Chemistry - Membrane Separation Systems

MEMBRANE
SEPARATION
SYSTEMS
Recent Developments and
Future Directions
R.W. Baker, E.L. Cussler, W. Eykamp,

W.J. Koros, R.L. Riley, H. Strathmann
NOYES DATA CORPORATION

Park Ridge, New Jersey, U.S.A.
Copyright @ 1991 by Noyes Data Corporation
Library of Congress Catalog Card Number: 90.23675
ISBN: O-8155-1270-8
Printed in the United States
Published in the United States of America by

Noyes Data Corporation
Mill Road, Park Ridge, New Jersey 07656
10987654321
Library of Congress Cataloging-in-Publication Data
Membrane separation systems : recent developments and future

directions I by R.W. Baker . . . [et al.1 .
p. cm.
Includes bibliographical references and index.
ISBN O-8155-1270-8 :
1. Membrane separation. I. Baker, R.W. (Richard W.)
TP248.25.M46M456 1991
660’2842.-dc20 9923675
CIP
Acknowledgments
This report was prepared by the following group of experts:
Dr. Richard W. Baker (Membrane Technology & Research, Inc.)

Dr. Edward Cussler (University of Minnesota)
Dr. William Eykamp (University of California at Berkeley)
Dr. William J. Koros (University of Texas at Austin)
Mr. Robert L. Riley (Separation Systems Technology, Inc.)
Dr. Heiner Strathmann (Fraunhofer Institute, West Germany).
Mrs. Janet Farrant and Dr. Amulya Athayde edited the report, and also served as
project coordinators.
The following members of the Department of Energy (DOE) made valuable

contributions to the group meetings and expert workshops:
Dr. Richard Gordon (Office of Energy Research, Division of Chemical Sciences)

Dr. Gilbert Jackson (Office of Program Analysis)
Mr. Robert Rader (Office of Program Analysis)
Dr. William Sonnett (Office of Industrial Programs)
The following individuals, among others, contributed to the discussions and

recommendations at the expert workshops:
Dr. B. Bikson (Innovative Membrane Systems/Union Carbide Corp.)

Dr. L. Costa (Ionics, Inc.)
Dr. T. Davis (Graver Water, Inc.)
Dr. D. Elyanow (Ionics, Inc.)
Dr. H. L. Fleming (GFT, Inc.)
Dr. R. Goldsmith (CeraMem Corp.)
Dr. G. Jonsson (Technical University of Denmark)
Dr. K.-V. Peinemann (GKSS, West Germany)
Dr. R. Peterson (Filmtec Corp.)
Dr. G. P. Pez (Air Products & Chemicals, Inc.)
Dr. H. F. Ridgway (Orange County Water District)
Mr. J. Short (Koch Membrane Systems, Inc.)
Dr. K. Sims (Ionics, Inc.)
Dr. K. K. Sirkar (Stevens Institute of Technology)
Dr. J. D. Way (SRI International)
The following individuals served as peer reviewers of the final report:
Dr. J. L. Anderson (Carnegie Mellon University)

Dr. J. Henis (Monsanto)
Dr. J. L. Humphrey (J. L. Humphrey and Associates)
Dr. S.-T. Hwang (University of Cincinnati)
Dr. N.N. Li (Allied Signal)
Dr. S. L. Matson (Sepracor, Inc.)
Dr. R. D. Noble (University of Colorado)
Dr. M. C. Porter (M. C. Porter and Associates)
Dr. D. L. Roberts (SRI International)
Dr. S. A. Stern (Syracuse University)
2
Acknowledgments 3
Additional information on the current Federal Government support of membrane

research was provided by:
Dr. D. Barney
Dr. R. Bedick
Dr. R. Delafield
Dr. C. Drummond
Dr. R. Gajemki
Dr. L. Jarr
Dr. R. Cortesi
Dr. G. Ondich
PJational Once Foundation
Dr. D. Bruley
Dr. D. Greenberg
Foreword
This book discusses recent developments and future directions in the field of
membrane separation systems. It describes research needed to bring energy-sav-
ing membrane separation processes to technical and commercial readiness for
commercial acceptance within the next 5 to 20 years. The assessment was con-
ducted by a group of six internationally known membrane separations experts
who examined the worldwide status of research in the seven major membrane
areas. This encompassed four mature technology areas: reverse osmosis, micro-
filtration, ultrafiltration, and electrodialysis; two developing areas: gas separa-
tion and pervaporation; and one emerging technology: facilitated transport.
Membrane based separation technology, a relative newcomer on the separation

scene, has demonstrated the potential of saving enormous amounts of energy
in the processing industries if substituted for conventional separation systems.
It has been estimated that over 1 quad annually, out of 2.6, can possibly be
saved in liquid-to-gas separations, alone, if membrane separation systems gain
wider acceptance. In recent years great strides have been made in the field and
even greater future energy savings should be available when these systems are
substituted for such conventional separation techniques as distillation, evapora-
tion, filtration, sedimentation, and absorption.
The book pays particular attention to identifying currently emerging innova-

tive processes, and to further improvements which could gain even wider accep-
tance for the more mature membrane technologies. In all, 38 priority research
areas were selected and ranked in order of priority, according to their relevance,
likelihood of success, and overall impact. Rationale was presented for all the
final selections; and the study was peer reviewed by an additional ten experts.
The topics that were pointed out as having the greatest research emphasis are
pervaporation for organic-organic separations; gas separations; microfiltration;
an oxidant-resistant reverse osmosis membrane; and a fouling-resistant ultra-
filtration membrane.
V
vi Foreword
The information in the book is from Membrane Separation Systems, Volume I-

Executive Summary, and Volume II-Final Report, by the Department of
Energy Membrane Separation Systems Research Needs Assessment Group,
authored by R.W. Baker, E.L. Cussler, W. Eykamp, W.J. Koros, R.L. Riley, and
H. Strathmann. The report was prepared by Membrane Technology & Research,
Inc. of Menlo Park, California, for the U.S. Department of Energy Office of
Program Analysis, April 1990.
The table of contents is organized in such a way as to serve as a subject index

and provides easy access to the information contained in the book.
Advanced composition and production methods developed by

Noyes Data Corporation are employed to bring this durably
bound book to you in a minimum of time. Special techniques
are used to close the gap between “manuscript” and “completed
book.” In order to keep the price of the book to a reasonable
level, it has been partially reproduced by photo-offset directly
from the original report and the cost saving passed on to the
reader. Due to this method of publishing, certain portions of the
book may be less legible than desired.
NOTICE
The studies in this book were sponsored by the U.S.

Department of Energy. On this basis the Publisher
assumes no responsibility nor liability for errors or any
consequences arising from the use of the information
contained herein.
Mention of trade names, commercial products, and

suppliers does not constitute endorsement or recommen-
dation for use by the Agency or the Publisher. Final
determination of the suitability of any information or
product for use contemplated by any user, and the man-
ner of that use, is the sole responsibility of the user.
The book is intended for informational purposes only.
The reader is warned that caution must always be ex-
ercised when dealing with hazardous materials in mem-
brane separation systems, and expert advice should be
obtained before implementation is considered.
Contents and Subject Index
VOLUME I
1. EXECUTIVE SUMMARY. ................................ .4

References. ...................................... .8
2. ASSESSMENT METHODOLOGY ........................... .9

2.1 Authors. ........................................ .9
2.2 Outline and Model Chapter. .......................... .I2
2.3 First Group Meeting. ............................... .I2
2.4 Expert Workshops. ................................ .I2
2.5 Second Group Meeting. ............................. .17
2.6 Japan/Rest of the World Survey. ........................ 17
2.7 Prioritization of Research Needs. ........................ 17
2.8 Peer Review. .................................... .I7
References. ..................................... .I8
3. INTRODUCTION. .................................... .19

3.1 Membrane Processes. ............................... .I9
3.2 Historical Development ............................. .27
3.3 The Future. ..................................... .29
3.3.1 Selectivity ................................ .29
3.3.2 Productivity. .............................. .30
3.3.3 Operational Reliability. ....................... .31
References. ..................................... .33
4. GOVERNMENT SUPPORT OF MEMBRANE RESEARCH .......... .34

4.1 Overview ....................................... .34
4.2 U.S. Government Supported Membrane Research ............ .37
4.2.1 Department of Energy ........................ .37
vii
viii Contents and Subject Index
4.2.1.1 Office of Industrial Programs/Industrial

Energy Conservation Program ............. .37
4.2.1.2 Office of Energy Research/Division of
Chemical Sciences. .................... .39
4.2.1.3 Office of Energy Research/Division of
Advanced Energy Projects. ............... .40
4.2.1.4 Office of Fossil Energy. ................. .41
4.2.1.5 Small Business Innovative Research Program. ... .42
4.2.2 National Science Foundation. ................... .45
4.2.3 Environmental Protection Agency ................ .46
4.2.4 Department of Defense. ....................... .47
4.2.5 National Aeronautics and Space Administration ....... .48
4.3 Japanese Government Supported Membrane Research ......... .48
4.3.1 Ministry of Education ........................ .49
4.3.2 Ministry of International Trade and Industry (MITI) .... .49
4.3.2.1 Basic Industries Bureau. ................. .49
4.3.2.2 Agency of Industrial Science and Technology
(AIST) ............................ .50
4.3.2.3 Water Re-Use Promotion Center (WRPC) ...... .51
4.3.2.4 New Energy Development Organization
(NEDO). ........................... .52
4.3.3 Ministry of Agriculture, Forestry and Fisheries ........ .52
4.4 European Government Supported Membrane Research. ........ .52
4.4.1 European National ...................
Programs. .53
4.4.2 EEC-Funded Membrane Research. ................ .54
4.5 The Rest of the World .............................. .55
5. ANALYSIS OF RESEARCH ........................

NEEDS .56
5.1 Priority ............................
Research Topics. .56
5.2 Research Topics by Technology Area .................... .65
5.2.1 Pervaporation .............................. .66
5.2.2 Gas Separation ............................. .68
5.2.3 Facilitated Transport. ........................ .70
5.2.4 Reverse Osmosis ............................ .72
5.2.5 Microfiltration ............................. .74
5.2.6 Ultrafiltration. ............................. .76
5.2.7 Electrodialysis ............................. .78
5.3 Comparison of Different Technology Areas ................ .79
5.4 General Conclusions. ............................... .81
References. ..................................... .84
APPENDIX A. PEER REVIEWERS’ COMMENTS ................. .86

A.1 General Comments ................................ .86
A.l.l The Report Is Biased Toward Engineering, or Toward
Basic Science .............................. .86
A.1.2 The Importance of Integrating Membrane Technology
Into Total Treatment Systems. .................. .87
A.1.3 The Ranking Scheme ......................... .88
Contents and Subject Index ix
A.1.4 Comparison with Japan ....................... .89

A.2 Specific Comments on Applications ..................... .90
A.2.1 Pervaporation .............................. .90
A.2.2 Gas Separation ............................. .91
A.2.3 Facilitated Transport. ........................ .91
A.2.4 Reverse Osmosis ............................ .92
A.2.5 Ultrafiltration. ............................. .92
A.2.6 Microfiltration ............................. .93
A.2.7 Electrodialysis ............................. .93
A.2.8 Miscellaneous Comments ...................... .93
VOLUME II
INTRODUCTION TO VOLUME II. ........................... .96

References. ..................................... .99
1. MEMBRANE AND MODULE PREPARATION. ................. 100

R.W. Baker
1 .I Symmetrical Membranes. ............................ 102
1.1.1 Dense Symmetrical Membranes .................. 102
1.1.1.1 Solution Casting. ..................... 102
1.1.1.2 Melt Pressing ........................ 102
1.1.2 Microporous Symmetrical Membranes. ............. 105
1.1.2.1 Irradiation .......................... 105
1.1.2.2 Stretching .......................... 105
1.1.2.3 Template Leaching .................... 109
1.2 Asymmetric Membranes ............................. 109
1.2.1 Phase Inversion (Solution-Precipitation) Membranes. .... 109
1.2.1 .I Polymer Precipitation by Thermal Gelation .... 110
1.2.1.2 Polymer Precipitation by Solvent Evaporation . 114
1.2.1.3 Polymer Precipitation by lmbibition of Water
Vapor ............................ .I14
1.2.1.4 Polymer Precipitation by Immersion in a
Nonsolvent Bath (Loeb-Sourirajan Process) .... 116
1.2.2 Interfacial Composite Membranes. ................ 118
1.2.3 Solution Cast Composite Membranes. .............. 121
1.2.4 Plasma Polymerization Membranes ................ 123
1.2.5 Dynamically Formed Membranes ................. 125
1.2.6 Reactive Surface Treatment. .................... 125
1.3 Ceramic and Metal Membranes. ........................ 126
1.3.1 Dense Metal Membranes ....................... 126
1 .3.2 Microporous Metal Membranes. .................. 126
1.3.3 Ceramic Membranes. ......................... 126
1.3.4 Molecular Sieve Membranes. .................... 127
1.4 Liquid Membranes. ................................ 131
1.5 Hollow-Fiber Membranes ............................ 132
1.5.1 Solution (Wet) Spinning. ...................... 134
1.5.2 Melt Spinning .............................. 134
x Contents and Subject Index
1.6 Membrane Modules ................................ 136

1.6.1 Spiral-Wound Modules ........................ 136
1.6.2 Hollow-Fiber Modules ........................ 140
1.6.3 Plate-and-Frame Modules ...................... 140
1.6.4 Tubular Systems. .......................... .I40
1.6.5 Module Selection. ........................... 140
1.7 Current Areas of Membrane and Module Research. ........... 144
References. ........................................ .I46
2. PERVAPORATION. .................................. .151

R. W. Baker
2.1 Process Overview. ................................. 151
2.1 .I Design Features. ............................ 154
2.1.2 Pervaporation Membranes. ..................... 156
2.1.3 Pervaporation Modules. ....................... 156
2.1.4 Historical Trends. ........................... 159
2.2 Current Applications, Energy Basics and Economics. .......... 160
2.2.1 Dehydration of Solvents. ...................... 161
2.2.2 Water Purification. .......................... 164
2.2.3 Pollution Control ........................... 169
2.2.4 Solvent Recovery ........................... 171
2.2.5 Organic-Organic Separations .................... 174
2.3 Industrial Suppliers ................................ 176
2.4 Sources of Innovation .............................. 180
2.5 Future Directions ................................. 182
2.5.1 Solvent Dehydration ......................... 182
2.5.2 Water Purification ........................... 184
2.5.3 Organic-Organic Separations .................... 184
2.6 DOE Research Opportunities. ......................... 185
2.6.1 Priority Ranking ............................ 186
2.6.1 .I Solvent Dehydration ................... 186
2.6.1.2 Water Purification. .................... 186
2.6.1.3 Organic-Organic Separations .............. 186
References. ..................................... 187
3. GAS SEPARATION. .................................. .I89

NJ. Koros
3.1 Introduction. .................................... 189
3.2 Fundamentals. .................................. .I91
3.3 Membrane System Properties. ......................... 193
3.4 Module and System Design Features ..................... 195
3.5 Historical Perspective. .............................. 199
3.6 Current Technical Trends in the Gas Separation Field. ......... 200
3.6.1 Polymeric Membrane ..................
Materials. 200
3.6.2 Plasticization .........................
Effects 203
3.6.3 Nonstandard Membrane Materials. ................ 205
3.6.4 Advanced Membrane Structures. ................. 205
3.6.5 Surface Treatment to Increase Selectivity. ........... 205
Contents and Subject Index xi
3.6.6 System Design and Operating Trends. .............. 206

3.7 Applications. ................................... .207
3.7.1 Hydrogen Separations ........................ 207
3.7.2 Oxygen-Nitrogen Separations. .................. .212
3.7.3 Acid Gas Separations ........................ .215
3.7.4 Vapor-Gas Separations ....................... .218
3.7.5 Nitrogen-Hydrocarbon Separations. ............... 219
3.7.6 Helium Separations .......................... 220
3.8 Energy Basics. .................................. .220
3.9 Economics. .................................... .225
3.10 Suppliers. ..................................... .225
3.11 Sources of Innovation ............................. .227
3.11 .I Research Centers and Groups. ................... 227
3.11.2 Support of Membrane-Based Gas Separation. ......... 230
3.11.2.1 United States ........................ 230
3.11.2.2 Foreign ............................ 230
3.12 Future Directions ................................ .230
3.12.1 Industrial Opportunities ....................... 230
3.12.2 Domestic Opportunities ....................... 231
3.13 Research Opportunities ............................ .231
3.13.1 Ultrathin Defect-Free Membrane Formation Process .... 231
3.13.2 Highly Oxygen-Selective Materials ................ 235
3.13.3 Polymers, Membranes and Modules for Demanding
Service ................................. .235
3.13.4 Improved Composite Membrane Formation Process. .... 235
3.13.5 Reactive Surface Modifications .................. 236
3.13.6 High-Temperature Resistant Membranes ............ 236
3.13.7 Refinement of Guidelines and Analytical Methods for
Membrane Material Selection. ................... 236
3.13.8 Extremely Highly Oxygen-Selective Membrane
Materials ................................ .237
3.13.9 Physical Surface Modification byAntiplasticization ..... 237
3.13.10 Concentration of Products from Dilute Streams ....... 237
References. .................................... .238
4. FACILITATED TRANSPORT. ........................... .242

E. L . Cussler
4.1 Process Overview. ................................ .242
4.1.1 The Basic Process .......................... .242
4.1.2 Membrane Features ......................... .244
4.1.3 Membrane and Module Design Factors. ............. 248
4.1.4 Historical Trends. ........................... 251
4.2 Current Applications ............................... 251
4.3 Energy Basics. .................................. .255
4.4 Economics. .................................... .257
4.4.1 Metals. ................................. .257
4.4.2 Gases.. ............................... ..25 8
4.4.2.1 Air Separation ....................... 258
xii Contents and Subject Index
4.4.2.2 Acid Gases. ......................... 258

4.4.2.3 Olefin-Alkanes and Other Separations ........ 259
4.4.3 Biochemicals ............................. .259
4.4.4 .................................
Sensors .260
4.5 Supplier ................................
industry .260
4.6 Research Centers and Groups. ......................... 260
4.7 Current Research. ................................. 261
4.8 Future Directions ................................. 262
4.8.1 Metal Separations ........................... 262
4.8.2 Gases .................................. .264
4.8.3 Biochemicals ............................. .267
4.8.4 Hydrocarbon Separations ...................... 268
4.8.5 Water Removal ............................ .268
4.8.6 Sensors ................................. .268
4.9 Research Opportunities: Summary and Conclusions .......... 269
References. .................................... .273
5. REVERSEOSMOSIS...................................27 6
R.L. Riley
5.1.1 The Basic Process .......................... .276
5.1.2 Membranes. .............................. .280
5.1.3 Modules. ................................ .280
5.1.4 Systems. ................................ .284
5.2 The Reverse Osmosis Industry ........................ .289
5.2.1 Current Desalination Plant Inventory .............. 289
5.2.1.1 Membrane Sales ...................... 289
5.2.2 Marketing of Membrane Products. ................ 291
5.2.3 Future Direction of the Reverse Osmosis Membrane
Industry ................................ .293
5.3 Reverse Osmosis Applications ......................... 293
5.4 Reverse Osmosis Capital and Operating Costs ............... 295
5.5 Identification of Reverse Osmosis Process Needs. ............ 299
5.5.1 Membrane Fouling. ......................... .299
5.5.2 Seawater Desalination ....................... .302
5.5.3 Energy Recovery for Large Seawater Desalination
Systems. ................................ ,305
5.5.4 Low-Pressure Reverse Osmosis ..........
Desalination 308
5.5.5 Ultra-Low-Pressure Reverse Osmosis Desalination ...... 310
5.6 DOE Research Opportunities. ........................ .312
5.6.1 Projected Reverse Osmosis Market: 1989-1994 ....... 312
5.6.2 Research and Development: Past and Present. ........ 312
5.6.3 Research and Development: Energy Reduction ...... .314
5.6.4 Thin-Film Composite Membrane Research ........... 314
5.6.4.1 Increasing Water Production Efficiency ....... 314
5.6.4.2 Seawater Reverse Osmosis Membranes. ...... .315
5.6.4.3 Low-Pressure Membranes ................ 316
5.6.4.4 Ultra-Low-Pressure Membranes ........... .316
Contents and Subject Index xiii
5.6.5 Membrane Fouling: Bacterial Adhesion to Membrane

Surfaces.. ... . . . . . . . . . . . . . . . . . . . . . . . . . . . .317
5.6.6 Spiral-Wound Element Optimization. . . . . . . . . . . . . . .318
5.6.7 Future Directions and Research Topics of interest for
Reverse Osmosis Systems and Applications. . . . . . . . . . . 319
5.6.8 Summary of Potential Government-Sponsored
Energy Saving Programs . . . . . . . . . . . . . . . . . . . . . . .325
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .327
6. MICROFILTRATION ................................. .329

William Eykamp
6.1 Overview ...................................... .329
6.2 Definitions and Theory. ............................ .330
6.3 Design Considerations .............................. 337
6.3.1 Dead-End vs Crossflow Operation. ............... .337
6.3.2 Module Design Considerations. ................. .339
6.3.2.1 Dead-End Filter ................
Housings 340
6.3.2.2 Crossflow Devices .................... .340
6.4 Status of the Microfiltration Industry .................... 342
6.4.1 Background .............................. .342
6.4.2 Suppliers. ............................... .342
6.4.3 Membrane Trends .......................... .344
6.4.4 Module Trends ............................ .348
6.4.5 Process Trends ............................ .348
6.5 Applications for Microfiltration Technology. ............... 348
6.5.1 Current Applications ........................ .348
6.5.2 Future Applications. ........................ .349
6.5.2.1 Water Treatment. ..................... 351
6.5.2.2 Sewage Treatment. ................... .351
6.5.2.3 Clarification: Diatomaceous Earth
Replacement ....................... .351
6.5.2.4 Fuels. ............................ .352
6.5.3 Industry Directions ......................... .352
6.6 Process Economics. ............................... .353
6.7 Energy Considerations ............................. .355
6.8 Opportunities in the industry ........................ .356
6.8.1 Commercially-Funded Opportunities. .............. 356
6.8.2 Opportunities for Governmental Research
Participation. ............................. .356
References. .................................... .359
7. ULTRAFILTRATION. ................................ .360

W. Eykamp
7.1.1 The Gel Model ............................ .362
7.1.2 Concentration Polarization ..................... 367
7.1.3 Plugging. ................................ .367
7.1.4 Fouling. ................................ .368
xiv Contents and Subject Index
7.1.5 Flux Enhancement ......................... .369

7.1.6 Module Designs. ........................... .370
7.1.7 Design Trends. ............................ .37 1
7.2 Applications. ................................... .372
7.2.1 Recovery of Electrocoat Paint. ................. .372
7.2.2 Fractionation of Whey ....................... .372
7.2.3 Concentration of Textile Sizing ................. .372
7.2.4 Recovery of Oily Wastewater. .................. .375
7.2.5 Concentration of Gelatin ..................... .375
7.2.6 Cheese Production. ......................... .375
7.2.7 Juice. .................................. .375
7.3 Energy Basics. .................................. .377
7.3.1 Direct Energy Use vs Competing Processes ........... 378
7.3.2 Indirect Energy Savings ...................... .378
7.4 Economics. .................................... .379
7.4.1 Typical Equipment Costs ..................... .379
7.4.2 Downstream .........................
Costs. .381
7.4.3 Product Recovery .......................... .382
7.4.4 Selectivity ............................... .383
7.5 Supplier Industry ................................ .384
7.6 Sources of Innovation ............................. .385
7.6.1 Suppliers. ............................... .385
7.6.2 Users. .................................. .386
7.6.3 Universities .............................. .387
7.6.4 Government. ............................. .387
7.6.5 Foreign Activities ........................... 387
7.7 Future Directions ................................ .387
7.8 Research Needs. ................................. .390
7.9 DOE Research Opportunities. ......................... 393
References. .................................... .394
8. ELECTRODIALYSIS. ................................. .396

H. Strathmann
8.1 Introduction. ................................... .396
8.2.1 The Principle of the Process and Definition of Terms. ... 396
8.2.1.1 The Process Principle ................... 397
8.2.1.2 Limiting Current Density and Current
Utilization. ........................ .397
8.2.2 Design Features and Their Consequences ........... .403
8.2.2.1 The Electrodialysis Stack ................ 404
8.2.2.2 Concentration Polarization and Membrane
Fouling. .......................... .404
8.2.2.3 Mechanical, Hydrodynamic, and Electrical
Stack Design Criteria .................. .406
8.2.3 Ion-Exchange Membranes Used in Electrodialysis ..... .408
8.2.4 Historical Developments ...................... .412
8.3 Current Applications of Electrodialysis ................... 413
Contents and Subject index xv
8.3.1 Desalination of Brackish Water

by Electrodialysis .. . . .413
8.3.2 Production of Table . .. . .. . . . . . .. . ,. . . .
Salt. 415
8.3.3 Electrodialysis in Wastewater Treatment . . .. . .. . . : :415
8.3.3.1 Concentration of Reverse Osmosis Brines. . . . .415
8.3.4 Electrodialysis in the Food and Pharmaceutical
Industries.. .. . .. . .. . .. . . . .. . . . .. . .. . .. . 416
83.5 Production of Ultrapure Water. . . . . . . . . . . . . . . . : :416
8.3.6 Other Electrodialysis-Related Processes . . . . . . . . . . . . 418
8.3.6.1 Donnan-Dialysis with Ion-Selective
Membranes. . . . ..
. .. . .. .. . . . . . . , . . . 418
8.3.6.2 Electrodialytic Water Dissociation . .. . . . . . . 418
8.4 Electrodialysis Energy Requirement ..................... 421
8.4.1 Minimum Energy Required for the Separation of
Water from a Solution ....................... .421
8.4.2 Practical Energy Requirement in Electrodialysis
Desalination. ............................. .421
8.4.2.1 Energy Requirements for Transfer of Ions
from the Product Solution to the Brine. ...... 422
8.4.2.2 Pump Energy Requirements ............. .423
8.4.2.3 Energy Requirement for the Electrochemical
Electrode Reactions. ................... 423
8.4.3 Energy Consumption in Electrodialysis Compared with
Reverse Osmosis ........................... .423
8.5 Electrodialysis System Design and Economics. .............. 425
8.5.1 Process Flow Description ...................... 425
8.5.2 Electrodialysis Plant Components. ............... .425
8.5.2.1 The Electrodialysis Stack ................ 425
8.5.2.2 The Electric Power Supply .............. .427
8.5.2.3 The Hydraulic Flow System .............. 427
8.5.2.4 Process Control Devices ................ .427
8.5.3 Electrodialysis Process Costs ................... .427
8.5.3.1 Capital ........................... .427
8.5.3.2 Operating Costs. ..................... ,430
8.5.3.3 Total Electrodialysis Process Costs .......... 430
8.6 Supplier Industry ................................ .433
8.7 Sources of Innovation-Current Research. ................ .435
8.7.1 Stack Design Research ........................ 435
8.7.2 Membrane Research. ........................ .436
8.7.3 Basic Studies on Process Improvements. ........... .437
8.8 Future Developments. ............................. .439
8.8.1 Areas of New Opportunity .................... .439
8.8.2 Impact of Present R&D Activities on the Future Use
of Electrodialysis. ........................... 439
8.8.3 Future Research Directions. ................... .444
References. .................................... .446
9. GLOSSARY OF SYMBOLS AND ABBREVIATIONS ............ .449

Volume I
1. Executive Summary
The Office of Program Analysis in the Office of Energy Research of the

Department of Energy (DOE) commissioned this study to evaluate and prioritize
research needs in the membrane separation industry.
One of the primary goals of the U.S. Department of Energy is to foster and
support the development of energy-efficient new technologies. In 1987, the total
energy consumption of all sectors of the U.S. economy was 76.8 quads, of which
approximately 29.5 quads, or 38%. was used by the industrial sector, at a cost of
$100 billion.’ Reductions in energy consumption are of strategic importance,
because they reduce U.S. dependence on foreign energy supplies. Improving the
energy efficiency of production technology can lead to increased productivity and
enhanced competitiveness of U.S. products in world markets. Processes that use
energy inefficiently are also significant sources of environmental pollution.
The rationale for seeking innovative, energy-saving technologies is,

therefore, very clear. One such technology is membrane separation, which offers
significant reductions in energy consumption in comparison with thermal
separation techniques. Membranes separate mixtures into components by
discriminating on the basis of a physical or chemical attribute, such as molecular
size, charge or solubility. They can pass water while retaining salts, the basis of
producing potable water from the sea. They are used for passing solutions, while
retaining bacteria, the basis for cold sterilization. They can separate air into
oxygen and nitrogen. There are numerous applications for membranes in the
world today. Total sales of industrial membrane separation systems worldwide are
greater than $1 billion annuaily.2 The United States is a dominant supplier of
these systems. United States dominance of the industry is being challenged,
however, by Japanese and, to a lesser extent, European competitors.
Some membranes are used in circumstances where energy saving is an

important criterion. Others are used in small-scale applications where energy
costs are relatively unimportant. This report looks at the major membrane
processes to assess their status and potential, particularly with regard to energy
4
Executive Summary 5
saving. Related technologies, for example the membrane catalytic reactor,

although outside the scope of this study, are believed to have additional potential
for energy savings.
This report was prepared by a group of six membrane experts representing

the various fields of membrane technology. Based on group meetings and review
discussions, a list of five to seven priority research topics was prepared by the
group for each of the seven major membrane technology areas: reverse osmosis,
ultrafiltration, microfiltration, electrodialysis, pervaporation, gas separation and
facilitated transport. These items were incorporated into a master list, totaling
38 research topics, which were then ranked in order of priority.
The highest ranked research topic was pervaporation membranes for organic-
organic separations. Another pervaporation-related topic concerning the
development of organic-solvent-resistant modules ranked seventh. The very high
ranking of these two pervaporation research topics reflects the promise of this
rapidly developing technology. Distillation is an energy-intensive operation and
consumes 28% of the energy used in all U.S. chemical plants and petroleum
refineries.s The total annual distillation energy consumption is approximately 2
quads.’ Replacement or augmentation of distillation by pervaporation could
substantially reduce this energy usage. If even 10% of this energy could be saved
by using membranes, for example in hybrid distillation/pervaporation systems, this
would represent an energy savings of 0.2 quad, or IO6 barrels of oil per day.
Three topics relating to the development of gas-separation membranes ranked

in the top 10 of the master list. Membrane-based gas separation is an area in
which the United States was a world leader. The dominant position of U.S.
suppliers, and U.S. research, is under threat of erosion because of the increased
attention being devoted to the subject by Japanese and European companies,
governments and institutions. Increased emphasis on membrane-based gas-
separation research and development would increase the probability that the new
generation technology for high-performance, ultrathin membranes will be
controlled by the United States. The attendant benefits would be that membrane-
based gas separation would become competitive with conventional, energy-intensive
separation technologies over a much broader spectrum. The energy savings that
6 Membrane Separation Systems
might be achieved by membrane-based gas-separation technology are exemplified

by two potential applications. If high-grade oxygen-enriched streams were
available at low cost, as a result of the development of better oxygen-selective
membranes, then combustion processes through industry could be made more
energy efficient. Various estimates have placed the energy savings from use of
high-grade oxygen enriched air at between 0.06 and 0.36 quads per year.’ It is
estimated that using membranes to upgrade sour natural gas will result in an
energy savings of 0.01 quads per year.
The second highest priority topic in the master list was the development of
oxidation-resistant reverse osmosis membranes. The current generation of
reverse-osmosis membranes have adequate salt rejection and water flux.
However, they are susceptible to degradation by sterilizing oxidants. High-
performance, oxidation-resistant membranes could displace existing cellulose
acetate membranes and open up new applications of reverse osmosis, particularly
in food processing. The energy use for evaporation in the food industry has been
estimated at about 0.09 quads.’ Reverse osmosis typically requires an energy
input of 20-40 Btu/lb of water removed.’ Assuming an average energy
consumption for conventional evaporation processes of 600 Btu/lb, the substitution
of reverse osmosis for evaporation could result in a potential energy savings of
0.04-0.05 quads.
In general, facilitated-transport related topics scored low in the master

priority list, reflecting the disenchantment of the expert group with a technology
with which membrane scientists have been struggling for the last 20 years
without reaching the point of practical viability. The development of facilitated-
transport, oxygen-selective, solid-carrier membranes was, however, given a high
research priority ranking of four. If stable, solid facilitated-transport membranes
could really be developed, they might offer much higher selectivities than polymer
membranes, and have a major effect on the oxygen and nitrogen production
industries.
The principal problem in ultrafiltration technology is membrane fouling. The

development of fouling-resistant ultrafiltration membranes was given a research
priority ranking of six. The development of fouling-resistant ultrafiltration
Executive Summary 7
membranes would have a major impact on cost and energy savings in the milk and
cheese production industries, for example.
Two high-priority topics cover research opportunities in the microfiltration

area, namely, development of low-cost microfiltration modules and development of
high-temperature solvent resistant membranes and modules. Microfiltration is a
well developed and commercially successful industry, whose industrial focus has
been in the pharmaceutical and food industries. Drinking water and sewage
treatment are new, but non-glamorous applications for microfiltration, requiring
membranes and equipment whose design concept and execution may be
incompatible with the mission of the private industry participants. The potential
for societal impact in this area is great, but existing microfiltration firms may
not find the opportunity appealing, because of technical risks, regulatory
constraints or competition from conventional alternatives.
Reverse osmosis, ultrafiltration and microfiltration are all technologies with

significant energy-savings potential across a broad spectrum of industry. For
example, a significant fraction of the wastewater streams from the food,
chemical and petroleum processing industries are discharged as hot streams and
the energy lost is estimated at 1 - 2 quads annually.a The development of low-
cost, chemically resistant MF/UF/RO membrane systems that could recover the
hot wastewater and recycle it to the process would result in considerable energy
savings. If only 25% of the energy present in the wastewater were recovered,
this would result in an energy savings of 0.25 to 0.50 quads.g
Many of the top 10 ranked priority research topics spotlighted technology

and engineering problems. In the view of the authors of the report, it appears
that emerging membrane separations technologies have reached a level of maturity
where progress toward competitive, energy-efficient industrial systems will be
most effectively expedited by increasing DOE support of engineering or
technology-based research programs. Applications-related research was viewed as
equally worthy of support as fundamental scientific studies. This view was not
shared unanimously by the reviewers, however. Two reviewers objected that the
list of research priorities was too much skewed toward practical applications and
gave a low priority to the science of membranes, from whence the long-term
innovations in membrane technology will come. One reviewer, on the other hand,
felt strongly that there was too much emphasis on basic research issues, and that
most of the top priority items identified in the report did not adequately address
engineering issues.
During the course of the study, government support of membrane-related

research in Japan and Europe was investigated. The Japanese government and the
European governments each spend close to $20 million annually on membrane-
related topics. Federal support for membrane-related research and development
through all agencies is currently about SIO-11 million per year. The United
States is. therefore, in third place in terms of government assistance to membrane
research. There was concern among some members of the group that this level
of spending will ultimately result in loss of world market share.
REFERENCES
1. W.M. Sonnett, personal communication.
2. A.M. Crull, “The Evolving Membrane Industry Picture,” in The 1998 Sixth
Ann ua I M e m b r a n e T ec h no Ionv/Planninn Conference Proceedin (?s, Business
Communications Company, Inc., Cambridge, MA (1988).
3. Bravo, J.L., Fair, J.R. J.L. Humphrey, CL. Martin, A.F. Seibert and S. Joshi,
“Assessment of Potential Energy Savings in Fluid Separation Technologies:
Technology Review and Recommended Research Areas,” Department of Energy
Report DOE/LD/12473--I (1984).
4. Mix, T.W., Dweck, J.S., Weinberg, M., and Armstrong, R.C., “Energy
Conservation in Distillation - Final Report”, DOE/CS/40259 (1981).
5. The DQE Industrial Energy Program: Research and Development in Separation

Technology. DOE publication number DOE/NBM - 80027730.
6. Parkinson, G., “Reverse Osmosis: Trying for wider applications,” Chemical

Engineering, p.26, May 30, 1983.
7. Mohr, C.M., Engelgau, D.E., Leeper. S.A.. and Charboneau, B.L., Membrane
ADoiications and Research in Food Processing, Noyes Data Corp., Park Ridge,
NJ 1989.
a. Bodine. J.F.. (ed.) Industrial Enerav Use Databoa, ORAU-160 (1980).
9. Leeper, S.A.. Stevenson, D.H., Chiu, P.Y.-C., Priebe, S.J., Sanchez, H.F., and
Wikoff, P.M., “Membrane Technology and Applications: An Assessment,” U.S.
DOE Report No. DE84009000, 1984.
2. Assessment Methodology
Industrial separation processes consume a sigmficant portion of the energy

used in the United States. A 1986 survey by the Office of Industrial Programs
estimated that about 2.6 quads of energy are expended annually on liquid-to-
vapor separations alone.’ This survey also concluded that over 1.0 quad of
energy could be saved if the industry adopted membrane separation systems more
widely.
Membrane separation systems offer significant advantages over existing

separation processes. In addition to consuming less energy than conventional
processes, membrane systems are compact and modular, enabling easy retrofit of
existing applications. This study was commissioned by the Department of Energy,
Office of Program Analysis, to identify and prioritize membrane research needs in
order of their impact on the DOE’s mission, such that support of membrane
research may produce the most effective results over the next 20 years.
2.1 AUTHORS
This report was prepared by a group of senior researchers well versed in

membrane science and technology. The executive group consisted of Dr. Richard
W. Baker (Membrane Technology & Research, Inc.), Dr. William Eykamp (University
of California at Berkeley) and Mr. Robert L. Riley (Separation Systems
Technology, Inc.), who were responsible for the direction and coordination of the
program. Dr. Eykamp also served as Principal Investigator for the program.
The field of membrane science was divided into seven general categories
based on the type of membrane process. To ensure that each of these categories
was covered by a leading expert in the field, the executive group was
supplemented by three additional authors. These additional group members were
Dr. Edward Cussler (University of Minnesota), Dr. William J. Koros (University of
Texas at Austin), and Dr. Heiner Strathmann (Fraunhofer Institute, West
Germany). Each of the authors was assigned primary responsibility for a topic
area as shown in Table 2-1.
9
Table 2- 1. List of Authors
Innir, Author
Membrane and Module Preparation Richard Baker

Microfiltration and Ultrafiltration William Eykamp
Reverse Osmosis Robert Riley
Pervaporation Richard Baker
Gas Separation William Koros
Facilitated and Coupled Transport Edward Cussler
Electrodialysis Heiner Strathmann
The role of the group of authors was to assess the current state of
membranes in their particular section, identify present and future applications
where membrane separations could result in significant energy savings and suggest
research directions and specific research needs required to achieve these energy
savings within a 5-20 year time frame. The collected group of authors also
performed the prioritization of the overall research needs.
As program coordinator, Dr. Amulya Athayde provided liaison between the

authors and the contractor, Membrane Technology & Research. Inc (MTR). Ms.
Janet Farrant (MTR) was responsible for the patent information searches and the
editing and final assembly of this report. The overall plan for preparation of
the report is shown in Figure 2- 1.
outlin First panel Prepare drafts
-) moating of sections
Int6rnatlonal
L---c mombrme research Peer
SUNOy
rovlow
0RIpofi
P
Figure 2-l. Overall plan for conducting the study of research needs in membrane separation systems
2.2 OUTLINE AND MODEL CHAPTER
The first major task of this program was to develop an outline for the
report and draft a model chapter. The outline was prepared by the executive
group and submitted to the authors of the individual sections for consideration.
A patent and literature survey was conducted at MTR in each of the topic areas
(listed in Table 2-1) to assess the state of the art as represented by recent
patents, product brochures and journal articles. This information was provided to
the group of authors.
Projects accomplished by committees are proverbially characterized by poor

cohesion and a lack of direction. To circumvent such criticism of this report
the section on reverse osmosis was selected as a model chapter for the rest of
the report. A draft prepared by Mr. Robert Riley was circulated among the
other authors to illustrate the desired format. The goal of this exercise was to
ensure that the report had a uniform style and emphasis. with the individuai
chapters in accord with each other.
2.3 FIRST GROUP MEETING
The first group meeting was held at MTR on December 26-27, 1988, and was
attended by the authors and the ex-officio group members representing the DOE:
Mr. Robert Rader and Dr. Gilbert Jackson (Office of Program Analysis), Dr.
William Sonnett (Office of Industrial Programs) and Dr. Richard Gordon (Office
of Energy Research, Division of Chemical Sciences).
The authors presented draft outlines of their sections. which were reviewed
by the entire group. The model chapter was discussed and revisions for the
outlines of the other chapters were drawn up.
2.4 EXPERT WORKSHOPS
A series of “expert workshops” was held upon completion of the draft

chapters to discuss the conclusions and recommendations of the authors with
membrane energumena drawn from the U.S. and international membrane
Assessment Methodology 13
communities. These workshops consisted of closed-panel discussions, organized in

conjunction with major membrane research conferences.
Two or three experts in the particular area were invited to review the draft
chapters and respond with their comments and criticism. The workshops
provided an opportunity for the authors to update the information on the state
of the art, as well as to obtain an informed consensus on the recommended
research directions and needs.
The workshops for the Reverse Osmosis, Ultrafiltration, Microfiltration,

Coupled and Facilitated Transport, Gas Separation and Pervaporation sections
were held on May 16-20, 1989, during the North American Membrane Society
Third Annual Meeting in Austin, Texas. The workshop on Electrodialysis was held
on August 4, 1989. during the Gordon Research Conference on membrane
separations in Plymouth, New Hampshire. A special workshop was also held at
the Gordon Research Conference during which all of the authors were present and
the list of research needs was discussed with the conference attendees. The lists
of workshop attendees are given in Table 2-2.
Table 2-2. Workshop Attendees
WORKSHOP ON ULTRAFILTRATION AND MICROFILTRATION
Attendee Affiliation
W. Eykamp (Author) University of California, Berkeley

G. Jackson DOE
R. Rader DOE
J. Short Koch Membrane Systems, Inc.
G. Jonsson Technical University of Denmark
A. L. Athayde MTR, Inc.
WORKSHOP ON REVERSE OSMOSIS
R. L. Riley (Author) Separation Systems Technology. Inc.

W. Eykamp University of California, Berkeley
R. Rader DOE
D. Blanchfield DOE Idaho Operations Office
D. Cummings EG&G Idaho
R. Peterson Filmtec Corp.
H. F. Ridgway Orange County Water District
A. L. Athayde MTR. Inc.
WORKSHOP ON GAS SEPARATION
W. J. Koros (Author) University of Texas. Austin

W. Eykamp University of California. Berkeley
R. W. Baker MTR, Inc.
R. Rader DOE
D. Cummings EG&G Idaho
R. Goldsmith CeraMem Corp.
B. Bikson Innovative Membrane Systems/ Union
Carbide Corp.
G. P. Pez Air Products & Chemicals, Inc.
WORKSHOP ON COUPLED AND FACILITATED TRANSPORT
E. L. Cussler (Author) University of Minnesota

R. Rader DOE
G. Jackson DOE
D. Haefner EC&G Idaho
J. D. Way SRI International
K. K. Sirkar Stevens Institute of Technology
G. P. Pez Air Products & Chemicals, Inc.
WORKSHOP ON PERVAPORATION
R. W. Baker (Author) MTR, Inc.

K.-V. Peinemann GKSS, West Germany
R. Rader DOE
G. Jackson DOE
H. L. Fleming GFf, Inc.
WORKSHOP ON ELECTRODIALYSIS
H. Strathmann (Author) Fraunhofer Institute, West Germany

W. J. Koros University of Texas, Austin
R. L. Riley Separation Systems Technology, Inc.
D. Elyanow Ionics, Inc.
L. Costa Ionics. Inc.
K. Sims Ionics. Inc.
T. Davis Graver Water, Inc.
P. M. Gallagher Alcan International. U.K.
W. Gudernatsch Fraunhofer Institute, West Germany
GENERAL WORKSHOP HELD AT THE GORDON RESEARCH CONFERENCE

W. J. Koros University of Texas, Austin
R. L. Riley Separation Systems Technology, Inc.
H. Strathmann Fraunhofer Institute, West Germany
E. L. Cussler University of Minnesota
J. Beasley Consultant
C. H. Lee AMT
T. Lawford EG&G Idaho
A. Allegrexa Millipore
L. Zeman Millipore
G. Blytas Shell Chemical Co.
D. Fain Martin Marietta Energy Systems
J. D. Way Oregon State University
K. Murphy Permea - Monsanto
I. Roman E. I. Du Pont de Nemours, Inc.
E. Sanders Dow Chemical Corp.
G. Tkacik Millipore
W. Robertson PPG
R. L. Hapke SRI International
J. Pellegrino NIST
L. Costa Ionics. Inc.
2.5 SECOND GROUP MEETING
The second group meeting was held during the Gordon Research Conference
and was attended by all of the authors. The final format of each chapter was
discussed and format revisions, based on comments from the expert workshops,
were adopted.
2.6 JAPAN/REST OF THE WORLD SURVEY
This study contains a review of the state of the art of membrane science
and technology in Japan, Europe and the rest of the world. Particular emphasis
is placed on support of membrane research by foreign governments and sources of
innovation in other countries. Two of the authors (Eykamp and Riley) visited
Japan to collect information on membrane research in that country. Information
on Europe was provided by Dr. Strathmann.
2.7 PRIORITIZATION OF RESEARCH NEEDS
The expert workshops identified over 100 research needs in membrane

separations. Although these items had been rated in terms of importance and
prospect of realization, they had been ranked within the individual sections of
membrane technology. To facilitate the prioritization process, the research needs
were condensed into a short list of 38 items, with the 5-7 highest ranked items
selected from each of the individual sections.
The short list of research needs was submitted to the group of authors,
who were asked to rank each of the items on the basis of energy-saving
potential and other objectives related to DOE’s mission.
2.8 PEER REVIEW
The report was submitted to a group of 10 reviewers selected by the DOE.

Table 2-3 is a list of the reviewers. The reviewers comments. along with
rebuttals or responses as appropriate, are presented in Appendix A.
Table 2-3. List of Peer Reviewers
Name Affiliation
Dr. J. L. Anderson Carnegie Mellon University

Dr. J. Henis Monsanto
Dr. J. L. Humphrey J.L. Humphrey and Associates
Dr. S.-T. Hwang University of Cincinnati
Dr. N. N. Li Allied Signal Corp.
Dr. S. L. Matson Sepracor, Inc.
Dr. R. D. Noble University of Colorado
Dr. M. C. Porter M. C. Porter and Associates
Dr. D. L. Roberts SRI International
Dr. S. A. Stern Syracuse University
REFERENCES
1. The DOE Industrial Energy Program: Research and Development in Separation

3. Introduction
3.1 MEMBRANE PROCESSES
Seven major membrane processes are discussed in this report. They are
listed in Table 3-1. There are four developed processes, microfiltration (MF),
ultrafiltration (UF), reverse osmosis (RO), and electrodialysis (ED). These are all
well established and the market is served by a number of experienced companies.
The first three processes are related filtration techniques, in which a solution
containing dissolved or suspended solutes is forced through a membrane filter.
The solvent passes through the membrane; the solutes are retained.
Table 3- 1. Membrane Technologies Addressed in This Report
Process Status
Developed Microfiltration Well established unit processes.

technologies Ultrafiltration No major breakthroughs seem
Reverse Osmosis imminent
Electrodialysis
Developing Gas separation A number of plants have been

technologies Pervaporation installed. Market size and
number of applications served
is expanding rapidly.
To-be-developed Facilitated transport Major problems remain to be

technologies solved before industrial systems
will be installed
Microfiltration, ultrafiltration, and reverse osmosis differ principally in the

size of the particles separated by the membrane. Microfiltration is considered to
refer to membranes that have pore diameters from 0.1 pm (1,000 A) to IO pm.
Microfiltration membranes are used to filter suspended particulates. bacteria or
large colloids from solutions. Ultrafiltration refers to membranes having pore
19
diameters in the range 20-1.000 A. Ultrafiltration membranes can be used to

filter dissolved macromolecules, such as proteins, from solution. Typical
applications of ultrafiltration membranes are concentrating proteins from milk
whey, or recovery of colloidal paint particles from electrocoat paint rinse waters.
In the case of reverse osmosis, the membrane pores are so small, in the
range of 5-20 A in diameter, that they are within the range of the thermal
motion of the polymer chains. The most widely accepted theory of reverse
osmosis transport considers the membrane to have no permeant pores at all.’
Reverse osmosis membranes are used to separate dissolved microsolutes, such as
salt, from water. The principal application of reverse osmosis is the production
of drinking water from brackish groundwater. or the sea. Figure 3-1 shows the
range of applicability of reverse osmosis, ultrafiltration, microfiltration and
conventional filtration.
The fourth fully developed membrane process is electrodialysis, in which

charged membranes are used to separate ions from aqueous solutions under the
driving force of an electrical potential difference. The process utilizes an
electrodialysis stack, built on the filter-press principle. and containing several
hundred individual cells formed by a pair of anion and cation exchange
membranes. The principal application of electrodialysis is the desalting of
brackish groundwater. However, industrial use of the process in the food
industry, for example to deionize cheese whey, is growing, as is its use in
pollution-control applications. A schematic of the process is shown in Figure 3-2.
Introduction 21
Psuedomonas
NIW dlmlnuta
(3.7H) Iniluenza 0.28 p
SUtXOSa Hem0 lobln
“s”
(2 ) (lO& 3%
cl0 I
1 I 1 1
0 0 0 0
Pore diame1.r
Figure 3-1. Reverse osmosis, ultrafiltration, microfiltration and conventional

filtration are all related processes differing principally in the
average pore diameter of the membrane filter. Reverse osmosis
membranes are so dense that discrete pores do not exist.
Transport in this case occurs via. statistically distributed free
volume areas. The relative size of different solutes removed by
each class of membrane is illustrated in this schematic.
Pick-up ~olullon salt solution
Cathode -
fed
To negative pole
of rectlfler+
i
Concultreted enlwnt
Figure 3-2. A schematic diagram of an electrodialysis process.

Introduction 23
There are two developing processes: gas separation with polymer

membranes and pervaporation. Gas separation with membranes is the more
developed of the two techniques. At least 20 companies worldwide offer
industrial, membrane-based gas separation systems for a variety of applications.
Two companies currently offer industrial pervaporation systems. The potential for
each process to capture a significant slice of the separations market is large. In
gas separation, a mixed gas feed at an elevated pressure is passed across the
surface of a membrane that is selectively permeable to one component of the
feed. The membrane separation process produces a permeate enriched in the more
permeable species and a residue enriched in the less permeable species. The
process is illustrated in Figure 3-3. Major current applications are the separation
of hydrogen from nitrogen, argon and methane in ammonia plants, the production
of nitrogen from air and the separation of carbon dioxide from methane in
natural gas operations. Gas separation is an area of considerable current research
interest and it is expected that the number of applications will expand rapidly
over the next few years.
Pervaporation is a relatively new process that has elements in common with

reverse osmosis and gas separation. In pervaporation, a liquid mixture is placed
in contact with one side of a membrane and the permeate is removed as a vapor
from the other. The mass flux is brought about by maintaining the vapor
pressure on the permeate side of the membrane lower than the partial pressure
of the feed liquid. This partial pressure difference can be maintained in several
ways. In the laboratory, a vacuum pump is used. Industrially, the low pressure
is generated by cooling and condensing the permeate vapor. A schematic of a
simple pervaporation process using a condenser to generate the permeate vacuum
is shown in Figure 3-4. Currently, the only industrial application of
pervaporation is the dehydration of organic solvents, in particular, the
dehydration of 90-95% ethanol solutions, a difficult separation problem because of
the ethanol-water axeotrope at 95% ethanol. However, pervaporation processes
are being developed for the removal of dissolved organics from water and the
separation of organic solvent mixtures. If the pervaporation of organic mixtures
becomes commercial, it will replace distillation in a number of very large
commercial applications.
module
Pressurized
Residue
feed gas -
7
Permeate
Figure 3-3. Schematic of a membrane gas separation process.
Permeate depleted
solution
Two component
liquid I&
Condenser
Uquld
-9
Figure 3-4. Schematic of a pervaporation process.

introduction 25
The final membrane process studied in the report is facilitated transport.

