general steps are
1. Maintain similar geometry throughout the scale-up process. For most of
the pharmaceutical granulation equipment, this can be achieved from
either 10 or 25 L nominal batch size to full scale. Be wary, however, in
some cases key geometric parameters do vary with scale on a particular
design, e.g., relative chopper size, relative fill height. Manufacturers should
be lobbied hard to provide geometrically similar designs at all scales.
2. Set key dimensionless groups to maintain similar powder flow during scaleup.
In particular, avoid changes of flow regime during scale-up that make
maintaining granule attributes during scale-up impossible.
3. Use your experience and an understanding of your process to decide which
product attributes are the most important, and which granulation rate
process is most dominant in controlling these attributes. This is difficult
to do a priori, but with good characterization of your formulation and process,
the regime map approaches described earlier are very useful.
4. Use your remaining degrees of freedom in the choice of process parameter
values to keep the most important one or two rate process dimensionless
groups constant.
This approach is most easily demonstrated on a particular type of equipment
(Section 5 for fluidized beds and Section 6 for high-shear mixers).
4. SCALE-DOWN, FORMULATION CHARACTERIZATION, AND
FORMULATION DESIGN IN PHARMACEUTICAL GRANULATION
In the development of a new pharmaceutical product, important decisions about the
manufacturing process are made with a few grams or tens of grams of formulation.
To provide, the drug product for clinical trials and the final design at large scale,
granulations are often conducted at several laboratory-and pilot scales as well. Typical
nominal batch sizes are 1, 10, 25, and 65 L scaling to commercial operation at 300
or 600 L.
Small-scale granulations up to 1L are often done by hand, and certainly performed
in equipment that is very different from the equipment that will be used for
scales from 10 L and larger. At this level, the general scale-up approach described in
Section 3.4 does not hold. How do we scale down to make the best use of data from
granulation of these small amounts?
scaling up from 10 L, formulation properties cannot be varied. Only process
parameters can be used to keep key granule attributes in the target range. Therefore,
very small-scale-experiments should target major formulation design decisions
and attempts to mimic completely different geometries at larger scale should be
avoided.
of these require relatively small amounts of material and can be measured at this
Scale-Up Considerations in Granulation 473
The key is to consider granulation as a particle design process (Fig. 12). During
Table 3 summarizes key formulation properties that should be measured. Most
© 2005 by Taylor & Francis Group, LLC
level. By using these data to help estimate key controlling groups for the granulation
rate processes in the larger-scale equipment, appropriate changes to the formulation
can be made. This avoids major headaches at a later stage. Good communication
between the technologists who design the formulations and the process engineers
who scale the process and transfer the product to manufacturing is an essential part
of this paradigm.
Some of the questions that can be addressed at this stage of formulation development
and scale-up include:
1. Wetting and nucleation
 Contact angle: Are the active and all the excipients easily wetted by the
liquid binder?
 Drop penetration time: Is the liquid phase too viscous, or the particle
size too small to achieve fast drop penetration?
2. Growth and consolidation
 What is the dynamic yield stress of the formulation?
 How much liquid binder is required for granule growth?
 What is the likely growth regime?
 What range of granule density is likely?
3. Attrition and breakage
 Will extensive granule breakage occur in the granulator?
 What are dry granule strength (fracture toughness) and porosity?
 Are attrition and dust formation during handling likely?
4. Downstream processing issues
 Does the formulation compress well for tableting?
 Can the desired dissolution profiles be met?
Details of how to measure key formulation properties are described in more
detail in Litster and Ennis (5).
Figure 12 Granulation as an example of particle design. Both formulation properties and
process parameters influence granule attributes.
474 He et al.
© 2005 by Taylor & Francis Group, LLC
5. SCALE-UP OF FLUIDIZED BED GRANULATORS
There are many different variations of fluidized bed granulators including bubbling
fluidized bed, draft tube fluidized beds, and spouted beds (5). However, in this section,
we limit ourselves only to the scale-up of the most commonly used fluidized bed
granulator, i.e., bubbling fluidized bed granulator. In particular, as most fluidized
bed granulators used in the pharmaceutical industry are operated in batch mode,
we will concentrate on the scale-up of batch bubbling fluidized bed granulators.
5.1. Bed Hydrodynamics and Scale-Up
Particle growth in a fluidized bed is closely related to the particle mixing and the flow
pattern in the bed. This dictates that the hydrodynamics of the scaled bed should be
the same as the small unit, i.e., hydrodynamic similarity. Basic fluidized bed hydro-
In bubbling fluidized beds, bed expansion, solids mixing, particle entrainment,
granule growth, and attrition are intimately related to the motion of bubbles in the
bed (Fig. 13). The volume flow rate of bubbles in the bed Qb, the bubble size db, and
the bubble rise velocity ub are the key parameters that characterize the bubbly flow.
There are numerous correlations relating these bubble parameters to process condi-
In general, Qb is a strong function of the excess gas velocity u 
 umf. Growing granules
are usually Geldart type B powders, or perhaps type A powders at the start of
the batch. For group B powders, db increases with bed height and is a function of
Figure 13 Effect of bubbles on (A) solids mixing and (B) solids entrainment. (From Ref. 13.)
Scale-Up Considerations in Granulation 475
dynamics are described in Chapter 9.
© 2005 by Taylor & Francis Group, LLC
tions [see, for example, Kunii and Levenspiel (13) and Sanderson and Rhodes (14)].
excess gas velocity. The bubble-rise velocity is directly related to db. For the simplest
models for group B powders we can write
Qb ? ?u 
 umf ?pD2
F ?5:1?
ub ? 0:71 ffiffiffiffiffiffiffi gdb p ?5:2?
db / ?u 
 umf ?0:4L0:8 ?5:3?
Thus, the excess gas velocity u 
 umf and the bed height L are the key process
parameters that control bubbling behavior in the bed.
Several rules exist for scaling up a bubbling fluidized bed under the condition
of hydrodynamic similarity. Fitzgerald and Crane (15) proposed that the following
dimensionless numbers be kept constant during scale-up:
 Particle Reynolds number based on gas density (dpurG)/m
 Spolid particle to gas density ratio rs/rG
 Particle Froude number u/(gdp)0.5
 Geometric similarity of distributor, bed, and particle L/dp
where dp is the particle diameter, u is the fluidization velocity (superficial gas velocity),
m is the viscosity of fluidizing gas, rG is the density of fluidizing gas, g is the
gravitational acceleration, and L is the fluidized bed height.
In this approach, experiments on the smaller scale are performed with model
materials, i.e., model gas (different from the larger-scale one) and model solid particles
(different particle density, size, and size distribution). For readers interested in
following Fitzgerald’s scale-up rules, a detailed calculation procedure can be found
in Kunii and Levenspiel’s book (13), illustrated with an example.
In a series of publications, Glicksman et al. (16–18) divided the scale-up into
two regimes, namely, inertia-dominated and viscous-dominated flow regimes. In
viscous-dominated flow regime, where particle Reynolds number based on fluid
density is 4, i.e., when, the (dpurG/m4, dimensionless numbers that need to be
kept constant are
u
?gdp?0:5 ;
dpurs
m
;
L
dp
;
DF
dp
; f; particle size distribution; bed geometry ?5:4?
where DF is the fluidized bed diameter.
In contrast, in inertia-dominated flow regime, (dpurG)/m	400, scale-up of the
process demands that the following dimensionless numbers are kept constant:
u
?gdp?0:5 ;
rG
rs
;
dpurs
m
;
L
dp
;
DF
dp
; f; particle size distribution; bed geometry ?5:5?
In the intermediate region, where 4(dpurG)/m400, both the viscous and the
inertial forces are important to the fluid dynamics, and all the dimensionless numbers
for the two regions mentioned earlier will need to be kept constant during scale-up, i.e.,
rsrGd3
pg
m2 ;
u
?gdp?0:5 ;
rG
rs
;
L
dp
;
DF
dp
; f; particle size distribution; bed geometry
?5:6?
476 He et al.
© 2005 by Taylor & Francis Group, LLC
Experimenting with only ambient air and particles made of the same material
but different sizes, Horio et al. (19,20) developed what has been lately defined as the
simplified scaling law. They demonstrated that, with similar bed geometry (ratio of
bed height to diameter), using particles of different mean sizes but the same distribution
characteristics, and operating the bed in proportional superficial gas velocities
would ensure that the hydrodynamic conditions of the two beds remain similar.
Expressed in mathematical terms
u2 
 umf;2 ? ffiffiffiffi m p ?u1 
 umf;1?
umf;2 ? ffiffiffiffi m p umf;1
m ?
L2
L1
bed geometry
?5:7?
where 1 and 2 refer to the small-scale and large-scale beds, respectively.
Experimental results from Roy and Davidson (21) suggest that when (dpurG)/
m< 30, the criteria of Horio et al. (19,20) are sufficient to give similarity in behavior.
However, when (dpurG)/m > 30, the more restrictive approach of Fitzgerald and
Crane has to be used.
Unfortunately, few of the previously mentioned scaling rules for bubbling
fluidized beds have been strictly followed for the scaling up of fluidized bed granulators.
This is largely because the scaling rules require model materials to be
used at the smaller scale, whereas in pharmaceutical granulation, the formulation
is unchanged during scale-up. However, the simplified rules presented by Horio
combined with our understanding of granulation rate processes do provide some
guide.
5.2. Granulation Rate Processes in Fluidized Beds
nucleation, and layered growth occur in the spray zone of the fluidized bed. Most
consolidation and coalescence also occur in or near the spray zone because fluidized
bed granulators are also driers. The drying process ‘‘freezes’’ the granule structure
and prevents further growth. Thus, good design of the spray zone is very important
and liquid flow rate is a critical process parameter. Beds should be designed to keep
dimensionless spray flux low (drop-controlled regime). If this is not done, the formation
of large clumps leads to rapid wet quenching and defluidization with likely
closely related to the design of the spray zone. The x-axis variable (spray surface area
per mass in granulator) is closely related to our definition of dimensionless spray
flux.
Due to the simultaneous drying, our consolidation and growth models for nearelastic
granules is usually appropriate and the viscous Stokes number is a key controlling
group [Eq. (3.11)]. This model predicts that in batch granulation, granules will
grow toward a maximum size corresponding to the critical Stokes number and transivelocities
are set by the flow of bubbles in the fluid bed and are a function of bubble
Fluidized beds produce porous granules because the consolidation time is
limited to granule drying time, which is of the order of seconds, rather than minutes.
Scale-Up Considerations in Granulation 477
Figure 14 shows the rate processes occurring during fluidized granulation. Wetting,
loss of the batch. Figure 15 shows how the product granule size distribution is
tion to the coating regimes (e.g., Fig. 10). The average and maximum granule collision
velocity and size (Table 3).
© 2005 by Taylor & Francis Group, LLC
Thus, process changes that reduce drying time (higher bed temperature, lower liquid
flow rate, and smaller drop size) will decrease granule density (increase granule porosity).
Increasing the liquid binder viscosity decreases granule voidage by increasing
the resistance of the granule to deformation.
Dry granule attrition in the fluid bed is an important source of fines. Eq. 3.14
quantifies the attrition rate in terms of granule properties and process conditions
two reasons: (1) it increases the effective ‘‘fluid’’ pressure on granules in the bed
and (2) it increases the average bubble size in the bed, leading to more vigorous
mixing and higher-velocity granule collisions.
Figure 15 Geometric standard deviation of granule size in an agitated fluid-bed granulator
as a function of gas fluidization velocity and binder dispersion measured using spray surface
area to mass in mixer. (From Ref. 10.)
Figure 14 Important granulation processes in the fluidized bed. (From Ref. 5.)
478 He et al.
(Fig. 11). Increasing fluid bed height increases both consolidation and attrition for
© 2005 by Taylor & Francis Group, LLC
5.3. Suggested Scaling Rules for Fluid Bed Granulators
Given this understanding of fluidized bed hydrodynamics and granulation rate process,
we suggest the following guidelines for scaling fluidized bed granulators:
1. Maintain the fluidized bed height constant. Granule density and attrition
rate increase with the operating bed height
L2 ? L1 ?5:8?
2. If L is kept constant, then batch size scales with the bed cross-sectional
area:
M2
M1 ?
D2
F;2
D2
F;1 ?5:9?
3. Maintain superficial gas velocity constant to keep excess gas velocity, and
therefore bubbling and mixing conditions similar:
Q2
Q1 ?
u2
u1 ?
D2
F;2
D2
F;1 ?5:10?
Note that the scaling rules defined by Eqs. 5.8–5.10 are consistent with
Horio’s simplified scaling rules (Eq. 5.7).
4. Keep dimensionless spray flux constant on scale-up. This is most easily
achieved by increasing the area of bed surface under spray (usually by
increasing the number of nozzles). By doing this, the liquid flow rate can
be increased in proportion to batch size without changing critical spray
zone conditions. Thus, batch times at small and large scale should be
similar:
V&
2 ? V&
1 ?5:11?
Aspray;2
Aspray;1 ?
D2
F;2
D2
F;1 ?5:12?
5. Keep viscous Stokes number constant. By adhering to the scaling rules
described earlier, Stv should automatically be similar at small and large
scale leading to similar consolidation and growth behavior.
There are also some cautionary notes relating to the minimum scale for the
laboratory scale studies. Slug flow, a phenomenon where single gas bubbles as large
as the bed diameter form in regular patterns in the bed, significantly reduces solid
mixing. It occurs in tall and narrow beds. Stewart (22) proposed a criterion for
the onset of slugging
u 
 umf
0:35 ffiffiffiffiffiffiffiffiffi gDF p ? 0:2 ?5:13?
To ensure that the bed is operating in bubbling mode without risking slugging,
the ratio in Eq. 5.13 must be kept below 0.2. In addition, both the bed height to bed
diameter and the particle diameter to bed diameter ratios should be kept low. For
pilot fluidized bed, the diameter should be >0.3 m.
To avoid the gas entry effect from the distributor (gas jet), there is also a
requirement on minimum fluidized bed height. The jet length depends on the gas
velocity and the size of the opening on the distributor. For the same opening size,
Scale-Up Considerations in Granulation 479
© 2005 by Taylor & Francis Group, LLC
jet length increases with gas velocity through the hole; for a given gas velocity through
the hole, small holes give shorter jets but are accompanied by a larger pressure drop
across the distributor. Even at a superficial gas velocity as low as 0.2 m/sec with a
hole size of 9.5mm diameter, jet length as long as 0.6m has been reported (23).
The amount of fluidization gas required to maintain constant fluidization velocity
changes linearly with the cross-section area of the bed. However, for large
fluidized beds, one of the major concerns is the even distribution of the fluidization
gas across the whole area of the bed. In addition to the use of a plenum chamber and
an even distribution of flow channels across the distributor, the distributor should be
designed in such a way that the pressure drop across it is at least 20% of the total.
If these scaling rules are applied, there is a good chance to keep granule properties
within the desired range on scaling. If fine tuning is needed at the large scale,
minor adjustments to the liquid spray rate can be used to adjust granule properties,
as all the granulation rate processes in fluidized beds are very sensitive to this
parameter.
6. SCALE-UP OF HIGH-SHEAR MIXER GRANULATORS
Effective scale-up of mixer granulators is more difficult than in fluidized beds. There
are several reasons for this:
 The geometric and mechanical design of mixer granulators varies enormously,
as do the powder flow patterns in the mixer. There is no such thing
as a generic high-shear mixer and caution is needed when transferring scaling
rules from one design to another.
 Even with the same series from the same manufacturer, geometric similarity
is not always maintained between different scales, e.g., impeller size in
relation to bowl size.
 Powder flow in high-shear mixers is not fluidized and powder flow patterns
are much harder to predict than in a fluidized bed.
 All three rate processes, i.e., wetting and nucleation, growth and consolidation,
and breakage and attrition, are taking place simultaneously in the
mixer granulator of all scales. However, the relative dominance of each
of the rate processes can vary significantly on different scales of the same
series, let alone in granulators of different series.
In this section, we will focus mainly on vertical shaft mixer, e.g., Fielder,
Diosna designs. Some of the suggested approaches may be used with caution for
other mixer designs.
6.1. Geometric Scaling Issue
For a simple vertical mixer design, the key dimensions are the impeller diameter D,
which is usually equal to the bowl diameter, the chopper diameter Dc, and the fill
height Hm. The dimensionless groups that need to be held constant for geometric
similarity are
Dc
D
;
Hm
D
In addition, the shape and positioning of the impeller and chopper should be
the same on scale-up. Unfortunately, manufacturers do not always adhere to these
480 He et al.
© 2005 by Taylor & Francis Group, LLC
rules. It is common for the absolute size of the chopper to be invariant, meaning its
relative influence is much larger in the small-scale granulator.
Relative fill height is also often varied with scale. This often reflects the smallsize
batches required for early-stage clinical trials and the desire to maximize production
rate (by maximizing batch size) at full scale. Varying relative fill height is very
dangerous, as it can have a major impact on powder flow patterns.
6.2. Powder Flow Patterns and Scaling Issues
There are two flow regimes observed in a vertical shaft mixer granulator, namely,
bumping and roping regimes (24). At low impeller speeds in the bumping regime,
the powder is displaced only vertically as the blade passes underneath leading to a
slow, bumpy powder motion in the tangential direction. There is almost no vertical
turnover of the powder bed, as shown in Figure 16A.
At a higher impeller speed in the roping regime, material from the bottom of
the bed is forced up the vessel wall and tumbles down at an angle of the bed surface
toward the center of the bowl. There is both good rotation of the bed and good
vertical turnover (Fig. 16B).
The transition from bumping to roping is due to a change in the balance
between centrifugal force and gravity. The centrifugal force, which is caused by
the rotational movement of the powder from the spinning of the blades, pushes
the powder outward toward the wall of the bowl, while gravity keeps the powder
tumbling back toward the center of the bowl from the buildup at the wall region.
This balance between rotational inertia and gravity is given by the Froude number:
Fr ?
DN2
g ?6:1?
where N is the impeller speed and g is the gravitational acceleration.
When the Froude number exceeds a critical value, transition from bumping to
roping takes place:
Fr > Frc ?6:2?
Frc will be a function of relative fill height (Hm/D), impeller design (size and geometry),
and powder flow properties.
Roping flow is more difficult to achieve as relative fill height increases because
the centrifugal force is only imparted to powder in the impeller region. This region
becomes a smaller fraction of the total powder mass as fill height increases. Schaefer
(25) also showed that impeller design had a significant effect on both Frc and bed
turnover rate.
Figure 16 Powder flow regimes in Fielder mixer granulators: (A) bumping and (B) roping.
Scale-Up Considerations in Granulation 481
© 2005 by Taylor & Francis Group, LLC
Cohesive powders transfer to roping at lower values of Fr because momentum
from the spinning impeller is more effectively transferred into the powder mass. Note
that powder flow properties generally change with the addition of the liquid binder
and, therefore, flow patterns will probably change significantly during a batch
granulation.
Figure 17 shows dry lactose powder surface velocity data in a 25 L fielder granulator
(24). In the bumping flow regime, the powder surface velocity increases in proportion
to the impeller speed. In the roping regime, the surface velocity stabilizes and
is less sensitive to impeller speed. In all cases, the surface velocity of the powder is
only of order of 10% of the impeller tip speed. Knight et al. (26) showed that dimensionless
torque T is a direct function of Froude number and effective blade height
heff:
T ? T0 ? kFr0:5 where k ? b
2heff
D  a
?6:3?
Thus, to maintain a similar powder flow pattern during scale-up, the Froude
number should be kept constant, i.e.,
N2
N1 ? ffiffiffiffiffiffi D1
D2 r ?6:4?
In addition, the dimensionless bed height should also be kept constant, i.e., the
same fraction of the bowl is filled at all scales:
Hm;2
Hm;1 ?
D2
D1 ?6:5?
Historically, mixer granulators have been more commonly scaled up using
constant tip speed or constant relative swept volume (25,27). Maintaining constant
Figure 17 Powder surface velocities as a function of impeller tip speed. (From Ref. 24.)
482 He et al.
© 2005 by Taylor & Francis Group, LLC
impeller tip speeds leads to the scaling rule:
N2
N1 ?
D1
D2 ?6:6?
This scale-up rule leads to Fr decreasing, as scale increases. Combined with the
common practice of overfilling full-scale granulators, this approach to scaling can
often lead to a change in operating regime from roping to bumping on scale-up.
The constant swept volume approach to scale-up was introduced partly to
account for variations in geometry on scale-up. The relative swept volume is defined as
V&
R ?
V&
imp
Vmixer ?6:7?
whereV&
R is the relative swept volume,V&
imp is the rate of swept volume of impeller, and
Vmixer is the mixer volume.
On scale-up,
V&
R;1 ? V&
R;2 ?6:8?
This approach is useful for comparing granulators where geometry changes
with scale. For geometrically similar granulators, Eq. 6.8 is equivalent to scale-up
with constant tip speed (Eq. 6.6).
6.3. Granulation Rate Processes and Related Scaling Issues
In high-shear mixer granulation, all three classes of rate process can have a signifi-
cant effect on the granule size distribution. Section 3.1 describes conditions for good
nucleation in the drop-controlled regime and uses examples from mixer granulation.
For good nucleation, the granulator should be operated in the roping regime for
good bed turnover and the dimensionless spray flux ca should be kept low. This
implies careful choice of the liquid flow rate, nozzle design, and positioning in the
granulator.
To maintain similar nucleation behavior and equivalent liquid distribution, the
dimensionless spray flux ca should be kept constant on scale-up. If drop size from
the full-scale nozzle is similar to the small scale, this implies
V&
2
A&2 ?
V&
1
A&1 )
V&
2
V&
1 ?
A&2
A&1 ?6:9?
A common scale-up approach is to keep the same total spray time and still use
a single nozzle at large scale. Thus, V& is proportional to D3. Despite the fact that the
powder area flux will increase slightly with scale, this approach generally leads to a
substantial increase in dimensionless spray flux. To keep dimensionless spray flux
constant, multiple spray nozzles or longer spray times should be used at a large scale.
It should be noted that consolidation, growth, and breakage processes are
controlled by Stdef. This can lead to quite complicated growth behavior in mixer
ing liquid viscosity and increasing impeller speed increase the rate of granule growth
but decrease the final equilibrium granule size. Both these effects increase Stdef. In
the early stage of granulation, this increases the probability of successful coalescence.
However, as the granules grow, the critical value of Stokes number for breakage may
be exceeded—at least near the impeller blade leading to a balance of breakage and
growth and an equilibrium granule size. This example also highlights that most
Scale-Up Considerations in Granulation 483
granulators. Figure 18 illustrates some of this complex behavior (27). Both decreas-
© 2005 by Taylor & Francis Group, LLC
high-shear mixers have a very wide range of collision velocities in different parts of
the bed. Granule coalescence will occur in regions of low collision velocity, while
breakage and consolidation are more likely near the impeller. To properly quantify
and predict this behavior, we need more sophisticated models that divide the granulator
into at least two regions and incorporate better understanding of powder flow
than we currently have.
Nevertheless, we can make some intelligent comments with regard to scale-up.
In a mixer granulator, the maximum collision velocity for a granule will be of the
order of the impeller tip speed. To maintain constant Stdef, the impeller tip speed
should be kept constant, i.e., Eq. 6.6. If a constant Fr rule is used (Eq. 6.4),
Stdef;2
Stdef;1 ?
U2
c;2
U2
c;1 ?
N2
2D22
N2
1D21
?
D2
D1 ?6:10?
Thus, Stdef increases with scale. This will lead to an increase in the maximum
achievable granule density and a decrease in the maximum achievable particle size.
The actual granule density and size may also depend on the kinetics of consolidation
and growth and are difficult to predict without more sophisticated quantitative modeling.
As such, the variation in Stdef with scale potentially leads to changes in granule
attributes that are difficult to predict.
The liquid saturation s (Eq. 3.7) should be kept constant on scaling. This
implies a similar liquid content on a kilogram per kilogram dry powder basis
provided the granule density does not change with scale. For operation in the steady
growth regime, this is a reasonable assumption. However, for operation in the induction
growth regime the change in density with scale is harder to predict.
6.4. Recommended Scaling Rules and a Case Study Example
The complexity of powder flow and granulation rate processes makes it impossible to
recommend a single definitive set of scaling rules. It is important to know which
granule attribute is of most importance during scaling and the main granulation rate
process that controls this attribute.
Figure 18 Variations in granule growth rate and extent of growth in a mixer granulator with
changes to (A) binder viscosity and (B) impeller speed. (From Ref. 27.)
484 He et al.
© 2005 by Taylor & Francis Group, LLC
Overall, we recommend the following approach:
1. Keep granulators geometrically similar during scale-up where manufacturer’s
designs permit. In particular, keep dimensionless fill height constant
during scale-up (Eq. 6.5).
2. To ensure similar powder mixing, keep Froude number constant during
scale-up by adjusting the impeller speed according to (Eq. 6.4). At the very
least, make sure Fr > Frc at all scales.
3. To achieve good binder distribution, ca should be kept constant on scaleup.
This is likely to mean multiple spray nozzles at the large scale to give
sufficient spray zone area (Eq. 6.9).
4. To keep Stdef constant for consolidation, breakage, and growth, constant
impeller tip speed should be maintained. This is in conflict with scaling rule
2 mentioned earlier. Scale-up with constant tip speed is possible, provided
that at large scale Fr > Frc.
5. Spray time during the batch and total batch time scaling rules require a
sound understanding of how the kinetics of growth and consolidation vary
with scale. We do not know these rules yet, and they are likely to be different
for operation at different growth regimes. As a starting point, keeping batch
times constant during scaling is probably reasonable provided this does not
conflict with other scaling rules (especially rule 3 mentioned previously).
Conflicting scale-up goals lead us to consider more sophisticated operating strategies
at a large scale including programming impeller speed to change during the
batch operation. For example, begin the granulation with high impeller speed (constant
Fr) to induce good dry powder turn over. This helps ensure good wetting and
nucleation at the beginning of the batch when it is most important. Later, reduce
the impeller speed to give a similar tip speed to smaller-scale operation to control
granule density or size. As the powder mass is now wet, it will be more cohesive and
operation above the critical Froude number for rolling flow will be easier to maintain.
and Ennis (5) give a case study for scale-up of a lactose granulation that is useful for
illustrating these scaling rules and conflicts. It is represented in the next section.
6.4.1. Scale-Up of a Lactose Granulation from 25 to 300 L
A lactose-based granulation in a 25 L granulator has given granules with acceptable
properties. The operating conditions for the 25 L granulator are summarized as
follows:
Parameter Value
Nominal volume (L) 25
Powder charge (kg) 5
Impeller speed (rpm) 330
Spray time (min) 8
Drop size (mm) 100
emin 0.3
W 0.15
V& (m3/sec) 1.610
6
Spray width W (m) 0.13
Powder surface velocity (m/sec) 0.85
ca 0.22
Scale-Up Considerations in Granulation 485
© 2005 by Taylor & Francis Group, LLC
The dimensionless spray flux ca was calculated by
ca ?
3V&
2
2A&2 dd ?6:11?
This granulation is to be scaled to 300 L using the following rules and heuristics:
 Keep Fr constant
 Keep spray time constant
 Spray from a single nozzle at large scale.
How do ca and Stdef change on scale-up?What are the implications from granulation
rate processes at full scale?
Scaling to 300 L granulation:
Assuming geometric similarity,
D2=D1 ? 121=3
Keeping Fr constant,
N2 ? ?D1=D2?0:5 N1 ? 218 rpm
Assuming spray width scales with impeller diameter,
W2 ? ?D2=D1?W1 ? 0:3m
powder surface velocity scales with tip speed,
v2 ? ?D2N2=D1N1?v1 ? 1:28m=sec
Keeping spray time constant with one nozzle,
V&
2 ? 12V&
1
Thus, the dimensionless spray flux at 300 L is
ca;2 ?
3V&
2
2W2v2d ?
3?12V&
1 ?
2?121=3W1?121=6v1d ? 3:41ca;1 ? 0:75
There has been a substantial increase in ca on scale-up taking the granulation
from nearly drop controlled into the mechanical dispersion regime. This could result
in a much broader granule size distribution at a large scale. A similar spray flux
could be achieved by using an array of four nozzles spaced at 90
 intervals around
the granulator (all positioned so the spray fan is at right angles to the direction of
powder flow).
We cannot calculate the value of Stdef because the dynamic yield stress Y for
the lactose/binder system is not given. However, if we neglect changes in Y due to
the larger strain rate, then Stdef will increase as
Stdef;2 ?
U2
c;2
U1
c;2
Stdef;1 ? ?D2N2?2
?D1N1?2 Stdef;1 ? 2:3Stdef;1
There is a significant increase in Stdef with scale-up that could impact on
the granule density and maximum size. It is not possible to scale with constant
486 He et al.
© 2005 by Taylor & Francis Group, LLC
Stdef while simultaneously maintaining constant Fr. Scale-up summary data are
given in the following table:
Parameter 25 L 300 L
Nominal volume (L) 25 300
Powder charge (kg) 5 60
Impeller speed (rpm) 330 218
Spray time (min) 8 8
Drop size (mm) 100 100
emin 0.3 0.3
W 0.15 0.15
V& (m3/sec) 1.610
6 19.210
6
Spray width W (m) 0.13 0.3
Powder surface velocity (m/sec) 0.85 1.28
ca 0.22 0.75
Stdef/Stdef, 25 L 1 2.3
7. CONCLUDING REMARKS
Scaling of granulators using the traditional chemical engineering dimensional analysis
approach of complete similarity is not possible due to the complexity of the
process and the constraints on formulation changes during scaling pharmaceutical
processes. Nevertheless, scale-up using partial similarity that strives to keep some
key dimensionless groups invariant is possible. It is very important to understand
the powder flow phenomena in the granulator of choice and to maintain the same
flow regime during scaling (bubbling vs. slugging, bumping vs. roping).
The second important requirement is to maintain constant key dimensionless
groups that control the important granulation rate process of most interest during
scale. This is somewhat easier to do in fluidized beds than in high-shear mixers.
Very small-scale tests, which have no geometric similarity to pilot-and fullscale
tests should be used to focus on formulation design and measurement of key
formulation properties that influence the granulation rate processes.
Insightful understanding of the granulation processes is essential for the identi-
fication of key variables and parameters for the dimensional analysis and scale-up considerations.
While development of definitive mathematical models for the granulation
processes is incomplete, the scaling approaches recommended in this chapter help
reduce uncertainty during new product development and transfer to industrial sites.
NOMENCLATURE
A& area flux of powder through the spray zone
dd liquid drop size (diameter)
di indentor diameter
dp particle or granule size
db bubble size
Ddrum drum granulator diameter
D impeller diameter of mixer granulators
Dc chopper diameter of mixer granulators
DF Fluidized bed diameter
Scale-Up Considerations in Granulation 487
© 2005 by Taylor & Francis Group, LLC
Fr Froude number
g gravitational acceleration
H hardness of granules
Hm fill height of mixer granulators
L characteristic length of a fluidized bed
Kc fracture toughness of granules
M1, M2 mass of particles in the fluidized bed
m scaling ratio
N impeller speed
Stdef Stokes deformation number
Smax granule pore saturation
tp drop penetration time
u superficial fluidization velocity
u1 fluidization velocity on the smaller bed
u2 fluidization velocity on the larger bed
umf,1 minimum fluidization velocity on the smaller bed
umf,2 minimum fluidization velocity on the scaled bed
ub bubble rise velocity
Uc particle collision velocity
V&
R relative swept volume
V&
imp rate of swept volume of impeller
Vmixer mixer volume
V& volumetric spray rate
w liquid to solid mass ratio
W spray zone width
Y dynamic yield stress of granules
m viscosity of binder
rg granule density
rG density of fluidizing gas
rs particle density
rl binder liquid density
y solid liquid contact angle
gLV liquid surface tension
d dimensionless bubble space, defined as the ratio
of bubble space over bubble radius
o drum peripheral speed
oi impeller peripheral speed
oc chopper peripheral speed
ca dimensionless spray flux
emin minimum porosity of granule
etap granule bed tap density
REFERENCES
1. Hileman GA. Regulatory issues in granulation processes. Parikh DM, ed. Handbook of
Pharmaceutical Granulation Technology. New York: Marcel Dekker, Inc., 1997.
2. Skelly JP, et al. Scaleup of immediate release oral solid dosage forms. Pharm Res 1993;
10:2–29.
3. Zlokarnik M. Dimensional Analysis and Scale-up in Chemical Engineering. Berlin:
Springer-Verlag, 1991.
4. Munson BR, Young DF, Okiishi TH. Fundamentals of Fluid Mechanics. 2nd ed. New
York: John Wiley & Sons, Inc., 1994.
488 He et al.
© 2005 by Taylor & Francis Group, LLC
5. Litster JD, Ennis B. The Science and Engineering of Granulation Processes. Dordrecht:
Kluwer Academic Publishers, 2004.
6. Hapgood KP, Litster JD, Smith R. AIChE J 2003; 49(2):350–361.
7. Iveson SM, Wauters PAL, Forrest S, Litster JD, Meesters GMH, Scarlett B. Powder
Technol 2001; 117(1,2):83–97.
8. Iveson SM, Beathe JA, Page NW. Powder Technol 2002; 127:149–161.
9. Ennis BJ, Tardos GI, Pfeffer R. Powder Technol 1991; 65:257.
10. Tardos GI, Irfran-Khan M, Mort PR. Powder Technol 1997; 95:245.
11. Evans AG, Wilshaw TR. Acta Metall 1976; 24:939.
12. Ennis BJ, Sunshine G. Tribol Int 1993; 26:319.
13. Kunii D, Levenspiel O. Fluidization Engineering. 2nd ed. Boston: Butterworth-
Heinemann, 1991.
14. Sanderson J, Rhodes M. Hydrodynamic similarity of solids motion and mixing in
bubbling fluidized beds. AIChE J 2003; 49:2317–2327.
15. Fitzgerald TJ, Crane SD. Cold fluidized bed modelling. Proceedings of the International
Conference of Fluidized Bed Combustion. Vol. III. Technical Sessions, 1985:85–92.
16. Glicksman LR. Scaling relationships for fluidized beds. Chem Eng Sci 1984; 39:1373–1379.
17. Glicksman LR. Scaling relationships for fluidized beds. Chem Eng Sci 1987; 43:1419–1421.
18. Glicksman LR, Hyre M, Woloshun M. Simplified scaling relationships for fluidized beds.
Powder Technol 1993; 77:177–199.
19. Horio M, Nonaka A, Sawa Y, Muchi I. A new similarity rule for fluidized-bed scale-up.
AIChE J 1986; 32:1466–1482.
20. Horio M, Takada M, Ishida M, Tanaka N. The similarity rule of fluidization and its
application to solid mixing and circulation control. Proceedings of Fluidization V.
New York: Engineering Foundation, 1986:151–156.
21. Roy R, Davidson JF. Similarity between gas-fluidized beds at elevated temperature and
pressure. Proceedings of Fluidization V. New York: Engineering Foundation, 1986:
293–299.
22. Steward PSB, Davidson JF. Powder Technol 1967; 6:61–80.
23. Werther J. Influence of the distributor design on bubble charcteristics in large diameter
gas fluidized beds. Davidson JF, Keairns DL, eds. Fluidization. New York: Cambridge
University Press, 1978.
24. Litster JD, Hapgood KP, Kamineni SK, Hsu T, Sims A, Roberts M, Michaels J. Powder
Technol 2002; 124:272–280.
25. Schaefer T. Ph.D. thesis, The Royal Danish School of Phamacy, 1977.
26. Knight PC, Seville JPK, Wellm AB, Instone T. Chem Eng Sci 2001; 56:4457–4471.
27. Kristensen HG, Schaefer T. Drug Dev Ind Pharm 1987; 13:803.
Scale-Up Considerations in Granulation 489
© 2005 by Taylor & Francis Group, LLC
17
Sizing of Granulation
Gurvinder Singh Rekhi and Richard Sidwell
Elan Drug Delivery Inc., Gainesville, Georgia, U.S.A.
1. INTRODUCTION
Tablets are the most frequently administered solid oral dosage forms in contemporary
practice. Tablets consist of a mixture of powders or granules that are compacted in
the die of a tablet press. Even though the popularity of directly compressible materials
has increased, many powders are granulated to overcome the difficulties in obtaining
an acceptable tablet dosage form and meeting the product specifications. The most
challenging task in a tableting process is to achieve a constant volume of homogenous
mixture to flow into the tablet die cavity. Unfortunately, most powder materials do
not have inherent flow properties. This, in turn, places demands on changing the physical
characteristics of the powder or improving the design of the tablet press (1).
Therefore, granulation becomes an integral part of a pharmaceutical process that
attempts to improve powder flow characteristics.
The granule properties play a pivotal role in the final performance of a tablet;
for example, granule size can affect the flowability and, hence, the average tablet
weight and weight variation, and drying rate kinetics of wet granulations. The effect
of granule size and size distribution on final blend properties and tablet characteristics
is dependent on formulation ingredients and their concentration, as well as the
type of granulating equipment and processing conditions employed. Therefore, granulation
and sizing of granulation become critical unit operations in the manufacture
of oral dosage forms (2,3). To some extent, the same requirements are necessary for
capsule manufacture, especially when the drug is bulky or has poor flow properties,
or in the newer high-speed capsule-filling machines, where limited compaction occurs.
Few materials used in the manufacture of pharmaceutical dosage forms exist in
the optimum size, and most materials must be reduced in size at some stage during
production. The advantages of sizing of granules in tablet formulation development
are as follows:
1. Mixing and blending of pharmaceutical ingredients are easier and more uniform
if the ingredients are of approximately the same size and distribution.
2. Improving color or active ingredient dispersion. Milling may reduce the
tendency for mottling, and hence improve the uniformity of color from
batch to batch.
491
© 2005 by Taylor & Francis Group, LLC
3. Wet milling produces uniformly sized wet granules, which promotes
uniform and efficient drying.
4. Improving uniformity of dosage units by virtue of uniformity of particle
size distribution and reduction in the segregation of the mix.
5. Enhancing flow properties reduces the weight variation and improves
content uniformity.
6. Increasing surface area due to particle size reduction may enhance the
dissolution rate and thereby, the drug’s bioavailability.
7. Reduction of dust, thereby reducing workers’ exposure.
Size reduction alone is not the panacea for all tableting problems. There are
some disadvantages to size reduction that may affect the final characteristics of a
dosage form, such as degradation of the drug or a change in the polymorphic form
as a result of the excessive heat generated, or increase in surface energies leading to
agglomeration, and so on. Hence, in optimizing the manufacture of pharmaceutical
dosage forms, it is important not only to characterize the formulation ingredients,
but also to study their effect on the manufacturing process (i.e., whether a granulation
should be milled and to what extent based on the final product specifications).
The objective of this discussion is to focus on sizing of granulation after drying
in a wet granulation process. However, a process of wet milling for obtaining uniformly
sized granules for uniform drying will also be addressed. A full discussion
of the theories of comminution or equipment description is beyond the scope of this
chapter. However, the text of this chapter will address the various types of equipment
used in the size reduction process, their merits and demerits and the variables
affecting the size reduction process, scale-up factors, and relevant case studies to be
considered in the development and optimization of tablet and capsule manufacture.
2. THEORY OF COMMINUTION OR SIZE REDUCTION
Comminution, or size reduction, is the mechanical process of reducing the size of
particles or aggregates. There is, as yet, only a basic understanding of the mechanism
and quantitative aspects of milling (4,5). Reduction of particle size through fracture
requires application of mechanical stress to the material to be crushed or ground.
Materials respond to this stress by yielding, with consequent generation of strain.
In the case of a brittle substance, complete rebound occurs on release of applied
stress at stresses up to the yield point, at which fracture would occur. In contrast,
plastic material would neither rebound nor fracture. The vast majority of pharmaceutical
solids lie somewhere between these extremes and thus possess both elastic
and viscous properties.
The energy expended by comminution ultimately appears as surface energy
associated with newly created particle surfaces, internal free energy associated with
lattice changes, and heat. For any particle, there is a minimum energy required which
will fracture it; however, conditions are so haphazard that many particles receive
impacts that are insufficient for fracture, and are eventually fractured by excessively
forceful impact. As a result, most efficient mills use <1–2% of the energy input to
fracture particles and to create new surfaces. The rest of the energy is dissipated
in the form of heat from the plastic deformation of the particles that are not fractured,
by friction, and in imparting kinetic energy to the particles. The greater the
rate at which the force is applied, the less effectively the energy is utilized, and the
higher is the proportion of fine material produced.
492 Rekhi and Sidwell
© 2005 by Taylor & Francis Group, LLC
A flaw in a particle is any structural weakness that may develop into a crack
under strain. The Griffith theory (4) of cracks and flaws assumes that all solids contain
flaws and microscopic cracks, which increase as the applied force increases,
according to the crack length and focus of the stress at the crack apex.
A granule is an aggregation of particles that are held together by bonds of finite
strength, and the ultimate strength of a wet granule depends on the surface tension of
the granulating liquid and capillary forces. After drying, granules develop stronger
bonds owing to fusion and recrystallization of particles, and curing of adhesives or
binding agent. The final strength of a granule depends on the base material, the type
and the amount of granulating agent used, and the granulating equipment employed.
A granule or particle may be subjected to one or more of the following four
forces during milling:
1. Shear (cutting forces)
2. Compression (crushing force)
3. Impaction (the direct, high-velocity collision force)
4. Tension (the force that works to elongate or pull a particle apart)
The mechanism by which sizing of dried granules occurs is similar to that of
crystalline materials. Cleavage occurs at the weakest point or points in the granule
and it could be at (2):
1. the binder–particle interface
2. the bridge of binder between the individual ingredient particles being
granulated
3. flaws in the individual ingredient particles within the granules, or
4. a combination of any of these.
Granules held together with lower binding strength agents such as povidone
will require less severe grinding conditions because the fractures take place primarily
at the binder bridge or the binder–particle interface.
Although the milling process can be described mathematically (6–8), its theory
has not been developed to the point at which the actual performance of a mill can be
predicted quantitatively. Three fundamental laws (Kick’s Law, Rittinger’s Law, and
Bond’s Law) have been proposed to relate size reduction to a single variable, the
energy input to the mill. None of the energy laws apply well in practice (9). Generally,
laboratory testing is required to evaluate the performance of a particular piece
of equipment; however, a work index and grindability index have been used to evaluate
the mill performance (5). The efficiency of milling process is influenced by the
nature of the force, as well as by its magnitude. The rate of application of force
affects comminution because there is a lag time between the attainment of maximum
force and fracture. Often, materials respond as a brittle material to fast impact, and
as a plastic material to a slow force.
3. PROPERTIES OF FEED MATERIALS AFFECTING
THE SIZING PROCESS
The milling or sizing process is affected by a variety of factors and has a direct effect
on the quality of the final product. The properties of feed material and the finished
product specifications determine the choice of equipment to be used for the process
of comminution. The properties of feed material include melting point, brittleness,
Sizing of Granulation 493
© 2005 by Taylor & Francis Group, LLC
hardness, and moisture content. The desired particle size, shape, and size distribution
must also be considered in the selection of milling equipment.
Materials can be classified as hard, intermediate, soft, or fibrous materials (e.g.,
glycyrrhiza and rauwolfia) based on Moh’s Scale. Fibrous materials require cutting
or chopping action and usually cannot be reduced in size effectively by pressure or
impact techniques. Before selecting and optimizing a size reduction process one
needs to know the properties of the material and the characteristics of a mill. The
important material properties (5,10) are as follows:
1. Toughness: The material’s resistance to the propagation of cracks. Reduction
of the particle size of tough material is difficult, but can sometimes
be made easier by cooling the material, thereby diminishing its tendency
to exhibit plastic flow, and making it more brittle.
2. Brittleness: The opposite of toughness. Size reduction poses no problems
except if the amount of fines is to be controlled.
3. Abrasiveness: This is an important factor because abrasive materials can
wear mill parts and screens; hence, metal contamination may be a problem.
4. Cohesive/adhesiveness: Particles sticking together or to machine surfaces
are often dependent on moisture content and particle size. Problems with
moisture content can be mitigated by drying the material or avoided by
using a wet size reduction process.
5. Melting point: This is critical because considerable heat is generated in
size reduction. High temperatures generated can cause melting of the
drug, blinding of screen, or can degrade heat-sensitive materials.
6. Agglomeration tendency: This tendency can be counteracted by drying the
material, either before or during the size reduction operation. In some
cases, mixing with other ingredients during milling might be helpful. Generally,
materials having a strong tendency to agglomerate are wetted
prior to milling.
7. Moisture content: A moisture content above 5% can often lead to agglomeration
or even liquefaction of the milled material. Hydrates will often
release their water of hydration under high temperatures and may require
cooling or low-speed milling.
8. Flammability and explosiveness: A measure of how readily a material will
ignite or explode. Explosive materials must be processed in an inert-gas
atmosphere.
9. Toxicity: This has little influence on the selection of the mill itself; however,
it must be considered in determining operator safety, containment,
and setup for this type of material.
10. Reactivity: The possibility of materials chemically reacting with the materials
of construction of the mill (including liners and gaskets) and cleaning
solutions must be considered.
4. CRITERIA FOR SELECTION OF A MILL
The selection of equipment is determined by the characteristics of the material, the
initial particle size, and the degree of the desired size of the milled product, that
is, coarse, medium, or fine.
The criteria for selection of a mill include the following (4):
494 Rekhi and Sidwell
© 2005 by Taylor & Francis Group, LLC
1. Properties of feed material: Size, shape, moisture content, physical and
chemical properties, temperature sensitivity, grindability, and material
compatibility
2. Product specifications: Size, particle size distribution, shape
3. Versatility of operation: Wet and dry milling, rapid change of speed and
screen, safety features
4. Scale-up: Capacity of the mill and production rate requirements
5. Repeatability: Ability to meter material to the mill to insure consistent
process
6. Dust control: Loss of costly drugs, health hazards, contamination
7. Sanitation: Ease of cleaning (clean in place, CIP) and sterilization (sterilization
in place, SIP)
8. Auxiliary equipment: Cooling system, dust collectors, forced feeding,
stage reduction
9. Batch or continuous operation
10. Economic factors: Equipment cost, power consumption, space occupied,
labor cost
After consideration of the foregoing factors for a specific milling problem, it is
suggested that a variety of mills should be evaluated for optimum product results
such as shape of granules and/or scalability from laboratory to production. In addition
to the standard adjustments of the milling process (e.g., screen, speed, rotor
design, feed rate), other techniques of milling may be considered for special materials.
Hygroscopic materials can be milled in a closed system supplied with dehumidified
air. As the bulk of the energy used in milling is converted into heat, heat-sensitive
materials, or hard materials that build up in the milling chamber may melt, decompose,
or explode. A two- or multistep milling process can be used for harder and dif-
ficult-to-grind materials. Materials can be milled using a coarser screen, and the
material can then be recycled by screening the discharge and returning the oversized
material for a second milling (closed circuit mill). Alternatively, one may chill the air
or gas (carbon dioxide or nitrogen) that transports the product, cool the product prior
to processing, or cool the comminuting chamber through which the product passes. A
chiller is necessary for all of these options and will add to the cost of processing (11). If
this not sufficient to embrittle the material, it may be fed to the mill simultaneously
with dry ice.
5. CLASSIFICATION OF MILLS
The majority of size reduction equipment may be classified according to the way in
which forces are applied; namely, impact, shear, attrition, and shear-compression
mill may be used to prepare a 16-mesh granulation and to mill a crystalline material
to a 120-mesh powder. The mills used for size reduction of the granules can be
divided into two primary categories based on the energy input into the process. Even
though there are several high-energy mills available for size reduction, only a few are
used in the pharmaceutical industry for the wet or dry sizing process. Milling is an
extremely inefficient unit operation with only 1–2% of the applied energy being
utilized in the actual size reduction. Milling efficiency is dependent on the characteristics
of the material used and the type of mill employed.
Sizing of Granulation 495
(Table 1). A given mill may operate successfully in more than one class: a hammer
© 2005 by Taylor & Francis Group, LLC
Table 1 General Characteristics of Various Types of Mills
Mechanism of
action Example Product size Type of material Not used for
Impact Hammer mill Moderate to fine Brittle and dry material Fibrous, sticky, low-melting
substances
Shear Extruder and hand
screen
Coarse Deagglomeration, wet
granulation
Dry material, hard, abrasive
materials
Attrition Oscillating granulator Coarse to moderate Dried granulation Wet granulation, abrasive
materials
Shear-compression Conical screening mill Moderate to coarse Wet, dry granulation Abrasive materials
496 Rekhi and Sidwell
© 2005 by Taylor & Francis Group, LLC
5.1. Low-Energy Mills
5.1.1. Hand Screen

 Size reduction occurs primarily by shear.

 They are made of brass or stainless steel and consist of a woven wire cloth
stretched in a circular or rectangular frame.

 They are available in sizes ranging from 4 to 325 mesh; however, for granulation,
primarily mesh sizes from 4 to 20 are used.

 They are most widely used for sieve analysis or for size reduction of wet and
dry granules in the early stages of formulation development.
5.1.2. Oscillating/Rotary Granulator

 They consist of an oscillating bar contacting a woven wire screen, and the
material is forced through the screen by the oscillating-rotary motion of the
bar (Figs. 1A and B).

 Size reduction is primarily by shear with some attrition

 Speed, rotary or oscillatory motion, and screen size are important variables
to be considered during the sizing process.

 They are used primarily for size reduction of wet and dry granulations and,
to some extent, for milling tablets and compacts that must be reprocessed.

 The narrow size distribution and minimum amount of fines are advantages
during the size reduction of dry granulation (2).

 Heat-sensitive and waxy materials can be milled owing to the low heat
generated during the sizing process.

 Low throughput rates and possible metal contamination from wearing
down or broken screen are some of its limitations.
5.1.3. Extruder

 It is primarily used for continuous wet granulation.
Figure 1 Oscillating granulator: (A) Frewitt MF line and (B) rotor, screen, and tensioning
spindles.
Sizing of Granulation 497
© 2005 by Taylor & Francis Group, LLC

 Wet material is forced through a screen and the extruded material is dried in a
tray or fluid bed dryer or can be spheronized to produce granules with a high
degree of sphericity and then dried for controlled-release applications.

 Less dust generation and more uniformgranules are some of the advantages.


5.2. High-Energy Mills
5.2.1. Hammer Mill
The hammer mill is one of the most versatile and widely used mills in the pharmaceutical
industry. The principle of size reduction in the hammer mill is one of
high-velocity impact between the rapidly moving hammers mounted on a rotor
and the powder particles (Figs. 2A and B). These mills can produce a wide range
of particle sizes, even down to micrometer size. The particle shape, however, is
generally sharper and more irregular than that produced by compression methods
(5). The force imparted by the hammers and the screen opening size and shape control
the degree of particle size reduction.

 They can be used for size reduction of wet or dry granulations and milling
of raw materials.

 There is a wide range of interchangeable feed throats and variable feed
screw systems available to optimize the feed rate (12).

 Hammers can rotate horizontally or vertically, based on the rotor configuration,
and at variable speeds.

 Hammers can be fixed or free-swinging.

 Hammers with blunt or impact edges are preferred for pulverizing and knife
or sharp edges are preferred for chopping or sizing of granules (12).

 Screen openings generally vary from 0.3 to 38 mm, with round or square
perforations, diagonal or straight slots, or with a rasping surface.

 Feed rate and dryness of the granules are important variables relative to the
material.

 Type of hammers, rotor speed, screen type, thickness, and opening size are
important variables relative to the machine.
Figure 2 Hammer mill: (A) Fitzmill model L1A and (B) principle of operation.
498 Rekhi and Sidwell
More information on extrusion may be found in Chapter 11.
© 2005 by Taylor & Francis Group, LLC

 Ease of setup, clean-up, minimum scale-up problems, and ability to handle
a wide variety of size and type of feedstock are some advantages.

 Heat buildup, screen wear, and potential clogging of screens are some of the
limitations.

 Integrated designs are available for dust containment.

 Examples include the Granumill and Fitzmill.
5.2.2. Conical Screening Mill

 It is effective for dry (deagglomeration–delumping) and wet milling of soft
to medium hard materials.

 The comminution chamber consists of an impeller rotating at variable
speed, imparting a compression or shear force inside a conical screen.

 The impeller imparts a vortex flow pattern to the feed material, and the centrifugal
acceleration forces the particles to the screen surface and up the


 The space between the impeller and the screen can be adjusted.

 The size and shape of the screen holes, screen thickness, impeller configuration,
and mill speed are important variables.

 It is used for difficult-to-mill, heat-sensitive material and hard granules.

 Low heat and lower amounts of fines are produced compared with the hammer
mill; hence, it produces a narrower particle size distribution.

 The impeller does not touch the screen; hence, chances of screen breakage
and metal contamination are greatly reduced compared with an oscillating
granulator.

 The dual action of conical screening mills (size reduction and mixing)
makes this equipment more desirable than the use of traditional oscillators
(14,15).

 Integrated designs are available that are attached to a high-shear granulator
discharge, which provides a deagglomerated, lump-free product for the