This process falls. under the heading of “to be developed” technology. Facilitated
transport usually employs liquid membranes containing a complexing or carrier
agent. The carrier agent reacts with one permeating component on the feed side
of the membrane and then diffuses across the membrane to release the permeant
on the product side of the membrane. The carrier agent is then reformed and
diffuses back to the feed side of the membrane. The carrier agent thus acts as a
shuttle to selectively transport one component from the feed to the product side
of the membrane.
Facilitated transport membranes can be used to separate gases; membrane

transport is then driven by a difference in the gas partial pressure across the
membrane. Metal ions can also be selectively transported across a membrane,
driven by a flow of hydrogen or hydroxyl ions in the other direction. This
process is sometimes called coupled transport. Examples of facilitated transport
processes for ion and gas transport are shown in Figure 3-5.
Because the facilitated transport process employs a reactive carrier species,

very high membrane selectivities can be achieved. These selectivities are often
far larger than the selectivities achieved by other membrane processes. This one
fact has maintained interest in facilitated transport for the past 20 years. Yet no
significant commercial applications exist or are likely to exist in the next
decade. The principal problem is the physical instability of the liquid membrane
and the chemical instability of the carrier agent.
Facilitated
transport
Op HEM -- [HEM 91 [HEM021 -HEM+Q
Coupled
transport
cu++ cu++
H+
Cu+*+PHR - CuFt2+2H+ CUR 2 +2H+-Cu+++2HR
Figure 3-5. Schematic examples of facilitated transport of ions and gas. The
gas-transport example shows the transport of 0, across a membrane
using hemoglobin as the carrier agent. The ion-transport example
shows the transport of copper ions across the membrane using a
liquid ion-exchange reagent as the carrier agent.
Introduction 27
3.2 HISTORICAL DEVELOPMENT
Systematic studies of membrane phenomena can be traced to the eighteenth

century philosopher scientists. The Abbe Nolet, for example, coined the word
osmosis to describe permeation of water through a diaphragm in 1748. Through
the nineteenth and early twentieth centuries, membranes had no industrial or
commercial uses. However, membranes were used as laboratory tools to develop
physical/chemical theories. For example, the measurements of solution osmotic
pressure with membranes by Traube’ and Pfeffe? were used by van? Hoff in
1887’ to develop his limit law, explaining the behavior of ideal dilute solutions.
This work led directly to the van? Hoff equation and the ideal equation of state
of a perfect gas. The concept of a perfectly selective semipermeable membrane
was also used by Maxwell and others at about the same time when developing the
kinetic theory of gases.
Early investigators experimented with any type of diaphragm available to

them, such as bladders of pigs, cattle or fish and sausage casings made of animal
gut. In later work collodion (nitrocellulose) membranes were preferred, because
they could be produced accurately by recipe methods. In 1906 Bechhold devised a
technique to prepare nitrocellulose membranes of graded pore size, which he
determined by a bubble-test method.’ Later workers, particularly Elford6,
Zsigmondy and Bachman7, and Ferrys, improved on Bechhold’s technique. By the
early 1930s microporous coilodion membranes were commercially available. During
the next 20 years this early microfiltration membrane technology was expanded to
other polymers, particularly cellulose acetate, and membranes found their first
significant applications in the filtration of drinking water samples at the end of
World War II. Drinking water supplies serving large communities in Germany and
elsewhere in Europe had broken down and there was an urgent need for filters to
test the water for safety. The research effort to develop these filters, sponsored
by the U.S. Army, was later exploited by the Millipore Corporation, the first and
largest microfiltration membrane producer.
By 1960, therefore, the elements of modern membrane science had been

developed. But membranes were used in only a few laboratories and small,
specialized industrial applications. There was no significant membrane industry
and total sales of membranes for all applications probably did not exceed $20
million per year. Membranes suffered from four problems that prohibited their
widespread use: they were too unreliable, too slow, too unselective, and too
expensive. Partial solutions to each of these problems have been developed during
the last 30 years, and as a result there is a surge of interest in membrane-based
separation techniques.
The seminal discovery that transformed membrane separation from a

laboratory to an industrial process was the development, in the early 1960s. of
the Loeb-Sourirajan process for making defect-free, high-flux, ultrathin reverse
osmosis membranes.g These membranes consist of an ultrathin, selective surface
film supported on a microporous support that provides the mechanical strength.
The first Loeb-Sourirajan membranes had fluxes 10 times higher than any
membrane then available and made reverse osmosis a practical technology. The
work of Loeb and Sourirajan, and the timely infusion of large sums of research
dollars from the U.S. Department of Interior, Office of Saline Water (OSW),
resulted in the commercialization of reverse osmosis and was a major factor in
the development of ultrafiltration and microfiltration. The development of
electrodialysis was also aided by OSW funding.
The 20-year period from 1960 to 1980 produced a tremendous change in the
status of membrane technology. Building on the original Loeb-Sourirajan
membrane technology, other processes were developed for making ultrathin, high-
performance membranes. Using such processes, including interfacial
polymerization or multilayer composite casting and coating, it is now possible to
make membranes as thin as 0.1 pm or less. Methods of packaging membranes
into spiral-wound. hollow-fine-fiber, capillary and plate-and-frame modules were
also developed, and advances were made in improving membrane stability. As a
result, by 1980 microfiltration. ultrafiltration, reverse osmosis and electrodialysis
were all established processes with large plants installed around the world.
introduction 29
The principal development of the last 10 years has been the emergence of
industrial membrane gas-separation processes. The first major development was
the Monsanto Prism@ membrane for hydrogen separation, in 19g0.10 Within a few
yeam, Dow was producing systems to separate nitrogen from air and Cynara and
Separex were producing systems for the separation of carbon dioxide from
methane. Gas-separation technology is evolving and expanding rapidly and
further substantial growth will be seen in the 1990s. The final development of
the 1980s was the introduction by GFT, a small German engineering company, of
the fit commercial pervaporation systems for dehydration of alcohol. By 1988,
CR had sold more than 100 plants. Many of these plants are small, but the
technology has been demonstrated and a number of other major applications are
at the pilot-plant scale.
3.3 THE FUTURE
In 1960, the dawn of modern membrane technology. the problems of

membranes were selectivity, productivity/cost, and operational reliability. These
problems remain the focus of membrane research today.
3.3.1 Selectivity
The problem of selectivity i.e., the ability of the membrane to make the
required separation, has been essentially solved in some processes, but remains the.
key problem in others. For example, in 1960, no membranes were known with a
high enough flux to make reverse osmosis an economically viable technology. The
first Loeb-Sourirajan membranes, produced in l%O-63. had high fluxes and were
able to remove 97-98% of the dissolved salt. This development made the process
commercial. By the early 1970s. Riley, at Gulf General Atomic, had improved the
salt-removal capability to 99.5%. l1 By the 1980s. Cadotte had produced interfacial
composite membranes able to remove 99.8-99.9% of the dissolved salt.12 Further
improvements in the selectivity of reverse osmosis membranes are not required.
Similarly. current microfiltration. ultrafiltration and electrodialysis membranes are
generally able to perform the selective separation required of them. On the other
hand, good membrane selectivity remains a generally unsolved problem in the
cases of gas separation and pervaporation. But here too, dramatic strides are
being made. For example, the first commercial air-separation membranes used
conventional polymers such as silicone rubber, ethylcellulose or poly-
trimethylpentene, with oxygen/nitrogen selectivities in the range 2-4. The next
generation of air-separation membranes now entering the marketplace uses
polymers specifically designed for oxygen/nitrogen separation application. These
membranes have selectivities of 6-8. ls More advanced materials. with even higher
selectivities. have already been reported in the literature.
3.3.2 Productivity
It is usually possible to design a membrane system to perform a given

separation. The problem is that a large, complex system, performing under
energy-expensive operating conditions may be required. Thus, productivity, or
separation performance per unit cost, is an issue in all membrane-separation
processes.
There are a number of components to the problem of productivity and cost

of membrane systems, including membrane materials, membrane configuration and
membrane packaging efficiency. Membrane materials with higher intrinsic
permeabilities clearly improve productivity. Similarly thinner, and thus higher-
flux membranes, will reduce overall process costs, as will more economical ways
of packaging these membranes into efficient modules. Having said this, there is a
limit to the reduction in costs that can be achieved. For example, in a modern
reverse-osmosis plant, membrane module costs generally represent only 25-35% of
the total capital cost of the plant, and module replacement costs are not more
than about 10% of the total operating cost. Even major reductions in
membrane/module costs will, therefore. not change the economics of the reverse
osmosis process dramatically. In the case of reverse osmosis, cost reductions
may be more easily achieved by improving nonmembrane parts of the process, for
example, the water pretreatment system. However, in some processes such as
microfiltration, membrane and module costs are more than 50% of the operating
cost. Cost reductions in the membrane/module area would, therefore, be useful in
these processes.
introduction 31
3.3.3 Operational Reliability
Operational reliability is the third and the most generally significant

problem in membrane processes. The causes of reliability problems vary from
process to process. Fouling is a critical factor in ultrafiltration and
microfiltration and therefore dominates the entire membrane operation. Fouling is
also a major factor in reverse osmosis. In gas separation, fouling is usually not a
problem and only minimal pretreatment of the feed stream is required. On the
other hand, in a typical membrane gas-separation process, it is only necessary to
develop one defect per square meter of membrane to essentially destroy the
efficiency of the process. The ability to make, and maintain, defect-free
membranes is, therefore, a key issue in gas separation.
Another factor that leads to operational unreliability is poor membrane

stability. In facilitated-transport membranes, instability is such a problem that
the process has never become commercial. Membrane instability has also proved
to be a major problem area in reverse osmosis, gas separation and pervaporation.
There is no panacea for system reliability. The solution usually appears to

be a combination of a number of factors, such as better membrane materials,
better module designs, improved cleaning and antifouling procedures, and better
process designs. A summary table outlining the relative magnitude of these
problem areas for the seven membrane technologies discussed in this report is
shown in Table 3-2 below.
Table 3-2. Development Status of Current Membrane Technologies
Problems
Process Major Minor Mostly solved Comments
Micro- Reliability cost Selectivity Better fouling control could

filtration (fouling) improve membrane lifetime
significantly.
Ultra- Reliability cost Selectivity Fouling remains the principal

filtration (fouling) operational problem of
ultrafiltration. Current fouling
control techniques are a
substantial portion of process
costs.
Reverse Reliability Selectivity Cost Incremental improvements

osmosis in membrane and process design
will gradually reduce costs.
Electro- Fouling cost Selectivity Process reliability and

dialysis Temperature Reliability selectivity are adequate for
stability current uses. Improvements
could lead to cost reduction,
especially in newer applications.
Gas Selectivity cost Reliability Membrane selectivity is the

separation Flux principal problem in many
gas separation systems.
Higher permeation rates would
help to reduce costs.
P9rvaporation Selectivity cost Membrane selectivities must

Reliability be improved and systems
developed that can reliably
operate with organic solvent
feeds before major new
applications are commercialized.
Coupled and Reliability Membrane stability is an

Facilitated (membrane unsolved problem. It must be
Transport stability) solved before this process can
be considered for commercial
applications.
Introduction 33
REFERENCES
I. H.K. Lonsdale. U. Merten and R.L. Riley, “Transport Properties of Cellulose Acetate
Osmotic Membranes.” -1. Polv. u, 1344 (l%5).
2. M. Traube, Arch.. .
_ PhvUpL. Leipzig (1867).
3. W. Pfeffer, pUntenuchunnen, Leipzig (1877).
4. J.H van? Hoff, Z. . 481 (I 887).
5. H. Bechhold, IJollid 7.-I,, 107 (1906) and Biochem., 379 (1907).
6. W.J. Elford. Trans.. 1094 (1937).
7. Zsigmondy and Bachmann. wInorn. I I9 (1918).
a. J.D. Ferry, “Ultrafiltration Membranes and Ultrafiltration,” &~pjgal Rev. 18, 373 (1935).
9. S. Loeb and S. Sourirajan, “Sea Water Demineralization by Means of an Osmotic

Membrane.’ in ,$&ie Water Con &cut-II. Ad ws in -Series Numbe r 28,
American Chemical Society, Washiigton. D.C. (19;[3).
10. J.M.S. Henis and M.K. Tripodi, “A Novel Approach to Gas Separation Using Composite
Hollow Fiber Membranes,’ .&Q, Sci. & Tech, u. IO59 (1980).
Il. R.L. Riley, H.K. Lonsdale, D.R. Lyons and U. Merten, “Preparation of Ultrathin Reverse
Osmosis Membranes and the Attainment of the Theoretical Salt Rejection,” 1. ADDI. Polvm.
u. 2143 (1967).
12. J.E. Cadotte and R.J. Petersen, “Thin-Film Composite Reverse-Osmosis Membranes: Origin,
Development, and Recent Advances,” American Chemical Society, &~&tic Membranes
.
I De&lg&pg, A.F. Turbak, Ed., Washington, D.C. (1981).
13. J.N. Anand. S.E. Bales, D.C. Feany and T.O. Janes, U.S. Patent 4,840.646, June (1989).
4. Government Support of Membrane Research
4. I OVERVIEW
Membrane science originated in Europe and many of the fundamental laws

and equations of membrane science bear the names of European scientists,
Graham’s Law, Fick’s Law, the van? Hoff equation, the Donnan effect and so on.
European dominance of the field lasted until the early 19X& when a new
membrane industry, centered in the United States, began. Federal research
support played a critical role in the early growth of this industry. Millipore. now
the world’s largest microfiltration company, got its start out of a U.S. Army
contract to develop membrane filters. The reverse-osmosis and electrodialysis
industries received even more significant levels of support from the Office of
Saline Water from 1960 to 1975. Poor drinking-water quality in the southern and
southwestern states. plus the possibility of increasing water supplies to arid
regions, were seen as problems that could be addressed by the newly emerging
membrane technology. Despite the fact that no membrane industry as such
existed, the U.S. Government made a far-sighted commitment to the new
technology. As a result, the industry received an average of between 520-40
million per year (in 1990 dollars) for membrane research over a period of I5
years.
During this “Golden Age”, hollow fibers, spiral-wound modules, asymmetric

membranes, thin-film composites and all the other basic components of current
membrane technology were developed. Not only did reverse osmosis and
electrodialysis research rely almost completely on the flow of Federal research
monies, but the ultrafiltration industry, and to a lesser extent the microfiltration
industry. also received considerable assistance from the fallout of this support.
Finally a significant invention, the spiral-wound module, tightly patented and
licensed gratis by the Government to U.S. companies, was decisive in maintaining
U.S. dominance over reverse-osmosis markets through the 1970s. Few outside the
industry appreciate the importance of these patents in blocking non-U.S. firms.
34
Government Support of Membrane Research 35
In 1975 the Office of Saline Water closed and there was a substantial
reduction in the level of Federal membrane research support, from $40 million
per year ( 1990 dollars) to the present level of SlO- I 1 million. The demise of the
Office of Saline Water coincided with a surge of interest in the membrane
industry in Japan and Europe.
In Europe and Japan there is a significant amount of government research

support to academic institutions and to private industry. Furthermore, the level
of support appears to be growing. The approximate levels of support in the
United States, Japan and Europe are summarized in Table 4-1.
Table 4-I. Government Membrane Support
Level of Support
(S millions/vear)
united !j&&g DOE: Office of Industrial Programs 1.5

Office of Basic Energy Research 1.0
Office of Fossil Energy 1.0
SBIR Programs 1.0
NSF 4.0
EPA 1.5
NASA
DOD g
Total 11.0
Ministry of Education:
Membrane Support to Universities 2.0 (est.)
MITI: Basic Industries Bureau 2.0 (est.)

AIST - Jisedai Project 2.0 (est.)
Aqua Renaissance ‘90 5.0
WRPC
NEDG %
Total 19.0
EuroDe National Programs for University Support 10.0 (est.)

National Membrane Programs:
Holland 2.0
U.K. 1.5
IdY
EEC (BRITE) Program z (est.)
Total 20.0
The numbers in this table should be treated with caution. The U.S.
numbers are fairly reliable, as are the numbers for the foreign, individually
designated programs. Numbers labeled “estimated” are however, just that and are
not reliable to better than 30%. Currently it appears that total U.S. Government
funding for membrane research is approximately SlO-11 million per year. compared
to approximately S19 million per year in Japan and S20 million per year in Europe.
In part because of the significant amount of research support that Japanese

and European companies have received, the dominant position that the U.S.
membrane industry enjoyed in world markets until 1980 has been eroded.
Japanese companies have largely recaptured their domestic markets in reverse-
osmosis, ultrafiltration and electrodialysis. Japanese companies now compete
strongly with U.S. suppliers in the areas of reverse osmosis and electrodialysis in
the Middle East. In the U.S. and Europe, Japanese companies have been less
successful. After failing to establish their own subsidiaries in the U.S., they are
beginning to enter the market by acquiring U.S. companies. For example, Nitto
Denko, a major Japanese reverse-osmosis and ultrafiltration company, recently
acquired Hydranautics, the third or fourth biggest U.S. reverse-osmosis company.
Toray Industries, another large Japanese firm, has also tried to acquire a U.S.
reverse osmosis company.
European companies have been less successful than the Japanese in

capturing their home markets and in competing overseas. There are a number of
significant European membrane companies, but they have not succeeded in
displacing American companies from their dominant position in ultrafiltration,
reverse osmosis and electrodialysis.
In gas separation and pervaporation, which represent the emerging membrane

industry, the commercial markets are still fluid. In gas separation, U.S.
companies are ahead. In pervaporation, European and Japanese companies lead,
with the United States trailing significantly behind. The extent of future
government support to the universities and to industry will have a significant
effect on the final U.S. position in these technologies.
4.2 U.S. GOVERNMENT SUPPORTED MEMBRANE RESEARCH
The current level of support of membrane-related research by the U.S.

Government is of the order of Si 1 million annually. The Department of Energy,
which funds energy-related membrane research and development. is one of the
significant sources of U.S. Government support. The NationaI Science Foundation
is the other major source of support, particularly for academic institutions and
others carrying out fundamental research in membrane science. Other sources of
funding include the Environmental Protection Agency, the Department of Defense,
the National Aeronautics and Space Administration and the Department of
Agriculture, which support research and development of membrane separation
systems that are relevant to the specific mission of the department or agency.
4.2.1 Department of Energy
The U.S. Department of Energy supports membrane separations research and

development via several programs. The emphasis in all of these programs is on
devices and processes that have the potential for high energy savings. The
current level of funding of the DOE’s membrane research and development
programs is between 54.3-4.5 million annually. The most significant of these
programs is the Industrial Energy Conservation Program. This program is
sponsored by the Division of Improved Energy Productivity of the Office of
Industrial Programs in the Office of Conservation and Renewable Energy.
4.2.1.1 Office of Industrial Programs/Industrial Energy Conservation Program
The mission of the Office of Industrial Programs is to increase the end-use

energy efficiency of industrial operations. The Industrial Energy Conservation
Program, administered by this office, is designed to fund research and
development of high-risk, innovative technologies to increase the energy
efficiency of industrial operations. Federal funding can accelerate industry’s
acceptance of a new technology by alleviating some of the risk associated with
commercialization. Research and development of membrane separation processes
for the paper, textile, chemical and food-processing industries have been funded
by this program since 1983. The current level of support is of the order S1.5
million per year. Table 4-2 contains a list of the specific projects and the
contractors.
Table 4-2. Membrane R&D Funded through the Office of

Industrial Programs since 1983
contractor Topic
Air Products & Chemicals, Inc. Active transport membranes
Alcoa Separations Catalytic membrane reactor
Allied-Signal Corp. Fluorinated membranes
Allied-Signal Corp. Membranes for petrochemical applications

with large energy savings
American Crystal Sugar Co. and Concentrating hot, weak sugar-beet juice
the Beet Sugar Develop. Found.
Bend Research, Inc. Membrane-based industrial air dryer
Bend Research. Inc. Membrane separation system for the corn

sweetener industry
Carre, Inc. Dynamic membranes to reclaim hot dye

rinse water
EGBG. Inc. Polyphosphazene membranes
EG&G. Inc. Assessment of membrane separations in the

food industry
Filmtec Corp. Temperature-resistant, spiral-wound

elements
HPD. Inc. Electrolysis of Kraft Black Liquor
Ionics. Inc. An electro-osmotic membrane process
Mavdil Corp. Membrane for concentrating high solubles

in water from corn wet milling
Membrane Technology & Research, Removal of heat and solvents from

Inc. industrial drying processes
National Food Processors Assn. Develop energy-efficient separation,

concentration and drying processes for
food products
Table 4-2 continued
Contractor Topic
National Food Processors Assn. Hyperfiltration as an energy conservation

technique for the renovation and recycle
of hot, empty container wash water
Physical Sciences, Inc. Reduced energy consumption for the

production of chlorine and caustic soda
SRI International, Inc. Piexoelectric membranes
SRI International, Inc. Hybrid membrane systems
State University of New York Energy-efficient, high-crystalline, ion-

exchange membranes for the separation of
organic liquids
State University of New York Membrane dehydration process for producing

high grade alcohols
University of Maine Ultrafiltration of Kraft Black Liquor
University of Wisconsin Colloid-chemical approach to the design of

phosphate-ordered ceramic membranes
UOP. Inc. A membrane oxygen-enrichment system
4.2.1.2 Office of Energy Research/Division of Chemical Sciences
The Division of Chemical Sciences of the Office of Basic Energy Sciences in

the Office of Energy Research funds fundamental research into membrane
materials and membrane transport phenomena. The objective of this support is
to add to the available knowledge regarding membrane separations. The funds
are primarily directed towards research at universities and the National
Laboratories. The Division of Chemical Sciences spends $500,000 per year on
membrane-specific research and another $500,000 per year on peripheral research
fundamental to the understanding of membrane transport. Some industrial
research is also supported, but is administered through the Small Business

Innovative Research Program (SBIR). Table 4-3 is a list of typical projects
supported by the Division of Chemical Sciences.
Table 4-3. Membrane R&D Funded through the Division of Chemical Sciences
Contractor Topic
Brigham Young University Novel macrocyclic carriers for proton-

coupled liquid-membrane transport
Lehigh University Perforated monolayers
University of Oklahoma A study of micellar-enhanced ultrafiltration
Syracuse University Mechanisms of gas permeation through

polymer membranes
University of Texas Synthesis and analysis of novel polymers

with potential for providing both high
permselectivity and permeability in gas
separation applications
Texas Tech University Metal ion complexation by ionizable crown

ethers
4.2.1.3 Office of Energy Research/Division of Advanced Energy Projects
The Division of Advanced Energy Projects within the Office of Energy

Research complements the role of the Division of Chemical Sciences. Most of
the projects supported involve exploratory research on novel concepts related to
energy. The typical project has both very high risk and high payoff potential,
and consists of concepts that are too early lo qualify for funding by other
Department of Energy programs. The support is sufficient to establish the
scientific feasibility and economic viability of the project. The developers are
then encouraged to pursue alternative sources of funding to complete the
commercialization of the technology. The Division does not support ongoing,
evolutionary research. Table 4-4 is a list of projects supported by the Division of
Advanced Energy Projects during the past five years. At present, the Division is
funding one membrane project on the separation of azeotropes by pervaporation at
a level of 5150.000 per year.
Table 4-4. Membrane R&D Funded through the

Division of Advanced Energy Projects since 1983
Contractor Topic
Bend Research, Inc. The continuous membrane column; a low

energy alternative to distillation
Bend Research, Inc. Liquid membranes for the production of

oxygen-enriched air
Membrane Technology & Research, Pervaporation: A low-energy alternative to

Inc. distillation
Membrane Technology 8 Research, Separation of organic axeotropic mixtures

Inc. by pervaporation
Portland State University Thin-film composite membranes for

artificial photosynthesis
4.2. I .4 Office of Fossil Energy
The Office of Fossil Energy supports research and development related to

improving the energy efficiency of fossil-fuel production and use. The projects
are typically administered through the Morgantown and Pittsburgh Energy and
Technolosy Centers (METC & PETC). Membrane projects related to improved
combustion processes and fuel and flue-gas cleanup are supported by the Gas
Stream Cleanup and Gasification programs at METC and by the Flue Gas Cleanup
program at PETC. The support for these programs amounts to about S1.0 million
per year. Representative research projects are listed in Table 4-5.
Table 4-5. Membrane R&D Funded through the Office of

Fossil Energy since 1983
Contractor Topic
Air Products & Chemicals, Inc. High-temperature, facilitated-transport

membranes
Alcoa Separations Alumina membrane for high temperature

separations
California Institute of Technology Silica membranes for hydrogen separation
Jet Propulsion Laboratory Zirconia cell oxygen source
Membrane Technology & Research, Low-cost hydrogen/Novel membrane

Inc. technology for hydrogen separation from
synthesis gas
Membrane Technology & Research, Development of a membrane SO,/NO,

Inc. treatment system
METC (in- house) Ceramic membrane development
National Institute for Standards & Gas separation using ion-exchange

Technology membranes
Oak Ridge National Laboratory Gas separation using inorganic membranes
SRI International Catalytic membrane development
SRI/PPG Industries Development of a hollow fiber silica

membrane
Worcester Polytech. Institute Catalytic membrane development
4.2. I .5 Small Business Innovative Research Program
The Small Business Innovative Research Program (SBIR) was initiated by

Congress in 1982 to stimulate technological innovation in the private sector and
strengthen the role of small business in meeting Federal research and
development needs. A greater return on investment from Federally funded
research as well as increased commercial application are the other expected
benefits from this program. The program consists of three phases and is open
only to small businesses. Phase I is typically a six-month feasibility study with

funding up to S50,OOO. If approved for follow-on funding, the project enters a
two-year Phase II stage, of further development and scale-up, with support of up
to S500.000. A final non-funded stage, Phase III, consists of commercial or third-
party sponsorship of the technology and represents the entry of the technology
into the marketplace.
This program encompasses topics of interest to a number of subdivisions of

the Department of Energy, including the Office of Fossil Energy (METC & PETC),
the Office of Energy Research and the Office of Conservation and Renewable
Energy. During 1989, the DOE-SBIR program supported two Phase II projects and
five Phase I projects, totalling S750.000 per year. Table 4-6 contains a list of
the projects that have been supported under this program.
Table 4-6. Membrane-related SBIR Projects since 1983
Year Phase Contractor Topic

initiated I II
1983 X Membrane Technology & Research, Novel liquid ion-exchange extraction

Inc. process
1983 xx Rend Research, Inc. Concentration of synfuel process

condensates by reverse osmosis
1983 xx Bend Research, Inc. Solvent-swollen membranes for the removal

of hydrogen sulfide and carbon monoxide
from coal gases
1984 xx Rend Research, Inc. Thin-film composite gas separation

membranes prepared by interfacial
polymerization
1984 X Membrane Technology & Research, Improved coupled transport membranes

Inc.
1985 X Magna-Seal, Inc. Perfluorinated crosslinked ion-exchange

membranes
1985 X Membrane Technology & Research, Plasma-coated composite membranes

Inc.
1985 X Merix Corp. Improved hydrogen separation membranes
1985 X Process Research & Development. Separation of oxygen from air using
Inc. amine-manganese complexes in
membranes
1985 xx Bend Research, Inc. A membrane-based process for flue gas

desulfuriaation
1986 X Foster-Miller, Inc. A high-performance gas separation

membrane
1987 xx Bend Research, Inc. Novel high-flux antifouling membrane

coatings
1987 xx Bend Research, Inc. High-flux. high-selectivity cyclodextrin

membranes
1988 xx Spire Corp. Novel electrically conductive membranes

for enhanced chemical separation
Table 4-6. continued
Year Phase Contractor Topic

initiated I II
1988 X Texas Research Institute Synthesis of new polypyrrones and their

evaluation OS gas separation membranes
1988 X X CeraMem Corp. A ceramic membrane for gas separations
1989 Cape Cod Research, Inc. A molecular recognition membrane
1989 CeraMem Corp. Low-cost ceramic support for high-

temperature gas separation membranes
1989 CeraMem Corp. Low-cost ceramic ultrafiltration

membrane module
1989 KSE. Inc. Chlorine-resistant reverse osmosis

membrane
1989 Coury & Associates Novel surface modification approach to

enhance the flux/selectivity of polymeric
membranes
1989 Membrane Technology & Research, Membranes for a flue gas treatment
Inc. process
1989 Membrane Technology & Research, Novel membranes for natural gas liquids
Inc. recovery
4.2.2 National Science Foundation
The National Science Foundation (NSF) supports fundamental research in

membrane separations both at universities and in industry through research
grants and SBIR awards. The level of funding of the NSF membrane research
program is comparable to that of the DOE (54 million dollars annually) although
the mission of these two programs is quite different. Unlike the DOE, which
funds energy-related research with an emphasis on the development of viable
technology. the NSF funds exploratory research and fundamental studies that
increase the understanding of the transport phenomena in membranes.
The Division of Chemical and Thermal Systems currently funds 50-60

projects per year in membrane-related research. The average project receives
about $60,000 per year and the total value of the program is S3.5 million. The
projects funded are fundamental studies of the basics of membrane science and
membrane materials. Although this work is important to the understanding and
use of membrane separation processes, not all of it is relevant to the energy
conservation issues addressed in this report.
A new program jointly administered by the Divisions of Life Sciences and

Chemical and Thermal Systems, has been set up to fund membrane-related
research in biotechnology at a rate of S500.000 per year. Most projects will
receive $60,000 per year. with one or two group awards of $200,000 per year.
Research in polymer and inorganic materials funded by the Division of

Materials Research also contributes to the body of knowledge on membranes.
4.2.3 Environmental Protection Agency
The Environmental Protection Agency (EPA) supports membrane separation

system research and development primarily through the SBIR and the Superfund
Innovative Technology Evaluation (SITE) programs. The research funded is
related to EPA’s mission of reduction, control and elimination of hazardous
wastes discharged to the environment. The current level of funding for
membrane-related research is of the order of $1.3 million per year.
The SBIR program currently supports projects investigating the use of

membranes for the removal of organic vapors from air and the removal of
volatile organic contaminants from aqueous streams. The present level of
funding in the SBIR program is on the order of $750,000 per year.
The SITE program was set up as part of the Superfund Amendments and
Reauthorization Act of 1986 (SARA). It is administered by the EPA’s Office of
Solid Waste and Emergency Response and the Office of Research and Development.
The Emerging Technologies Program (ETP), a component program of SITE, is
designed to assist private developers in commercializing alternative technologies
for site remediation. The research projects funded through the ETP are typically
bench- and pilot-scale testing of new technologies and are funded at a level of
5150,000 per year. The three membrane-related projects being funded through the
ETP are listed in Table 4-7.
Table 4-7. Membrane R&D Funded through the Emerging Technologies Program
Contractor Topic
Atomic Energy of Canada Ltd. Ultrafiltration of metal/chelate complexes

from water
Membrane Technology & Research, Removal of organic vapor from

Inc. contaminated air streams using a membrane
process
Wastewater Technology Center Cross-flow pervaporation system for the

removal of Voc’s from aqueous wastes
4.2.4 Department of Defense
The Department of Defense (DOD) funds a small number of membrane

separations research projects through its SBIR program. These projects address
specific strategic and tactical needs of the DOD, but are also applicable to
industrial separations. Examples of such research are:
. Chlorine-resistant hollow-fiber reverse osmosis elements for portable
desalination units
. Membranes for on-board water generation from vehicular exhausts
. Membrane oxygen extraction units for providing breathable air in chemically
contaminated environments
. Polymeric and liquid membranes for the extraction of oxygen from seawater
As the type of research and level of support is governed by the current

needs of the DOD, there is no specific program for membrane research.
Consequently. funding is small and intermittent.
4.2.5 National Aeronautics and Space Administration
The National Aeronautics and Space Administration (NASA) has funded a few
membrane-related projects through the SBIR program. These projects are
oriented toward NASA’s mission in space and consist of new technology for life
support systems in space. The areas of research supported are:
. Membrane systems for removal and concentration of carbon dioxide in the
space vehicle cabin atmosphere
. Membrane systems for water recovery and purification
Since the type of research and level of support is governed by the current
needs of NASA, there is no specific program for membrane research.
Consequently, funding is small and intermittent.
4.3 JAPANESE GOVERNMENT SUPPORTED MEMBRANE RESEARCH
Although the dates of origin of the membrane industries in Japan and the
U.S. differ by about 20 years, in many ways the experiences of the two countries
are similar. ‘The Japanese government continues to support a large research
effort in membranes that began in the 1970s. A number of programs will begin
to expire in the early 1990s. but will undoubtedly be replaced by others,
although their size may decrease and their focus change. A reduction in
government support would reflect the current size and status of the Japanese
membrane industry. Some leading Japanese companies no longer participate
directly in Government-sponsored programs. They prefer to support research
efforts with their own funds, in this way maintaining an edge over their
competition. Having said this, the total level of Government membrane research
support is currently twice the U.S. Government level.
Japan sponsors a variety of programs that support membrane research and

development. A few are direct most are indirect. The Ministry of Education,
for example, does not have a membrane program per se, but membrane programs
are included in the support of educational research. Aqua Renaissance ‘90, an
agency of the Ministry of International Trade and Industry (MITI). supports work
on membranes as a means to achieve its goals in the area of water re-use. Other
agencies also support membrane research and development as an opportunity to
develop domestic products that will displace foreign imports and will ultimately be
exported to world markets.
4.3.1 Ministry of Education
Academic research is sponsored by general grants to faculty. and by specific

research programs with relevance to the membrane field. Ministry of Education
programs are said to be primarily for the training of students, with little regard
for the utility of the research in the near term. Pervaporation membrane
research has been a particularly active area recently. The general level of this
support is estimated at $2 million annually.
4.3.2 Ministry of International Trade and Industry (MITI)
MIT1 sponsors research and development projects thought to have medium-

term practical significance. Several membrane-related projects are included in
the program. Some of the agencies and departments of MIT1 known to be
sponsoring membrane work are listed below.
4.3.2.1 Basic Industries Bureau
This agency sponsors a project on membrane dehydration of alcohol

(dehydration of azeotropes). The program began when GFT started selling
pervaporation plants in Japan. Many separations are potential candidates for
pervaporation technology. The program’s goal is to develop superior technology.
Its main focus has been on membranes, particularly those derived from chitosan,
to make water-permeable dehydration membranes. Recently. three companies,
Sasakura Engineering, Tokuyama Soda and Kuraray, announced that they had
independently developed chitosan-based pervaporation membranes, whose properties
are said to be competitive with GFf membranes. Details have not yet been
revecled. although some of this work is now beginning to appear in the U.S.
patent literature.
4.3.2.2 Agency of Industrial Science and Technology (AIST)
AIST is responsible for the National Laboratories, two of which have active
membrane programs. The Government Industrial Research Institute (Osaka) is
often mentioned in reports of membrane research. Programs are also under way
at the National Chemical Laboratory for Industry (Tsukuba). AIST also conducts a
project for revolutionary basic technologies, formally known as Research and
Development Project of Basic Technologies for Future Industries, popularly known
as the Jisedai Project. One of fourteen categories targeted for development is
“Synthetic Membranes for New Separation Technology.” Included are efforts to
develop high-performance pervaporation and gas-separation membranes. This work
is the responsibility of National Chemical Laboratory for Industry (Tsukuba), and
has been performed at the Research Institute for Polymers and Textiles (AIST),
the Industrial Products Research Institute (AIST), and at the Research Association
for Basic Polymer Technology, an organization of 10 private companies and two
universities.
Another AIST-sponsored project is the National Research and Development

Program. Nine projects considered particularly important and urgent for the
nation are under development. One of these is the New Water Treatment
Program, known generally as Aqua Renaissance ‘90. The annual budget of the
membrane program is in the region of 54-5 million. This project is aimed at
developing new ways to treat wastewater from a variety of sources (municipal,
starch processing, etc.) in the Japanese context. One very important
consideration in any Japanese waste-treatment facility is the plant footprint.
Land in Japan is at a premium, so conventional secondary sewage treatment was
eliminated at the outset of Aqua Renaissance ‘90 as requiring too much land.
Membranes fit well into plans to build a new type of waste-treatment facility.
Japan’s lack of indigenous fuel also makes the production of methane from its
wastes attractive. Thus the combination of anaerobic digestion and membrane
concentration looked particularly attractive. The effort is funded at a level high
enough to work out the problems and try the needed equipment. Whether this
work will result in a new way of treating wastes remains to be seen. What is
obvious is that the state of the membrane art generally has been advanced
significantly as a result of the program.
The Aqua Renaissance ‘90 idea is not a novel concept. Dorr-Oliver worked
on essentially the same approach for many years. They did not have the
resources to solve all the problems and achieve commercial success. The problem
proved too big and too complex for that one company to solve. If the project is
a significant success, Japan stands to gain substantial external markets, because
wastewater treatment is a ubiquitous problem. Many large cities throughout the
world would be interested in replacing their existing sewage-treatment plants
with high-efficiency, low-land-use alternatives.
4.3.2.3 Water Re-use Promotion Center (WRPC)
The WRPC is an incorporated foundation, chartered by, and partially funded

by, MI-l-I. It was set up in 1980 to promote water saving. Its activities involve
desalination, water re-use, training and performance testing of membrane systems.
It lists approximately 100 members, including local government and water
authorities, engineering companies, manufacturing companies, banks and insurance
companies. It has about 20 permanent employees and about 33 more on temporary
assignment from their employers. These people are paid by WRPC and do training
assignments as well as assessments sponsored by Japan International Cooperation
Agency, usually as part of Japan’s foreign aid program. The annual budget is
approximately 66 million. Major membrane-related projects conducted by WRPC in
the year ending March, 1988, included:
Experiments for establishing seawater desalination technology by reverse

osmosis.
Using solar cells to power reverse osmosis desalination systems.
Electrodialysis for seawater desalting utilizing solar cells.
Experiments for establishing a new technology for ultra-pure water

production.
Experiments on removal of malodor and color using activated carbon fiber.
Studies on effective use of industrial water.

4.3.2.4 New Energy Development Organization (NEDO)
NED0 is an MITI-funded foundation established in 1980. Its charter is to

consider alternatives to petroleum for energy supply. Recently, NED0 activities
have been enlarged to involve all industrial technology. One of NEDO’s
programs. the Alcohol Biomass Energy Program, contains a project for
development of membranes to maintain high densities of methanogenic bacteria,
and development of modules for employing them. Their interest extends beyond
this project to the broader area of water re-use.
4.3.3 Ministry of Agriculture, Forestry and Fisheries
This ministry is active in the membrane area through promotion of the use
of membranes, particularly reverse osmosis and ultrafiltration membranes, in the
food industry. There is a current program on chemical conversion of biomass
involving membranes and a completed project on wastewater treatment for the
food industry.
4.4 EUROPEAN GOVERNMENT SUPPORTED MEMBRANE RESEARCH
Europe is a significant importer of industrial membrane separation

equipment and a major market for U.S. industrial membrane manufacturers,
particularly microfiltration and ultrafiltration equipment suppliers. Pall, Millipore
and Koch Membrane Systems all derive significant benefit from their activities in
Europe. There are also strong European companies, however, in the areas in
which the Americans have traditionally been most successful (DDS, Sartorius, PCI.
S & S. Rhone Poulenc). The U.S. position could change.
In the emerging field of pervaporation. GlT, the German subsidiary of a

French company, is the undisputed world leader at present.
4.4.1 European National Programs
European membrane research groups receive support from their owo national
governments and from the multinational groupings such as the European Economic
Community (EEC). The amount of support given by national programs is difficult
to track because it is hidden in the general funds given to the universities. A
recent survey by the European Membrane Society identified a total of 79
universities and institutes where there were significant membrane science and
technology research programs. Some of these groups are very large, for example,
the groups at the University of Twente (Holland), GKSS Geesthacht (West
Germany) and the Fraunhofer Institute of Stuttgart (West Germany). These
groups each have more than 20 research students and staff and budgets of
several million dollars. Other groups are undoubtedly smaller and may consist of
a professor and one or two students, with a budget of SlOO,OOO or less. We
believe that an estimate of SIO million disbursed by various national Government
Ministries of Education and Science to support membrane research in academia is
conservative. This estimate is in accord with an intuitive sense of the relative
size of the European and American academic interests in membranes. In addition
to this general support, there are some specific national membrane programs aimed
at industry and academic groups. The more important of these groups are
discussed below.
l The Dutch Innovative Research Program on Membrane Technology. This

project funded over seven years at S2 million per year is aimed at producing new
membranes for gas separation, pervaporation and ultrafiltration. Membrane fouling
is another topic area.
. The United Kingdom Science and Engineering Council Program. This five-year
program has an annual budget of Sl.5 million. Research is aimed at a wide range
of basic and applied membrane topics.
l The Italian National Project in Fine Chemicals. This program, with a budget
of S2.5 million annually, supports 20 academic and industrial teams working in the
membrane area.
4.4.2 EEC-Funded Membrane Research
In addition to the national membrane programs, membrane-research support

is available through the EEC. The most important program to the membrane
community is the Basic Research in Industrial Technologies for Europe (BRITE)
program. The BRITE program is now in its second term. Within select research
areas, projects within the Community may be subsidized up to 50%. All projects
must have a sponsor in at least two member states, one of which must be
industrial. Membranes were one of the areas selected for particular emphasis.
The countries participating are France (F), the Netherlands (NL), the United
Kingdom (UK). Italy (I), West Germany (D), Denmark (DK) and Spain (E).
Amongst the topics funded were:
. Gas separation membranes for upgrading methane containing gases to
pipeline quality. [Gerth (F) and Nederlandsse Gasunie (NL).]

l Gas separation membranes for separation of CO, and H&S from natural gas.
[Akzo (NL) and Elf Aquitaine (F) and University of Twente (NL).]
l Development of cross-flow microfiltration membranes for the
biotechnology industry. [Tech Sep (F) and Advanced Protein Products (UK)
and University of Loughborough (UK).]
. Development of inorganic and ceramic membranes for gas separations.
[Eniricerche SPA (I) and Enichem (I), Harwell Laboratory (UK), Esmill
Water Systems (NL), Hoogovens Groep (NL) and ECN (NL).]
l Application of membranes to the textile industry. [Separem (I),
Peignage D’Auchel (F), Texilia (I). Fraunhofer Institute (D) and University of
Calabria (I).]
l Integrated ultrafiltration and microfiltration membrane processes. [DDS (DK),
Sot. Lyonnaise des Eaux (F). University College Wales (UK), Technical
University of Denmark (DK) and Imperial College, London (UK).]
l The use of membranes to treat olive oil wastewater. [Inst. Ricerche Breda (I),
Separem SPA (I), Labein (E), Pridesa (E) and Centro Richerche Bonomo (I).]
l Acid-stable pervaporation membranes. [BP Chemicals (UK), GFf (D),
RWTH (D). University of Twente (NL) and University of Kaln (D).]
4.5 THE REST OF THE WORLD
The portion of the industrial membrane industry outside of the U.S., Europe
and Japan is negligible, except for a surprisingly vigorous program in Australia.
There are three Australian-based membrane companies, Memtec, Syrinx and
Aquapore. Of these, Memtec is the largest, with about 130 Australian employees
and, since their acquisition of Brunswick Filtration Division in 1988, a substantial
presence in the U.S. Memtec produces microfiltration equipment largely centered
on water pollution control applications.
The Australian government is sponsoring membrane research at the level of

about $1 million annually.
5. Analysis of Research Needs
5.1 PRIORITY RESEARCH TOPICS
Based on group meetings and review discussions, a list of five to seven

important research topics was selected for each of the seven major membrane
technology areas. This list, totaling 38 research topics, was then ranked in
order of priority. The list is shown in Table 5-l. together with the ranking
scores assigned by each group member. A topic ranked number 1 received 38
points in the score column, a topic rated number 2 received 37, and so on.
Since the review group had six members and there were 38 topics, the maximum
possible score for any topic was 6x38 or 228.
A few points should be made about this priority list. First, although the
research interests of the six author group members are completely different
(this, in fact, was the basis for their selection), the priority rankings that they
assigned were remarkably similar. Most of the group members had one or two
topics, out of the 38, that they ranked particularly high or low compared to the
average ranking. The deviations of the group member’s individual rankings from
the average ranking were, however, generally small. The standard deviation
shown in the last column reflects the scatter between the individual group
member’s ranking of each topic. In general, the scatter was least at the top and
bottom of the tables, reflecting good agreement between the group members on
the most and least significant research topics. Not unexpectedly, there was most
scatter in the middle range. Based on these scores, the top 10 priority research
topics were selected. These topics are listed, with brief descriptive comments, in
Table 5-2.
56
Analysis of Research Needs 57
Table 5-1. Important Research Topics for the Seven Membrane Technology Areas,
Ranked in Priority Order
Table 5-2. Priority Research Topics in Membrane Separation Systems
Rank Research Topic CornmenU Score
Pervaporation If sufficiently selective membranes could be made, 201

membranes for organic- pervaporation could replace distillation in many
organic separations separations
Reverse Osmosis Commercial polyamide reverse osmosis membranes 187

oxidation-resistant rapidly deteriorate in the presence of oxidizing agents
membrane such as chlorine, hydrogen peroxide, etc. This
deficiency has slowed the acceptance of the process in
some areas.
Gas Separation Would allow broad usage of advanced materials - even 184
development of better if done in hollow fibers
generally applicable
method for producing
membranes with ~OOA
skins
Facilitated Transport Air separations of higher selectivity are a target I82

oxygen-selective solid common to all types of membranes
facilitated transport
membranes
Gas Separation Selectivity of 8-10 and permeability of 10 Barrer is 180

higher 0,/N, selectivity required. Experimental materials approach these, but
productivity polymer no ability to spin form them in hollow-fiber form has
been reported. Most valuable as hollow fibers.
Ultrafiltration Fouling is a ubiquitous problem in UF. Its elimination 179

fouling-resistant would boost total throughput >30% and reduce capital
membranes costs by 15% on top of eliminating cleaning. Better
fractionation would also result, expanding UF use
significantly.
Pervaporation Current modules cannot be used with organic solvents 167

better solvent- and are also very expensive
resistant modules
Rank Research Topic Comments Score
8 Gas Separation Only small amounts of the valuable selective material 164
development of a are required
generally applicable
method for forming com-
posite hollow fibers
with <SWA skins
9 Microfiltration Opportunity for ceramic or inorganic membranes. 151

high-temperature, Potential uses include removal of particulates from
solvent-resistant coal liquids and replacement of bag houses in flue gas
membranes and modules treatment
IO Microfiltration Huge potential applications will require commodity IS1

low-cost membrane pricing, far from today’s reality.
modules
Tbe highest ranked research topic was pervaporation membranes for organic-
organic separations. A closely related topic, solvent-resistant pervaporation
modules, ranked seventh in the priority list. The very high ranking of these two
pervaporation research topics reflects the promise of this rapidly developing
technology. The separation of organic mixtures by distillation consumes two
quads of energy in the U.S. annually.’ In principle, pervaporation could be used
to supplement many existing distillation operations, for example by treating the
top or bottom fractions from the distillation column. In some applications, such
as ethanol/water separation or separation of organic/organic mixtures that form
azeotropes at certain concentrations, pervaporation might displace distillation if
appropriate membranes and equipment were available. If even a conservative 10%
of the present energy expenditure on distillation were saved, this would represent
0.2 quads annually, or IO5 barrels of oil daily.
The principal problem hindering the development of commercial pervaporation

systems is the lack of membranes and modules able to withstand solvents at the
elevated temperatures required for pervaporation. These problems can be solved.
The development of membrane modules for a few special applications, for example,
the removal of methanol from isobutene methyltertbutyl ether (MTBE) mixtures, is

already at the pilot-plant stage. Research breakthroughs in pervaporation appear
imminenS widespread applications of the process could occur within the next
decade if adequate research support were available. The impact on the nation’s
energy usage by the year 2010 could be substantial.
The second priority topic is the development of oxidation-resistant reverse

osmosis membranes. The current generation of polyamide, high-performance
reverse osmosis membranes have salt rejections of greater than 99.5% and fluxes
three to five times higher than the cellulose acetate membranes developed in the
1970s. However, these membranes have not displaced cellulose acetate membranes
because of their susceptibility to degradation by oxidizing agents such as
chlorine, hydrogen peroxide or ozone. These oxidants are used to sterilize the
membrane system. Periodic sterilization with high concentrations of chlorine is a
requirement in food applications; low levels of chlorine are added to the
feedwater of other reverse osmosis plants to prevent bacterial growth fouling the
membrane surface. Methods of reducing the exposure of the membrane to
chlorine have been developed, but these methods have reliability and cost
problems. A number of groups are trying to solve the membrane degradation
problem by modifying the chemistry of the polymer membrane. Progress has been
made over the past 10 years, but membrane chlorine sensitivity remains a largely
unsolved problem. The industry is also moving away from chlorine sterilization to
ozonation. This emphasizes the need for a membrane with broad spectrum
oxidation resistance rather than just chlorine resistance. If high-performance
oxidation-resistant membranes were available, they could displace cellulose acetate
membranes industry-wide, and a number of new applications for membranes would
open up.
Development of ultrathin-skinned, gas separation membranes was ranked

third in the priority research list; development of ultrathin, composite membranes
was ranked eighth. The selection of these two closely related topics in the top
IO priority research list reflects the major impact that the development of
generally applicable methods of making ultrathin membranes would have on the
gas-separation industry. Development of this technology would also be of value in
other membrane areas, particularly pervaporation.
The ultrathin-skinned, gas-separation membrane topic ranked number three

refers to asymmetric membranes composed of one material. This would include
membranes made by the phase inversion process, for example. The ultrathin
composite membrane topic ranked number eight refers to multilayer membranes, in
which a high-flux support is overcoated by an ultrathin permselective layer.
Membranes of both types, made from a variety of polymers, by a variety of
different techniques, are currently in commercial use. Asymmetric and composite
membranes can be produced with a skin or permselective layer thickness down to
about 0.5-I pm. Membranes with permselective layers with thickness in the range
of 0.1-0.5 pm are also made commercially, but the number of materials that can
be formed into membranes of this type is very limited. Finally, there are a few
claims in the literature of defect-free membranes being made in the range of 0.05
pm (500 A) or less. These claims must be treated with caution and it is certain
that no generally applicable technique exists for forming this type of membrane.
New polymer materials are now being developed that do not lend themselves
to fabrication into membranes by either the phase-inversion or the solution-
coating technique, especially when very thin, <500 A, permselectiva layers are
required. The development of membrane preparation methods, either for integral-
skinned, asymmetric membranes or for multilayer composite membranes, that could
be used to fabricate ultrathin membranes from any polymer material would
therefore have a major effect on the entire gas-separation industry.
The energy impact of improved gas-separation technology is likely to be

substantial. For example, if improved membranes for making oxygen-enriched air
were available. it has been estimated that up to 0.36 quads of energy per year
could be saved.’ Removal of acid gases from sour natural gas could result in an
energy savings of 0.01 quads per year in the processing of the gas a1one.s If the
process enables the processing of very sour natural gas reserves that could not be
exploited by other means, then the energy savings would be very large.
The development of facilitated-transport, oxygen-selective solid carrier
membranes was given a research ranking of four. Liquid, oxygen-selective
facilitated-transport membranes have been an area of research since the 1960s and
some high-performance membranes have been produced in the laboratory. For
example. liquid carrier-containing membranes have been reported with

oxygen/nitrogen selectivities of 20. compared to selectivities of 6 for the best
commercial polymeric membranes. ’ The permeability of the liquid membranes is
also very high. Unfortunately, these liquid membranes are too unstable to be
used in any commercial process. This instability problem has not been solved
despite 20 years of research.
Recently, workers in Japan and West Germany have developed facilitated

transport membranes using solid carriers.6*6 In these membranes the carrier is
either physically dispersed in a polymer matrix or covalently bonded to the
polymeric backbone of the matrix material. Contrary to accepted wisdom, these
membranes exhibit substantial facilitation of the permeating species. The long-
term stability of the membranes has not been demonstrated, nor have they been
formed into high-performance, ultrathin membranes, but the solid carrier approach
has merit. Although producing stable facilitated-transport membranes appears to
be a high-risk research topic, the reward if this membrane can be made is
correspondingly large. Stable membranes with an oxygen/nitrogen selectivity of
20, for example, would probably displace cryogenic processes as the production
method for oxygen and nitrogen. Since nitrogen and oxygen are the first and
third most important industrial chemicals in the United States, this would be a
breakthrough of tremendous significance. Even more importantly. with these
membranes it would become possible to produce oxygen-enriched air containing
40-80% oxygen at low cost. Availability of this oxygen-enriched air would
dramatically alter the economics of many combustion processes. Although topic
four is centered on the production of oxygen-selective carriers, it is likely the
same technology, if successfully developed, could be applied to other separations,
for example, the separation of acid gases from methane or alkane-alkene
separations.
The production of highly oxygen-selective polymers was given a research

priority ranking of five. The objective of this research topic is similar to topic
four above. The target is, however, a good deal more modest and the prospects
for success higher. The best commercially available oxygen-selective membranes
have an oxygen/nitrogen selectivity in the range 6-7 and permeabilities of 2-10
Barrer. Systems based on these membranes are competitive for the production of
95-98% nitrogen on a small scale, up to 20-50 tons/day. They are not competitive
for larger plants, where the economics of cryogenic separation are more favorable.
Even small incremental improvements in membrane performance could, however,
substantially increase the market share of membrane processes. If slight
improvements in membrane performance were achieved, such that an
oxygen/nitrogen selectivity of 7-10, with a permeability of 10 Barrer or more
were possible, commercial production of oxygen-enriched air by membrane systems
would become viable.
Reaching an oxygen/nitrogen selectivity target of 7-10 and a permeability

target of 10 Barrer or more appears to be within sight. A number of materials
with properties close to these values have already been reported. If they can be
fabricated into high-performance membranes and modules, they could have a
significant impact on the energy used in gas separation technology.
The principal problem in ultrafiltration technology is membrane fouling. For

this reason, the development of fouling-resistant ultrafiltration membranes was
given a research priority ranking of six. Fouling in ultrafiltration generally
occurs when materials dissolved or suspended in the feed solution are brought in
contact with, and precipitate on, the membrane surface. The precipitated
material forms a secondary barrier to flow through the membrane and drastically
lowers the flux through the membrane. The fouling layer becomes more dense
with time, rapidly at first and then more slowly. The flux through the
membrane declines correspondingly.
Fouling is usually controlled by rapid circulation of the feed solution across

the membrane surface. The turbulence this produces in the feed solution slows
the deposition of material on the membrane. Rapid feed circulation uses large
amounts of energy, however, so a balance is struck between energy consumption
and the amount of acceptable fouling. Fouling eventually reaches a point where
even rapid feed solution circulation no longer maintains the flux at an acceptable
level. The ultrafiltration system is then taken out of service and cleaned.
Cleaning, however, almost never restores the system to its original performance
and after some time, varying from 9 months to 5 years, the ultrafiltration modules
must be replaced.
The development of fouling-resistant ultrafiltration membranes would

decrease capital and operating costs and increase membrane lifetime. This is a
difficult problem and no one solution is likely to be generally applicable. The
basic mechanics of membrane fouling are undefined, so basic, as well as
engineering, research is required. Promising approaches under development
include modifying the membrane surface by making it more hydrophilic or adding
charged groups to the surface. Membrane pretreatment with additives that coat
the membrane to inhibit fouling is also used.
Solvent-resistant pervaporation modules, the seventh priority research topic,

was discussed in conjunction with solvent-resistant pervaporation membranes
(topic number one) and thin, composite gas separation membranes, the eighth
priority research topic, was discussed in conjunction with thin-skinned, gas-
separation membranes (topic number three).
Priority research topics nine and ten cover two research opportunities in the
microfiltration area, namely. development of low-cost microfiltration modules and
development of high-temperature, solvent-resistant membranes and modules
Development of low-cost modules was selected as a priority topic because a
number of extremely large potential applications exist for microfiltration if Costs
can be reduced. These applications include numerous possibilities in water-

pollution control applications. For microfiltration to move from its current role
as an effective but relatively expensive technology, microfiltration modules will be
need to be produced as a commodity with drastically lower costs. The authors
believe this goal is desirable and achievable.
The development of high-temperature and solvent-resistant membranes and

modules, the tenth priority research topic, would allow microfiltration to be used
in a number of applications where the limitations of current membrane modules
are a problem. These applications include filtration of hot wash-waters for
recycling. filtration of refinery oils, and removal of particulates from various hot
fluid streams. Ceramic membranes, which could be used in this type of
application, are just entering the market. These first generation ceramic
membranes are far too costly to be widely used, but as development efforts
progress, lower cost, more efficient ceramic filters may become available.
Ultrahigh .temperature performance polymers could also be used in this type of
application.
5.2 RESEARCH TOPICS BY TECHNOLOGY AREA
In the preceding section, the 38 priority research topics were addressed by

rank order. The same topics arranged by technology areas are listed in Tables 5-
3 to 5-9. These tables were produced after several revisions suggested at the
group meetings and by the external reviewers. Each table lists the top five to
seven priority research topics in its area. A brief description of the research
topic and the priority ranking is given, together with the prospect for realization
of each particular topic. Topics with relatively low prospects for commercial
success within IO-20 years were given a fair ranking in terms of prospects for
realization. Topics where the prospects were considered better, but where the
technology is still very undeveloped, or where major problems exist, were ranked
good. Topics with a relatively high probability for successful commercialization,
with minor problems, were ranked very good. Topics marked excellent were
considered very highly likely to succeed within the next ten years with adequate
research support.
Following each table is a summary of the relative merits and importance of

the various items. A detailed discussion of the individual topic areas is given in
the appropriate chapters in Volume 2.
5.2.1 Pervaporation
Table 5-3. Priority Research Topics in Pervaporation
Prospect for Rank out