 Examples include the Comil, Glatt sieve GS or GSF, and FitzSiv.
5.2.3. Centrifugal-Impact Mills
Centrifugal-impact mills and sieves are useful to minimize the production of fine particles,
because their design combines sieving and milling into a single operation.
Unlike the conical screening mills, these consist of a nonrotating bar or stator which
is fixed within a rotating sieve basket. This action produces a very low product agitation
and impact; hence, no heat is generated. The particles that are smaller than the
holes of the sieve can pass through the mill without comminution; however, the larger
particles are directed by centrifugal force to impact the stator. Older designs are
not preferred because the likelihood of sieve-to-stator contact can result in metal
particulates in the product. Newer designs eliminate metal-to-metal contact. The
6. WET MILLING
The discussion so far has been focused on dry milling. These mills can also be used
for wet milling or coarse milling. There are several reasons for wet milling, including
the following (16):
Sizing of Granulation 499
cone (360 ) in a spiraling path (13) (Figs. 3A and B).
dryer (Fig. 4).
Frewitt SG line (Figure 5) is an example of this type.
© 2005 by Taylor & Francis Group, LLC
1. To increase surface area for more efficient drying
2. To improve size uniformity
3. To improve granule formation
4. To prevent large particles that will shatter to ‘‘fines’’ on dry milling
5. For further mixing or blending because ingredients are approximately of
the same size
As discussed for low-shear mills, extruders can be used as a continuous wet granulation
method.Wet milling is necessary with low-shear mixers, such as planetary, ribbon,
or sigma mixers, but with high-shear mixers, the combination of high impeller
speed and built-in choppers produces a product ready for drying. Also, integrated
designs are available such that the wet milling step in no longer a separate operation.
Figure 3 Conical-screening mill: (A) Glatt model GSF 180 and (B) principle of operation.
500 Rekhi and Sidwell
© 2005 by Taylor & Francis Group, LLC
Figure 4 An integrated pharmaceutical manufacturing facility (high-shear granulator–
conical screening mill–fluid bed drier).
Figure 5 Centrifugal-impact mill: Frewitt SG Line.
Sizing of Granulation 501
© 2005 by Taylor & Francis Group, LLC
Finally, there are continuous granulators available (M6 NICA granulator) for
product applications that do not require extensive kneading treatment, and can be
used for both batch and continuous operation. The wetted product is discharged
in continuous stream through an adjustable opening in the turbine cover. A homogenous
mix is produced in a few minutes and further milling may not be necessary.
7. VARIABLES AFFECTING THE SIZING PROCESS
7.1. Process Variables
As discussed in the introduction, the granule properties can dictate the properties of
the final tablet. Some of the problems faced during the tableting process are flow of
granules, maintaining uniform density in the granule bed, and the particle size
distribution. Each of the stages of the granulation can be critical and can affect
tableting. In addition to the wet granulation process, the sizing process can be critical
for the particle size distribution that, along with the amount of fines, dictates the
flow properties. These, in turn, influence the packing and density of the granules.
Reproducibility of batches depends not only on the properties of the unmilled dry
granules, but also on the mill and milling parameters. Finally, the dry-milling stage
is important because of the excessive heat generated that might affect the stability of
the final product.
The characteristics of the granules after size reduction depend mainly on the
type of mill, impeller type and speed, screen size, and thickness.
7.2. Type of Mill
The type of mill chosen can affect the shape of the granules and throughput. The
shape of the milled granules can affect the flow properties. An impact mill produces
sharp, irregular particles that may not flow readily, whereas an attrition mill produces
free-flowing spheroidal particles. An oscillating granulator uses shear and
attrition as the main mechanisms for size reduction. The granules produced are more
spheroidal, because size reduction takes place by surface erosion. If the same material
is subjected to impact by hammers in a hammer mill, the granules will shatter
and cause irregularly shaped granules. If a conical screening mill is used for the same
material for size reduction, it imparts some shear and some compression between the
rotating impeller and the screen.
7.2.1. Hammer Mill
There are a number of variables in a hammer mill that can influence comminution
(12,17–19). The following section discusses five operating variables in detail:
1. Rotor shaft configuration: The hammers may be mounted on a vertical or
mill) have feed inlets at the top and material is fed perpendicular to the
swing of the hammers. In the case of horizontal shaft mills (Fitzpatrick–
Fitzmill), the material is fed tangentially to the hammer swing. Rotor con-
figuration can influence the particle size distribution of granules. In the vertical
configuration, the screen is placed 360 around the hammers and this
provides more screen open area and less time for the granules to stay in the
milling chamber when compared with the horizontal shaft mills.
502 Rekhi and Sidwell
horizontal shaft (Figs. 6A and B). The vertical shaft mills (Stokes-Tornado
© 2005 by Taylor & Francis Group, LLC
2. Material feed rate: The feed rate controls the amount of the feed material
that enters the comminutor and prevents overfeeding (slugging) or underfeeding
(starving) in the milling chamber. Although both affect the particle
size distribution, overfeeding is relatively more detrimental. If the rate of
feed is relatively slow, the product is discharged readily, and the amount
of undersize material, or fines, is minimized. On the other hand, overfed
material stays in the milling chamber for a longer time, because its
discharge is impeded by the mass of material. This leads to a greater reduction
of particle size, overloads the motor, and the capacity of the mill is
reduced. The rule of thumb is to keep the feed rate equal to the rate of
discharge. The feed rate can be controlled using variable-feed screws,
vibratory feeders, or dischargers controlled by gravity. In addition to controlling
the flow, the feed throat must allow the material to enter at a
proper angle. There are more than 50 feed throat designs available that
one needs to consider for optimizing the milling process. Most mills used
in the pharmaceutical operations are designed so that the force of gravity
is sufficient to give free discharge, generally from the bottom of the mill.
3. Blade type: Comminution is effected by the impact of the material with the
fast moving blades and attrition with the screen. Generally, the blades of a
hammer mill have a blunt or flat edge on one side and a sharp or knife-edge
on the other side. The desired particle size range determines which blades
to use. Many models of hammer mills have a rotor that may be turned
180, so that the blunt edges can be used for fine grinding or the knife-edge
can be used for cutting or granulating. The blunt edge offers the surface
and the impact during milling, generating smaller granules. The knife-edge,
because the sharper edge causes cutting of the granules, thereby generates
larger granules. Individual blades—blunt, sharp, or reversible—are
the material being ground, while swinging blades lie back and depend on
the centrifugal force for movement. Fixed blades are preferred over swinging
blades because they are easier to clean, and work better than swinging
blades at low rotor speeds, when grinding fibrous material or if carefully
Figure 6 Different types of hammer mills: (A) vertical shaft (B) horizontal shaft.
Sizing of Granulation 503
installed either fixed or swinging (Fig. 7A–C). Fixed blades plough through
© 2005 by Taylor & Francis Group, LLC
controlled grinding is needed. The material to be ground determines the
configuration of the blades on the motor shaft, as well as the blade density.
The shape of the blades (straight, stepped, sickle, or other) is largely a matter
of designer preference; little empirical evidence exists to establish the
superiority of one shape over another. The size of the grinding chamber
generally determines the number of blades (e.g., a 6 in. grinding chamber
will have 16 blades).
4. Rotor speed: The size of the product is markedly affected by the speed of the
hammers. As a general rule, and with all other variables remaining constant,
the faster the rotor’s speed, the finer the grind. Usually, three speed settings
are available: slow (1000 rpm), medium (2500 rpm), and fast (4000 rpm).
Changes in rotor speed are accomplished by variable-speed drive, or by
manually changing the hammer drive and motor pulley ratio. Rotor speeds
of 2500–4000 rpm are typically used with blunt edges in fine grinding applications,
whereas speeds of 1000–2500 rpm are typically used with knifeedges
for coarse grinding. Particle size distributions are wider at low speed
than at medium and high speeds (12). Below the critical rotor speed, the
Figure 7 Different types of hammer mill rotors (Fitzmill): (A) cast rotor, (B) bar rotor and
(C) swing-blade rotor.
504 Rekhi and Sidwell
© 2005 by Taylor & Francis Group, LLC
material experiences attrition, rather than impact action, which causes
more spheroidal granules and may result in overheating of the material.
5. Screen size and type: The screen is usually an integral part of the hammer
mill and does not act as a sieve. The particle size of the product depends
on the openings in the screen, the thickness of the screen, and the speed
of the hammer. The particle size of the output granules will be much smaller
than the size of the screen used, because particles exit at an angle, with high
velocity.
Screens can be perforated, woven wire type, or with a slot configuration. The
screen openings may range in size and open area based on screen configuration.
Because of the large forces that the screens are subjected to, the perforated screens
are preferred over the woven-type screens. However, if the raw material fuses from
the heat generated, or if the material is difficult to mill, woven-type screens are
preferred for their increased open area. The herringbone design and cross-slot are
preferred for grinding amorphous and crystalline materials (Fig. 8).
7.2.2. Conical Screening Mill
Similarly, for conical screening mills, the operating variables affecting particle size
distribution are type of impeller, impeller speed, and screen size and type:
1. Material feed rate: In contrast to hammer mills, conical screening mills perform
with greater efficiency when the comminution chamber is kept relatively
full. Underfeeding results in low efficiency and reduced throughput.
2. Impeller: There are several types of impellers available (13); however, the
Knife-edge: Its principle mode of operation is shear, and hence, it is used for
compression-sensitive, heat-sensitive materials.
Round-edge: Its principle mode of operation is compression, and it provides
high throughput and low retention. It is mainly used for wet or dry deagglomeration–
delumping.
Figure 8 Different types of hammer mill screens.
Sizing of Granulation 505
four main types used frequently are as follows (Fig. 9A).
© 2005 by Taylor & Francis Group, LLC
Round-edge with teeth: It is the same as the round-edge impeller except that
it has teeth on one side, providing aggressive, high throughput. It reduces
fines in milling compacted materials by prebreaking with teeth and reducing
retention time. It is often used for tablet rework.
Knife-edge low-intensity impeller: It is used where a shear or cut is required;
it gives a scissor-like action, for fibrous materials or capsule rework.
3. Speed: The speed of the impeller can affect the particle size of the product.
Conical screening mills available have variable- or fixed-speed drives, however,
the number of revolutions per minute vary depending on the size of
the impeller. It is suggested that one keep the tip speed of the impeller
the same on scale-up to achieve the same particle size distribution.
4. Screen size and type: Screens are available in various sizes (Fig. 9B), based
on thickness, open area, and hole configuration such as round, square,
slotted, and grater-type openings. Only perforated screens are available.
Various researchers have performed extensive studies of the effects of the foregoing
variables on granulation and milling processes (20–22). Motzi et al. (21), based
on their observations of significant interaction effects, concluded that effects of mill
speed, screen size, and impeller shape on particle size distribution cannot be evaluated
individually, but must be evaluated at a level that is a combination of all three.
7.2.3. Hybrid Designs
utilize a one-piece cantilever rotor with heavy blades mounted parallel to the center
Figure 9 Conical screening mill (Comil): (A) impellers and (B) screens.
506 Rekhi and Sidwell
Hybrid designs, such as the Granumill (Fig. 10A and B), are now available which
© 2005 by Taylor & Francis Group, LLC
shaft. Incorporating a variable-speed drive, these mills operate as screening mills
when run at low speed, and as impact mills when run at high speed.
7.3. Other Variables
There are other variables that can affect the sizing process, such as feed material properties,
granulation process, and drying process. The properties of materials have been
discussed in Section 3. The type of granulation [i.e., dry (roller compaction), planetary,
high-shear, or fluid-bed] can determine the strength of the granules and, hence,
the sizing process. Furthermore, the drying process, whether tray or fluid-bed, can
also be important. Tray-dried granules are usually case hardened and difficult to mill,
whereas the fluid-bed process yields more porous and friable granules. Similarly,
granules produced by high-shear granulators are harder and are therefore more diffi-
cult to mill than those manufactured using low-shear or fluid-bed processes.
8. SCALE-UP
8.1. Hammer Mill
same screen size and type used on the lower scale, keeping the rotor tip speed constant
is one of the most important considerations in scale-up of a milling process.
Vertical and horizontal rotor configurations may affect throughput and also particle
size distribution.
8.2. Conical Screening Mill
the same impeller type, screen size, and screen type used on the lower scale, the tip
Figure 10 Hybrid design: (A) Granumill and (B) detail of Granumill rotor and screen.
Sizing of Granulation 507
Table 2 shows the various sizes of Fitzmills available (12). In addition to having the
Table 3 shows the scale-up parameters for various Comils (13). In addition to having
© 2005 by Taylor & Francis Group, LLC
Table 2 Scale-up Parameters for Fitzmill
Chamber Rotor
Model
Capacitya
factor
Nominal
width
Screen area
(in.2)
Diameter of
chamber
Rotor
configuration
No. of
blades
Tip speed
factorb
Maximum
rpm
Maximum
horsepower
Homoloid 0.4 2.5 43.0 6.625 Horizontal 12 1.73 7200 10.0
M5A 0.7 4.5 76.0 8.0 Horizontal 16 2.09 4600 3.0
D6A 1.0 6.0 109.0 10.5 Horizontal 16 2.75 4600 5.0
DAS06 1.0 6.0 109.0 10.5 Horizontal 16 2.75 4600 15.0
aThroughput relative to Model D6 at the same tip speed.
bTip speed?factoroperating speed.
Source: From Ref. 12.
Table 3 Scale-up Parameters for Comil
Model
Capacity
factor
Impeller
diameter
(in.)
Screen size (in.)a
Impeller speed scale-up comparison (rpm)
Motor
horsepower Infeed opening (in.) A B C
197 1 4.375 5 1.5 3 1200 2400 3600 4800 6000 7200 — 1 or 2 3 Round
194 5 7.625 8 2.5 5 700 1400 2100 2800 3500 4200 5600 5 6 Round
196 10 11.125 12 4 7 450 900 1350 1800 2250 2700 3600 10 or 15 8 Round
198 20 23.250 24 16 7 225 450 675 900 — — — 20 1122 Rectangular
199 40 29.469 30 16 12 180 360 540 — — — — 30 1224 Rectangular
Tip speed
(ft/min)
1400 2800 4200 5600 7000 8400 11,200
aA, screen upper diameter; B, screen lower diameter; C, screen height.
Source: From Ref. 13.
508 Rekhi and Sidwell
© 2005 by Taylor & Francis Group, LLC
speed of the impeller is one of the key variables in scale-up; thus, it should be kept
constant.
9. CASE STUDIES
9.1. Comparison of Fitzmill vs. Comil
It is often difficult to predict the results from similar pieces of equipment having the
same operating principle at two different scales, let alone using two pieces of equipment
having different operating principles. Many times in the development of a pharmaceutical
dosage form the equipment used during formulation development and
that used in production are quite different. Apelian et al. (22) studied the effect of particle
size distribution on chlorpheniramine maleate granules using a Fitzmill and
Comil. For the Fitzmill, various screens sizes (1, 2, 3, and so on) at medium speed
were evaluated, and for the Comil, impellers (1601 and 1607) at two speeds (1680
and 3420 rpm) using various screen sizes (039, 045, 055, and 055G) were studied. They
reported that milling the granulation using a Fitzmill with a screen size of 2, at medium
speed, gave a particle size distribution similar to the granulation milled using a
Comil (1601 impeller, 055 screen at 1680 rpm) (Fig. 11). The results of this study suggest
that in making a major change in the milling process, one needs to optimize the
critical processing variables in order to achieve a similar particle size distribution.
9.2. Comparison of Hand Screen vs. Comil
The effect of changing the dry milling from a hand screen operation to a conical
Figure 11 Particle size distribution of chlorpheniramine maleate granulations milled using
FitzMill and Comil.
Sizing of Granulation 509
screening mill is shown in Figure 12. Naproxen granulations (0.5 and 4 kg) were
© 2005 by Taylor & Francis Group, LLC
manufactured in a fluid-bed granulator using PVP K-90 as a binder (23). The particle
size distribution of granules (0.5 kg), passed manually through an 18-mesh screen,
was much coarser than that of the granules (4 kg) that were milled using a Comil
(Model 197S). A flat-faced impeller (1607) at an impeller speed of 2500 rpm with a
spacer setting of 0.25 in. and screen number 2A055 (14-mesh) were used for the
milling operation. Even though the granulations were prepared by the same procedure,
the milling conditions drastically affected the particle size distribution. As a
general rule, during a switch over from a low-energy milling operation to a highenergy
milling operation the screen size should be coarser in the high-energy mill.
The particle velocity is higher and therefore the size of the granule exiting out of
the screen is much smaller than the screen opening. As seen from Figure 12, when
the screen size was increased to 14-mesh for the conical-screening mill, the amount
of fines generated was higher. Hence, during scale-up, optimization of milling conditions
may be necessary to achieve the same particle size distribution.
10. LIST OF EQUIPMENT SUPPLIERS
1. The Fitzpatrick Company, 832 Industrial Drive, Elmhurst, IL 60126, USA;
2. Fluid Air, Inc., 2550 White Oak Circle, Aurora, IL 60504, USA;
3. Frewitt Ltd., P.O. Box 61, CH-1706 Fribourg, Switzerland;
4. Glatt Air Techniques Inc., 20 Spear Road, Ramsey, NJ 07446, USA;
Figure 12 Particle size distribution of naproxen granulations milled using hand screen and
Comil.
510 Rekhi and Sidwell
www.fitzmill.com
www.fluidairinc.com
www.frewitt.com
www.glattair.com
© 2005 by Taylor & Francis Group, LLC
5. Niro, Inc., 9165 Rumsey Road, Columbia, MD 21045, USA;
6. Quadro Inc., 55 Bleeker Street, Millburn, NJ 07041, USA;
7. Vector Corporation, 675 44th Street, Marion, IO 52302, USA;
ACKNOWLEDGMENTS
The authors would like to thank Dr. Murali K. Vuppala of Praecis Pharmaceuticals
Inc. for coauthorship of the first edition of this chapter, Mr. Scott Wennestrum from
The Fitzpatrick Company, Mr. John Bender from Fluid Air, Inc., Mr. David Adams
and Mr. Patrick Arthur from Quadro, Inc., and Mr. Jeff Montross from Glatt Air
Techniques, Inc. for providing the photographs shown in this chapter.
REFERENCES
1. Prescott JK, Hossfeld RJ. Maintaining product uniformity and uninterrupted flow to
direct-compression tableting presses. Pharm Tech 1994; 18:98–114.
2. Lantz RJ Jr. Size reduction. In: Lieberman HA, Lachman L, Schwartz JB, eds. Pharmaceutical
Dosage Forms: Tablets. Vol. 2. New York: Marcel Dekker, Inc., 1990:107–157.
3. Fonner DE, Anderson NR, Banker GS. Granulation and tablet characteristics.
Lieberman HA, Lachman L, eds. Pharmaceutical Dosage Forms: Tablets. Vol. 2. New
York: Marcel Dekker, Inc., 1981:201.
4. Parrot EL. Milling. Lachman L, Lieberman HA, Kanig JL, eds. The Theory and Practice
of Industrial Pharmacy. Philadelphia: Lea & Febiger, 1986:21–46.
5. O’Conner RE, Rippie ED, Schwartz JB. Powders. Gennaro AR, ed. Remington’s Pharmaceutical
Sciences. Easton, PA: Mack Publishing Company, 1990:1615–1617.
6. Carstensen JT, Puisieux F, Mehta A, Zoglio MA. Milling kinetics of granules. Int
J Pharm 1978; 1:65–70.
7. Steiner G, Patel M, Carstensen JT. Effect of milling on granulation particle—size distribution.
J Pharm Sci 1974; 63:1395–1398.
8. Motzi JJ, Anderson NR. The quantitative evaluation of a granulation milling process.
III. Prediction of output particle size. Drug Dev Ind Pharm 1984; 10:915–928.
9. Snow RH, Kaye BH, Capes CE, Sresty GC. Size reduction and size enlargement.
Perry RH, Green D, eds. Perry’s Chemical Engineers’ Handbook. New York:
McGraw-Hill Inc., 1984:8-9–8-20.
10. Prior MH, Prem H, Rhodes MJ. Size reduction. Rhodes MJ, ed. Principles of Powder
Technology. New York: John Wiley & Sons, 1990:237–240.
11. Kukla RJ. Strategies for processing heat-sensitive materials. Powder Bulk Eng 1988;
2:35–43.
12. FitzMill Technical Bulletin. The Fitzpatrick Company, Elmhurst, IL.
13. Quadro Inc., CoMil Product Literature, Millburn, NJ .
14. Poska RP, Hill TR, van Schaik JW. The use of statistical indices to gauge the mixing effi-
ciency of a conical screening mill. Pharm Res 1993; 10:1248–1251.
15. Fourman GL, Cunningham DL, Gerteisen RL, Glasscock JF, Poska RP. Improved color
uniformity in tablets made by the direct compression method: a case study. Pharm Tech
1990; 14:34–44.
16. Schwartz JB. Theory of granulation. Kadam KL, ed. Granulation Technology for Bioproducts.
Boca Raton, FL: CRC Press, 1990:17.
17. Johnson C. Comminution variables and options. Powder Bulk Eng 1989; 3:40–44.
Sizing of Granulation 511
www.niro.com
www.quadro.com
www.vectorcorporation.com
© 2005 by Taylor & Francis Group, LLC
18. Owens JM. How to correct common hammermill problems. Powder Bulk Eng 1991;
5:38–43.
19. Hajratwala BR. Particle size reduction by a hammer mill I: effect of output screen size,
feed particle size, mill speed. J Pharm Sci 1982; 71:188–190.
20. Byers JE, Peck GE. The effect of mill variables on a granulation milling process. Drug
Dev Ind Pharm 1990; 16:1761–1779.
21. Motzi JJ, Anderson NR. The quantitative evaluation of a granulation milling process. II.
Effect of output, screen size, mill speed and impeller shape. Drug Dev Ind Pharm 1984;
10:713–728.
22. Apelian V, Yelvigi M, Zhang GH, et al. Comparison of quadromill and fitzmill used in
milling process of granulation. Pharm Res 1994; 11:S-142.
23. UMAB/FDA Collaborative Agreement RFP # 223–91–3401.
512 Rekhi and Sidwell
© 2005 by Taylor & Francis Group, LLC
18
Granulation Characterization
Raj Birudaraj
Roche Palo Alto, Palo Alto, California, U.S.A.
Sanjay Goskonda
Durect Corp., Cupertino, California, U.S.A.
Poonam G. Pande
Synthon Pharmaceuticals Inc., Research Triangle Park, North Carolina, U.S.A.
1. INTRODUCTION
Granulation is a process selected by formulation scientists to prevent segregation of
formulation components in a powder mix, improve blend flow, bulk volume of
granulation, content uniformity, compressibility, and other properties. Granulation
minimizes the technical risks associated from batch to batch variability in raw materials
that could impact the manufacturing process and performance. Various granulation
techniques such as high- and low-shear granulation, roller compaction, spray
drying, fluid-bed granulation, extrusion speronization, and melt granulation are used
in solid dosage form development. The choice of granulation technique depends on
various factors such as chemical and physical stability of the final dosage form,
intended biopharmaceutical performance, and is occasionally limited due to available
equipment. Dosage form performance is assessed through a characterization
program in which drug dissolution, bioavailability, chemical stability, or manufacturing
ruggedness is taken into account. The scientist can rely on many different tools
for granulation characterization. These tools can probe the physical and chemical
attributes of the material. In this chapter many characterization techniques that
are applied to granulation will be reviewed. Most of the techniques used for granulation
characterization are conducted during the research and development stage of
product development. A significant portion of this chapter was adapted from the
previous edition of Handbook of Pharmaceutical Granulation Technology (1).
2. PHYSICAL AND CHEMICAL CHARACTERIZATION OF GRANULES
Physical property characterization of pharmaceutical granulations has been extensively
reported in literature. Chemical properties are equally important due to their
impact on specifications of a dosage form such as content uniformity, chemical
513
© 2005 by Taylor & Francis Group, LLC
purity, and in vitro performance. The interaction between physical characteristics of
a formulation and chemical-based performance should not be understated. In vivo
performance such as bioequivalence is the ultimate performance test because it is this
‘‘characterization’’ that determines whether a pivotal bioequivalency batch passes or
fails. The effect of granule size on the dissolution performance, for example, could
ultimately affect the outcome of such a bioequivalence study. The complexity of this
kind of relation underscores how something seemingly simple, such as particle size
and its dependence on granulation process parameters, can influence dissolution
and ultimately in vivo performance.
Physical characterization can be performed at molecular, particulate, or bulk
(macroscopic) levels. From the terminology cited by Brittain et al. (2), molecular
properties are associated with individual molecules, particulate properties are considered
as properties that pertain to individual solid particles, and bulk properties
are those that are associated with an assembly of particulate species. Most reports
in pharmaceutical literature cover characterization of bulk properties.
2.1. Particle Morphology
Particle morphology can be assessed using optical microscopy. Samples of granulation
can be evaluated directly under a microscope or sorted using a device proposed
by Ridgeway and Rupp (3) in which the granulation is fed onto a triangular metal
deck and vibrated. Particles of different shapes segregate on this deck and are
collected for analysis by microscopy.
Another technique to study particle morphology is scanning electron microscopy.
Yoshinari et al. (4) studied morphological changes in granule shape with
the addition of different amounts of granulating fluid (Fig. 1).
Figure 1 Scanning electron micrographs of mannitol granules obtained after treating
d-crystal with various ratio of water: (A) 5% w/w, (B) 10% w/w, (C) 15% w/w, and
(D) 25% w/w. (From Ref. 4.)
514 Birudaraj et al.
© 2005 by Taylor & Francis Group, LLC
Particle shape can be quantified by different methods. One popular method is
through the use of Heywood coefficients (5). The Heywood shape coefficient is defined
as the ratio of the surface shape coefficient (p for a sphere) to the volume shape coeffi-
cient (p/6 for a sphere); hence, the shape coefficient for a sphere would be 6.0. Applying
this to a cube and using its projected area in its most stable position, the shape
coefficient is 6.8. Cutting the cube in half in one dimension increases the shape factor
to 9.0, whereas it increases to 26.6 if that cube was sliced one-tenth in one dimension.
Further details of these types of calculations are provided by Rupp (5).
The effect of particle shape on bulk powder properties has been illustrated by
Rupp (5). The effect of particle size and shape on the bulk density and flow rate is
in the bulk becomes more efficient as the shape factor or loss in sphericity increases.
The flow rate becomes worse with loss in sphericity.
2.2. Particle Size Distribution
Particle size distribution can be measured by sieve analysis, laser light scattering, or
optical microscopy (2). Light-scattering techniques are generally not applied to
granulations due to the large size distribution of granules. Dry-sieve analysis and
Figure 2 Bulk density as a function of shape factor: triangles, 302 mm; squares, 461 mm; and
circles, 805 mm. (From Ref. 3.)
Granulation Characterization 515
illustrated in Figures 2 and 3. As illustrated in these examples, packing of powder
© 2005 by Taylor & Francis Group, LLC
microscopy are generally the most popular methods for determining size distribution
of granules. Microscopy provides a more exact measurement of size, although
it is the most labor-intensive method. Computer-aided image analysis techniques
have been employed to simplify the method, but some problems, such as threedimensional
surface effects could be misinterpreted by the computer, resulting in
errors in the calculated distribution.
Dry-sieve analysis is the easiest and the most convenient method for measuring
granule size. The granulation is placed on top of a stack of five to six sieves which
have successively smaller-sized openings from top to bottom. The stack is vibrated,
and the particles collect on top of the sieves. The data are usually represented in
terms of percentage retained on the sieve, or percentage that is undersize or oversize
Two factors—particle size distribution and shape—can bias a distribution
obtained by sieve analysis. According to Shergold (7), variations in particle size
distribution can be obtained due to the loading of the stack, which in turn, is affected
by the particle size distribution and shape of the powder being analyzed. In a more
extensive evaluation, Fonner and co-workers (8) studied the effect of loading, shaker
speed, and time on the data obtained for particle size distribution of model granulations.
Particle breakage occurred at increasing shaker speeds and times. According to
Figure 3 Effect of particle shape on flow rate for an orifice diameter of 6.34 mm: triangles,
302 mm; squares, 461 mm; and circles, 805 mm. (From Ref. 3.)
516 Birudaraj et al.
vs. screen-opening size (Fig. 4) (6).
© 2005 by Taylor & Francis Group, LLC
the authors, particle attrition causes the loading of material to influence the time to
reach equilibrium. In this article, time to reach equilibrium is defined as the time at
which the material passes through the sieve. The authors finally concluded that a
nonbiased analysis of particle size distribution should include an analysis of equilibrium
times for each sieve.
An example for use of sieve analysis was offered by Stevens (9), who addressed
the effect of granule size distribution on the weight variation of tablets by constructing
‘‘isovariational curves.’’ The intent of this evaluation was to define a geometric
mean diameter and distribution that would result in an acceptable tablet weight
variation. The geometric standard deviation was inverted and named the ‘‘R-value.’’
Granulations with various mean diameters and distributions were prepared by sieving,
recombining, and running on a rotary tablet press. The coefficient of variation
(CV) for tablet weight was recorded for each run. R-values and CV for each geometric
mean diameter were obtained from CV–R plots and then placed on a contour
plot containing geometric mean diameter (abscissa), R-value (ordinate), and CV
(isovariational contour lines). Although the author claimed this must be done on
a case-by-case (and press-by-press) basis, these curves can be used to select the
optimal mean granule size and distribution to minimize tablet weight variation.
Figure 4 Sieve analysis of lactose granules formed by massing and force screening showing
effect of binder fluid level: circles, 15.3% v/v; squares, 18.4% v/v; upward triangles, 23% v/v;
and downward triangles, 30.6% v/v. (From Ref. 6.)
Granulation Characterization 517
© 2005 by Taylor & Francis Group, LLC
2.3. Powder X-Ray Diffraction
A detailed description of powder x-ray diffraction is provided by Brittain et al. (2)
and the references cited therein. This technique is mostly used to determine the crystalline
form of a solid. Understanding the effect of granulation on the form of bulk
drug has gained increasing attention in the past few years (10). X-ray techniques provide
the formulator with a first line of analysis in the determination of polymorph
changes as a result of processing. In a joint American Association of Pharmaceutical
Science–Federal Drug Administration (AAPS–FDA) report, an experimental rationale
‘‘where both the applicant (industry) and the FDA are in better position to
assess the possible effects of any variations in the solid-state properties of the drug
substance’’ was provided. Both x-ray and solid-state nuclear magnetic resonance
were recommended techniques for assessing the crystalline form of drug substance
in a drug product (8).
Yoshinari et al. (4) used x-ray techniques to report polymorphic transition of
mannitol during wet granulation. These authors stated that after vacuum drying
the wet granulated d-crystal of mannitol, a polymorphic change occurred to b-crystal
polymorph (Fig. 5). Concomitant morphological changes were noted in the scanning
electron micrographs which showed that d-granules consisted of small
primary particles of the b-form of mannitol.
Modulated temperature x-ray powder diffraction (XRPD) is being used
increasingly in the pharmaceutical industry both at preformulation and at formulation
stages. Airaksinen et al. (11) studied polymorphic transitions during drying
using two methods: a multichamber microscale fluid-bed dryer or a variable temperature
powder x-ray diffractometer. Relative amounts of different polymorphic
Figure 5 Powder X-ray patterns of mannitol: (A) d-crystal of mannitol after granulation
and vacuum drying, (B) d-crystal of mannitol before granulation, and (C) b-crystal. (From
Ref. 4.)
518 Birudaraj et al.
(Fig. 6)
© 2005 by Taylor & Francis Group, LLC
forms of theophylline remaining in the dried granules were determined by XRPD.
The authors concluded that metastable anhydrous theophylline predominated when
the granules were dried at 40–50
C. Temperature >50
C produced mostly anhydrous
theophylline, more than 20% of the metastable form remained even at 90
C
Morris and co-workers (12) reported polymorphic changes when hydroxymethylglutarate
coenzyme A reductase inhibitor was wet-granulated with water.
The starting material was in the anhydrous form, which then converted to an
amorphous form during the wet-granulation process. The loss in crystallinity
was experimentally determined using powder x-ray diffraction. Exposure of this
Figure 6 Scanning electron micrographs of d-crystal of mannitol: (A) before and (B) after
granulation. (From Ref. 4.)
Granulation Characterization 519
(Fig. 7).
© 2005 by Taylor & Francis Group, LLC
granulation to an environment of >33% relative humidity, caused a form conversion
to its crystalline hydrate. This series of experiments demonstrated the usefulness of
a sophisticated technique, such as XRPD, in the assessment of the physical stability
of bulk drug during granulation.
Figure 7 (A) XRPD patterns of theophylline granules dried using a multichamber microscale
fluid-bed dryer (MMFD) at temperatures ranging from 30
C to 90
C using dry inlet
air (under 0.5 g/m3). (B) XRPD patterns of theophylline granules dried using MMFD at temperatures
ranging from 30
C to 90
C using ambient inlet air (7.6 g/m3). Note the characteristic
peaks of theophylline anhydrate metastable form (). XRPD patterns of theophylline
monohydrate and theophylline anhydrate at room temperature are shown below as controls.
(From Ref. 11.)
520 Birudaraj et al.
© 2005 by Taylor & Francis Group, LLC
2.4. Thermal Analysis
Standard thermal methods of analysis include differential scanning calorimetry
(DSC), differential thermal analysis, and thermogravimetric analysis. A common
application of thermal analysis in characterization of granulations is based on thermogravimetry
and is known as loss on drying (LOD). In a LOD analysis, a sample of
the granulation is heated at a temperature near the boiling point of water or solvent.
The weight loss, recorded directly on an analytical balance, is due to the evaporation
of water or solvent and is considered the residual moisture content of a granulation
(13). This technique is extensively used to establish both granulation and drying
parameters for wet-granulation unit operations (14–16).
DSC was used in the evaluation of a phase change of chlorpromazine hydrochloride
after wet granulation (17). The starting drug substance (form II) converted
to a hemihydrate (form I-H) after it was wet-granulated with water–ethanol binder
fluid. Form I-H converted to a partially dehydrated form (I-H0) after drying. An investigation
of this form led the authors to another anhydrous polymorph (form I). Extensive
characterization by DSC showed the transition from hemihydrate (form I-H) to
anhydrous form (form I) at 40–50
C, followed by a conversion from form I to form II
at135
C, and finally, melting at 185–189
C (17). By solution calorimetry, form I was
found to be the most stable based on their heat of solution (DH) values. Although the
dissolution profiles of both anhydrous forms were equivalent, the two forms had
different tableting characteristics. The more stable form I did not exhibit the severe
capping problems originally observed for form II. Granulations made with ibuprofen
and b-cyclodextrin were evaluated using DSC. Lower DH values for oven-dried
granulations indicated better complexation and faster dissolution than air-dried
granulations (18).
DSC is a valuable tool in characterizing granulations made by hot melt
extrusion for determining glass transition temperatures and crystalline or amorphous
nature of the formulation. Perissutti et al. (19) used this technique to characterize
carbamazepine and polyethylene glycol 4000 granules obtained after hot melt
granulation.
2.5. Near-Infrared Spectroscopy
Near-infrared (NIR) spectroscopy is a fast, nondestructive method and requires no
sample preparation (20) In the NIR region, absorption bands are mainly caused by
overtones and combination vibrations of CH, NH, and OH groups. Intact and rapid
NIR spectroscopy offers advantageous possibilities of in-line measurements during
granulation processes to measure moisture content and detect any physicochemical
changes (21,22). Rasanen et al. (20) studied the polymorphic conversion of theophylline
during wet granulation using NIR. The authors found that at a low level of
granulation liquid (0.3 mol of water per mole of anhydrous theophylline), water
absorption maxima in the NIR region occurred first at around 1475 and 1970 nm.
These absorption maxima were identical to those of theophylline monohydrate. At
higher levels of granulation liquid (1.3  2.7 mol of water per mole of anhydrous
theophylline), increasing absorption maxima occurred at 1410 and 1905 nm due to
Granulation Characterization 521
OH vibrations of free water molecules (Fig. 8).
© 2005 by Taylor & Francis Group, LLC
2.6. Electrostatic Charge
Static charge is generated when two bodies come into intimate contact and then are
separated. The surface of one of the contacting bodies attracts electrons from the
other surface, thereby resulting in a negative charge. Forces of attraction and repulsion
can cause significant problems in powder handling. Gold and Palermo (23)
described a technique that measures the static charge of a powder as it passes out
of a hopper. The surface separation between particles occurs frequently in this
dynamic environment. In the system proposed by these authors, powder was allowed
to flow out of a hopper onto a glass receptacle. Directly beneath this receptacle was
a copper disk that was attached to another copper disk beneath an ionostat. The
Figure 8 Second derivative of absorbance of anhydrous theophylline, theophylline monohydrate,
and theophylline granules at 1380–1520 nm: (A) anhydrous theophylline transformation
into theophylline monohydrate at around 1475 nm; and (B) effect of free water molecules at
around 1410 nm (number indicating mole(s) of water per mole of anhydrous theophylline).
(From Ref. 20.)
522 Birudaraj et al.
© 2005 by Taylor & Francis Group, LLC
ionostat recorded voltage transmitted by the first disk. It was found that acetaminophen,
when granulated with either starch paste or syrup, exhibited a much lower static
charge than ungranulated powder. This reduction in static charge correlated with
improved flow out of the hopper.
2.7. Surface Area
Granulation properties are mainly dependent on the size and surface area of particles
and granules (24,25) The surface area of a granule or particle can also affect the dissolution
rate of a solid. Gas adsorption is the most common method to determine
surface area, although liquid penetration methods have also been proposed (26).
In one of the methods developed by Brunauer, Emmet, and Teller, called the BET
method (27), an inert gas is adsorbed onto the surface of a solid at low temperature
and then desorbed at room temperature (1). Either nitrogen or krypton is used as the
adsorbate, and helium is usually used as a carrier gas for the adsorbate. Various concentrations
of adsorbate in carrier gas are used in this analysis to determine the
volume of gas that is adsorbed in a monolayer on to the solid. Eq. 2.1 is used to
determine this value
P
V?P0  P? ?
1
VmC ? ?C  1?P
VmCP0 ?2:1?
where V ? volume of gas adsorbed at pressure P, P ? partial pressure of adsorbate,
C ? a constant relating the heats of adsorption and condensation of the adsorbate,
P0 ? saturation pressure of adsorbate at experimental temperature, and Vm ? volume of gas adsorbed in monolayer of solid.
A plot of P/V (P0 – P) vs. P/P0 yields a straight line. Vm is calculated from
both the slope and the intercept of this line (27). The specific surface area (SSA),
in units of square meters per gram is calculated using Eq. 2.2
SSA ? ?VmN0Acs?
M ?2:2?
where N0 ? Avogadro’s number, Acs ? cross-sectional area of adsorbate, and M ? mass of solid sample.
Another method that has been proposed for measuring the surface area of
powders is known as air permeability (28). A column packed with powder is subjected
to a stream of air. The system is sealed off and the pressure drop is measured
across the bed. To take into account the ‘‘slip’’ of gas along walls of pores a modification
of the governing equation is employed (28) This correction minimizes the
apparent dependence of SSA on air pressure. Although this method has not been
extensively used on granulations or powders in the pharmaceutical sciences, it has
been applied to compressed tablets (29).
2.8. Granule Porosity
Mercury intrusion methods are routinely applied in the determination of pore size
and distribution to both granulations and tablet compacts (30). In this method,
excess pressure is applied to a nonwetting liquid, such as mercury, for the smallest
Granulation Characterization 523
© 2005 by Taylor & Francis Group, LLC
pores to be filled. A relation between the applied pressure and pore radius, known as
the Washburn equation Eq. 2.3 is then used to calculate pore size opening
DP ? 
2g
r 
 cosy ?2:3?
where DP?pressure difference across interface, g ? surface tension of the penetrating
liquid, y?contact angle of the penetrating liquid, and r ? pore radius.
In this analysis (2), the sample is introduced into the chamber, degassed, and
then completely covered with mercury. Pressure is incrementally applied, and the
volume of mercury that penetrates into the pores is recorded. A plot of cumulative
volume intruded vs. pressure is obtained and converted to cumulative volume vs.
pore radius. The pore size distribution is obtained by differentiating the latter curve.
Wikberg and Alderborn (31) have suggested that increase in intragranular porosity,
increases the propensity of the granules to fragment leading to formation of
stronger tablets. For granulations, which are less prone to fragmentation, Johansson
et al. (32) showed that increased intragranular porosity increased the degree of
deformation, resulting in formation of a closer intragranular pore structure during
compression and stronger tablets.
Ganderton and Hunter (6) used mercury intrusion to compare the intragranular
porosity of granulations manufactured by different processes. Calcium phosphate
was granulated with 10% dextrose–water binder and the comparison was made
between a pan granulator and a Z-blade mixer. A plot of intragranular porosity
decrease in intragranular porosity was observed when massed in a Z-blade mixer
and then screened. An opposite trend was noted for lactose granulated with water
as a binder. The results of this study suggest that the physicochemical properties
of the powder being granulated can have a significant influence on the sensitivity
of granule properties to granulation process.
Farber et al. (33) studied the porosity and morphology of granules by two different
techniques, x-ray computed tomography (XRCT) and mercury porosimetry
of total porosity when compared to mercury porosimetry. However, XRCT provided
detailed morphological information such as pore shape, spatial distribution,
2.9. Granule Strength
The strength of granules is a property that can be measured in the development of a
formulation. In a classic treatment of the topic, Rumpf (34) has described numerous
mechanisms that contribute to granule and agglomerate strength: (a) solid bridges
between particles, (b) interfacial forces and capillary pressure in moveable liquid
surfaces, (c) adhesional and cohesional forces in bonding, (d) attraction between solid
particles, and (e) particle shape-influenced mechanical interlocking. Rumpf (34)
considered the tensile strength of the particle bond to be the most important factor
in the determination of agglomerate strength. He proposed a direct test of the tensile
strength of agglomerates; although the size of the granules that were tested was
2.5 cm (1 in.) in diameter.
Several methods have been proposed in the pharmaceutical literature for the
measurement of the strength of granules. One way of measuring granule strength is
524 Birudaraj et al.
vs. binder level (denoted as moisture content) is shown in Figure 9. A significant
(Fig. 10). These authors concluded that XRCT is less accurate in the determination
and connectivity (Fig. 11).
© 2005 by Taylor & Francis Group, LLC
by directly crushing them, a test proposed by Harwood and Pilpel (35) and subsequently
modified by both Ganderton and Hunter (6) and Gold et al. (36). The
force required to crush the granule is recorded when a platen is moved at a constant
strain rate. Deflections in the load profile are interpreted as break points and the
strength is recorded in units of mass or force (36). A direct measurement of tensile
strength is difficult owing to the lack of understanding of the surface area on which
the applied load acts.
Mehta and co-workers (37) have criticized this method because of the inherent
variability of the properties of granules as well as for the difficulty in measuring the
strength of granules smaller than 40 mesh. Many samples must be measured and
results averaged to make the granule-crushing measurement meaningful. In their
method, ball milling is used to estimate granule strength by following granule attrition
(37). Attrition is monitored using dry sieve analysis. Their method correlated
very well with the Harwood-Pilpel method with a much better coefficient of variation
for their technique (1–5% for the ballmilling method and 50–80% for the directcrushing
method) (37)
Another attrition method used in determining granule strength is determination
of the friability of granulations (38,39). In this measurement, a friabilator is
charged with the granulation to be tested and then rotated a set number of times.
The percentage loss of mass for a particular size is usually the value that is
represented in a granule friability analysis (30,39). This method is useful to identify
Figure 9 The effect of moisture content on the porosity of –12?16 mesh calcium phosphate
granules processed for 10 min: circles, pan granulated; squares, massed and screened. (From
Ref. 6.)
Granulation Characterization 525
© 2005 by Taylor & Francis Group, LLC
properly the differences as a function of granule size within a granulation (39). An
2.10. Granule Flowability and Density
Flow behavior of granules is affected by multiple variables such as physical properties
of the granulation and the equipment design used for handling during a given
process. Prescott and Barnum (40) show the value of selecting flow property measurements
that are meaningful to actual processing. Rump and Herrmann (41)
and Podczeck (42) provide more information on underlying particle properties that
contribute to powder flow.
Specific volume is one of the properties of a powder that is believed to affect
powder flowability. Specific volume is determined by pouring a known mass of blend
into a graduated cylinder. The volume is read off the cylinder and the specific volume
is calculated by dividing the volume by the mass of the blend (43). Bulk density is
calculated by dividing the mass by volume. The compressibility of a blend can also
be determined at this time. The granulated cylinder is vibrated on a shaker for a time
period. This vibration reduces the volume that the blend occupies in the graduated
cylinder. The percentage compressibility (6) is calculated as in Eq. 2.4
% Compressibility ?
100  ?P  A?
P ?2:4?
where P ? packed density (after vibration) and A ? bulk density (untapped).
Figure 10 The granule growth of the one of the batch. Each dot (.) shows the data point for
the particle mean size measured from the surface image information. Each numbered dot
corresponds to the numbered surface images. Additionally, three end-point data points and
images are shown together with three replicate data points from end-point sieve analysis (.).
The spraying and drying phases of the process are separated with a dashed line. (From
Ref. 33.)
526 Birudaraj et al.
example is given for a sulfadiazine granulation in Figure 12.
© 2005 by Taylor & Francis Group, LLC
The percentage compressibility is commonly known as Carr’s index (44). This
index is interpreted in the following way: the higher the compressibility, the poorer
the flowability. An example of how this is used in the evaluation of a granulation
process was reported by Ertel et al. (45). In this report, the authors evaluated effects
of the scale of preparation of a sucrose–lactose–starch granulation using Lodige
granulators at the 4 and 30 kg scale. The compressibility of the granulations was
dependent on the kneading time as well as the scale of preparation.
Powder flowability can be directly measured using the ‘‘flow through an ori-
fice’’ technique (12,46). This test is close to an actual ‘‘use test,’’ because a hopper
is charged with the blend and the flow rate during discharge is measured. A variation
of this test is achieved by determining the relation between the flow rate of a blend
and the diameter of the orifice. Harwood and Pilpel (35) describe various methods of
Figure 11 Tomographic images and pore size distributions of granules produced under high
shear in the Fukae mixer. (A) X-ray shadow image; (B) typical reconstructed cross-sectional
image; (C) high-magnification image of area in the box in 9b; (D) pore size distribution calculated
from tomographic image; and (E) pore size distribution as measured by mercury porosimetry.
(From Ref. 33.)
Granulation Characterization 527
© 2005 by Taylor & Francis Group, LLC
data analysis for this technique in detail. These methods include empirical equations,
dimensional analysis, and energy considerations.
A more fundamental evaluation of powder flowability can be obtained by
assessing interparticulate friction. The angle of repose is a parameter that is dependent
on interparticulate friction and cohesion (47–49). The static angle of repose is
obtained by filling an open-ended cylinder with the blend and then slowly raising
the cylinder, allowing the powder to flow out. The angle of the heap that is formed
is called as the static angle of repose (46). In general, the higher the angle of repose
the poorer the flowability of the blend. A dynamic, or kinetic, angle of repose can be
obtained by charging a hollow drum with blend and rotating it. The angle of repose
can be calculated by recording how high the blend travels up the wall of the drum
(46,50). This angle of repose will be 1–5
 lower than the static angle (46). However,
there are some problems with this technique. Danish and Parrott (47) found that the
angle of repose of blends did not correlate with the flow through an orifice data, a
result that is in agreement with other authors (36).
A better method for determining the cohesive and frictional effects of particles
is by using a shear cell (48,51,52). There are various cell configurations, the most
and then compressed by twisting the lid of the cell. The number of twists required
to load the powder to the point at which the resistance to shear (measured as stress
applied to ring around the bed) is constant. This phase of the test is known as ‘‘shear
consolidation.’’ The load is reduced and the resistance to shear is then recorded. A
‘‘yield locus’’ of this shear stress vs. the reduced load is obtained and used to calculate
various flow-related parameters (47,48,51). Numerous parameters can be
Figure 12 Friability of sulfadiazine granulations. (From Ref. 39.)
528 Birudaraj et al.
popular proposed by Jenike (51). In the Jenike cell (Fig. 13), a powder is loaded
© 2005 by Taylor & Francis Group, LLC
derived from the yield locus: flow factor, shear index, cohesion, tensile strength,
effective angle of internal friction, and unconfined yield stress (48). All of these parameters
can be used in the characterization of powder flowability.
Other shear cells have been used to characterize the flowability of blends. Carr
and Walker describe an annular shear cell that measures the resistance to the angular
movement of the shoe that is placed on top of the powder (53). The advantage to this
type of design is that unlimited travel of the shoe provides the opportunity to
measure successive initial consolidation loads without reloading the powder (53)
Hiestand (52) and later on, Amidon and Houghton (54), describe a plate-type shear
cell that is similar to the Jenike shear cell, with the exception that the powder bed is
unconstrained at the edge of the bed.
Sinko and co-workers (55) have demonstrated the usefulness of the plate-type
shear cell in directly measuring the improvement in flowability that is afforded
by wet granulation. In this study, the measurement of flowability estimated by the
effective angle of internal friction, a parameter that estimates the powder’s resistance
to shear stress as a function of applied load, was used to track the potential
improvement in flowability achieved through dry- and wet-granulation. A comparison
of the effective angle of internal friction for a direct compression, wetgranulated,
and dry-granulated formulation of near equivalent composition is
flowability of the formulation than a similarly sized dry granulation.
2.11. Moisture Control in Granulations
Control of moisture content in granulations is very important and it could affect
the physical and chemical performance of final dosage forms. Moisture could affect
flow of granules, tablet compression, tablet disintegration, crystal habit, capsule
Figure 13 Schematic of Jenike shear cell. (From Ref. 48.)
Granulation Characterization 529
provided in Figure 14. Wet-granulation resulted in a greater improvement in the
© 2005 by Taylor & Francis Group, LLC
brittleness, chemical stability, and many other properties. Moisture content is generally
measured using moisture analyzer during product development; a thin layer of
sample is heated at a set temperature until it reaches a constant weight and the
results are expressed as LOD. Chowhan (56) showed that tablets made using low
moisture content granulations when exposed to higher humidity increased in hardness
and disintegration. Badawy et al. (57) showed in a processing study the moisture
content had the largest effect on compressibility of the granulation compared to
seven other process parameters. Within the tested levels increasing moisture content
increased the granulation compressibility. It should be noted that due to the great
diversity of granulation formulation, one has to characterize each of their formulation
and process to evaluate the moisture effect on physical and chemical properties.
Some polymorphic transitions in granulations are moisture mediated.
Wostheinrich and Schmidt (58) showed that thiamine hydrochloride granulations
were caking on storage and the tablets were increasing in hardness due to conversion
of form I crystal to hydrate forms due to exposure to humidity at room temperature.
Williams et al. (59) indicated that the active ingredient which is highly crystalline
hydrochloride form was dissociated in minute quantities (1% of tablet weight) into
amorphous free form due exposure to moisture and altering tablet properties. Minimizing
moisture exposure during process and storage was recommended.
It is ideal to develop equilibrium moisture isotherms for granulations to understand
the moisture content at different humidities. To develop moisture isotherms
granulations are exposed to different relative humidities at a set temperature and
the equilibrium moisture content is determined. This information could be used to
develop specifications for the moisture content of the granulation and would help
device ideal processing and packaging conditions. One application of moisture
Figure 14 The effect of process on the flowability of lactose based granulations: DC, direct
compression, RC, roller compaction; Wet, wet granulated. (From Ref. 55.)
530 Birudaraj et al.
© 2005 by Taylor & Francis Group, LLC
isotherms data could be applied to the formulation development of capsules. Capsules
show brittleness at low relative humidity and a tendency to cross-link at high
humidity and high temperature. Asorption–desorption moisture transfer model
was developed by Zografi et al. (60) to predict moisture transfer between excipients,
and this model was used later to control brittleness of capsules successfully by Kotny
et al. (61) and Chang et al. (62).
3. CONCLUSION
The diverse nature of formulations and processes are used to manufacture granulations
calls for various tools that are needed to help characterize granule properties.
The tools mentioned in this chapter could be used as a guide for the formulationprocess
scientist during research and development stages. There are also several
other sophisticated tools (like Raman imaging) still being evaluated for specific
characterization of the granulation like uniformity of drug distribution.
REFERENCES
1. Sinko C. Granulation characterization: methods and significance. Parikh DM, ed. Handbook
of Pharmaceutical Granulation Technology. New York: Marcel Dekker, 1997.
2. Brittain HG, Bogdanowich SJ, Bugay DE, DeVincentis J, Lewen G, Newman AW.
Physical characterization of pharmaceutical solids. Pharm Res 1991; 8:963.
3. Ridgeway K, Rupp R. J Pharm Pharmacol 1969; 21:30S.
4. Yoshinari T, Forbes RT, York P, Kawashima Y. The improved compaction properties of
mannitol after a moisture-induced polymorphic transition. Int J Pharm 2003; 258:121.
5. Rupp R. Flow and other properties of granulates. Bol Chim Farm 1977; 116:251.
6. Ganderton D, Hunter BM. A comparison of granules prepared by pan granulation and
by massing and screening. J Pharm Pharmacol 1971; 23:1S.
7. Shergold FA. The effect of sieve loading on the results of sieve analysis of natural sands.
J Soc Chem Ind 1946; 65:245.
8. Fonner DF, Banker GS, Swarbrick J. Micromeritics of granular pharmaceutical solids II:
factors involved in the sieving of pharmaceutical granules. J Pharm Sci 1966; 55:576.
9. Stevens PTA. The relationship between particle size distribution of granules and weight
variation of tablets. Acta Pharm Technol 1980; 26:304.
10. Byrn S, Pfeiffer R, Ganey M, Hoiberg C, Poochikian G. Pharmaceutical solids: a strategic
approach to regulatory considerations. Pharm Res 1995; 12:945.
11. Airaksinen S, Karjalainen M, Ra?sa?nen E, Rantanen J, Yliruusi J. Comparison of the
effects of two drying methods on polymorphism of theophylline. Int J Pharm 2004;
276:129.
12. Morris KR, Newman AW, Bugay DE, Randive SA, Singh AK, Szyper M, Varia SA,
Brittain HG, Serajuddin ATM. Characterization of humidity-dependent changes in crystal
properties of a new HMG-CoA reductase inhibitor in support of its dosage form
development. Int J Pharm 1994; 108:195.
13. Carstensen JT. Pharmaceutical Principles of Solid Dosage Forms. Lancaster, PA: Technomic
Publishing, 1993.
14. Helle?n L, Yliruusi J, Merkku P, Kristofferson E. Process variables of instant granulator
and spheroniser: I. Physical properties of granules, extrudate, and pellets. Int J Pharm
1993; 96:197.
15. Lipps DM, Sakr AM. Characterization of wet granulation process parameters using
response surface methodology. I. Top-spray fluidization. J Pharm Sci 1994; 83:937.
Granulation Characterization 531
© 2005 by Taylor & Francis Group, LLC
16. Schaefer T, Worts O. Control of fluidized bed granulation. III. Effects of inlet air distribution.
Control of moisture content of granules in the drying phase. Arch Pharm Chem
Sci Ed 1978; 6:1.
17. Wong MWY, Mitchell AG. Physiochemical characterization of a phase change produced
during the wet granulation of chlorpromazine hydrochloride and its effects on tableting.
Int J Pharm 1992; 88:261.
18. Ghorab MK, Adeyeye MC. Enhancement of ibuprofen dissolution via wet granulation
with beta-cyclodextrin. Pharm Dev Technol 2001; 6:305.
19. Perissutti B, Rubessa F, Moneghini M, Voinovich D. Formulation design of carbamazepine
fast-release tablets prepared by melt granulation technique. Int J Pharm 2003;
256:53.
20. Ra?sa?nen E, Rantanen J, Jorgensen A, Karjalainen M, Paakkari T, Ylirussi J. Novel identification
of pseudopolymorphic changes of Theophylline during wet granulation using
near infrared spectroscopy. J Pharm Sci 2001; 90:389.
21. Aldridge PK, Evans CL, Ward HW II, Colgan ST, Boyer N, Gemperline PJ. Near-IR
detection of polymorphism and process-related substances. Anal Chem 1996; 68:997.
22. Buckton G, Yonemochi E, Hammond J, Moffat A. The use of near infra-red spectroscopy
to detect changes in the form of amorphous and crystalline lactose. Int J Pharm
1998; 168:231.
23. Gold G, Palermo BT. Hopper flow electrostatics of tableting material I. Instrumentation
and acetaminophen formulations. J Pharm Sci 1965; 54:310.
24. Chalmers AA, Elworthy PH. Oxtetracycline tablet formulations: effect of variations in
binder concentration and volume on granule and tablet properties. J Pharm Pharmacol
1976; 28:228.
25. Harwood CF, Pilpel N. Granulation of griseofulvin. J Pharm Sci 1968; 57:478.
26. Carmen PC. The determination of the specific surface of powders. I. J Soc Chem Ind
1938; 57:225.
27. Brunaeuer S, Emmett PH, Teller E. Adsorption of gases in multimolecular layers. J Am
Chem Soc 1938; 60:309.
28. Rigden PJ. The specific surface of powders. A modification of the theory of the airpermeability
method. J Soc Chem Ind 1947; 66:130.
29. Alderborn G, Duberg M, Nystrom C. Studies on direct compression of tablets X.
Measurement of tablet surface area by permeametry. Powder Technol 1985; 41:49.
30. Hayes JM, Rossi-Doria P. Principles and Applications of Pore Structural Characterization.
Bristol, UK: JW Arrowsmith, 1985.
31. Wikberg M, Alderborn G. Compression characteristics of granulated materials. IV. The
effect of granule porosity on the fragmentation propensity and compactibility of some
granulations. Int J Pharm 1991; 69:239.
32. Johansson B, Wikberg M, Ek R, Alderborn G. Compression behaviour and compactibility
of microcrystalline cellulose pellets in relationship to their pore structure and mechanical
properties. Int J Pharm 1995; 117:57.
33. Farber L, Tardos G, Michaels JN. Use of X-ray tomography to study the porosity and
morphology of granules. Powder Technol 2003; 132:57.
34. Rumpf H. The strength of granules and agglomerates. Knepper WA, ed. Agglomeration.
New York: Wiley Interscience, 1962; 379.
35. Harwood CF, Pilpel N. Granulation of griseofulvin. J Pharm Sci 1968; 57:478.
36. Gold G, Duvall RN, Palermo BT, Slater JG. Powder flow studies II: effect of glidants on
flow rate and angle of repose. J Pharm Sci 1996; 55:1291.
37. Mehta A, Zoglio MA, Carstensen JT. Ball milling as measure of crushing strength of
granules. J Pharm Sci 1978; 67:905.
38. Hill PM. Starch paste granulations: binder dilution effects on granulations and tablets.
J Pharm Sci 1976; 65:313.
39. Marks AM, Sciarra JJ. Effect of size on other physical properties of granules and their
corresponding tablets. J Pharm Sci 1968; 57:497.
532 Birudaraj et al.
© 2005 by Taylor & Francis Group, LLC
40. Prescott JK, Barnum RA. On powder flowability. Pharm Technol Oct 2000; 60–84.
41. Rumpf H, Herrmann W. Properties, bonding mechanisms, and strength of agglomerates.
Process Prep 1970; 11:117.
42. Podczech F. Particle-particle adhesion. Pharmaceutical Powder Handling. London:
Imperial College Press, 1998:256.
43. Lachman L, Lieberman HA, Kanig JL, eds. Theory and Practice of Industrial Pharmacy.
Philadelphia, PA: Lea & Febiger, 1986.
44. Gordon RE, Rosanske TW, Fonner DE, Anderson NR, Banker GS. Granulation technology
and tablet characterization. Lieberman HA, Lachman L, Schwartz JB, eds. Pharmaceutical
Dosage Forms: Tablets. New York: Marcel Dekker, 1990.
45. Ertel KD, Zoglio MA, Rischel WA, Carstensen JT. Physical aspects of wet granulation
IV: effect of kneading time on dissolution rate and tablet properties. Drug Dev Ind
Pharm 1990; 16:963.
46. Pilpel N. The flow of powders and granular solids. Br Chem Eng 1966; 11:699.
47. Danish FQ, Parrott EL. Flow rates of solid particulate pharmaceuticals. J Pharm Sci
1971; 60:548.
48. York P. Powder failure testing—pharmaceutical applications. Int J Pharm 1980; 6:89.
49. Train D. Some aspects of the property of angle of repose of powders. J Pharm Pharmacol
1958; 10:127T.
50. Carr RL. Evaluating the flow properties of solids. Chem Eng 1965; 72:163.
51. Jenike AW. Gravity Flow of Solids, Bulletin 108. Salt Lake City, UT: Utah Engineering
Experimental Station, University of Utah, 1961.
52. Hiestand EN. The determination of some important physical properties of powders.
Pharm Ind 1972; 34:262.
53. Carr JF, Walker DM. An annular shear cell for granular materials. Powder Technol
1968/1969; 1:369.
54. Amidon GE, Houghton ME. Powder flow testing in preformulation and formulation
development. Pharm Manuf 1985.
55. Sinko CM, Carlson GT, Beckwith BA. Mechanical property modification of powder.
Proceedings of the First International Particle Technology Forum, Denver, CO,
1994; 121.
56. Chowhan ZT. Moisture, hardness, disintegration, and dissolution interrelationships in
compressed tablets prepared by wet granulation process. Drug Dev Ind Pharm 1979;
5:41.
57. Badawy SI, Menning MM, Gorko MA, Gilbert DL. Effect of process parameters on
compressibility of granulation manufactured in a high-shear mixer. Int J Pharm 2000;
198:51.
58. Wostheinrich K, Schmidt PC. Polymorphic changes of thiamine hydrochloride during
granulation and tableting. Drug Dev Ind Pharm 2001; 27:481.
59. Williams AC, Cooper VB, Thomas L, Griffith LJ, Petts CR, Booth SW. Evaluation of
drug physical form during granulation, tabletting, and storage. Int J Pharm 2004; 275:29.
60. Zografi G, Grandolfi GP, Kotny MJ, Mendenhall DW. Prediction of moisture of moisture
transfer in mixtures of solids: transfer via vapor phase. Int J Pharm 1988; 42:77.
61. Kontny MJ, Mulski CA. Gelatin capsule brittleness as a function of relative humidity at
room temperature. Int J Pharm 1989; 54:79.
62. Chang RK, Raghavan KS, Hussain MA. A study of gelatin capsule brittleness: moisture
transfer between the capsule shell and its content. J Pharm Sci 1988; 87:556.
Granulation Characterization 533
© 2005 by Taylor & Francis Group, LLC
19
Bioavailability and Granule Properties
Sunil S. Jambhekar
South University School of Pharmacy, Savannah, Georgia, U.S.A.
1. INTRODUCTION
The most important property of a dosage form is its ability to deliver the active
ingredient to the ‘‘site of action’’ in an amount sufficient to elicit the desired pharmacological
response. This property of a dosage form has been variously referred to as
its physiological availability, biological availability, or bioavailability.
Bioavailability may be defined more accurately as the rate and extent of
absorption of a drug from its dosage form into the systemic circulation. Accordingly,
the absorption of a drug following intravenous administration is extremely rapid and
complete. However, due to convenience and stability problems, drugs are often
administered orally in a tablet or capsule dosage form. Therefore, it is imperative
that their rate and extent of absorption in individual be known accurately. Furthermore,
it is equally important that the factors that influence the rate and extent of
absorption of drugs be also known and understood by formulators.
The subject of bioavailability began to receive growing attention as studies
showed that the therapeutic effectiveness of a drug from the dosage form depends,
to a large extent, on the physiological availability of their active ingredient(s), and
is a function of the drug concentration in the patient’s blood or plasma. The importance
of bioavailability in drug therapy, therefore, stems from the fact that the rate
and extent of absorption of a drug from a dosage form can, in fact, affect the patient’s
response to a drug. In light of these facts, the determination of bioavailability has
become one of the ways to assess the in vivo performance of a dosage form following
its formulation development. It must, however, be remembered that bioavailability
studies, very often, are conducted in normal, fasted, and a small number of subjects
and, therefore, the results of these studies may not always reflect the true efficacy
relationship in patients under treatment conditions.
For many years it was assumed that if a dosage form contained the labeled
amount of a drug, its performance could be taken for granted. However, it is
now evident for some time that many factors acting individually or in concert
may produce the therapeutic failure.
535
© 2005 by Taylor & Francis Group, LLC
2. BIOAVAILABILITY PARAMETERS
In assessing the bioavailability of a drug from a dosage form, three parameters are
measured following the administration of a drug through a dosage form and obtaining
the drug blood concentration time profile (Fig. 1):
1. Peak concentration (Cp) max
2. Peak time, tmax
3. The area under the concentration time curve, ?AUC?10
The parameters tmax and (Cp) max are the measure of the rate of absorption and
the ?AUC?10
is a measure of the extent of absorption.
2.1. Peak Time
This parameter represents the length of time required to attain the maximum concentration
of drug in the systemic circulation. The parameter reflects the onset of action
and, hence, can be utilized as a measure of the rate of absorption. The faster the rate
of absorption, the smaller the value for the peak time and the quicker the onset of
action of the drug. The peak time is determined by using the following equation:
tmax ?
In?Ka=K?
?Ka 
 K? ?2:1?
where Ka and K are the first-order absorption and elimination rate constants, respectively.
Eq. 2.1 clearly indicates that larger the value of the absorption rate constant
(Ka), the smaller is the value of peak time (tmax) and the quicker is likely to be the
onset of action.
Figure 1 A graphical representation of plasma/serum drug concentration data following the
administration of a drug by an extravascular route.
536 Jambhekar
© 2005 by Taylor & Francis Group, LLC
The elimination rate constant (K) is constant for a drug in normal healthy individuals
and it changes when organs responsible for the elimination of the drug (i.e.,
kidney and liver) exhibit abnormalities. The absorption rate constant (Ka), on the
other hand, depends on the route of administration, the dosage form, and the formulation
of a drug. And, for hydrophobic drugs or when the absorption is dissolution
rate limited, the faster dissolution is generally reflected in the higher value for the
absorption rate constant. Therefore, by changing the formulation of a drug or route
of administration, one can alter the peak time and, therefore, the rate of absorption
and time for the onset of action.
2.2. Peak Plasma Concentration (Cp) max
This parameter represents the highest drug concentration in the systemic circulation
or the plasma concentration that corresponds to the peak time. Furthermore, this
parameter is often associated with the intensity of the pharmacologic response of
the administration of a dosage form should be above the minimum effective concentration
and below the minimum toxic concentration. The peak plasma concentration
can depend on the absorption rate constant (Ka) and the fraction of the administered
dose that eventually reaches the systemic circulation. The higher the absorption rate
constant and fraction that reaches the general circulation, the greater is the peak
plasma concentration for the administered dose. The route of administration, the
dosage form, and the formulation can therefore influence the peak plasma concentration.
It is determined by using the following method:
?Cp?max ?
KaF?Xa?0
V?Ka 
 K? ?e
ktmax 
 e
katmax? ?2:2?
where F is the fraction of the dose that eventually reaches the systemic circulation,
(Xa)0 is the administered dose, V is the apparent volume of distribution of a drug,
and tmax is the peak time.
Since the term ?KaF?Xa?0?=?V?Ka 
 K?? in Eq. 2.2 constitutes the intercept of
the plasma concentrations against the time profile, Eq. 2.3 can be written as
?Cp?max ? I?e
ktmax 
 e
katmax? ?2:3?
where I is the intercept (mg/mL) of the plasma concentrations against time profile.
2.3. Area under the Plasma Concentration Time Curve
This parameter represents the extent of absorption of a drug following the administration
of a dosage form. The greater the fraction of the dose that reaches the general
circulation the greater is the extent of the absorption and, hence, ?AUC?10
. The term
?AUC?10
, expressed as mg/mL/hr, for a drug following its administration by various
extravascular routes or various dosage forms that are administered extravascularly,
is determined by employing the following equation:
?AUC?10
?
KaF?Xa?0
V?Ka 
 K?
1
K 