Research Topic Realization Comments of 38
Membranes for Very Good If sufficiently selective membranes 1

organic-organic could be made, pervaporation could
separations replace distillation in many separations.
Better solvent- Excellent Current modules cannot be used with 7

resistant modules organic solvents and are also very
expensive
Better membranes for Very Good More solvent selective membranes are I5
the removal of organic required. especially for hydrophilic
solvents from water solvenu (phenols, acetic acid, methanol,
ethanol, etc.)
Dehydration membranes Good Would be of use in breaking many common I7

for acidic, basic, and aqueous-organic azeotropes.
concentrated aqueous
solvent streams
Plant designs and Good Pervaporation will probably be used in 33

studies hybrid systems for organic-organic
separations. System design studies are
needed to guide research.
Four of the five pervaporation research topics listed in Table 5-3 were ranked in
the top half of the priority research list. Two topics relating to the development
of pervaporation membranes and modules for the separation of organic mixtures
were ranked in the top ten. This high ranking reflects the potential
pervaporation has to replace or augment distillation in a number of significant
applications in the chemical processing industry. Distillation is an energy-
intensive operation that consumes 28% of the energy used in all U.S. chemical
plants and petroleum refineries. 7 The total annual distillation energy consumption
is approximately 2 quads, or 3% of the entire national energy usage.’ The top
IO distillation separations ( Crude oil; Intermediate hydrocarbon liquids; Light
hydrocarbons; Vacuum oil; Sour water; Ammonia/water; Styrene/Ethyl-benzene;
Ethylene glycol/water; Methanol/water; Oxygen/nitrogen) together consumed 1.0

quads of energy in 1981. Many of the major distillation separations consume more
than 2,000 Btu/lb of product.
Pervaporation systems are currently used for breaking the water-alcohol

axeotrope in the preparation of anhydrous alcohol. Process experience indicates
that the steam requirement of pervaporation is 20% of that for axeotropic
distillation for the preparation of 99.7% isopropanol from a feed stream containing
87% isopropanok6 Pervaporation does have other energy requirements, which
include electrical energy for vacuum pumps and chillers. However pervaporation
still offers a 60% energy savings over axeotropic distillation in the dehydration of
ethanoLg It is likely that similar savings could be achieved in other separations
where azeotropes are involved. Pervaporation could also be used to supplement
many existing distillation operations, for example by treating the top or bottom
fractions from the distillation column. A 10% reduction in energy consumption for
distillation would save 0.2 quads of energy per year.
The other two pervaporation topics ranking in the top half of the priority
list both relate to removal of solvents from aqueous streams. This type of stream
is very common and pervaporation systems could be widely used in solvent-
recovery and pollution-control situations. However, current membranes are best
suited to recovery of relatively hydrophobic solvents. Developments of membranes
able to treat more hydrophilic polar solvents and acidic or basic solvent streams
would allow the process to be much more widely used.
5.2.2 Gas Separation
Table 5-4. Priority Research Topics in Gas Separation

Development of Very Good Would allow broad usage of advanced 3

generally applicable materials - even better if done in
method for producing in hollow fibers
membranes with <SOOA
skins
Higher Os/Ns selectivity Very Good Experimental materials approach these 5

(~~97-10) and produc- intrinsic o and P numbers, but no ability
tivity polymer to spin form them in hollow fiber form has
(PmZ-3 Barrer for 0,) been reported. Most valuable as hollow
fibers.
Development of a Good Only small amounts of the valuable 8

generally applicable selective material are required.
method for forming
composite membranes
with <500A skins
Membrane material with Good Will become more important as the acid 12
high selectivity CO, and gas partial pressure in the feed from EOR
HsS separations from CH, projects increases.
(a>49 and H, (-20) at
high CO, and HsS
partial pressures
Reactive treatments Good Attractive if it is generally applicable. 24

for increasing the Both photochemical and fluorination
selectivity of a processes have been demonstrated on dense
preformed ultrathin films and on a relatively thick (Irm)
skin without excessive composite membrane, but not on thin
flux losses (<l.OOOA) membranes.
High oxygen selective Fair Carbon fiber, inorganic or facilitated 25

membrane (um 12- I5 for transport membranes may meet Q and
0,/N,) with good flux goals.
stability and an 0s flux
0.5- 1x10-’ ems (STP)/
cm*scmHg
Guidelines to stream- Good Much progress has been made. but steady, 21
line selection of long-term building of this capability
polymers for high provides a good basis for opening
efficiency potential new markets and preventing
separations displacement by foreign products.
Gas separation was considered to be a high priority area for membrane

research. Three of the seven gas separation topics in Table 5-7 were listed
among the top ten priority research topics. Gas separation research topics are
divided into two areas; the first dealing with methods of making better, high-
performance membranes, and the second dealing with development of membrane
materials with improved selectivity and permeability.
Topics covering methods of making high-performance gas separation

membranes were ranked 3, 8 and 24 in the priority research list. To fully exploit
the potential of gas separation materials now available, membranes that are
essentially defect-free and have permselective layers on the order of SOOA thick
or less must be mass produced. Techniques have been developed that come close
to this target with a few materials. However, generally applicable techniques are
not available. A number of approaches are being explored and the prospects of
success are good to very good.
The second major area of current gas separation research is the

development of better membrane materials. In the past, membranes were prepared
from polymers developed for other uses. The new generation of gas separation
membranes just now entering the market all use membranes made from polymers
specially designed and synthesized for their permeability properties. This area of
research will continue to grow. Particularly important target applications are the
separation of oxygen and nitrogen from air and the separation of acid gases, such
as carbon dioxide and hydrogen, from natural-gas and chemical-process industry
streams. Development of these new membrane materials has been aided by basic
ongoing research aimed at understanding the effects of polymer membrane
structure on permeability.
Estimates for the energy savings from oxygen-selective membranes vary

widely, depending on the oxygen enrichment possible. Low grade oxygen
enrichment (35%-50%) has been shown to be sufficient to improve the energy-
efficiency of combustion processes. However if high grade (~75%) oxygen-enriched
streams were available at low cost, then the process modifications and resultant
energy savings would occur throughout industry. Various estimates have placed
the energy savings from the production of oxygen-enriched air at between 0.06
and 0.36 quads per year.’
Upgrading of 200,000 SCFD of sour natural gas (17% HsS, 45% CO,) to
remove 3096 of the acid gas present using a membrane system will result in an
estimated savings of 0.01 quads per year. s The total energy savings will depend
on the economic feasibility of producing gas from sour gas wells and are
potentially huge.‘O
5.29 Facilitated Transport
Table 5-5. Priority Research Topics in Facilitated Transport

Research Topic RealiwCon Comments of 38
Oxygen-selective Fair Air separations of higher selectivity 4

solid facilitated are a target common 10 all types of
transport membranes membranes
Olefin-selective solid Fair Membrane life is the key question 23

facilitated wansport especially with sulfide contaminants.
membranes
Optimal design of Excellent AS membranes get better, module design 30

membrane contac1ors maximizing mass transfer per dollar
becomes key.
Membrane contactors Excellent Dramatic success for drugs can be 32

for copper and uranium repeated with metals.
Membrane contacfors Good Success in the field is uncertain. 35

for flue gas and
aeration
The five facilitated transport research topics are divided into two groups:
research on oxygen-selective solid facilitated transport membranes, which was
ranked very high, and ail the other topics, which were ranked relatively low.
Separation of oxygen and nitrogen from air continues to interest membrane
research groups around the world. Facilitated transport membranes have been
made in the laboratory with selectivities for oxygen from nitrogen of 20 or
more.’ If this selectivity could be achieved in a stable industrial membrane it
would be a major breakthrough with an enormous economic impact. In principal,

this membrane would allow oxygen-enriched air to be used in a large number of
combustion processes to produce the same amount of useful energy, but use
significantly less fuel. Having said this, the production of these membranes is
likely to prove extremely difficult, although recent work by the Japanese has been
encouraging.
The four other facilitated-transport membrane topics were ranked low

because generally the applications did not seem large, were too far in the future.
or did not appear to offer a major advantage over competing technologies.
5.2.4 Reverse Gamoais
Table 5-6. Priority Research Topics in Reverse Osmosis

Research Topic Realization C0mment.s of 38
Oxidation-resistant Good CommerciaI polyamide reverse osmosis 2

membrane membranes rapidly deteriorate in the
presence of oxidizing agents such as
chlorine, hydrogen peroxide, etc. This
deficiency has slowed the acceptance of
the process in some areas.
Improved pretreatment GOOd Improvement of classical pretreatment 16

methods that will enhance the reduction
of suspended solids in feed streams to
reverse osmosis systems is desired.
Bacterial attachment Excellent Bacterial fouling of membrane surfaces 18

to membrane surfaces reduces productivity. Affinity of micro-
organisms for different membranes is
markedly different. Elucidation of attach-
ment mechanism is required to select
optimal membrane material and surface
morphology.
Increased water flux Excellent Commercial thin-film composite 21

membranes operate at 30% of theoretical
efficiency because ol flow restrictions
within the membrane. Modest improve-
ment could reduce the energy consumption
of the reverse osmosis process
significantly.
Cleaning improvements Excellent Membrane cleaning is not always suc- 31

cessful; it remains a trial and error
operation.
Disinfectants Good Disinfectants that do not produce tri- 38

halomethanes are needed to control
membrane fouling by microorganisms.
Five of the six reverse osmosis priority research topics related to problems
associated with membrane-fouling and addressed various ways of tackling this
problem. For example. chlorination of reverse osmosis feed waters is now
required to prevent bacterial fouling of the membranes. However, chlorine
degrades interfacial composites, the best membranes currently available.

Development of an interfacial composite membrane resistant to not just chlorine,
but other oxidants, such as ozone or hydrogen peroxide, was ranked very high.
Improved methods of pretreating the feed or preventing bacterial attachment to
the membrane in the first place also ranked in the top half of the priority
research list. Finally, better membrane cleaning methods and a search for
alternatives to chlorine as a disinfectant were included on the list, although
ranked of lesser importance.
The focus on the operating problem of membrane fouling reflects the

importance of this problem to the reverse osmosis industry. It also reflects the
very high performance of current membranes. The best membranes available
have salt (NaCI) rejections of greater than 99.5% with corresponding water fluxes
of 0.5 ms/m2 day. The development of membranes with better salt rejections
and/or higher fluxes would enable reverse osmosis operations to operate at lower
pressures, but the impact on costs would not be dramatic. For this reason,
development of higher flux reverse osmosis membranes was included as a research
topic, but ranked in the lower half of the list.
52.5 Microfiltration
Table 5-7. Priority Research Topics in Microfiltration

Low-cost membrane Excellent Hugs potential applications will 9

modules require commodity pricing, far
from today’s reality.
High-temperature, Good Opportunity for cenmic or inorganic IO

solvent resistant membranes. Potoncial uses include
membranes and modules removal of particulates from coal and oil
liquids and replacement for bag houses
in flue gas treatment
Non-fouling, cleanable. Good Critical for abattoin. dairies, 22

long-life membranes breweries and wineries. Mut be
tolerant of the industry-approved
sanitizer.
Continuous integrity Good Applications where biological integrity 34

resting is required need evidence of continued
compliance. especially for remote and
automatic operation.
Cheap, fouling- Fair Current modules foul rapidly, especially 37

resistant module with solutions having high loadings of
designs particulates. Better module designs are
required.
Microfiltration is a well-developed membrane process. Commercially, it is

the largest and most deveioped of any studied. It has a high rate of investment
and a high level of success. The profitable products developed by this industry
concentrate on high value applications such as pharmaceuticals, foods, chemicals
for making semiconductor integrated circuits, etc. These applications are
exacting, demanding and do not require commodity pricing. There are important
applications at the mass usage end of the spectrum; perhaps even potable water
and sewage treatment. These applications require a different sort of thinking
about product design, manufacturing and pricing.
The other research topics on the microfiltration list were aimed at

developing specific membrane modules that could expand the applications of
microfiltration. Development of high-temperature and solvent-resistant
membranes was considered to be a high-priority topic because it could open up
significant markets for microfiltration in the petrochemical industry and in the
filtration of hot gas streams. Similarly, development of a method of continuously
monitoring the integrity of membranes would allow increased market penetration
of microfiltration into the cold sterilization of foods, beverages and
pharmaceutical products.
5.2.6 Ultrafiltration
Table 5-8. Priority Research Topics in Ultrafiltration

Fouling-resistant GOOd Fouling is ubiquitous in UF. Its 6

membranes elimination would boost total through-
put ~30% and reduce capital costs by 15%
on top of eliminating cleaning. Better
fractionation would also result, expanding
UF use significantly.
Lower-cost, longer- Excellent Lower cost modules with better fouling 13

life modules control are required.
Low-energy module Excellent Current module designs use large amounts I4

designs of energy in feed recirculation to con-
trol concentration polarization and
fouling. More efficient module designs
would use less energy.
Solvent-resistant Fzir Petroleum applications of ultrafiltration 20

membranes and modules could be large. Will require high temp-
perature. solvent resistant membranes and
modules. Ceramic membranes would fir here.
High-temperature, Good Current membranes cannot treat important 28

high-pH and oxidant- industrial streams because of temperature,
resistant membranes pH and oxidant sensitivity; another
potenrial application for ceramic membranes.
Of the deveioped membrane processes, ultrafiltration was ranked highest as

an area for increased research attention. This reflected the opportunities for
further growth of this technology if unsolved problems are addressed. The
biggest ultrafiltration research problem is membrane fouling; three of the five
ultrafiltration research topics, ranked 6, I3 and 14, addressed various aspects of
this problem. Fouling-resistant membranes is clearly a preferred research topic,
but improved modules which are lower in cost and inherently more fouling-
resistant, or modules which use less energy to control fouling, were other
approaches given high priority.
Finally, the development of membranes and modules able to treat solutions at

high temperatures, at high and low pHs, and containing solvents was considered to
be a significant opportunity for ultrafiltration research, but of less importance
than fouling-control research. Current membranes and modules are almost ail
polymer based and cannot be exposed to harsh environments. Ceramic
membranes are being developed that have promise and are finding niche
applications. If the cost and reliability of these modules could be improved, a
number of significant opportunities for large-scale use of ultrafiltration would
develop.
Both ultrafiltration and microfiltration could find new or broader applications

in the food industry with attendant energy savings. The food industry uses 1.5
quads of energy per year.” Areas where the use of membranes could result in
energy savings include:
l Concentration of corn steepwater and potato byproduct water
l Degumming, refining and bleaching of edible oils
l Clarification and concentration of beet sugar juice
l Bioprocessing of potato and dairy wastes
l Solvent recovery in edible oil processing
The potential energy savings in these areas are estimated at 0.13 quads
annually.”
5.2.7 Electrodialysis
Table 5-9. Priority Research Topics in Eiectrodialysis

Membranes with better Excellent Current ED systems are limited by II

temperature stability operating temperature. Temparature-
resistant modules would lower the
electrical resistance and reduce
energy use.
Spacer design for Good Concentration polarization remains a 19

better flow problem in electrodialysis. Better
distribution spacers would help.
Better bipolar Very Good Bipolar membranes could be a major 23

membranes growth area in electrodialysis if
better membritnes can be made.
Steam-sterilizable Very Good Electrodialysis is making inroads into 29

membranes the food and drug industry, bur steril-
ization remains a problem.
Fouling-resistant Very Good Fouling remains a problem in some 36

membranes electrodialysir applications.
Electrodialysis is an established membrane separation process which has

changed little in the last ten years. For this reason, the five priority research
topics in the eiectrodialysis area all addressed specific engineering problems. The
highest priority rankings in Table 5-9 are both aimed at improving the current
major application of electrodialysis. namely desalination of brackish waters.
Membranes with better temperature stability and spacers with improved flow
distributions would produce incremental improvements in brackish water
desalination systems. Almost a billion dollars worth of electrodialysis systems are
installed worldwide. Consequently, an incremental reduction in operating cost, of
as little as 10%. by retrofitting better membranes and spacers, would produce a
substantial savings.
The remaining three priority electrodialysis research topics were aimed at

making electrodialysis more useful for various niche applications. For example,
the application of electrodialysis to the food and pharmaceutical industries would
be helped by more fouling-resistant membranes and stream-sterilizable membranes.
Better bipolar membranes would be useful in the production of low grade acid and
alkali. All of these applications were ranked fairly low, principally because the
importance of the particular applications they addressed was not large.
5.3 COMPARISON OF DIFFERENT TECHNOLOGY AREAS
As was shown clearly by Table 5-1, the relative importances of the research
priorities in different technology areas were ranked very differently. For
example, the highest priority topic in electrodialysis, temperature-stable
membranes, ranked almost equal with the fourth highest priority topic in gas
separation, membranes for acid-gas separations. All but the highest priority item
in facilitated transport ranked about level with, or below, the lowest priority
items in ultrafiltration or gas separation.
Averaged rankings of the topics in each technology area are given in Table
5-10.
Table 5-10. Overall Ranks of the Seven Membrane Technology Areas
Membrane Average Research

Technology Area Topic Priority Ranking
-
Pervaporation 14.6
Gas separation 14.9
Ultrafiltration 16.2
Reverse Osmosis 21.0
Microfiltration 22.4
Electrodialysis 24.2
Facilitated transport 24.8
Clearly, research in the general areas of pervaporation and gas separation

was ranked substantially higher than the other technology areas. This high
ranking reflects the general feeling of the group that these two technologies
offer the best opportunities for research breakthroughs that would have a major
effect on energy consumption and costs in U.S. industry.
The three established membrane filtration processes, ultrafiltration, reverse

osmosis and microfiltration were grouped together in the center of the list
spanning the average ranking. As a group, the ultrafiltration-related topics were
ranked most important, followed by reverse osmosis, then microfiltration. All
three topics scored one entry in the top 10 rankings, and microfiltration scored
two. The priority topics in each area were remarkable similar. All of the areas
included priority research topics covering fouling-resistent membranes and
modules, membranes and modules that can withstand harsh environments, and
lower cost modules.
Module fouling is a continual problem in all membrane filtration processes,

and the high priority given by the author group to ways of reducing fouling
reflects the importance of the problem. Fouling-resistant membranes for
ultrafiltration ranked seventh out of 38, improved pretreatment to reduce fouling
and reduction of bacterial fouling, both for reverse osmosis, ranked sixteenth and
eighteenth, and nonfouling microfiltration membranes ranked in position twenty-
two. Methods of reducing the cost of modules and improving module design also
ranked high.
Electrodialysis and facilitated transport were both marked at the bottom end
of the research priority list about equal in level of importance. In the case of
electrodialysis, the authors generally felt that electrodialysis is a well-developed
process with a few established large applications. Electrodialysis does not appear
to be as widely applicable to problem separations as other membrane technologies,
such as reverse osmosis, ultrafiltration or microfiltration. For this reason it was
ranked low. The low rank of facilitated transport reflected general
disenchantment with the process. Liquid facilitated-transport membranes with
very high selectivities and fluxes have been available for more than 20 years, but
there are no commercial plants in operation. The problems of membrane and
carrier instability have just proven too intractable.
5.4 GENERAL CONCLUSIONS
One of the primary goals of the U.S. Department of Energy is to foster and
support the development of energy-efficient new technologies. The primary
objective for energy-efficient technology is a strategic one: to reduce U.S.
energy consumption, thereby reducing the oil trade deficit and the dependence on
foreign sources of oil. The energy costs of an industrial process directly affect
the cost of the goods produced. Therefore energy-efficient production
technology can result in higher productivity gains, an increase in the
international competitiveness of U.S. industry and a reduction of the current
trade deficit. Processes that use energy inefficiently are also significant sources
of environmental pollution. Environmental concerns have added impetus to the
search for energy-efficient, environmentally safe technologies. One such
technology is membrane separation, which offers significant reductions in energy
consumption in comparison with conventional separation techniques.
Membrane separation processes are widely used in many major industries.

Total sales of industrial membrane separation systems are more than $1 billion
annually.‘2 The United States is the dominant supplier of these systems.
United States dominance of the industry is being threatened, however, by
Japanese and, to a lesser extent, European companies.
The focus of this project was to report to the U.S. Department of Energy on
recommendations for priority research needs in membrane separation science and
technology. These specific aspects are discussed in the previous sections. Set
out here are some general conclusions relating to DOE’s support of membrane
research.
Conclusion I. DOE and other Federal soendina on membrane-related research is

small and fragmented: Current total Federal support for membrane-related
research is on the order %10-l I million/year. Of this total, approximately $4-5
million is provided by the National Science Foundation (NSF) to support basic
membrane research, mostly in the universities. A further 52-3 million is used by
the Environmental Protection Agency (EPA), the Department of Defense (DOD),
and the National Aeronautics and Space Administration (NASA) to support various
membrane activities that relate directly to their missions. The final $4 million is
used by the Department of Energy (DOE) to sponsor energy-related research
programs. Various offices within the DOE support programs in their own
particular area of interest. The Office of Industrial Programs funds research at
about the $1.5 million/year level; the Office of Basic Energy Research funds about
$1 million/year, and the Office of Fossil Energy about Sl-1.5 million/year.
In contrast, Federal research support was at a much higher level in the

1960s and 1970s. The lead agency was the Office of Saline Water (later the
Office of Water Research and Technology), which sponsored S20-40 million/year of
membrane-related research activities for many years. This high-risk investment
reaped handsome rewards, going far beyond the originally contemplated scope of
the program and impacting several different areas of membrane technology. which
are still being enjoyed by the U.S. economy.
Current U.S. Government membrane-related research programs, from all

agencies together, are approximately half of the corresponding Japanese and
European efforts. Other governments have attached greater importance to
furthering the advance of membrane science and technology. Without increased
commitment and support to membrane-related topics. the United States may begin
to lose markets in the existing membrane technologies, and may be a junior player
in world markets for the emerging membrane technologies.
Conclusion 2. Enaineerinn oroblems are holdins the U.S. membrane industry

m A noteworthy aspect of the research priority list was the heavy emphasis
on membrane technology and engineering, rather than membrane science.
Engineering- or technology-related problems ranked in positions 3, 5, 7. 8, 9 and
10 in the top 10 priority list. Other items that have an engineering component
include development of high-performance oxygen/nitrogen separation membranes
and modules, for which some suitable polymer materials are already known, but
where the technology to form them and use them is lacking. Even an item such
as the first-ranked priority topic, pervaporation membranes for organic/organic
separations, which at the moment requires basic membrane development and
testing studies, will not be able to be exploited industrially, with the attendant
major energy-savings benefits. unless the membrane development goes hand-in-

hand with the ability to form modules and design systems able to handle the
environment in which the pervaporation process is performed.
At present, a large portion of the total monies provided by Federal sources

is devoted directly to basic scientific research programs. As is right and proper,
essentially all of the S4 million support for research from the NSF is devoted to
fundamental membrane science. The projects funded by the DOD and NASA,
together amounting to no more than about Sl million annually, are a mix of
basic and engineering items, but highly specialized and out of the mainstream of
membrane development. EPA spends Sl.5 million/year, mostly on applications- and
engineering-oriented programs. The DOE’s $4 million annual expenditure on
membrane research is diverse. The Division of Chemical Sciences of the Office
of Energy Research, for example, typically funds fundamental programs, whereas
the other branches of DOE fund a spectrum of programs ranging from theoretical
or modeling studies to heavy engineering. In total, it appears that, of the SlO-11
million available annually to membrane topics, less than 54 million is probably
spent on engineering-related projects.
The emphasis of the expert group on technology and engineering issues reflects
the current developed status of the membrane industry. The state-of-the-art in
the emerging, as well as the established technologies, shows that engineering
issues are central to the ability to achieve practical, economically viable, energy-
efficient membrane systems.
Qat n V ion and nas seoaration:

If pervaporation could displace or supplement distillation in sectors of chemical
processing, the effects on energy consumption and competitiveness of U.S.
industry would be substantial. At present, the United States trails third in the
world in pervaporation research effort and capabilities. It is apparent that both
the Europeans and the Japanese have recognized the important potential of the
technology. In gas separation, where the United States is still first in the field,
ground may be lost as other countries step up their efforts. A focused effort in
gas separation technology is needed if the United States is to be a leader in the
new generation technology. The attendant benefits would be that membrane-based
gas separation will become competitive with conventional, energy-costly separation

technologies over a much broader spectrum.
Conclusion 4. Govern-t suooort is irclppLtzLat: Federal support remains

crucial to the membrane industry, both developing and developed. In the United
States, innovation typically comes from universities, small companies, or small
groups within companies. This has been especially true in the membrane
industry. The microfiltration industry, the area that currently commands more
than half of the total revenues generated by membrane sales, has been built up
by dedicated companies, a number of which, such as Gelman, Gore, Amicon and
Pall. started literally as one-man bands. The same is true in reverse osmosis,
where companies like Desalination Systems and Osmonics were built on the new
technology. In both of these industries, early U.S. Government support was a
key factor in future success. Membrane research is being conducted in a number
of large companies, but in general the research effort is fragmented, and a
sizeable portion of the R&D effort is coming from small innovators. It was felt
that, in the emerging technologies in particular, the leadership, focusing and
commitment roles played by Federal agencies in the past are still essential if
progress across a broad front is to be stimulated and maintained.
REFERENCES
1. Mix, T.W., Dweck, J.S., Weinberg, M., and Armstrong, R.C.. “Energy
Conservation in Distillation - Final Report”, DOE/X/40259 (1981).

3. Funk, E.. “Acid Gas Removal,” Proceedings of the 1988 Sixth Annual
Membrane Technology/Planning Conference Proceedings, Cambridge, MA,
November 1-3. 1988.
4. B.M. Johnson, R.W. Baker, S.L. Matson, K.L. Smith, I.C. Roman, M.E. Tuttle
and H.K. Lonsdale, “Liquid Membranes for the Production of Oxygen-Enriched
Air. II. Facilitated-Transport Membranes,” J. Memb. Sci. 3l_ 31-67 (1987).
5. Nishide, H., M. Ohyanagi, 0. Okada, and E. Tsuchida. “Dual Mode Transport

of Molecular Oxygen in a Membrane Containing a Cobalt Porphyrin Complex
as a Fixed Carrier,” MacromoleculeS. 20, 417-422, (1987).
6. Nishide, H.. and E. Tsuchida, “Facilitated Transport of Oxygen Through the

Membrane of Metalloporphyrin Polymers,” paper at &ond Annual Nation4
Meetinn of the North American Membrane So&y 9 Syracuse, N.Y., June,
1988.
7. Bravo, J.L., Fair, J.R. J.L. Humphrey, CL. Martin, A.F. Seibert and S. Joshi,
“Assessment of Potential Energy Savings in Fluid Separation Technologies:
Technology Review and Recommended Research Areas,” Department of Energy
Report DOE/LD/12473--1 (1984).
8. Asada, T., “Dehydration of Organic Solvents. Some actual results of

pervaporation plants in Japan,” Proceedings of Third International Conference
on Pervaporation Processes in the Chemical Industry, Nancy, France,
September 19-22. 1988.
9. Cogat, P.O., “Dehydration of Ethanol: Pervaporation compared with

azeotropic distillation,” Proceedings of Third International Conference on
Pervaporation Processes in the Chemical Industry, Nancy, France, September
19-22, 1988.
10. Henis, J., personal communication - review comments, 1990.
11. Mohr, C.M.. Engelgau, D.E., Leeper, S.A., and Charboneau, B.L., Membrane
Aoohcations and Research in Food Processing, Noyes Data Corp., Park Ridge,
NJ 1989.
12. A.M. Crull, “The Evolving Membrane Industry Picture,” in The 1998 Sixth
Ann ua IMmre b ane T ec hnlo onv/Plannina Conference Proceed in as , Business
Communications Company, Inc., Cambridge, MA (1988).
Appendix A. Peer Reviewers’ Comments
A draft final version of this report was sent to ten outside reviewers. The
reviewers were chosen for their experience and background in membrane science
and technology and their knowledge of the membrane industry. The following
people served as peer reviewers of this report
Dr. J. L. Anderson (Carnegie Mellon University)

Dr. J. Henis (Monsanto)
Dr. J. L. Humphrey (J. L. Humphrey and Associates)
Dr. S.-T. Hwang (University of Cincinnati)
Dr. N.N. Li (Allied Signal)
Dr. S. L. Matson (Sepracor, Inc.)
Dr. R. D. Noble (University of Colorado)
Dr. M. C. Porter (M. C. Porter and Associates)
Dr. D. L. Roberts (SRI International)
Dr. S. A. Stern (Syracuse University)
As far as possible, the reviewers’ comments, particularly those dealing with

specific changes or corrections, were incorporated directly into the report.
Excerpts from the reviews, covering general comments, policy recommendations
and dissenting views are presented in this section along with the authors’
rebuttals.
A.1 GENERAL COMMENTS
Three features of the report drew comments from many reviewers. The first
concerns the balance of the report between emphasis on basic science and
emphasis on engineering issues. The second concerns the importance of
integrating membrane technology into hybrid treatment systems. The last
concerns the merits or demerits of the ranking scheme that was adopted by the
group.
A. I I The report is biased toward engineering, or toward basic science.

Dr. Alex Stern commented that “the list of research priorities is too much
skewed toward practical applications”. Dr. Stern expressed concern at the
“decline in long-range fundamental research in this country”. His opinion was
that “applied research and development can solve many operational problems and
86
Appendix A: Peer Reviewers’ Comments 87
improve the efficiency of existing membrane separation processes. However, only

fundamental research can generate the new concepts which will produce the
membrane processes of the future.”
Dr. John Anderson pointed out that “the major emphasis of this report is on
the research needs for membrane engineering and technology..... The panel were
composed primarily of industrial researchers with a few academic persons
scattered throughout. The science of membranes (how they work, structure
versus function) was given low priority for this study”.
Dr. Steve Matson also observed that “high priority is given in the study to
engineering and product oriented research”.
Dr. Jav Henix expressed a completely opposite view. Dr. Henis said that
there was too much emphasis on basic research issues, and stressed that the
research topics need a greater engineering emphasis. He believed that most of
the top priority items have not adequately addressed engineering issues, and that,
if engineering input had been included in the analysis, the priorities might have
been different.
A.1.2 The importance of integrating membrane technology into total treatment

systems.
Dr. Steve Matson said that “it is very difficult to dispute the essential
conclusion of the study that pervaporation and membrane gas separations are two
areas in which increased federal funding would likely have great and relatively
near-term impact on energy consumption in the chemical process industry. This
reviewer might have put hybrid membrane processes (not-just pervaporation-
based) a bit higher on the priority list, for example, and he might have lobbied
for more consideration of important problems in biotechnology that are
addressable with membranes and which have important energy and environmental
implications”.
Dr. John Anderson stated that the “concept of systems design with membrane
technology integrated into the design is ignored. No persons active in design
research were on any of the panels. This omission significantly weakens the
statements made on behalf of the potential of membranes, for the real potential
of membrane separations will only be achieved when they are formally integrated
into process design methodology”.
REBU-ITAL: The expert group acknowledges that hybrid designs are very
important. The advantages of combining distillation and membrane separation,
for example, are discussed in Volume II, Chapter 2, Pervaporation. However,
hybrid systems are only useful where the membrane process will complement an
existing separation operation to provide technical or economic advantage. Such
opportunities clearly exist for the emerging technologies of pervaporation and
membrane gas separation, particularly in the process industries. The mature
membrane technologies, however, tend to be stand-alone, for example desalination
by reverse osmosis, and many microfiltration and ultrafiltration applications, or
their potential for inclusion in an integrated separation process has already been
recognized and is not likely to be substantially changed by improvements in the
membrane process.
A.1.3 The ranking scheme.

Dr. John Anderson was bothered by the rankings. Besides some possible
vested interest by panel members”, he believed that the rankings are ‘too loosely
assigned and might lead to biased funding in one area at the expense of another
equally important area. I strongly recommend that the top 10 or I5 areas be
listed without a priority ranking” but rather “be viewed as a collection of equally
important individual topics”.
Dr. Richard Noble accepted the ranking scheme, but would have preferred
that the ranked items be grouped together by according to theme. His point was
that “there are common themes or research needs that *permeate” this field.
Advances in a particular theme in one membrane area can have a synergistic
effect in other areas.” Dr. Noble advocated DOE support of the following
general themes: Membranes with Improved Resistance, Membrane Fouling. Thinner
Membranes, Membrane Materials and Use of Reaction Chemistry. He deprecated
support of themes relating to Membrane Treatment. Modules, and Standards.
Criteria and Testing.
Qg, Steve Mahpn preferred to rank the 38 items only in terms of high,
medium or low priority. His high-priority items all fell within the top ten
rankings, and his low-priority items all fell below ranking 17.
&g, Sun-Tak Hww Mark Port= also provided their own rankings, both
of which were in very good agreement with the consensus of the expert group.
Dr. Hwang ranked most ultrafiltration topics a little higher than the report
rankings; Dr. Porter ranked gas separation topics generally higher, and reverse
osmosis and microfiltration topics generally lower than the report rankings.
REBUTTAL: The goat of the study, and, therefore, the objective of the group,
was to prepare a prioritized list of research needs. All of the 38 topics
considered were significant enough to enter the analysis. The rankings were
prepared by secret vote of the group of authors, whose personal biases, if any,
were mitigated by the rest of the group. While one may disagree with the
concept of ranks. examination of the scores in Table 5.1 shows that there is a
clear consensus on certain definite levels of priority that should be assigned.
A.1.4 Comparison with Japan

Two reviewers. Dr. Jav Henis and Dr. Richard Noble, drew comparisons
between membrane technology in the United States and Japan. Dr. Noble urged
that “Government funding of membrane-related research is important and
essential’. His view was that “DOE should facilitate partnerships and/or
collaborative efforts between universities and industrial companies to make
fundamental advances and rapidly transfer the knowledge to the private sector so
it can be implemented and commercialized. This is the approach being taken in
Japan and Europe and uses the talents and resources of everyone who can aid in
advancing the knowledge base and implementing the knowledge”.
Dr. Henis was concerned that the Japanese have been producing better
products with our basic science. He felt that what the United States needs is a
strongly practical approach. He stressed that good science should not be
restricted to fundamental issues, but should also include engineering and
applications considerations.
A.2 SPECIFIC COMMENTS ON APPLICATIONS
A.2.1 Pervaporation
Pervaporation research ranked number one in the priority chart and not
surprisingly, therefore, attracted comments from all the reviewers.
Dr. Jimmv Humohree called for more emphasis on hybrid applications. He

said that high purity distillation requires high reflux ratios, which in turn
increase the steam requirements. He pointed out that, for instance,
pervaporation could be used in a hybrid arrangement to treat the overhead
product from a distillation column to produce a high purity stream.
Dr. John Anderm said that the case for pervaporation is overstated, or at
least not supported. His opinion was that recent advances in multicomponent
distillation with respect to energy conservation and azeotrope breaking will
reduce the impact of pervaporation. He believed that pervaporation will not
replace distillation over the next 50 years, although it may prove valuable in
supplementing distillation in the separation of organic liquids.
Dr. Richard Noble expressed the view that the development of solvent-
resistant modules is not worthy of DOE support and is best left to funding by
venture capital.
REBU-ITAL: If pervaporation is to be used either as an alternative to

distillation or to complement distillation, then both membranes and modules that
can handle the environment in which organic/organic separations take place will
be required. For DOE to support membrane development but not module
development is inconsistent, and creates a risk of the membrane technology being
either wasted or taken up and developed outside the United States. The effort
supported by the Office of Saline Water to develop reverse osmosis technology
embraced both membranes and modules, and proved very successful.
Dr. Jav HeniS. like Dr. Humphrey, took the view that current distillation
technology, with best available energy recovery systems, should be considered in
evaluating the relative merits of pervaporation. He felt that new pervaporation
units would be used in the basic chemical industries, which are presently in
decline in the United States and are increasingly located off-shore, so that the
domestic energy savings resulting from pervaporation will not be large.
REBUTTAL: The report recognizes that hybrid systems may be where the real
potential for certain pervaporation applications lies. For strategic and practical
reasons, the United States will always have a large petrochemical industry, and
this is an industry segment where pervaporation will both find applications and
result in energy savings. Besides the basic chemical industries, pervaporation
could be used in the chemical process industries, food processing, wastewater
treatment and many other specific applications.
A.2.2 Gas Separation

Several reviewers made specific comments expressing their own ideas as to
the most significant areas on which to focus. Dr. Richard Noble thought that
the breakthrough will be in new materials, such as inorganic membranes, zeolites
and molecular sieve membranes. He felt that most of the limitations of present
gas separation technology arise from the polymeric membrane materials.
Dr. Alex Stern stressed the importance of fundamental research into

molecular dynamics, which would lead to the ability to predict diffusion
coefficients from basic physico-chemical properties, and the design and synthesis
of new materials created exclusively for their permeation properties.
Dr. Jav HeniS believed that the development of a membrane to remove

carbon dioxide and hydrogen sulfide from low-grade natural gas, ranked 12 in the
priority list, has been rated too low. and urged that such a membrane could have
a measurable, instantaneous impact on U.S. energy reserves. Dr. Henis also
questioned the importance of the development of ultrathin-skinned membranes.
His view was that the problem of membrane productivity could be addressed by
other means, such as increasing the free volume of the polymer.
A.2.3 Facilitated Transport

Most reviewers concurred with expert group opinion that the general
prospects for facilitated transport are not bright. However, Dr. John Anderson
believed “a major breakthrough is possible with facilitated transport; however, new

research concepts are needed here. Thus, I would argue that with respect to
this topic, membrane science should be supported by DOE, and the science should
be truly gg& (i.e., not just another species of mobile carrier in a liquid film)“.
Dr. Richard Noble thought that there will be niche applications for facilitated
transport in 10 years time, and he would have liked to see oxygen/nitrogen
selective facilitated transport membranes included in the discussion of gas
separation membranes.
Dr. Jav Henis said that facilitated trwport deserves a very low or zero
priority, because the combination of requirements is impossible for a real system.
He pointed out that solid carriers are active species, not unlike catalyst
molecules, and are subject to the same poisoning processes, and that liquid
membranes require an infinite partition coefficient for the carrier between the
membrane and the process streams to prevent the carrier from being leached out.
A.2.4 Reverse Osmosis

pL Jav Henis wanted clarification that oxidation-resistant membranes, ranked
2 in the priority list, should cover membranes that will resist oxidants other
than chlorine. He stated that the industry trend is toward ozonation, and that
membrane research should, therefore, be directed at membranes that could
withstand various oxidants.
Dr. Noble felt that most of the research needs identified for reverse osmosis
were more appropriately within the province of the Department of the Interior,
and should not be funded by DOE.
A.2.5 Ultrafiltration
Dr. John Anderson. commented that “work on fouling-resistant membranes is
certainly needed, but the scope of this research should include development of
easily cleanable and restorable ultrafiltration membranes. These might not be
polymer-based.”
Dr. Richard Noble thought that more research is needed on ceramic and
inorganic membranes. “They can be cleaned, sterilized, and put in hostile
environments much more easily than polymer films. They are a high-cost item
now but research will inevitably lead to lower costs and materials suited to
various applications.”
Dr. Alex Stern believed more fundamental research should be supported, such
as using Monte Carlo techniques to calculate particle trajectories and predict gel
layer buildup and fouling rate. He felt that these basic insights can contribute to
more efficient membrane and module design and low-energy operation.
A.2.6 Microfiltration
Dr. Richard Noble was of the opinion that low-cost module development is
best left to market forces and should not be supported by the DOE.
A.2.7 Eiectrodialysis
Dr. Richard Noble stressed that bipolar membranes and better module design
are important.
A.2.8 Miscellaneous Comments

Dr. John Anderson and Dr. Steve Matson were both concerned about the
scope of the study. Dr. Anderson said “The entire area of biochemical/biomedical
membrane separations is omitted. This promises to be a big dollar item, and
energy will certainly play some role here on products of modest volume. In my
mind, it is not inconsistent for DOE to consider supporting research on large-
scale bioseparations by membrane methods.” Dr. Matson expressed himself
“somewhat distressed by the scope of the present study: i.e., by what is not
covered by the study as opposed to what is. Its limitation to relatively well-
developed membrane technologies and industries is a very significant one,
especially in the context of a “research needs” assessment. While the study sets
out to consider four “fully-developed” membrane processes and two “developing”
processes, it examined only one “to-be-developed” technology -- namely faciiitated
transport -- and that a technology which is over 20 years old. Thus, the study
deals primarily with an assessment of the state of the art and with what can
reasonably be expected to advance it.” Dr. Matson suggested a follow-on study
focused on “embryonic or emerging membrane technologies (e.g., the use of
sorbent membranes in high-flux adsorption processes, the use of catalytically
active membranes in reaction processes, the exploitation of attributes of

membranes other than the permselectivity, and the like.’ Dr Richard Noble also
would have liked to see catalytic membrane reactors included in the study. and
would have liked to see more discussion of the use of facilitated transport
membranes in sensors.
An additional study was also an idea broached by Dr. Norman Li. who felt
that “the discussions of the effect on environmental quality were diffused and not
very clear. Since this is an important issue, perhaps a separate volume to
discuss air and water purification via various types of membranes would be a
more focused and useful approach.”
Both Pr. Norman G and Qr. Jimmv Hua asked for a detailed breakdown
of NSFs programs in membrane research.
Volume II
Introduction to Volume II
Industrial separation processes consume a significant portion of the energy

used in the United States. A 1986 survey by the Office of Industrial Programs
estimated that about 4.2 quads of energy are expended annually on distillation,
drying and evaporation operations.' This survey also concluded that over 0.8
quads of energy could be saved in the chemical, petroleum and food industries
alone if these industries adopted membrane separation systems more widely.
Membrane separation systems offer significant advantages over existing

separation processes. In addition to consuming less energy than conventional
processes, membrane systems are compact and modular, enabling easy retrofit to
existing industrial processes. The present study was commissioned by the
Department of Energy, Office of Program Analysis, to identify and prioritize
membrane research needs in light of DOE'S mission.
This report was prepared by a group of six experts representing various

fields of membrane technology. The group consisted of Dr. Richard W. Baker
(Membrane Technology & Research, Inc.). Dr. Edward Cussler (University of
Minnesota), Dr. William Eykamp (University of California at Berkeley), Dr.
William J. Koros (University of Texas at Austin), Mr. Robert L. Riley
(Separation Systems Technology, Inc.) and Dr. Heiner Strathmann (Fraunhofer
Institute, West Germany). Dr. Eykamp served as Principal Investigator for the
program. Dr. Amulya Athayde (MTR) served as program coordinator for the
project. Dr. Athayde and Ms. Janet Farrant (MTR) edited the report.
Seven major membrane processes were covered: four developed processes,

microfiltration (MF), ultrafiltration (UF), reverse osmosis (RO), and
electrodialysis (ED); two developing processes, gas separation with polymer
membranes and pervaporation; and one yet-to-be-developed process, facilitated
transport.
96
Introduction t o Volume I I 97
The first three processes, microfiltration, ultrafiltration, and reverse osmosis,

are related filtration techniques, in which a solution containing dissolved or
suspended solutes is forced through a membrane filter. The solvent passes
through the membrane; the solutes are retained. These processes differ
principally in the size of the particles separated by the membrane.
Microfiltration is considered to refer to membranes that have pore diameten

from 0.1 pm (I.OO0 A) to IO pm. Microfiltration membranes are used to filter
suspended particulates, bacteria or large colloidr from solutions. Ultrafiltration
refers to membranes having pore diameters in the range 20-1.000 A.
Ultrafiltration membranes can be used to filter dissolved macromolecules, such as
proteins, from solution. In the case of reverse osmosis, the membrane pores are
so small. in the range of 5-20 A in diameter, that they are within the range of
the thermal motion of the polymer chains. Reverse osmosis membranes are used
to separate dissolved microsolutes. such as salt, from water. The principal
application of reverse osmosis is the production of drinking water from brackish
groundwater, or the sea.
The fourth fully developed membrane process is electrodialysis, in which

charged membranes are used to separate ions from aqueous solutions under the
driving force of an electrical. potential difference. The process utilizes an
electrodialysis stack. built on the filter-press principle. and containing several
hundred individual cells formed by a pair of anion and cation exchange
membranes. The principal application of electrodialysis is the desalting of
brackish groundwater.
Gas separation with membranes is the more mature of the two developing
technologies. In gas separation. a mixed gas feed at an elevated pressure is
passed across the surface of a membrane that is selectively permeable to one
component of the feed. The membrane separation process produces a permeate
enriched in the more permeable species and a residue enriched in the less
permeable species. Current applications are the separation of hydrogen from
argon, nitrogen and methane in ammonia plants, the production of nitrogen from
sir and the separation of carbon dioxide from methane in natural gas operations.
Pervaporation is a relatively new process that has elements in common with

reverse osmosis and gas separation. In pervaporation, a liquid mixture is placed
in contact with one side of a membrane and the permeate is removed as a vapor
from the other. The mass flux is brought about by maintaining the vapor
pressure on the permeate side of the membrane lower than the potential pressure
of the feed liquid. Currently, the only industrial application of pervaporation is
the dehydration of organic solvents. However, pervaporation processes are being
developed for the removal of dissolved organics from water and the separation of
organic solvent mixtures.
The final membrane process studied in the report is facilitated transport.