1
Ka 
  ?2:4?
All the terms of the Eq. 2.4 have been defined previously. Eq. 2.4 can further be
reduced to
Bioavailability and Granule Properties 537
the drug. Therefore, the peak plasma concentration (Fig. 1) of a drug following
© 2005 by Taylor & Francis Group, LLC
?AUC?10
? Intercept
1
K 

1
Ka 
  ?2:5?
the intercept in Eq. 2.5 being the intercept of the plasma concentration time profile.
The extent of absorption can also be determined by using the following equation:
?AUC?10
?
F?Xa?0
VK ?2:6?
where, the term VK is the systemic clearance of the administered drug. This parameter
being independent of the route of administration, the formulation, and the
extravascularly administered dosage form, it is ostensible that the extent of absorption
[i.e., ?AUC?10
] is controlled by the product of the fraction of the administered
dose reaching the general circulation and the administered dose [i.e., F(Xa)0].
2.4. Factors Affecting the Bioavailability
There are a number of factors responsible for the variation in bioavailability.
Broadly speaking, these factors can be classified as patient related or dosage form
related. Patient related factors include age, disease state, abnormal genetic characteristics,
and gastrointestinal physiology. The detailed discussion on these factors is
beyond the scope of the objectives of this chapter.
Dosage form related factors include formulation and manufacturing related
variables such as particle size, type, and quantity of excipient used, method of manufacturing,
compression pressure, derived properties of the powder, and many other
factors.
The fact that the bioavailability of a drug may be significantly affected by its
physical state of the drug and the dosage forms via which it is administered has been
unequivocally demonstrated. And, because drugs are administered through dosage
forms, these dosage forms should have adequate stability, consistent bioavailability,
and uniform composition.
Following the administration of a drug through a solid dosage form, a
sequence of steps are required before the drug reaches the systemic circulation. As
tion and deaggregation, followed by the dissolution of the drug. The dissolved drug
molecules must penetrate the gastrointestinal membrane and picked up by the blood.
Each of the steps involved may limit how fast the drug molecules reach the general
circulation and, therefore, site of action. The step that offers the maximum resistance
is referred to as the rate-limiting step. Which step will be rate limiting, on the other
hand, will depend on the physicochemical properties of the dosage form and the physiology
of the gastrointestinal tract. The focus of the discussion here, however, will
be on the physicochemical properties of the dosage form.
As illustrated in Figure 2, solid dosage form must disintegrate or deaggregate
before much of the drug is available for absorption. Drug dissolution subsequently
occurs from the resulting granules. Therefore, the properties of granules are important
in understanding how dissolution is influenced by these properties. Following
the ingestion of a solid dosage form, whether or not a drug is deaggregated, it will
not be absorbed until it has dissolved into the luminal fluids of the gastrointestinal
tract. Because of the effects of disintegration and deaggregation on the dissolution,
the remaining discussion will focus on the factors influencing the dissolution of
the drug.
538 Jambhekar
shown in Figure 2, an orally administered solid dosage form undergoes disintegra-
© 2005 by Taylor & Francis Group, LLC
2.5. Dissolution and Granule Properties
Amongst available dosage forms, compressed tablets are the most widely used dosage
form. Tablets are generally obtained by using either wet granulation or direct
compression process. Wet granulation process consists of mixing a drug with other
powdered material and wetting the mixture with an aqueous or hydroalcoholic solution
of a suitable binder such as gelatin, starch, or polyvinylpyrrolidone. The damp
mass is passed through the screens of 8–12 mesh and dried to produce cohesive
granules. Each granule, in theory, is a blend of an active ingredient and excipients.
The granules flow easily through the hopper into the tablet press and are easily
compressed.
Many derived properties of the powder greatly influence the granule properties,
which, in turn, influence the dissolution of an active ingredient from the dosage
form. These derived properties include powder density, porosity, specific surface,
particle number, and powder flow. These derived properties are, in essence, determined
by the particle size and size distribution. Consequently, particle size and size
distribution play a vital role in influencing the bioavailability of drugs, particularly,
when dissolution is the rate-limiting step in the absorption process. The important
role these properties play in influencing the bioavailability must, therefore, be recognized
and taken into consideration during optimization of a dosage form formulation.
For example, a smaller particle size is desirable, if the drug is hydrophobic,
to improve the drug dissolution due to increased specific surface; however, too small
a particle size may adversely affect the powder flow and content uniformity of a
dosage form. Other derived powder properties like true and bulk density and particle
size will play an important role in the mixing of powder blends, prior to granulation
and compression. Powder flow is another derived property of importance. Flow of
the powder and granules can present difficulties in the manufacturing of a tablet
Figure 2 Schematic representation of the process of the drug dissolution and its entry into
the general circulation.
Bioavailability and Granule Properties 539
© 2005 by Taylor & Francis Group, LLC
dosage form, which, in turn, can affect the content uniformity of a drug and the
bioavailability.
Many processes used in the tablet manufacturing greatly influence the dissolution
rates of the active ingredient. The method of manufacture, the size, the moisture
content, age and the flow property of the granules, the order of mixing of ingredients
during the granulation, as well as the compression force employed in the tableting
process, all contribute to the dissolution characteristics of the final product and,
therefore, may be the bioavailability of a drug from the finished product.
Several studies have demonstrated that the granulation process, in general,
enhances the dissolution rate of poorly soluble drugs (1). The use of diluents and fillers
such as starch (2), anhydrous (3) and spray dried lactose (4,5), microcrystalline
cellulose (6), and compression force, particle size and lubricants (7) tends to enhance
the hydrophilicity of the active ingredients and improve their dissolution characteristics.
In this regard, the wet granulation procedure was considered a superior
method compared to other methods. However, newer tableting machines and excipients,
accompanied by careful formulation and proper mixing sequence, will permit
preparation of tablets with good dissolution characteristics and not dictated by
the method of preparation per se.
Marlow and Shangraw (8) reported that sodium salicylate tablets prepared by
direct compression with spray dried lactose uniformly exhibited more rapid and
complete dissolution compared to those prepared by wet granulation. Furthermore,
it was reported that the presence of disintegrate in the dry compression was essential
for good dissolution. Finholt et al. (9) reported, in a separate comparative study, utilizing
phenobarbital tablets that were manufactured by both wet and dry granulation
that both procedures yielded comparable dissolution rates provided a disintegrate
was incorporated and mixed with drug before dry granulation. However, the incorporation
of disintegrate following the dry granulation of a drug resulted in slower
dissolution rates.
In the manufacturing of tablets by the conventional wet granulation method,
there are many independent factors that affect the property of granules and, therefore,
the dissolution rate. Recent advances in granulation technology and the
employment of high-shear mixers and fluid bed granulating equipment have helped
to identify several critical in-process variables. The systematic control of variables
such as the type and time of mixing of the granules, time and temperature of drying,
blending time with lubricant, age of the granules, moisture content of the granules at
the time of compression, and the tablet crushing strength are of importance to ensure
the consistency in the dissolution and, hence, bioavailability.
In early studies on the physics of tablet compression, Higuchi et al. (10) recognized
the influence of compressional forces employed in the tableting process on the
apparent density, porosity, hardness, disintegration time, and average particle size of
the compressed tablets. Hardness is a measure of resistance of a dosage form to the
mechanical deforming. It is a function of high compression forces used in the manufacturing,
and it may change with the aging of granules. Higuchi et al. (10) reported
a linear relationship between hardness and the logarithm of the compressional force,
and the specific surface of the compressed tablets was found to undergo marked
changes during the compressional process. The high compression may increase the
specific surface and, hence, may enhance the dissolution. On the other hand, the high
compression may also inhibit the wettability of a tablet due to the formation of
a firmer and more effective sealing layer of the lubricant due to high pressure and
temperature that accompany a strong compressional force. Levy et al. (2) reported
540 Jambhekar
© 2005 by Taylor & Francis Group, LLC
that salicylic acid tablets, when prepared by double compression, showed an increase
in the dissolution with an increase in the pre-compression pressure due to fracturing
of drug particles at higher pressure. The higher compression may also produce
slower dissolution, at least in the initial period, due to an increased difficulty of fluid
penetration into the compressed tablets. Luzzi et al. (11) and Jalsenjak (12) observed
the dissolution rate of sodium phenobarbital to be inversely proportional to the
hardness from tablet and microcapsule, respectively.
Another important granule property that influences the dissolution of drug is
the moisture content of the granule at the time of compression. Chowhan et al.
(13–17) studied the effect of moisture content and crushing strength on ticlopidine
hydrochloride tablet friability and dissolution. It was observed (16) that at the moisture
content of 1–2%, the drug dissolution was inversely related to the tablet crushing
strength. However, at the moisture content level of 3–4%, there was no clear relationship
between the dissolution and the crushing strength.
In later studies by Chowhan et al. (14,17), it was reported that the granules prepared
by high-speed shear mixer were less porous than those prepared by planetary
mixer, and the porosity of the tablet may improve the dissolution of drug by facilitating
solvent penetration provided the entrapment of air in the pores is minimized
or avoided.
In yet another important study, Levy et al. (2) studied the effect of the granule
size on the dissolution rate of salicylic acid tablets and found that the dissolution rate
increased with a decrease in the granule size; the increase in dissolution rate, however,
was not proportional to the increase in the apparent surface area of the granules.
Furthermore, it was also reported that the dissolution rate decreased
significantly with the increase in the age of the granules.
The chemical components of the formulation have also been shown to prolong
disintegration time, which subsequently affect the drug dissolution and bioavailability.
Inert fillers have been found to potentiate the chemical degradation of active
ingredient causing alteration in the disintegration and dissolution time of compressed
tablets to change with storage. Alam and Parrott (18) have shown that hydrochlothiazide
tablets, granulated with acacia and stored at temperatures ranging from
room temperature to 80C increased in hardness with time. This was reflected in
increased disintegration and dissolution time. On the other hand, tablets granulated
with starch and polyvinylpyrrolidone did not show any change in disintegration and
dissolution time.
3. CONCLUSION
Drug availability following the oral dosing may be thought of as the result of the
following steps:
1. Getting the drug into solution.
2. Moving the drug molecules through the membrane of the gastrointestinal
tract.
3. Moving the drug away from the site of administration into the general
circulation.
It is clear from the discussion that the bioavailability of drugs, particularly
poorly soluble drugs, mainly depends on the ability of the drug to dissolve at the site
of administration. The dissolution, in turn, especially from solid dosage forms such
Bioavailability and Granule Properties 541
© 2005 by Taylor & Francis Group, LLC
as tablet and capsule, depends on the powder properties, granule properties, and the
processing variables used in the manufacture of the dosage forms. The granule properties
and other variables, which determine and influence the granule properties, will
serve as major topics of discussion in subsequent chapters. Knowledge of these factors
and their role in influencing the bioavailability of a drug will allow the formulators
to develop an optimum drug dosage form by selecting the process and
preparation variables involved in a rational manner.
REFERENCES
1. Solvang S, Finholt P. Effect of tablet processing and formulation factors on dissolution
rates of active ingredient in human gastric juice. J Pharm Sci 1970; 59:49–52.
2. Levy G, Antkowiak J, Procknal J, White D. Effect of certain tablet formulation factors
on dissolution rate of the active ingredient II: granule size, starch concentration, and
compression pressure. J Pharm Sci 1963; 52:1047–1051.
3. Batuyios N. Anhydrous lactose in direct tablet compression. J Pharm Sci 1966; 55:
727–730.
4. Gunsel W, Lachman L. Comparative evaluation of tablet formulations prepared by
conventionally processed and spray dried lactose. J Pharm Sci 1963; 52:178–182.
5. Duvall R, Koshy K, Dashiell R. Comparative evaluation of dextrose and spray dried
lactose in direct compression systems. J Pharm Sci 1965; 54:1196–1200.
6. Reier G, Shangraw R. Microcrystalline cellulose in tableting. J Pharm Sci 1966; 55:
510–514.
7. Iranloye T, Parrott E. Compression force, particle size, and lubricants on dissolution
rates. J Pharm Sci 1978; 67:535–539.
8. Marlow E, Shangraw R. Dissolution of sodium salicylate from tablet matrices prepared
by wet granulation and direct compression. J Pharm Sci 1967; 56:498–504.
9. Finholt P, Pedersen P, Solvang R, Wold K. Medd Norsk Farm Selsk 1966; 28:238.
10. Higuchi T, Rao A, Busse E, Swintosky J. The physics of tablet compression II: the influence
of degree of compression on properties of tablets. Am Pharm Assoc Sci Ed 1953;
42:194–200.
11. Luzzi L, Zoglio M, Maulding H. Preparation and evaluation of the prolonged release
properties of nylon microcapsules. J Pharm Sci 1970; 59:338–341.
12. Jalsenjak I, Nicolaidou C, Nixon J. Dissolution from tablets prepared using ethylcellulose
microspheres. J Pharm Pharmacol 1977; 29:169–172.
13. Chowhan ZT, Palagyi L. Hardness increase induced by partial moisture loss in compressed
tablets and its effect on in vitro dissolution. J Pharm Sci 1978; 67:1385–1389.
14. Chowhan ZT. Moisture, hardness, disintegration and dissolution interrelationships in
compressed tablets prepared by the wet granulation process. Drug Dev Ind Pharm
1979; 5(1):41–62.
15. Chowhan ZT. Role of binders in moisture-induced hardness increase in compressed
tablets and its effect on in vitro disintegration and dissolution. J Pharm Sci 1980;
69:1–4.
16. Chowhan Z, Yang I, Amaro A, Chi L. Effect of moisture and crushing strength on tablet
friability and in vitro dissolution. J Pharm Sci 1982; 71:1371–1375.
17. Chowhan Z, Chatterjee B. A method for establishing in process variable controls for
optimizing tablet friability and in vitro dissolution. Int J Pharm Technol Prod Manuf
1984; 5(2):6–12.
18. Alam AS, Parrott EL. Effect of aging on some physical properties of hydrochlorothiazide
tablets. J Pharm Sci 1971; 60:263–266.
542 Jambhekar
© 2005 by Taylor & Francis Group, LLC
RECOMMENDED READING
Abdou HM. Dissolution, Bioavailability and Bioequivalence. Easton, PA: Mack Publishing
Company, 1989.
Blanchard J, Sawchuk R, Brodie B, eds. Principles and Perspectives in Drug Bioavailability.
Basel, Switzerland: S. Krager AG, 1979.
Gibaldi M. Biopharmaceutics and Clinical Pharmacokinetics. 4th. Philadelphia, PA: Lea and
Febiger, 1991.
Jambhekar S. Micromeritics and rheology. In: Ghosh and Jasti, eds. Theory and Practice of
Contemporary Pharmaceutics. Boca Raton, FL: CRC Press. In press.
Leeson LJ, Carstensen J, eds. Dissolution Technology. Washington, DC: Academy of Pharmaceutical
Sciences, 1974.
Stavchansky SA, McGinity JW. Bioavailability. In: Lieberman HA, Lachman L, Schwartz JB,
eds. Tablet Technology Pharmaceutical Dosage Forms: Tablets. Vol. 2. 2nd ed. New
York: Marcel Dekker, Inc., 1990.
Bioavailability and Granule Properties 543
© 2005 by Taylor & Francis Group, LLC
20
Process Analytical Technology
D. Christopher Watts and Ajaz S. Hussain
Office of Pharmaceutical Science, Center for Drug Evaluation and Research,
Food and Drug Administration, Bethesda, Maryland, U.S.A.
1. INTRODUCTION
Pharmaceutical granulation is a critical unit operation that is frequently utilized to
modulate attributes of powder mixtures to aid in further processing (e.g., compaction
or encapsulation), as today few pharmaceutical products are granules. The granulation
processes must be designed to impart a high degree of control on many
important physical attributes, such as granule size distribution, shape, content uniformity,
moisture content and distribution, porosity, density, tensile strength, and
surface morphology. These physical attributes are often critical to process-ability
of granulations and for final product quality and performance (e.g., content uniformity,
dissolution, stability, and bioavailability). An optimally designed granulation
unit operation can be an excellent tool for minimizing variability and thereby reducing
the risk of poor quality. Traditionally, granulation has been utilized to manage
lot–lot and supplier–supplier variability in the physical attributes of pharmacopieal
materials—excipients and drug substances.
Over the last three decades significant scientific and technological advances
have occurred in pharmaceutical granulation processes. During this period, the
practice of pharmaceutical granulation, especially wet granulation, has made significant
progress in moving from an art to a more science and engineering foundation.
If we use ‘‘granulation end point’’ in wet granulation as a gauge to
measure this progress, the practice of granulation end point moved from an operator
dependent evaluation (i.e., observing the behavior of a hand compacted mass)
to a time dependent end point and, in some cases, a more predictive end-point
criteria. The next few steps in this evolutionary process can be envisioned to be
robust metrics of granule functionality, the ability to predict optimal granulation
processing conditions and end points from an understanding of the attributes of
raw materials and feed-forward control. Furthermore, a mechanistic understanding
the granulation process and the ability to predict the impact of granule properties
on product performance seems to be a worthy and achievable goal in the near
future.
The objectives of this chapter are to discuss the FDA’s Process Analytical
Technology (PAT) framework using selected literature examples on granulation
545
© 2005 by Taylor & Francis Group, LLC
operations. This discussion is primarily to illustrate certain aspects of the PAT
definition and tools. As outlined in the draft guidance, ‘‘Process Analytical Technology
(PAT) is defined as a system for designing, analyzing, and controlling manufacturing
through timely measurements (i.e., during processing) of critical quality and
performance attributes of raw and in-process materials and processes with the goal
of ensuring final product quality’’ (1). It should be noted that the primary motivation
of PAT is to understand manufacturing processes through identification of all relevant
sources of variability and their impact on product quality, in order to design
and develop monitoring and control strategies to minimize variability in the final
product. Therefore, the term analytical in PAT is viewed broadly to include chemical,
physical, microbiological, mathematical, and risk analysis conducted in an
integrated manner. Simply, analytical is interpreted as ‘‘analytical thinking,’’ not
only chemical analysis.
The focus on process understanding comes from the basic tenet of building, not
testing, quality into products. This is a fundamental tenet championed by Deming,
Juran, and others (2,3). It is important to emphasize that considering any unit operation
(granulation or otherwise) in isolation from the entire manufacturing process
would be incomplete in meeting the desired product quality attributes, as each
procedure in the manufacturing process should be directed at meeting the needs
of subsequent manufacturing operations and, ultimately, final product quality.
For example, how do attributes of the raw materials and the blend (assuming this
unit operation precedes granulation), and the inherent variability in these attributes
impact an optimal granulation process and end point? This is a critical question to
be addressed as one designs a granulation process and its control strategy. Equally
important is the question: what are the critical attributes of the granules for their
subsequent processing and final product quality?
2. BACKGROUND
At the July 2001 meeting of the FDA’s Advisory Committee for Pharmaceutical
Science (ACPS), the Office of Pharmaceutical Science proposal to facilitate application
of modern pharmaceutical engineering principles and new technologies in
pharmaceutical development and manufacturing received an enthusiastic endorsement
by the ACPS (4). This proposal was further elaborated and discussed at
the November 2001 and April 2003 meetings of the Science Board to the FDA
(5,6), and once again, received strong support. These meetings were the beginning
of the FDA’s PAT Initiative and the broader CGMP’s for the 21st Century Initiative:
A Risk-Based Approach (7,8). The PAT Initiative was further developed
through public discussions at the PAT—Subcommittee of ACPS and numerous
scientific conferences and workshops such as the American Association of Pharmaceutical
Scientists (AAPS) Arden House, International Forum on Process Analytical
Chemistry (IFPAC), International Society of Pharmaceutical Engineers (ISPE),
and Parenteral Drug Association (PDA). These discussions led to the development
of the final FDA guidance ‘‘PAT—A Framework for Innovative Pharmaceutical
Manufacturing and Quality Assurance’’ (1). The following section from the final
guidance provides an excellent summary of the issues FDA is trying to address
with its initiatives.
Conventional pharmaceutical manufacturing is generally accomplished using
batch processing with laboratory testing conducted on collected samples to ensure
546 Watts and Hussain
© 2005 by Taylor & Francis Group, LLC
quality. This conventional approach has been successful in providing quality
pharmaceuticals to the public. However, today significant opportunities exist for
improving the efficiency of pharmaceutical manufacturing and quality assurance
through the innovative application of novel product and process development, process
controls, and modern process analytical chemistry tools. Unfortunately, the
pharmaceutical industry generally has been hesitant to introduce new technologies
and innovative systems into the manufacturing sector for a number of reasons.
For example, one reason often cited is regulatory uncertainty, which may result from
the perception that our existing regulatory system is rigid and unfavorable to the
introduction of new technologies. In addition, a number of scientific and technical
issues have been raised as possible reasons for this hesitancy. Nonetheless, industry’s
hesitancy to broadly implement new pharmaceutical manufacturing technologies is
undesirable from a public health perspective. The health of our citizens and animals
in their care depends on the availability of safe, effective, and affordable medicines.
Efficient pharmaceutical manufacturing is a critical part of an effective U.S. health
care system.
In the future, pharmaceuticals will have an increasingly prominent role in
health care. Pharmaceutical manufacturing will need to employ innovation, cutting
edge scientific and engineering knowledge, along with the best principles of
quality management to respond to the challenges of new discoveries (e.g., novel
drugs and nanotechnology) and ways of doing business (e.g., individualized
therapy, genetically tailored treatment). Regulatory policies must also rise to the
challenge.
In August 2002, recognizing the need to free industry from its hesitant perspective,
the FDA launched a new initiative entitled Pharmaceutical cGMPs for the 21st
Century: A Risk-Based Approach. This initiative has several important goals, which
ultimately will help improve the American public’s access to quality health care
services. The goals are intended to ensure that:

 The most up-to-date concepts of risk management and quality systems
approaches are incorporated into the manufacture of pharmaceuticals
while maintaining product quality.

 Manufacturers are encouraged to use the latest scientific advances in pharmaceutical
manufacturing and technology.

 The Agency’s submission review and inspection programs operate in a
coordinated and synergistic manner.

 Regulations and manufacturing standards are applied consistently by the
Agency and the manufacturer, respectively.

 Management of the Agency’s Risk-Based Approach encourages innovation
in the pharmaceutical manufacturing sector.