This process falls under the heading of "to be developed" technology. Facilitated
transport usually employs liquid membranes containing a complexing or carrier
agent. The carrier agerrt reacts with one permeating component on the feed side
of the membrane and then diffuses across the membrane to release the permeant
on the product side of the membrane. The carrier agent is then reformed and
diffuses back to the feed side of the membrane. The carrier agent thus acts as a
shuttle to selectively transport one component from the feed to the product side
of the membrane.
Each of the authors was assigned primary responsibility for a topic area as
shown below.
m Author
Pervaporation Richard Baker
Gas Separation William Koros
Facilitated and Coupled Transport Edward Cussler
Reverse Osmosis Robert Riley
Microfiltration and Ultrafiltration William Eykamp
Electrodialysis Heiner Strathmann
The role of the authors was to assess the current state of membranes in
their particular section, identify present and future applications where membrane
separations could result in significant energy savings and suggest research
directions and specific research needs required to achieve these energy savings
Introduction to Volume I1 99
within a 5-20 year time frame. The authors as a group also performed the
prioritization of the overall research needs.
Volume I of this report is devoted to a description of the methodology used

in preparing the report, an introductory outline of membrane processes, a
breakdown of current Federal support of membrane-related research in DOE and
other Government agencies, and an analysis of the findings and recommendations
of the group.
Volume 11 supports Volume I by providing detailed source information in

light of which the findings of Volume I can be interpreted. Volume I1 contains
eight chapters. The first discusses membrane and module preparation techniques.
The same general methods can be used to prepare membranes for use in diverse
separation applications, so it was decided to make a separate chapter of this
topic, rather than repeating essentially the same information in several individual
application chapters. Chapters Two through Eight take the areas of membrane
technology one by one, and provide an overview of the process fundamentals, a
discussion of the present technology and applications, and the limitations and
opportunities for the future. For each area, a list of high-priority research items
was identified and the relative importance of the items, in terms of impact on the
industry, energy savings and likelihood for realization, was prepared. The top
five or seven items in these individual lists were used by the group of authors to
prepare the master table of 38 priority research needs that forms the basis of
much of the discussion in Volume I.
REFERENCES

Technology, 1987. DOE Publication number DOE/NBM-8002773.
1. Membrane and Module Preparation
by
R.W. Baker, Membrane Technology and Research, Inc.

Menlo Park, CA
The present surge of interest in the use of membranes to separate gases and
liquids was prompted by two developments: first, the ability to produce high
flux, essentially defect-free membranes on a large scale; second, the ability to
form these membranes into compact, high-surface-area, economical membrane
modules. These major breakthroughs took place in the 1960s and early 1970%
principally because of a large infusion of US. Government support into reverse
osmosis research via the Office of Saline Water. Adaption of reverse osmosis
membrane and module technology to other membrane processes took place in the
1970s and 1980s.
In this section, we will briefly review the principal methods of

manufacturing membranes and membrane modules. Membranes applicable to
diverse separation problems are made by the same general techniques. In fact,
selecting a self-consistent membrane organization method based on either
-
application, structure or preparation method, is impossible. Figure I 1 illustrates
the problem.' We see that gas separation membranes can be symmetrical,
asymmetric or even liquid. Porous membranes can be used for ultrafiltration,
microfiltration or even molecular sieving. Hollow-fiber modules can be used for
gas separation, pervaporation, reverse osmosis, dialysis and so on. For the
purposes of this section. we have organized the discussion by membrane structure:
symmetrical membranes, asymmetric membranes, ceramic and metal membranes, and
liquid membranes.
Symmetrical membranes are uniformly isotropic throughout. The membranes

can be porous or dense, but the permeability of the membrane material does not
change from point to point within the membrane. Asymmetric membranes, on the
other hand, typically have a relatively dense, thin surface layer supported on an
open, often microporous substrate. The surface layer generally performs the
separation and is the principal barrier to flow through the membrane. The open
support layer provides mechanical strength.
Ceramic and metal membranes can be both symmetrical or asymmetric.

However, these membranes are grouped separately from polymeric membranes
because the preparation methods are so different.
Liquid membranes are the final membrane category. The permselective

barrier in these membranes is a liquid phase, often containing dissolved carriers
which selectively react with specific permeants to enhance their transport rates
through the membranes. Because the membrane barrier material is a liquid,
unique preparation methods are used. In their simplest form, liquid membranes
consist of an immobilized phase held in the pores of a microporous support
membrane; in another form the liquid is contained as the skin of a bubble in a
liquid emulsion system.
100
Expanded films
Template leachlng Mlcroflltratlon
- Phase Inversion Pore membrane Ultraflltratlon
Nucleatlon track Dlalysls
Compressed powders
Ion-relectlve Electrodlalyslr
membrane
The membrane organization scheme described above works fairly well.

However, a major preparation technique, solution precipitation, also known as
phase inversion, is used to make both symmetrical and asymmetric membranes. To
avoid needless repetition, the technique has been described under asymmetric
membranes only. The major membrane structures are shown in Figure 1-2.
I . 1 SYMMETRICAL MEMBRANES
1.1. I Dense Symmetrical Membranes
Dense symmetrical membranes are widely used in R & D and other laboratory
work to characterize membrane properties. These membranes are rarely used
commercially, however, because the transmembrane flux is too low for practical
separation processes. The membranes are prepared by a solution casting or a
thermal melt-pressing process.
I . I . I . 1 Solution casting
Solution casting uses a casting knife or drawdown bar to draw an even film
of an appropriate polymer solution across a casting plate. The casting knife
consists of a steel blade, resting on two runners, arranged to form a precise gap
between the blade and the plate on which the film is cast. A typical hand-
casting knife is shown in Figure 1-3. After the casting has been drawn, i t is left
to stand, and the solvent evaporates to leave a thin, uniform polymer film.
A good polymer casting solution is sufficiently viscous to prevent the

solution from running over the casting plate. Typical casting solution
concentrations are in the range 15-20 wt% polymer.
Solvents having high boiling points are inappropriate for solution casting,
because their low volatility demands long evaporation times. During an extended
evaporation period, the cast film can absorb sufficient atmospheric water to
precipitate the polymer, producing a mottled, hazy surface. If limited polymer
solubility dictates the use of a high boiling solvent, a spin casting technique
proposed by Kaelble can be employed.2 The apparatus used in this technique is
shown in Figure 1-4. It consists of a hollow aluminum cylinder with a lip, fitted
with a removable plastic liner. The cylinder is spun and the casting solution is
poured into it. Centrifugal force pushes the solution to the cylinder walls. AS
the solvent evaporates, it leaves a dense film at the surface of the casting
solution; this more concentrated polymer material is then forced back into the
solution by the high centrifugal force, bringing fresh unevaporated solvent to the
surface. This action results in rapid solvent evaporation, more than ten times
faster than normal plate-cast films, and allows solution-cast films to be prepared
from relatively nonvolatile solvents.
I . I . 1.2 Melt pressing
Many polymers d o not dissolve in suitable casting solvents. Such polymers,

including polyethylene, polypropylene, and nylons can be made into membranes by
melt pressing.
Membrane and Module Preparation 103
Nucleation Track Membrane Microporous Phase-Inversion

Membrane
Asvmmetrical Membranes
Loeb-Sourirajan Phase-Inversion Silicone-Rubber Coated

Membrane Composite Membrane
Electrolytically-Deposited Multilayer Ceramic Membrane

Membrane
Figure 1-2. Types of membrane structures.

L-_jd,
Figure 1-3.
=r -. --ll
A typical hand-casting knife. (Courtesy of Paul N. Gardner

Company, Inc. Pompano Beach, FL.)
Casting
cylinder
Ld\ Rotating
spindle
Figure 1-4. Schematic of a spin-casting mold. (From Kaelble.)

Melt pressing, or melt forming, involves sandwiching the polymer at high

pressure between two heated plates. A pressure of 2.000-5,000 psi is applied for
0.5 to 5 minutes, while holding the plates just above the melting point of the
polymer. The optimum pressing temperature is the lowest temperature that yields
a membrane of completely fused polymer of the desired thickness. If the
polymer contains air or is in pellet form, massing the material on a rubber mill
before pressing improves results.
To prevent the membrane from sticking to the press plates, the polymer is
usually placed between two sheets of Teflon-coated foil or non-waterproof
cellophane. Cellophane does not stick to the plates and, if dampened with water
after molding, it strips easily from the membranes. To make films thicker than
100 pm, metal shims or forms of an appropriate thickness are placed between
the plates. The best technique for making thinner membranes is free-form
pressing. A typical laboratory process used to make melt-formed films is shown
in Figure 1-5.
1 . l . 2 Microporous Symmetrical Membranes
A number of proprietary processes have been developed to produce

symmetrical microporous membranes. The more important are outlined below.
1.1.2.1 Irradiation
Nucleation track membranes were first developed by the Nuclepore

Corporation.3** The two-step process is illustrated in Figure 1-6. A polymer film
is first irradiated with charged particles from a nuclear reactor. As the particles
pass through the film, they break polymer chains in the film and leave behind
sensitized tracks. The film is then passed through an etch solution bath. This
bath preferentially etches the polymer along the sensitized nucleation tracks,
thereby forming pores. The length of time the film is exposed to radiation in the
reactor determines the number of pores in the film; the etch time determines the
pore diameter. This process was initially used on mica wafers, but is now used to
prepare polyester or polycarbonate membranes. A scanning electron micrograph of
a nucleation track membrane is shown in Figure 1-7.
I . 1.2.2 Stretching
Microporous membranes can also be made from crystalline polymer films by

orienting and stretching the film. This is a multistep process. In the first step,
a highly oriented crystalline film is produced by extruding the polymer at close to
its melting point coupled with a very rapid drawdown.SoB
After cooling, the film is stretched a second time, up to 300%. in the

machine direction of the film. This second elongation deforms the crystalline
structure of the film and produces slit-like voids 200 to 2,500 A wide. A
scanning electron micrograph of such a film is shown in Figure 1-8. The
membranes made by Celanese Separation Products, sold under the trade name
Celgard@. and the membranes made by W. L. Gore, sold under the trade name
Gore-Tex", are made by this type of p r o c e ~ s . ~
Figure 1-5. A typical laboratory press used to form melt-pressed

membranes. (Courtesy of Fred S. Carver, Inc., Menomonee
Falls, WI.)
I I I
I I
\ \ 1 1
\ \
\\ '%d@/
.';
'----e@' \
.
Figure 1-7. Scanning electron micrograph of a typical nucleation track

mem brane.
H
3 2 un;
Figure 1-8. Scanning electron micrograph of a typical expanded

polypropylene film membrane, in this case Celgard. (Reprinted
with permission from Cierenbaum et al., f n d . Enn. Chem. Proc,
Res. Develoo. 13, 2. Copyright 1974 Amercian Chemical
Society.)
1.1.2.3 Template leaching
Template leaching is not widely used, but offers an alternative

manufacturing technique for insoluble polymers. A homogeneous film is prepared
from a mixture of the membrane matrix material and a leachable component.
After the film has been formed, the leachable component is removed with a
suitable solvent and a microporous membrane is f ~ r r n e d ? * ~ The
~ ' ~ same general
method is used to prepare microporous glass."
1.2 ASYMMETRIC MEMBRANES
In industrial applications, symmetrical membranes have been almost

completely displaced by asymmetric membranes, which have much higher fluxes.
Asymmetric membranes have a thin, finely microporous or dense permselective
layer supported on a more open porous substrate. Hindsight makes it clear that
many of the membranes produced in the 1930s and 1940s were of the asymmetric
type, although this was not realized at the time. The asymmetric structure was
not recognized until Loeb and Sourirajan prepared the first high-flux,
asymmetric, reverse osmosis membranes by what is now known as the Loeb-
Sourirajan technique.I2 Loeb and Sourirajan's work was a critical breakthrough
in membrane technology. The reverse osmosis membranes they produced were an
order of magnitude more permeable than any symmetrical membrane produced up
to that time. More importantly, the demonstration of the benefits of the
asymmetric structure paved the way for developers of other membrane separation
processes. Improvements in asymmetric membrane preparation methods and
properties were accelerated by the increasing availability in the late 1960s of
scanning electron microscopes, the use of which enabled the effects of structural
modifications to be easily assessed.
1.2. I Phase Inversion (Solution-Precipitation) Membranes
Phase inversion, also known as solution precipitation or polymer

precipitation, is the most important asymmetric membrane preparation method.
In this process, a clear polymer solution is precipitated into two phases: a solid,
polymer-rich phase that forms the matrix of the membrane and a liquid, polymer-
poor phase that forms the membrane pores. If the precipitation process is rapid,
the pore-forming liquid droplets tend to be small and the membranes formed are
markedly asymmetric. If precipitation proceeds slowly, the pore-forming liquid
droplets tend to agglomerate while the casting solution is still fluid, so that the
final pores are relatively large. These membranes have a more symmetrical
structure.
Polymer precipitation from a solution can be achieved in several ways, such

as cooling, solvent evaporation, or imbibition of water. Preparation of
membranes by all of these techniques has been reviewed in the literature.
The simplest technique is thermal gelation, in which a hot one-phase

casting solution is used to cast a film. The cast film cools, reaching the point at
which polymer precipitation occurs. As the solution cools further, precipitation
continues. The process can be represented by the phase diagram shown in
Figure 1-9. The pore volume in the final membrane is largely determined by the
initial composition of the cast film, because this determines the ratio of the
polymer to liquid phase in the cooled film. However, the spatial distribution and
size of the pores is largely determined by the rate of cooling and, hence,
precipitation, of the film. In general, rapid cooling leads to membranes with
small pores.
Solvent evaporation is another method of forming asymmetric membranes.

This technique uses a solvent casting solution consisting of a polymer dissolved in
a mixture of a good volatile solvent and a less volatile nonsolvent (typically water
or alcohol). When a film of this solution is cast and allowed to evaporate, the
good volatile solvent leaves first. The film becomes enriched in the nonsolvent
and finally precipitates. This precipitation process can be represented as a
pathway through a three-component phase diagram, as shown in Figure 1-10.
Finally, the casting solution can be precipitated using imbibition of water,

either as vapor from a humid atmosphere or by immersion in a water bath. This
precipitation process can also be represented as a pathway through a phase
-
diagram. as shown in Figure 1 1 1. This is the Loeb-Sourirajan process.
The use of phase diagrams of the type shown in Figures 1-9, 1-10 and 1-11
to understand the formation mechanism of phase-inversion membranes has been an
area of considerable research in the last 10-15 years.1S'16 The models that have
been developed are complex, and will not be reviewed here. Although the
theories are well developed, it is still not possible to predict membrane
structures only from the phase diagram and solution properties. However, phase
diagrams do allow the effects of principal membrane preparation variables to be
rationalized and are of considerable value in designing membrane development
processes.
1.2.1.1 Polymer precipitation by thermal gelation
The thermal gelation method of making microporous membranes was first

commercialized on a large scale by Ak~o.".'~ Akzo markets microporous
polypropylene and polyvinylidene fluoride membranes under the trade name
Accurel@.
Polypropylene membranes are prepared from a solution of polypropylene in

N,N-bis(2-hydroxyethyl)tallowamine. The amine and polypropylene form a clear
solution at temperatures above 100-150". Upon cooling, the solvent and polymer
phases separate to form a microporous structure. If the solution is cooled slowly,
an open cell structure of the type shown in Figure 1-12 results. The
interconnecting passageways between cells are generally in the micron range. If
the solution is cooled and precipitated rapidly, a much finer structure is formed,
as shown in Figure 1-13. The rate of cooling is therefore a key parameter
determining the final structure of the membrane.
initial temperature
and composition of
casting solution
Cloud point
(point of
first precipitation)
Two-phase region
Composition
of membrane
pore phase
ofpo'ymer 0 20 40 60 80 100
matrix phase
Solution composition (% of solvent)
Figure 1-9. Phase diagram showing the composition pathway traveled by the
casting solution during thermal gelation.
11 2 Membrane Separation Systems
Polymer
Non-solvent
(water)
Figure 1-10. Phase diagram showing the composition pathway traveled by a

casting solution during the preparation of porous membranes by
solvent evaporation.
Polymer
Non-solvent
(water)
Figure 1 - 1 1 . Phase diagram showing the composition pathway traveled by a

coating solution during the preparation of porous membranes by
water imbibition.
Membrane and Module Preparation 11 3
Figure 1-12. Type I "Open Cell" structure of polypropylene formed at low

cooling rates (2,400X).
Figure 1-13. Type I1 "Lacy" structure of polypropylene formed at high

cooling rates (2,OOOX).
A simplified process flowsheet of the membrane preparation process is shown

in Figure 1-14. The hot polymer solution is cast onto a water-cooled chill roll,
which cools the solution causing the polymer to precipitate. The precipitated
film is passed to a n extraction tank containing methanol, ethanol or isopropanol,
which extracts the solvent. Finally, the membrane is dried, sent to a laser
inspection station, trimmed and rolled up. The process shown in Figure 1-14 is
used to make flat-sheet membranes. The preparation of hollow-fiber membranes
by the same general technique has also been described.”
I .2.1.2 Polymer precipitation by solvent evaporation
This technique was one of the earliest methods of making microporous

membranes to be developed, and was used by Bechhold, Elford, Pierce, Ferry
and others in the 1920s and 1930s.20-22 In the method’s simplezt form, a polymer
is dissolved in a two-component solvent mixture consisting of a volatile good
solvent such as methylene chloride, and an involatile poor solvent, such as water
or ethanol. This two-component polymer solution is cast on a glass plate. As
the good, volatile solvent evaporates the casting solution is enriched in the poor,
non-volatile solvent. The polymer precipitates. forming the membrane structure.
The solvent evaporation-precipitation process may be continued until the
membrane has completely formed, or the process can be stopped and the
membrane structure fixed by immersing the cast film into a precipitation bath of
water or other nonsolvent.
Many factors determine the porosity and pore size of membranes formed by
the solvent evaporation method. In general, increasing the nonsolvent content of
the casting solution, or decreasing the polymer concentration, increases porosity.
It is important that the nonsolvent be completely incompatible with the polymer.
If partly compatible nonsolvents are used, the precipitating polymer phase contains
sufficient residual solvent to allow it to flow and collapse as the solvent
evaporates. The result is a dense rather than microporous film.
I .2. I .3 Polymer precipitation by imbibition of water vapor
Preparation of microporous membranes by simple solvent evaporation alone

is not widely practiced. However, a combination of solvent evaporation with
precipitation by imbibition of water vapor from a humid atmosphere is the basis
of most commercial phase-inversion processes. The processes involve proprietary
technology that is not normally disclosed by membrane developers. However,
during the development of composite reverse osmosis membranes at Gulf General
Atomic, Riley et al. prepared this type of membrane and described the technology
in some detail in a series of Office of Saline Water Reports.2s These reports
remain the best published description of this technique.
Extraction and
aftertreatment
liquid
Figure I - 14. Simplified process flowsheet for membrane

preparation by thermal gelation.
The type of equipment used by Riley et al. is shown in Figure 1-15. The
casting solution typically consists of a blend of cellulose acetate and cellulose
nitrate dissolved in a mixture of volatile solvents, such as acetone, and involatile
nonsolvents, such as water, ethanol or ethylene glycol. The polymer solution is
cast onto a continuous stainless steel belt. The cast film then passes through a
series of environmental chamber:. Hot humid air is usually circulated through the
first chamber. The film loses the volatile solvent by evaporation and
simultaneously absorbs water from the atmosphere. The total precipitation process
is slow, taking about 10 minutes to complete. As a result the membrane structure
is fairly symmetrical. After precipitation, the membrane passes to a second oven
through which hot dry air is circulated to evaporate the remaining solvent and
dry the film. The formed membrane is then wound on a take-up roll. Typical
casting speeds are of the order of 1-2 ft/min.
1.2.1.4 Polymer precipitation by immersion in a nonsolvent bath (Loeb-

Sourirajan process)
The Loeb-Sourirajan process is the single most important membrane-

preparation technique. A casting machine used in the process is shown in Figure
1-16?‘ A solution containing approximately 20 wt% of dissolved polymer is cast
on a moving drum or paper web. The cast film is precipitated by immersion in a
bath of water. The water rapidly precipitates the top surface of the cast film,
forming an extremely dense permselective skin. The skin slows down the entry of
water into the underlying polymer solution, which thus precipitates much more
slowly, forming a more porous substructure. Depending on the polymer, the
casting solution and other parameters, the dense skin varies from 0.1-1.0 pm
thick. Loeb and Sourirajan were working in the field of reverse osmosis.”
Later, others adapted the technique to make membranes for other applications,
including ultrafiltration and gas s e p a r a t i ~ n . ~ ’ - ~ ~Ultrafiltration and gas
separation membranes have the same asymmetric structure found in reverse
osmosis membranes, but the skin layer of ultrafiltration membranes contains small
pores 50-200 A in diameter.
A major drawback of gas separation membranes made by the Loeb-Sourirajan

process is the existence of minute membrane defects. These defects, caused by
gas bubbles, dust particles and support fabric imperfections, are almost impossible
to eliminate. Such defects do not significantly affect the performance of
asymmetric membranes used in liquid separation operations, such as ultrafiltration
and reverse osmosis, but can be disastrous in gas separation applications. In the
early 1970s. Browall showed that membrane defects could be overcome by coating
the membrane with a thin layer of relatively permeable material.28 If the coating
was sufficiently thin, it did not change the properties of the underlying
permselective layer. but it did plug membrane defects, preventing simple Poiseuille
gas flow through the defect.
Flowmeter
Later, Henis and Tripodi, at Mon~anto,~'applied this concept to sealing

defects in polysulfone Loeb-Sourirajan membranes with silicone rubber. The form
of the Monsanto membrane is shown in Figure 1-17. The silicone rubber layer
does not function as a selective barrier but rather plugs up defects, thereby
reducing non-diffusive gas flow. The flow of gas through the portion of the
silicone rubber layer over the pore is very high compared to the flow through the
defect-free portion of the membrane. However, because the total area of the
membrane subject to defects is very small, the total gas flow through these
plugged defects is negligible. Because the polysulfone selective layer no longer
has to be completely free of defects, the Monsanto membrane can be made
thinner than is possible with an uncoated Loeb-Sourirajan membrane. The
increase in flux brought about by decreasing the thickness of the permselective
skin layer more than compensates for the slight reduction in flux due to the
silicone rubber sealing layer.
A scanning electron micrograph of a Loeb-Sourirajan membrane is shown in

Figure 1-18. In this example, the permselective film has deliberately been made
thicker than normal to better show the dense surface layer. Loeb-Sourirajan
membranes can have skin layers thinner than 0.1 pm.
Cellulose acetate Loeb-Sourirajan reverse osmosis membranes were introduced

commercially in the 1960s. Since then, many other polymers have been fabricated
into asymmetric membranes in attempts to improve membrane properties. In the
reverse osmosis area, these attempts have had limited success, the only significant
example being DuPont's polyamide membrane. In gas separation and ultrafiltration
a number of membranes with useful properties have been made. However, the
early work on asymmetric, membranes has spawned numerous other technologies in
which a microporous membrane is used as a support to carry another thin, dense
separating layer. The preparation of such membranes is now discussed.
1.2.2 Interfacial Composite Membranes
In the early 1960s, Cadotte at North Star Research developed an entirely

new method of making asymmetric membranes that produced reverse osmosis
membranes with dramatically improved salt rejections and water fluxes.29 In this
method an aqueous solution of reactive prepolymer, such as polyamine, is first
deposited in the pores of a microporous support membrane, typically a
polysulfone Loeb-Sourirajan ultrafiltration membrane. The amine loaded support is
then immersed in a water immiscible solvent solution containing a reactant. for
example, a diacid chloride in hexane. The amine and acid chloride then react at
the interface of the two solutions to form a densely crosslinked, extremely thin
membrane layer. This type of membrane is shown schematically in Figure 1-19.
The first membrane made by Cadotte was based on the reaction of
polyethylenimine crosslinked with toluene-2,4-diisocyanate, shown in Figure 1-20.
The process was later refined by Cadotte et aLSO and Riley et aLs1 in the U.S.
and Nitto in Japan.s2
Loeb-Sourirajan Membrane
Permselective Layer
Microporous Support Layer
Defects
Monsanto Prism@ Membrane
Seal Layer
Permselective Layer
Microporous Support Layer
Figure 1 - 17. Schematic of Loeb-Sourirajan and Monsanto gas

separation membranes.
Dense Microporous
skin layer support layer
Figure I - 18. Scanning electron micrograph of an asymmetric

Loeb-Sourirajan membrane
-
Hexane-Acid
Surface of Polysul-
fone S u p p x t F i l m
Crosslinked .Mine Sone
Figure I - 19. Schematic of the interfacial polymerization

procedure.
I
)&:
c‘13
p w cI?‘0‘1 \ NH
/”
-NH4
/” +-tN
n/N-v‘l
vN-NH
I Nli
I
+ c:o
N -N XN I
1 I\, \ N H
NH
I
iH CH37-J-
NH \ /NH NH
I “ NH+--+~~
c=o
I
I AN*
\N*NH-N-Nt+
P N H r
iH
\
1
Jf’ NH
Hid-C-NH
Nf
NH2 NH-C- NH
II II
0 0
C H ~ C HGROUPS
~ REPRESENTED e y ++
Figure 1 - 20. Idealized structure of polyethylenimine crosslinked with

toluene 2.4-diisocyanate.
Membranes made by interfacial polymerization have a dense, highly

crosslinked interfacial polymer layer formed on the surface of the support
membrane a t the interface of the two solutions. A less crosslinked, more
permeable hydrogel layer forms under this surface layer and fills the pores of
the support membrane. Since the dense crosslinked polymer can only form at the
interface, it is extremely thin, on the order of 0.1 pm or less, and the flux of
permeate is high. Because the polymer is highly crosslinked, its selectivity is also
high. The first reverse osmosis membranes made this way were u p to an order of
magnitude less salt-permeable than other membranes with comparable water fluxes.
When used for gas separation, interfacial polymerization membranes are less
interesting. This is because of the water-swollen hydrogel that fills the pores of
the support membrane. In reverse osmosis. this layer is highly water-swollen and
offers little resistance to the flow of water. When the membrane is dried and
used in gas separations, however, the gel becomes a rigid glass with very low gas
permeability. This glassy polymer fills the membrane pores and, as a result,
defect-free interfacial composite membranes usually have low gas fluxes although
their selectivities can be good.
1.2.3 Solution Cast Composite Membranes
Another very important type of composite membrane is formed by solution

casting a thin (0.5-2.0 pm) film on a suitable microporous film. The first
membranes of this type were prepared by Ward, Browall and others at General
Electric using a type of Langmuir trough system.= In this system a dilute
polymer solution in a volatile water-insoluble solvent is spread over the surface of
a water filled trough. The thin polymer film formed on the water surface is then
picked up on a microporous support. This technique was developed into a
semicontinuous process at General Electric but has not proved reliable enough for
large-scale commercial use.
Currently, most solution cast composite membranes are prepared by a

technique pioneered by Riley and others at UOP.34 In this technique, a polymer
solution is directly cast onto the microporous support film. It is important that
the support film be clean, defect-free and very finely microporous, to prevent
penetration of the coating solution into the pores. If these conditions are met, a
liquid layer 50-100 pm thick can be coated onto the support, which after
evaporation leaves a thin permselective film, 0.5-2-pm thick. This technique was
used by Hennis and Tripodi to form the Monsanto Prism@ gas separation
membranesz7 and at Membrane Technology & Research to form pervaporation and
organic solvent/air separation membra ne^.^^ A scanning electron micrograph of
this type of membrane is shown in Figure 1-21.58
Figure 1- 21. Scanning electron micrograph of a silicone rubber composite

membrane (courtesy of H. Strathmann).
I .2.4 Plasma Polymerization Membranes
Plasma polymerization of films was first used to form electrical insulation

and protective coatings, but a number of workers have also prepared
permselective membranes in this way.37-4o However, plasma polymerization
membranes have never been produced on a commercial scale. A typical, simple
plasma polymerization apparatus is shown in Figure 1-22. Most workers have
used R F fields at frequencies of 2-50 MHz to generate the plasma. In a typical
plasma experiment helium, argon, or other inert gas is introduced at a pressure of
50-100 millitorr and a plasma is initiated. Monomer vapor is then introduced to
bring the total pressure to 200-300 millitorr. These conditions are maintained for
a period of 1-10 minutes. During this time, an ultrathin polymer film is deposited
on the membrane sample held in the plasma field.
Monomer polymerization proceeds by a complex mechanism involving ionized

molecules and radicals and is completely different from conventional
polymerization reactions. In general, the polymer films are highly crosslinked
and may contain radicals which slowly react on standing, The stoichiometry of
the film may also be quite different from the original monomer due to
fragmentation of monomer molecules during the plasma polymerization process.
It is difficult to predict the susceptibility of monomers to plasma polymerization
or the nature of the resulting polymer film. For example, many vinyl and acrylic
monomers polymerize very slowly, while unconventional monomers such as benzene
and hexane polymerize readily. The vapor pressure of the monomers, the power
and voltage used in the discharge reaction, the type and the temperature of the
substrate all affect the polymerization reaction. The inert gas used in the plasma
may also enter into the reaction. Nitrogen and carbon monoxide, for example, are
particularly reactive. In summary, the products of plasma polymerization are ill
defined and vary according to the experimental procedures. However, the films
produced can be made very thin and have been shown to be quite permselective.
The most extensive studies of plasma polymerized membranes have been

performed by Yasuda, who has tried to develop high-performance reverse osmosis
membranes by depositing plasma films onto microporous polysulfone f i l m ~ . ~ ~ ~ ~ *
Other workers have also studied the gas permeabiiity of plasma polymerized films.
For example, Stancell and Spencer39 were able to obtain a plasma membrane with
a selectivity aH2/CH, of almost 300, while Kawakami et ai.4o have reported
plasma membranes with a selectivity aOz/,Nz of 5.8. Both of these selectivities
are very high when compared to the selectivity of other membranes. The plasma
films were also quite thin and had a high flux. However, in both cases the
plasma film was formed on a substrate made from thick, dense polymer films 25-
100 wn thick. As a result, the flux through the overall composite membrane was
still low.
To
vacuum
Pump
Glass
Film Brass
electrode
Figure I - 22. Simple plasma coating apparatus.

1.2.5 Dynamically Formed Membranes
In the late 1960s and early 1970s. a great deal of attention was devoted to
preparing dynamically formed anisotropic membranes, principally by Johnson, Kraus
and others a t Oak Ridge National Laboratories."*" The general procedure used
is to form a layer of inorganic or polymeric colloids on the surface of a
microporous support membrane by filtering a solution containing suspended
colloid through the support membrane. A thin layer of colloids is then laid
down on the membrane surface and acts as a semipermeable membrane. Over a
period of time the colloidal surface layer erodes or dissolves and the membrane's
performance falls. The support membrane is then cleaned and a new layer of
colloid is deposited. In the early development of this technique a wide variety of
support membranes were used, ranging from conventional ultrafiltration membranes
to fire hoses. In recent years microporous ceramic or carbon tubes have emerged
as the most commonly used materials. Typical colloidal materials are polyvinyl
methylether, acrylic acid copolymers, or hydrated metal oxides such as zirconium
hydroxide.
Dynamically formed membranes were pursued for many years as reverse

osmosis membranes because they had very high water fluxes and relatively good
salt rejection, especially with brackish water feeds. However. the membranes
proved to be unstable and difficult to reproduce reliably and consistently. For
these reasons, and because high-performance interfacial composite membranes
were developed in the meantime, dynamically formed membranes gradually fell out
of favor. Gaston County Corporation, using microporous carbon tube supports, is
the only commercial supplier of dynamic membrane equipment. The principal
application is polyvinyl alcohol recovery in textile dyeing operations.
I .2.6 Reactive Surface Treatment
Recently a number of groups have sought to improve the properties of

asymmetric gas separation membranes by chemically modifying the surface
permselective layer. For example, Langsam at Air Products has treated films of
poly( trimethylsilyl- 1-propyne) with diluted fluorine gas."-" This treatment
chemically reacts fluorine with the polymer structure in the first 100-200 A-
thick surface layer. This treatment produces a dramatic improvement in
selectivity with a modest reduction in permeability. The surface treated polymer
is superior to conventional polymers, even though the permeability is less than
that for untreated poly(trjmetbylsilyl- I-propyne).
Calculations of the permeability and selectivity for the treated surface layer
indicate that permeation through this layer is a diffusion controlled process.
Since the untreated poly(trimethylsilyl- I -propyne) has extremely high free volume
and correspondingly high permeant diffusivity, it is evident that the fluorination
is reducing the free volume of the surface layer and thereby decreasing the
diffusivity of the permeant.
1.3 CERAMIC A N D METAL MEMBRANES
1.3.1 Dense Metal Membranes
Palladium and palladium alloy membranes can be used to separate hydrogen

from other gases. Palladium membranes were extensively studied during the
1950s and 1960s and a commercial plant to separate hydrogen from refinery off-
gas was installed by Union Carbide.46*47 The plant used palladium/silver alloy
membranes in the form of 25 pm-thick films. The technique used to make the
membrane is not known, but films of this thickness could be made by standard
metal casting and rolling technology. A number of problems, including long-term
membrane stability under the high temperature operating conditions, were
encountered and the plant was replaced by pressure-swing adsorption systems.
Small-scale systems, used to produce ultrapure hydrogen for specialized

applications, are currently marketed by Johnson Matthey and Co. These systems
also use palladium/silver alloy membranes. based on those developed by
H ~ n t e r . ' ~ *Membranes
~~ with much thinner effective palladium layers than were
used in the Union Carbide installation can now be made. One technique that has
been used is to form a composite membrane, with a metal substrate, onto which is
coated a thin layer of palladium or palladium alloy.50 The palladium layer can be
applied by spraying, electrochemical deposition, or by vacuum methods, such as
evaporation or sputtering. Coating thicknesses on the order of a few microns or
less can be achieved. Johnson Matthey and Co. holds several patents covering
this type of technology, including techniques in which composites are coated on
both sides with palladium and then rolled down to form a very thin composite
film.5' The exact nature of the membranes used in their present systems is
proprietary.
1.3.2 Microporous Metal Membranes
Recently, Alcan I n t e r n a t i ~ n a l ~and

~ . ~A~ l u ~ u i s s ehave
~ ~ ~begun
~ ~ to produce
microporous aluminum membranes. The production methods used are proprietary,
but both use an electrochemical technique. In the Alcan process aluminum metal
substrate is subjected to electrolysis using an electrolyte such as oxalic or
phosphoric acid. This forms a porous aluminum oxide film on the anode. The
pore size and structure of the oxide film is controlled by varying the voltage.
Once formed, the oxide film is removed from the anode by etching. The
membranes have a tightly controlled pore size distribution and can be produced
with pore diameters ranging from 0.02-2.0 pm.
1.3.3 Ceramic Membranes
The major advantage of ceramic membranes is that they are chemically

extremely inert and can be used at high temperatures, where polymer films fail.
Ceramic membranes can be made by three processes: sintering, leaching and sol-
gel methods. Sintering involves taking a colloidal suspension of particles,
forming a coagulated thin film, and then heat treating the film to form a
continuous, porous structure. The pore sizes of sintered films are relatively
large, of the order of 10-100 pm. In the leaching process, a glass sheet or
capillary incorporating two intermixed phases is treated with an acid that will
dissolve one of the phases. Smaller pores can be obtained by this method, but
the uniformity of the structure is difficult to control. The preparation of ceramic

membranes by a sol-gel technique is the newest approach, and offers the greatest
potential for making membranes suitable for gas separation.
~ process on
Figure 1-23 summarizes the available sol-gel p r o c e s s e ~ . ~The
the right of the figure involves the hydrolysis of metal alkoxides in a water-
alcohol solution. The hydrolyzed alkoxides are polymerized to form a chemical
gel, which is dried and heat treated to form a rigid oxide network held together
by chemical bonds. This process is difficult to carry out, because the hydrolysis
and polymerization must be carefully controlled. If the hydrolysis reaction
proceeds too far, precipitation of hydrous metal oxides from the solution starts to
occur, causing agglomerations of particulates in the sol.
In the process on the left, complete hydrolysis is allowed to occur. Bases or

acids are added to break u p the precipitate into small particles. Various reactions
based on electrostatic interactions a t the surface of the particles take place. The
result is a colloidal solution. Organic binders are added to the solution and a
physical gel is formed. The gel is then heat treated as before to form the
ceramic membrane.
The sol-gel technique has mostly been used to prepare alumina membranes to
date. Figure 1-24 shows a cross section of a composite alumina membrane made
by slip casting successive sols with different particle sizes onto a porous ceramic
support. Silica or titanic membranes could also be made using the same
principles. Anderson et al.56 have made unsupported titanium dioxide membranes
with pore sizes of 50 A or less by the sol-gel process shown in Figure 1-25.
1.3.4 Molecular Sieve Membranes
A common, albeit. inaccurate analogy used to describe a membrane is that

of a sieve, with smaller species passing through a porous barrier while the larger
species are retained. In practice, most membranes rely on the permeant-
selective properties of a non-porous barrier, whether it be liquid, polymer or
metal, to separate the components of a mixture. One could use a porous barrier,
if it were sufficiently finely porous to achieve a sieving at the molecular level.
However, until recently membranes with the necessary pore size and pore
distribution were not known.
Recent developments in the field of inorganic membranes offer promise of

the development and commercialization of molecular-sieve type membranes in the
next decade, especially for gas separation applications. A substantial amount of
work on molecular sieve membranes is being performed in Israel, where the
membranes are closest to scale-up. Soffer, Koresh and co-workers have been
working on carbon membranes obtained by heating polymeric hollow fibers to
between 500-800°C in an inert atmosphere or in a vacuum, which reduces the
polymer to carbon.57 The membranes are reported to have pore diameters
between 2-5 A.
Suspension Sol
Thin layer Sol layer Sol layer
layer
Figure 1-24. Cross-sectional scanning electron micrograph of a three-

layered alumina membrane/support (pores of 0.2, 0.8 and
IO pm, respectively, in each layer).
I Clear solutlon Hydroxide preclpltallon
I Polymeric sol Suspension
12 hours heating 85-95'C
Colloidal suspension
- Cool to room temperature
4 Stable coiloidal sol
Dipping
___c
Pouring
1
Sol in plastic container
LrlSupported membrane
Gelling
Controlled
relative humidity
Q Clear supported
gel membrane
Firing 400-5OO'C
Ceramic membrane
Controlled
Transparent unsupported
membrane
In the United States, PPG Industries is active in the area of molecular

sieve membranes and is believed to have developed IOA-diameter glass hollow-
fiber membranes with carbon dioxide/methane selectivity of 50." Beaver
describes the process used for producing microporous glass hollow-fiber
membranes by a leaching process. The starting material for the membranes is a
glass containing at least 30-70% silica, as well as oxides of zirconium, hafnium or
titanium and extractable materials. The extractable materials comprise one or
more boron-containing compounds and alkali metal oxides and/or alkaline earth
metal oxides. Glass hollow fibers produced by melt extrusion are treated with
dilute hydrochloric acid at 90'C for 2-4 hours to leach out the extractable
materials, washed to remove residual acid and then dried. This technique has
been used to produce hollow fibers membranes with 8-10 A-diameter pores.
I .4 LIQUID MEMBRANES
Liquid membranes containing carriers to facilitate selective transport of

gases or ions were a subject of considerable research in the 1960s and 1970s.
Although still being worked on in a number of laboratories, the development of
much more selective conventional polymer membranes in recent years has
diminished interest in these films.
The earliest liquid membrane experiments were done by biologists, using

natural carriers contained in cell walls. As early as 1890. Pfeffer postulated
transport properties in membranes using carriers. Perhaps the first coupled
transport experiment was performed by Osterhout, who studied the transport of
ammonia across algae cell walls.58 By the 1950s the carrier concept was well
developed, and workers began to develop synthetic biomembrane analogues of the
natural systems. For example, in the mid-1960s. Sollner and Shean5' studied a
number of coupled transport systems using inverted U-tubes, shown in Figure I -
26a. At the same time, Bloch and Vofsi published the first of several papers in
which coupled transport was applied to hydrometallurgical separations, namely the
separation of uranium using phosphate esters.60 Because phosphate esters were
also plasticizers for polyvinyl chloride (PVC), Bloch and Vofsi prepared
immobilized liquid films by dissolving the esters in a PVC matrix. Typically, the
PVC/ester film was cast on a paper support. Researchers actively pursued this
work until the late 1960s. At that time, interest in this approach lagged,
apparently because the fluxes obtained did not make the process competitive with
conventional separation processes. Some workers are continuing to apply these
membranes to metal separations, but most current interest in PVC matrix
membranes is in their use in ion selective membrane electrodes.61
Following the work of Bloch and Vofsi, two other methods of producing
immobilized liquid films were introduced. Both are still under development. In
the first approach. the liquid carrier phase is held by capillarity within the pores
of a microporous substrate, as shown in Figure 1-26b. This approach was first
used by Miyauchi62 and further developed by Baker et al." and by Largman and
Sifniade~.~' The principal objective of this early work was the recovery of
copper and other metals from hydrometallurgical solutions. Despite considerable
effort on the laboratory scale, the process has never become commercial. The
major problem is instability of the liquid carrier phase in the microporous

membrane support.
The second type of immobilized liquid carrier is the emulsion or "bubble"

membrane. in this technique a surfactant-stabilized emulsion is produced as
shown in Figure 1-26c. The organic carrier phase forms the wall of the emulsion
droplet separating the aqueous feed from the aqueous product solutions. Metal
ions are concentrated in the interior of the droplets. When sufficient metal has
been extracted, emulsion droplets are separated from the feed and the emulsion is
broken, liberating a concentrated product solution and an organic carrier phase.
The carrier phase is decanted from the product solution and recycled to make
more emulsion droplets. The principal technical problem is the stability of the
liquid membrane. Ideally, the emulsion membranes must be completely stable
during the extraction step, to prevent mixing of the two aqueous phases, but must
be completely broken and easily separated in the stripping step. Achieving this
level of control over emulsion stability has proved difficult.
The emulsion membrane technique was popularized and fully developed by Li

and his co-workers at E X X O ~ Starting
. ~ ~ ~in ~the~ late 1950s and continuing for
more than twenty years, the Exxon group's work led to the installation of the
first pilot plant in 1979. However, the process is still not commercial, although a
number of pilot plants have been installed, principally on hydrometallurgical feed
streams. Another important group working independently on the problem at about
the same time was Cussler et al. at Carnegie Mellon U n i v e r ~ i t y . ~More
~ recent
workers in the field include Halwachs and Schurgerlee in West Germany, Marr and
Kapp in Austria," Stelmaszek et al. in Poland,70 Martin and Davis in the United
Kingdom,71 and Danesi et al..72 Noble et ai.,75 and Lamb, Christensen and Izatt7'
in the United States.
1.5 HOLLOW-FIBER MEMBRANES
Most of the techniques described above were originally developed to

produce flat-sheet membranes. However, most techniques can be adapted to
produce membranes in the form of thin tubes or fibers. Formation of membranes
into hollow fibers has a number of advantages. one of the most important of
which is the ability to form compact modules with very high surface areas. This
advantage is offset, however, by the lower fluxes of hollow-fiber membranes
compared to flat-sheet membranes made from the same materials. Nonetheless,
the development of hollow-fiber membranes by Mahon and the group at Dow
Chemical in 1960,75*7s and their later commercialization by Dow Chemical,
DuPont, Monsanto and others, represents one of the major events in membrane
technology.
Hollow fibers are usually on the order of 25-200 pm in diameter. They can
be made with a homogeneous dense structure, or more preferably as a
microporous structure having a dense permselective layer on the outside or inside
surface. The dense surface layer can be integral. or separately coated. The
fibers are packed into bundles and potted into tubes to form a membrane module.
More than a kilometer of fibers is required to form a membrane module with a
surface area of one square meter. Since no breaks or defects are allowed in a
module, this requires very high standards of reproducibility and quality control.
Organic carrier phase
Feed solution Product solution
(a) U-Tube Liquid Membranes
Feed solution Product solution
Microporous Organic
polymeric carrier
membrane phase
(b) Supported Liquid Membranes
Product solution
Organic carrier
phase
<
t
Feed solution
(c) Emulsion Liquid Membranes
Figure 1-26. A schematic illustration of three types of liquid membranes.

Hollow-fiber fabrication methods can be divided into two classes. The most
common is solution spinning, in which a 20-30% polymer solution is extruded and
precipitated into a bath of nonsolvent. Solution spinning allows fibers with the
asymmetric Loeb-Sourirajan structure to be made. An alternative technique is
melt spinning. in which a hot polymer melt is extruded from an appropriate die
and is then cooled and solidified in air or a quench tank. Melt-spun fibers are
usually dense and have lower fluxes than solution-spun fibers, but, because the
fiber can be stretched after it leaves the die, very fine fibers can be made. Melt
spinning can also be used with polymers such as poly(trimethy1pentene) which are
not soluble in convenient solvents and are therefore difficult to form by wet
spinning.
I 3 . 1 Solution (Wet) Spinning
The most widely used solution spinnerette system was first devised by
M a h ~ n . ~ The
~ , ~spinnerette
~ consists of two concentric capillaries, the outer
capillary having a diameter of approximately 400 pm and the central capillary
having an outer diameter of approximately 200 pm and an inner diameter of 100
pm. Polymer solution is forced through the outer capillary while air or liquid is
forced through the inner one. The rate at which the core fluid is injected into
the fibers relative to the flow of polymer solution governs the ultimate wall
thickness of the fiber. Figure 1-27 shows a cross-section of this type of
spinnerette. Unlike solution-spun textile fibers, solution-spun hollow fibers are
not stretched after leaving the spinnerette.
A complete hollow-fiber spinning system is shown in Figure 1-28. Fibers are

formed almost instantaneously as the polymer solution leaves the spinnerette. The
amount of evaporation time between the solution’s exit from the spinnerette and
its entrance into the coagulation bath has been found to be a critical variable by
the Monsanto group. If water is forced through the inner capillary, an
asymmetric hollow fiber is formed with the skin on the inside. If air under a few
pounds of pressure, or an inert liquid, is forced through the inner capillary to
maintain the hollow core, the skin is formed on the outside of the fiber by
immersion into a suitable coagulation bath.
Wet-spinning of this type of hollow fiber is a well-developed technology,

especially in the preparation of dialysis membranes for use in artificial kidneys.
Systems that can spin more than 100 fibers simultaneously on an around-the-clock
basis are in operation.
1.5.2 Melt Spinning
In melt spinning, the polymer is extruded through the outer capillary of the
spinnerette as a hot melt, the spinnerette assembly being maintained at a
temperature between 100-300°C. The polymer can either be used pure or loaded
with up to 50 wt% of various plasticizers. Melt-spun fibers can be stretched as
they leave the spinnerette, facilitating the production of very thin fibers. This is
a major advantage of melt spinning over solution spinning. The dense nature of
melt-spun fibers leads to lower fluxes than can be obtained with solution-spun
fibers, but because of the enormous membrane surface area of hollow-fiber
systems this need not be a problem.
Silver solder
tube
Figure 1-27. A spinnerette design used in solution-spinning of hollow-fiber

membranes.
Evaporation gap
Coagulation bath
Figure 1-28. A hollow-fiber solution-spinning system.

Substantial changes in the fiber properties can be made by varying the

added plasticizer. A particularly interesting additive is one which is completely
miscible with the melt at the higher temperature of the spinnerette, but becomes
incompatible with the polymer a t room temperature. If the additive is soluble, it
can then be leached from the fiber, leaving minute voids behind. This produces a
mechanism for making microporous melt-spun fibers.
I .6 MEMBRANE MODULES
A useful membrane process requires the development of a membrane module

containing large surface areas of membrane. The development of the technology
to produce low cost membrane modules was one of the breakthroughs that led to
the commercialization of membrane processes in the 1960s and 1970s. The earliest
designs were based on simple filtration technology and consisted of flat sheets of
membrane held in a type of filter press. These are called plate and frame
modules. Systems containing a number of membrane tubes were developed at
about the same time. Both of these systems are still used, but because of their
relatively high cost they have been largely displaced by two other designs, the
spiral-wound and hollow-fiber module. These designs are illustrated in Figure 1 -
29.
1.6.1 Spiral-Wound Modules
Spiral-wound modules were originally used for artificial kidneys, but were
fully developed for reverse osmosis systems. This work, carried out by UOP
under sponsorship of the Office of Saline Water (later the Office of Water
Research and Technology) resulted in a number of spiral-wound design^.^^-^' The
design shown in Figure 1-30 is the first and simplest design, consisting of a
membrane envelope wound around a perforated central collection tube. The wound
module is placed inside a tubular pressure vessel and feed gas is circulated
axially down the module across the membrane envelope. A portion of the feed
permeates into the membrane envelope, where it spirals towards the center and
exits via the collection tube.
Simple laboratory spiral-wound modules 12-inches-long and 2-inches in

diameter consist of a simple membrane envelope wrapped around the collection
tube. The membrane area of these modules is typically 2-3 ft'. Commercial
spiral-wound modules are typically 36-40 inches long and have diameters of 4, 6,
8, and I2 inches. These modules typically consist of a number of membrane
envelopes, each with an area of approximately 20 ft', wrapped around the central
collection pipe. This type of multileaf design is illustrated in Figure 1-31.77
Multileaf designs are used to minimize the pressure drop encountered by the
permeate gas traveling towards the central pipe. If a simple membrane envelope
were used this would amount to a permeate spacer length of 5-25 meters,
producing a very large pressure drop, especially with high flux membranes.
--*
"
"
PB -"
C
aJ
i
3
.e
g
L
n
4
f
4
.-OD
D
2 5 ri
v
2 al
P
0
M
.d
L
0
lA 0
0
m
L.
.r(
-
B
C
.I
_
C
3..
-0
o"d 82
:; 0;
- aJ
-
N
I
5m c t!
- -c
8
.
3
.-LL
00
c
0 D
2 E
a
r(
1
3
P
L.
P)
D
.r(
L
$
r(
--F!
7 4
z
D P
U
/
-
Modu'e housing
Residue flow
Permeate flow
Residue flow
Spacer
Membrane
Spacer
Permeate flow
afier passing through
membrane
Figure 1-30. Spiral-wound membrane module.
Membrane envelope
Membrane envel
Membrane envelope
Figure 1-31. Schematic of a four-leaf, spiral-wound module.

I .6.2 Hollow-Fiber Modules
Hollow-fiber membrane modules are formed in two basic geometries. The

first is the closed-end design illustrated in Figure 1-32 and used, for example, by
Monsanto in their gas separation systems or DuPont in their reverse osmosis
fiber systems. In this module, a loop of fiber or a closed bundle is contained in
a pressure vessel. The system is pressurized from the shell side and permeate
passes through the fiber wall and exits via the open fiber ends. This design is
easy to make and allows very large fiber membrane areas to be contained in an
economical system. Because the fiber wall must support a considerable hydrostatic
pressure, these fibers usually have a small diameter, on the order of 100 pm ID
and 150-200 pm OD.
The second type of hollow-fiber module is the flow through system

illustrated in Figure 1-33. The fibers in this type of unit are open at both ends.
In this system, the feed fluid can be circulated on the inside or the outside of
the fibers. To minimize pressure drops in the inside of the fibers, the fibers
often have larger diameters than the very fine fibers used in closed loop
systems. These so-called spaghetti fibers are used in ultrafiltration,
pervaporation and in some low to medium pressure gas applications, with the feed
being circulated through the lumen of the fibers. Feed pressures are usually
limited to less than 150 psig in this type of application.
1.6.3 Plate-and-Frame Modules
Plate-and-frame modules were among the earliest types of membrane systems

and the design has its origins in the conventional filter-press. Membrane feed
spacers and product spacers are layered together between two end plates, as
illustrated in Figure 1-34. A number of plate-and-frame units have been
developed for small-scale applications, but these units are expensive compared to
the alternatives and leaks caused by the many gasket seals are a serious problem.
The use of plate-and-frame modules is now generally limited to electrodialysis
and pervaporation systems or small ultrafiltration and reverse osmosis systems.
I .6.4 Tubular Systems
Tubular modules are now used in only a few ultrafiltration applications

where the benefit of resistance to membrane fouling because of good fluid
hydrodynamics overcomes the problem of their high capital cost.
1.6.5 Module Selection
The choice of an appropriate module design for a particular membrane

separation is a balance of a number of factors. The principal factors that enter
into this decision are listed in Table 1-1.
Residue stream f
--
1 ,
Feed stream
Permeate stream -f
The paper clip illustrates the relative

size of the hollow fibers inside a
Prism Alpha separator.
Figure 1-32. Schematic of a Monsanto Prism@ closed-end module.

kbllw fiber a l e shell
Pmducr
Figure 1-33. Schematic of a capillary membrane module.
Membrane support plate
Membrane support plate
Figure 1-34. Schematic of plate-and-frame membrane system.