 Agency resources are used effectively and efficiently to address the most significant
health risks.
Pharmaceutical manufacturing continues to evolve with increased emphasis on
science and engineering principles. Effective use of the most current pharmaceutical
science and engineering principles and knowledge—throughout the life cycle of a
product—can improve the efficiencies of both the manufacturing and regulatory processes.
This FDA initiative is designed to do just that by using an integrated systems
approach to regulating pharmaceutical product quality. The approach is based on
science and engineering principles for assessing and mitigating risks related to poor
Process Analytical Technology 547
© 2005 by Taylor & Francis Group, LLC
product and process quality. In this regard, the desired future state of pharmaceutical
manufacturing may be characterized as follows:
1. Product quality and performance are ensured through the design of effective
and efficient manufacturing processes.
2. Product and process specifications are based on a mechanistic understanding
of how formulation and process factors affect product performance.
3. Continuous real-time quality assurance.
4. Relevant regulatory policies and procedures are tailored to accommodate
the most current level of scientific knowledge.
5. Risk-based regulatory approaches recognize:

 The level of scientific understanding of how formulation and manufacturing
process factors affect product quality and performance.

 The capability of process control strategies to prevent or mitigate the
risk of producing a poor-quality product.
3. PAT AND PROCESS UNDERSTANDING
The PAT framework is intended to support innovation and efficiency in pharmaceutical
development, manufacturing, and quality assurance. It utilizes process understanding
as its foundation to facilitate innovation by industry and risk-based
regulatory decisions. For processes that are well understood, significant reduction
in regulatory requirements for both CMC review and cGMP inspection may be
realized. In the context of granulation one can describe three levels of process
understanding: (1) all critical sources of variability in quality and performance are
identified and explained; (2) material and process variability is managed by the
process design; and (3) quality attributes of the granulation and the final product
can be accurately and reliably predicted. Ultimately, the ability to accurately and
reliably predict product quality attributes, for example via mathematical models,
may be considered to reflect a high degree of process understanding.
Obtaining an increased understanding of the granulation process requires a
multidisciplinary approach, combining expertise and scientific principles from
several fields, including material science, pharmaceutics, engineering, and mathematics.
The process knowledge should grow when organized knowledge management
programs are instituted to collect appropriate information over the life cycle
of a specific product and from information gathered from development and manufacturing
experience of other products. The following expression is an initial attempt
that brings many of the key elements (i.e., design, predictability, and capability)
that provide a basis for assessing process understanding (PU) in a summary form.
The triple integration function is utilized to symbolize the need for integration
across: (1) scientific disciplines, (2) across organizations, and (3) over time (life
cycle).
PU ? ZZZ ?Design??Predictability??Capability?
Successful implementation of the PAT framework will require cooperation and
collaboration across scientific disciplines and divisions within an organization such
as R&D, manufacturing, quality, and information/knowledge management. Such
548 Watts and Hussain
© 2005 by Taylor & Francis Group, LLC
cooperation and collaboration within the FDA organization is also necessary for
science and risk-based regulatory decisions; hence, the FDA established a team
approach for review and inspection for PAT (the FDA PAT Team).
4. PAT TOOLS AND THEIR APPLICATION
There are a variety of tools available to facilitate increased understanding and
provide an integrated, systems approach to measuring, analyzing, optimizing,
controlling, and modeling of the granulation process. Many of these tools are well
established and utilized frequently in many industrial sectors, and are beginning to
find wider utility in the pharmaceutical industry.
4.1. Multivariate Tools for Design, Data Acquisition, and Analysis
Pharmaceutical products are complex multifactorial systems; consequently, the
‘‘one-factor-at-a-time’’ approach to experimentation is often insufficient for identifying
and addressing interactions between product and process variables. When
attempting to identify the critical product and process variables with significant
effects on final product quality, a statistically sound experimental design is essential.
Having a sound experimental design not only increases the efficiency of experimentation,
but also identifies interactions of the variables, providing opportunities to
efficiently and accurately identify the ‘‘best’’ product and process parameters with
various optimization techniques.
For example, pharmaceutical companies that utilize structured design of
experiments are not only likely to develop robust product and processes, but are also
able to clearly articulate why certain variables are critical and justify acceptable
ranges for these critical variables to ensure quality. This approach is often essential
to minimize manufacturing problems during routine production, especially for
processes that have a fixed time as end point. Although the frequently used ‘‘fixed
granulation time’’ provides a means to demonstrate ‘‘process validation’’ in the
conventional ‘‘three-batch’’ mode, it presupposes that the fixed time is optimal for
addressing the raw material variability encountered in routine production. This
source of variability is often underestimated during product development and during
demonstration of ‘‘process validation’’ under very controlled conditions (e.g., use of
the same lot of raw materials, etc.). Manufacturing difficulties and out of specification
results, with no clear root cause, for ‘‘validated’’ processes increases production
cycle time, results in rejection or recalls, and erodes regulatory and public
confidence.
Sahni et al. (9) have described a general framework for utilizing an appropriate
experimental design to identify critical process parameters and using multivariate
techniques for data analysis. Specifically, the authors investigated two material
streams and five different process variables. Using a full factorial design to probe
the effects of these variables and their interaction(s) would require 27, or 128,
experiments. In order to take a more efficient, yet structured approach, a fractional
factorial design (using 32 experiments) was employed to identify critical material and
process attributes.
Because the system described above is multifactorial, a simple univariate
approach to data analysis would likely not elicit the critical factors. In that regard,
Process Analytical Technology 549
© 2005 by Taylor & Francis Group, LLC
principal component analysis (PCA) and partial least squares (PLS) regression,
typical multivariate analysis techniques, were used for analyzing data from the 32
experiments. Several sources are available for in-depth understanding of these techniques
(10,11).
Briefly, PCA models the data in terms of the significant factors, or ‘‘principal
components,’’ which describe the systematic variability of the data. PCA also
describes the data in terms of ‘‘residuals’’ that represent the noise in the system.
PLS may be described as a method for constructing predictive models from data sets
with many collinear factors. Both have received considerable attention in the analysis
of multivariate data.
A group of investigators from the Netherlands employed similar multivariate
techniques for modeling the manufacture of tablets, using wet granulation for
optimization and in-process control (12). Again, material and process attributes were
investigated using a systematic approach to identify the critical process parameters/
material attributes having a significant effect on tablet quality. Once identified, the
authors developed a strategy for in-process control of the critical variables, such that
tablet quality could be ensured and predicted via the developed models, throughout
the production process. Consequently, a test of the final product/tablet would
confirm the prediction of product quality, thereby validating both the process
control strategy and the model.
In order to adequately evaluate these complex systems, multivariate data
analysis techniques must be used. The examples previously discussed highlight only
a few approaches to multivariate data acquisition and analysis. There are many
approaches for using these tools to obtain an increased understanding of granulation,
and to elicit critical process/material attributes, as well as their relationship
with final product quality attributes, for controlling the manufacturing process.
4.2. Process Analyzers
In terms of innovation, some of the most significant advances have occurred in the
area of process analyzers, and there are numerous examples of the application of
these tools to obtain a greater understanding of the granulation process. Certainly,
once critical process and material attributes have been identified, an appropriate tool
for the analysis of these attributes is crucial for timely analysis, as well as ultimately
controlling the granulation process.
Many studies have detailed the use of various devices for monitoring granulation
process conditions, including drive shaft torque (13) and power consumption of
the motor (14) for high-shear granulation processes. Although other investigations
have demonstrated the relationship between these measurements and granule growth
(15), they are indirect measures of in-process material quality. As such, any change in
the process or material (including raw material) would affect these measurements.
Consequently, a measurement of the process parameter would not accurately
represent the true value of the quality attribute of interest.
Spectroscopic sensors, most notably near infrared (NIR), have received
significant attention in the analysis and control of the granulation process. Rantanen
et al. have published several examples on the use of NIR for in-line measurement of
moisture and particle size during fluid-bed granulation (16,17), as well as automation
of the granulation process (18). Using sensors in this manner provides insight into
the process, allowing the direct measurement and nondestructive analysis of
‘‘samples’’ in the product stream.
550 Watts and Hussain
© 2005 by Taylor & Francis Group, LLC
These sensors have also been used to elicit physical information about the
granules, such as particle size, to understand (and subsequently control) the impact
of other material attributes and process parameters on product quality. For example,
Watano et al. developed a system for monitoring granule growth via an imaging system,
which employed, among other items, a CCD camera (19) and an image probe
inserted into a high-shear granulator. The system was used not only for monitoring
granule growth, but also for control of the high-shear granulation process.
4.3. Process Control Tools
Currently, the end point of most granulation processes are based on measurements of
time (i.e., add binder for t1 min, mass for t2 min, etc.) or process parameters (such as
power or torque as described earlier). However, a direct measure of the critical/
desired product quality attribute (particle size, density, moisture, etc.) would be ideal
for its control. Subsequent determination of granulation end point would therefore be
based on a direct measure of the desired quality attribute, rather than it being inferred
from process parameters, and later confirmed through laboratory measurements.
Materials entering the granulation process are widely known to vary, especially
in their physical attributes. The granulation process should therefore manage this
variability, and consequently mitigate its risk to product quality by an integrated
approach to process control. Additionally, it is important to note that the functionality
of incoming material must be measured and understood. This information
should ideally be ‘‘fed forward’’ to adjust/control the following unit operations to
ensure optimal processing and end-point establishment, such that starting variability
is minimized and not transferred to the final product. Limits are often necessary on
incoming materials to ensure their variability does not exceed the design space used
for developing a given control strategy, lending data from the sensors open to
erroneous interpretation/analysis and thus introducing a failure risk to the control
strategy. However, such excursions from the design space, if and when they occur,
can provide an opportunity for its expansion, as well as expand the general
knowledge base of the process.
Many examples of controlling the granulation process can be found in the
literature. Following development of the on-line imaging system for monitoring granule
growth, Watano et al. incorporated a fuzzy control system for granule growth (20).
Granule growth was measured via the imaging system and fed into the fuzzy control
system. Based on measurements of granule size, the fuzzy control system adjusts
various process parameters in order to obtain the desired particle size distribution.
Several other investigations have also focused on the use of NIR for moisture
analysis and control during granulation, including Frake et al. (21). In their study,
the investigators employed the use of an NIR sensor for real-time determination
of the moisture content and particle size distribution of a fluid-bed granulation. This
control strategy permits any necessary modifications to process conditions, as well as
the identification of a process end point, once the desired moisture content and
particle size are obtained.
Although a few examples of traditional ‘‘feed-back’’ control for granulation
processes were described, the concept of ‘‘feed-forward’’ control has been investigated
as well. Paul Mort et al. discuss the aspects of feed-forward control associated
with a continuous granulation process (22). Based on the amount of recycled fines
reintroduced into the system, and attributes of the incoming materials, the binder
solution is modified to obtain the desired granule density and particle size.
Process Analytical Technology 551
© 2005 by Taylor & Francis Group, LLC
Appropriate strategies for process control provide an assurance, with every
product manufactured, that the desired product quality will be obtained. A test of
the final product thereby ‘‘validates’’ that the process is indeed under control.
4.4. Continuous Improvement and Knowledge Management Tools
Continuous learning of the production process also allows for optimization of process
parameters that may not have been fully explored during development. Additionally,
any changes to a given process, granulation or otherwise, could be considered lowrisk
because of the level of process understanding achieved and demonstrated,
thereby reducing the regulatory burden to change control/management. Ultimately,
an inverse relationship exists between the level of process understanding and the risk
of producing a poor-quality product. Thus, a focus on process understanding can
facilitate risk-based regulatory decisions and innovation (1).
Continuous learning through data collection and analysis over the life cycle of
a product is crucial. However, information gleaned from the production process of
one product should not be considered in isolation. Management and generalization
of knowledge from one product to another can allow for the development of expert
systems and a large knowledge base that can be used to facilitate future process
development, scale-up, optimization, etc.
For example, expert systems such as neural networks can be employed to
model various stages of the production process, including granulation. Investigators
in Finland have noted the advantages of such systems in modeling the fluid-bed
granulation process (23). Results from this investigation demonstrated that the
networks were able to more accurately predict responses to the granulation process
than multilinear stepwise regression analysis.
5. CONCLUSION
Implementing the PAT framework and having a greater understanding of the
manufacturing process has obvious advantages to the pharmaceutical industry, the
regulators, and the public health. The PAT Initiative provides a regulatory environment
that encourages and facilitates pharmaceutical companies to innovate and
employ the tools necessary to achieve an in-depth understanding of the manufacturing
process. There are many examples in the literature that suggest that active control of
the granulation is not only possible, but may soon become the rule, rather than the
exception. This paper was written prior to the FDAs final report on CGMPs for the
21st century. The readers are requested to review this report
Paper entitled ‘‘Innovation and Continuous Improvement in Pharmaceutical
REFERENCES
1. Guidance for Industry: PAT—A Framework for Innovative Pharmaceutical Development,
Manufacturing, and Quality Assurance. September 2004.
2. Aguayo R. Dr. Deming: the American who taught the Japanese about quality. Fireside
1991; September 15.
552 Watts and Hussain
(http://www.fda.
gov/cder/gmp/gmp2004/GMP_finalreport2004.htm), in particular the FDA’s White
Manufacturing’’ available at http://www.fda.gov/cder/gmp/gmp2004/manufSciWP.pdf.
http//www.fda.gov/
cder/guidance/6419fnl.htm.
© 2005 by Taylor & Francis Group, LLC
3. Juran JM. Juran on quality by design: the new steps for planning quality into goods and
services. Free Press 1992; May 4, revised edition.
4. Advisory Committee for Pharmaceutical Science. Transcript of Meeting, July 2001.
5. Science Board to the Food and Drug Administration. Transcript of Meeting, November
6. Science Board to the Food and Drug Administration. Transcript of Meeting, April 2003.
7. Food and Drug Administration. Unveils New Initiative to Enhance Pharmaceutical
Good Manufacturing Practices, August 21, 2002.
8. Food and Drug Administration. Announces New Progress toward ‘‘21st Century’’ Regulation
of Pharmaceutical Manufacturing, September 3, 2003.
9. Sahni NS, Isaksson T, Tormod Naes. The use of experimental design methodology and
multivariate analysis to determine critical control points in a process. Chemometrics
Intelligent Lab Syst 2001; 56:105–121.
10. Brereton RG. Multivariate Signal Processing in Chemometrics. Application of
Mathematics and Statistics to Laboratory Systems. Chichester, U.K.: Ellis Horwood,
1990:209–237.
11. Wold S, Albano C, Dunn WJ, Edlund U, Esbensen K, Geladi P, Hellberg S, Johansson E,
Lindsberg W, Sjostrom M. Multivariate Data Analysis in Chemistry. Chemometrics:
Mathematics and Statistics in Chemistry. ReidelDDordrecht, The Netherlands 1984:17–95.
12. Westerhuis JA, Coenegracht PMJ, Lerk CF. Multivariate modeling of the tablet
manufacturing process with wet granulation for tablet optimization and in-process
control. Int J Pharm 1997; 156:109–117.
13. Lindberg NO, Leander LL, Reenstierna B. Instrumentation of a Kenwood Major domestic
type mixer for studies of granulation. Drug Dev Ind Pharm 1982; 8:775–782.
14. Leuenberger H. Scale-up of granulation processes with reference to power consumption
monitoring. Pharm Acta Helv 1983; 29:274–280.
15. Holm P, Schaefer T, Kristensen HG. Granulation in high-speed mixers. Part V: Power consumption
and temperature changes during granulation. Powder Technol 1985; 43:213–223.
16. Rantanen J, Rasanen E, Tenhunen J, Kansakoski M, Mannermaa J, Yliruusi J. In-line
moisture measurement during granulation with a four-wavelength near infrared sensor:
an evaluation of particle size and binder effects. Eur J Pharm Biopharm 2000; 50:271–276.
17. Rantanen J, Jorgensen A, Rasanen E, Luukkonen P, Airaksinen S, Raiman J, Hanninen
K, Antikainen O, Yliruusi J. Process analysis of fluidized bed granulation. AAPS
PharmSciTech 2001; 2(4):21.
18. Rantanen J, Kansakoski M, Suhonen J, Tenhunen J, Lehtonen S, Rajalahti T,
Mannermaa JP, Yliruusi J. Next generation fluidized bed granulator automation. AAPS
PharmSciTech 2000; 1(2):10.
19. Watano S, Numa T, Koizumi I, Osaka Y. Feedback control in high shear granulation of
pharmaceutical powders. Eur J Pharm Biopharm 2001; 52:337–345.
20. Watano S, Numa T, Miyanami K, Osaka Y. A fuzzy control system of a high shear
granulation using image processing. Powder Technol 2001; 115:124–130.
21. Frake P, Greenhalgh D, Grierson SM, Hempenstall JM, Rudd DR. Process control and
end-point determination of a fluid bed granulation by application of near infrared
spectroscopy. Int J Pharm 1997; 151:75–80.
22. Mort PR, Capeci SW, Holder JW. Control of agglomerate attributes in a continuous
binder-agglomeration process. Powder Technol 2001; 117:173–176.
23. Murtoniemi E, Yliruusi J, Kinnunen P, Merkke P, Leiviska K. The advantages by the use
of neural networks in modeling the fluidized bed granulation process. Int J Pharm 1994;
108(2):155–164.
Process Analytical Technology 553
http://www.fda.gov/ohrms/dockets/ac/01/transcripts/3763t1.htm.
2001. http://www.fda.gov/ohrms/dockets/ac/01/transcripts/3799t1.htm.
http://www.fda.gov/ohrms/dockets/ac/03/transcripts/3944t1.DOC.
http://www.fda.gov/bbs/topics/
NEWS/2002/NEW00829.html.
http://www.fda.gov/
cder/gmp/index.htm.
© 2005 by Taylor & Francis Group, LLC
21
Granulation Process Modeling
I. T. Cameron and F. Y. Wang
Particle and Systems Design Centre, School of Engineering,
The University of Queensland, Queensland, Australia
1. MODELING OF GRANULATION SYSTEMS
In this section, we introduce the background to granulation modeling by posing the
question, ‘‘Why model?’’ and a further one, ‘‘How are models used in granulation
systems technology?’’ The following sections seek to answer these questions and
demonstrate the benefits that can be derived from appropriate granulation process
modeling.
1.1. Motivation for Modeling
There are many motivations for modeling granulation systems that are common to
all process-related modeling. This is an area, that has grown enormously over the last
50 years. Michaels (32) has pointed out that despite the change of particle technology
from an underfunded and widely scattered research enterprise to a thriving globally
recognized engineering discipline over the past 25 years, design and analysis of industrial
particulate processes remain rooted in empiricism. Without exception, granulation
processes, like most solid-handling operations, continue to be one of the least
understood and hence inefficient in the process industries. Thus, granulation
remained more of ‘‘an art than a science’’ until a decade ago, as stated by Litster
(24). Granulation operations were performed by employing popular practice rather
than through systematic scientifically-based strategies. The ineffectiveness of this
approach led researchers on a quest to represent the dynamic or steady-state characteristics
of systems through a deeper understanding of the relevant phenomena of the
physico-chemical phenomena being studied. Likewise in granulation systems, there
has been a growing interest in the building of models and their deployment to
address a range of applications.
1.1.1. Benefits
The benefits from the use of modeling include:

 An increased understanding of the governing mechanisms through endeavoring
to represent them in the model description.
555
© 2005 by Taylor & Francis Group, LLC

 An increased understanding of the relative importance of mechanistic
contributions to the outputs of the process.

 Capturing of insight and knowledge in a mathematically usable form.

 Documentation of research findings in accessible form for various applications.

 Application of models for improved control performance and process diagnosis.

 Potential reuse of model components for a variety of applications from
design through to process diagnosis.

 As a vehicle for new, novel designs of processing equipment.

 As a means to direct further experimentation and process data generation.
1.1.2. Costs
There are several important and not insubstantial costs involved in process modeling
including:

 Time to plan, develop, test, and deploy models.

 Personnel with the requisite discipline background to generate effective
models through insight and modeling skills.

 The effort in laboratory-scale or plant scale-trials to elucidate process
behavior and the cost of doing so.

 The cost of poor modeling practice in terms of inadequate documentation
through the modeling phases and loss of corporate memory.
1.2. Process Modeling Fundamentals
Process modeling is purpose driven in that a model is developed for a particular area.
These areas could include:

 Improved control performance through the use of process model-based
control algorithms.

 Optimal performance of granulation systems through model-based optimization
of production parameters such as shortest batch time or optimal
product size distribution.

 Improved production scheduling using models to generate improved batch
times estimates.

 Plant diagnosis for real-time plant operator guidance systems (OGS).

 Extraction of parameter estimates such as rate constants and granulation
kernel parameters.

 Improved design of equipment or development of new designs based on
better understanding and use of mechanistic phenomena.
The resultant model must be ‘‘fit-for-purpose,’’ and this is achieved by having
clearly stated goals for the modeling that are used to help assess the appropriateness
of the model form and the model fidelity required for the job. In particular, modeling
requires a methodology, especially one that is generic in nature.
1.2.1. A Systems Perspective
Models need to be built on a clear systems engineering understanding of the process.
define the inputs (u), and disturbances (d), to our system as well as the outputs (y)
556 Cameron and Wang
A typical system schematic is seen in Figure 1. For the system (S), we need to clearly
© 2005 by Taylor & Francis Group, LLC
and the states of interest (x). The system S converts inputs and disturbances to
outputs and is expressed as:
y ? S?u; d ?1:1?
A model M is a representation of the system that transforms inputs to predicted
outputs y(M) in the form:
y?M? ? M?u; d ?1:2?
How close y and y(M) are is a key question in model validation.
Approaching modeling from a systems perspective provides a clear framework
for developing models and identifying the key issues to be considered. Four principal
classes of variables play particular roles in the modeling. They are:
Inputs:
Inputs u are the variables that are manipulated to ‘‘drive’’ the system or maintain
its condition in the face of changes from disturbances. Typically, we consider such
aspects as binder addition rate or mixing intensity as inputs that we can manipulate.
Disturbances:
Disturbances d are variables over which we do not have clear control. They arise
from raw material properties that might change from batch to batch. They could
be environmental factors such as ambient temperature and humidity. They might be
fluctuations in input voltage to motors or steam pressure that produce temperature
disturbances in heated systems. Principal disturbances in all relevant categories need
to be identified.
Outputs:
Outputs y are the variables of interest for the designer, operator, or manager.
They might be quality variables that are related to granule properties such as size
distribution, granule moisture, or granule hardness. Other outputs of interest could
be related to product temperature, flowrates, and composition. The outputs are
necessarily measurable in some way, either directly on-line such as PSD or moisture,
or via laboratory analysis such as composition.
States:
States x of the system represent the internal variables that characterize the system
behavior at any point in time.
Finally the system S is a major consideration in modeling because of the
various ways the real system can be represented by the model M of that system.
Numerous forms of the model M are available. They can have a structure based
on capturing fundamental phenomena from physics and chemistry (‘‘white box’’
Figure 1 Schematic of a process system.
Granulation Process Modeling 557
© 2005 by Taylor & Francis Group, LLC
models) to internal structures based on purely empirical approaches known as
‘‘black box’’ models. For black box models, the form is simply a convenient equation
that captures the relationships amongst inputs, disturbances, and outputs.
1.2.2. Modeling Methodology and Workflow
Modeling should not be a haphazard activity. It is essential that a consistent and
defensible methodology be adopted. One such methodology is given by Hangos &
Cameron (2001) and is seen in Figure 2. Each of the seven key steps is a vital part
of any modeling activity, emphasizing that modern process modeling is not just
Figure 2 Modeling methodology and workflow.
558 Cameron and Wang
© 2005 by Taylor & Francis Group, LLC
about generating a set of equations. It is a much more holistic activity. It is also itera-
The key aspects can be summarized as:

 Goal-set definition: making clear the reason for the modeling and the goals
to be addressed in the modeling.

 Model conceptualization: clarifying the conserved quantities and the governing
mechanisms to be included. Clearly setting out the assumptions underlying
the model.

 Modeling data: generating or referencing physical property data or plant
data relevant to model building and model validation.

 Model building and analysis: putting the model together and then analyzing
the model for properties relevant to solution and dynamic properties.

 Model verification: ensuring that the coded model in the simulation environment
is correctly represented and bug-free.

 Model solution: solving the model numerically or analytically in some
limited circumstances. This can be a challenging task.

 Model calibration and validation: performing parameter estimation and then
validating the model against plant or laboratory data.
1.2.3. The Modeling Goal
The modeling goal plays a vital role in the development of the model. Here we
consider the most important general goals and describe briefly what is achieved.
Dynamic simulation problem
The model is developed to predict the system behavior in time. We wish to predict
the outputs y, given the inputs u, the disturbance pattern d, the model structure
M with the model parameters p. This is the most common goal.
Design problem
We are interested in calculating certain parameters or design from the parameter
set given all the other inputs, model form, disturbances, and a set of desired outputs.
Process control
Regulation and state-driving problems
We aim at designing or computing the inputs u for a prescribed response y of
the system.
System identification problem
We seek to determine the structure of the model M and its parameters p, using
input and output information.
State estimation problem
We attempt to estimate the internal states x of a particular model M.
Fault detection and diagnosis problem
We bring to light the faulty modes and/or system parameters which correspond
to the measured input and output data.
These three key areas show the various goals that are often given for modeling
of processing systems.
1.3. Approaches to Modeling
There are several approaches to modeling process systems. At one extreme, we have
mechanistic modeling that seeks to incorporate fundamental physics and chemistry
Granulation Process Modeling 559
tive in nature as seen in Figure 2.
We make reference to the general process system illustrated in Figure 1.
© 2005 by Taylor & Francis Group, LLC
into the model. This is the so-called ‘‘white box’’ approach. At the other end of
the spectrum, we have the fitting of an arbitrary function to the input–output data—
the empirical model. In between, we have the so-called ‘‘gray box’’ models that are
normally what is developed. Some relevant comments follow.
1.3.1. Empirical or Black Box Methods
These models are based on actual plant data, typically in the form of time series of
input and output data at fixed time intervals. The model is built by selecting a model
structure M and then fitting the model parameters to get the best fit of the model to
the data. Model forms such as auto-regressive moving average (ARMA) or autoregressive
moving average exogenous (ARMAX) are typically used.
In most cases, techniques are used to vary both the structure and the model
parameters to obtain the ‘‘best’’ or simplest model that gives the ‘‘best’’ fit. Various
information criteria like Akaike’s or Bayesian measures can be used that essentially
get the simplest model for the best fit. It is a form of parsimony in model building.
Many packages such as the MATLABTM Identification Toolbox help in such modeling.
This approach can be very useful if no significant insight is needed into the
model, but when a model is needed quickly to be used for a control application. The
structural form of the model and the parameter values normally have no physical
significance. The application range of such a model is limited to the range of data,
and hence this can be a significant limitation. Extrapolation is dangerous.
1.3.2. Mechanistic and Gray-Box Models
Mechanistic models incorporate the underlying understanding of physics and chemistry
into the models. Typically, we can identify three major aspects in mechanistic
modeling that cover conservation and constitutive aspects:

 Application of thermodynamic conservation principles for mass, energy,
and momentum and,

 Application of population balances that track particle size distributions as
various particulate phenomena take place.

 Development of appropriate constitutive relations that define intensive
properties, mass and heat transfer mechanisms, as well as particle growth
and breakage mechanisms.
The development of mechanistic models is far more complex and time-consuming
than it is for empirical models and is only justified when time permits. The model is to
be used over a wide operating range and the relevant insight in establishing the constitutive
relations is available.
Inevitably, even the best mechanistic models require some data fitting, leading
to the concept of ‘‘gray box’’ models. This is the normal practice in industrial modeling
of such systems. It means that adequate data must be available to carry out the
validation studies. This task is particularly difficult for the validation of dynamic
models.
2. KEY FACTORS IN GRANULATION MODELING
The modeling of granulation system from a mechanistic perspective inevitably means
representing the conservation principles and the constitutive relations that reflect the
560 Cameron and Wang
© 2005 by Taylor & Francis Group, LLC
key factors in granulation. We briefly consider these in turn but refer the reader to
the relevant chapters in this handbook for detailed descriptions of the phenomena.
2.1. Conservation Principles
The conservation of mass, energy, momentum, and particle number can be important
aspects of granulation modeling.
Mass conservation is crucial and is the fundamental concept for any granulation
system. Key factors here will be solids or slurry feed rates, any out flows, and the
addition of binders and additives to the granulation device. Accompanying the mass
balance over the device will be the energy balance, from which the intensive property
of temperature can be estimated.
Of particular importance in granulation systems is the factor of particle populations.
The particle size distribution within granulation devices is crucial, and there
are a number of mechanisms that simultaneously occur in the device depending on
the powder properties and operating regime. The challenge is in the description of
these mechanisms that occur. The following section outlines those mechanisms that
require consideration. Section 3 deals in detail with the development of population
balance representations and their variants.
2.2. Principal Constitutive Mechanisms
There are three principal mechanisms that need to be considered.
2.2.1. Nucleation
Nucleation refers to the formation of initial aggregates that are typically a result of
the interaction between the binder spray droplets and the powder in the device. This
mechanism provides the initial stage for further growth through a number of
mechanisms. A number of nucleation models have been proposed in the literature.
2.2.2. Growth
Granule growth occurs through two key mechanisms that can be distinguished for
discussion purposes. The topic is discussed more fully in ‘‘One-Dimensional Population
Balance Models’’.
2.2.2.1. Layering. Layering refers to the taking up of fine particles onto the
surface of larger granules. It is often induced by rolling action and is a means of
granule growth that creates hard, compact granules. A practical layering model is
proposed in ‘‘Optimization and Open-Loop Optimal Control Equations’’ for solving
optimal control of granulation processes.
2.2.2.2. Agglomeration. Agglomeration or coalescence refers to the successful
collision of two particles that result in a composite particle. The success of collisions
can be a function of particle size, binder and powder properties, and operational
factors such as bed height, powder velocity, and shear for mixer granulators. See
2.2.3. Breakage
Breakage in high shear and drum granulation is a significant issue, and is especially
more important in high-shear devices. There are various forms of breakage from
cleavage of particles to particle surface attrition where granule is chipped by collision
with other particles, wall, or impeller. Complexity of breakage models extends from
Granulation Process Modeling 561
‘‘Coalescence Kernels’’ for more details.
© 2005 by Taylor & Francis Group, LLC
binary breakage models to full particle distributions represented by breakage and
selection functions, or empirical models (38,45,54).
The following sections develop in detail some of the important aspects of granulation
process modeling, through the use of population balances and alternative
approaches.
3. REPRESENTING GRANULATION PROCESSES THROUGH
POPULATION BALANCES
The particulate nature of solids is characterized by a number of properties, such
as size, shape, liquid and gas content, porosity, composition, and age. These are
denoted as internal coordinates, whereas Euclidian coordinates, such as rectangular
coordinates (x, y, z), cylindrical coordinates (r, f, z), and spherical coordinates (r, y, f)
used to specify the locations of particles, are defined as external coordinates.
The most important property for the characterization of particles is particle
size. Randolph and Larson (36) pointed out that: ‘‘As no two particles will be exactly
the same size, the material must be characterized by the distribution of sizes or
particle-size distribution (PSD).’’ If only size is of interest, a single-variable distribution
function is sufficient to characterize the particulate system. If additional properties
are also important, multivariable distribution functions must be developed.
These distribution functions can be predicted through numerical simulations using
population balance equations (PBE).
Ramkrishna (35) provided a brief explanation on the population balance equation:
‘‘The population balance equation is an equation in the foregoing number
density and may be regarded as representing a number balance on particles of a particular
state. The equation is often coupled with conservation equation for entities in
the particles’ environmental (or continuous) phase.’’
In this chapter, both single-variable and multivariable population balances are
described. However, emphasis will be placed on the single-variable population
balance equations with size as the only internal coordinate.
3.1. General Population Balance Equations
A population balance for particles in some fixed subregion of particle phase space
can be conceptually represented as follows:
density function change
in class; location&time 
 ?
disperse in
through boundary 
 
disperse out
through boundary 
 
?
flow in
through boundary 
 
flow out
through boundary 
 
?
grow in
from lower classes 
 
grow out
from current class 
 
?
birth due to
coalescence 
 
death due to
coalescence 
 
?
breakup in
from upper classes 
 
breakup out
from current class 
 
?3:1?
562 Cameron and Wang
© 2005 by Taylor & Francis Group, LLC
The superstructure of the general population balance equation can be represented as
follows:
@
@t
f ?x; r; t? ?Hr 
 rmHr Drf ?x; r; t? ? Hr 
 _RRf ?x; r; t?Hx 
 _XXf ?x; r; t?
?Bc?x; r; t?Dc?x; r; t??Bb?x; r;t?Db?x; r; t?
?3:2?
where f is the multivariant number density as a function of properties and locations;
r is the external coordinate vector (also known as spatial coordinate vector) for the
determination of particle locations; x is the internal coordinate vector for the identi-
fication of particle properties, such as size, moisture content, and age; Dr is the dispersion
coefficient; R? is the velocity vector in the external coordinate system; X? is the
rate vector in the internal coordinate system; Bc and Dc are birth and death rates for
coalescence, respectively; Bb and Db are birth and death rates for breakage, respectively.
The first and second terms in the right-hand side of Eq. 3.2 represent dispersion
and convection particle transport, respectively, whereas the third quantifies the
growth of particles with respect to various properties, such as size and moisture. The
birth and death rates for coalescence are given by:
Bc?x; r; t? ? ZOx
dVx0 ZOr
1
d
b?~x; ~r; x0; r0?f ?x; r; t?f ?x0; r0; t?
@?~x; ~r?
@?x; r?
dVr0
Dc?x; r; t? ? f ?x; r; t? ZOx
dVx0 ZOr
b?x0; r0; x; r?f?x0; r0; t? dVr0
?3:3?
where b is the coalescence kernel, Ox and Or are integration boundaries for internal
and external coordinates, respectively, d represents the number of times identical
pairs have been considered in the interval of integration so that 1/d corrects for
the redundancy, the term @?~x;~r? @?x;r?
stands for the coordinate transformation such that
the colliding pair with original coordinates ?~x; ~r and ?x0; r0, respectively, before collision
should be identified by the coordinates [x, r] after coalescence. Mathematically,
this requires that the density with respect to coordinates ~x?x; rjx0r0?; ~r?x; rjx0r0? ?  must be transformed into one in terms of (x, r) by using the appropriate Jacobian
of the transformation. Ramkrishna (35) showed that the determinant of the Jacobian
of the transformation satisfies the following equation:
@ ~x; ~r ? ?
@ x; r ? ??
@~x1
@x1   
@~x1
@xn
@~x1
@r1
@~x1
@r2
@~x1
@r3
 .
@~xn
@x1   
@~xn
@xn
@~xn
@r1
@~xn
@r2
@~xn
@r3
@~r1
@x1   
@~r1
@xn
@~r1
@r1
@~r1
@r2
@~r1
@r3
@~r2
@x1   
@~r2
@xn
@~r2
@r1
@~r2
@r2
@~r2
@r3
@~r3
@x1   
@~r3
@xn
@~r3
@r1
@~r3
@r2
@~r3
@r3


?3:4?
Granulation Process Modeling 563
© 2005 by Taylor & Francis Group, LLC
The birth and death rates for breakage are described as:
Bb?x;r; t? ? ZOr
dVr0 ZOx
b?x0;r0; t?P?x;rjx0;r0; t?S x0;r0; t ? ?f x0;r0; t ? ?dVx0 ?3:5?
Db?x; r; t? ? S x; r; t ? ?f x; r; t ? ? ?3:6?
where b(x0, r0, t) is the average number of particles formed from the breakage of a
single particle of state (x0, r) at time t, P(x, r,jx0, r0, t) is the probability density
function for particle from the breakage of state (x0, r0) at time t that have state (x, r),
S(x, r, t) is the selection function, which represents the fraction of particles of state
(x, r) breaking per unit time.
Equations 3.3 and 3.4 and involve three different locations: and ~r and r0 for the
colliding pair of particles and r for the agglomerated particle. Although this treatment
is general and mathematically rigorous, it could be unnecessarily complicated for engineering
applications. A common practice is to assume that these three locations are
very close to each other during the particle collision and granule formation. That is:
~r  r0  r ?3:7?
This assumption requires that the phenomenon of fast particle jumps in the system is
not severe, which is achievable in most industrial granulation processes. If Eq. 3.7
holds, Eq. 3.3 and 3.4 can be simplified considerably to obtain:
Bc?x; r; t? ?
1
2 ZOx
b ~x; x0; r ? ?f ~x; r; t ? ?f x0; r; t ? ?
@ ~x ? ?
@ x ? ?
dVx0
Dc?x; r; t? ? f x; r; t ? ?ZOx
b x0; x; r ? ?f x0; t ? ?dVx0
?3:8?
@ ~x ? ?
@ x ? ? ?
@~x1
@x1   
@~x1
@xn
.. .
.. .
.. .
@~xn
@x1   
@~xn
@xn


?3:9?
Similarly, Eq. 3.5 becomes:
Bb?x; r; t? ? ZOx
b?x0; r; t?P?xjx0; r; t?S x0; r; t ? ?f x0; r; t ? ?dVx0 ?3:10?
In the following, breakage effects have been considered negligible and Eq. 3.7
is always assumed to be valid.
3.2. One-Dimensional (1-D) Population Balance Models
One-dimensional population balance models for both batch and continuous systems
are described in this section as special cases of the generalized population balance
model stated in ‘‘General Population Balance Equations.’’
3.2.1. Batch Systems
For a well-mixed batch system with only one internal coordinate v (particle size), is
Eq. 3.2 reduced to:
564 Cameron and Wang
© 2005 by Taylor & Francis Group, LLC
@
@t
n?v; t? ? 
@
@v
Gn?v; t? ? 
?
1
2 Z v
0
b v  v0; v0 ? ?n v  v0; t ? ?n v0; t ? ?dv0  n?v; t?
Z 1
0
b v; v0 ? ?n v0; t ? ?dv0
?3:11?
where n is the one-dimensional number density, G is the growth rate. For notational
clarity, we use f and n to denote the multidimensional and one-dimensional number
density functions, respectively. Both notations bear the same physical significance.
Through a comparison of Eq. 3.11 with Eq. 3.2, 3.8, and 3.9, it is easy to observe
the following membership relationships:
v 2 x; v0 2 x0; v  v0 ? ?2~x; G ?
dv
dt 2 _XX;
@ ~x ? ?
@ x ? ? ?
@ v  v0 ? ?
@v ? 1 ?3:12?
Equation 3.11 is more frequently applied to industrial granulation processes than its
generalized format described in Eq. 3.2.
3.2.2. Continuous Systems
The population balance equation for continuous systems with one internal coordinate
and one external coordinate is given by:
@
@t
n?v; z; t? ? 
@
@z
_Z
Zn?v; z; t?  
@
@v
Gn?v; z; t? ? 
?
1
2 Z v
0
b v  v0; v0 ? ?n v  v0; z; t ? ?n v0; z; t ? ?dv0
 n?v; z; t? Z 1
0
b v; v0 ? ?n v0; z; t ? ?dv0
?3:13?
where the special velocity is defined as:
_Z
Z ?
dz
dt 2 _RR ?3:14?
Although continuous granulation processes are commonly encountered in the fertilizer
and mineral processing industries, they are also performed in the pharmaceutical
industry as batch processes employing either high-shear mixers or batch
fluidized-bed granulators. Consequently, most modeling studies on pharmaceutical
granulation have focused on batch processes. However, it is important to obtain complete
knowledge of both batch and continuous granulation processes for improved
design and operations.
3.2.3. Coalescence Kernels
It is easy to see that a coalescence kernel is affected by two major factors: (1) collision
probability of the specified pair of particles, and (2) successful coalescence or
rebounding after collision. The first mainly depends on the particle sizes, granulator
configurations, particle flow patterns, and operating conditions. The second has been
intensively studied by Liu et al. (25) who identified the following four most important
aspects affecting the success of coalescence: elasticplastic properties, viscous
fluid layer, head of collision, and energy balance. The authors have also observed that
there are two types of coalescence distinguished by particle deformations. That is,
Granulation Process Modeling 565
© 2005 by Taylor & Francis Group, LLC
Type I coalescence is not associated with any particle deformation during the
collision, whereas the Type II is accompanied by particle deformations. Liu and
Litster (26) further proposed a new physically based coalescence kernel model based
on the criteria developed earlier [Liu et al., (25)]. From these fundamental studies, it
can be determined qualitatively that the coalescence kernels should depend on particle
sizes, energy consumption, particle deformability, and most importantly, the
moisture content (viscous fluid layer). A historical summary of the proposed coalescence
kernels is given in Table 1, which is an extension of the table originally presented
by Ennis and Litster (11) with the new coalescence kernel developed by Liu and Litster
(26) and another kernel from aerosol dynamics [Friedlander, (13)].
3.3. Two-Dimensional (2-D) Population Balance Models
In this section, we study a perfect mixing, batch granulation system with two internal
(property) coordinates: particle value v and liquid value vL. Because of the perfect
Table 1 Summary of the Proposed Coalescence Kernel in the Literature
Kernel References
b ? b0 (18)
b ? b0 ?u?v?a
?uv?b (19)
b ? b0
u2=3?v2=3 ? ? 1=u?1=v
(40)
b ? a?u ? v? (14)
b ? a ?uv?2
?u?v?
(14)
b ?
k; t < ts
a?u ? v?; t > ts 
 (1)
k: constant, ts: switching time
b ?
k; w < w
0; w > w 

w ? ?u ? v?a
?uv?b
(2)
k, a, b: constants
w: critical granule volume
b ? b0 1=u ? 1=v ? ?1=2 u1=3 ? v1=3  2
b ? b0 u1=3 ? v1=3  u1=3 ? v1=3  
(13)
b u;v  ?
b1 Types I & II without permanent deformation
b2 Type II with permanent deformation
0 rebound ( Liu and Litster (2001)
566 Cameron and Wang
© 2005 by Taylor & Francis Group, LLC
mixing feature, there is no spatial coordinate in the model. However, the proposed
modeling strategy can be easily extended to continuous processes with both internal
and external coordinates. The 2-D population balance equation for a batch granulation
process is:
@
@t
f ?v; vL; t? ? 
@
@v
dv
dt
f ?v; vL; t?  	
@
@vL
dvL
dt
f ?v; vL; t?  	
?
1
2 Z v
0 Z min vL;vv0 ? ?
0
b v  v0; vL  v0L; v0; v0L 
 
f v v0; vL  v0L; t 
 f v0; v0L; t 
 dv0Ldv0
f ?v; vL; t? Z 1
0 Z vL
0
b v; vL; v0; v0L 
 f v0; v0L; t 
 dv0Ldv0
?3:15?
The relationship between the bivariant number density function f and single-variant
number density function n is determined as:
n?v; t? ? Z v
0
f v; vL; t ? ?dvL ?3:16?
For the aggregation-only processes, the first two terms on the right-hand side of
representing convective particle transport and particle growth by layering are negligible.
Equation 3.15 is reduced to:
@
@t
f ?v; vL; t? ??
1
2 Z v
0 Z min vL;vv0 ? ?
0
b v  v0; vL  v0L; v0; v0L 
 
f v v0; vL  v0L; t 
 f v0; v0L; t 
 dv0Ldv0
 f ?v; vL; t? Z 1
0 Z vL
0
b v; vL; v0; v0L 
 f v0; v0L; t 
 dv0Ldv0
?3:17?
Under certain mathematical assumptions, a two-dimensional population balance
equation can be reduced to two single-dimension population balance equations that
are described in the following section.
3.4. Reduced Order Models
3.4.1. Reduced Order Models Using the Concept of Lumped Regions in Series
When particle populations are spatial-dependent, such as that in a long rotating
drum granulator, the population balance model is described in Eq. 3.13 with spatial
variable z included in the model equation. In many industrial applications, the concept
of lumped regions in series is used to reduce the model order. By using this
method, a granulator is divided into a number of sections with an assumption that
perfect mixing can be achieved in each section. The basic idea is schematically
In Figure 3, Q denotes the number flow -rate, the subscripts F and P represent
the feed and product streams, respectively, and NR is the total number of regions
Granulation Process Modeling 567
depicted in Figure 3.
© 2005 by Taylor & Francis Group, LLC
used to approximate the granulator. The reduced order model Eq. 3.13 for using the
method of lumped regions in series is given by:
@
@t
n?v; i; t? ? 
@
@v
Gin?v; i; t? ? ?Q?i  1?
n?v; i  1; t?
nt?i  1; t?  Q?i?
n?v; i; t?
nt?i; t?
?
1
2 Z v
0
b v  v0; v0 ? ?n v  v0; i; t ? ?n v0; i; t ? ?dv0
 n?v; i; t? Z 1
0
b v; v0 ? ?n v0; i; t ? ?dv0
i ? 1; 2; . . . ;NR
?3:18?
where i represents the ith region, nt is the total number density, and Q(0)?QF.
3.4.2. Model Order Reduction for Multi-Dimensional Population Balances
Biggs et al. (7) have developed the concept of binder size distribution (BSD) to
correlate moisture content with particle size. Based on BSD, the mass of binder in
the size range (v, v?dv) is quantified as dM?M(v) dv and:
M?t; v? ? rL Z v
0
vLf v; vL; t ? ?dvL ?3:19?
where rL is the binder density.
They showed that if the assumption that at a given size all granules have the
same liquid content, the 2-D population balance equation given by Eq. 3.17 can
be reduced to a set of two, one-dimensional equations described as follows:
@
@t
n?v; t? ?
1
2 Z v
0
b v  v0; v0 ? ?n v  v0; t ? ?n v0; t ? ?dv0
 n?v; t? Z 1
0
b v; v0 ? ?n v0; t ? ?dv
?3:20?
@
@t
M?v; t? ?
1
2 Z v
0
b v  v0; v0 ? ?M v v0; t ? ?n v0; t ? ?dv0
M?v; t? Z 1
0
b v; v0 ? ?n v0; t ? ?dv
?3:21?
In their experiments, pharmaceutical materials were granulated in a high-shear
mixer. Good agreements between experimental and simulation results were achieved
Figure 3 Concept of lumped regions in series.
568 Cameron and Wang
© 2005 by Taylor & Francis Group, LLC
enabling the granulation rates to be defined by two parameters: the critical binder
volume fraction and the aggregation rate constant.
3.4.3. Reduced Order Models Using the Method of Moments
The moments are defined as:
Mj ? Z 1
0
vjn?v?dv
mj ? Mj=M0
j ? 0; 1; 2; . . .
?3:22?
Because of the variety of coalescence kernels, it is impossible to develop a generalized
structure for reduced order models using the method of moments. A special kernel
model is assumed in this work. The methodology can be extended to the development
of moment models with different kernel structures. The sample kernel model
is assumed as:
b?v; v0? ? b0
vb ? v0b
?vv0?a ? b0
v0?ba?
va ?
v?ba?
v0a  	 ?3:23?
The discretized format of Eq. 3.23 is given by:
bi;j ? b0
v?ba? j
va
i ?
v?ba? i
va
j " # ?3:24?
The one-dimensional aggregation-only population balance equation (PBE) described
by Eq. 3.20 with the kernel model given by Eq. 3.24 can be reduced to a set of
ordinary differential equations as follows:
d
dt
M0 ? b0 m?ba?ma  M0
d
dt
M1 ? 0
d
dt
Mr ?
1
2
b0X
r1
k?1
r
k  
 m?ka?m?rk?ba? ? m?k?ba?m?rka? h iM2
0 ;
r ? 2; 3; . . .
?3:25?
where m is defined in Eq. 3.22, for example, m(ka)?M(ka)/M0. Equation 3.25
involves the determination of fractional and negative moments. If the type of particle
size distribution is more or less known, such as log-normal or G-distribution,
Eq. 3.25 is solvable with the incorporation of interpolation and extrapolation
techniques. For more general solution techniques, fractional calculus enabling the
computation of fractional differentiations and integrations should be used, which
exceeds the scope of this chapter.
3.4.4. Multitimescale Analysis
It is often the case that in an interconnected process situation where several processes
are being simulated simultaneously that processes operate on distinct timescales.
Such is the case when combinations of prereaction units are combined
with granulation devices, dryers, and screening in full process flowsheet simulations.
Granulation Process Modeling 569
© 2005 by Taylor & Francis Group, LLC
It can also be the case within a particular processing unit that incorporates a range
of mechanisms.
It can be observed that these processes often operate on different timescales
covering the range of microseconds to minutes or even hours. This time separation
in scales provides opportunity to make assumptions that can simplify the modeling
by separating the phenomena into at least three classes:

 Slow modes (long time constant behavior)

 Medium modes

 Fast modes (short time constants)
When we do this analysis, we can often use qualitative methods based on our
general understanding of physics, chemistry, and the rate processes such as heat and
mass transfer. The alternative and more complex analytical approach is through the
use of eigenvalue and eigenvector analysis (37), which is based on the underlying
models of the processes. This analysis often allows us to simplify complex models
when we model for particular goals by making the following assumptions:

 Slow modes can be treated as being constant over the timeframe of interest.

 Medium modes are modeled in detail

 Fast modes are regarded as pseudosteady states, being represented by algebraic
equations.
This timescale approach can simplify significantly the complexity of the process
models depending on the timeframe of interest in the simulation and the approach
has general application to all forms of models.
4. SOLVING AND USING POPULATION BALANCES
4.1. Solution of Population Balance Equations
4.1.1. Discretization Methods
4.1.1.1. Hounslow Discretization. Hounslow et al. (16) developed a relatively
simple discretization method by employing an M-I approach (the mean value theorem
on frequency). The population balance equations, such as Eq. 3.20, are normally
developed using particle volume as the internal coordinate. Because of the identified
advantages of length-based models, Hounslow et al. (16) performed the coordinate
transformation to convert the volume-based model described by Eq. 3.20 to a
length-based model as follows:
d
dt
n?L; t? ?
L2
2 Z L
0
b L3  l3 
 1=3
; l h in L3  l3 
 1=3
; t h in?l; t?
L3  l3 
 2=3 dl
 n?L; t? Z 1
0
b?L; l?n?l; t? dl
?4:1?
in which L and l denote the characteristic length of particles. The Hounslow method
is based on a geometric discretization with the following ratio between two successive
size intervals:
Li?1=Li ? ffiffiffi 2 3 p ; or vi?1=vi ? 2 ?4:2?
570 Cameron and Wang
© 2005 by Taylor & Francis Group, LLC
where L and v represent the chacteristic length and volume of particles, respectively,
the subscripts i?1 and i denote the size classes. The continuous population balance
equation described by Eq. 4.1 is converted into a set of discretized population balance
equations in various size intervals by using this technique. That is, the change
of number density in the ith size interval is given by:
d
dt
ni ? ni1X
i2
j?1
2ji?1bi1; jnj 
 ?
i
2
bi1;i1n2
i1
 niX
i1
j?1
2jibi; jnj 
  niX
imax
j?1
bi; jnj 
 
i ? 1; 2; . . . ; imax
?4:3?
The continuous binder size distribution model described by Eq. 3.21 can also be
discretized using a similar numerical scheme as follows (7):
d
dt
Mi ? Mi1X
i 2
j?1
2ji?1bi1; jnj 
 ? ni1X
i 2
j?1
2ji?1bi1; jnj 
 
? niX
i1
j?1
1  2ji 
 bi; jMj  ? bi1;i1ni1Mi1
MiX
i 1
j?1
2jinj 
 MiX
imax
j?1
bi; jnj 
 
i ? 1; 2; . . . ; imax
?4:4?
4.1.1.2. Kumar and Ramkrishna’s Discretization Technique. Kumar and
Ramkrishna (22) developed a discretization method by using a grid with a more general
and flexible pattern with fine or coarse discretizations in different size ranges.
The size range between two sizes vi and vi?1 is called the ith section, and the particle
size in this section is simply denoted by xi (grid point) such that vi < xi < vi?1 as seen
in Figure 4.
A particle of size v in the size range xi and xi ? 1 can be represented by two
fractions a(v, xi) and b(v, xi ? 1) associated with the two grid points xi and xi ? 1,
respectively. For the conservation of two general properties f1(v) and f2(v), these fractions
satisfy the following equations:
a v; xi ? ?f1 xi ? ??b v; xi?1 ? ?f1 xi?1 ? ??f1?v?
a v; xi ? ?f2 xi ? ??b v; xi?1 ? ?f2 xi?1 ? ??f2?v? ?4:5?
Figure 4 General grid used with Kumar & Ramkrishna numerical technique.
Granulation Process Modeling 571
© 2005 by Taylor & Francis Group, LLC
By using this composition technique for particle properties, discrete equations for
coalescence-only population balance model given by Eq. 3.20 have been formulated
as follows:
dni
dt ? X
jk
j;k
xi1	 xj?xk ? ?	xi?1
1 
1
2
dj;k  
Zb?j; k?nj?t?nk?t?  ni?t?X
imax
k?1
b?i; k?nk?t? ?4:6?
In Eq. 4.6, ni, b, imax are defined previously, dj,k is the Dirac-delta function, and Z is
defined as follows:
Z ?
xi?1  v
xi?1  xi
; xi 	 v 	 xi?1
Z ?
v  xi1
xi  xi1
; xi1 	 v 	 xi
?4:7?
The first and second terms on the right-hand side of Eq. 4.6, respectively, represent
the birth rate and death rate of particles because of the ith size interval coalescence.
Attention should be paid to the selection of the internal coordinates. The original
Kumar–Ramkrishna discretization should be applied to volume-based models
rather than length-based models. Although both are interconvertible, it is important
to check consistency in numerical computations.
4.1.2. Wavelet-Based Methods
The wavelet-based methods are relatively new numerical schemes for solving population
balance equations consisting of both differential and integral functions (28).
Again, the volume-based population balance equations with particle volume as the
internal coordinate are used to demonstrate the main characteristics of the wavelet
methods. The most important advantage of these methods over other numerical
techniques is their ability to effectively deal with steep moving profiles. Here, we only
explain the basic algorithms of the wavelet collocation method for practical applications
using the Daubechies wavelets rather than provide mathematical insights for
general wavelet techniques.
Similar to other collocation methods, the coordinates should be normalized
within the interval [0, 1]. For the 1-D population balance equation given by Eq. 3.11,
this can be done by introducing the linear transformation x?v/vmax, where x is the
dimensionless particle volume and vmax is the maximum particle size in the system.
The original integral intervals [0, v] and [0, 1] are transformed to [0, x] and [0, 1],
respectively. Consequently, Eq. 3.11 becomes:
@
@t
n?x; t? ? 
@
@x
G?x?n?x; t? ? 
?
vmax
2 Z x
0
b x  x0; x0 ? ?n x x0; t ? ?n x0; t ? ?dx0
 vmaxn?x; t? Z 1
0
b x; x0 ? ?n x0; t ? ?dx0
?4:8?
where G(x) is defined as dx/dt rather than dv/dt. For a broad class of engineering
problems, the approximate solution of a general function w(x) with J-level resolution
572 Cameron and Wang
© 2005 by Taylor & Francis Group, LLC
can be written in terms of its values in the dyadic points:
wJ ?x? Xm
wJ ?2Jm?y?2Jx  m? ?4:9?
where y(x) is denoted as the autocorrelation function of scaling function. We first
solve the coalescence-only PBE with G(x)?0. If J-level wavelet method is used,
the matrix representation at the ith dyadic point is given by:
@ni
@t ?
vmax
2
n0 n1    n2J ? M3;i
n0
n1
.. .
n2J
26664
37775
 vmaxniM2
i
n0
n1
.. .
n2J
26664
37775
?4:10?
where ni is the number density at the ith dyadic (collocation) point. The operational
matrix M3,i and vector Mi
2 are constructed as follows. M3,i are (2J?1)  (2J?1)
operational matrices at volume points are represented as:
M3;i ?
M3;i
0;0 M3;i
0;1    M3;i
0;2J
M3;i
1;0 M3;i
1;1    M3;i
1;2J
.. .
.. .
. .
. .. .
M3;i
2J ;0 M3;i
2J ;1    M3;i
2J ;2J
2666664
3777775
?4:11?
Mi
2 are 1(2J?1) operational vectors at volume points and are described by:
M2
i ? M2
i;0 M2
i;1    M2
i;2J   Elements in matrix M3,i are developed as:
M3;i
k1;k2 ?
1
2JX2
J
l?0
b?xi  xl ; xl?
 Olk2;ik1k2 ?i  k2?  Olk2;ik1k2?k2? 
 
?4:12?
Elements in the operational vectors Mi
2 are given by:
M2
i;k ?
1
2J X2
J
l?0
b?xi; xl?Hkl?k? " # ?4:13?
The required two-term integral of autocorrelation function Hk (x) and three-term
integral of autocorrelation function Oj,k(x) in Eqs. 4.12 and 4.13 are defined as:
Hk?x? ? Z x
1
y?y  k?y?y?dy ?4:14?
Oj;k?x? ? Z x
1
y?y  j?y?y  k?y?y?dy ?4:15?
The autocorrelation function y (k) and its derivatives y(s)(k) are represented
as follows:
y?k? ? Z ?1
1
f?x?f?x  k? dx
y?s??k? ? ?1?s Z ?1
1
f?x?f?s??x  k? dx
?4:16?
Granulation Process Modeling 573
© 2005 by Taylor & Francis Group, LLC
where j(x) is the scaling function. y(s)(k) can be evaluated by using the following
recursive algorithm with y(k)?y(0)(k):
y?s??2J1k? ? 2sy?s??2Jk? ? 2s1XN
l?1
a2l1?y?s??2Jk  2l ? 1?
? y?s??2Jk  2l  1?? ?4:17?
The differential operators in the PBE can also be evaluated at the collocation
points as:
@n
@x ? A?y?1??n
@2n
@x2 ? B?y?2??n
?4:18?
where n?[n1, n2, . . . , nN]T is a vector in which ni is the number density at the ith
collocation point, N?2J?1 is the number of collocation points, T stands for vector
transpose, A and B are square matrices computed using the values y(1)(k) and y(2)(k),
respectively. An algorithm for the computation of the matrices A and B was
described in Liu et al. (27). Consequently, the growth term in Eq. 4.8 can be approximated
using Eq. 4.18.
It should be pointed out that after coordinate normalization, functions of interest
are evaluated in the closed intervals [0, 1], rather than in [1, 1] or other intervals.
In this case, some modified interpolation functions can be constructed to
interpolate the values in dyadic points outside [0, 1] to the desired interval (6,25).
Three population balance equations with different kernel models have been
successfully solved by using the wavelet collocation method (28). These kernel models
were: (1) size-independent kernel b?b0?constant, (2) linear size-dependent
kernel b?x; x0? ? b0?x ? x0?, and (3) nonlinear size-dependent kernel Q?x; x0? ? b?x1=3 ? x01=3??x1=3 ? x01=3?. Simulation results have shown that the wavelet collocation
methods are able to achieve fast convergence with high accuracy if adequate
resolution levels are selected. The methods are particularly effective for the processes
with steep moving profiles that are difficult to solve by using other numerical schemes.
Here, emphasis has been placed on the introduction of basic techniques for the
resolution of population balance equations using wavelets, which is required to solve
2J?1 ordinary equations for each population balance equation given by Eq. 4.8.
Recently, Liu and Cameron (29) have developed a new wavelet-based adaptive technique,
which enables a dramatic reduction of the number of ordinary differential
equations to be solved. Furthermore, this adaptive method allows the automatic
selection of the minimum wavelet level J with acceptable accuracy. With the background
knowledge described in this chapter, readers may understand the adaptive
technique through studies on the original papers without major difficulties.
Website on the wavelet collocation methods for the computations of population
balance equations. The operational matrices M3,i and Mi
2, matrices A and B,
together with the integral functions H and O at various resolution levels are available
4.1.3. Solving Differential-Algebraic Equation DAE Systems
Many of the previously mentioned numerical methods lead to large sets of differential
equations coupled with sets of nonlinear algebraic equations. These are the
574 Cameron and Wang
at http://www.cheque.uq.edu.au/psdc.
© 2005 by Taylor & Francis Group, LLC
so-called DAE systems. A number of approaches are available to solve these equation
sets, mainly based on implicit or semi-implicit methods such as the backward
differentiation formulae (BDF) (34) or variants of Runge–Kutta methods (9,53).
The Mathworks package MATLABTM also contains useful DAE solvers in the
latest versions that are based primarily on implicit BDF formulae. Solutions of these
types of problems are generally straightforward. Some issues still remain in obtaining
consistent initial conditions for the solution to commence.
4.2. Monte Carlo Methods
There is a long history in studies on application of Monte Carlo methods to process
engineering. The first serious research paper on aMonte Carlo treatment for systems
involving population balances could be credited to Spielman and Levenspiel (42).
Since then, a significant number of publications have appeared in the literature on
the resolution of population balance equations using Monte Carlo methods (35).
Comprehensive Monte Carlo treatments are described in the literature (20,35). Only
selected issues on basic techniques are addressed in this section.
4.2.1. Classification of Monte Carlo Methods
Monte Carlo methods can be used in two ways for engineering applications.
1. Direct evaluation of difficult functions. For example, the integral given by:
I ? Z b
a
f ?x? dx ?4:19?
can be evaluated as:
I ? E?Y? ? E ?b  a?g?X? ? ?E 
YY?n? ? 

Y
Y?n? ?    ?b  a?Pn
i?1
g Xi ? ?
n
?4:20?
where X1, X2, . . . , Xn are random variables defined in the closed interval [a, b],
and E(Y) denotes the mathematical estimation of function Y.
2. Artificial realization of the system behavior (35). This method is commonly
applied to complex particulate processes, which are described in some
detail here. In the artificial realization, the direct evaluation of integral
and differential functions is replaced by the simulation of the stochastic
behavior modeled by using a randomness generator to vary the behavior
of the system (20). The important probabilistic functions in the original
model equations, such as coalescence kernels for granulation processes,
are still essential in Monte Carlo simulations and are shown later.
Monte Carlo methods for the artificial realization of the system behavior can
be divided into time-driven and event-driven Monte Carlo simulations. In the former
approach, the time interval Dt is chosen, and the realization of events within this time
interval is determined stochastically. Whereas in the latter, the time interval between
two events is determined based on the rates of processes. In general, the coalescence
rates in granulation processes can be extracted from the coalescence kernel models.
The event-driven Monte Carlo can be further divided into constant volume methods
Granulation Process Modeling 575
© 2005 by Taylor & Francis Group, LLC
in which the total volume of particles is conserved, and the constant number method
in which the total number of particles in the simulation remains constant. The main
advantage of the constant number method for granulation processes is that the
population remains large enough for accurate Monte Carlo simulations (41,52).
An additional advantage associated with the constant number methods is its ability
to reduce the renumbering effort. Consequently, the constant number method is
recommended and is further explained.
4.2.2. Key Equations for Constant Number Monte Carlo Simulation
Key equations needed in Monte Carlo simulations include the interevent time Dtq
representing the time spent from q1 to q Monte Carlo steps, coalescence kernel
Kij, normalized probability pij for a successful collision between particles i and j,
and a number of intermediate variables. The coalescence kernel can be divided into
particle property independent part Kc and dependent part kij (Xi, Xj) as follows:
Kij ? Kckij Xi;Xj 
  i; j ? 1; 2; . . . ;N ?4:21?
where X denotes the vector of internal coordinates representing particle properties,
such as size and moisture content, and N is the total number in the simulation system.
It can be seen that Eq. 4.21 is similar to the coalescence kernel given by bij?b0
kij (vi, vj) described in the previous sections for one-dimensional systems. However, it
should be pointed out that i and j in Eq. 4.21 are used to identify the individual particles,
whereas that in bij, i, j?1,2, . . . , imax are size classes rather than particle identity
numbers. To avoid confusion, bij and b0 are replaced by Kij and Kc, respectively,
in Monte Carlo simulations. The normalized probability for successful collision is
given by:
pij ?
kij
kmax ?4:22?
where kmax is the maximum value of the coalescence kernel among all particles. The
final result of the interevent time is given by:
Dtq ?
2tc
kij  
1
N
N
N  1  
q
?4:23?
with
tc ?
1
KcC0 ?4:24?
and
kij  ? PN
i?1 PN
j?1;i6?j
kij
N?N  1? ?4:25?
In Eq. 4.25, C0 is the total number concentration at t?0 defined by C0?N/V0
where V0 is the volume of particles at the initial time. We presented only the final
results of the needed equations here. Interested readers are referred to Smith and
Matsoukas (41) for detailed mathematical derivations.
576 Cameron and Wang
© 2005 by Taylor & Francis Group, LLC
4.2.3. Simulation Procedure
The simulation procedure for the constant number Monte Carlo method applied to
coalescence processes consists of the following key steps:
1. Initialization of the simulation system. This includes the determination of
sample size (normally 10,000–20,000 particles) followed by assigning the
identity number and properties to each particle. The properties must satisfy
the initial property distributions, such as particle size and moisture distributions.
Set t0?0 and q?1.
2. Acceptance or rejection of coalescence. In this step, two particles, i and j,
are randomly selected and the coalescence kernel kij with normalised probability
pij given by Eq 4.22 are computed, followed by the generation of a
random probability prq. If pij < prq, the coalescence is rejected, and a new
pair of particles is selected again to repeat the calculation until pij > prq,
which implies a successful coalescence. When the coalescence is successful,
the new agglomerated particle holds the identity number i, and another
particle randomly selected from the rest of the system is copied as particle
j and go to Step 3.
3. Computation of the interevent time. The interevent time for step q is computed
using Eqs 4.23–4.25. Total operational time is given by:
t ? t0 ?X
q
m?1
Dtm ?4:26?
4. Set q?q?1 and return to Step 2.
5. Simulation termination and result validation. As t reaches the prespecified
termination time tf, check the acceptance of simulation results. If acceptable,
stop the simulation; otherwise, modify model parameters and start a
new simulation process.
It can be seen that the Monte Carlo methods are applicable to both one-dimensional
and multidimensional coalescence processes without any theoretical and
algorithmic hurdles. However, most reported results with good agreement with
experimental data are limited to one-dimensional systems except that reported by
Wauters (52). This is mainly because of the lack of reliable multidimensional kernel
models rather than the applicability of Monte Carlo methods.
5. APPLICATION OF MODELING TECHNIQUES
5.1. Modeling for Closed-Loop Control Purposes
5.1.1. Development of Control Relevant, Linear Models
As the linear control theory and techniques are better developed and easier to implement
than their nonlinear counterparts, it is highly desirable to use linear models for
control purposes. In process engineering, the nonlinear models are frequently linearized
around certain operating points. The linearization technique is described briefly
here. Let the general nonlinear system be described as:
dx
dt ? f?x;u?
y ? h?x? ( ?5:1?
Granulation Process Modeling 577
© 2005 by Taylor & Francis Group, LLC
where x?[x1, x2, . . . , xp]T, y?[y1, y2, . . . , yq]T, and u?[u1, u2, . . . , us]T are vectors
of state, output,and control variables, respectively, f?[f1, f2, . . . , fp]T and h? [h1, h2, . . . , hq]T are vectors of smooth functions, in which p, q, and s are dimensions
of the vectors of state, output and control variables, respectively. In the population
balance equations given by Eqs. 4.3, 4.6 and 4.10, x?[n1, n2, . . . , np], p?imax. The
conventional linearization method is based on the first-order Taylor series expansion
around certain operational points. The resulting linear model is given by:
ddx
dt ? Adx ? Bdu
dy ? Cdx
A ?
@f?x;u?
@xT x ? xo
u ? uo