Cost, although always important, is a difficult factor to quantify. The

actual selling price of membrane modules varies widely depending on the
application. Generally, high-pressure modules are more expensive than low-
pressure or vacuum systems. The selling price also depends on the volume of the
application and the pricing structure adopted by the industry. For example,
membranes for reverse osmosis of brackish water are produced by many
manufacturers. As a result, competition is severe and prices are low. Similar
modules used in other separations are much more expensive. Our estimate of the
cost of producing membrane modules is given in Table 1-1; the selling price is
typically two to five times higher.
Table 1 - 1. Characteristics of the Major Module Designs
Hollow Capillary Spiral- Plate-

Fine Fibers Fibers Wound and-Frame Tubular
Manufactur-
ing cost
($/m2) 5 - 20 20 - 100 30 - 100 100 - 300 50 - 200
Packing
density high moderate moderate low low
Resistance
to fouling very poor good moderate good very good
Parasitic
pressure
drops high moderate moderate moderate low
Suitable for can be can be

high pressure done with done with
operation? yes no Yes difficulty difficulty
Limited
to specific types
of membranes? yes Yes no no no
A second major factor determining module selection is resistance to fouling.

Membrane fouling is a particularly important problem in liquid separations such as
reverse osmosis and ultrafiltration. In gas separation applications fouling is more
easily controlled.
A third factor is the ease with which various membrane materials can be
fabricated into a particular module design. Almost all membranes can be formed
into plate and frame modules, for example, but relatively few materials can be
fabricated into hollow fine fibers or capillary fibers. Finally, the suitability of
the module design for high pressure operation and the relative magnitude of
pressure drops on the feed and permeate sides of the membrane can sometimes be
important considerations.
In reverse osmosis, most modules are of the hollow fine fiber or spiral-
wound design. Plate-and-frame and tubular modules are used in a few
applications where membrane fouling is particularly severe, for example. food
applications or processing of heavily contaminated industrial water. Currently,
spiral-wound modules appear to be displacing hollow fiber designs because they
are inherently more fouling resistant and feed pretreatment costs are therefore
lower. Also. the thin film interfacial composite membranes that are the best
reverse osmosis membranes now available, cannot be fabricated into hollow fine
fiber form.
In ultrafiltration applications, hollow fine fibers have never been seriously

considered because of their susceptibility to fouling. If the feed solution is
extremely fouling, tubular or plate-and-frame systems are used. In recent years,
however, spiral modules with improved resistance to fouling have been developed
and these modules are increasingly displacing the more expensive plate-and-frame
and tubular systems. Capillary systems are also used in some ultrafiltration
applications.
In high pressure gas separation applications, hollow fine fibers appear to

have a major segment of the market. Hollow-fiber modules are clearly the lowest
cost design per unit membrane area, and the poor resistance of hollow-fiber
modules to fouling is not a problem in gas separation applications. High pressure
gas separations also require glassy rigid polymer materials such as polysulfone,
polycarbonate, and polyimides. all of which can easily be formed into hollow fine
fibers. Of the major companies servicing this area only Separex and W.R. Grace
use spiral-wound modules.
Spiral-wound modules are much more commonly used in low pressure or

vacuum gas separation applications, for example, the production of oxygen-
enriched air, or the separation of organic vapors from air. In these applications,
the feed gas is at close to ambient pressure and a vacuum is drawn on the
permeate side of the membrane. Parasitic pressure drops on the permeate side of
the membrane and the difficulty in making high-performance hollow fine fiber
membranes from the rubbery polymers used in these membranes both militate
against the hollow fine fiber designs.
Pervaporation has the same type of operational constraints as low pressure

gas separation. Pressure drops on the permeate side of the membrane must be
small and many pervaporation membrane materials are rubbery. For this reason,
spiral-wound modules and plate-and-frame systems are both being used. Plate-
and-frame systems are competitive in this application despite their high cost
primarily because they can be operated at high temperatures with relatively
aggressive feed solutions, where spiral-wound systems would fail.
1.7 CURRENT AREAS OF MEMBRANE AND MODULE RESEARCH
During the past 40 years membrane technology has moved from laboratory to
industrial scale. Annual sales of the four developed membrane processes,
microfiltration, ultrafiltration, reverse osmosis and electrodialysis, are of the order
of SI billion. More than one million square meters of membranes are installed in
industrial plants using these processes around the world. Because these processes
are now developed, opportunities for technological breakthroughs in new

membranes and modules with dramatic consequences are limited. A survey of the
patent literature shows this. Currently between 80-100 U.S. patents on various
improvements to microfiltration, ultrafiltration, reverse osmosis or electrodialysis
membranes and modules are issued annually. Almost all of these patents cover
small improvements to existing techniques or systems. Progress in these areas
will continue in the future but is likely to occur through a large number of
small, incremental advances.
Areas in which breakthroughs in membrane and module technology are

occurring are gas separation and, to a lesser extent, pervaporation and facilitated
transport. These are developing processes and a number of innovative concepts
are being explored that could produce major changes. The total number of U.S.
patents on membrane and module preparation in these areas in 1988-9 was
between 60-80. Although nearly equal in number to the established process
patents, the flavor of these patents taken together is quite different. They are,
in general, more innovative; some cover substantial advances. Gas separation,
therefore, currently represents the advancing frontier of modern membrane
technology.
REFERENCES
1. R. Rautenbach and R. Albert, Membrane Processes, John Wiley & Sons,

Chichester, England (1989).
2. D.H. Kaelble, "Spin Casting of Polymer Films," J, A D D ~Polvm.

. Sci. 9, 1209
( 1965).
3. R.L. Fleischer, P.R. Price and R.M. Walker, "Solid-state Track Detectors:
Applications to Nuclear Science and Geophysics," Ann. Review Nucl. Sci. 15,
1 (1965).
4. R.L. Fleischer, P.B. Price and R.M. Walker, Nuclear Tracks in Solids," S&
American 22Q (June 30, 1969).
5. H.S. Bierenbaum, R.B. Isaacson, M.L. Druin and S.G.Plovan, "Microporous

Polymeric Films." Ind. Ena. Chem.. Prod. Res. Develoo. 13(1), 2 (1974).
6. Okugama, U.S. Patent 4,585,604.
7. R.W. Gore, U.S. Patent 4,187,390 (Feb. 1980).
8. U.S. Patent 3,532,527.
9. K. Gerlach, E. Kressler and W. Henne, U.S. Patent 4,708,799 (1987).
10. T. Ichikawa, K. Takahara, K . Shimoda, Y. Seita and M. Emi, U.S. Patent

4,708,800 (1987).
11. R.P. Beaver, U.S. Patent 4,778,499 (Oct. 1988).
12. S. Loeb and S . Sourirajan, "Sea Water Demineralization by Means of an

Osmotic Membrane," Saline Water Conversation. Vol. 2, R.F. Could (Ed.),
ACS Symposium Series 38, American Chemical Society, Washington, w, p.
117 (1963).
13. H. Strathmann, K. Kock, P. Amar and R.W. Baker, "The Formation

Mechanism of Asymmetric Membranes," Desalination 16, 179 (1975).
14. H. Strathmann and K. Kock, "The Formation Mechanism of Phase Inversion

Membranes," Desalination 2 1 , 241 (1977).
15. J.G. Wijmans and C.A. Smolders, "Preparation of Asymmetric Membranes by

the Phase Inversion Process," in Synthetic Membranes: Science, Engineering
Lnd ADDlicationg, P.M. Bungay. H.K. Lonsdale and M.N. de Pinho (Eds.), D.
Reidel, Dordrecht, The Netherlands (1986).
16. M.H.V. Mulder, J. Oude Hendrikman, J.G. Wijmans and C.A. Smolders, "A
Rationale for the Preparation of Asymmetric Pervaporation Membranes," I,
ADDI. Polvm. Sci. 3Q, 2805 (1985).
17. A.J. Castro, U.S. Patent 4,247,498 (1981).

18. W.C. Hiatt, G.H. Vitzhum, K.B. Wagener, K. Gerlach and C. Josefiak,
"Microporous Membranes via Upper Critical Temperature Phase Separation,"
in Materials Science of Synthetic Membrana , D.R. Lloyd (Ed.), American
Chemical Society, Washington, D.C. (1985).
19. W. Henne, M. Pelger, K. Gerlach and J. Tretzel, in Plasma S e wion and

Plasma Fractionation, M.J. Lysaght and H.J. Gurland (Eds.), S. Karger AG,
Basel, Switzerland (1983).
20. W.J. Elford. "Principles Governing the Preparation of Membranes Having

Graded Porosities, The Properties of "Gradocol" Membranes as Ultrafilters,"
Trans. Far. SOC.33, 1094 (1937).
21. R. Zsigmondy and W. Bachmann, "Uber neue filter," Z. Anorn. Chem. 101, 119
(1918).
22. J.D. Ferry, "Ultrafilter Membranes and Ultrafiltration," Chem Rev. 18, 373
( 1936).
23. R.L. Riley, H.K. Lonsdale, L.D. LaGrange and C.R. Lyons, "Development of
Ultrathin Membranes," Qffice of Saline Water Reoort No. 386, PB# 207036.
Jan. (1969).
24. B.J. Epperson and L.J. Burnett. "Development and Demonstration of a Spiral-
Wound Thin-Film Composite Membrane System for the Economical Production
of Oxygen-Enriched Air." Final ReDort DeDartment of Enerav No. C S / a .
November (1983).
25. A.S. Michaels. U.S. patent 3,615,024 (Oct. 1971).
26. J.M.S. Henis and M.K. Tripodi, "A Novel Approach to Gas Separations Using
Composite Hollow Fiber Membranes," a.Sci. & Tech. IS, 1059 (1980).
27. J.M.S. Henis and M.K. Tripodi, U.S.Patent 4,230,463 (1980).
28. W.R. Browall, U.S. Patent 3,980,456 (Sept. 1976).
29. L.T. Rozelle, J.E. Cadotte, K.E. Cobian and C.V. Kopp, Jr., "Non-
Polysaccharide Membranes for Reverse Osmosis, NS- 100 Membranes," s.
Sourirajan (Ed.), National Research Counc il of Canada Pub. NRCC 15627,
(1977).
30. J.E. Cadotte, "Evolution of Composite Reverse Osmosis Membranes," in

Materials Science of Svnthetic Membranes, D.R. Lloyd (Ed.), American
Chemical Society, Washington, D.C. (1985).
31. R.L. Riley, R.L. Fox, C.R. Lyons, C.E. Milstead, N.W. Seroy and M. Tagami,
Pesalination 19, I13 (1976).
32. Y. Kamiyama, N. Yoshioka, K. Matsui and E. Nakagome, "New Thin-Film

Composite Reverse Osmosis Membranes and Spiral- Wound Modules,"
. .
sailnation 51, 79 (1984).
33. W.J. Ward, W.R. Browall and R.M. Salemme, "Ultrathin Silicone Rubber
Membranes for Gas Separation," J. Membra ne Sci. 1, 99 (1976).
34. R.L. Riley and R.L. Grabowsky, "Preparation of Gas Separation Membranes,"
U.S. Patent 4,234.701 (January 1984).
35. R.W. Baker, U.S. Patent 4,553,983 (Nov. 1985).
36. H.S. Strathmann, private communication.
37. H. Yasuda, "Plasma Polymerization for Protective Coatings and Composite

Membranes," J. Memb. Sci. 18, 273. (1984).
38. H. Yasuda, "Composite Reverse Osmosis Membranes Prepared by Plasma

Polymerization," in Reverse Osmosis and Synthetic Membranes, S. Sourirajan
(Ed.), National Research Council of Canada (1977).
39. A.R. Stancell and A.T. Spencer, "Composite Permselective Membrane by

Deposition of an Ultrathin Coating from a Plasma," J. ADDI. Polv. Sci. 16,
1505 (1972).
40. M. Kawakami, Y. Yamashita, M. Iwamoto and S. Kagawa, "Modification of Gas

Permeabilities of Polymer Membranes by Plasma Coating," J. Memb. Sci. 19,
249 ( 1984).
41. K.A. Kraus, A.J. Shor and J.S. Johnson, Desalination. 2, 243 (1967).
42. J.S. Johnson, K.A. Kraus, S.M. Fleming, H.D. Cochran and J.J. Perona,
Desalination. 5, 359 (1968).
43. M. Langsam, U.S. Patent 4,657,564 (1987).
44. M. Langsam, A.C. Savoca, U.S. Patent 4,759,776 (July 1988).
45. M. Langsam, M. Anand, E.J. Karwacki, Presented at the 2nd Annual

National Meeting of the North American Membrane Society, Syracuse, NY,
(June 1-3, 1988).
46. R.B. McBride, D.L. McKinley, them. Enn. Proa. 61, 81 (1965).
47. R.B. McBride, R.T. Nelson and R.S. Hovey, U.S. Patent 3,336,730 (1967).
48. J.B. Hunter, U.S. Patent 2,773,561 (1956).
49. J.B. Hunter, Platinum Met, Rev. 4, 130 (1960).
50. R.W. Baker, J. Louie, P.H.Pfromm and J.G. Wijmans, U S . Patent 4,857,080,
(August 1989).
51. A.S. Darling, British Patent 1,292,025 (1972).
52. U.S. Patent 4,717,455.

53. U.S.Patent 4,722,771.
54. R.C. Furneaux et al., "The Formation of Controlled Porosity Membranes from
Anodically Oxidized Aluminimum," w e 337, NO. 6203, 147-9, (1989).
55. H.P. Hsieh, "Inorganic Membranes," in New Memb rane Materials and
Processes for m, K.K. Sirkar & D.R.Lloyd (Eds.). AIChE Symposium
Series, 84. p.1 (1988).
56. M.A. Anderson, M.S. Gieselmann, Q. Xu, J. Membr. Sci. 3,

243 (1988).
57. A. Soffer, J.E. Koresh and S.S. Saggy, U.S. Patent 4,685,940 (1987).
58. W.J.V. Osterhout, Phsioloev 21, 125-32 (1935).
59. G.M. Shean and K. Sollner, Anals of the New York Academy of Sciences.
m,759-76 (1966).
60. R. Bloch, A. Finkelstein, 0. Kedem and D. Vofsi, "Metal Ion Separation by
Dialysis Through Solvent Membranes," I&EC Process Desian and
DeveloDment. 6,231-37 (1967).
61. G.J. Moody and J.D.R. Thomas, "Polyvinyl Chloride Matrix Membrane Ion-
Selective Electrodes" in Ion-Selective Electrode Methodoloqy, A. K.
Covington (Ed.), CRC Press: Boca Raton, FL, 1 I 1 (1979).
62. T. Miyauchi, US. Patent 4,051,230, (1977).
63. R.W. Baker, M.E. Tuttle, D.J. Kelly and H.K. Lonsdale, "Coupled Transport
Membranes," J. Membr, Sci. 2, 213- 233 (1977).
64. T. Largman and S. Sifniades, "Recovery of Copper from Aqueous Solutions by

Means of Supported Liquid Membranes," Hvdrometallurnv 1, 153- I62 (1978).
65. N.N. Li and A.L. Shrier, "Liquid Membrane Water Treating" in Recent
DeveloDments in SeDaration Science, Vol. I , CRC Press: Cleveland, OH
(1972).
66. R.P. Cahn and N.N. Li, "Hydrocarbon Separation by Liquid Membrane
Processes" in Membrane Seoaration Processes, P. Meares (Ed.), Elsevier:
Amsterdam, 328 (1975).
67. E.L. Cussler, MulticomDonent Diffusion, Elsevier: Amsterdam (1976).
68. W. Volkel, W. Poppe, W. Halwachs and K. Schurgerl, " Extraction of Free

Phenols from Blood by a Liquid Membrane Enzyme Reactor," J. Membr. Sci.
-
1 I , 333 (1982).
69. R. Marr and A. Kopp, &m. Inn. w. , 399 (1980).

2. 197 (1977).
-
70. J. Stelmaszek, J. Membr. Sci..
71. T.P. Martin and G.A. Davies, "The Extraction of Copper From Dilute
Aqueous Solutions Using P Liquid Membrane Process," Hvd rometai 1. 2, 315
(1976/1977).
72. P.R. Danesi, ' , 231 (1984).

73. A.L. Bunge and R.D. Noble, "A Diffusion Model for Reversible Consumption in
Emulsion Liquid Membranes," J. Membr. Sci. 21. 55 (1984).
74. R.M. Izatt, R.M. Haws, J.D. Lamb, D.V. Dearden, P.R. Brown, D.W. McBride,
Jr. and J.J. Christensen, ,J Membr. Sci.. 24, 273 (i984).
75. H.I. Mahon, U.S. Patent 3,228,876 (1966).
76. H.I. Mahon, U.S.Patent 3,228,877 (1966).
77. D.T. Bray, U.S. Patent 3,417,870 (1968).
78. J. Westmoreland, U.S. 3,367.504 (1968).
79. S.S. Kremen, "Technology and Engineering of ROGA Spiral Wound Reverse
Osmosis Membranes," in Reverse Osmosis and Svnthetic Membranes, S.
Sourirajan (Ed.), Natinal Research Council of Canada, Ottawa (1977).
2. Pervaporation
R.W. Baker, Membrane Technology and Research, Inc.

Menlo Park CA
2.1 PROCESS OVERVIEW
Pervaporation is a membrane process used to separate mixtures of dissolved

solvents. The process is shown schematically in Figure 2-I. A liquid mixture
contacts one side of a membrane; the permeate is removed as a vapor from the
other side. Transport through the membrane is induced by the difference in
partial pressure between the liquid feed solution and the permeate vapor. This
partial-pressure difference can be maintained in several ways. In the laboratory a
vacuum pump is usually used to draw a vacuum on the permeate side of the
system. Industrially, the permeate vacuum is most economically generated by
cooling the permeate vapor, causing it to condense. The components of the feed
solution permeate the membrane at rates determined by their feed solution vapor
pressures, that is, their relative volatilities and their intrinsic permeabilities
through the membrane. Pervaporation has elements in common with air and steam
stripping, in that the more volatile contaminants are usually, although not
necessarily, preferentially concentrated in the permeate. However, during
pervaporation no air is entrained with the permeating organic, and the permeate
solution is many times more concentrated than the feed solution, so that its
subsequent treatment is straightforward.
The separation factor, ,9pervsp. achieved by a pervaporation process can be

defined in the conventional way as
c”i/c”j
9 PcnaP =c’i/c’j ’ (1)
where c’~ and c’~ are the concentrations of components i and j on the feed liquid
side and c”~ and c”~ are the concentrations of components i and j on the
permeate side of the membrane. Because the permeate is a vapor, c”~ and C’j can
be replaced by p’; and P”~, the vapor pressures of components i and j on the
permeate side of the membrane. The separation achieved can then be expressed
by the equation
P”i/P”j
BP-.-P =“‘i/c; (2)
151
Liquid (C’i/C’j)
\
Feed
Condenser
Vapor (P”i/P”j)
Total quantity passed through membrane

Flux =
Membrane area time
ratio of components in permeate vapor

Separation factor (&ervap) =
ratio of components in feed solution
P”i/P”j
BPcnaP =c’i/c’j
Figure 2-I. Schematic of pervaporation process.

Pervaporation 153
The most convenient mathematical method of describing pervaporation is to

divide the overall separation processes into two steps. The first is evaporation of
the feed liquid to form a (hypothetical) saturated vapor phase on the feed side of
the membrane. The second is permeation of this vapor through the membrane to
the low pressure permeate side of the membrane. Although no evaporation
actually takes place on the feed side of the membrane during pervaporation, this
approach is mathematically simple and is thermodynamically completely equivalent
to the physical process. The evaporation step from the feed liquid to the
saturated vapor phase produces a separation, ,9_,, which can be defined as the
ratio of the components’ concentrations in the feed vapor to their concentrations
in feed liquid
P’i/P’j
Bm.p = - 9 (3)
C’i/C’j
where P’~ and p; are the partial vapor pressures of the components i and j in
equilibrium with the feed solution.
The second permeation step of components i and j through the membrane is

directly related to conventional gas permeation. The separation achieved in this
step. pm.,, can be defined as the ratio of components in the permeate vapor to
the ratio of components in the feed vapor,
P”i/P”j
Bmorn= (4)
P’JP’j
From the definitions given in Equations (I-4). we can write the equality,
This equation shows that the separation achieved in pervaporation is equal to the
product of the separation achieved by evaporation of the liquid and the
separation achieved by permeation of the components through a membrane. To
achieve good separations both terms should be large. It follows that, in general,
pervaporation is most suited to the removal of volatile components from relatively
involatile components, because BevSp will then be large. However, if the
membrane is sufficiently selective and &,,,, is large, nonvolatile components can
be made to permeate the membrane preferentially.
It can be shown that Equation (5) can be rewritten as
[-I
P’i/P”i P’j
_-. (6)
P’j/P”j P’i
where a,,, is the membrane selectivity, that is the ratio of the individual
component permeabilities,
pi
%n.nl- - . (7)
‘j
Equation (6) is useful in defining the contributions to the separation

achieved in a pervaporation process. The first contribution is the simple
evaporation term Bavap. This term can be obtained from the vapor-liquid
equilibrium diagram and is the same factor that determines the efficiency of a
distillation process. The second term a,,,.,,, reflects the intrinsic permeation
selectivity of the membrane. The third term
P’i - P”t P’j
[- P’j - P”j I . -
p’i
reflects the particular operating conditions of the pervaporation process.
To obtain the maximum possible separation for a particular mixture the

membrane permeation selectivity should be much greater than I, and there should
be the maximum possible difference between the feed and permeate vapor partial
pressures.
2.1.1 Design Features
As described above, transport through pervaporation membranes is produced

by maintaining a vapor pressure gradient across the membrane. The vapor
pressure gradient used to produce a flow across a pervaporation membrane can be
generated in a number of ways. Figure 2-2 illustrates a number of possible
processes.
In the laboratory, the low vapor pressure required on the permeate side of
the membrane is most conveniently produced with a vacuum pump as shown in
Figure 2-2a. On a commercial scale, however, the vacuum pumps required would
be impossibly large. An attractive alternative to vacuum operation, illustrated in
Figure 2-Zb, is to cool the permeate vapor to condense the liquid. The feed
solution may also be heated. In this process, sometimes called thermo-
pervaporation, the driving force is the difference in vapor pressure between the
hot feed solution and the cold permeate liquid at the temperature of the
condenser. Because the cost of providing the cooling and heating required is
much less than the cost of a vacuum pump, as well as being operationally more
reliable, this type of system is preferred in commercial operations.
A third possibility, illustrated in Figure 2-2~. is to sweep the permeate side

of the membrane with a carrier gas (normally air). In the example shown, the
carrier gas is cooled to condense and recover the permeate vapor and the gas is
recirculated. This mode of operation has little to offer compared to temperature
gradient-driven pervaporation, since both require cooling water for the
condenser. However, if the permeate has no value and can be discarded without
condensation (for example, in the pervaporative dehydration of an organic solvent
with an extremely selective membrane), this is the preferred mode of operation.
In this case, the permeate would contain only water plus a trace of organic
solvent and could be discharged or incinerated at a low cost. No cooling water is
required.’ An alternative carrier gas system uses a condensable gas, such as
steam, as the carrier sweep fluid. This system is illustrated in Figure 2-2d.
Pervaporation 155
.) P.r”.por.bo” WI”, . h”ro+.h.” ,a.r”“.l. .“d ,,.rW rryc,.
Llquld
liquid
I P”rn.h I
“W soecmcl
,rrtlon hscno”
Figure 2-2. Schematics of possible pervaporation process configurations.

Low-grade steam is often available at very low cost and if the permeate were
immiscible in the condensed carrier water, it could be recovered by decantation.
The condensed water would contain some dissolved solvent and could be recycled
to the evaporator and then to the permeate side of the module. This mode of
operation is limited to water-immiscible permeates and to feed streams in which
contamination of the feed liquid by water vapor permeating from sweep gas is not
a problem. A possible example would be the removal of low concentrations of
toluene and similar organic solvents from aqueous feed streams.
The final two pervaporation processes illustrated in Figures 2-2e and 2-2f
are systems of particular interest for removing low concentrations of dissolved
organic compounds from water. The system shown in Figure 2-2e could be used
when the permeating solvent has a limited solubility in water. In this case, the
condensed permeate liquid will often separate into two phases: an organic phase,
which can be treated for reuse, and an aqueous phase saturated with organic
solvent, which can be recycled to the feed stream for reprocessing. The system
shown in Figure 2-2f shows two condensers in series. In this case, a second,
partial separation can be achieved, in addition to the pervaporation separation.
This would be useful, for example, in the separation of dilute alcohol-water feeds.
Thus, the alcohol-enriched permeate would be separated into a water-rich fraction
that might be recycled to the pervaporation unit and a second, highly enriched
alcohol fraction.2
2.1.2 Pervaporation Membranes

The selectivity of pervaporation membranes can vary substantially and has a
critical effect on the overall separation obtained. The range of results that can
be obtained for the same solutions and different membranes is illustrated in
Figure 2-3 for the separation of acetone from water. The figure shows the
concentration of acetone in the permeate as a function of the concentration in
the feed. The two membranes shown have dramatically different properties. The
pervaporation selectivity varies with concentration. At 20 wt% acetone in the
feed, silicone rubber has a pervaporation selectivity of approximately I6 favoring
acetone; the crosslinked polyvinyl alcohol (PVA) membrane has a pervaporation
selectivity of only 0.3 and selectively removes water.
The acetone-selective, silicone rubber membrane is best used to treat dilute

acetone feed streams, concentrating most of the acetone in a small volume of
permeate. The water-selective, polyvinyl alcohol membrane is best used to treat
concentrated acetone feed streams containing only a few percent water. Most of
the water is then removed and concentrated in the permeate. Both membranes
are more selective than distillation, which relies on the vapor-liquid equilibrium to
achieve a separation.
2. I .3 Pervaporation Modules
Pervaporation applications often involve hot, organic-solvent-containing feed
solutions. These solutions can degrade the seals and plastic components of
membrane modules. As a result, the first-generation commercial pervaporation
modules installed have been made from stainless steel and are of the plate-and-
frame design. Figure 2-4 shows a single cell of a multicell pervaporation stack.
These cells can be made in various sizes. The largest currently produced
incorporates membranes with a membrane area of I.5 m2.
PervaPoration 157
80
1
I
I
60
Permeate I
acetone
:
concentration
I
wtw t
I
I
I
I
-I.
I
I
GFT-PVA membrane I
I
A
20 40 60 80 100
Feed acetone concentration (wt%)
Figure 2-3. The pervaporation separation of acetone-water mixtures achieved

with a PVA, water-selective membrane, and an acetone-selective,
silicone rubber membrane (source: GFf literature).
Feed 1
Gasket -
Membrane-
Membrane
Gasket -
t
Product
Figure 2-4. A pervaporation plate-and-frame module.

Pervaporation 159
In recent years attempts have been made to switch to lower cost module
designs. Membrane Technology and Research (MTR) and Air Products have both
used spiral-wound modules for pervaporation. British Petroleum is also said to be
developing a tubular module system.
2.1.4 Historical Trends
The origins of pervaporation can be traced to the 19th century, but the
process was first studied in a systematic fashion by Binning, Lee and co-workers
at American Oil in the early 1950s. 3 Binning and Lee were interested in applying
the process to the separation of organic mixtures. Although this work was
pursued for a number of years and several patents were obtained, the process was
not commercialized, because membrane fabrication technology available then did
not allow the high-performance membranes and modules required for a
commercially competitive process to be made. By the 1980s. however,
advancements in membrane technology made it possible to prepare economically
viable pervaporation systems. The surge of interest in the process starting in
1980 is illustrated in Figure 2-5, which tabulates the increase in the quantity of
pervaporation literature over the last 40 years.
Pervaporation has now been commercialized for two applications. The first
and by far the most important is the separation of water from concentrated
alcohol solutions. GFT of Hamburg, West Germany, the leader in this field,
installed their first major plant in 1982. ’ Currently, more than 100 plants have
been installed by GFf for this application.5 The second application is the
separation of small amounts of organic solvents from contaminated waters, for
which the technology has been developed by MTR.6 Both of the current
commercial processes concentrate on the separation of organics from water. This
separation is relatively easy, because organic solvents and water, due to their
difference in polarity, exhibit distinct membrane permeation properties. The
separation is also amenable to membrane pervaporation because the feed solutions
are relatively non-aggressive and do not chemically degrade the membrane.
No commercial systems have yet been developed for the separation of the
more industriaily significant organic/organic mixtures. However, current
technology now makes development of pervaporation for these applications possible
and the process is being actively researched in a number of laboratories. The
first pilot-plant results for an organic-organic application, the separation of
methanol from methyltertbutyl ether/isobutene mixtures, was recently reported by
Air Products. 7J Texaco is also working on organic-organic separations.g This is
a particularly favorable application and currently available cellulose acetate
membranes give good separation. It can only be a matter of time, however,
before much more commercially significant organic-organic separations are
attempted using pervaporation.
2.2 CURRENT APPLICATIONS. ENERGY BASICS AND ECONOMICS
There are three current applications of pervaporation; dehydration of

solvents, water purification, and organic-organic separations as an alternative to
distillation. Currently dehydration of solvents, in particular ethanol and
isopropanol, is the only process installed on a large scale. However, as the
technology develops, the other applications are expected to grow. Separation of
organic mixtures, in particular, could become a major application. Each of these
applications is treated separately below. The energy considerations and system
economics vary from application to application and are discussed in the individual
application sections.
100
90
80
70
Pervaporation 60
citations
50
per year
40
30 Binning, Lee, et al.
20
10
0
1960 1965 1970 1975 1980 1985
Time (yr)
Figure 2-S. Pervaporation citations in Chemical Abstracts for

the years 1959- 1988.
Pervaporation 161
2.2.1 Dehydration of Solvents
More than 100 plants have been installed for the dehydration of ethanol by
pervaporation. This is a particularly favorable application because ethanol forms
an azeotrope with water at 95% and there is a need for a 99.5% pure product.
A comparison of the separation of ethanol and water obtained by a

pervaporation membrane (GFT polyvinyl alcohol composite membrane) and the
separation obtained by distillation is shown in Figure 2-6. Because of the
ethanol-water azeotrope at 95% ethanol, the concentration of ethanol from
fermentation feeds to high degrees of purity requires rectification with a benzene
entrainer, some sort of molecular-sieve drying process, or a liquid-liquid
extraction process. All of these processes are expensive. The existence of
extremely water-selective pervaporation membranes, however, means that
pervaporation systems can produce almost pure ethanol (~99.9% ethanol from a 90%
ethanol feed). The permeate stream contains approximately 50% ethanol and can
be recirculated back to the distillation column.
An energy and cost comparison of a very small ethanol/water separation

plant from GFT is shown in Table 2-l. Pervaporation is less capital and energy
intensive than distillation or adsorption processes. These savings produce a cost
reduction of 3-Se/gallon of ethanol produced. However, because of the modular
nature of the process, the costs of pervaporation are not as sensitive to
economies of scale as are the costs of distillation and adsorption processes.
Figure 2-7 shows GFT cost data for ethanol dehydration plants varying from 100
to 10,000 L/h. Costs of plants above 100 L/h capacity scale almost linearly with
capacity. Distillation costs, on the other hand, scale at a rate proportional to
0.6-0.7 times the power consumption. Thus, distillation remains the most
economical process for large plants. The cross-over point at which distillation
becomes preferable to pervaporation from an energy and economic point of view
currently appears to be about 10 x lo8 gal/year (5,000 L/h) processing capacity.
Table 2-I. Separation Options for Small Scale EtOH/Water

(Basis: 1,000 L/day, 99.5 wt% EtOH)’
Pervaooration Distillation Adsorution
System cost f 75,000 $140,000 S 90,000

Pumps 3 kW 2 kW 2 kW
Steam 45 kg/h @ 1.8 bar 70 kg/h @ 7.3 bar 90 kg/h @ 7.3 bar,
22o’c
Entrainer __ 3 L/day __
60
60
Permeate ethanol
concentration (wt%)
40
: / GFT-PVA-
I_, ,
20 +
I
OV
0 20 40 60 80 100
Feed ethanol concentration (win/.)
Figure 2-6. Separation of ethanol/water mixtures by distillation and GFT’s

polyvinyl alcohol pervaporation membrane.
Pervaporation 163
1,000,000 -
System cost
(9 100,000 -
10,000 -
1000 1 I I I
10 100 1000 10,000
Product output (Uh)
Figure 2-i’. Cost of ethanol dehydration plants as a function of plant

capacity. lo Costs in S/liter/hour capacity are shown.
A flow scheme for an integrated distillation-pervaporation plant operating on

5% ethanol feed from a fermentation mash is shown in Figure 2-8. The
distillation column produces an ethanol stream containing 8590% ethanol, which is
fed to the pervaporation system. To maximize the vapor pressure difference
across the membrane, the pervaporation module usually operates at a temperature
of 105-13o’C with a corresponding feed stream vapor pressure of 2-6 atmospheres.
Despite these harsh conditions, the membrane lifetime is good and qualified
guarantees for up to four years are given.
Figure 2-8 shows a single-stage pervaporation unit. In practice, at least

three pervaporation stages are usually used in series, with additional heat being
supplied to the ethanol feed between each stage. This compensates for
pervaporative cooling of the feed and maintains the feed at 8o’C. The heat
required is obtained by thermally integrating the pervaporation system with the
condenser of the final distillation column. Most of the energy used in the
process is therefore low grade heat. Generally, about 0.5 kg of steam is required
for each kilogram of ethanol produced. The energy consumption of the
pervaporation process is, therefore, about 2,000 Btu/gal of product, less than 20%
of the energy used in azeotropic distillation, which typically is in the range
I I ,OOO-12,000 Btu/gal. Moreover, pervaporation uses very low-grade steam, which
is available in most industrial plants at very little cost.
Although most of the installed solvent dehydration systems have been for
ethanol dehydration, dehydration of other solvents including isopropanol. glycols,
acetone and methylene chloride, has been considered. Schematics of pervaporation
processes for these types of separation are shown in Figure 2-9. These processes
are all in the pilot-plant or demonstration-plant stage.
2.2.2 Water Purification
A number of applications exist for pervaporation in the removal or recovery

of organic solvents from water. If the aqueous stream is very dilute, pollution
control is the principal economic driving force. However, if the stream contains
more than l-2% solvent, recovery for eventual reuse can significantly enhance the
process economics.
A number of membranes have been used for the separation of solvents from
water and are discussed in the literature.6*10-” Usually the membranes are
made from rubbery polymers such as silicone rubber, polybutadiene, natural
rubber, polyether copolymers and the like. Some typical results for a commercial
organic-solvent-selective membrane are shown in Figure 2-10. This membrane has
high selectivities for hydrophobic volatile solvents such as benzene, 1,1,2-
trichlorethylene and chloroform. It has high selectivities for solvents of
intermediate hydrophobic character such as ethyl acetate, methyl ethyl ketone,
and acetone, but only modest selectivities with relatively hydrophilic solvents such
as ethanol, methanol and acetic acid. This pattern of behavior is typical of most
membranes, although there are substantial differences between different membrane
materials. Some comparative data for the separation of toluene and
trichloroethylene from dilute aqueous solutions are shown in Figure 2-l I. In
these particular experiments, the ethene-propene terpolymer membrane was
significantly more selective than, for example, silicone rubber.
Pervaporation 165
Distillation
columns
I I 1
Mash
t
1
Pervaporation
unit
40230% ethanol
Figure 2-8. Integrated distillation/pervaporation plant for ethanol recovery from

fermenters.
Dehydration of elhylcnedlchloride (EDC)
Figure 2-9. A schematic of various types of pervaporation solvent dehydration

systems.
Pervaporation 167
60 I I I I
50
Trlchloroethane
Permeate Elhyl
concenlratlon acetate
(WI%)
/
10
0
0 0.2 0.4 0.6 0.6 1.0
Feed concentration (wt%)
Figure ?-IO. Separation of organic solvents from water by pervaporation using a

commercial membrane (MTR CODE-100. Membrane Technology and
Research, Inc.).
60000
m Toluene
_ 0 Trichloroethylene
60000
Pervaporation
separation 40000
factor
20000
Figure 2-I 1. Comparative selectivities for toluene and trichloroethylene from water
and various rubber membranes.
Pervaporation 169
There is very little data on membranes with high selectivities for polar
solvents such as ethanol and methanol. The most promising results to date have
been obtained with silicone rubber membranes containing dispersed zeolite
particles.r5 It appears that ethanol preferentially permeates through the pores of
the zeolite particles and membranes have been produced in the laboratory with
ethanokwater selectivities of 40 or more. Membranes with these characteristics
could find a substantial application in fermentation and solvent recovery
applications if they can be made on a large scale.
2.2.3 Pollution Control
The costs of pervaporation treatment for pollution control vary widely,

depending on the solvent and the fractional removal of solvent required. A
sample cost calculation is given below for the application of a pervaporation
system to a polluted ground-water. The process is illustrated in Figure 2-12. In
this example, a 20,000 gal/day groundwater stream is assumed to be polluted with
I.000 ppm (0.1%) of I, I ,2-trichioroethane. The pervaporation separation factor
assumed for this solvent is approximately 200. with a transmembrane flux of I.0
L/m?h. Both of these figures are easily achievable with current membranes.
Using these values and a computer model, it can be shown that 99% removal of
the solvent can be achieved with 85 m2 of membrane, producing a permeate
stream containing 4. I% I, I ,2_trichloroethane. ’ This solvent is relatively insoluble
in water. Therefore, on condensation, the permeate spontaneously separates into
a pure solvent stream which can be decanted for reuse or incineration, and an
aqueous stream containing 0.4% solvent, which can be returned to the incoming
feed for further treatment. The economics of this process are summarized in
Table 2-2. based on data from MTR. These costs appear to be very attractive.
The capital cost of the plant is equivalent to a cost of S7.75/gal-day capacity.
The operating cost is %18.8/l ,000 gal feed. These costs compare favorably with
other waste treatment methods.
Table 2-2. Capital and Operating Costs of a 20,000-gal/day PerVap System to

Remove I, I ,2-Trichloroethane from Wastewater
Capital Cost S155.000
Operating Cost
Depreciation at 15% S 22,800
Labor (10% of capital) 15,500
Module Replacement (3-year life) 15,000
Energy 60.00Q
f I I3,300/yr
Operating Cost $18.8/ I.000 gal feed

Residue
0.001%
1,1,2-trichloroethane
Feed
0.1% Membrane unit
1 ,l ,Ptrichloroet
20,000
gallday
1 ,1,2-trichloroethane
Phase 1,1,2-trichloroethane
separator
E!sd Perme* Residue
Flow (gal/day) 20,000 500 19,980

Concentration (%) 0.1 4.1 0.001
Membrane pervaporation enrichment 200

Membrane flux 1.0 (L/mz.h)
Membrane area 85 (m2)
Figure 2- 12. Flow diagram and design parameters of an MTR PerVap system to
treat a 0.1% I, 1,2- trichloroethane stream.
Pervaporation 171
2.2.4 Solvent Recovery
If the concentration of soivent in the feedstream is sufficiently high,

solvent recovery for its own sake becomes economically attractive. An application
of this type is shown in Figure 2-13 .e This stream contains 2% ethyl acetate and
the object is to recover 90% of the solvent. As shown, the pervaporation system
produces a permeate containing 45.7% solvent, which spontaneously separates into
an organic phase of 96.7% solvent suitable for reuse, and an aqueous phase. The
aqueous phase from the permeate is recycled to the feed stream.
The capital cost of a 20.000 gallon/day plant is S140.000 and the operating
costs, as shown in Table 2-3, are approximately S107,000, or S0.98/gal of solvent
recovered. The current cost of technical grade ethyl acetate is approximately
U/gal, so the system has a pay-back time of only a few months.
Table 2-3. Capital and Operating Costs of a 20,000 gal/day Pervaporation

System to Recover Ethyl Acetate
Capital Cost s140,000
Operating Costs
Depreciation at 15% 21,000
Labor (10% of capital) 14,000
Module Replacement (3 year life) 12,000
Energy 60.00Q
S107,000/yr
Operating Cost S17.8/1.000 gal feed

Operating Cost S 0.98/gal solvent recovered
Pervaporation would be more widely applied in solvent recovery operations if

membranes more selective for hydrophilic polar solvents were available. Existing
membrane materials, such as silicone rubber, have pervaporation selectivities,
B , for this type of solvent typically in the range S-10. If materials having
sik:?vities in the range 40-50 could be made for solvents such as acetic acid,
formic acid, ethanol or methanol, pervaporation could easily compete with
distillation or solvent extraction. Membranes with this type of selectivity have
been reported on a laboratory-scale, but have yet to be tested industrially.
Calculations of the type given above suggest that pervaporation can be

economically applied to a wide range of aqueous industrial streams. The process
is still very new and its ultimate competitive position compared to more
conventional techniques is uncertain. Figure 2-14 shows the solvent concentration
range in which pervaporation is believed most competitive. When the feedstream
contains less than 100 ppm solvent, carbon adsorption beds become small and this
is probably the preferred technique. Similarly, when the stream contains more
than 5-10% solvent, distillation, steam stripping or incineration will probably be
less expensive than pervaporation. However, in the intermediate range of
between 100 ppm and 5% solvent, pervaporation will find a number of important
applications.
Residue
0.2%
Ethyl acetate
Feed Membrane unit

2%
(7Om2)
Ethyl acetate
20,000
gal/day
I Vacuum I
Condenser
Permeate
45.7%
separator
Es!22 Permeate Residue
Flow (gal/day) 20,000 800 19,600

Concentration (%) 2.0 45.7 0.2
Membrane pervaporation enrichment 100

Membrane flux, pure water I.0 (kg/m?h)
.2% ethyl acetate 2.0 (kg/m”,h)
Membrane area 70 (m”)
Figure 2-13. Flow diagram and design parameters of an hlTR PerVap system to
treat a 2?4 ethyl acetate stream.
Pervaporation 173
25
20
15
Water
treatment cost
(1,000gal teed)
10
0.001 0.01 0.1 1 10 100
Feed stream solvent concentration (%)
Figure 2-14. Costs of pervaporation water purification systems compared to

those of competitive technologies.
2.2.5 Organic-Organic Separations
The final application of pervaporation membranes is the separation of organic

solvent mixtures. Currently this is the least developed application of
pervaporation. However, if problems associated with membrane and module
stability under the relatively harsh conditions required for these separations can
be solved, this could prove to be a major membrane application around the year
2000.
In 1976 distillation energy consumption was estimated to be two quads (one

quad equals 10” Btu). which equals 3% of the entire national energy usage.16 In
principle, pervaporation could be used as a substitute or supplement to distillation
processes with substantial cost and energy savings. This is definitely true for the
separation of axeotropes and close boiling liquids, which are difficult to separate
by conventional distillation techniques, A list of major distillation separations
that consume more than 2,000 Btu/lb of product to perform is given in Table 2-4.
Replacement of even some of these separations by a low-energy pervaporation
process could have a major impact on national energy use.
Pervaporation 175
Table 2-4. Major Distillation Separation Processes Using More Than 2,000 Btu/lb
of Product
U.S. Specific
Distillation Diitilhtion Key
Energy Energy Components
Process Consumption Consumption Light/
Description (QuaWyr) (Btu/lb product) Heavy
Ethylbenzene/ 0.00355 1 I.582 Ethylbenzene/

gasoline p-xyiene
fractionation
Adipic Acid: 0.00336 2.210 __

air oxidation of
cyclohexane
Methyl ethyl ketone 0.00279 5.470 Set-butanol/water

from n-butylene
Glycerine 0.00492 Il.618 Dilsobutyl-ketone/

using peracetic acid heavy impurities,
water/glycerine
Ethanol: 0.0105 8,784 Ethanol azeotrope/

direct hydration water
of ethylene
Vinyl acetate monomer 0.00453 3,061 Vinyl acetate-water

from ethylene and azeotrope/acetic
acetic acid acid water
azeotrope
Phenol 0.00437 2.005 Cumene/phenol

from cumene
o-Xylene 0.00603 5,689 m-xylene/o-xylene
Acetic acid 0.00574 2,362 Water/acetic acid

acetaldehyde
oxidation
It is unlikely that a pervaporation process would perform the entire

separation required for any item on the list. In practice, a hybrid distillation-
pervaporation process would be used. Such a process is shown in Figure 2-15. In
the process shown in Figure 2-15, a SO/SO azeotropic mixture is fractionated into
two approximately equal streams, each containing 85% of one of the components.
These streams are then sent to distillation columns to produce pure components
and a recycled azeotrope stream. A key feature of this process is that only a
relatively modest separation of the azeotropic mixture by the membrane is
required to make the separation feasible. Of course, membranes with very high
selectivities are more efficient than those with lower selectivities, but in this
application a membrane with a selectivity as low as 5-10 might be economically
viable, depending on the solvent mixture and the alternative treatments available.
A number of workers have studied membranes for the separation of organic

mixtures and membranes have been developed which appear to be remarkably
selective, even for relatively similar compounds. Some typical results from Aptel
are shown in Figure 2-16.” The principal problem hindering the development of
commercial systems for this type of application is the lack of membranes and
modules able to withstand long-term exposure to organic compounds at the
elevated temperatures required for pervaporation. ~- -- .. __
Membrane and module stability problems do not appear to be insurmountable,

however, as shown by the successful demonstration of a pervaporation process for
the separation of methanol from an isobutene/methyltertbutyl ether (MTBE)
mixture. This mixture occurs during the production of MTBE; the product from
the reactor is an alcohol/ether/hydrocarbon raffinate stream in which the
alcohol/ether and the alcohol/hydrocarbon both form azeotropes. The product
stream is treated by pervaporation to yield a pure alcohol permeate, which can be
recycled to the etherification process. This process has been developed by
Separex, a division of Air Products (currently owned by Hoechst-Celanese Corp.),
using cellulose acetate membranes. The separation factor for methanol from
MTBE is 1,000 or more with cellulose acetate membranes. Two alternative ways
of integrating the pervaporation process into a complete reaction scheme are
illustrated in Figure 2- I7.‘*s These membranes are reported to work well for
feed methanol concentrations up to 6%. Above this concentration, the membrane
becomes plasticized and selectivity is lost.
2.3 INDUSTRIAL SUPPLIERS
The current industrial suppliers are listed in Table 2-5. GFf is the industry
leader with more than 90% of the current market, primarily alcohol dehydration
plants in Europe. However, this is a rapidly evolving technology and the industry
structure may be very different in a few years time. The table only shows
companies that have announced the development of pilot-scale or full industrial
systems. The list is probably incomplete, because a number of major oil
companies are believed to be evaluating pervaporation systems, particularly for
organic-organic separations, although no announcements have appeared in the
patent or technical literature. The current pervaporation market is probably less
than %8 million/year but is growing.
Pervaporation 177
Feed
50% A. 50% B
Distillation
COlUW
Pure B Pure A
Figure 2- 15. Schematic of a combined pervaporation-rectification process for

the separation of a SO/SO azeotropic mixture.
60
Benzene
(cyclohexene)
In product (wt%)
40
0
0 20 40 60 60 100
Benzene (cyclohexene) in feed (wt%)
Figure 2- 16. Fraction of benzene or cyclohexane in product VS. feed mixture

composition for pervaporation at reflux temperature of two
binaries (benzene/cyclohexane and cyclohexene/cyclohexane). A
2O-pm, crosslinked alloy membrane was used.16
Pervaporation 179
MTBE Reaction Chemistry
CH3 CH3
I I
CHBOH + C = CH2 Z CH3OH - C - CH3
I I
CH3 CH3
Methanol lsobutene Methyl Tea-l-butyl ether
Pervaporatlon Process before Debutanizer
Pervaporation on Debutanizer Sidedraw
Figure 2- 17. Methods of integrating pervaporation membranes in the recovery

of methanol from the MTBE production process.7s8
Table 2-5. Industrial Suppliers
Company Application Membtana/Modula Comment3
GFT solvent Crosslinked PVA composite Founded by G.F. TGsel, now

dehydration membranes. Plate-and- a division of Ceraver. GFf
frame modules. the major pcrvaporation
company. Concentrating on
alcohol dehydration, they
have over 100 plants
installed. Mitsui is their
Japanese distributor.
Lurgi Solvent GFT PVA membranes. Use GFf membranes, but their
dehydration Plate-and-frame own plate-and-frame modules
modules. based on existing heat
exchanger technology. Have
installed a number of alcohol
dehydration plants.
Membrane Solvent Composite membranes. Sells plants in the S-IO

Technology recovery Spiral-wound modules. thousand gallon/day range for
Sr Research water pollution control.
British Solvent Composite membranes based on Some small plants operating.