;B ?
@f?x;u?
@uT x ? xo
u ? uo

;C ?
@h?x?
@xT x ? xo
u ? uo

?5:2?
In the control literature, the symbol d in front of x, y, and u is normally omitted for
simplicity. The readers should be aware that in the models developed this way, x, y,
and u denote deviations from their respective values at the specified operational
point rather than the real values. That is, the linearized model used in control studies
is represented as:
dx
dt ? Ax ? Bu
y ? Cx
?5:3?
The discretized population balance equations given by Eqs. 4.3, 4.6 and 4.10, and the
binder size distribution model described by Eq. 4.4 can be linearized to obtain the
models with the format given by Eq. 5.3. The control variables are normally connected
with the coalescence kernels (55).
5.1.2. ARX and ARMAX Models for Linear Model Predictive Control
For model predictive control purposes, there are two commonly used black box
models: ARX model with autoregressive (AR) part and extra (X) input, and
ARMAX model with additional moving average (MA) part accounting for disturbances.
The method for the development of ARX and ARMAX models is well
explained by Ljung (30). The single input, single output ARX is given by:
y?t? ? a1y?t  1??  ?anay?t  na? ? b1u?t  1??  ?bnbu?t  nb? ? e?t?
?5:4?
and the ARMAX model is represented as:
y?t? ? a1y?t  1??  ?anay?t  na? ?
b1u?t  1??  ?bnbu?t  nb? ? e?t? ? c1e?t  1??  ?cnce?t  nc?
?5:5?
In Eqs. 5.4 and 5.5, y is the output (controlled) variable; u is the input (manipulative)
variable, e is the disturbance; a, b, and c are time varying coefficients identified
on-line; na, nb, and nc are defined as prediction, control, and disturbance horizons.
578 Cameron and Wang
© 2005 by Taylor & Francis Group, LLC
The matrix format for multivariable RRX and ARMAX models is described by:
A?q?y?t? ? B?q?u?t? ? e?t? ?ARX? ?5:6?
A?q?y?t? ? B?q?u?t? ? C?q?e?t? ?ARMAX? ?5:7?
In Eqs (5.6) and (5.7), matrices A, B, and C are defined as:
A?q? ? Iny ? A1q1 ?  ?Anaqna
B?q? ? B0 ? B1q1 ?  ?Bnbqnb
C?q? ? C0 ? C1q1 ?  ?Cncqnc
?5:8?
where qk is the delay operator representing ‘‘delayed by k time intervals,’’ for
example:
A?q?y?t? ? y?t? ? A1y?t  1??  ?Anay?t  na? ?5:9?
The compact format of ARX and ARMAX models given by Eqs 5.6 and 5.7 can be
easily converted into more intuitive, expanded format exemplified by Eq. 5.9. With
input (u and e) and output (y) data, the matrices A, B and C can be readily identified
employing the System Identification Toolbox for Use with MATLABTM (31). An
ARX model for a pan, granulation process was developed by Adetayo et al. (3) with
a successful application for effective control of the plant.
5.1.2. Nonlinear Model Predictive Control Structure
Nonlinear model predictive control (NMPC) schemes consist of simultaneous determinations
of manipulative variables and uncertain parameters. In some cases, the
open-loop dynamic optimization is carried out for the determination of desired trajectories
(set-points). This integrated control strategy was developed by Miller and
Rawlings (32a) in a study on model identification and control for batch cooling crystallizers,
in which the population balance described by partial differential equation
was reduced to a low-dimensional model using method of moments. That is, the
control objective was of average size rather than the size distribution in Miller and
Rawlings’ work. Detailed studies on modeling and model predictive control
(MPC) of particle size distribution in emulsion co-polymerization processes using
population balance models have been carried out by Immmanuel et al. (17) and
Crowley et al. (10). The reported research results have shown that the NMPC
schemes using population balance equations should also be applicable to pharmaceutical
granulation processes because of the similar model structure. A general
an integration of the modeling strategy originally proposed by Sanders et al. (39)
with the model based control scheme.
A full implementation of NMPC in industrial granulation plants using physically
based models has not been reported yet. A simulated study has been carried out
by Zhang et al. (55) to control an industrial-scale fertilizer plant using a physically
based model. The main limitation of the study was that the physically based population
balance model was used to generate output data without real on-line measurements.
This implies that if a severe plant-model mismatch occurs, the proposed
control strategy may fail. Further work is required to modify the model on-line,
based on the measurement data.
Granulation Process Modeling 579
structure for nonlinear model predictive control is shown in Figure 5, which depicts
© 2005 by Taylor & Francis Group, LLC
5.1.3. On-Line Measurement-Based Control Schemes
In addition to model-based control schemes using population balance equations,
there are a number of practical control schemes in the pharmaceutical industry, that
do not rely on mathematical models. These include simple feedback control with or
without feed-forward compensation, and fuzzy-logic control systems.
5.1.3.1. Simple Feedback Control with Feed-Forward Compensation. One of
the most important issues for the effective control of granulation processes is the
development of fast and reliable measurement techniques for the characterization
of particulate systems. Because of the difficulties associated with the direct measurement
of particle characteristics, such as particle size distribution, moisture content,
and deformability, some indirect monitoring parameters have been adopted as
the indicators of particle characteristics. A commonly accepted monitoring parameter
in the pharmaceutical industry is power consumption, which has been successfully
used to control the particle size in high-shear mixers at the end-point (12,23).
Based on a series of investigations carried out by Leuenberger, (23), the energy
dissipated per unit volume in a high-shear mixer, dW/dV, can be approximately
represented as:
dW
dV ? msck /
1  e
e ?5:10?
where W is the power consumption, V is the granulator volume, m is the apparent
coefficient of friction, sc is the cohesive stress, k is the dimensionless shear rate,
and e is the porosity of the powder mass. It is easy to show that power consumption
is related to the saturation level S and is defined as follows:
S ?
H?1  e?
e
r ?5:11?
where H is the mass ratio of liquids to solids, and r is the density of the particle relative
to that of the liquid (r?rS/rL). Furthermore, Kristensen and Schaefer (21)
pointed out that the saturation level defined by Eq. 5.11 could be related back to
the average granule size. Consequently, the power consumption, the saturation level,
and the granule particle size are inter-related, which forms a technical basis to use
Figure 5 General structure of NMPC using physically based models.
580 Cameron and Wang
© 2005 by Taylor & Francis Group, LLC
power consumption as a monitoring parameter for the characterization of particles
within the high-shear mixer. A detailed description of the control strategy using.
Mort et al. (33) pointed out that: ‘‘With recent development in particle sizing
technology, the agglomerate size distribution can be measured in-line at any number
of points in the process.’’ The main measurement technique is image analysis by
mounting high-speed cameras and lighting systems in appropriate locations. As
the direct measurement data of particle sizes are available, the controller design
can be based on these data without relying on the indirect indicators under the condition
that the rate of binder addition is sufficiently slow to allow for image data to
be collected, processed, and fed back. This concept has been used for batch granulation
processes in fluidized beds. The same authors also proposed a feed-forward control
strategy to compensate the fluctuation of the recycle rate. The simple feedback
control with feed-forward compensation scheme is shown in Figure 6.
The measurement data in Figure 6 could be the indirect monitoring parameters
(23) or the explicit particle size distribution (33), depending on the relative speed of
the measurement system and process dynamics.
5.1.3.2. Fuzzy-Logic Control of High-Shear Granulation. Watano et al.
(50,51) have developed a novel system to control granule growth in a high-shear mixer.
The system basically consisted of image processing and a fuzzy controller as shown in
Figure 7.
In Figure 7, D(t) is the deviation between the desired value (Dd) and the measured
value (Dm) of granule size, and DD(t) denotes the change rate of measured
values that are mathematically represented as follows:
D?t? ? Dd  Dm?t?
DD?t? ? Dm?t?  Dm?t  1? ?5:12?
Other notations in Figure 7 are as follows. V(t) is the result of fuzzy reasoning used
to control the output power of liquid feed pump, while K1 and K2 represent gains of
the input variables.
Figure 6 Simple feedback control scheme with feed-forward compensation.
Figure 7 Block diagram of granule size control system (after Watano et al., 2001).
Granulation Process Modeling 581
© 2005 by Taylor & Francis Group, LLC
In the methodology developed by Watano et al. (50,51), four fuzzy variables
were used, namely ZR (zero), PS (positive small), PM (positive medium) and PL
(positive large). The values of D(t), DD(t) and V(t) were all classified into these four
categories. Ten rules were proposed to relate measured D(t) and DD(t) with V(t).
Consequently, V(t) can be quantified using the if-then statement. An example is
given as follows:
If D(t)?PS and DD(t)?PL then V(t)?ZR [Rule 2 in Table 2 of Watano et al. (51)].
In this way, all the combinations of D(t) and DD(t) can be connected with V(t)
for the effective control of the process. The technique can be considered as highly
successful with the experimental justifications.
5.2. Modeling for Optimal Design, Operation, and Open-Loop
Optimal Control
Process optimization and open-loop optimal control of batch and continuous drum
granulation processes are described here as another important application example
of population balance modeling. Both steady-state and dynamic optimization studies
are carried out that consist of: (i) Construction of optimization and control relevant,
population balance models through the incorporation of moisture content, drum
rotation rate, and bed depth into the coalescence kernels; (ii) Investigation of optimal
operational conditions using constrained optimization techniques; and (iii)
Development of optimal control algorithms based on discretized population balance
equations. The objective of steady-state optimization is to minimize the recycle rate
with minimum cost for continuous processes. It has been identified that the drum
rotation rate, bed depth (material charge), and moisture content of solids are practical
decision (design) parameters for system optimization. The objective for the optimal
control of batch granulation processes is to maximize the mass of product-sized
particles with minimum time and binder consumption. The objective for the optimal
control of the continuous process is to drive the process from one steady state to
another in minimum time with minimum binder consumption, which is also known
as the state-driving problem. It has been known for some time that the binder sprayrate
is the most effective control (manipulative) variable. Although other process
variables, such as feed flow rate and additional powder flow rate can also be used
as manipulative variables, only the single input problem with the binder spay rate
as the manipulative variable is addressed here to demonstrate the methodology. It
can be shown from simulation results that the proposed models are suitable for control
and optimization studies, and the optimization algorithms connected with either
steady-state or dynamic models are successful for the determination of optimal
operational conditions and dynamic trajectories with good convergence properties.
It should be pointed out that only open-loop optimal control issues for granulation
processes without uncertainty are addressed here. The integration of openloop
optimal control with closed-loop, nonlinear model predictive control (NMPC)
for uncertain processes is reported elsewhere by the authors (49).
5.2.1. Statement of Optimization and Open-Loop Optimal Control Problems
are two operational strategies: (1) premix the fine particles with the proper amount
of liquid binder followed by the rotating operation until the desired size distribution
582 Cameron and Wang
A typical batch drum granulation process is schematically shown in Figure 8. There
© 2005 by Taylor & Francis Group, LLC
is achieved; and (2) simultaneous mixing and granulating by spraying liquid binder
(and fine powders in some cases) on the moving surface of particles inside the rotating
drum. The first strategy involves system optimization without any control action.
The optimization problem can be stated as: to determine the optimal moisture content,
initial size distribution, rotating rate, and bed depth (drum charge), such that
the desired size distribution can be obtained within a minimum time tf. Optimal control
techniques can be applied to the second strategy, which can be stated as: for the
specified initial conditions, maximize the mass of product-sized particles in minimum
time with minimum energy consumption by adjusting the manipulative variables,
such as binder spray rate and drum rotation speed. We discuss the optimal control
problem with the binder spray rate as the single manipulative variable in detail.
A slightly modified continuous drum granulation process with an additional
fine powder stream is shown in Figure 9. As mentioned previously, the additional
fine powder stream is used to improve the controllability of the process, which is
not seen in the conventional design. Our studies on continuous drum granulation
include the steady-state optimization and optimal state driving from one steady state
to another. The objective for steady-state optimization is to achieve minimum recycle
rate with minimum cost through the determination of optimal operational conditions,
such as rotating rate, binder spray rate, feed flow rate, bed depth, and drum
inclination angle. The optimal state driving attempts to drive the system from one
steady state to another in a minimum time with minimum energy consumption by
adjusting the time-dependent manipulative variables, such as binder spray rate, feed
flow rate, and optionally additional fine powder flow rate.
5.2.2. Optimization and Open-Loop Optimal Control Equations
The optimization and open-loop optimal control equations consist of model equations
and objective functions.
Figure 8 Schematic diagram of batch drum granulation.
Figure 9 Schematic diagram of continuous drum granulation.
Granulation Process Modeling 583
© 2005 by Taylor & Francis Group, LLC
5.2.2.1. Optimization and Control Relevant-Model Equations. The discretized
population balance equation for batch system can be described as follows:
d
dt
ni ? 
@
@L
Gni ? ??Bi  Di
i ? 1; 2; . . . ; imax
?5:13?
where ni, Bi, and Di stand for the particle number, birth rate, and death rate in the ith
size interval, respectively; and i?1,2, . . . , imax, in which imax is the total number of
size intervals. Similarly, continuous processes can also be represented as:
d
dt
ni ? 
@
@L
Gni ? ??Bi  Di ? Fin nin
i
nin
t  Fout ni
nt
i ? 1; 2; . . . ; imax
?5:14?
where F is the number flow -rate, the subscript t indicates the total value, and the
superscripts identify the inlet and outlet streams. Using Hounslow’s discretization
methods, the relevant terms in the right-hand side of Eqs. 5.13 and 5.14 are given by:
Bi ? ni1X
i2
j?1
2ji?1bi1; jnj 
 ?
i
2
bi1;i1n2
i1 ?5:15?
Di ? niX
i1
j?1
2jibi; jnj 
  niX
imax
j?1
bi; jnj 
  ?5:16?
@Gni
@L ? 
2G
?1 ? r?Li  r
r2  1
ni1 ? ni 
r
r2  1
ni?1

r ? Li?1=Li ? ffiffiffi 2 3 p ?5:17?
where bi,j is equivalent to the representation b(Li, Lj). Consequently, an original
population balance equation described by a partial differential-integral equation is
converted into a set of ordinary differential equations. It is more convenient to convert
the number-based population balance equations described by Eqs. 5.13–5.17 to
mass-based ones, which are demonstrated by the authors (48).
A control relevant model was developed by Zhang et al. (55), in which the coalescence
kernel is a function of the moisture content. In the newly developed kernel
models reported by Balliu (5) and Wang et al. (47,48), in addition to moisture content,
the bed depth and drum speed are also incorporated. Two kernel models,
namely size-independent kernel and size-dependent kernel, are used in optimization
and control simulations. The size-independent kernel is given by:
bi;j ? b0 ? a0  ?xm?n1ea1xm ? ?Bd ?n2ea2Bd   Sn3
d ea3Sd 
  ?5:18?
where xm is the moisture content in particles, Bd is the bed depth, Sd is the drum
rotating rate, a0–a3 and n1–n3 are constants determined through parameter identi-
fication techniques based on the measurement data. The size dependent kernel
584 Cameron and Wang
© 2005 by Taylor & Francis Group, LLC
is represented as (13):
bi;j ? b0
Li ? Lj 
 2
LiLj ?5:19?
where b0 is also defined in Eq. 5.18.
As the main mechanism determining the growth rate G in Eqs. 5.13 and 5.14
is the layering of the fine powders on the surface of particles, it can be deduced that
the growth rate is a strong function of the powder fraction and moisture content.
The following correlation is used to calculate the growth rate:
G ? Gm 
Mpowder
k PMi ?Mpowder  exp a?xw  xwc?2 h i ?5:20?
where Gm is the maximum growth rate, Mpowder is the mass of fine powder below
the lower bound of the particle classes, Mi is the mass of particles in the ith size class,
xwc is the critical moisture, k and a are fitting parameters. Studies on powder mass
balance lead to the following equations for batch processes:
dMpowder
dt ? Fin
powder  3G Z 1
0
M?L?
L
dL ?5:21?
and continuous processes:
dMpowder
dt ? Fin
powder 
Mpowder
tR  3G Z 1
0
M?L?
L
dL ?5:22?
where Fin
powder represents the flow rate of additional powder stream in both batch and
continuous cases. It can be used as an additional manipulative variable.
The liquid mass balance for batch processes is given by:
dxw
dt ?
1
Mt
Rw ?5:23?
where Mt is the total mass of solids in the drum and Rw is the binder spray rate.
Similarly, we can develop the liquid mass balance for the continuous process as:
dxw
dt ?
1
Mt
Fin
Mxin
w  FMxw ? Rw   ?5:24?
where and Fin
M FM are inlet and outlet mass flow rates, respectively, and Xin
w is the
moisture content in the feed solids.
In summary, the equations in the control relevant model for batch systems are
discretized population balance equations given by Eq. 5.13, powder dynamics
described by Eq. 5.21, and liquid dynamics represented by Eq. 5.23. The corresponding
equations for continuous processes are Eqs. 5.14, 5.22 and 5.24. Both cases share
the same kernel models given by Eqs. 5.18 and 5.19, and the growth-rate model
described by Eq. 5.20.
5.2.2.2. Objective Functions for System Optimization and Open-Loop Optimal
Control. The objective function for system optimization of batch granulation is:
Minimize
Sd ;Bd ;xw
J ? w1Mp tf 
  tf 
 
Subject to : Equation 5:13
?5:25?
Granulation Process Modeling 585
© 2005 by Taylor & Francis Group, LLC
The objective function for batch granulation with the binder spray rate as the
only manipulative variable is given by:
Mimimize
Rw
w1Mp tf 
 ? w2 Rtf
0 Rwdt
tf ( )
Subject to : Equations 5:13; 5:21; and 5:23
?5:26?
In Eqs. 5.25 and 5.26 , Mp is the mass of product sized particles, w1 and w2 are
weighting functions.
The objective function for steady-state optimization of continuous granulation
is:
Minimize
Sd ;Bd ;Fin ;Rw w1Fp ? w2Rw   Subject to :
Equations 5:14; 5:22; 5:24 with left hand sides replaced by zero ?5:27?
where Fp is the mass flow rate of product sized particles.
For the state-driving study, we carry out steady-state optimizations for
two different product specifications: the product range for steady state 1 (SS1) is
2.0–3.2 mm, whereas that for steady state 2 (SS2) is 3.2–5.0 mm. The objective function
for this optimal state-driving problem is described as:
Minimize
Rw
J ?X w1;i Mi tf 
 MSS2
i 
 2 h i? w2 Z tf
0
Rwdt ? w3tf 
 
Subject to :
Equations 5:14; 5:22; 5:24 and zero derivatives at final time
?5:28?
where Mi(tf) and Mi
ss2denote the mass of particles in the ith size interval at the final
time and for steady state 2, respectively.
5.2.3. Dynamic Optimization Algorithm
It is not difficult to solve the steady-state optimization problems with constraints
represented by algebraic equations by using commercial software packages. We
mainly explain the dynamic optimization methods used in this work. The basic struc-
In the dynamic optimization algorithm depicted in Figure 10, a control parameterization
technique (44) is used to discretize the originally continuous control
variables. That is, a control (manipulative) variable u(t) is represented by a set of
piece-wise constants, ui, i?1, 2, . . . , q. These constants are treated as parameters
to be determined by using dynamic optimization algorithms.
As the MATLAB software packages with Optimisation Toolbox provides both
effective ordinary differential equation (ODE) solvers as well as powerful optimization
algorithms, the dynamic simulations reported in this paper are carried out by
using the MATLAB Optimisation Toolbox (8).
5.2.4. Selected Simulation Results and Discussion
Simulations for both batch and continuous granulation processes are based on
a pilot plant drum granulator with the following parameters: length?2 m, diameter
?0.3 m, nominal hold up?40 kg, rotation rate?25–40 rpm, retention time
586 Cameron and Wang
ture of the algorithm employed in this paper is shown in Figure 10.
© 2005 by Taylor & Francis Group, LLC
range?6–10 min. Other process parameters are available in a recent paper by the
authors (48).
(a–c) with two data sets with and without constraints on control action. The control
constraints restrict lower and upper bounds on the control variables (lower
bound?0 kg/s, upper bound?0.015 kg/s), as well as the gradient of the control
actions (jdRwjdtj 	.0.0003 kg/s2). It can be seen from Figure 11d that if the normal
constraints on the control variable are replaced by a high upper bound of control
variable (0.036 kg/s) as the only constraint, very high spray rates at the early operating
stage with very short spray time leads to the minimum objective function given
by Eq. 5.26. However, if the normal constraints are activated, the control variable
moves smoothly rather than suddenly with the price of a longer operational time.
The difference between final times in the two cases is about 104 seconds (283–179 s),
which is quite significant. The results clearly have implications on equipment design
and specifications that could allow the constraints to be moved out, thus approaching
the best operating policy.
Through steady-state optimizations using the objective function described by
Eq. 5.27, optimal binder spray rates for two different specifications on product size
ranges are obtained. These are: Rw?0.050 kg/s for 2.0–3.2mm as the product size
range, and Rw
and 12b show the profiles using an optimal control policy and a constant spray rate
policy. The change of the cumulative mass between initial and final times under optimal
control policy is shown in Figure 12c. The control profiles are depicted in Figure
12d. The optimal control policy leads to about 50% reduction on the objective function
given by Eq. 5.28. The optimal spray policy can be stated as: ‘‘Gradually
Figure 10 Basic structure of the dynamic optimization algorithm.
Granulation Process Modeling 587
The simulated optimal profiles for the batch processes are shown in Figure 11
?0.075 kg/s for 3.2–5.0mm as the product size range. Figures 12a
© 2005 by Taylor & Francis Group, LLC
Figure 12 Optimal control of continuous drum granulation.
Figure 11 Optimal control of batch drum granulation.
588 Cameron and Wang
© 2005 by Taylor & Francis Group, LLC
increase the spray rate from the first steady state (0.005 kg/s) to achieve a relatively
high spray rate (0.0084 kg/s) followed by gradual reduction of the spray rate until
the spray rate of the second steady-state value (0.0075 kg/s) is reached, which will
cance of optimal control studies can be demonstrated by observing the fact that
the optimal profiles approach the second steady state faster, and the optimal control
strategy is easy to implement with smooth movement. It should be pointed out that
the small difference between two control policies shown in Figure 12 is because of
the small difference between two product specifications (product ranges from
2.0–3.2mm to 3.2–5.0). It can be predicted that if the two steady states are far away,
profound economic benefit can be achieved. Optimal control strategies are particularly
important to plant start up and shut down operations.
Figure 13 shows the dynamic profiles of optimal state driving from steady state
1 to steady state 2 with different levels of constraints. Dynamic changes of product
mass, undersized mass, and moisture content are shown in Figure 13 a, b, and c,
respectively, under two constraint levels. Figure 13d depicts control profiles for these
two cases. In addition to the constraints on control actions, the final time constraints
to ensure the final steady-state status is imposed on the system. That is, the left-hand
side of Eqs 5.14, 5.22, and 5.24 should be zero at the final time. However, it is not
necessary to achieve zero exactly for the derivatives at the final time. We normally
impose the final time constraints as jdx?tf ?=dtj < e in which x represents general
state variables such as the number of particles, mass of powder, and moisture
Figure 13 Effects of constraint tightness on optimal control of drum granulation.
Granulation Process Modeling 589
be maintained for the rest of the operational period.’’ From Figure 12, the signifi-
© 2005 by Taylor & Francis Group, LLC
content; and e is a very small positive number for practical applications with the
value depending on the tightness of constraints. The e values are chosen as 106
and 103
be shown from Figure 13 that the control strategy with loose constraints leads to
shorter operational time than that with tight constraints (1827 vs. 1925s). However,
the moisture dynamics shows severe offset and oscillation. In optimization simulations,
only final time constraints are changed for the two cases. It is interesting to
note that the program with tight constraints leads to small and smooth controller
movements even though the constraints on the control variable are not altered explicitly.
It seems that the loose constraints allow too much manipulative variation that
drives the system into a region (Xw  0.1) where moisture variations have significant
impact on the granulation performance. A marginal benefit identified by 5% time
reduction is achievable using loose constraints with a price of process oscillations.
Consequently, control strategy with tight final time constraints is superior to that
with loose constraints in this particular application.
Through an analysis on the simulation results, the following conclusions can be
drawn.
1. Population balance modeling provides an important basis for optimal
design and operations for both batch and continuous granulation processes.
2. The effects of liquid content, bed depth, and drum rotation rate on the coalescence
behavior can be quantified through the development of new kernel
models with the structure described by Eqs 5.18 and 5.19. The simulation
results are qualitatively consistent with industrial experience in large-scale
fertilizer production.
3. An optimal control strategy and algorithm using commercial optimization
software packages connected to reliable DAE/ODE solvers are successful
for the determination of optimal trajectories with good convergence properties.
This implies that under certain conditions, the more complicated
optimal control algorithms, such as that based on the well-known Pontryagin’s
maximum principle, could be avoided.
4. As start-up and shutdown operations are frequently encountered in granulation
plants with huge financial impacts, studies on optimal control strategies
can lead to significant economic benefits.
6. CONCLUSION
Granulation modeling is an area of growing importance. It is dominated by the
population balance approach for developing mechanistic models. However, it
requires an improved understanding of the key factors involved in particle growth
and breakage. This is currently improving. The growing importance of particulate
flow patterns is being addressed through approaches such as discrete element methods
(DEM), which will hopefully provide a microscale view of particle motions in the
granulation device. The challenge is in addressing the multiscale nature of granulation
modeling that spans from particle interactions up to the plant level.
The development of empirically based models has provided a simple means of
addressing quickly a number of control-related applications. This will continue to be
a useful approach for such problems.
590 Cameron and Wang
for tight and loose constraints indicated in Figure 13, respectively. It can
© 2005 by Taylor & Francis Group, LLC
Application of models to design, advanced control, and diagnosis will require
mechanistic models that continue to incorporate the latest understanding of the
underlying mechanisms. Much work is currently underway in these areas and the
incorporation into existing models of new knowledge will help extend the applicability
of process models for granulation.
REFERENCES
1. Adetayo AA, Litster JD, Pratsinis SE, Ennis BJ. Population balance modelling of drum
granulation of materials with wide size distribution. Powder Technol 1995; 82:37–49.
2. Adetayo AA, Ennis BJ. A unifying approach to modelling granulation processes coalescence
mechanisms. AIChE J 1997; 43(1):927–934.
3. Adetayo AA, Pottman M, Ogunnaike B. Effective control of a continuous granulation
process, Proc. Control of Particulate Processes IV, 23–28, Engineering Foundation,
NY, 1997.
4. Adetayo AA, Ennis BJ. A new approach to modelling granulation processes for simulation
and control purposes. 2000; 108:202–209.
5. Balliu N. An object oriented approach to the modelling and dynamics of granulation circuits,
PhD Thesis, School of Engineering, The University of Queensland, Australia 2004.
6. Bertoluzza S. A wavelet collocation method for the numerical solution of partial differential
equations. Appl Comput Harmonic Anal 1996; 3:1–9.
7. Biggs CA, Sanders C, Scott AC, Willemse AW, Hoffman AC, Instone T, Salman AD,
Hounslow MJ. Coupling granule properties and granulation rates in high-shear granulation.
Powder Technol 2003; 130:162–168.
8. Branch MA, Grace A. MATLAB Optimization Toolbox User’s Guide, The Math Works
Inc., Natick, 1996.
9. Cameron IT. Solution of Differential–Algebraic Systems using Diagonally Implicit
Runge-Kutta Methods. IMA J of Numerical Analysis 1983; 3:273–289.
10. Crowley TJ, Meadows ES, Kostoulas A, Doyle III FJ. Control of particle size distribution
described by a population balance model of semi-batch emulsion polymerisation.
J Process Control 2000; 10:419–432.
11. Ennis BJ, Litster JD. Size enlargement, in Perry’s Chemical Engineering Handbook, 7th
Ed., Chapter 8. McGraw-Hill, NY, 1997.
12. Faure A, York P, Rowe RC, Process control and scale-up of pharmaceutical wet granulation
processes: a review. European J Pharm Biopharm 2001; 52:269–277.
13. Friedlander SK, Smoke, Dust and Haze, 1st Ed., Wiley, NY, 1977; 2nd Ed., Oxford
University Press, NY, 2000.
14. Golovin AM. Sov. Phys. Dokl 1963; 8:191.
15. Hangos KM, Cameron IT. Process Modelling and Model Analysis, Academic Press,
London, ISBN 0–12–156931–4, 2001.
16. Hounslow MJ, Ryall RL, Marshall VR. A discrete population balance for nucleation,
growth and aggregation. AIChE J 1998; 34(11):1821–1832.
17. Immanuel CD, Cordeiro CF, Meadows ES, Crowley TJ, Doyle FJ. Modelling of particle
size distribution in immulsion co-polymerisation: Comparison with experimental data and
parametric sensitivity studies. Computers & Chemical Engineering 2002; 26:1133–1152.
18. Kapur PC, Fuerstenau DW. Coalescence model for granulation. Ind Eng Chem Pro Des
Dev. 1969; 8:56–62.
19. Kapur PC. Kinetics of granulation by non-random coalescence mechanism, Chem Eng
Sci 1972; 27:1863–1869.
20. Kaye BH, Powder Mixing, Chapman & Hall, London, 1997.
21. Kristensen HG, Schaefer T. Granulation: A review on pharmaceutical wet granulation.
Drug Dev Ind Pharm 1987; 13:803–872.
Granulation Process Modeling 591
© 2005 by Taylor & Francis Group, LLC
22. Kumar S, Ramkrishna D. On the solution of population balance equations by discretization
– I: A fixed pivot technique. Chem Eng Sci 1996; 51(8):1311–1332.
23. Leuenberger H. Moist agglomeration of pharmaceutical processes. In: Chulia D, Deleuil
M, Pourcelot Y, Eds. Powder Technology and Pharmaceutical Processes, Handbook of
Powder Technology. Elsevier: Amsterdam, 1994 9:337–389.
24. Litster JD. Scale-up of wet granulation processes: science not art, Powder Technol, 2003;
130:5–40.
25. Liu LX, Litster JD, Iveson SM, Ennis BJ. Coalescence of deformable granules in wet
granulation processes. AIChE J 2000; 46(3):529–539.
26. Liu LX, Litster JD. Population balance modelling of granulation with a physically based
coalescence kernel, Chem Eng Sci 2002; 57:2183–2191.
27. Liu Y, Cameron IT, Wang FY. The wavelet collocation method for transient problems
with steep gradients. Chem Eng Sci 2000; 55:1729–1734.
28. Liu Y, Cameron IT. A new wavelet-based method for the solution of the population
balance equation. Chem Eng Sci 2001; 56:5283–5294.
29. Liu Y, Cameron IT. A new wavelet-based adaptive method for solving population
balance equations. Powder Technol 2003; 130:181–188.
30. Ljung L. System Identification: Theory for the user. Prentice Hall, NJ: Upper Saddle
River, 1987.
31. Ljung L. System Identification Toolbox for Use with MATLAB. The Math Works,
Natick, MA, 2000.
32. Michaels JN. Toward rational design of powder processes. Powder Technol 2003; 138:
1–6.
32a. Miller SM, Rawlings JP. Model identification and control strategies for batch cooling
crystallisers. AIChE J 1994; 40(8):1312–1326.
33. Mort PR, Capeci SW, Holder JW. Control of agglomerate attributes in a continuous
binder-agglomeration process. Powder Technol 2001; 117:173–176.
34. Petzold L. A description of DASSL: A differential-algebraic system solver, Proc. IMACS
World Congress, Montreal, Canada, 1982.
35. Ramkrishna D. Population Balances: Theory and Applications to Particulate Systems in
Engineering. San Diego: Academic Press, 2000.
36. Randolph AD, Larson MA. Theory of Particulate Processes: Analysis and Techniques of
Continuous Crystallization, 2nd Ed., Academic Press, San Diego 1988.
37. Robertson GA, Cameron IT. "Analysis of Dynamic Process Models for Structural
Insight and Model Reduction–Part 1. Structural Identification Measures’’. Computers
& Chemical Engineering 1997; 21(5):455–473.
38. Salman AD, Fu J, Gorham DA, Hounslow MJ. Impact breakage of fertilizer granules.
Powder Technol 2002; 130:359–236.
39. Sanders CFW, Willemse AW, Salman AD, Hounslow MJ. Development of a predictive
high-shear granulation model. Powder Technology 2003; 138:18–24.
40. Sastry KVS. Similarity size distribution of agglomerates during their growth by coalescence
in granulation or green pelletization. Int J Mineral Process 1975; 2:187–203.
41. Smith M, Matsoukas T. Constant number Monte Carlo simulation of population
balances. Chem Eng Sci 1998; 53(9):1777–1786.
42. Spielman LA, Levenspiel O. A Monte Carlo treatment for reaction and coalescing
dispersed systems. Chem Eng Sci 1965; 20:247.
43. Tardos GI, Khan MI, Mort PR. Critical parameters and limiting condition in binder
granulation of fine powders. Powder Technol 1997; 94:245–258.
44. Teo KL, Goh CJ, Wong KH. A Unified Computational Approach for Optimal Control
Problems. Longman Scientific and Technical, New York, 1991.
45. van den Dries K, de Vegt O, Girard V, Vromans H. Granulae breakage phenomena in
a high shear mixer: influence of process and formulation variables and consequences
on granule homogeneity. Powder Technol 2003; 130:228–236.
592 Cameron and Wang
© 2005 by Taylor & Francis Group, LLC
46. Wang FY, Cameron IT. Review and future directions in the modelling and control of
continuous drum granulation. Powder Technol 2002; 124:238–253.
47. Wang FY, Ge XY, Balliu N, Cameron IT. Optimal control and operation of drum
granulation, Proc. 2nd International Conference on Population Balance Modelling
(PBM 2004). Valencia, Spain, May 2004.
48. Wang FY, Ge XY, Balliu N, Cameron IT, Optimal control and operation of drum granulation.
Chem Engi Sci In press (2005).
49. Wang FY, Cameron IT. Multi-level optimal control of particulate process with on-line
identification, to be presented at The 7th World Congress of Chemical Engineering
(WCCE 7), Glasgow, Scotland 10–14 July (2005).
50. Watano S, Numa T, Koizumi I, Osako Y. Feedback control in high shear granulation
of pharmaceutical powders. Eur J Pharm Biopharma 2001a; 52:337–345.
51. Watano S, Numa T, Miyanami K, Osako Y. A fuzzy control system of high shear
granulation using image processing. Powder Technol 2001b; 115:124–130.
52. Wauters PAL. Modelling and Mechanisms of Granulation, PhD thesis, The Delft
University of Technology The Netherlands, 2001.
53. Williams R, Burrage RK, Cameron IT, Kerr M. A four-stage index 2 diagonally implicit
Runge-Kutta method. Applied Numerical Mathematics 2002; 40:415–432.
54. Yekeler M, Ozkan A. Determination of the breakage and wetting parameters of calcite
and their correlations. Part Part Syst Charact 2002; 19:419–425.
55. Zhang J, Lister JD, Wang FY, Cameron IT. Evaluation of control strategies for fertiliser
granulation circuits using dynamic simulation. Powder Technol 2000; 108:122–129.
Granulation Process Modeling 593
© 2005 by Taylor & Francis Group, LLC
22
Regulatory Issues in Granulation
Prasad Kanneganti
Quality Operations, Pfizer Global Manufacturing, Pfizer Asia Pacific Pte. Ltd., Singapore
1. INTRODUCTION
The pharmaceutical industry is one of the most regulated consumer industries today.
Concerned with the safety of its citizens, governments across the world have set up regulations
that govern the manufacturing and distribution of finished pharmaceuticals for
human consumption. Like the industry, the regulations are also evolving to meet current
business needs and future technological challenges. With the widespread use of
the Internet, consumers have becomebetter informed about drugs, therapies, and health
risks. Current Good Manufacturing Practices (cGMPs) that were initially well established
within the pharmaceutical industry inUSAare now widely used across the globe.
Globalization of the pharmaceutical industry has taken place rapidly in recent times
fueled by mergers and acquisitions within the industry and also by economic, political,
and regulatory factors. Harmonization initiatives seeking to develop common regulatory
standards have become increasingly relevant and necessary.
2. PHARMACEUTICAL QUALITY MANAGEMENT
Quality management is that aspect of management function that establishes and
implements the quality policy formally authorized by senior management. The fundamental
elements of quality management are an appropriate infrastructure or quality
system and systematic actions known as quality assurance taken to ensure adequate
confidence that the product or service will satisfy established requirements for quality.
Thus, quality assurance is a management tool covering all matters that individually or
collectively influence the quality of a product. It incorporates cGMP as well as other
factors such as product design and development. Quality control is a subset of cGMP
and is concerned mainly with sampling, specifications, and testing of raw materials
and finished pharmaceutical products. The concepts of quality assurance, cGMP,
and quality control are thus interrelated aspects of quality management.
2.1. Current Good Manufacturing Practices
cGMP is that part of the quality assurance system which ensures that medicinal products
are consistently produced and controlled to the quality standards appropriate
595
© 2005 by Taylor & Francis Group, LLC
to their intended use and as required by the marketing authorization provided by the
regulatory agencies. The production of pharmaceutical products involves some risks,
e.g., cross-contamination, label mix-ups, etc., that cannot be prevented entirely
through end product testing and cGMPs diminish such risks. This quality assurance
element is mandated by law around the world for the manufacturing, storage, and
distribution of pharmaceuticals. Although the standards set by the United States
Food and Drug Administration (FDA) (1,2) are well recognized as an industry
benchmark, standards from other countries and regions are also gaining global
recognition (3,4).
One of the cGMP standards that is gaining worldwide recognition is that set up
by the Pharmaceutical Inspection Convention (PIC) and the Pharmaceutical Inspection
Co-operation Scheme (PIC Scheme) commonly known as PIC/S. The purpose
of PIC/S is to facilitate the networking between participating authorities and the
maintenance of mutual confidence, the exchange of information and experience in
the field of GMP and related areas, and the mutual training of GMP inspectors
(5). PIC/S became operational in November 1995 when the PIC Scheme commenced
operating in conjunction with PIC, which had already been operating since 1970.
The need to form the PIC Scheme became necessary when it was realized that
an incompatibility between PIC and European law did not permit individual EU
countries that were members of PIC to sign agreements with other countries seeking
to join PIC. Only the European Commission was permitted to sign agreements with
countries outside Europe, and the Commission itself was not a member of PIC.
Therefore, a less formal and more flexible cooperation scheme was developed to continue
and enhance the work of PIC. Unlike PIC, which is a legal treaty between
countries, the PIC Scheme is a cooperative arrangement between Health authorities.
2.2. International Conference on Harmonization
The International Conference on Harmonization (ICH) of technical requirements
for registration of pharmaceuticals for human use is a project that brings together
the regulatory authorities of Europe, Japan, and the United States and experts from
the pharmaceutical industry in these regions to discuss the scientific and technical
aspects of product registration. With its secretariat in Geneva, Switzerland, ICH
aims to achieve greater harmonization in the interpretation and application of technical
guidelines and product registration requirements thus avoiding unnecessary
duplication of testing carried out during the R&D phase for new medicines (6).
Guidelines issued by ICH are very useful reference documents for both industry
and regulatory bodies. The topics for these guidelines are divided into four major
categories, each with a specific topic code:

 Quality Topics (Q), e.g., Stability Testing (Q1), Impurity Testing (Q3)

 Safety Topics (S), e.g., Carcinogenicity Testing (S1), Genotoxicity Testing
(S2)

 Efficacy Topics (E), e.g., Dose Response Studies (E4), Good Clinical
Practices (E6)

 Multidisciplinary Topics (M), e.g., Medical Terminology (M1), The
Common Technical Document (M4).
There is no specific ICH guideline for granulation; however, guidelines such
as the Common Technical Document (CTD) highlight the requirements for specifi-
cations, testing, impurities, stability, and validation in drug product regulatory
596 Kanneganti
© 2005 by Taylor & Francis Group, LLC
submissions. In view of the wide international acceptance of these guidelines, it
would be prudent to check compliance with requirements specified in them while
putting together documentation dossiers to support regulatory filings or technology
transfers.
2.3. ISO 9000 Standards
The International Organization for Standardization (ISO) is the world’s largest
developer of standards. ISO’s principal activity is the development of technical
standards. These are very useful to the industry, regulatory bodies, trade officials,
suppliers and customers of products and services. With its Central Secretariat in
Geneva, Switzerland, ISO coordinates a network of the national standards institutes
of 148 countries (7). Each country is represented by one member that, unlike in the
case of the United Nations, need not be a delegation of the national government.
ISO has gained wide acceptance internationally as a commonly understood
baseline for quality, safety, and environment standards. It ensures fair play and
facilitates cross-border trade. ISO standards are voluntary and being a nongovernmental
organization, it has no legal authority to enforce their implementation. This
is an essential difference with cGMPs that have been legislated into law in several
countries.
One of the most popular standards is ISO 9000, which is a generic management
system standard that has become an international reference for quality requirements
in business-to-business dealings. The latest standard is the ISO 9000: 2000 series,
which was launched on December 15, 2000, and constitutes the most thorough overhaul
of this standard since it was first published in 1987. Businesses complying with
the older standards of ISO 9001 through 9003 were provided 3 years to make the
transition to the new version. ISO 9000: 2000 takes into account the developments
in the field of quality management and introduces eight quality management
principles on which the quality management system standards are based. Senior
management can use these principles as a framework to guide their organizations
toward improved performance. The eight principles are customer focus, leadership,
involvement of people, process approach, system approach to management, continual
improvement, factual approach to decision making, and mutually beneficial
supplier relationship.
Several ingredients in a drug product formulation may be common chemicals
also used in the food and cosmetic industries. For manufacturers of such chemicals,
if cGMP is not mandated by law, compliance with ISO 9000 is generally expected by
pharmaceutical manufacturers as part of their supplier management program.
3. POSTAPPROVAL CHANGE CONSIDERATIONS
Scale-up of manufacturing is required during transfer of production processes from
drug product development laboratories to commercial manufacturing centers or
between manufacturing centers. Much of the industry is undergoing mergers and
acquisitions that is leading to the globalization of manufacturing and supply chain.
To make manufacturing processes efficient, companies are grouping key products
into regional manufacturing centers. In the past, global trade was hindered by
regulatory and importation barriers. With the recent success of global initiatives
to improve trade, such barriers are eroding away, giving way to large-scale global
manufacturing facilities.
Regulatory Issues in Granulation 597
© 2005 by Taylor & Francis Group, LLC
During the manufacture of clinical batches, the amount of active ingredient
available is limited and the process equipment available are often scaled-down versions
of those used for the production of commercial batches. Batch sizes, thus,
are smaller than those used during the manufacture of routine commercial batches.
Process scale-up and commercial manufacturing are expedited as the industry
attempts to maximize the commercial benefits afforded by patent protection for
new drug molecules.Most companies formally record the scientific data that are generated
into product development reports. These form the basis for establishing the
manufacturing process, specifications, in-process controls, and validation acceptance
criteria used during commercial production of the drug product. Product development
reports also provide a link between the biobatch/clinical batch and commercial
process through development and scale-up. Information from the development
phase is used to prepare the chemistry, manufacturing, and controls (CMC) section
of an application such as a new drug application (NDA) filed with the FDA (8).
Where applicable, reference is also provided to other documents such as drug master
files (DMFs) submitted earlier to the FDA by the manufacturer or their vendors (9).
The FDA with input from the industry developed guidance for scale-up and
postapproval changes (SUPAC) for drug products. SUPAC covers components or
composition, site of manufacture, scale of manufacture, and manufacturing
process/equipment. These guidelines represent the agency’s current thinking on
the topic and are not binding on the industry or agency, with alternative approaches
being acceptable. The guidance documents have been found by the pharmaceutical
industry to enhance its ability to plan and implement change and manage resources
efficiently.
The SUPAC-IR guidance for immediate release solid oral dosage forms (10)
provides recommendations to sponsors of NDAs, abbreviated new drug applications
(ANDAs), and abbreviated antibiotic applications (AADAs) who intend, during the
postapproval period, to make changes. This guidance was the result of a workshop
on the scale-up of immediate release products conducted by the American Association
of Pharmaceutical Scientists (AAPS) in conjunction with the United States
Pharmacopoeial Convention (USP) and the FDA (11). It defines the levels of change,
recommended CMC tests for each level of change, in vitro dissolution tests, and in
vivo bioequivalence tests for each level of change and filing documentation that
Notification to FDA of postapproval changes to NDAs are made using change
documentation known as supplements (12). The regulations describe the type of
changes that require prior approval from the FDA before the change can be implemented
(preapprovable changes). Under some circumstances, changes can be made
before approval from FDA [changes being effected (CBEs)] or described in the
annual report to the FDA. In the case of CBE supplements, the FDA may, after
a review of the information submitted, decide that the changes are not approvable.
The SUPAC guidance documents list information that should be provided to the
FDA to ensure that product quality and the performance characteristics of the drug
products are not adversely affected by the changes proposed to be carried out.
3.1. Component and Composition Changes
The SUPAC guidance focuses on changes in excipients in the drug product. Changes
in the amount of drug substance are not addressed by this guidance. The changes are
598 Kanneganti
should support the change (Figures 1 and 2).
© 2005 by Taylor & Francis Group, LLC
categorized into three levels according to the increasing impact on product quality
and performance expected.
3.1.1. Level 1 Changes
Level 1 changes are those that are unlikely to have any detectable impact on formulation
quality and performance. Examples of such changes are deletion or partial
deletion of an ingredient intended to affect the color or flavor of the drug product,
changes in the composition of the printing ink to another approved ingredient, etc.
Changes in excipients, expressed as percentages (w/w) of total formulation, less than
additive effect of all excipients changes should not be more than 5%.
The documentation necessary to support this type of change are application/
compendial release requirements and stability data for one batch on long-term stability.
No in vivo bioequivalence data or additional dissolution data other than those
Figure 2 Format of SUPAC guidance documents.
Figure 1 Development of SUPAC guidance documents.
Regulatory Issues in Granulation 599
or equal to the percent ranges shown in Table 1 are also Level 1 changes. The total
© 2005 by Taylor & Francis Group, LLC
required by the application/compendia are necessary for this submission. The entire
documentation package including long-term stability data for the Level 1 change is
filed with the FDA through the annual report mechanism.
3.1.2. Level 2 Changes
Level 2 changes are those that could have a significant impact on formulation quality
and performance. The testing and filing requirements for Level 2 changes vary
depending on three factors—therapeutic range, solubility, and permeability. Therapeutic
ranges are defined as narrow or non-narrow and drug solubility and drug permeability
are defined as either low or high. A list of narrow therapeutic range drugs
is provided in the guidance document. Solubility is calculated based on the minimum
concentration of drug, milligram/milliliter (mg/mL), in the largest dosage strength,
determined in the physiological pH range (pH 1–8) and temperature (37 0.5C).
Permeability (Pe, cm/sec) is defined as the effective human jejunal wall permeability
of a drug and includes an apparent resistance to mass transport to the intestinal
membrane.
An example of a Level 2 change is change in the technical grade of an excipient,
e.g., Avicel PH102 vs. Avicel PH200. Changes in excipients, expressed as a percentage
(w/w) of the total formulation, greater than those listed earlier for Level 1
changes but less than or equal to a percent range representing a twofold increase
over Level 1 changes (Table 1) are also deemed as Level 2 changes. The total additive
effect of all excipients changes should not be more than 10%.
The documentation necessary to support this type of change are application/
compendial release requirements, batch records, and stability data for one batch
with 3 months’ accelerated stability data in supplement and one batch on long-term
stability. Dissolution data requirements depend on three scenarios known as cases
that cover a high/low permeability and a high/low drug solubility, as shown in
Table 1 SUPAC-IR: Component or Composition Change Levels
% Excipient (w/w of total dosage unit)
Excipient Level 1 Level 2 Level 3
Filler 5 10 >10
Disintegrant
Starch 3 6 >6
Other 1 2 >2
Binder 0.5 1 >1
Lubricant
Ca or Mg stearate 0.25 0.5 >0.5
Other 1 2 >2
Glidant
Talc 1 2 >2
Other 0.1 0.2 >0.2
Film coat 1 2 >2
Total drug excipient change (%) 5 10 N/A
Source: From Ref. 10.
600 Kanneganti
Table 2. No in vivo bioequivalence data are necessary for this submission if the
© 2005 by Taylor & Francis Group, LLC
situation falls within one of the cases shown in Table 2. A prior approval supplement
that contains all information including accelerated stability data is to be filed. The
long-term stability data are filed through the annual report mechanism.
3.1.3. Level 3 Changes
Level 3 changes are those that are likely to have a significant impact on formulation
quality and performance. Similar to Level 2 changes, the testing and filing documentation
requirements vary depending on therapeutic range, solubility, and permeability.
Examples of Level 3 changes are any qualitative and quantitative excipients
change to a narrow therapeutic drug beyond the ranges stated for Level 1 changes
(Table 2).
A change in granulating solution volume is not covered under SUPAC-IR as it
is a minor change to a normal operating procedure and should be included in the
batch record after undergoing validation and manufacturer’s site change control
procedure. A change in the granulating solvent, e.g., from alcohol to water, can
be expected to alter the composition of the drug product even though it may be
removed during manufacturing and hence it is a Level 3 change that requires a prior
approval supplement.
The documentation required to support Level 3 changes are application/
compendial release requirements and batch records. If a significant body of information
is available, one batch with 3 months’ accelerated stability data is to be included
in the supplement and one batch on long-term stability data reported in the annual
report. Where a significant body of information is not available, up to three batches
with 3 months’ accelerated stability data are to be included in the supplement and
one batch on long-term stability data reported in the annual report.
Dissolution data requirements for Level 3 changes are as specified for Case B in
Table 2. In addition, a complete in vivo bioequivalence study is required. This study
may be waived if an acceptable in vivo/in vitro correlation has been verified. A prior
approval supplement that contains all information including accelerated stability
data is to be filed.
Table 2 SUPAC-IR: Dissolution Testing Categories
Category Nature of drug
Dissolution
medium
Time points
(min) Specification
Case A High permeability,
high solubility
0.1N HCl 15 85%
Case B Low permeability,
high solubility
As stated in
application/
compendia
15, 30, 45, 60,
120, or until
asymptote is
reached
Dissolution profile
similar to
current
formulation
Case C High permeability,
low solubility
Water, 0.1N HCl,
USP buffer media
at pH 4.5, 6.5, and
7.5 (plus
surfactant if
justified)
15, 30, 45,
60, 120
90% or asymptote
is reached;
profile similar to
current product
Source: From Ref. 10.
Regulatory Issues in Granulation 601
(Table 1) and all drugs not meeting the dissolution criteria listed for Level 2 changes
© 2005 by Taylor & Francis Group, LLC
3.2. Site Changes
Site changes consist of changes in location of the site of manufacture for both
company-owned and contract manufacturing facilities. These do not include any
scale-up changes, changes in manufacturing process or equipment, or changes in
components or composition. The new manufacturing locations are expected to have
a satisfactory cGMP inspection. Similar to those for component and composition
changes, site changes are also categorized into three different levels that require
differing depth of test and filing documentation. Level 1 changes are site changes
within a single facility while Level 2 changes are site changes within the same campus.
Level 3 changes consist of a change in manufacturing site to a different campus, i.e.,
the facilities are not on the same original contiguous site or in adjacent city blocks.
3.3. Changes in Batch Size
Postapproval changes in the size of a batch (scale-up/scale-down) from the pivotal/
pilot scale biobatch material to a larger or smaller production batch require additional
information to be submitted with the change application. Scale-down below
100,000 dosage units is not covered by the SUPAC guidance. All scale-up changes
are required to undergo suitable process validation and regulatory inspection. There
are two levels of batch size changes that cover batch size increases up to and including
a factor of 10 times the size of the pilot/biobatch and increases beyond a factor
of 10 times, respectively (Table 3).
3.4. Manufacturing Equipment/Process Changes
Equipment changes consist of changes from nonautomated or nonmechanical equipment
to automated or mechanical equipment or changes to alternative equipment
either of the same or different design and operating principles or of a different capacity.
Process changes include changes such as mixing time and operating speeds
either within or outside application/validation ranges. A change in the process used
in the manufacture of the drug product, e.g., change from wet granulation to direct
compression is also included. Table 3 provides a summary of the documentation
requirements to file changes to manufacturing equipment and process.
3.5. Modified Release Solid Dosage Form
Modified release solid dosage forms include both delayed and extended release drug
products. Delayed release is the release of a drug (or drugs) at a time other than
immediately following oral administration. Extended release products, on the other
hand, are formulated to make the drug available over an extended period after ingestion
so that a reduction in the dosing frequency compared to an immediate release
dosage form is achieved.
Following the successful release of the guidance document for immediate
release solid oral dosage forms (SUPAC-IR), the FDA issued a specific guidance,
the SUPAC-MR, for scale-up and postapproval changes affecting modified release
solid dosage forms (13,14) in 1997. This guidance covers postapproval changes for
modified release solid oral dosage forms that affect components and composition,
scale-up/scale-down, site change, and manufacturing process or equipment changes.
602 Kanneganti
These requirements are summarized in Table 3.
© 2005 by Taylor & Francis Group, LLC
It permits less burdensome notice of certain postapproval changes within the meaning
of 21CFR 314.70.
In the case of components and composition, SUPAC-MR covers changes in
non-release controlling excipients and release controlling excipients separately. The
criticality of excipients to drug release is to be established and appropriate justifi-
cations provided if an excipient is claimed as a non-release controlling excipient in
the formulation of the modified release solid dosage form. The change level classi-
fication, therapeutic range, test and filing documentation for components and
composition changes, site changes, changes in batch size (scale-up/scale-down),
manufacturing equipment changes, and manufacturing process changes for extended
release solid dosage forms and delayed release solid dosage forms are summarized in
this guidance document.
3.6. Changes to Granulation Equipment
The FDA released in January 1999 another guidance document that specifically
addressed documentation requirements for filings addressing changes to pharmaceutical
manufacturing equipment (15). This is the manufacturing equipment addendum
developed with the assistance of the International Society of Pharmaceutical Engineering
(ISPE) and is used in conjunction with the SUPAC-IR and SUPAC-MR guidance
documents. It includes a representative list of equipment commonly used in
the industry but does not include equipment modified by a manufacturer to meet
specific needs. Definitions and classification for broad categories of unit operations
such as blending and mixing, drying, particle size reduction/separation, granulation,
unit dosage, coating, printing, and soft gelatin encapsulation are provided. For each
unit operation, a table categorizing process equipment by class (operating principle)
and subclass (design characteristics) along with examples of commercially available
equipment is presented.
Granulation is defined as the process of creating granules either by using a
liquid that causes particles to bind through capillary forces or by dry compaction
forces. Granulation is stated to impact on one or more of the powder properties such
as enhanced flow; increased compressibility; densification; alteration of physical
appearance to attain more spherical, uniform, or larger particles; and or enhanced
hydrophilic surface properties.
The operating principles listed in the SUPAC manufacturing equipment
addendum (15) are
1. Dry granulation: Dry powder densification and/or agglomeration by direct
physical compaction.
2. Wet high-shear granulation: Powder densification and/or agglomeration
by the incorporation of a granulation fluid into the powder with
high-power-per-unit mass, through rotating high-shear forces.
3. Wet low-shear granulation: Powder densification and/or agglomeration
by the incorporation of a granulation fluid into the powder with lowpower-
per-unit mass, through rotating low-shear forces.
4. Low-shear tumble granulation: Powder densification and/or agglomeration
by the incorporation of a granulation fluid into the powder with lowpower-
per-unit mass, through rotation of the container vessel and/or
intensifier bar.
Regulatory Issues in Granulation 603
© 2005 by Taylor & Francis Group, LLC
Table 3 SUPAC-IR: Site Equipment and Process Change Requirements by Category
Documentation
Type/level Change permitted Exclusions Chemistry Dissolution Bioequivalence Filing
Component/composition
Level 1 Table 1: total change
5%
No change beyond
approved target ranges
LTSSa commitment Application/
compendial only
None Annual report
Level 2 Table 1: total change
10%
No narrow therapeutic
range drugs
Accelerated stability data
plus LTSS commitment
Varies (Table 2) None Prior approval
supplement
Level 3 Table 1 None Accelerated stability data
plus LTSS commitment
Case B (Table 2) Full Prior approval
supplement
Site change
Level 1 Single facility No scale or process changes None Application/
compendial only
None Annual report
Level 2 Contiguous campus No scale or process changes None Application/
compendial only
None CBE
supplement
Level 3 Different campus No scale or process changes Accelerated stability data
and LTSS commitment
Case B (Table 2) None CBE
supplement
Scale-up/scale-down
Level 1 10-fold increase in
batch size
No change in site, controls,
or equipment
LTSS commitment Application/
compendial only
None Annual report
Level 2 >10-fold increase in
batch size
No change in site, controls,
or equipment
Accelerated stability data
and LTSS commitment
Case B (Table 2) None CBE
supplement
604 Kanneganti
© 2005 by Taylor & Francis Group, LLC
Manufacturing equipment
Level 1 Non-to-automated/
non-to-mechanical;
new equipment
design w/wo same
capacity
No change in operating
principle
LTSS commitment Application/
compendial only
None Annual report
Level 2 New design or
operating principle
None Accelerated stability data
and LTSS commitment
Case C (Table 2) None Prior approval
supplement
with change
justification
Manufacturing process
Level 1 Operating within
validation ranges
None None Application/
compendial only
None Annual report
Level 2 Operating outside
validation ranges
None LTSS commitment Case B (Table 2) None CBE
supplement
Level 3 New process
(e.g., wet-to-dry
granulation)
None Accelerated stability data
and LTSS commitment
Case B (Table 2) Full Prior approval
supplement
with change
justification
aLTSS, long-term stability study.
Source: From Ref. 10.
Regulatory Issues in Granulation 605
© 2005 by Taylor & Francis Group, LLC
5. Extrusion granulation: Plasticization of solids or wetted mass of solids and
granulation fluid with linear shear through a sized orifice using a pressure
gradient.
6. Rotary granulation: Spheronization, agglomeration, and/or densification
of a wetted or nonwetted powder or extruded material. This is accomplished
by centrifugal or rotational forces from a central rotating disk,
rotating walls, or both. The process may include the incorporation and/
or drying of a granulation fluid.
7. Fluid bed granulation: Powder densification and/or agglomeration with little
or no shear by direct granulation fluid atomization and impingement on
solids, while suspended by a controlled gas stream, with simultaneous drying.
8. Spray dry granulation: A pumpable granulating liquid containing solids (in
solution or suspension) is atomized in a drying chamber and rapidly dried
by a controlled gas stream, producing a dry powder.
The classification of granulation equipment in the SUPAC manufacturing
equipment addendum (15) is as follows:
1. Dry granulator: Dry granulator subclasses primarily are distinguished by
the densification force application mechanism:

 Slugging

 Roller compaction
2. Wet high-shear granulator: Wet high-shear granulator subclasses primarily
are distinguished by the geometric positioning of the primary impellers;
impellers can be top, bottom, or side driven.

 Vertical (top or bottom driven)

 Horizontal (side driven)
3. Wet low-shear granulator: Wet low-shear granulator subclasses primarily
are distinguished by the geometry and design of the shear inducing components;
shear can be induced by rotating impeller, reciprocal kneading
action, or convection screw action.

 Planetary

 Kneading

 Screw
4. Low-shear tumble granulator: Although low-shear tumble granulators may
differ from one another in vessel geometry and type of dispersion or intensifier
bar, no low-shear tumble granulator subclasses have been identified.
5. Extrusion granulator: Extrusion granulator subclasses primarily are
distinguished by the orientation of extrusion surfaces and driving pressure
production mechanism:

 Radial or basket

 Axial

 Ram

 Roller, gear, or pelletizer
6. Rotary granulator: Rotary granulator subclasses primarily are distinguished
by their structural architecture. They have either open top architecture,
606 Kanneganti
© 2005 by Taylor & Francis Group, LLC
such as a vertical centrifugal spheronizer, or closed top architecture, such as
a closed top fluid bed dryer:

 Open

 Closed
7. Fluid bed granulator: Although fluid bed granulators may differ from one
another in geometry, operating pressures, and other conditions, no fluid
bed granulator subclasses have been identified.
8. Spray dry granulator: Although spray dry granulators may differ from one
another in geometry, operating pressures, and other conditions, no spray
dry granulator subclasses have been identified.
ment within the same class or subclass would be considered to have the same design
and operating principle under SUPAC-IR and SUPAC-MR. As an example, a
change from one type of wet high-shear granulator (e.g., vertical type from manufacturer
A) to another type of wet high-shear granulator (e.g., vertical type from
manufacturer B) generally would not represent a change in operating principle
and would, therefore, be considered to be the same under either SUAPC-IR or
SUPAC-MR.
A change from equipment in one class to equipment in a different class would
usually be considered a change in design and operating principle. Thus, a change
from a wet high-shear granulator to a fluid bed granulator demonstrates a change
in the operating principle from powder densification by wet agglomeration using
high shear to powder densification with little or no shear. Such a change would be
considered to be different under either SUPAC-IR or SUPAC-MR.
The FDA advises change applicants to carefully consider and evaluate on a
case-by-case basis changes in equipment that are in the same class, but different
subclass. For example, a change from a horizontal (side driven) wet high-shear granulator
to a vertical (top or bottom driven) wet high-shear granulator represents a
change within a class and between subclasses. This change would not require a preapproval
supplement provided the manufacturing process with the new equipment is
validated. The data and rationale used to make this determination can be reviewed
by the FDA at its discretion. In the event a single piece of equipment is capable of
performing multiple discrete unit operations, e.g., mixing, granulation, drying, etc.,
the unit is evaluated solely for its ability to granulate.
3.7. International Change Notification
The manufacturing process and equipment change notification outside the United
States varies from region to region. A brief description of the manufacturing process
is required as part of the filing requirements for marketing a drug product. Some
countries require master batch records to be filed but most others do not require
much detail. In addition, a site master file that provides information on the production
and control of the manufacturing operations including major process equipment
at the site is sometimes required (16).
Regulatory agencies in countries that form the European Community (EC)
have adopted a common approach to the procedures for variations to the terms of
a marketing authorization. Variations can be by notification such as Type IA and
Type IB that are categorized as minor variations that fulfill the conditions set forth
Regulatory Issues in Granulation 607
Table 4 shows a listing of granulation equipment classes and subclasses. Equip-
© 2005 by Taylor & Francis Group, LLC
Table 4 Unit Operation—Granulation
Class Subclass Examples
Dry granulator Slugging Various
Roller compaction Alexanderwerk
Bepex (Hosokawa)
Fitzpatrick
Freund
Vector
Wet high-shear granulator Horizontal (side driven) Littleford Day
Lodige
Processall
Vertical
(top or bottom driven)
Aeromatic-Fielder (GEA-Niro)
APV
Baker-Perkins
L.B. Bohle
Dierks & Shone
Diosna (Fluid Air)
GEI-Collette (GEI International)
Key International
Littleford Day
Lodige
Powrex (Glatt)
Processall
Werner & Pfeiderer
Zanchetta (Romaco)
Wet low shear granulator Planetary Aaron
Aeschbach
AMF
GEI-Collette (GEI International)
Hobart
Jaygo
Littleford Day
Ross
Vrieco
Kneading Aaron
Paul O. Abbe
Custom Metal Craft
Dynamic Air
Jaygo
Kemutec
Littleford Day
Processall
Ross
Sigma
Teledyne Readco
Screw Vrieco-Nauta (Hosokawa)
Low-shear tumble
granulator
Slant cone, or double cone,
or V-blender
Paul O. Abbe
Gemco
Patterson-Kelley
Extrusion granulator Radial or basket Alexanderwerk
GEA Niro
(Continued)
608 Kanneganti
© 2005 by Taylor & Francis Group, LLC
in an annex to the EC regulations (17). A major variation of Type II is a variation
which cannot be deemed to be a minor variation or an extension of the marketing
authorization and requires prior approval. A downscaling of batch size by 10 times
or an increase that results in up to 10 times the original batch size approved at the
grant of the marketing authorization is a Type IA change (18). All other batch size
decreases/increases or minor changes in the manufacturing process for the finished
product require a Type IB filing. A minor change to the manufacture is one where
the overall manufacturing principle remains the same and the new process leads to
an identical product regarding all aspects of quality, safety and efficacy.
4. VALIDATION OF GRANULATION PROCESSES
Validation is defined by the FDA as establishing documented evidence which provides
a high degree of assurance that a specific process will consistently produce a
product meeting its predetermined specifications and quality attributes (19). Process
Table 4 Unit Operation—Granulation (Continued )
Class Subclass Examples
LCI
Luwa
Ross
Axial Bepex (Hosokawa)
Gabler
LCI
Ram LCI
Roller, gear, or pelletizer Alexanderwerk
Bepex (Hosokawa)
Rotary granulator Open Freund (Vector)
GEA Niro
LCI
Luwa
Closed Aeromatic-Fielder (GEA Niro)
Glatt
LCI
Luwa
Fluid bed granulator None identified Aeromatic-Fielder (GEA Niro)
APV
BWI Hu? ttlin (Thomas Engineering)
Diosna
Fitzpatrick
Fluid Air
Glatt
Heinen
Vector
Spray dry granulator None identified Allgaier
GEA Niro
Glatt
Heinen
Source: Ref. 15.
Regulatory Issues in Granulation 609
© 2005 by Taylor & Francis Group, LLC
validation is required both in general and specific terms by cGMPs for finished
pharmaceuticals—21 CFR Parts 210 and 211. The WHO defines validation as the
collection and evaluation of data, beginning at the process development stage and
continuing through the production phase, which ensure that the manufacturing
processes, including equipment, buildings, personnel, and materials, are capable of
achieving the intended results on a consistent and continuous basis (20).
For a manufacturing facility, process knowledge is provided through technology
transfer dossiers. Granulation is a critical process step that has a direct impact
on the quality of the drug product manufactured and hence requires validation. The
overall validation activity at a manufacturing facility is detailed in a document
known as the validation master plan (VMP). The validation of the granulation process
is described in the VMP. Critical process parameters for granulation such as the
rate and amount of granulation fluid added, impeller and chopper speed, and mixing
time are identified and in-process controls such as moisture content and granulation
end point measurement are established during the product development phase.
4.1. Equipment/Utilities Qualification
The qualification of the manufacturing equipment and control instrumentation is a
prerequisite to the validation of the granulation process. Critical utilities such as
purified water, compressed air, gaseous nitrogen, etc., required for granulation are
also validated to ensure that they meet the required quality specification at the point
of delivery to the granulation equipment.
The qualification of granulation equipment is carried out sequentially beginning
with design qualification (DQ) followed by installation qualification (IQ) and
on the effort put into its design and DQ provides evidence that quality is built into
the design of the equipment. Quite often a design rationale instead of a DQ is prepared.
This document addresses why a specific piece of equipment was chosen highlighting
its quality and safety considerations and provides evidence of the assessment
carried out to judge its suitability for the manufacturing of the drug product.
IQ provides documented evidence that the equipment is installed as designed
and specified and correctly interfaced with other systems such as electrical supply
and utilities. During this phase of qualification, equipment manuals/drawings, specifications,
manufacturers test records, etc., together with installation documents and
‘‘as-built’’ drawings are compiled and verified. Calibration of instrumentation and
maintenance checks are also established. OQ is a documented demonstration that
the process equipment as installed operates well. At this stage, generally a manufacturing
process simulation is carried out using a placebo formulation instead of the
actual drug product recipe. For each qualification phase, a protocol detailing the
activity and acceptance criteria is prepared. At the conclusion of the testing activity,
a summary report that discusses the results and the readiness to proceed to the next
phase of qualification is issued.
4.2. Performance Qualification
Performance Qualification (PQ) is a documented program that demonstrates that the
granulation process when carried out within defined parameters will consistently perform
its intended function to meet its pre-established acceptance criteria. Thus, PQ is
610 Kanneganti
operational qualification (OQ) (Fig. 3). The quality of process equipment depends
© 2005 by Taylor & Francis Group, LLC
dynamic testing that combines the equipment, utilities, and manufacturing process to
produce the product under routine operational conditions.
Product specifications that become the basis for the acceptance criteria at PQ
stage are established during the development of the process with the biobatch or
pivotal clinical batch serving as the reference batch. Prospective validation of the
granulation process is generally carried out for new products and the data included
in regulatory submissions, if necessary. The norm is to manufacture at least three
consecutive PQ batches; however, a process capability study can establish the actual
number of batches required based on the natural variability of a process (21). Revalidation
may be required after changes that significantly impact product quality are
made or on a periodic basis at scheduled intervals.
A FDA field inspection guide for validation of oral solid dosage forms lists
granulation/mix analysis as a major area for investigation (22). It discusses various
types of mixers and granulation equipment and highlights their design features as
well as problems associated with their efficiency and validation. Blending validation
and content uniformity failures due to poor mixing are of chief concern for most
fluid bed dryer is superior to the oven dryer as it yields a more uniform granulation
with spherical particles.
4.3. Computer Validation
Granulation equipment is supported by computer control systems that are getting
increasingly sophisticated. Most commercial equipment has programmable logic
controllers or embedded microprocessors. International forums with representation
from users and the vendors of equipment and software have been set up to address
the software life cycle documentation requirements. GAMP 4 is a globally accepted
Figure 3 Documentation hierarchy for pharmaceutical process validation.
Regulatory Issues in Granulation 611
conventional mixers (Table 5). This guide also compares dryers and notes that the
© 2005 by Taylor & Francis Group, LLC
guidance document developed by ISPE and the GAMP Forum to address computer
validation (23). The PIC/S Guide to Good Practices for computerized systems in
regulated ‘‘GXP’’ environments developed by international regulatory agencies is
also a useful reference document for manufacturers and other users (24).
Electronic records and electronic signatures that have cGMP implications are
generally expected by regulatory agencies to be equivalent to paper records and
handwritten signatures executed on paper (25). A new guidance that represents
FDA’s current thinking on electronic records and electronic signatures was released
in August 2003 (26). The agency took a narrower interpretation of the requirements
stated in 21 CFR Part 11 following feedback from the pharmaceutical industry and
vendors that the regulations could stifle technological advances by restricting the use
of electronic technology and increasing the cost of compliance. PIC/S also requires
the regulated user to validate the system for storage of the information electronically
for the required time and to ensure that the data are protected from damage or loss
and can be easily retrieved in a legible form (24).
Table 5 Typical Problems Associated with Mixing Equipment
Mixer Type Design Feature Limitations/problems
Planetary (pony pan) Open pan/pot Dusty operation
Horizontal blending Cross-contamination problem
Poor vertical mixing
Segregation or nonmixing of
components
Difficult to validate
Ribbon blender Top loading Moderately dusty operation
Horizontal and vertical
blending
Discharge valve
Blade clearance
Cross-contamination problem
‘‘Dead spot/zone’’ at the discharge
valve
Poor mixing at ends of the center
horizontal mixing bar and shell
wall
Cleaning problems with seals/
packing
Risk of overfill leading to poor
mixing
Tumble blender Twin shell/double cone Mild mixing action
Mild mixing action Powder lumps will not break up
Low humidity results in static
charge buildup
High humidity leads to lumping
High shear High-energy chopper Different mixing time compared to
conventional mixers
Drug substance may partially
dissolve or recrystallize
Charring due to heat generation
Cleaning requires disassembly of
chopper
Source: From Ref. 22.
612 Kanneganti
© 2005 by Taylor & Francis Group, LLC
5. CONCLUSION
The globalization of pharmaceutical manufacturing is proceeding rapidly, driven by
mergers and acquisitions within the industry and the freeing up of cross-border trade.
In addition to known global quality standards such as ISO 9000, harmonization of
cGMPs and regulatory filing requirements is making progress driven by ICH and
regional forums such as PIC/S. Simplification of existing regulations for notification
of postapproval changes to regulatory filings has also been achieved through workshops
representing regulatory bodies such as the FDA, the pharmaceutical industry,
and academia. Such guidance has been found to be very useful by manufacturers as it
gives them greater flexibility of operations, cost savings, and efficient management of
resources. The manufacturer is however, expected to carry out a thorough scientific
review to evaluate the impact of all changes on product quality and performance.
Granulation is a critical process step and hence requires to be validated as part of
the overall validation of the manufacturing process. Computer validation becomes
increasingly relevant with the use of control systems in granulation equipment,
electronic records, and electronic signatures. Process engineers and product development
researchers today require a sound understanding of regulations governing drug
product approval, validation, and change management.
REFERENCES
1. 21 CFR Part 210. Current Good Manufacturing Practice in Manufacturing, Processing,
Packing, or Holding of Drugs; General.
2. 21 CFR Part 211. Current Good Manufacturing Practice for Finished Pharmaceuticals.
3. PIC/S Guide to Good Manufacturing Practice for Medicinal Products (PE 009-1), Sep 1,
2003.
4. WHO expert committee on specifications for pharmaceutical preparations, 32nd report.
Good Manufacturing Practices for Pharmaceutical Products (Annex 1). Geneva: World
Health Organization, 1992.
5.
6.
7.
8. FDA Guideline for the Format and Content of the Chemistry, Manufacturing, and Controls
Section of an Application, Feb 1987.
9. FDA Guideline for Drug Master Files, Sep 1989.
10. FDA Guidance for Industry: Scale-up and Postapproval Changes for Immediate Release
Solid Oral Dosage Forms, Nov 1995.
11. Skelly JP, et al. Workshop report: scale up of immediate release oral solid dosage forms.
Pharm Res 1993; 10(2):313–316.
12. 21 CFR Part 314.70. Supplements and Other Changes to an Approved Application.
13. FDA Guidance for Industry: Scale-up and Postapproval Changes for Modified Release
Solid Oral Dosage Forms, Sep 1997.
14. Skelly JP, et al. Workshop report: scale-up of oral extended-release dosage forms. Pharm
Res 1993; 10(12):1800–1805.
15. FDA Guidance for Industry: Scale-up and Postapproval Changes for Immediate Release
and Modified Release Solid Oral Dosage Forms (Manufacturing Equipment Addendum),
Jan 1999.
16. Health Sciences Authority (Singapore). Guidance Notes on Preparation of a Site Master
File, May 1999.
17. European Commission regulation (EC) No. 1084/2003, Jun 3, 2003.
Regulatory Issues in Granulation 613
Pharmaceutical Inspection Co-operation Scheme (PIC/S). http://www.picscheme.org.
International Conference on Harmonization (ICH). http://www.ich.org.
International Organization for Standardization (ISO). http://www.iso.ch.
© 2005 by Taylor & Francis Group, LLC
18. Guideline on Dossier Requirements for Type IA and Type IB Notifications, European
Commission, Jul 2003.
19. FDA Guideline on General Principles of Process Validation, May 1987.
20. WHO expert committee on specifications for pharmaceutical preparations, 34th report.
Good Manufacturing Practices: Guidelines on Validation of Manufacturing Processes
(Annex 6). Geneva: World Health Organization, 1996.
21. Kieffer R, Torbeck L. Validation and process capability. Pharm Technol 1998:66–76.
22. FDA Guide to Inspections of Oral Solid Dosage Forms Pre/Post Approval Issues for
Development and Validation, Jan 1994.
23. Good Automated Manufacturing Practice Guide (GAMP 4). International Society of
24. PIC/S Guide to Good Practices for Computerized Systems in Regulated ‘‘GXP’’ Environments
(PE 011-1), Aug 20, 2003.
25. 21 CFR Part 11. Electronic Records; Electronic Signatures.
26. FDA Guidance for Industry on Part 11, Electronic Records; Electronic Signatures—
Scope and Application, Aug 2003.
614 Kanneganti
Pharmaceutical Engineering (ISPE), Dec 2001. http://www.ispe.org.
© 2005 by Taylor & Francis Group, LLC