Petroleum dehydration ion-exchange polymers. Tubular Mostly water/isopropsnol.
(Kalsep) and plate-and-frame modules.
Air Products Organic-organic Cellulose acetate membranes. Small pilot system

(Separex) separation. Spiral-wound modules. operating at field site
mrthanol/MTBE (2 gal/h) on an MTBE
vent stream.
Texaco Solvent Their own composite membranes

dehydration in GFT plate-and-frame modules.
especially
ethylene glycol,
isopropyl alcohol,
organic/organic
separation
2.4 SOURCES OF INNOVATION
Pervaporation is still one of the less developed areas of membrane

technology and the number of research groups of any size is small. Table 2-6
lists the major institutions and key personnel, where known.
Pervaporation 181
Table 2-6. Major Membrane Technology Institutions and Key Personnel
Company/Institution Key Personnel
Industrial Comoanieg
Air Products and Chemicals, M.S. Chen

Allentown, Pennsylvania
British Petroleum, Cleveland, Ohio J. Davis
Exxon, Baton Rouge, Louisiana
GFT, Boonton, New Jersey H.E.A. Bruschke. H.L. Fleming
Koninklijke/Shell Laboratory,
Amsterdam, The Netherlands
Lurgi GmbH, Frankfurt, West Germany V. Sander
Membrane Technology & Research, R.W. Baker, J.G. Wijmans

Menlo Park, California
Academic and Nonorofit InstitutionS
University of Twente, M.H.V. Mulder, C.A. Smolders

Enschede, The Netherlands
University of Maine, Orono, Maine E. Thompson
University of Syracuse, S.A. Stern

Syracuse, New York
State University of New York, I. Cabasso

Syracuse, New York
University of Cincinnati, S-T. Hwang

Cincinnati, Ohio
GKSS, Geesthacht K.W. Boddeker, K.-V. Peinemann

West Germany
Fraunhofer IGB H. Strathmann

Stuttgart, West Germany
Ecole Nationale. J. Neel

Nancy, France
TNO, Zeist, The Netherlands F.F. Vercanteren

2.5 FUTURE DIRECTIONS
2.51 Solvent Dehydration
Currently solvent dehydration is the major commercial application and

research area of interest for pervaporation. For example, at the Third
International Pervaporation Conference in Nancy, France, in 1988, more than two
thirds of the papers presented dealt with solvent dehydration applications. Of
these, more than half dealt with the dehydration of ethanol. Nonetheless,
although of considerable current interest, this application is to a large extent a
solved problem.
The current industry leader is GFf, who use a polyvinyl alcohol-based

membrane. This membrane is extremely selective and future improvements in
module selectivity are unlikely to change the overall dehydration process
economics dramatically. However, a number of other membranes have been
reported, particularly those based on ion-exchange polymers. Figure 2-18 shows a
permeate composition vs. feed composition performance curve for one of these
membranes.”
Comparison of this figure with data for the GFT polyvinyl alcohol membrane
illustrated in Figure 2-6 shows that the membrane selectivities are comparable or
a little less. With newer versions of the membranes, these results could probably
be improved. The principal advantage these membranes may offer is higher
fluxes, which could reduce process costs by reducing the amount of membrane
area required and hence the module costs. However, current plate-and-frame
modules already have heat and mass transfer problems, even at existing fluxes. It
may be difficult to exploit new higher-flux membranes without making
corresponding improvements in module design and performance.
Development of better modules is the principal area where an investment of

research resources would lead to significant cost savings. Current commercial
plate-and-frame modules are made from stainless steel. These modules are
reliable, resistant to solvent attack and can withstand high pressure and
temperature operations. However, they are extremely expensive, almost ten times
the cost of spiral-wound modules, for example. Development of spiral-wound or
hollow-fiber modules able to withstand the operating conditions encountered in
solvent dehydration would lower costs significantly. allowing much wider use of
the process.
Pervaporation 183
100
,’
,’
__._....-
..--
I I
Vapor-liquid
equilibria
___...- .I..,
0
0
I
0
0
0
0
0
0
I
_.;J
Product
0
ethanol 0
concentration -0 **’
wt%o)
20 40 60 60 100
Feed ethanol concentration (WI%)
Figure 2-18. Separation of ethanol-water feed solutions with Nafion@ ion-

exchange, hollow-fiber membranes with various counter-ions.”
2.52 Water Purification
Water purification by pervaporation has yet to be demonstrated on an

industrial scale. Application and process development is, therefore, a high-
priority research area. Major applications exist in pollution control, solvent
recovery and waste reduction. Aroma and flavor recovery from wastewaters in
the food and flavor and fragrance industries is another promising opportunity.
Current water-purification membranes are sufficiently permselective for

hydrophobic solvents, but are inadequate for handling hydrophilic solvents such as
ethanol, methanol, acetic acid and formic acid. These solvents, being freely
water miscible, are found in many industrial process and waste streams. Many
industrial water streams that would be candidates for pervaporation contain
significant quantities of hydrophilic solvents. Development of more selective
membranes for these solvents is urgently required. The inclusion of zeolites or
other molecular sieving materials into polymer membranes, discussed in Section
2.2.2 may also be worth pursuing further.
2.5.3 Organic-Organic Separations
The separation of organic solvent mixtures represents by far the largest

opportunity for major energy and cost savings. It is estimated that 40% of the
total energy consumed by the chemical processing industry is used in distillation
operations. Not only does distillation use large amounts of energy, but it is not
an effective separation method in many cases. In particular, distillation is not
well-suited to separating mixtures of components with similar boiling points or
mixtures that form azeotropes. Currently, when standard distillation is unsuitable,
extractive distillation, liquid/liquid separation or some other technique must be
used. Azeotropic mixtures are an obvious target application for a pervaporation
process. However, if suitable membranes and modules can be developed, more
widespread applications to the entire chemical industry could be considered. It is
unlikely that pervaporation would be used to perform complete separations;
pervaporation systems are likely to find their best applications in hybrid processes
where an optimized distillation/pervaporation operation is performed. The
pervaporation units would perform a first, crude, low-cost separation, leaving the
distillation columns to perform a polishing operation.
The lack of membranes and modules able to withstand aggressive solvent

mixtures under the operating conditions of pervaporation is the key problem
hindering the application of pervaporation to organic mixtures. Very little work
on membranes able to separate organic mixtures has been reported in the
literature. This is a good research opportunity. If membranes were available,
they could probably be used with plate-and-frame modules, but these modules are
so expensive they will only be used commercially in few, very favorable
situations. Widespread application of organic-selective pervaporation membranes
requires the development of solvent-resistant, hollow-fiber or spiral-wound
modules.
In addition to module and membrane development, research must be

performed on overall system design. Pervaporation units will be used in
combination with other processes. Effective ways of integrating these processes
are required. Heat integration is a particularly important and undeveloped area.
Pervaporation 185
Support of research in the broad topic areas listed above is likely to

produce an across-the-board improvement in organic-organic pervaporation
systems. However, to encourage industrial acceptance of the process, focused
research aimed at specific, particularly favorable applications of pervaporation
should be encouraged. If pervaporation systems. could be developed for one or
two applications, even if they are small, niche applications, it would encourage
further development of the process into broader, much more economically
significant areas.
2.6 DOE RESEARCH OPPORTUNITIES
Based on the discussion above, a list of recommended research topics has

been developed. The topics are listed in Table 2-7 and ranked by order of
priority within application.
Table 2-7. Pervaporation Priority Research Topics
Research Topic Prospect For Importance Comments

Realization
Membranes for Very Good 10 If sufficiently selective membranes

organic-organic could be made, pervaporation could
separations replace distillation in many
separations
Better solvent- Excellent IO Current modules cannot be used

resistant modules with organic solvents and are also
very expensive
Better membranes for Very Good 7 More solvent-selective membranes

the removal of organic are required, especially for
solvents from water hydrophilic solvents
Plant designs and Good 7 Pervaporation will probably be

studies used in hybrid systems for
organic-organic separations.
System design studies are needed
to guide research
Dehydration membranes Good 7 Would be of use in breaking many

for acidic, basic. and common aqueous-organic
concentrated aqueous azeotropes
solvent streams
2.6.1 Priority Ranking
2.6.1.1 Solvent dehydration
Generally. this is a relatively developed area. Several companies are active

and commercializing plants. Better membranes would help to reduce costs, but
the principal problem is the high cost of plate-and-frame modules.
. Development of new module design would reduce costs.
2.6. I .2 Water purification
This process has been demonstrated at the pilot level, but not in the field.
. Current membranes are inadequate for ethanol, methanol, and other

hydrophilic solvents. Several areas of current membrane research look
promising. This could have a big impact in the fermentation industry.
. Current membranes are inadequate for high-boiling solvents, phenols,

glycols, etc. Improved membranes are also required here.
. Results from demonstration plants are required to encourage industry

acceptance.
2.6.1.3 Organic-organic separations
This area is now only being developed for niche applications. The big
potential application is azeotropes and closely-boiling solvent mixtures.
Current membranes are inadequate. Some research on improved

membrane material is underway, but few materials have been
formulated into high-performance membranes. The theory behind the
materials selection process is weak.
It is pointless to develop better materials and membranes unless they

can be formulated into high-surface-area modules. Research is required
here.
Development of a few new niche opportunities would be useful to

encourage the early acceptance of the program.
System design and applications development of these membranes is also

unexplored. Research is required here.
Pervaporation 187
REFERENCES
1. R. DeVellis et al., U. S. Patent 4,846,977 (July, 1989).
2. West German Patent P 3610011.0 (1986).
3. R.C. Binning, R.J. Lee, J.F. Jennings and E.C. Martin, “Separation of Liquid
Mixtures by Permeation,” Ind. & m Chem. 53. 45 (1961).
4. A.H. Ballweg, H.E.A. Brtischke. W.H. Schneider and G.F. Tiisel,

“Pervaporation Membranes,” in Proceedinas of the Fifth International
Alcohol Fuels Svmoosium, Auckland, New Zealand, 13-18 May 1982, John
McIndoe, Dunedin, New Zealand (1982).
5. H.E.A. Briischke, “State of the Art of Pervaporation,” in Proceedinas of Third

International Conference on Pervaooration in the Chemical Industry, Nancy,
France, R. Bakish (Ed.) (1988).
6. I. Blume, J.G. Wijmans and R.W. Baker, “The Separation of Dissolved

Organics from Water by Pervaporation,” ,I. Memb. Sci,, in press.
7. M.S.K. Chen, G.S. Markicwicz and K.G. Venugopal, “Development of

Membrane Pervaporation Trim Process for Methanol Recovery from
CH,OH/MTBE/C, mixtures,” AIChE Spring Meeting, Houston (1989).
8. M.S. Chen, R.M. Eng, J.L. Glazer and C.G. Wensley, “Pervaporation Processes
for Separating Alcohols from Ethers,” U.S. Patent 4,774,365 (September,
1988).
9. V.M. Shah, C.R. Bartels, M. Pasternak and J. Reale, “Opportunities for

Membranes in the Production of Octane Improvers,” paper presented at
AIChE Spring National Meeting, Houston, TX, April 2-6 (1989).
10. H.L. Fleming, “Membrane Pervaporation Separation of Organic/Aqueous

Mixtures,” International Conference on Fuel Alcohols and Chemicals,
Guadalajara. Mexico, (1989).
11. Y.M. Lee, D. Bourgeois and G.Belfort, “Selection of Polymer Materials for
Pervaporation,” in Proceedinas of Seco d International Conference on
Pervaooration, San Antonio, TX, R. Bakish’(Ed.) (1987).
12. H.H. Nijhuis, M.H.V. Mulder and C.A. Smolders, “Selection of Elastomeric
Membranes for the Removal of Volatile Organic Components from Water,” in
Proceedinas of the Third International Conference on Pervaooration in the
Chemical Industrv, Nancy, France, R. Bakish (Ed.) (1988).
13. C.-M. Bell, F.-J. Gerner, and H. Strathmann, “Selection of Polymers for
Pervaporation Membranes,” ,I. Memb. Sci. 36, 3 I5 (1988).
14. G. Bengston and K.W. B(Mdeker, “Pervaporation of Low Volatiles from

Water,” in Proceedinns of the Third In- Confwe on
Pervaooration in the Chemical Indum. Nancy, France, R. Bakish (Ed.)
(1988).
15. H.J.C. Hennepe, D. Bargemann, M.H.V. Mulder and C.A. Smolders, “Zeolite
Filled Silicone Rubber Membranes, Part I”, J. * 39 (1987).
16. T.W. Mix. J.S. Dweak, M.Weinberg and R.C. Armstrong, “Energy Conservation
in Distillation,” a report to the U.S. Department of Energy (DGE/CD/40259),
(July. 1987).
17. P. Aptel, N. Challard, J. Cuny and J. Neel, “Application of the Pervaporation

Process to Separate Axeotropic Mixtures,” J. Sci. 4, 271 (1978).
18. I. Cabasso, Z.Z. Lin, “The permselectivity selectivity of ion-exchange

membranes for non-electrolyte liquid mixtures,” I. Memb. Sci. 24, 101 (1988).
3. Gas Separation
by W. J. Koros, University of Texas, Austin, Texas
3.1 INTRODUCTION
The separations of oxygen and nitrogen from air, and hydrogen from carbon
monoxide, methane or nitrogen, are large consumers of energy in the chemical
processing industry. In general, purified gases are more valuable than arbitrary
mixtures of two or more components, since pure components provide the option of
formulating an optimum mixture for particular applications. Moreover, in some
cases, such as the inert blanketing of chemically sensitive products, a pure gas
stream is desired.
Energy-intensive compression of feed streams is often needed to provide the

driving force for permeation in membrane-based separations. In their simplest
ideal forms, represented schematically in Figure 3-1, membranes appear to act as
molecular-scale filters that take a mixture of two gases, A and B, into the feed
port of the module and produce a pure permeate containing pure A and a
nonpermeate containing pure B. Real membranes can approach the simplicity
and separation efficiency of such idealized devices, but, more usually, complex
recycling of some of the permeate or nonpermeate stream may be needed because
perfect selection of all of the A and B molecules cannot be achieved in a single
pass. Optimization of the properties of the membrane to achieve the most
efficient separation is important to avoid expensive and cumbersome systems that
fail to achieve the attractive simplicity of ideal membranes.’
In principle, as shown in Figure 3-2, three basic types of membranes can

be used for gas separations. The third type, solution-diffusion membranes (Figure
3-Zd), is dominant in current devices. In these membranes, the gas dissolves in
the membrane material and diffuses through the membrane down a gradient of
concenfration. Ultramicroporous (pores <7A in diameter) molecular-sieve
membranes (Figure 3-2~) have received increasing attention recently because of
claims that they may offer higher productivities and selectivities than solu-
tion-diffusion membranes.2 Many questions remain to be answered concerning the
durability, freedom from fouling and ease of large-scale manufacture of
ultramicroporous membranes, and further investigation does appear to be justified.
So-called “Knudsen” diffusion membranes (Figure 3-2b) cannot compete

commercially with the two preceding types due to their low selectivity between
molecules. Pores exist in the barrier layer of these membranes, but the pores
are smaller in diameter than rhe distance a molecule would travel in the gas
phase between collisions, on the order of 50 to 100 A in diameter or more. The
relative rates of flow across the membrane for an equimolar feed for typical gases
are given by the inverse square root ratio of the molecular weights of the gases.
For a 50/50 mixture of nitrogen and hydrogen, this would mean that for every 3.7
molecules of hydrogen that permeated, I molecule of nitrogen would permeate.
This separation factor is too low to be commercially attractive. As indicated in
Figure 3-2a. if pores are even larger than those that permit Knudsen separation,
nonselective permeation by viscous flow occurs, and this is obviously of no use in
separation processes.
189
Feed stream Non permeate

(A & B) m-earn (B)
1 Permeate stream
(A)
Figure 3-1. Generalized representations of an ideal membrane separation process.
cunuaum
(a) Viscous flow- b) Knudsen rlow- ) ulmmicropomus (d) Solution-Diffusion-
no sqnraion separation is molecular sieving- separation is based on
is achieved based on the separation is based bolh solubilily and
invcne square primarily on the mobility factors in
mot ratio of the much higher diffiasion csswttially all ca.ws.
moleular weights “es of lhe smallen Difikrivity vlecdviry
of A & 8. molecule. but sorption favor the smallat
level differences may molecule. Solubiliry
be important facfors wlectivi1y favon lhe
for similarly sized mos, condensable
pncvrnu like 02 a$. molecule.
Figure 3-2. Membrane-based gas transport and separation mechanisms.

Gas Separation 191
3.2 FUNDAMENTALS
High-performance nonporous polymeric and ultramicroporous carbon
membranes can achieve relative passage rates favoring hydrogen over nitrogen
exceeding 100, and they form the basis for simple, convenient processes to
separate a wide spectrum of gas pairs. The ultramicroporous membranes
(Figure 3-2~) are believed to have a tortuous, but continuous network of passages
connecting the upstream and the downstream face of the membrane.s
Solution-diffusion membranes have no continuous passages, but rely on the
thermally agitated motion of chain segments comprising the polymer matrix to
generate transient penetrant-scale gaps that allow diffusion from the upstream to
the downstream face of the membrane .s The penetrants undergo random jumps,
but because there is a higher concentration of them at the upstream face than
the downstream face, a diffusive flow occurs toward the downstream face.
By varying the chemical nature of the polymer, one can change the size
distribution of the randomly occurring gaps to retard the movement of one
species, while allowing the movement of the other. This is said to be a “mobility
selectivity” mechanism. A similar mobility control can be exercised in ultra-
microporous materials by controlling the size distribution in the network of
available passages to favor one of the components relative to the other. If one
could perfectly control this distribution, a true molecular-sieving process would
occur and infinite selectivity would be achieved. The essential impossibility of
such a situation is suggested by the data in Table 3-1, which shows the effective
sieving dimensions of various important penetrants as measured by exclusion from
crystalline cages in zeolites.’ It is amazing that, for both ultramicroporous and
solution-diffusion membranes, mobility selectivities of 2,000 have been reported
for helium over methane which differ by only 1.2A in minimum dimension.’ The
ability to regulate the distribution of transient-gap sizes in solution-diffusion
membranes is achieved by the use of molecules with highly hindered segmental
motions and packing. Typically, these materials are noncrystalline and are
referred to as glassy polymers.“ Indeed, they are organic analogs to ordinary
inorganic glass. The rates of diffusion through glassy polymers are sufficiently
high to allow their use as membrane materials, because the segmental motions are
more pronounced than in the inorganic materials. If one heats a glassy polymer
to a sufficiently high temperature, it eventually begins to behave as a rubbery
material and loses its mobility selectivity.
The mobility selectivity mechanism is not the only factor determining

membrane operation. Selectivity is determined, not only by the relative jumping
frequency of molecules A and B, but also by the relative concentrations available
for jumping. These concentrations are determined by the relative sorptivity of
gases A and B in the nonporous polymer or in the ultramicroporous network. One
can speak of a “solubility selectivity” analogous to the mobility selectivity
mentioned above,7 and the product of the solubility and mobility selectivities
determines the overall selectivity of the membrane. A good measure of the
relative solubilities of two gases is given by their respective boiling point.’ For
instance, helium and nitrogen have normal boiling points of 4°K and I l2”K,
respectively, indicating that helium is less condensable and will tend to have a
much lower sorption level in polymers and ultramicroporous media compared to
nitrogen. Subtle effects due to interactions between the membrane material and
gases are also seen sometimes, but they tend to be small and can be ignored in
most cases.
Table 3-1. Miminum kinetic (sieving) diameters of various penetrants taken from
the zeolite literature.’
It is possible for the mobility selectivity and the solubility selectivity to

complement each other or to oppose each other. For the polymer materials cited
above that have a mobility selectivity of over 2,000 for helium from methane, the
solubility selectivity for helium relative to methane is l/16, so the product of
these two gives an overall selectivity well above 100 for helium relative to
methane.’ For other gas pairs, such as carbon dioxide and methane, the two
factors both favor carbon dioxide over methane since carbon dioxide is smaller
and also more condensable. Typically, rubbery and glassy materials with similar
chemical natures (i.e., similar percentages of polar VS. nonpolar moieties in the
chain) have similar sorption properties for gases. In fact, glassy materials often
have higher sorption capacities than their flexible rubbery analogs, because of the
presence of unrelaxed packing defects present in the glassy polymers that are not
present in the rubbery state.‘v8
The relative rates of permeation of two components in an actual gas

mixture are usually within 10 to 15% of the rates measured with pure gases, SO
long as equivalent pressure differences of the two components exist between the
upstream and downstream faces of the membrane.g This approximation breaks
down at the point of “transport plasticization”, where the presence of a penetrant
affects the local rate of diffusion of another nearby penetrant by altering polymer
segmental motions. As the sorbed penetrants increase the ease of segmental
motions, the size range of the transient gaps tends to be less sharply controlled,
and the mobility selectivity begins to fall. In a loose fashion, one can say that
the previously glassy material starts to behave in a more “rubberlike” fashion;
however, it is really still a glassy material in most cases. In operating a
membrane, one must be on the lookout for such selectivity losses compared to the
pure component case, since they signal the onset of problem conditions for that
membrane material.
Another factor that affects the selectivity significantly is the operating

temperature of the membrane. Increasing the operating temperature gives rise to
a strong increase in polymer segmental motions, causing exponential increases in
molecular diffusion rates. However, higher temperatures tend to generate larger,
less size-discriminating gaps in the polymer matrix. Penetrants also become less
condensable as the temperature rises. The net results of these changes are
generally lower selectivities, but higher permeabilities. Permeability is the
product of the diffusivity and solubility parameters, and characterizes a
penetrant’s ability to move across a membrane of given thickness under a given
differential pressure drive force. Permeability is a fundamental property of the
Gas Separation 193
membrane material and permeant. lo In a practical membrane separation process,

one is generally forced to accept lower selectivities if higher productivities are
achieved by increasing the temperature of operation.
3.3 MEMBRANE SYSTEM PROPERTIES
It is convenient to divide the discussion of important properties for gas

separation membranes into technical and practical categories. Technical
requirements refer to those characteristics that must be present for the system to
even be considered for use in gas separation applications. Practical requirements,
on the other hand, relate to the characteristics that are important in making a
technically acceptable system competitive with alternative technologies, such as
cryogenic distillation or pressure-swing adsorption.
Because the solution-diffusion permeation process is slow relative to

Knudsen and nonselective viscous-flow transport mechanisms, an area fraction
containing nonselective pores even as small as 10W6 in the membrane surface is
sufficient to render a membrane effectively useless. The most critically
important technical requirement for solution-diffusion membranes, therefore, is to
attain a perfect pore-free selective layer. The selective layer must, of course,
also be able to maintain its defect-free character over the entire working life of
the membrane in the presence of system upsets and long-term pressurization.
For molecular-sieve membranes, a similar standard of perfection must be met

to ensure that essentially no continuous pores with sizes greater than a certain
critical size exist between the upstream and downstream membrane faces. The
exact size limit is not clear at present, since it may be that the effective
constriction sizes responsible for the molecular-sieving mechanism may be
influenced by the species that are, themselves, permeating. Adsorption on the
walls of the passageways may reduce the effective openings well below that of
the “dry” substrate. Unsubstantiated claims suggest that pore sizes as large as
10A in the dry substrate may be closed down to the 3-4A level likely to be
needed for molecular sieving of materials like carbon dioxide and methane.
Uitramicroporous membranes, like nonporous membranes, must adhere to rigorous
standards, because the pores must remain sufficiently clear of condensable
molecules or contaminants that would tend to inhibit the rapid passage of desired
product to the downstream side.
Because industrially important gas streams contain condensable and

aggressively sorptive or even reactive components, it is often desirable to pretreat
the gas to remove such components prior to feeding the gases to the module.
Pretreatment is not a major problem and competitive separation processes such
as pressure swing adsorption (PSA) also use feed pretreatments. However, the
more robust the membrane system is in its ability to accept unconditioned feeds,
the more attractive it is in terms of flexibility and ease of operation.
Besides the above technical requirements, practical requirements demand that

a membrane should offer commercially attractive fluxes. Even for materials with
relatively high intrinsic permeabilities, commercially attractive fluxes require that
the effective thickness of the membrane be made as small as possible without
introducing defects that destroy the intrinsic selectivity of the material. For
practical purposes, therefore, even high productivity polymers are generally not
used in a dense film form because of the enormous membrane areas that would be
required to treat commercial-size gas streams.
To achieve sufficiently thin selective layers (+oooA) and yet maintain

mechanical integrity, “asymmetric” membrane structures with complex
morphologies, such as those shown in Figure 3-3b-d, are favored over the simple
dense film shown in Figure 3-3a. Membranes comprised of cellulose acetate or
aromatic polyamides typify the “integrally skinned” membranes shown in
Figure 3-3b. Cellulose acetate and the aromatic polyamides can be formed with
nonporous skins using a process based on the original Loeb-Sourirajan techniques
sometimes called wet phase inversion.“~t2 Recently, a wide spectrum of other
materials have also been shown to be formable in this manner.‘s
Figure 3-3. Membrane types useful for solution-diffusion separations of

gas mixtures.
Polysulfone and many other attractive polymers can easily be made into
almost thin skin structures, but until recently it has been difficult to prepare a
truly perfect defect-free skin. To permit the use of such materials, therefore,
the “caulked composite membrane structure shown in Figure 3-3~ was
introduced.” The caulking procedure is useful, because it allows rapid production
rates of the basic membrane without extreme attention to the elimination of
defects. Defects are patched in a later processing step by coating the membrane
with a dilute solution of silicone rubber. Because the silicone rubber layer is
relatively thin and has a high permeability. it seals defects in the base membrane
without affecting the intrinsic permeability properties of the membrane.
Gas Separation 195
The membrane sealing approach described above derives from an earlier type
of membrane in which a porous nonselective support is coated with a second
material with desirable solution-diffusion separation properties.16 The support is
produced by wet phase inversion (Loeb-Sourirajan process), and the ultrathin,
selective skin can be produced by a variety of techniques, of which dip coating,
doctor-knife application or plasma polymerization are the most common. This
technique permits one to decouple the support function of the membrane from the
basic separation function of the skin.
In addition to flux, a practical membrane system must achieve certain

upstream or downstream product stream compositions. The ideal membrane
separation factor, that is the ratio of the intrinsic permeabilities of the two
penetrants, should be as high as possible to allow flexibility in setting
transmembrane pressure differences, while still meeting product-purity
requirements. The ideal separation factor also affects the energy used in
compressing the feed gas. Membrane selectivity also determines if multistage
system designs are needed to meet the required product compositions.
Unfortunately, high ideal membrane permselectivities often correlate with low
intrinsic membrane permeabilities, and this forces a compromise between
productivity and selectivity of the membrane material used. The trade-off
between intrinsic membrane permeability and selectivity is the major issue
concerning scientists searching for new polymers from which to make the
separating layers of gas-separation membranes. This important issue will be
summarized in a later section.
3.4 MODULE AND SYSTEM DESIGN FEATURES
The total gas flow from the upstream to the downstream side in a membrane
module (Figure 3-1) is determined by the integrated product of the flux and the
membrane area in the module. In principle, one can achieve satisfactory
production rates by simply increasing the area density (membrane area per unit
volume) in the module. Means of packaging gas-separation membranes, illustrated
in Figure 3-4, include plate-and-frame, spiral-wound and hollow-fiber modules.
The origins and attributes of the various module designs are discussed fully in
Chapter One.
(a) Plate and frame module of CKSS.
F..d .tr..m
0, rn1X.d p.s..
Figure 3-4. Gas separation module types.

Gas Separation 197
The plate-and-frame approach gives the lowest surface area/unit-volume

ratio. GKSS, of West Germany, has developed this technology for almost fifteen
years, and as much as 200 fts of membranes per fts of module can be achieved
with their efficient arrangement of the central collection tube, support frames and
permeate channels. This is roughly twice the area density available in early
plate-and-frame systems.
The hollow-fiber and spiral-wound module configurations, shown in

Figures 3-4b and 342, achieve substantially higher area densities. For example,
a IS-fold increase in permeation area per unit volume (to 3,000 ft’ of membrane
per fts) can be achieved by replacing even the advanced GKS plate-and-frame
module with 200-pm outside diameter fibers packed at a 50% void factor in a IO-
inch-diameter shell. Even higher area densities can be achieved by using smaller
diameter fibers and higher packing densities, the limiting factor being pressure
drop in the fibers and shell.
The spiral-wound configuration provides area densities intermediate between

the plate-and-frame and typical hollow-fiber modules. with 1.000 ft* of membrane
per fts of module being typical. As in the plate-and-frame case, this number
might be increased somewhat by optimization, but it will not approach the roughly
3,000 ft* of membrane per fts of module in the 200-pm hollow-fiber example
cited above.
Based on the need to maximize production rates, it appears that hollow-fiber

modules will eventually dominate the gas separation field. However, other factors
are also important. Until recently, it was possible to form the permselective
skins on flat membranes two to three times thinner than the skins on asymmetric
hollow-fibers. This meant the overall productivities of spiral-wound and hollow-
fiber membrane modules were similar. Recently, however, hollow fibers of
roughly 400-pm outside diameter were introduced by the Monsanto in their Prism@
Alpha module that had a caulked thin-skin of 5OOA thick. A SOOA skin
corresponds roughly to that achievable in spiral-wound systems. If other, more
advanced, membrane materials can be produced in equally thin selective layers in
hollow-fiber form, this module configuration will likely dominate the field. In the
absence of such capability, the spiral-wound and hollow-fiber modules will both
remain popular.
In high-pressure gas-separation applications such as hydrogen and carbon

dioxide separations, the pressurized feed gas is usually introduced on the shell
side of the hollow fiber. Fibers are much stronger under compression than
expansion and if individual fibers fail, they do so by being crushed closed. This
means they no longer contribute to the total membrane area, but rhey do not
serve as bypasses to contaminate the permeate with feed gas. The flows in
such modules have aspects of both countercurrent and crossflow patterns.’
For low-pressure applications where fiber failures are less likely, for
example the production of nitrogen from air, bore-side feed is common, even if
the skin layer is on the outer fiber surface. An advantage of this configuration
is that channeling and other flow distribution problems are eliminated.
Approximate crossflow or countercurrent flow patterns can be achieved with bore
feed by using a central collection tube (cross flow) or by removing product from
the shell at the same end as the feed entered (countercurrent).
Detailed proprietary computer models are used by suppliers to simulate the

behavior of their particular system under conditions of varying composition,
pressure, upstream to downstream pressure ratio and stage-cut. The stage-cut is
the fraction of the incoming feed stream that passes through the membrane as
permeate. Higher membrane selectivities minimize the stage-cut by limiting the
amount of the undesired component that passes through the membrane. For
nitrogen enrichment of air, where the residue is the desired stream, this means
lower compression costs are incurred to produce a given quantity of product gas
at the required purity. For hydrogen separation, where the permeate is the most
valuable stream, higher selectivities allow higher fractional recoveries of valuable
product, without falling below the product purity specification.
Gas separation membrane processes, in common with all rate-determined unit

operations, suffer depletion of the driving force, in this case, the partial pressure
difference along the unit. Except with an extraordinarily permselective
membrane, high-purity permeate cannot be produced unless the partial pressure of
the faster permeating gas in the feed stream is maintained at a relatively high
level. For cases such as helium recovery from natural gas. where the initial
concentration of helium is <I%. a single-stage unit will not produce high-purity
helium with current membranes. Moreover, in many cases, production of a
high-purity permeate and high-purity residue is required simultaneously. In these
cases, more than one module stage is required.
Systems requiring multiple compression operations are unattractive because

of high capital and operating costs, so most system designs are limited to two or
three module stages.16 Nevertheless, to complete an adequate engineering and
economic analysis of even two or three stage systems is a complex problem.
Spillman notes that due to their novelty, early publications concerning the use of
membranes often ignored the importance of optimization of the system design.17
A more mature consideration of recycle and other optimization options can make
membranes the clearly favored choice for a particular separation, even though a
membrane process may have appeared marginally viable previously.
Spillman further notes that working with membrane design requires an

objective consideration of economically acceptable recovery rates. The value of
lost product should be considered an operating cost, similar to other expenses.
In some cases, membranes do not neatly replace existing separation processes and
a rethinking of the process logic may be beneficial. The nature of the
separation is different than, say, for PSA. so process changes are sometimes
necessary, and the specific impact of these process changes must be considered.
A typical situation is illustrated by the separation of hydrogen in refinery

streams. In a membrane process the purified hydrogen is produced at low
pressure. PSA, on the other hand, delivers the hydrogen at close to the feed
pressure. Which process is selected will depend on the pressure at which the
product hydrogen is to be used. Increased familiarity of plant designers with
membranes, coupled with increasingly sophisticated computer software, should
promote the use of membranes while avoiding inappropriate applications.
Gas Separation 199
3.5 HISTORICAL PERSPECTIVE

Gas separation using polymeric membranes was first reported over 150 years
ago by Mitchell in a study with hydrogen and carbon dioxide mixtures.‘s In 1866,
Graham made the next important step in understanding the permeation process.lg
He postulated that permeation involves a solution-diffusion mechanism by which
penetrants first dissolved in the upstream face of the membrane were then
transported through it by the same process as that occurring in the diffusion of
liquids. Graham demonstrated that atmospheric air could be enriched from 21% to
41% oxygen using a natural rubber membrane, and that increasing the thickness of
a pinhole-free membrane reduces the permeation rate, but does not affect the
selectivity.
The above advances were not matched by a practical means of applying the
permeation process to large-scale separation of gas streams. Early attempts by
Union Carbide to apply plate-and-frame technology for helium separation were not
successful, because the fluxes were too low to be of commercial interest.” The
discoveries of Loeb and Sourirajan in developing integrally-skinned asymmetric
membranes (Figure 3-3b) were not applied immediately to gases, because the
porous substructure collapsed upon drying and multiple defects were introduced in
the selecting skin.21
In the 197Os, discovery of the extraordinarily high permeability of silicone

rubber encouraged a group at General Electric (GE) to develop a novel process
for ultrathin membranes based on silicone rubber or silicone rubber/polycarbonate
copolymers.22~2s Pinholes were eliminated by laminating multiple layers to yield
separating layers of 1,OOOA supported on a microporous substrate. The membranes
were housed in plate-and-frame forms to yield modules with orders of magnitude
higher productivity than found in the original work by Union Carbide. This
research at GE laid the groundwork for later developments in composite mem-
branes discussed earlier (Figure 3-3d). The Oxygen Enrichment Company was
started as a result of the work at GE to produce oxygen-enriched air systems
for home medical use. This company continues to supply products based on this
technology. The technology did not develop to the industrial production of
oxygen-enriched air because the membrane fabrication process and system
performance were not adequate for large-scale, less-costly applications.
In 1970, a process that allowed asymmetric cellulose acetate membranes to be

dried and used in gas separation applications was developed, but not pursued.”
The field was dormant until 1977, when a group at Du Pont applied melt-spinning
technology to produce polyester hollow-fine-fibers with inside diameters of 36 pm
for high-pressure hydrogen applications.a6 Instead of seeking to increase flux
by minimizing membrane thickness, Du Pont accommodated the low flux using in-
credible area densities, approaching 10,000 ft2 of membrane/fts. Because of their
robust, dense walls, the fibers were fed on the bore side without danger of fiber
failure.
A patent by Du Pont in the 1970~~~ also identified many glassy polymer ma-
terials with unusually high intrinsic permeabilities. These materials had much
better potential than the low productivity polyester for gas separation in hollow-
fine-fiber melt-spun form. In asymmetric membrane forms, these are attractive
even today. The technology was mothballed in 1979, as Du Pont focused on
reverse osmosis.
Coincident with Du Pont’s withdrawal from the gas separation field,

Monsanto announced the revolutionary concept of the “resistance composite”, or
“caulked membrane” (Figure 3-3~). 14.sr The clever approach involved applying a
thin coating over, or applying a vacuum to suck silicone rubber into, surface
defects present in asymmetric hollow fibers. The coating eliminates Knudsen and
viscous flow that undermined the intrinsic solution-diffusion selectivity of the
skin. Although used commercially only for the polysulfone material, it was clear
that the approach and patent coverage was generaI and would also have been
applicable to materials such as those described in the Du Pont patent. Monsanto
chose to work with standard polysuifone because of its ready availability,
adequate properties and easy processability. Timing was excellent, because of
the “energy crisis”, and Monsanto moved strongly into the hydrogen separation
field. Initial applications were hydrogen recovery from ammonia purge streams,
refinery and petrochemical off-gases.
Most gas-separation work to date has been performed with polymeric

membranes. However, membranes made from palladium-silver alloys have been the
subject of substantial research and limited commercial development.2s These
membranes have extremely high hydrogen/methane selectivities, and can be
operated at elevated pressures and temperatures above 4OO’C. These systems are
still available today, but have not been a large scale commercial success.
3.6 CURRENT TECHNICAL TRENDS IN THE GAS SEPARATION FIELD
3.6.1 Polymeric Membrane Materials
Identification of advanced materials has been an important driving force

propelling gas separation membrane technology in recent years. An indication of
the advances in performance that may be possible with new membrane materials is
given in Figure 3-5. The figure shows “best-case” permeability/selectivity curves
attainable with present commercial polymers. in fact, most individual commercial
materials fall below the lines, either in permeability, selectivity or both. For
example, commercially available polyvinyl fluoride has a selectivity of 170 for
helium/methane but a helium permeability of 1 Barrer. On the other hand,
silicone rubber and polyphenylene oxide have helium permeabilities of 300 and 105
Barrers, respectively, but helium/methane selectivities of 0.4 and 24. Similar
situations apply for the other two gas pairs in Figure 3-5.
Several studies have reported materials that deviate favorably from the
standard correlation line defined by the properties of commercially available
materials. although generalized correlations of polymer properties and separation
are not yet possible.s~2Q~sa Within a given family of polymers, however, rules can
be identified to guide the optimization process. Figure 3-5 shows data points for
two classes of materials. the polycarbonates and polyimides, to which such
optimization procedures have been applied. The polymers and their structures are
identified in Table 3-2. The results in this figure are typical of the new-
generation polymers, which offer higher permeability at equivalent selectivity, or
higher selectivity and higher permeability, than the points on the present “best-
case” lines. The field of membrane material selection is complex. The most
important points can be distilled into two rules of thumb, summarized below and
discussed in more detail in recent publications.s’
Gas Separation 201
O2 P.r...billty
Figure 3-5. Typical trade-off between selectivity and permeability for gas
separation systems. The solid lines represent best case expectations
based on commercial materials. The numbers and letters refer to
material in Table 3-2.
I. Structural modifications that simultaneously inhibit chain packing and the

torsional mobility around flexible linkages in the polymer backbone tend to
cause either: a) simultaneous increases in both the permeability and
selectivity, or b) a significant increase in permeability with negligible loss
in selectivity.
2. Modifications that reduce the concentration of the most mobile linkages in a

polymer backbone tend to increase selectivity without undesirable
reductions in permeability so long as the modification does not
significantly reduce intersegmental d-spacings.
Table 3-2. Systematically varied polyimides and polycarbonates illustrating the

value of structure-property optimization for improving the trade-off
between selectivity and permeability.
Polyimides Polycarbonata
Standard
bisphenol-A
Polycarbonate _&$&I_
(a) PC >
Hcxafluorinatcd
bisphcnol-A
Polycarbonate _++J+!_
2
(b) HFPC
Tctramcthyl
bisphenol-A
_+f$gZ_
Polycarbonatc 5”
i
(c) -rMPC
Hexaftourinated
Tetramcthyl
bisphenol-A
Polycarbonatc
~FDA-DAP (d) HFl-MPC

Gas Separation 203
The application of membranes at high temperatures in aggressive chemical

environments may require the use of structures with intractable skins comprised
of exotic polymers. This area is likely to be of increasing interest as experience
is gained in module formation for such demanding applications. If an organic
membrane existed now with the ability to sustain extended operation above ZOO’C,
the potting and other components in the module would need to be upgraded for
this service, and virtually no research seems to be underway in this regard. The
polyimides and aramides are probably the best materials studied to date for
potential high-temperature applications. Data for permeation of carbon dioxide
and nitrogen as a function of temperature indicate a much lower temperature
dependence of permeation for rigid polyimides, as compared to most commercial
glassy materiaIsz2 This indicates that rigid materials like the polyimides are
less affected by increases in temperature than the standard glassy polymers with
lower glass transition temperatures. The ability to maintain glassy-state
separation properties at elevated temperatures may find application in
membrane-reactor hybrid processes.=
3.6.2 Plasticization Effects
As previously discussed, plasticixation of the membrane polymer by a sorbed

permeant can affect the permeability and selectivity properties significantly. An
illustration of apparent plasticixation behavior is shown in Figure 3-6. The
permeability of carbon dioxide in silicone rubber and various substituted
polycarbonates is seen to vary considerably with upstream pressure. In each case,
downstream pressure was maintained at less than 10 mm Hg.sl
For standard unconditioned poiycarbonate, the upturn point in permeability

is not reached until roughly 600 psia carbon dioxide pressure. For
hexaflourinated tetramethyl bisphenol-A polycarbonate, HFTMPC, the point is
reached at only about 250 psia. due to the much higher solubility in this “open”,
packing-inhibited material. In terms of permissible carbon dioxide partial
pressure prior to the onset of plasticixation, tetramethyl bisphenol-A
polycarbonate, TMPC. and HFIMPC may be less attractive than standard
polycarbonate, because of their high sorption coefficient caused by larger
d-spacing. These data suggest that useful operating ranges for different polymer
materials will vary considerably. Further research is needed to better understand
these preliminary observations.
4 -
WOO-
I I I 1 6 ’ 1 ’ I I
P 0 200 400 600 800 0 200 400 600
E pressure [ psia]
m
z
36-l ” 1’ ’ -
p1 cssure [ psia]
pressure [ psia]
Figure 3-6. Carbon dioxide permeability at 35’C for silicone rubber and the
polycarbonate polymers in Table 3-2 as a function of upstream CO,
pressure.
PC: Standard bisphenol-A polycarbonate

HFPC: Hexafluorinated bisphenol-A polycarbonate
TMPC: Tetramethyl bisphenol-A polycarbonate
HFTMPC: Hexafluorinated tetramethyl bisphenol-A polycarbonate
Gas Separation 205
3.6.3 Nonstandard Membrane Materials
Carbon, molecular-sieve membranes formed by pyrolysis of unspecified

polymer precursors are being developed by Rotem. the commercial arm of the
Israeli nuclear research center in conjunction with the Swedish Gas Company,
Aga.2*M Helium fluxes through the membranes of 200x10-s
cm3(STP)/cm2+cmHg and hydrogen/methane selectivities above I.000, based on
pure gases, have been reported. 2 The materials, however, have not been tested
thoroughly in mixed gas streams. Due to their ultramicroporous structure, they
are likely to have serious problems with “plugging” by condensable agents such as
water. It may be possible to eliminate adsorption problems by operating at high
temperatures, without the need for major feed pretreatment. This area appears
to have considerable potential. if the inherently fragile nature of the membranes
can be dealt with.
In principle, ceramic and glass membrane materials offer similar properties,
although the minimum pore sixes attainable are currently around 25A for ceramics
and IOA for glass. These dimensions are 2-6 times larger than the minimum
dimensions of the various molecules that are of interest in gas separation. As
noted earlier, molecular adsorption onto pore walls may reduce the effective pore
size in some cases. Moreover, at low temperatures, the possibility of additional
transport modes, such as surface diffusion of condensable agents, playing a part
in the separation process may make ceramic or glass membranes viable.
Systematic, fundamental studies are needed to assess the true potential of the
approach.
3.6.4 Advanced Membrane Structures
Integrally-skinned asymmetric membranes, made by the Loeb-Sourirajan

process, are commonly used in commercial membrane gas separation. Although the
Loeb and Sourirajan approach has successfully been applied to defect-free
cellulosic-based asymmetric membranes, it has been found difficult to prepare
thin-skinned defect-free membranes from other polymers. The “caulking”
process used by Monsanto in the original Prism’@. and later Prism” Alpha, products
with standard polysulfone provides a solution to this problem, but patent coverage
precludes its generalized use by other companies in the field.27
The preparation of ultrathin, flat-sheet asymmetric gas-separation membranes

has recently been reported for polyethersulfone and polyetherimide by GKSS3’
The membranes showed some skin-layer defects that had to be post-treated, but
the Monsanto patent was avoided using a different post-treatment procedure.i3
An approach that eliminates the need for post-treatment, yet produces an
ultrathin, defect-free skin on flat-sheet membranes, has recently been reported.36
For optimum value, new skin-formation processes should be applicable to both
hollow-fiber and flat-sheet membranes. Work in this area is likely to be an
important activity in the future.
3.6.5 Surface Treatment to Increase Selectivity
Two types of surface treatment have been considered: (i) reactive and (ii)
physical (antiplasticization). Techniques in the reactive category include
fluorination, silylation and photochemical crosslinking.37ss Fluorination of
polytrimethyl-silylpropyne (PTMSP) has been studied at Air Products. Recent data
from Air Products suggest that fluorinated PTMSP exhibits stable properties under
exposure to organic materials such as vacuum pump oil. a result that conflicts
with earlier studies.3Q However, there are reports that a lower-flux, but perhaps
more reliable polymethylpentene material is the basis for a fluorinated membrane
that will be introduced in the near future. Presumably, the fluorination will
produce increases in selectivity as in the case of the PTMSP.
Photochemical crosslinking of labile benzophenone groups has been described

by Du Pont for polyimides based on benzophenone dianhydride.se This approach
produced large increases in the selectivity of a dense film, with small reductions
in apparent permeability. As in the case of fluorination, however, the reduction
in the permeability of the outer skin was probably very high. If the entire thin
skin of an asymmetric membrane were treated by either a fluorination or
photochemical process, very significant reductions in flux might be expected as
compared to the unmodified material. The key to producing useful materials
seems to be to use an intrinsically high-flux material to form a thin-skinned
(ml,OOOA) asymmetric membrane from that material and then to reactively surface
treat the skin layer. If fully developed, this technique could result in membranes
in which attractive productivities are maintained, along with significantly
improved selectivities.
The incorporation of physically bound components into polysulfone has been

reported to increase selectivity by an antiplasticization phenomenon.40*41 The
effect is believed to result from the inhibition of chain motions by the
antiplasticizers. It is not desirable to load the entire membrane with
antiplasticizers. because loss of mechanical properties occurs. Even if extremely
nonvolatile components are used as the antiplasticizing agents, calculations
suggest that the small skin volume and the extremely high passage rates of
permeating gases through the skins will effectively remove the antiplasticizing
agents in a relatively short period of time (<3-6 months), so this approach
appears much less attractive than the reactive ones noted above.
3.6.6 System Design and Operating Trends
Little activity appears to be under way with regard to the design and
operation of modules. A notable exception to this is a patent by A/G
Technology relating to the use of a bore-side feed for hollow-fiber modules.”
Both bore-side and tube-side feeds have been used since the beginning of
hollow-fiber membrane technology. Indeed, the first Du Pont hollow-fiber module
for hydrogen separation was based on tube-side feed. The current Du Pont,
Monsanto, UOP and Dow modules for hydrogen separation are all believed to use
tube-side feed.
For low-pressure applications such as nitrogen enrichment, where fiber

failures are unlikely, bore-side feed is common. The details of internal flow
paths are proprietary. Designs that give behavior from countercurrent to
crossflow and mixtures of the two are believed to exist. Differences in product
takeoff and feed introduction points also result in a range of possible
configurations.
Gas Separation 207
More significant differences exist in the internal and external design details
for gas separation membrane units than are found in reverse osmosis units, for
example. The standardization that came about in reverse osmosis technology as
a result of the information exchange provided by the Office of Saline Water
Research program is absent from the gas separation scene; instead each company
has taken its own divergent path. As the technology matures some measure of
standardization may emerge due in part to competition for the replacement
module market.
Module and system design for gas separation applications are a complex
issue. System designs require consideration of numerous possible scenarios
including single-stage, multistage, recycle and so on. The design is typically
done by the supplier, using a proprietary computer program to optimize the
system configuration and minimize the cost based on known or required selectivity
and productivity parameters. This trend promotes user dependence upon the
supplier for innovation. For some applications, for example, production of
nitrogen from air for blanketing or inerting, this situation may be perfectly
satisfactory. On the other hand, in petrochemical and refinery hydrogen
applications, the user may view the membrane system as one of many integral
and interacting unit operations in a plant. Sophisticated system-design programs,
able to consider multiple case studies, are likely to be developed by the user,
without concern for the details of module internals. Using module performance
information from suppliers, such programs allow independent consideration of
different vendor suggestions as well as user proposed flowsheet options. In this
respect, users may well become an important source of innovation.
3.7 APPLICATIONS
Table 3-3 summarizes currently commercial and precommercial applications of

membrane-based gas separations. The key features of the various separation
types are discussed below.
3.7.1 Hydrogen Separations
The first membrane-based, gas-separation systems were used for hydrogen

separation, from ammonia purge gas streams, and to adjust the hydrogen/carbon
monoxide ratio in synthesis gas.
Table 3-3. Current and Developing Applications of Membrane-Based Separations
Gas Components: Application Comments & Key Technical Problems
H,MI : Ammonia Purge Gas Successful.... umdensibles (H,O or NH,) must be rcmowd
HJ CH, : Rcfmery H, ruxvery Successful, but condensible hym undesilablc or fatal
Hi CO : Synthesis gas ratio adjustment Successful, but condensiblc methanol must be rattovcd
OJN, : N, enriched incrting medium Practical for 95%... marginai for 98%+, next generation membranes
(higher selectivity) will improve position vs PSA
Home medical oxygen enrichment Technology exists. no serious problems but small market
0, enriched burner gas Higher temperature burner designs needed.... new generation of membranes
possibly adequate productivitya for ecmmmics
Highly-O, (~40%) enriched gas Selectivity of polymeric membranes much too low . . a = 60 needed
Acid gases/ Hydrocarbons :

CO, recovery from bio and Successful.... condensible organics removed even in competitive PSA
kindiiU gases process and water scrub requires water cleanup for pollution control....
better productivitya tradeoff is desirable
CO, recovery from well Successful. but condensibles mustbenmovedand better productivity-a
injections and plasticizationresistance is desirable
H$ removal from sour gas No known installations, but current ox emerging membranes may Wok
[continued)
Gas Components: Application Comments & Key Technical Problems
H,O I Hydrocubons: Natural gas drying Effeetivc, but hydmcarbcm lossa lo pamc~c marginally acwpublc
\O / Air : Air dehydration fffcctivc. for iniumediue dewpoinu. canccnbaion polarization in permeate
must be overcome
Hydrocutnms / Air : Pollution control, Successful for chlorinsti hydmwbons. Pamuu. tends u) become oxygen
solvem re.uwcly enriched (flammability cunccm) and vacuum systan design is exotic
Hydroc&ns(CH,)/N,: Cutxnt membrane4 ue not selective enough to avoid excessive loss of

Upgrading of low BTU gas me4haoeintodupamutesuum
He4 Hydrocarbons : Helium recovay Low He fozd cnncentruio~ require multistage opauion. market is small
fmm gas wells
He4 N, : Helium recovery from diving Fusible. market is small

air mixtures
Hydrogen separations from highly supercritical gases, such as methane,

carbon monoxide, and nitrogen are easy to achieve by membranes, because of the
extremely high diffusion coefficient of hydrogen relative to all other molecules
except helium. Even though solubility factors are not favorable for hydrogen,
the diffusion contribution dominates and gives overall high selectivities. For
example, the hydrogen/methane selectivity of some of the new rigid polyimide and
polyaramide membranes is about ZOO. An example of Monsanto’s use of membranes
for synthesis gas composition adjustment is the production of methanol from
synthesis gas. Monsanto has published a study of a plant in Texas City producing
100~10~ gal/yr of methanol. ‘s The combined methanol/syngas process, shown in
Figure 3-7, requires a hydrogen:carbon dioxide ratio of 3:l for feed to the
reactor/interchanger. The overall flow diagram is shown in Figure 3-7a and the
separator flow diagram indicating the detailed process conditions is shown in
Figure 3-7b. The separator train involves two parallel banks of four IO-ft long,
8-in diameter modules arranged in series, to handle the 4,150 scfm of feed gas at
682 psig. Specific information about the 8-in-diameter modules is proprietary, but
if one assumes a standard 200 pm OD fiber and a 50% packing factor, roughly
80,000 ft2 of total membrane area is present in the compact separator train.
The feed stream leaving the methanol scrubber is saturated with water and
is preheated prior to entering the separator train to prevent condensation on the
membranes. The permeate gas, consisting largely of hydrogen and carbon dioxide
is sent back to the synthesis reactor, while the nonpermeate from the last
separator is sent to the fuel header to be burned. Roughly half of the available
hydrogen and carbon dioxide is recovered, leading to a net increase of 2.4% in
methanol production, simply by elimination of the wasted raw materials lost in
the purge. The slightly slower permeation rate of carbon dioxide relative to
hydrogen requires the addition of a small amount of carbon dioxide makeup gas to
the recycle stream to maintain the 3:l stoichiometry required. Accounting for
this supplemental carbon dioxide and a debit for the lost fuel value of the
reclaimed raw materials, the permeator installation permitted an impressive 13%
reduction in variable costs per gallon of methanol.
Monsanto’s success encouraged companies with technology in asymmetric

cellulose acetate, which could be dried out without the need for a caulking post
treatment, to enter the gas separation market. In the early 1980s Dow (Cynara),
Grace, Envirogenics and Separex began to market gas separation systems. Except
for the hollow-fiber Cynara product, all modules were spiral-wound. Except for
the Envirogenics module, all of these products are still being sold and have had a
measure of acceptance. Separex, which was recently purchased by Hoechst-
Celanese. has sold more modules for hydrogen separation than the rest of the
cellulose acetate manufacturers. Air Products showed that by using 18.5 million
standard cubic feet/day (mmscfd) of a typical synthesis gas feed stream comprised
of 73% hydrogen and 24% carbon monoxide, the feed to the methanol reactor
could be adjusted to match the stoichiometry needed in the synthesis reactor.
The permeate from the process was 98% hydrogen. and Air Products indicated that
cost of the membrane system was less than half the cost of a competitive
pressure swing adsorption (PSA) system.”
Gas Separation 211
I I Prism Srporotorr I I I
Q-y MrthonolCrudr
Srpototll.00
Figure 3-7a. Schematic representation of complete oxo-alcohol process showing

the relationship of the two banks of &in. diameter, IO-ft-long
Prism@ separators.
NonPrrmtOntCOS
-to furl Htodrf
n,. 41.3’4
co: 13%
Prlrm Srporotoo cot-t5 6%
cnr 2922
Pvrqc co, -
1 - JA-KM-IF- Ml: 0.4%
nz 3a4-k flo.‘: 2700 SCfN
co. 5.3% Prcssurc *
676 p,lq
co1 I4 0%
Hfdroqrn Ptrmton
CM, 213% .a.__
Nl. 0.2%
H* 79 6%
MtON. 0.6%
flowr 4150 SCfN
co 1.ax
coz.11 0%
Pwtrurc : 662 ps,q
cn,- 7 4’L
Nz:O%
flow I1450 SCf Y
~rerw :
3ao ps,q
uton 0%
nzo ait
flow a0 9 CPM
Figure 3-7b. Detailed flow, pressure and composition diagram for the Prism@
separator system in Figure 3-7a.
Spillman summarizes numerous other hydrogen recovery case studies iovolv-

iog ammonia synthesis and refinery applications.” A specific example is the
recovery of 98% pure hydrogen from a 15 mmscfd stream containing 75% hydrogen
by means of a Du Poot polyaramide hollow-fiber membrane system. The stream
ori-ginates from a hydrodesulfurixatioo purge stream in a Coooco refinery. The
investment and operating costs of the membrane system, compared with PSA,
cryogenic treatment and incremental reforming in a hydrogen plant are shown in
Figure 3-g.
High Purity(75%) HDS Purge
” Du Pont P.S.A Cryogenic Hydrogen ”

Membrane Plant
Figure 3-8. Comparison between process economics for hydrogen recovery for a
hydrodesulfurixation (HDS) purge gas stream.”
The excellent temperature stability of the newer, high-performance materials

should allow their use at temperatures above IOO’C in the future, so long as
potting materials capable of high-temperature operation can be developed.
3.7.2 Oxygen-Nitrogen Separations
Membranes can be used to separate air to yield either nitrogen- or oxygen-

enriched air as the final product. For production of nitrogen-enriched air, the
stage-cut (fraction of feed air passing through the membrane) is set to allow
sufficient oxygen to pass through the membrane to reduce its mole fraction to
whatever level is desired. In a perfectly permselective membrane, at zero
Gas Separation 213
downstream pressure, the stage-cut could be set at roughly 21%, pure oxygen
collected as permeate and pure nitrogen collected as residue. The current
generation of membranes have oxygen/nitrogen selectivities of 4-5, but higher
ideal separation factors are becoming available. The next generation of modules
will probably operate with stage-cuts of less than 45% to produce high purity
nitrogen.
By far the largest current application is to produce nitrogen-enriched air for

inerting of foods, fuels, etc. In 1987, Monsanto introduced their Prism Alpha@
membranes, asymmetric membranes with an improved structure that gives rise to a
thinner selective skin layer.
A/G Technology, a small company with experience in ethylcellulose, hollow-

fiber spinning, entered the air separation market and carved out a niche in small-
scale nitrogen inerting systems. Union Carbide’s hollow-fiber composite
membrane system was also introduced successfully for both nitrogen inerting and
hydrogen separation applications in 1987. A large membrane plant to produce
inerting nitrogen upon demand was started up at Johnson & Johnson in 1988 and
appears to have been successful.
The nitrogen-enriched air market represents a huge potential opportunity

for membrane systems, but entrenched gas companies can reduce liquid nitrogen
costs to inhibit market-share losses to membranes. Nevertheless, gas suppliers
have begun to accept that membranes will cause eventual loss of markets from
conventional technologies. The result is the formation of numerous joint ventures
between membrane suppliers and gas companies, for example Du Pant/Air Liquide,
Dow/British Oxygen, Akzo/Air Products, Rotem/Aga and Allied Signal/Union
Carbide.
The current generation of membranes seem to be extremely attractive for

95% nitrogen; however, PSA is a strong competitor for the 98%+ market. Recent
patents and announcements by Dow of higher selectivity polyester carbonate
materials suggest that the 99%+ market may soon be at least partially served by
membranes.
The evolving picture in the air separation field is illustrated by Figure 3-9
in which the shaded area indicates the improvement in capability of membranes
since 1985.17 Even prior to the introduction of the next generation membranes,
PSA is being challenged for the 98%+ market. A case quoted by Spillman shows
that a metal powder facility using 286.000 scf of nitrogen per month spends 50.53/
cscf for liquid nitrogen plus tank metal. A Prism@ Alpha system producing 99%
nitrogen is claimed to yield a nitrogen cost of $0.26 per cscf, amounting to over
a 50% savings compared to delivered liquid nitrogen, even when capital and
maintenance costs are included.17
A recent comparison between A/G Technology’s ethylcellulose hollow-fiber

module and PSA for the production of 95% nitrogen indicated roughly equivalent
costs of SO.l3/cscf. Replacing the ethylcellulose membranes with a higher
selectivity material, such as polysulfone, would favor the membrane option.
Moreover, membrane units are light and simple compared with PSA, so in
applications such as aircraft fuel tank inerting they would tend to be favored.
Mole Percent Nitrogen

100s
0.01 0.1 1 loo fmo

Fiowrati: C-SCFH
Figure 3-9. Updated comparison of air separation technologies and sources

for nitrogen gas. The shading reflects advances in gas
separation membranes made since the original figure was
prepared in 1985.17
Gas Separation 215
Except for home medical uses, membrane-generated, oxygen-enriched air is

not currently a significant market. A 45-50% oxygen-enriched air permeate is
at the limit of what can be achieved with present membranes. The Japanese have
identified oxygen-enriched air feeds to burners as an important priority. A
commercial installation by Osaka Gas Company to generate oxygen-enriched air
for improved methane combustion efficiency, described in a 1983 paper, is believed
to be the first of its kind in the world.‘d A four-year DOE-supported study of
the use of oxygen-enriched air in a copper tube annealing furnace was
demonstrated to be practical using even the low-selectivity materials available ten
years ago.” The use of membrane-produced, oxygen-enriched air for combustion
applications appears to be a sleeping giant. Improved membranes with suitable
productivity and selectivity will soon be available to make the oxygen-enriched
burner feed market attractive. A/G Technology’s assessment of the relative cost
of available oxygen in 35% oxygen-enriched air indicates a cost of only S28 per
ton for membranes compared with 542 per ton for PSA.” Ward and co-workers,
in an article published in 1986, indicate that even at S55/ton the economics of
oxygen enrichment for burners appear attractive. The article shows that, with
high-efficiency modules, it is technically feasible to save roughly 40% of natural
gas fuel costs by operating with 30% oxygen feeds.4g As burner designs evolve to
accommodate inexpensive oxygen-enriched air sources, this aspect of membrane
technology will likely burgeon.
The use of sterile, oxygen-enriched air for biotechnology is also a market

that is likely to develop as the systems for air processing become more familiar
to the public.
3.7.3 Acid Gas Seoarations
Currently, the principal application in this category comprises carbon

dioxide removal from a variety of gas streams containing primarily methane as the
second component. Such gases arise from landfills and from enhanced oil
recovery (EOR) projects. Enhanced oil recovery operations involve the injection
of high-pressure (2,000 psia) carbon dioxide into a reservoir via an injection well,
and removal of carbon dioxide, light gases and oil from production wells around
the periphery of the injection well. Monsanto, Dow and Grace have all been
active in the area of membrane systems to recover the injected carbon dioxide.
A Dow installation at a Sun Oil field has run successfully for over five years
without replacement modules, considerably longer than the original two- to
three-year lifetime expectations. The membrane modules are still owned by Dow,
so careful control of module exposure to adverse conditions has been maintained.
A complete study of a typical carbon dioxide recovery system in an EOR

application has been done by ShelLso In this study, the 95% carbon dioxide
permeate stream from the membrane can be used for reinjection; the nonpermeate
stream, containing 98.5% methane. forms pipeline quality natural gas. Spillman has
shown that multistage membrane systems can be competitive with recovery of
carbon dioxide by the conventional diethylamine (DEA) process, even at carbon
dioxide concentrations as low as S%.” By resorting to multistage processing, it
appears that membranes provide markedly less expensive processing costs than
amines for this case as well as the higher carbon dioxide concentration cases.
The difference between the single-stage membrane, the optimized multistage
membrane and the DEA process is illustrated in Figure 3-10 and Table 3-4.
Table 3-4. Detailed economic comparison between amine treatment and
optimized multistage membrane process for CO, removal from natural
gas.”
CO1 Content ol Feed 5%

- 10%
- IS%
- 20%
- 30%
- -40%
Amine
Capilrl(MillIons
ofDdlars) 3.35 4.54 5.45 6.21 7.50 8.56
fxpenses(Mllllons
ofhhrS/YeJr) 1.22 1.81 2.33 2.82 3.73 4.58
LostProduct(Mlllbns of Dollm/Year) 0.02 0.04 0.07 0.09 0.14 0.19
CapitalCharge(Milllonrof Dollrrs/Yerr) 0.91 1.23 1.48 1.68 2.03 2.32
ProcessingCOG (Dolfars/MSCF Feed) 0.17 0.24 0.30 0.36 0.46 0.55
Membrane (Mulllrtqe Process)

Capital
(MIllions
alDollars) 1.86 3.33 3.87 3.69 3.37 3.32
Eqases (MillIonsofDollars/Year) 0.53 0.85 0.97 1.00 0.98 0.96
LostProduct(MIllIons of Oollarsflcrr) 0.43 0.69 0.93 1.24 1.54 1.49
CapitalCharge(Millions 01 Dollars/Year) 0.51 0.90 1.05 1.00 0.91 0.90
Processing
Cost (Doffrrs/MSCF Feed) 0.11 0.19 0.23 0.25 0.27 0.26
so.40
SO30
so20
SO.10
Flow Rate = 37.2 -fd
Figure 3-10. Economic comparison between amine treatment, single-stage and

optimized multistage membrane processes for CO, removal from
natural gas.”
Spillman further notes that although the methane loss for membrane systems may
exceed that in amine systems, the fuel requirements to operate the amine unit
more than offset these losses in the overall economics.
Current indications are that using membrane separations in natural gases and
in EOR applications looks very promising. Even in the case of future, large-scale
EOR projects (which are currently on hold due to low oil prices), it may be wise
to build a small- to medium-scale startup cryogenic plant to handle the base load
and add additional capacity only as it is needed during the aging of the field as
carbon dioxide concentrations increase.
EOR is by no means the only opportunity for carbon dioxide selective

membranes. Naturally occurring gases containing up to 50% carbon dioxide are
also excellent candidates for membrane-based gas separation processes. The
range of feed pressures and compositions in these cases can cover a broad
spectrum; streams containing more than 15% carbon dioxide look most attractive
for membranes. Schendel et aL5r have recognized that, in many applications,
membranes will be able to compete with cryogenic systems, even when the cost
comparison favors cryogenics, because of the flexible and modular nature of
membrane systems.
In addition to the naturally occurring sources of methane, gases produced

from municipal waste or landfills can produce a methane-rich gas containing up
to 45% carbon dioxide with various impurities that must be dealt with. It has
been estimated that the municipal waste from U.S. cities, if properly processed,
could produce up to 200~10~ scf/year of methane.52 Moreover, since landfills
tend to be near urban centers, the cost of moving the gas is minimized.
In such applications, membranes will have to compete with a variety of other

technologies including water scrubbing, PSA and direct use of the contaminated
gas as a low-energy fuel. Typical landfill gases contain significant amounts
(100-200 ppm) of chlorinated hydrocarbons, as well as a broad spectrum of other
hydrocarbons. Thus, the clean-up and use of the gas via PSA or membranes is
preferable to direct combustion.
For both natural gas and landfill gas applications, more selective and
productive membranes would improve the economics by reducing hydrocarbon
losses to the permeate. Particularly for high-pressure applications, plasticization
of the membrane by carbon dioxide or hydrocarbons may reduce the membrane
performance. Identification and development of new materials that can resist
plasticization would be valuable.
3.7.4 Vapor-Gas Separations
Separation of water vapor from natural gas and other streams, and
separation of organic vapors from air and other streams, are both areas of
current and future opportunity for membranes.
Dehydration of natural gases occurs spontaneously during carbon dioxide

removal, because water vapor has a high intrinsic permeability and the
downstream mole fraction tends to be maintained at a low level by the high
downstream mole fraction of carbon dioxide. In this case, it is easy to meet the
Gas Separation 219
0.014% limit on allowable water in the residue gasI In the absence of carbon
dioxide, a tendency exists for downstream water vapor concentration polarization
to occur. Concentration polarization can be eliminated by allowing sufficient
hydrocarbon permeation, but this leads to undesirable product loss, so the use of
triethylene glycol solvent drying is generally preferred for relatively pure natural
gases without carbon dioxide.17
Small-scale dehydration of air is possible using a clever approach developed

by Monsanto in their Cactus@ system. The tendency for undesirable reductions in
driving force due to concentration polarization at the downstream membrane
surface is eliminated by allowing a small nonselective bypass of air through
uncaulked defects in the otherwise selective membrane skin. The system allows
economical production of low dew point (-IO’C) air without excessive losses due
to transmembrane air flows, and it provides a trouble-free source of dry air for
many small laboratory and instrumentation applications.
Removal of volatile organics, especially chlorinated and chlorofluorinated

hydrocarbons from air have attractive ecological as well as economic driving
forces. Spiral-wound membrane systems for this application are marketed by
Membrane Technology and Research, Inc. GKSS has worked in the same area,
using a plate-and-frame unit that minimizes vacuum losses in the permeate
channels. The system designs currently proposed for the application require
careful module design and pump selection.
3.7.5 Nitrogen-Hydrocarbon Separations
This separation need arises due to the presence of small amounts (515%) of
nitrogen impurities in some natural gas reservoirs. Also, the use of nitrogen as a
displacement fluid in enhanced oil recovery operations has been suggested. The
nitrogen/methane separation is a very difficult one to achieve using membranes.
For rubbery membrane materials, the higher condensability of methane favors it as
the permeate material. Even in glassy materials, the solubility selectivity
generally overcomes the slight mobility selectivity favoring the more compact
nitrogen and gives an overall selectivity near unity. Some glassy polymers with
highly restricted motions have given selectivities as high as three in favor of
nitrogen. The attractive aspect of the more selective glassy polymers is their
tendency to pass nitrogen over methane, so it is not necessary to lose
pressurization of the product gas. Such selectivities, however, are still too low
for commercial viability.
The possibility of incorporating nitrogen complexing groups to promote

nitrogen solubility selectivity has not been pursued as a means of complementing
the already favorable mobility selectivity for this pair of gases in highly rigid
glassy polymers. In fact, a serious potential difficulty exists in this case unless
the complexing interactions are very weak. If the complexing is not weak, the
diffusion coefficient of the nitrogen may be suppressed, thereby undermining the
nitrogen/methane mobility selectivity at the expense of the improved solubility
selectivity.
3.7.6 Helium Separations
Helium is a very small molecule and, in the case of hydrogen, a number of

high-selectivity and good-productivity membranes already exist. For example,
helium can readily be recovered from diving gases by means of a membrane unit.
Helium recovery from natural gases is also potentially possible, although the low
helium concentration (x1-I 5%) would favor a multistage unit.
3.8 ENERGY BASICS
Idealized and actual gas separation membrane processes are very different.
Figure 3-l 1 shows an idealized, reversible system. Two perfectly selective
membranes allow permeation of oxygen and nitrogen compared to the partial
pressures in the feed. Only reversible work is performed to raise the exit
pressure back to one atmosphere for the two pure-component streams.
Clearly, such a reversible process with differential driving forces would

require an infinite amount of membrane surface area. Actual membrane processes,
using finite membrane area and finite partial pressure driving forces require more
energy consumption than the ideal reversible amount. Nevertheless, some useful
statistics about membrane systems can be derived by knowing the theoretical
minimum energy requirement. The calculation below is for air at one atmosphere
and ZO’C.
For an open isothermal, steady state process, the energy balance, in terms
of the Gibbs free energy, G, of the inlet and outlet streams can be written:”
G out - Gin = $AW - $TA~, (1)
where AW is the net work and AS,, is the entropy change associated with the
process. For a reversible system, AS,, = 0 , so total work equals the sum of
the work done on the nitrogen and oxygen permeate streams:
G out - ‘4” = IAW, + AW, = W
Since the Gibbs free energy is an extensive property that is determined by the
number of moles (n) and the chemical potential (II) of the various streams:
G out = bN2 pN + n02 po2)P + cnN2 pN + not C(02)R 3 (3)

2 2
where P and R denote the permeate and residue streams, and
Gin = (n pN + n 02 ro,)p (4)

N2 2
where F denotes the feedstream.
0.79 atm pure N2
perfectly N2 selective membrane
Air @ 20°C
1 atm b
1 ton 02
3.76 ton N2
perfectly O2 selective membrane
Figure 3-I 1. Representation of an idealized isothermal reversible membrane

separator for producing pure oxygen and pure nitrogen at 1 atm
from a feed of air at 2o’C.
For an ideal gas, we have:
pi = &‘+ RT In pi = &‘+ RTIn yi + RTln p ; (5)
p = 1 atmosphere in all cases here, so
pi = @ + RT In yi (6)
Associated with one ton of oxygen, there are:
(2000/32)x454 gmol oxygen = 28,375 gmol oxygen
and
(0.79/0.21) 28,375 gmol nitrogen = 106,774 gmol nitrogen.
Substituting into equation (2). the n &’ terms cancel each other, SO:
W = 43.8 kWh/ton air separated
The individual work requirements to produce the 0.79/0.21 = 3.76 tons of pure
nitrogen, W,, and the I ton of pure oxygen, W,, respectively are:
W, = 17.14 kWh (or 4.56 kWh/ton nitrogen)

W, = 30.16 kWh per ton oxygen
For oxygen enrichment applications. W, is the pertinent value to consider.

Ward et al. suggest the use of the concept of “tons of equivalent pure oxygen”
oxygen (EPO,) to allow comparison of membrane and other approaches for
producing oxygen enriched air streams.” “Equivalent pure oxygen” is the amount
of pure oxygen needed to be blended with air to make a particular mixture of
oxygen-enriched air. For example, 100 moles of 35% oxygen enriched air can be
made by blending 17.7 mole of pure oxygen with 82.3 moles of air. The resulting
mixture, therefore, contains 17.7 moles of equivalent pure oxygen (EPO,).
To produce 10 tons of available oxygen in a 35% oxygen-enriched air,

therefore, IO/O.35 = 28.57 tons of air are delivered. Of the 10 tons of oxygen
in the air, 4.94 tons are supplied as EPO,. If this oxygen is obtained from a
reversible separation like that described above, an ideal energy cost of 30.16 x
4.94 kWh = 149 kWh for the 10 tons of available oxygen would result. The
actual energy cost to produce EPO, by standard cryogenic plants ranges from
275-375 kWh/ton.47 Assuming a midrange value of 325 kWh/ton EPO, gives the
1600 estimate shown in Table 3-5 for comparison to the ideal 149 kWh value.
PSA processes have been estimated to operate at 400 kWh/ton EPO,, thereby
yielding the energy cost of 1976 kWh indicated in the table.
Gas Separation 223
A/G Technology has estimated the cost to produce IO tons per day available
oxygen in 6 35% oxygen enriched air stream using their ethyl cellulose membranes
as opposed to a standard PSA approach (see Table 3-6).‘8 If operated in a
pressurized mode rather than with a vacuum downstream, the power costs for the
membrane unit approach those of the PSA unit. The vacuum mode of
operation is more energy efficient, because only the permeate must be
compressed as opposed to the entire feed in “pressurized” mode. Estimated
power costs for the membrane system were S86 for the membrane system and
$131 for the PSA system, suggesting that the membrane process is somewhat
more energy efficient than the PSA process in this case. If the 2,000 kWh
energy value noted in Table 3-5 is used with the $131 cost of power quoted by
A/G Technology for the PSA case, they indicate a reasonable energy cost of
6.6C/kWh at the time of the 1985 analysis. This number allows estimation of
the 1300 kWh value shown in Table 3-5 based on the $86 value quoted by A/G
Technology for production of IO tons of available oxygen using their membrane
system.
For all of the actual separation cases, the energy cost is well above the
theoretical minimum value of 149 kWh, because practical processes must be carried
out at finite rates and are less energy efficient than those carried out
reversibly. Further productivity increases in membranes may provide the
opportunity to reduce the driving forces required to achieve economical production
rates, thereby even further improving their position relative to the other
separation processes in terms of energy efficiency. As noted previously,
however, energy costs are only one aspect of the strong attraction of membrane
systems. Their convenience, flexibility and overall operating costs are even more
important. This point is made in Table 3-6 from the article by A/G Technology
in comparing the total cost to deliver ten tons per day of available oxygen as a
35% oxygen-enriched stream.
‘Vacuum Swing Adsorption” (VSA), which operates between essentially

atmospheric pressure and a vacuum, rather than between an elevated presstire
and one atmosphere, allows energy savings by requiring compression of only the
oxygen-enriched product gas rather than the entire feed. In this case, the energy
efficiency of the adsorption and membrane operations are probably similar.
Table 3-5. Comparison of energy consumption by different processes for tons

per day of available oxygen in a 35% oxygen enriched air stream. If
the adsorption process is run as VSA (vacuum swing adsorption) it
is likelv that it will be more similar in energy use to the membrane
than isthe case for PSA.”
Type of Process Energy Requirement

(kw-llr/ton EP@ )
Revenible isothermal 149

(blended pun 02 with air to make 35%)
Cryogenic production of 99.5% 1600

(blended with air to make 35%)
Prcssurc swing adsorption production 2ooo

90% 02 (blended with air to make 35%)
Typical u-rent generation membrane for 1300

direct production of 35% 02 containing
air with no post blending
Table 3-6. Cost comparisons for 35% oxygen-enriched air (10 tons/day available
oxygen).48
A/G Technology
Membranes PSA
Installed Capital Cost 288.00 552.00
(Thousands of Dollars)
Expenses ($/day)
Membrane Replacement 38.00 -
Power 86.00 131.00
Capital Charges 105.00 202.00
Depreciation 33.00 63.00
Other 18.00 27.00
Total (Dollars/Day) 280.00 423.00
Total (Dollars/Ton) 28.00 42.00
Gas Separation 225
3.9 ECONOMICS
Evolution of more advanced materials such as the polymers indicated in

Figure 3-5 and Table 3-2 will eventually reduce the need for multistage and
recycle approaches to achieving required product specifications for one or both
the permeate and nonpermeate streams. In general, however, selection of
optimized approaches for using membranes in a given application will remain
complex and is best done with the aid of detailed computer programs to allow
complete analysis of numerous candidate process layouts. The application
examples in Table 3-3 and the earlier case study results by Shell referred to in
the section on the historical development of the field illustrate the versatility of
gas separation membrane technology.50 It is this very versatility that places
membranes in competition with a diverse array of competitive technologies.
Spillman has presented a useful analysis emphasizing the importance of

objective consideration of economic factors when evaluating competitive
technologies such as membranes, PSA, cryogenic, chemical and physical solvent
candidates for a given application.” He notes that issues such as the pressure
level at which the products can be delivered, e.g., in nitrogen inverting or
hydrogen recovery, are sometimes deciding factors of equal or greater importance
to the absolute recovery of the desired component. Also, as illustrated by the
use of membranes to reduce carbon dioxide concentrations prior to feeding to a
cryogenic plant for enhanced oil recovery operations, hybrid approaches involving
membranes with another separation process may sometimes be the best strategy.
3.10 SUPPLIERS
Table 3-7 summarizes the principal suppliers of commercial-scale equipment

for gas-separation applications. As can be seen, the only company offering
equipment in all application areas at present is Monsanto. Even the other large
players have focused on two or three specific topics, and the field is made up of
a large number of smaller companies, or larger companies with relatively small
membrane groups, that are attacking the niche markets.
For the larger companies, it appears that joint ventures with companies that
supply gas by other means will continue to flourish. An across-the-board
capability to handle gas-separation problems regardless of the volume, pressure or
composition of the feed will obviously put a supplier in a strong position with
regard to the competition. Furthermore, the optimum solution to any particular
gas separation problem may well involve a hybrid process incorporating two or
more individual technologies. such as membranes plus pressure swing adsorption
for high-purity separations. or membranes plus compression-condensation for the
most cost-effective pollution control approach.
As the industry becomes more mature. it may also become increasingly

standardized as companies compete with one another for the replacement module
market.
Table 3-7. Commercial-scale membrane suppliers and gas separation module

types for each supplier.
Company CO, HZ 0, N, Other* Module Type**
A/G Technology (AVIR) X HF

Air Products (Separex) X SW
Asahi Glass (HISEP) X
Cynara (Dow) HF
Dow (Generon) X HF
DuPont X HF
GKSS X PF
Grace Membrane Systems X SW
Membrane Tech. & Resch. X SW
Monsanto X X HF
Nippon Kokan X
Osaka Gas X
Oxygen Enrichement Co. X PF

Perma Pure X HF
Techmashexport PF
Teijin Ltd.
Toyobo HF
Ube Industries X HF
Union Carbide (Linde) X HF
UOP/Union Carbide
l
Included solvent vapor recovery, dehumidification and/or helium recovery.
l * SW = spiral wound, HF = hollow fiber, PF = plate and frame.
Gas Separation 227
For the small companies, with limited resources and expertise to undertake
large-scale plant engineering, the trend to go after niche markets will continue.
The gas separation field is young enough and open enough to accommodate many
successful small companies, at least through the short to medium range future.
A/G Technology is very effectively marketing its ethylcellulose hollow-fiber
modules. Membrane Technology and Research, Inc. has announced membranes for
removing organic vapors from air, and for removing higher hydrocarbons from
natural gas, which may place it in a favorable position.
3.11.1 Research Centers and Groups
As indicated earlier, industrial users often rely upon the suppliers of systems
for innovation in the details of membrane materials, membrane structure and
system design. This trend is promoted by suppliers who have found it attractive
to either sell self-contained systems or, in some cases, to provide gas as a utility
to a customer who merely rents the system. Although complex computer programs
are usually needed to accurately assess the true potential of membranes compared
to competitive technologies, small users can still promote innovation in the
field. By proposing novel applications and challenging the existing state of the
art in membranes, users can help establish goals to drive the fundamental material
science and applied engineering research being supported in academia by both
government and industry. Technical meetings and conferences are valuable forums
for the interchange of such information and for keeping users apprised of the
rapidly expanding capabilities of gas separation membrane systems.
The field of gas separation membrane research is extremely active at the

present time. Tables 3-8a and 3-8b list locations where sufficient concentration
of effort exists to indicate that a significant contribution to the field is
occurring. Major groups are identified by an asterisk. Table 3-8a covers
industrial research centers; Table 3-8b covers academia. For the larger industrial
groups, over ten professional scientists and engineers are typically involved in the
indicated activities. Similarly, in academia, five to ten graduate students may be
involved in activities with multiple faculty at a given location. Activities in
facilitated transport, discussed in Chapter Four, are not included.
The entries in the “activities” column are based on publicly available

information; where activities are not indicated, they may exist but have not been
publicized. Most of the commercial suppliers listed earlier in Table 3-7, are
expected to have capabilities in at least the last three categories, and have been
so indicated in Tables 3-8a and 3-8b. Where a joint venture is in place, the
name of the principal partner with the major strength in gas membranes is given.
Multiple addresses are given in cases where specific activities are carried out at
different locations, but some overall coordination of effort exists. Often, but not
always, the second partner in a joint venture provides market access and
knowledge, rather than membrane expertise.
Table 3-8a. Industrial Research Centers and Groups Contributing to the

Development of Membrane-Based Gas Separation. NM refers to
organizations conducting research into new materials; F refers to
membrane formation; MDM to module design and modeling; ST to
system testing.
Activities
Organixalion & Localion NM F MDM ST
x
x x
x x
x x
x x
Gas Separation 229
Table 3-gb. Academic, Government and Private Research Institutes Contributing

to the Development of Membrane-Based Gas Separation. NM
refers to organizations conducting research into new materials; F
refers to membrane formation; MDM to module design and
modeling; ST to system testing.
Activities
Organization & Location MM F MDM ST
X X
X X
X X
X
3.11.2 Support of Membrane-Based Gas Separation
3.11.2.1 United States
To date, the bulk of the funding for gas separation research has come from
industrial sources that see long-term opportunities in this area. The major
companies identified in the preceding table have made multimillion dollar
investments to develop technology to promote their position in this rapidly
growing field. In addition to the industrial support, various U.S. private and
governmental organizations have provided limited support. Included in the above
funding sources are the National Science Foundation, the Department of Energy
and the Small Business Innovation Research (SBIR) program within these
organizations. Except for scattered fundamental grants, and highly
mission-oriented projects funded by Defense Advanced Research Projects Agency
(DARPA). the Department of Defense has not helped significantly to promote gas
separations using membranes. This is surprising, especially given the advantages
provided to a mobile military unit by potential fuel savings if oxygen-enriched
air is used in an advanced internal combustion engine. Moreover, weight and
space savings associated with the possibility of on-board generation of inert
atmospheres for aircraft and ship fuel tanks is an area that would seem to merit
much more attention by the military than it has received.
3.11.2.2 Foreign
Foreign industrial activities in the membrane-based gas separation area are

significant, with those in Europe and Japan being the most intense. As in the
United States, university and institute funding is much less apparent than in the
industrial arena.
The worldwide market for industrial or home uses of membranes wiil

ultimately be determined by the cost-to-service ratio achievable with these
products. Originally unavailable markets, such as nitrogen inerting, have
recently become available as the cost-to-service ratio decreases. The membrane
module, and sometimes the membrane material itself, are critical elements in
determining this ratio. Other system components and their complexities must be
engineered to accommodate for membrane deficiencies, thereby adding costs. First
generation products have achieved commercial success with essentially
non-tailored materials, but longer term success will require more attention to the
membrane material and membrane structure itself.
3.12.1 Industrial Opportunities
In the near future, industrial opportunities will fit into one of the existing
categories listed in Table 3-3. Existing products, however, are marginally useful
or unacceptable for some of these applications. The one factor that would do
most to speed commercial success would be improvements in membrane materials
and membranes. A listing of application areas in terms of anticipated future
importance is given in Table 3-9, with brief explanatory comments. The
“Prospects” column is intended to reflect the current assessment of the likelihood
Gas Separation 231
that problems preventing the implementation of membranes for the cited

application area have been. or will be resolved within the next S-20 year period.
The “Importance” column represents the consensus of opinions from the working
group meeting at the 1989 North American Membrane Society (NAMS) meeting. A
high rating on a scale from I-10 indicates that the application is likely to be
significantly important to the United States in the energy conservation field. The
“Comments” column mentions technical, economic or related constraints and
factors that may impact the implementation of the process.
3.12.2 Domestic Opportunities
In overall dollar value, applications of membrane-based gas separation will

likely be predominately in the industrial sphere, at least for the next S-10 years.
However, the simplicity of the membrane process presents some attractive
possibilities for new consumer markets for an increasingly environmentally
sensitive public. Potentially, industrial burner and furnace designs based on
oxygen-enriched feeds may find their way into the home-heating market as
energy costs rise and concern for carbon dioxide emissions increases. The
compactness and minimal service needs for such systems are attractive. Similarly,
less reliance on refrigeration to reduce spoilage of certain types of foods and
other oxygen-sensitive articles in shipment may ultimately find an analog in more
energy-efficient home storage. Development of practical, small-scale digesters for
generation of methane from wastes in rural or isolated installations appears
technically feasible, but dependent upon simple, efficient membrane system. Such
developments are likely to require time to be achieved, but on a society-wide
basis, they amount to very significant developments. Moreover, such potential
developments add an additional dimension and driving force for promoting
industrial scale membranes which will be the eventual source of technology for
such consumer products.
3. I3 RESEARCH OPPORTUNITIES
Potential areas of research emphasis are described in Table 3-10. The

topics are ranked semiquantitatively according to their prospects for realization as
well as their importance in addressing the application areas in Table 3-9. The
rankings are based on an integration of all of the preceding discussions of the
various sections as well as discussions of the panel convened at the 1989 NAMS
meeting.
3.13.1 Ultrathin Defect-Free Membrane Formation Process

An attractive. achievable opportunity would entail the development of a
convenient general process to produce truly defect-free, asymmetric membranes
with ultrathin skins less than 50oA thick. The availability of such membranes,
especially in hollow-fiber form, would provide major improvement in all gas
separation systems. Monsanto has a process for spinning ultrathin-skinned
standard polysulfone, but it is not readily applicable to other polymers. A
process that could handle more selective. existing commercial polymers, such as
polyetherimide (PEI), as well as new custom-made materials, would be a major
breakthrough. Moreover, such a process could be used to form membranes that
would serve as a basis for post-reaction modification of highly productive, low
selectivity materials (item #5 in Table 3-10).
Table 3-9. Future Directions for Membrane-Based Gas Separation
Application Area Prospects Importance Comments
I. N, enrichment WXll~l 10 Major m&t...share determined by cost- benefit position vs.

of air PSA &cryogenic. Higher selectivity and productivity uni!s
needed to cut compression costs and module size. resp.
2. Low level 0, excellent 8 Major market... similar considerations to Il. except possible
cNlchmcnt of air additional opportunities exist if one can separate 0, from a
compessed gas stream at high temperature (3CKK) for
cogenaation processes
3. Hydrogen separation eXCeUelIt 1 Large market. but diversely disnrbuted. Good materials exist
for super critical gases (CO. N,. & CH,) but not CO,. Higher
temperature module of interest.
4. Acid gases separation good 8 Sigticant market. but more plasticization- resistanr high
from hydrocarbons & H, selectivity materials needed to compete with cryogenrc in new
plants if goal is to use only membranes.
5. Helium recovery god 1 Workable systems exist. however, market is limited. PSA and
membranes appear competitive.
6. Hydrocarbon and chloro- fair 5 Silicone rubber works well. but system design needs a lot of
lluoro hydrocarbon work. Potential danger due to 0, concenrration in the
separation from air or permeate.
n,uogen
(continued)
Applicalion Area Prospects Importance Comments
7. An d”hydruion fair Small scale process for eliminating H,O conccmntion

polaruation using an internal recycle or an air sweep through
a con~olled porosity skin limils acceprable. but inefticient
for large scale.
Y. Natural gas dehydration fair [n absence of additional permeable impwlries like CO,. CH,
losses associated wilh H,O concentration polarization are LOO
high. A more CH, rqecting membrane mlghl work If water
permeability is not too low.
9. N: removal from poor Solubility selecuviry and mobility selcct~vity oppose each
narural 935 olher. and both are low. Nitrogen complexmg agenrs char
may allow increasing the solubility of N, relauve to CH,
may be useful if lhey do not destroy Ihe mobility selecuv~~y
associated with rigid glasses like Ihe lmldes in Table 3.
10. High level (90’%+) 0, Poor Much more selectwe membrane materials wll be needed to do
enrichment of air lhis with convenuonal polymers. It appears likely lhar
facilitated transport may be the melhod of choice if
membranes are to be used for this application.
1 I. Sclccwe medune poor Although size differences favor passage of methane.

removal from bulane condensibility favors heavier penetnn~~. and good selectivity
or edlane streams is not likely to be achievable al economical fluxes. Few data
I?. Acid gas removal from poor Difficult to do well with membranes. since concemraoon
flue gases driving force IS low. Only hope IS by facditated uanspon
membranes.
Table 3- 10. Research Topics of Future Interest for Membrane-Based Gas

Separation
Research Topic Importance Comments
10 Would ,lk.w b,o.d usage of dvmced malenals

-evaIbeacrifdmsinholk#wfitxrfonn
10 Expaimcnul muaids approach tbese intrinsic

(a-&lojmi productivity aandP numbers.bumabiLty10spolform
mlvmn
Cd
fP - 2-3 Burn for 0.1II thaninulwhinfcmnhubDmrcporwd.
cap&k of being famed with . Ag.,,,, more valuable in hollow fiber form.
SOOA defocr-free skin 10 p~vxic
oxygen P/l > 1x104 cc(STPJ I
s.zc cn+cm Hg
38. Membrane material wirh high 7 Will become mare important as he acid gas
selectivity for C(Y wd 4.5 p.nirl -sure in rhe fad from EOR prolea
sepuluon from CH, (a ~45) inausa.
and H, (a >20) II high CO, and
ys panill plurures.
3b. Improved module designs.
maunlls for mcduk
amwucocm II high CO, and
yS P=od PIurw
4. Dcvclopmcnl of a umvtialt 10 Sune a, #I. b,u even more valuable ifpess~ble

and gmerally applicable method since only a small amount of valuable advanced
fa forming comp0sll.e bouow maerials need u) be used if ady the sclscdng
tibm vld, ulcahin (doA) is CanpcMd of sufh IMlailL.
derm-bet skins
5. Renxivc matmcnu la incrurrng 6 Atmctivs if it is gmaally Ippliubls. Both

lhe sekcuvlty of. pizformut fllMxiM&M and pholocbemul crosslinkmg
uluwhm sclativc skin witboot have been dmKmsuNcd cm dnlse fti
cx.esslve flux losses and on I rclrtively thick (1 pm) composits
mrmbrtns.buno~onthin<l~Amembnxs
6s. Lxiusui.l survey 10 delime 5 Sboold be done bcfcm large cxpmdiwcs arc
needs fasxollc high mIdelo&veloph&hLcmpcnonmalmals.
tcmpennuc organic and since Ihe current market is unclear.
inagmuc memhanu High tanperaturc 0, s&ctive membrane for
6b. High rcmperarure membranes ? cogensntioo and membnne pucmn may be
impaunr
7. Retime & quanofy gudcliwa uxd 6 Much progrus hu tea ma&. bur stcdy
analytical methods u) soumline longterm building of this cap&lily pmwdcs *
YlcfUml of polymers fa high good basis for opaing pomaurl new markerr
cff~imcy -oats and pnvsntig displunnem by foreign p?cdocu
8. Highly oxygen selative

membrane (a-1%15 forOp,)
v&b good sldilicy md a 0,
P/f10_5- lrIwcc(%rP~
KS c.m%.m Hg
9. PbyslCal vc.mlalo (mupiasu- Unlikdy u) be ruble ova cxlcndcd periods

Clzao‘m) u) -lbcaaof ifpuau~yinti(<~OOOAselcctmg
a prfomled IdIn thin xlczuvc layer). uni dclsraiav u) mahuuul pmpenlcs
skin without cx-ve if pxsml lhrmghou whole manhanc.
miuclials m P/f
IO. Cancal- of poducD

frmn dllulc s-
Gas Separation 235
There seems to be nothing inherently preventing the achievement of this

goal, and lab-scale demonstrations with flat asymmetric membranes have already
been achieved. Elimination of defects by simple coatings is precluded by a
Monsanto patent, so for the field to advance in general for all suppliers,
intrinsically perfect skins or more elaborate post-treatments not covered by the
Monsanto “caulking” process to heal defects are needed. Such a project should
also involve an improved understanding of membrane formation fundamentals and
would require the implementation of more advanced spinning technology than
exists in the field at the present time.
3.13.2 Highly Oxygen-Selective Materials
Another achievable project of specific applicability involves the discovery of

a stable, intermediate permeability membrane with an oxygen/nitrogen selectivity
of 7-10. The membrane should be able to be formed into sufficiently thin skins
to provide transmembrane fluxes on the order of ~OOXIO-~ cmS(STP)/cmascmHg
in a high efficiency hollow-fiber form. Such a membrane would place strong
pressure on PSA, even for the 99% nitrogen market. Although liquid membranes
achieve this selectivity, their flux is not good and stability is poor. Assuming
that formation of ultrathin, defect-free skins can be achieved, polymers with
intrinsic properties almost good enough to achieve the above goals are known.
Therefore, it is likely that this goal will be achieved if sufficient resources are
focused on it.
3.13.3 Polymers. Membranes and Modules for Demanding Service
This item relates to the development of more robust membrane materials and
modules capable of operation in extreme environments. This objective is likely to
be important in enhanced oil recovery processes in the presence of strongly
interacting high-pressure carbon dioxide feeds. For this application microporous
ceramic or glassy membranes may have some real potential. It has been noted
that surface flow of the more condensable component seems to provide
selectivity of that component relative to methane. This is a surprising finding
that needs more careful scrutiny.
3.13.4 Improved Composite Membrane Formation Process

The development of a process for producing ultrathin, defect-free composite
membranes is important. If such a process could be developed in a generally
applicable form for use with most advanced materials, a high-productivity and
high-selectivity coating could be applied on an inexpensive support. It would
then be feasible to use even expensive materials to produce high performance
modules without sacrificing economic attractiveness.
This is a difficult problem. but one well worth pursuing. Work done at
Union Carbide, IPRI (Japan), Air Products and MTR with thin (<1,50OA) coatings
on fibers or flat membranes suggests the feasibility of the approach. The
surface-to-volume advantages of hollow-fibers compared to spiral units make
them the most attractive as a composite support; however, formation of a perfect
ultrathin composite hollow-fiber membrane is a formidable objective. Cabasso
suggests the use of an intermediate layer on which to place the selective layer
to help channel permeate flow to the pores of the underlying support.” The
issue of optimum composite membrane structure deserves considerable attention.
3.13.5 Reactive Surface Modifications
Surface modification by plasma, photochemical reactions, fluorination or

other reactive treatments is an attractive way to produce “asymmetric,
asymmetric” membranes. The procedure, in principle, allows treatment of just
the outer layer of otherwise relatively thick, but defect-free asymmetric
membrane produced by conventional technology. Surface modification may be
applicable to even commercially available polymers with unacceptably low intrinsic
selectivities. but with relatively high intrinsic permeabilities.
Laboratory demonstrations of various manifestations of the approach exist

(plasma, bromination, fluorination, photochemical), but a commitment is required
by industry to scale up the technology, and the expense and problems are likely
to be significant. Monsanto’s attempts to surface-treat polyphenylene oxide
(PPO) was abandoned because of technical difficulties. Nevertheless, there are no
inherent barriers preventing the development of surface-modification technology.
3.13.6 High-Temperature Resistant Membranes
The development of high-temperature resistant membranes is a topic of

considerable scientific interest. but it is not clear whether the availability of such
materials would really make a significant impact on energy consumption. A study
to look specifically at the energy implications of temperature-resistant membranes
and modules would be useful. As for Item 3, ceramic, carbon or glass membranes
may have application here if an economic and energy study justify costs.
Examples of such applications could include hydrogen separations where it

may ultimately be possible to link reactors and membranes to influence the
direction of reactions by removal or introduction of hydrogen to force the
products to a favorable form. Moreover, the possibility of high-temperature
oxygen-selective membranes for use in cogeneration processes has been
suggested.
3.13.7 Refinement of Guidelines and Analytical Methods for Membrane Material

Selection
A great deal of progress has been made in recent years in developing

optimization strategies for improving the trade-off between permeability and
selectivity for members within a given class of polymers. Little general
understanding exists. however, about general principles that would allow one to
decide on which family of polymers to perform such optimization procedures.
Moreover, the current approaches are very empirical and experimentally
intensive. As an investment in the long-term viability of the United States
membrane industry, steady improvements in the understanding of fundamental
factors governing the detailed permeation process need to be made. An example
of such a study is the evaluation of the free volume cavities available for
diffusion of molecular penetrants through glassy polymers by applying a Monte
Carlo technique to a model of the energy states of polymer segments within a
closed cube.s6
Gas Separation 237
3.13.8 Extremely Highly Oxygen-Selective Membrane Materials
A potentially very significant, but difficult to achieve, opportunity would be

offered by a stable, highly oxygen-selective, highly productive membrane. If an
oxygen/nitrogen selectivity of IZ- 15 could be achieved with a flux of
50-100~10‘~ cmS(STP)/cmZ+cmHg, membranes would have ready access to even
intermediate (50-60% oxygen) markets. Such a membrane would also potentially
displace PSA for all nitrogen production and all oxygen production except the
highest purity markets (~90%). Although liquid membranes offer this selectivity,
their flux and stability are poor. A more likely hope in this respect would be
the Rotem carbon fibers; however, they may be impractical due to handling
problems as well as potential “plugging” due to the presence of condensable
components competing for diffusion pathways.
3.13.9 Physical Surface Modification by Antiplasticixation
The use of antiplasticizing agents loaded into the selective layers of

membranes seems to be unreasonable for long term operation. Even if the agent
has an incredibly low vapor pressure, the large quantities of gas passing through
the membrane and the minuscule amount of agent present should tend to cause
unacceptably rapid removal even at ambient temperatures. Under elevated
temperature conditions the situation is even less attractive and not worthy of
support.
3.13.10 Concentration of Products from Dilute Streams
The ability to selectively collect dilute components as a concentrated

permeate product is typically very poor for most applications using conventional
membranes. An exception to this rule is, however, given by the case of .helium
separation from natural gas. Helium may be present at concentrations below 5%,
but the high selectivity of current glassy membranes for the helium/methane
system makes the separation relatively straightforward.
The larger and more important problem of removing acid gases from flue
gases illustrates the more commonly encountered situation involving dilute solution
processing. The limited selectivity relative to nitrogen that is available for
gases such as sulfur dioxide and carbon dioxide that are present in dilute
concentrations in flue gases precludes economical removal using conventional
membranes. Facilitated transport membranes, on the other hand, appear to be
appropriate for this type of application, but are typically not well suited for the
higher concentration ranges due to saturation of carrier capacity. Therefore,
facilitated transport and conventional membranes can be viewed as complementary
to each other in terms of covering a broad spectrum of feed compositions.
REFERENCES
I. Koros, W.J. and Chern, R.T., “Separation of Gaseous Mixtures Using Polymer
Membranes”, Chap.20 in Handbook of -on Process Technology, R. W.
Rousseau (ed), John Wiley and Sons, New York, (1987).
2. Koresh, J. and Soffer, A., SeDn. and Tecu 9 973 (1987).
3. Felder, R.M. and G.S. Huvard, Methods of Experimental Physics, 16c, 315
(1980).
4. Breck, D.W., aolite Molecular Sieves, John Wiley and Sons, New York,
(1974).
5. Kim, T.H., Koros, W.J. and Husk, G.R., an. Sci.. and Techn.. 23, 1611
(1988).
6. Billmeyer, F.W., Textbook of Polvmer Science, 2nd Edition, John Wiley and
Sons, New York (1971).
7. Koros. W.J., J. Polvm. Sci.. Polvm. Phvs. Ed.. 23, 1611 (1985).
8. Chern. R.T., Koros. W.J., Sanders, E.S., Chen, S.H. and Hopfenberg, H.B., in
Industrial Gas Separations, ACS Svmoos’um Series 221 . ed. by T.E. Whyte.
CM. Yon, and E.H. Wagener, pg. 47 , 1i83.
9. O’Brien, K.C.. Koros, W.J., Barbari, T.A., and Sanders, ES., ,I. Membr. Sci.,
2, 229 ( 1986).
10. Stannett. V.T.. Chapter 2 in Diffusion in Polvmerg, ed. by J. Crank and G.

Park, Academic Press (1968).
II. Gantzel, P.K. and Merten, U., &EC Proc.. Des.. Dev.. 8, 84 (1969).
12. H.H. Hoehn. “Aromatic Polyamide Membranes”, in Materials Science of

Synthetic Membranes, ed. D. R. Lloyd, ACS Svmposium Series 269, pg. 81 ,
1985.
13. Pinnau. I. and Koros, W.J.. ” Integrally-Skinned-Asymmetric Gas Separation

Membranes Based on Bisphenol-A Polymers”, paper presented at the
International Symposium in Suzdal, USSR, Feb., 1989.
14. Henis, J.M.S. and Tripodi. M.K., &pn. Sci.. and Techn.. 15, 1059 (1980).
IS. Cadotte J., “Composite Reverse Osmosis Membrane”, in Materials Science of

Synthetic Membranes. ed. D. R. Lloyd, ACS Svmoosium Series 269, pg. 273 ,
1985.
16. Matson, S.L., Lopez, J. and Quinn, J.A., Chem. Enar. Sci.. 38, 503 (1983).
Gas Separation 239
17. Spillman, R.W.. “Economics of Gas Separation by Membranes,” Chem. Ena.

Pron.. u. 41 (1989).
IS. Mitchell, J.K., Philam J. Med. sci.. 9 p. 36 (1831).
19. Graham, T., Philos. Mak,_lZ. 401 (1866).
20. Stern, S.A., Sinclair, T.F., Gareis, P.J., Vahldieck, N.P. and Mohr, P.H.,
“Performance of a Variable Volume Permeation Cell.” I. &. E.C.. 57, 49
(1965).
21. Loeb, S.. and Sourirajan, S.. “Sea Water Demineralization by Means of an
Osmotic Membrane.” Adv. Chem. Ser._& I17 (1962).
22. Ward, W.J.. Browall, W.R.. and R.M. Salemme, “Illtrathin

Silicone/Polycarbonate Membranes for Gas Separation Processes,” J. Membr,
sci.., 99 (1976).
23. Ward, W.J., U. S. Patent 4,279,855, July (1981).
24. U. Merten and P.K. Cantzel, U.S. Patent 3,415,038, December (1968).
25. Gardner, R.J., Crane, R.A. and Hannan, J.F., “Hollow Fiber Permeator for
Separating Gases,” Chem. Enn. Pron.. 71, 76 (1977).
26. Hoehn, H.H. and J.W. Richter, U.S. Patent Re. 30,351, July (1980).
27. Henis, J.M.S. and Tripodi, M.K., U.S. Patent 4,230,463, October (1980).
28. Hunter, J.B. et al., U.S. Patents 2,961,061 November (1960), and 3,254,956
June (1966).
29. Pye, D.G., Hoehn, H.H., and Panar, M., “Measurement of Gas Permeability of
Polymers, I. Permeation in Constant Volume/Variable Pressure Apparatus,” 1,
ADDI. Polvm. Sci.. ZQ, 287 (1976). Pye, D.G., Hoehn, H.H., and Panar, M.,
“Measurement of Gas Permeability of Polymers. II. Apparatus for
Determination of Mixed Gases and Vapors,” J. Ao~lv. Polvm. Sci. 20, 1921
(1976).
30. Pilate, L., Litz, L., Hargitay, B., Osborne, R.. C.. Farnham, A., Kawakami, J.,
Fritze, P., and McGrath. J., “Polymers for- Permselective Gas Separations,”
ACS Preprints, 16(2), 42 (1975).
31. Koros, W.J. and Hellums, M.W., “Gas Separation Membrane Material Selection
Criteria: Differences for Weakly and Strongly Interacting Feed Components”,
paper presented at the Fifth International Conference on Fluid Properties
and Phase Equilibria, May (1989).
32. Kim, T.H.. Koros, W.J., and Husk. G.R.. “Temperature Effects on Gas
Permselection Properties in Hexafluoro Aromatic Polyimides”, J. Membr. Sci.,
in press.
33. Shinnar, R., Thermodynamic Analysis in Chemical Process and Reactor

Design,” them. Enn., 2303 (1988).
34. Chu, W., “Hydrogen purifiers: Reliable technology in demanding times,”

American * Feb. (1989).
35. Peinemann, K.V., U.S. Patent 4.746.333, May (1988). and Peinemann, K.V.,
Pinnau, I. and Wind, J., “Polyethersulfone and Polyetherimide Membranes
with High Selectivities for Gas Separation”, International Workshop on
Membranes for Gas and Vapor Separation, Mar. 7-9, 1988, Qiryat Anivim,
Israel.
36. Hayes, R.A., U.S. Patent 4.717.393, January (1988).
37. Mohr, J. M., and Paul, D. R., “Surface Fluorination of Composite Membranes
for Gas Separations”, Meeting of North American Membrane Society, Austin,
Texas, May, 1989.
38. Langsam. M., U.S. Patent 4,657,564, April (1987).
39. Hopfenberg. H.B., Witchey, L.C., and Chern, R.T., “Sorption and Transport of
Organic Vapors in Poly( I -trimethylsilyl- I qropyne)“, paper presented at the
1987 Annual AIChE meeting, New York, November (1987).
40. Brooks, et al., U.S. Patent 4,575,385, March (1986).
41. Murphy, M.K., Beaver, E.R.. and Rice , A.W., “Post Treatment of Graded
Asymmetric Membranes for Gas Application”, paper presented at the AIChE
1989 Spring National Meeting in Houston, Texas, April (1989).
42. Gollan, A.. U.S. Patent 4,734,106, March (1988).
43. Burmaster, B.M. and Carter, D.C., “Increased Methanol Production Using
Prism Separators”, paper presented at AIChE Symposium, Houston, TX,
March (1983).
44. Schott, M.E., Houston, C.D., Glazer, J.L., and DiMartino, S.P., “Membrane
HZ/CO Ratio Adjustment,” paper presented at AIChE Symposium, Houston,
TX, April (1987).
45. Nakamura, A., and Minoru, H., “Novel Polyimide Membrane for Hydrogen
Separation,” them. Econ. and Ena. Rev.. 117, 41 (1985).
46. H. Ito and M. Watabe, “Oxygen Enriched Air Converstion System by

Membranes”, Maku 18 No. 5, 372 (1983).
47. Epperson, B.J.. and Burnett, L.J., “Development and Demonstration of a

Spiral-Wound Thin-Film Composite Membrane System for the Economical
Production of Oxygen-Enriched Air”, Final Report on Contract NO.
DE-ACOI-79CS40294, Report No. DOE/CS/40294.
Gas Separation 241
48. Gollan, A. and Kleper, M.H., “Membrane Based Air Separation,” AIChE SY~
Ser.250. 82, pp. 35, (1986).
49. Matson, S.L., Ward, W.J., Kimura, S.G., and Browall, W.R., “Membrane Oxygen
Enrichment, II. Economic Assessment,” L Membr. SCI.. 39, 79 (1986).
50. Youn. KC.. Blytas, G.C., and H. H. Wall. “Role of Membrane Technology in
the Recovery of Carbon Dioxide: Economics and Future Prospects”, paper
presented at Gas Processors Association Regional Meeting, Houston, TX,
Nov. (1983).
51. Schendel, R.L.. Mark, C.L., and Mak. J.Y., “Is Permeation Competitive,”
Hvdrocarbon Proced . 58 (1983).
52. Kumar, R., and Van Sloun, J.K.. “Purification by Adsorptive Separation,”
Chem. Enn. PrQg__& 34 (1989).
53. Denbigh, K., The Princioles of Chemical Eauilibrium. 3rd edition, Cambridge
University Press, New York, (1978).
54. Lundy, K. A. and Cabasso, I., “Analysis and Construction of Multilayer

Composite Membranes for the Separation of Gas Mixtures,” Ind. Ena. Chem,
Res.. 28, 742 (1989).
55. V.M. Shah, S.A. Stern and P.J. Ludovice, “Estimation of the Free Volume in
Polymers by Means of a Monte Carlo Technique,” Macromolecules. 22.4660 (1989).
4. Facilitated Transport
E.L. Cussler
Department of Chemical Engineering and Materials Science,
University of Minnesota
4.1 .l The Basic Process
Facilitated transport is a form of extraction carried out in a membrane. As

such, it is different from most of the other membrane processes described in this
report. Many of these, for example ultrafiltration, are alternative forms of
filtration. Others, especially gas separations and pervaporation, depend on
diffusion and solubility in thin polymer films. In contrast, facilitated transport
involves specific chemical reactions like those in extraction.’
Facilitated transport usually has four characteristics that make it different

from other membrane separations:
(1) It is highly selective.

(2) It reaches a maximum flux at high concentration differences.
(3) It can often concentrate as well as separate a given solute.
(4) It is easily poisoned.
Characteristics (2) and (3) are the most powerful evidence that facilitated
transport is occurring.
The way in which facilitated transport works is shown schematically in

Figure 4-l. The U-tube at the top of this figure contains two aqueous solutions
separated by a denser chloroform solution. The aqueous solution on the left
contains mixed alkali metal chlorides; the solution on the right is initially water.
The chloroform solution contains dibenzo- IS-crown-6, a macrocyclic polyether. An
additive like this polyether is called a “mobile carrier”.2 With time, salts in the
left-hand solution in Figure 4-1 dissolve in the chloroform and diffuse to the
right-hand solution. This dissolution is enhanced or “facilitated” by as much as a
million times by the mobile carrier. Moreover, the enhancement is selective: in
this case, the flux of potassium chloride is four thousand times greater than that
of lithium chloride.
242
Facilitated Transport 243
Feed solution stripping solution;

chloroform solution itlally pure water
Same chloroform
solution held In
membrane pores
Figure 4-1. Facilitated Transport as Extraction. The extraction system at the

top of the figure mimics the facilitated diffusion in the membrane at
the bottom of the figure.
The process at the top of Figure 4-1 is obviously an extraction. The

reduction from extraction to facilitated transport is made possible by reducing the
chloroform phase to a thin sheet, perhaps 30 pm thick, as suggested at the
bottom of Figure 4- 1. This thin sheet is stabilized by capillary forces within the
pores of a microporous hydrophobic polymer membrane. Celgard@ membranes,
made from microporous polypropylene are a common choice. As before, potassium
chloride is selectively extracted from the left-hand ‘feed’ solution into the
membrane; the mobile carrier greatly facilitates the salt’s diffusion, and the right
hand “stripping” solution removes the solutes from the membrane.
The example of facilitated transport illustrated in Figure 4-1 involves the

separation of metal ions with a carrier that selectively reacts with one ion and
not with the others in the solution. These carriers are widely used in solvent
extraction under the name liquid ion-exchange (LIX) reagents. Another type of
facilitated transport process uses carriers that will selectively react with and
transport one component of a gas mixture. One of the most widely studied gas
separation facilitated transport processes is the separation of oxygen from air.
Hemoglobin is a well known natural carrier for oxygen, and many other synthetic
carriers are also known.
In the past, most facilitated transport membranes used selective carriers

dissolved in an organic liquid solvent. In conventional forms of the process, the
carrier-solvent solution is immobilized in the pores of a microporous membrane.
If the solvent is nonvolatile and insoluble in the surrounding media, the pores of
the microporous membrane are made sufficiently small that the liquid is
immobilized by capillarity. Such immobilized liquid membranes (ILMs) are easily
made, and are normally stable for periods of days, weeks or even months, but not
longer.
A second type of facilitated transport membrane is the emulsion liquid

membrane (ELM), first developed by Li at Exxon.s In these membranes, the
solvent-carrier and the aqueous product solution are emulsified together to form
an oil-in-water emulsion. This emulsion is then re-emulsified in the feed solution,
forming a water-in-oil-in-water emulsion. The carrier-solvent phase is the oil
phase, and forms the wails of an emulsion droplet separating the aqueous feed
solution from the aqueous product solution. Permeate ions are concentrated in
the interior of the emulsion droplets. When sufficient permeate has been
extracted, the droplets are separated from the feed solution and the emulsion is
broken, liberating a concentrated product solution and an organic carrier phase.
The organic carrier phase is decanted from the product solution and recycled to
make more emulsion droplets.
4.1.2 Membrane Features
Facilitated membrane diffusion can be much more selective than other forms
of membrane transport.3-6 but the membranes used in the process are usually
unstable. This instability is a tremendous disadvantage, and is the reason why
this method is not commercially practiced.
The characteristics of facilitated diffusion are illustrated by the values in

Table 4-1, which compares diffusion coefficient, separation factor and thickness
for polymer membranes and for facilitated transport membranes. The speed of
any membrane separation is directly proportional to the membrane’s diffusion
coefficient and inversely proportional to the membrane’s thickness. The diffusion
coefficient in polymer membranes is much smaller than that in liquid membranes,
but the thickness of the polymer membranes is also much smaller. As a result,
polymer membranes may show separations with speeds comparable to those of
facilitated diffusion.
Table 4- 1. Representative Membrane Characteristics. Facilitated transport

membranes offer fast diffusion and good selectivity, but are thicker
than polymer membranes.
Diffusion
coefficient Separation Thickness
(cm2/sec) Factor (cm)
Glassy polymer membrane 10-s 4 10-s
Rubbery polymer membrane 10-s 1.3 10-4
Facilitated transport
membrane (liquid) 10-s 50 10-s
The high selectivity of facilitated transport comes from the chemical

reactions between the diffusing solutes and the mobile carrier. These chemical
reactions cover the spectrum of chemistry used in extraction and absorption. In
the system in Figure 4-l. both potassium chloride and lithium chloride are largely
insoluble in the chloroform membrane unless they are complexed. The potassium
ions are strongly complexed by the crown compound, but lithium ions are only
weakly complexed, this is why this membrane is much more selective for potassium
chloride than for lithium chloride.
The coupling between diffusion and chemical reaction gives facilitated

diffusion another unusual feature: the flux is not always proportional to the
solute concentration difference between the feed and permeate side. At low
solute concentration, doubling the concentration difference often doubles the flux,
but at high solute concentration, doubling the concentration difference may have
no effect on the flux at all. This unusual non-linearity occurs because at high
solute concentration, the reactive carrier is fully utilized. In other words,
essentially all the carrier molecules are complexed on the feed side of the
membrane. and essentially none are complexed on the permeate side of the
membrane. Increasing the feed concentration has no effect, therefore, because no
additional complexed solutes can be formed. The resulting data are sometimes
confused with dual mode sorption, a similar, but less well defined effect, which is
described in the section on gas separations.
Because facilitated transport involves coupled diffusion and chemical

reaction, it can sometimes concentrate as well as separate a specific solute. Such
concentration is sometimes called “coupled transport”. An example, given in
Figure 4-2. shows how a feed solution containing 100 ppm copper can produce a
stripping solution with 1,200 ppm copper. The energy for this separation usually
comes from a second reaction, often implicit and often involving a pH change. In
the copper case, it is the reaction of the carrier with acid. Again, the process is
like extraction, where such concentration is common.
Although facilitated transport membranes are highly selective, they suffer

from a major problem, namely instability, that has precluded industrial adoption.
Four main causes of membrane instability have been identified: solvent loss,
carrier loss, osmotic imbalances, and spontaneous emulsification.697 Each merits
more discussion:
Splvent Los Solvent loss occurs when the liquid membrane solvent dissolves in
the adjacent solutions and eventually ruptures. As a result, facilitated transport
solvents are usually chosen for their relative insolubility in the adjacent solutions.
In practice, rupture usually occurs for some other reason, before the solvent in
the membrane has dissolved significantly, so this mechanism is only infrequently a
significant cause of instability.
Carrier LOST: The mobile carrier makes the diffusing solutes more soluble in the
membrane. Unfortunately, the diffusing solutes also tend to make the mobile
carrier more soluble in the adjacent solutions. Moreover, many carriers are
chemically unstable, including many proposed for separating oxygen from air.
These disadvantages can sometimes be reduced by modifying the carrier
chemistry, for example reducing its water solubility by adding side chains to the
mobile carrier, just as is done in extraction.
Osmotic ImbalanceS: A facilitated transport system often involves concentration

differences between the feed and the permeate solution that are 1.0 M or higher.
These imply osmotic pressure differences of 40 atmospheres or more. Such
pressure differences can force the liquid membrane out of the pores of the
polymer support. This effect can be reduced by using supports with very small
pores or good wetting, but it remains a chief cause of membrane instability.
Soontaneous Emulsification: Most membrane additives are amphoteric. For

example, the oximes used to extract copper ions from aqueous solution into
kerosene can also extract some water. This water is sometimes “stranded” within
the membrane, eventually producing trails of water droplets which coalesce into
channels across the membrane. Such emulsification is less well studied than the
osmotic imbalances, but it may also be a serious cause of membrane rupture.
Other factors that may influence the membrane performance are membrane
viscosity, membrane density and support wetting. These issues are less significant
than those concerned with membrane stability.
Copper
Concentration (ppm)
in the acid phase
0 2 4 6
Time (min.)
Figure 4-2. Copper concentration by facilitated transport. While the basic

feed contains only 100 ppm copper, the acidic product has
over 1000 ppm copper.’
4. I .3 Membrane and Module Design Facton

. .
ed Llauld As described above, liquid membranes for
facilitated transport can be stabilized in the pores of a microporous polymer
membrane. Ideally, the microporous membrane supports should have very small
pores. so that the effects of any applied pressures are minimized. The pores
should also be as homogeneous as possible. In practice, many ultrafiltration and
microfiltration membranes work well. Hydrophobic microporous polypropylene
membranes are commonly used.
The membrane support should have a large surface area per volume to allow
a large flux per volume. Plate-and-frame and hollow-fiber supports, such as those
shown in Figures 4-3a and 4-3b. are often used. Capillary hollow-fiber systems
are generally preferred. The support is prepared simply by wetting it with the
membrane liquid. Sometimes, this wetting is enhanced with vacuum or pressure,
but this is rarely essential.
Emulsion Liauid Membranes: A second method of using facilitated transport

membranes is as water-in-oil-in-water emulsions, shown in Figure 4-3~.~~~~~ An
example of an emulsion liquid membrane would be droplets of water emulsified by
rapid stirring in a chloroform solution of surfactant and of mobile carrier, then
dispersed in the mixed salt feed solution by slow stirring. Salts like potassium
chloride diffuse from the mixed feed across the chloroform into the internal
phase. This method is also called “double emulsion membranes,” “liquid surfactant
membranes.” and “detergent liquid membranes.”
The advantage of emulsion liquid membranes is the large area per volume,
which enables separations to take place at great speed. The principal
disadvantage is the complexity. The membrane separation may be fast, but the
emulsion manufacture, the emulsion separation, and the recovery of the internal
phase from the centers of the droplets can be extremely difficult.“*”
Membrane ContactorS: A final method of employing facilitated diffusion is to

make a membrane contactor. A membrane contactor typically uses a hollow-fiber
module containing microporous membranes, as shown in Figure 4-3d. Feed flows
through the lumens of the hollow fibers, and extractant flows countercurrently
on the shell side of this module. Solute entering in the feed is removed by the
extractant. The solute is later stripped from the extractant under different
conditions in a second module. This process is really a hybrid, combining aspects
of both conventional extraction and liquid membranes. Alternatively, one can use
two sets of hollow fibers immersed in a “liquid membrane” solution. One set of
fibers carries the feed, and the other contains the product stream. In either
case, the membrane contactor performs the same function as a packed tower.
Membrane contactors exhibit the advantages of liquid membranes, but avoid

disadvantages of absorption and extraction.“*ls They have a large area per
volume and hence produce fast separations. They are unaffected by loading, by
flooding, or by density differences between feed and extract. In this sense, they
are a poor man’s centrifugal extractor. Membrane stability problems are avoided
by essentially putting one membrane interface in an extraction module and the
other interface in a stripping module. However, although the performance of
C
.-
-z
l?!
membrane contactors in the laboratory is excellent, their performance under

industrial conditions is only currently being explored. The energy requirements of
such a system have yet to be clarified.
4.1.4 Historical Trends
Studies of facilitated transport originated in biochemistry.10 Living

membranes often show fluxes that are faster and more selective than would be
expected to be achieved by the membrane components. Living membranes often
concentrate specific solutes, but are easily poisoned by particular drugs. These
observations led to the postulates of mobile carriers. Theories of mobile carriers
were later tested by studies on blood, where hemoglobin facilitates diffusion of
oxygen and of carbon dioxide.
The high fluxes, high selectivity and specific concentration achieved in

biological membranes spurred research into synthetic membranes and carriers. The
simultaneous development of microporous polymer films abetted this development
by providing membrane supports. By 1975, facilitated transport processes were
being studied for acid gas treatment, metal ion recovery, and drug purification.
None of these processes became commercial. Ironically, biochemical interest also
waned, deflected into studies of chemically selective pores.
Since 1980, the most important research studies in the field of facilitated
transport have focussed on membrane stability, an issue which remains unresolved.
4.2 CURRENT APPLICATIONS
Many fundamental studies of facilitated transport systems have been

performed. Table 4-2 summarizes a recent review by Noble et a1.s. and shows the
broad scope of facilitated diffusion research to date.
The solutes shown can be roughly divided into three groups. The largest
group includes metals, especially in their ionic forms. This group underscores the
similarity between facilitated transport and extraction. A second group, gases,
builds on parallels between facilitated transport and absorption. A third group
consists of organics of moderate molecular weight, with an emerging bias towards
pharmaceuticals. These fundamental studies have elucidated the detailed chemistry
responsible for many types of facilitated diffusion.
In contrast to these fundamental efforts, commercial successes are sparse.

The most serious, publicly detailed effort has been by Rolf Marr and his
associates at the University of Graz, in Austria.g These workers developed a
large-scale emulsion liquid membrane process for the selective extraction of zinc
from textile waste. The process is shown schematically in Figure 4-4. They
used liquid membranes of diamines dissolved in kerosene. The product solution is
first emulsified in the carrier medium. This emulsion is then re-emulsified with
the feed solution. In the permeation step the zinc migrates into the interior
product solution. The emulsion is then passed to a settler and the feed solution,
now stripped of zinc, is removed as a raffinate. In a final step, the carrier-
product emulsion phase is passed to a de-emulsifier where the emulsion is broken
to produce a carrier solution, which is recycled, and a product solution containing
Table 4-2. Fundamental Studia of Facilitated Transport. 8
PCWlldk solventer Carrier

specln 5urlaclanl Species SUPW
co OHS0 CuSCN. KMethyllmidazote Poty(trimethylvinylsilane~
co,,H&H,. co l&O. IMP EDA. Hlndrred Amlnes PFSA
coml, OMSO Cut3 and VCI, Poly(trlmethylvinytsllane~
CO,/N, HrO K&Or Porous Polypropylene
COJO, and II, Polyethytm ctycot Ca. Ida Bulk Llquid
HA co, Hz0 Carbonate
t&s/CO, Hz0 Carbonate. Ethanolamine Bulk Liquid
NO Ktycloheqt-2.pyrrolldone Fe(H). Zntlll Bulk Liquid
0, OMF Sillcon Carbonate Cellulose Acetate
0, and co OMC I-Vlnylpyrldlnc Porous PTFE Film
Wh OMSO (Co) Polyamine Chclatc Poly(trimethylvinylsllane)
0,/b OMSO (NH Pyridlne Complex Porous PTFE Film
0*/h OMSO A’. IY-bis(2.aminobenzal)EDA Polyarylene
0,/h OMSO Dlpropylcnetrlaminc Co Salt Bulb Liquid
Cd, Hg. And Zn PheMe mDlcyclohcnano-lB-crown-6 Bulk Liquid
Ac. Hr. Ba. Co. ftl. Cu. Zn. Ag CH,CI Polytluoroalkyl, -alkenyl Ion Bulk Liquid
Exchangers
Acetlc &Id Tertiary Amlna
Arettc Acid Trloctylphorineoride
Acetlc, Benzotc. and Phenylacetlc Acid: LIOH
Phenoi and n+Cresot
Ag’ Cumcne Solutions DEHPA 1 eflon
AgBrrt-1 loluene DClK, Bulk Liquld
AgBr,t-1 Macrocycles
All~ll Catlons Chloroform Polynactln. OB-18-C-6 Bulk Liquld
Alkall Metals CH,CI, Macrocycles Bulk Liquid
Am Aminer and Alkylphosphortc Acids
Am”. Eu” Oecallne/KOodecane Neutral/Acldlc Organic Phosphates Composite Porous Films
AmtIll) Olethylbeneene (Carbamoylmethyl) Phorphlne Polypropylene Film
(continued)
Table 4-2 continued
Permtrblt 5olvtn1or Carritr

Sptdtt Surfrclrnt Sptcits SUVpOII
Amino Acids DltIhylbtnxent Mxrocycla Bulk Liquid

Amlnotlk-yi Sulfonlc. -crrboxyllc Acids CHCI, Tttraoctylrmmonwm Bromidt 8ulk Liqwd
Btnxtnt/Cyclohtunt H,O A9N’% Porou, PTFE Film
Bllhbln Htxadtcyltrimtchylrmmonium Bromide Bulk Liqwd
C,H./CH, H&J A9* Porous Film
Cd" Aliquat 336
Cd.Ln, Hq. Au Mwrocyclt
cow/Mlw W Olalkyl~hor~hlnic Acld/lcR- Compmtr Porous tllm,
Alkyltmmonwm St11
co. u BIs(Ethylhtxyl)phosphatt
Cf KOtcrnt Trloctyltmmt
Cl OU Alrmmt 336 Porous Polyproplltnt
Film
CI" Ammt
CP. HqJ’ Alrmlnt 336
crcvn OrqrnlcSohnt Trloctylrmlnt Bulk LlquId
crcvn.RtWI) Orqtnlc Soivtnt Qwrttnrry Ammonium Chlorldn Bulk Liqwd
cr.Ln. Rt. Tc v SPAN 80 Aliqurl 336
0. 9. co,u. ct. Tc DEHPA
Q. Sr. co.u. ct. fc 8.Hydroxyqulnollnt
CU ECA so25. ECA 4360 2.Hydroxy-5~rlkylbtruophtnont
CU Ktrostnt Prlmrry Amint Porous Polypropyltnt
Film
CU Ktrostnt N-510 rnd P 204 Bulk Liquid
CU KDodtunt Dkthylhtxyl~hos~horoca Add
CU 2.Hydroxy-5.nonylbtnrophrnont oximt Film
CU Olsptrsol SMtX29
(continued)
Table 4-2 continued
PtrmtJbk Soirtntor CJrrirr

spwes Surlactmt Spears
CU BIs(Z-Ethylhcxylphosphorlc Acid Porous Hollow fibers

CU LIX64N
CU LIX65N
CU LlX64N. Acorga P5100. 300
Cd and H’ Toluene D-Hydroxy-orlmc Porous Polypropylene
Film
Cu”. NH; H,O StCJrlC Acid
Cu. Co. 2% Nl H*O Dl(2-EthylhexyDphosphorlc Acid Bulk Clquid
Cu. Nl. Ln CH,CI, Lipophlllc Tetraaminu Bulk Liquid
Enmtlomerlc Enrichment Hz0 Cyclodextrlns
Ethylene/Elhane B,O AgNDr Porous Film
Ft(III) CrHSt, l&Crown-6 Macrocycle Bulk Liquid
Fe.co Orgrnic Solvent hlethyttrlOc~lJmmOnium Bulk Llquld
n- 011 Trldodccylrmlne Electrode
H,PO.( -) Oxomolybdenum(~tetrJphenyl
Porphorln
HJllde5 1.2~Dlchloroethmc Ammonioimldater
Kg CHCI, Macrocycles Bulk Llquld
K9 Dlbutylbcnzoylthlourtr Bulk Liquld
HNO, DecJnc TBP. TM
HNO, Dlcthylbcnzenc TrllJurylamlne Bulk Llquld
Hydrocarbons B,O SulfolJne. TEC
K Jnd NJ Plcratn Crown Ethers Bulk Liquid
L-Amino Aclds ADOCEN 464
L-Ltu. L-Trp Me btm Dlbenzo-lB-crown-6 Bulk Liquld
LJ. Sm. fr. Fe. CA,Na Crown Ethers DE-18-C-6. B-IX.5 Bulk Liquid
LJnthJntdti. LJ.Sm. Cd. fr DEHPA
Li’. K’. NJ- OctJmethyltetraoxJquarterJne Bulk Llquld
Li’. NJ*. K’ CH,CI, MOnOJZJ CrOWn Ethers Bulk Liquid
M&l M( +). M(2 l) CatIons CHCI, Amino Acldr and Polypcptlda Bulk Liquld
MetJl N&rates. Pb. Ag. Tt Macrocycles
the separated zinc. The zinc process was installed in Lenzing, Austria, where it
processed 1,000 m3/hr for at least six months.
Other efforts to demonstrate industrial processes have been stillborn.

General Electric explored the possibility of acid gas removal using an aqueous
carbonate liquid membrane supported by cellulose acetate.’ The project was
abandoned at the pilot stage. Bend Research and General Mills Chemicals (now a
division of Henkel) both studied copper removal using an oxime solution. They
apparently abandoned the work because they found no compelling advantages over
conventional extraction. Recovery of uranium from ore leach solutions was also
developed by Bend Research to the pilot-plant stage, using amines immobilized in
hollow-fiber modules. This work was also abandoned because of stability
problems. In by far the largest effort, Exxon studied the recovery of metals and
of drugs using emulsion liquid membranes, a geometry on which they hold broad
patents, now beginning to expire.s~‘5~16*17 Both uranium and copper recovery
were investigated, but neither process became commercial. One additional effort,
a collaboration with Mitsui Chemical for recovering dissolved mercury, reached the
pilot scale, but has since been abandoned.
Many Chinese and Eastern European commercial efforts on facilitated

transport apparently continue, for reasons that are not clear. Pilot units for
processing chromium and other metals are reported to be operating in China. The
future prospects of these efforts are unknown.
4.3 ENERGY BASICS
Facilitated transport competes with gas absorption and with liquid extraction,
but not with evaporation. It uses pressure differences or chemical energy, not
thermal energy. As a result. facilitated transport membranes will have little
effect on the direct use of thermal energy.
Facilitated transport systems tend to require the same amount of energy as

conventional gas absorption and liquid extraction. For example, hydrogen sulfide
separated by facilitated diffusion from flue gas must use a high-pressure feed and
either a low-pressure product or a steam sweep. As a second example, copper
recovered from dilute solution by facilitated diffusion requires almost the same
acid per mole of copper as is required in liquid extraction. For the same
separation, a similar amount of energy is needed.
The advantage of facilitated transport membrane processes lies not in their

energy consumption, but in their speed. Facilitated membrane processes are much
faster than their conventional counterparts. Used in the form of membrane
contactors. facilitated transport processes are about eighty times faster than
absorption towers of equal volume, and six hundred times faster than extraction
columns of equal volume. This greater speed is primarily the result of the very
large surface area per volume possible in membrane devices. This surface area
can be achieved at a wide range of flows, independent of the constraints of
loading and flooding which can compromise conventional unit operations.
Although facilitated transport membranes are likely to be ten times more
expensive per volume than existing equipment, their speed and space efficiency
still offer some real potential for cost savings. This advantage alone, however,
may not provide sufficient motivation to displace fully depreciated equipment in a
L--------~------__---_
I
I
I
,
Emulslficatlon Permeation Settling Breaking

step step of emulsion
Figure 4-4. A schematic of the only commercialized facilitated transport

system that has been built. This Austrian process uses emulsion
liquid membranes to recover dissolved zinc.
stable chemical process industry. It could, however, have substantial impact in

other smaller-scale applications, especially pharqaceuticals.
4.4 ECONOMICS
Only one commercial-sized facilitated transport membrane system has been

installed, so no sensible discussion of the economics of the process is possible. It
is possible, however, to highlight product areas where commercialization has been
considered and identify the improvements necessary to make commercialization
attractive. Four such product areas are obvious: metals, gases, biochemicals, and
sensors.
4.4.1 Metals
Key metals of interest in the non-ferrous industry are copper and uranium.
Not surprisingly, this is where most facilitated transport systems have been
investigated.1*Q~‘6~1* Both immobilized liquid membranes and emulsion liquid
membranes have been used; both work effectively; both have been tried at the
pilot scale; both have been abandoned.
The full development of facilitated transport membrane systems for these

applications may have been inhibited by a general depression in the metals
industry. Recent increases in copper prices, and interest in superconductivity
research using rare earths, make it worth reviewing the possible advantages for
facilitated transport of metals.
Enernv Use: The energy use of facilitated membrane separations will be

similar to that of extractions. In some cases, facilitated transport can operate
with “dirty” feeds which require filtration before liquid-liquid extraction. There
is no energy advantage in using facilitated transport membranes rather than
conventional extraction processes.
Seoaration Soee& Membrane separations can be hundreds of times faster than

conventional extractions. For copper, this speed is not a significant advantage,
because existing conventional extractions take only about twenty minutes per
stage. For hazardous materials, however, including radioactive isotopes, it means
the equipment can be much smaller. For these hazardous materials, there is an
opportunity.
Extractant Costs The amount of extractant inventory is dramatically reduced.

Solvent entrainment in the raffinate is usually minor with supported liquid
membranes, but can be significant in conventional extraction. However,
extractant losses caused by dissolution in the feed solution are the same as those
in extraction. Advantages here are minor except for very expensive extractants.
Stable Membranes: One of the advantages of membrane technology in general

is that membrane systems run reliably with minimal operator attention. Currently,
all liquid membranes have major stability problems. The ability to perform
facilitated transport in stable membranes would be a major advance. Improved
membrane stability is the key to better economics for metal separations.
4.4.2 Gases
Gas separations using facilitated diffusion have focused on three areas:

oxygen separation from air, acid gas treatment and olefin-alkane separation. In
each case. the attraction of facilitated diffusion comes from high speed and
higher selectivity. Some processes also claim lower regeneration costs, and hence
energy savings. The chief concern again is membrane stability.
4.4.2.1 Air separation
Oxygen and nitrogen are produced industrially in greater quantities than all
chemicals except sulfuric acid, so their separation is of major importance and has
been studied for many years. The minimum energy consumption possible to
perform an oxygen/nitrogen separation, based on the free energy of unmixing, is
about 0.09 kWh/ms of oxygen. This separation is now carried out commercially by
cryogenic distillation. pressure swing adsorption, and polymer membranes. The
energy currently used for cryogenic distillation is 0.4 kWh/ms of oxygen, or less
than five times the thermodynamic minimum. The energy used by pressure swing
adsorption (PSA) is somewhat higher. The capital costs of a PSA system are,
however, lower than those of a cryogenic unit for small to medium sized
installations.
Commercial polymer membrane-based air-separation units have been

successful, especially in producing nitrogen-enriched air for inert blanketing. The
wide application of membrane separation has been inhibited by membrane
selectivity. Facilitated transport offers higher selectivity, so it could, in theory
at least, outperform diffusion through polymer materials. Thus the outlook for
facilitated transport in gas separation is unfocused but positive.
Past studies of facilitated transport have used solutes to complex oxygen,

and hence enhance its flux across the membrane.‘g-2a In most studies,
porphyrins or Schiff bases were used as complexing agents. The first successful
systems used these solutes in immobilized liquid membranes, which gave selective
separations but were not stable.24 In this case, the instability comes largely
from the chemical decay of the carriers, not from failure of the membrane films.
No published studies discussing the energy required for facilitated transport are
available.
Other studies have explored facilitated oxygen transport systems in which

the oxygen complexing solutes are contained in solid films. Recently, some
Japanese groups have succeeded in making oxygen-selective membranes in this
way.2ss24 These membranes are apparently not yet stable, and the basis for
their claimed success is frail. However, they do show characteristics of
facilitated diffusion, and may become a truly major advance.
4.4.2.2 Acid gases
Acid gas feed streams are best organized under three subheadings: flue-gas
desulfurization, industrial gas treating and organic sulfur removal. In flue-gas
desulfurization, the target is to remove 90% or more of the 0.2% sulfur dioxide
present in the gas. In gas treating, the goal is to reduce the concentration of
carbon dioxide and hydrogen sulfide to perhaps 100 ppm (4 ppm for hydrogen
sulfide in natural gas). In the third case, organic sulfur concentrations are to be
reduced to 1 ppm or less.
Currently, these separations are accomplished by gas absorption. When the

feed concentrations are high, the acid gases are dissolved in non-reactive
(“physical”) solvents. When the feed concentrations are lower, the acid gases are
more easily removed with reactive (“chemical”) solvents. Many of these chemical
solvents are synthesized to react reversibly; these are the best candidates for
Many workers have studied facilitated diffusion as a means of separating

acid gases.4*25~z6 Their efforts show that the diffusion process has the same
energy use and selectivity as conventional gas absorption. It is much faster per
volume of equipment. However, membrane stability and membrane fouling remain
major, unresolved concerns. Moreover, new chemical solvents of greater capacity
have essentially increased the capacity of existing gas absorbers, and hence
reduced the need for new technology. As in the case of metals, facilitated
transport has not shown compelling advantages over currently practiced
technology.
4.4.2.3 Olefin-alkanes and other separations
The separation of olefins from alkanes can only be accomplished effectively

by absorption or distillation. Thus facilitated transport can achieve savings in
energy and improvements in selectivity. The selectivities for facilitated
transport membranes should exceed those in conventional polymer membranes.
Past efforts at olefin/alkane separations have exploited the reactions of

olefins with metal ions, such as Ag+, Cu+, and Hg+.27n2s Originally, the ions
were used in aqueous nitrate solutions; later, in cation exchange membranes.
Within these membranes, ions are electrostatically tethered but still mobile, even
in the absence of water. As a result, such membranes avoid many of the stability
problems that plague facilitated diffusion.
Pez, et al., at Air Products, have recently developed molten salt membranes
for ammonia/hydrogen and other separations.2Q-s1 Their results suggest that the
transport mechanism through the molten salt may be similar to that in ion-
exchange membranes. Both the ammonia separations and olefin/alkane separations
will be especially attractive if the feed is already at high pressure. The
technology to perform these separations works well at laboratory scale. Given
that this is the one area of facilitated diffusion where stability is not an
overwhelming difficulty, the time appears ripe for scale-up and commercial
exploitation.
4.4.3 Biochemicals
Specialty chemicals, especially high value added materials, such as antibiotics

and proteins, are frequently mentioned as candidates for facilitated transport
separations. Although facilitated transport systems will work, the energy demands
and extractant costs will be comparable with those of conventional processes in
most cases.
Fast facilitated diffusion may be superior to slower conventional extraction

for unstable solutes. For example, penicillin is stable as the sodium salt but
unstable as the free acid. It is purified by adding acid to the salt to make the
free acid, extracting the free acid into a solvent such as amyl acetate, and
stripping the salt out of the acetate with base. The faster the extraction, the
greater the yield. However, although facilitated extraction has been a common
idea for fifteen years, it is not practiced commercially for penicillin purification.
4.4.4 Sensors
Facilitated diffusion is used in a variety of chemical sensors, especially for

biochemicals. It works well. offering high selectivity. However, the market for
biochemicai sensors is extremely fragmented, with literally hundreds of small.
highly specialized uses. As a result, the tendency is for instrument
manufacturers to produce generic electrode assemblies, into which the buyer
installs his own selective chemical system. There is little potential, within the
context of the present study, for recommending development efforts that would
have any impact in such a market.
4.5 SUPPLIER INDUSTRY
There are currently no suppliers of equipment for facilitated transport.
4.6 RESEARCH CENTERS AND GROUPS
Table 4-3 summarizes the research centers and investigators active in

facilitated transport research. The list is broken down by membrane type.
The first group, which is the largest and most diverse, includes work that
extends beyond the boundaries of liquid membrane studies. This group includes
the greatest numbers outside of the United States. Advances in this area may
well continue to be scientifically successful, but are unlikely to be practiced
commercially.
The second group, which focuses on membrane contacton, is now entering

commercial practice. The scientific development of this area is largely complete:
the design equations are known, the key operating steps are well established, and
some important applications have been identified. So far, much of the work has
been academic. What is needed is encouragement for more commercial
applications. This is especially true in the area of acid gases and antibiotics.
The third group lists those studying solid membranes capable of facilitated
diffusion. Interestingly, the group has a large number of industrial scientists, a
reflection of the major potential gain. At present, these industrial scientists are
aiming at specific separations, like those of air or of olefin-alkanes. Their
academic counterparts aim more at exposing mechanisms, rather than developing
processes.
Table 4-3. Facilitated Transport Researchers
TOPIC/Investigatop Area of Interest
WOUID MEMBRANES
Noble and Pm (NIST) Beat review article

Bunge (Colorado School of Mines) Metals
Danesi (Europ. Atomic Energy) Rare earths
Freisen (Bend Research) Air
Rolf Marr (U. of Graz) Commercial zinc process
Noble (U. of Colorado) Good theoretical perspective
Stroeve (UC Davis) Facilitation by micelles
Wasan (Illinois Inst. of Tech) Emulsion liquid membranes
Govind (U. of Cincinnati) Air
MEMBRANE CONTACTORS
m (Stevens Institute of Tech.) Liquid extraction

Aptel (Toulouse U.) Water
Callahan (Hoechst-Celanesz) Gold
Cussler (U. of Minnesota) Acid gases; proteins
Fane (U. of New South Wales) Membrane distillation
SOLID MEMBRANES
Cussler (U. of Minnesota) One of few analyses

Ho (Exxon) Olefin-alkane
Nishide (Waseda U.) Air
Noble (U. of Colorado) Theory
Pez (Air Products & Chemicals) Molten salt glasses
Way (Oregon State) Air
Yoshikawa (Kyoto U.) Acid gases
l The underlined author is a good starting point.
4.7 CURRENT RESEARCH
Facilitated transport remains a specialized research area, which is practiced

mostly in the academic sphere. A listing of published authors would show a
number from industry, but would be misleading, because many industrial scientists
only publish their work after it has been abandoned.
The most active industrial program in this area is at Air Products and
Chemicals, where facilitated diffusion of ammonia and other gases through molten
salts are being examined. The company has recently been awarded a contract of
$1.2 million from the DOE, Office of Industrial Programs, for further
development of the work.
At Air Products and Chemicals, viscous, extremely low volatility liquids and
molten salts are used as liquid membranes. These membranes can withstand
pressure differences and high temperatures. Although the first membranes of this
type to be developed were molten salts that needed to be operated at
temperatures in excess of 4OO’C, the newer systems are based on polyamine
liquids that can be used at room temperature.
Pez et al. have described a facilitated transport membrane for oxygen/

nitrogen separation based on molten lithium nitrate.se The molten salt is
supported in the pores of a stainless steel mesh and the membrane is operated at
429-C. Nitrite ions act as the carrier to selectively transport oxygen across the
membrane as nitrate ions. Nitrogen does not react with the salt and hence its
permeation rate is minimal. A similar approach i used for ammonia separation,
using zinc chloride as the molten salt.sl
The Air Products group has also studied membranes based on salts that are
molten at room temperature. These materials are also referred to as “ionic
liquids”. Examples are triethylammonium chlorocuprate and ammonium thiocyanate.
The former has been used as a liquid membrane for the separation of carbon
monoxide from nitrogen, a separation not yet achievable with polymer membranes.
The latter liquifies in the presence of ammonia to form an electrically
conducting liquid membrane.
A refinement is the use of molten salt hydrates, which resemble molten

salts. but include some bound water molecules. zs Molten salt hydrates typically
have lower melting points, and can dissolve larger volumes of gas, than molten
salts themselves. For example, Air Products used tetramethylammonium fluoride
tetrahydrate in a liquid membrane for the selective separation of carbon dioxide
from mixtures containing hydrogen and methane.
Potential applications for facilitated transport are listed and rated in Table
4-4. The most promising area in the metal separation category is copper
extraction, and in the gas separation category, oxygen/nitrogen separation from
air. The technology is not attractive for hydrocarbon separations or water
separations, both of which can be accomplished much more efficiently by other
membrane-based technologies.
4.8. I Metal Separations
The key areas of interest for the use of facilitated transport in metal
separation continue to be copper and uranium extractions. Recent sharp
increases in copper prices make improvements in existing copper producing
processes, and development of new processes, worthwhile. The market for
uranium is less predictable, because it will always depend on the acceptability of
nuclear power generation.
If superconducting materials begin to be developed on a large scale. there

will be a rapid growth in the demand for rare earths to make these materials.
Rare earths could be separated effectively by facilitated transport. However, any
research support for this area should wait until needs are clearer.
Table 4-4. Applications for Facilitated Transport
Prospects for
Application Area Importance Realization Comments
Metal Sepa&~B
Copper Extraction 8 Good Stable membranes

Uranium Purification 5 Fair with the speed and
Rare Earth Recovery 3 Fair selectivity of
facilitated diffusion
remain the key.
Air 10 Good This area is the most

Acid gases 6 Good varied and will need
Hydrogen/Methane 2 Poor different responses
to different challenges
Biochemicals
Commodities Fair Facilitated transport

Antibiotics Excellent will work, but it is
Flavors Fair unlikely to be used
Proteins commercially.
Membrane confactors will
become standard for
antibiotics and
flavors.
Hvdrocarbon Seoaratioq
Olefin-Alkane Good The olefin-alkane

Aromatic-Aliphatic Poor effort merits more
Linear-Branched Poor work. The other
Solvent Recovery Poor separations are
better done with
other membranes.
Water Remova
Ethanol-Water 2 Poor Pervaporation and

Gelatin-Water 2 Poor ultrafiltration have
greater promise.
Facilitated transport of metals will develop in two directions. First,

membrane stability must be improved. Efforts to date to improve liquid membrane
stability have been partially successful, but have not provided compelling reasons
to abandon extraction. Staging membranes currently seems impractical.s2
Another approach is to accept that the carrier liquid will be lost or degraded and
to replace it periodically. Such a technique seems cruder and less elegant than,
for example, the “contained liquid membranes” of Sirkar.‘s However, membrane
reloading may merit some further consideration.
The obvious alternative to improving liquid membrane stability is to foster

development of solid facilitated transport membranes containing reactive groups
responsible for selective separations.2s*24 The reactive groups are parallels of
the mobile carriers in liquid membranes; reflecting this parallel, they are
sometimes called “chained carriers” or “caged carriers.’ The mechanism by which
solid facilitated transport membranes work is not yet understood. Authors often
use words like “hopping” and “flexing,” and analogies with “bucket brigades” or
“chemically selective pores.” One group has whimsically discussed performance in
terms of Tarzan paddling a canoe VS. swinging from a vine, as shown in Figure
4-5.
Research into solid membranes capable of facilitated diffusion is a key area

that merits fundamental support. The public and academic research effort should
initially focus on gaining a better understanding of the transport mechanisms.
This will aid parallel industrial efforts to develop practical systems.
Facilitated transport of copper and uranium could be improved by using

membrane contactors, which do not suffer so severely as liquid membranes from
problems associated with carrier loss or degradation. A membrane contactor
typically uses a membrane module containing microporous hollow fibers to carry
out an existing liquid-liquid extraction. It can also use one module containing
two types of hollow fibers, one for feed and one for product. The result can be
a fast, efficient separation effected at a fraction of the cost of a centrifugal or
conventional extraction. However, the performance of a membrane contactor
under the conditions actually found in the metals industry remains unknown.
Investigating membrane contactors, at least for copper separations, deserves
support for development.
In summary then, for metal separations, the most worthwhile areas in which
to commit research resources, are:
1. Understanding and developing solid facilitated transport membranes.
2. Developing membrane contacton.
A lower priority, but of some merit, is to find improved ways to reload
membranes with liquid carriers.
4.8.2 Gases
Facilitated transport of gases continues to be dominated by questions of
membrane stability. Improved membrane stability is being actively pursued by
various uncoordinated strategies, of which three stand out. First and most
dramatically, several groups are using ionic mobile carriers tethered
electrostatically within ion-exchange membranes. So long as the ions within the
membrane are protected from side reactions. the membranes wiil be stable.
Second. other groups are using non-porous glasses or gels as membranes. The
most promising membrane materials are the molten salts developed by Air
Products. Third, some workers are using membranes with smaller, well defined
pore sizes. including microporous ceramics. All of these efforts are in the
relatively early stages of development, and it is too early to assess their
ultimate success.
Interestingly, efforts to improve membrane stability are without definite

targets. Everyone believes that improved stability is vital, but there is no
consensus as to whether an acceptable membrane stability would be measured in
weeks, months or years.
The key gas separations for facilitated transport are those of air and of acid
gases. The most important gas separation is that of air, where polymer
membranes are only modestly selective, and where significant improvement in
performance might be achieved by facilitated transport membranes. The second
important separation is that of acid gases, which can probably be improved by
facilitated diffusion. Other important gas separations, like hydrogen-nitrogen and
carbon dioxide-methane, are poor candidates for the method.
The success of polymer membranes for air separation has had a major impact
throughout the chemical and chemical engineering industries, and has been a
factor in the increased interest in membrane separations by large corporations and
the U.S. government. Polymeric membranes do not exhibit high oxygen/nitrogen
selectivities. The increased chemical selectivity possible with facilitated diffusion
has spurred development of reactive solutions which complex oxygen. Reactive
solutions make good liquid membranes, but the research effort has stalled for two
reasons. First, the solutions are often fpn reactive: they complex oxygen and
dimerize so strongly that stripping is difficult. This excess reactivity can
probably be solved by new chemistry, in a comparable way to that in which
temperature- and pressure-sensitive amines were developed for acid gas
absorption. The second reason is, of course, the lack of membrane stability, an
area where there continues to be ongoing research.
Facilitated transport of oxygen in solid membranes continues to attract

major research efforts, many in industry. Unfortunately. failures are rarely
published, so that many unknowingly repeat earlier unsuccessful ventures. The
situation can be almost comic; for example, at least five different groups have
found that hemoglobin analogues in rubbery polymer films do not facilitate
oxygen transport. In spite of this, major effort for oxygen facilitation in solid
membranes will and should continue, spurred by recent reports of Japanese
success. The problem is too important to ignore, and the rewards for the
development of stable facilitated transport membranes for oxygen/nitrogen
separation would be substantial.
The separation of acid gases by means of facilitated diffusion is technically

feasible, but has not been vigorously pursued and is not practiced commercially.
The lack of interest may reflect direct competition with polymer membranes or
with gas absorption, where new reactive liquids have increased the capacity of
old equipment. Solid facilitated transport membranes suitable for carrying out
acid gas separations might be possible, but do not seem to be under active
development.
A new direction for facilitated diffusion of acid gases uses membrane

contacton. The fundamentals of the process are in place, but the practical
utility haa yet to be demonstrated. Demonstrating this utility merits more
technical effort. A second direction in acid gas treatment is the use of ceramic
membranes as liquid membrane supports. Ceramic membranes presently have a low
surface area per volume, but they do have small pores and remain stable at high
temperatures. They can remove low concentrations of gases such as hydrogen
sulfide without feed recompression, though they are less effective than absorption.
This direction has yet to show compelling advantages over gas absorption.
In gas separation, research into oxygen facilitation in solid membranes is

most important. Work on solid facilitated transport membranes for other
separations has merit, as does further development work on membrane contactors
4.8.3 Biochemicals
Facilitated transport continues to be viewed as a natural separation

technique for unstable, costly biochemicals, where fast, high-purity separations are
imperative. This view is oversimplified.
Some interesting biochemical examples of facilitated transport aim at chiral

separations. These require mobile carriers selective for, for example, a d-isomer.
Existing carriers are not sufficiently selective to be commercially attractive.
Alternatively. some workers have used a “membrane reactor,” which uses an
enzyme to oxidize the d-isomer and a membrane to separate the l-isomer and the
oxidized form of the d-isomer. Membrane reactors of this type, although they do
not use facilitated diffusion, may be appropriate to perform the same types of
biochemical separations as could be done by facilitated transport.
However, despite the superficial match between biochemical products and

facilitated transport processes, a major research effort in this area is not
justified. To understand why, a comparison should be made between the
production methods for commodity chemicals, such as those made by Exxon
Chemical Company, and specialized pharmaceuticals. as manufactured by the
Upjohn Company, for example. Exxon’s 23 well defined chemical products are
made on the large scale, each by dedicated equipment working 24 hours a day.
The equipment is optimized for each product. An efficient production line of this
type is basic to profitability in commodity chemicals. Upjohn’s 343 products, on
the other hand, are made on the small scale. For specialized drugs and products,
the annual demand may be 20 kg or less. As a result, production is carried out
batchwise. on non-dedicated units that may be operating under far from optimal
conditions for that particular application. However, Upjohn’s strategy is entirely
appropriate to the market, where highly sophisticated equipment lying idle for
most of the time would be completely impractical.
The new biotechnology products now beginning to come onto the market also
fall into the specialized, limited market category. For example, the anticipated
demand for tissue plasminogen activator (TPA), an agent proven to dissolve blood
clots, is around 80 kg/year. Factor VIIIC, used in treating hemophilia, is

consumed at the rate of about 60 grams/year in the United States. and is supplied
by five competing companies. Even bovine growth hormone, expected to be a
major biotechnology product. has an expected demand of about 20 tons/year. It
is not feasible to devote a major effort to develop facilitated transport systems to
be used in making this type of product, when the advantages to be gained over
other already well developed membrane and non-membrane processes are marginal.
For example, ultrafiltration. a fully mature membrane technology, can perform a
variety of biological separations adequately. For solvent recovery in the
pharmaceutical industry, gas separation membranes or pervaporation membranes
would work as well, or better than facilitated transport membranes.
In the biotechnology area, therefore, a facilitated transport research effort

is not recommended.
4.8.4 Hydrocarbon Separations
As discussed in Section 4.4.2.3, facilitated transport can effectively separate

olefins and alkanes. The membranes used for this separation, which are cation
exchangers with mobile carrier metal ions, are stable. However, this promising
direction for facilitated diffusion has stalled. Industrial development has ceased,
and academic research has tended to shift to air separations. It is not clear why
this opportunity is not being pursued, particularly as energy and cost savings
could probably be achieved by a facilitated transport process.
These stalled olefin separations are the only promising application for
facilitated transport of hydrocarbons. Aliphatic and aromatic hydrocarbons are
better separated by distillation. When this fails, pervaporation is a better choice
than facilitated diffusion. The exception might be if axeotropes inhibit distillation
and if highly selective complexing of aromatics is possible. Linear vs. branched
hydrocarbons are more effectively separated by adsorption. Solvent recovery,
potentially a large area, is better approached with polymer membranes. Except
for oiefins, facilitated diffusion is a bad choice for hydrocarbon separations.
4.8.5 Water .Removal
One can imagine mobile carriers which would react with molecules of water
and facilitate their transport. These carriers might even produce greater speed
and selectivity than can be attained by other means. However, these advantages
would probably be outweighed by membrane stability problems. At least at
present, facilitated transport is a poor second choice compared with pervaporation
and ultrafiltration.
4.8.6 Sensors
Although facilitated transport is not commonly used for commercial

separations, it is often used for chemical sensors. For example, to measure the
concentration of glucose in the presence of other sugars, one can cover a pH
electrode with a liquid membrane containing glucose oxidase. This enzyme
oxidizes glucose, producing a pH change sensed by the electrode. Such
facilitated transport electrodes are common and effective, especially when they
concentrate the solute of interest, and hence amplify the basic signal.
The development of these sensors will continue. It merits technical effort,

though not necessarily by DOE. For example, a sensor for the AIDS virus seems
better suited to NIH sponsorship, but a sensor for plastic explosive might be of
DOE interest. Each of these sensors might be made more sensitive by use of
4.9 RESEARCH OPPORTUNITIES: SUMMARY AND CONCLUSIONS
Facilitated transport is more selective and less stable than other types of
membrane separations. It does not merit dramatically increased technical effort.
However, it does provide some opportunities which, if selectively pursued, can
yield major advances.
Eight research opportunities are given in order of importance in Table 4-5.

These opportunities echo the conclusions of the earlier sections. Three are
concerned with solid membranes showing facilitated transport capabilities; three
focus on membrane contactors; and only two deal with conventional liquid
membranes. Each group merits more discussion.
The group loosely defined as solid membranes includes molten salt

membranes, which are much less volatile than ordinary liquid membranes, and
gels, which have higher viscosity and a more pronounced structure. The three
projects on solid facilitated diffusion parallel similar efforts that might be carried
out with other types of membranes. That for air, which is most important, will
compete with cryogenic distillation and with pressure swing adsorption. That for
olefins competes with adsorption and extraction. The study of solid transport
mechanisms is more fundamental and longer term.
Interestingly, all three solid membrane topics probably require more

chemistry than engineering. For example, for air separations, the goal should be
the synthesis of still more oxygen-complexing species stable under process
conditions. At present, the goal need not include optimal membrane geometry,
mass transfer and concentration polarization, and mathematical modeling. We
need a process which has the potential of working well before we can demand its
optimization.
The second group of topics, dealing with membrane contactors, have a

bastardized background but a bright future. The modules used for membrane
contactors are based on cheap, simple hollow fibers, not expensive, exotic units.
Such modules can contain liquid membrane solutions with attractive properties.
Membrane contactors will be a significant growth area in the next five

years. As designs are optimized, many applications will appear where substantial
energy savings are possible. Membrane contactors seem immediately valuable for
drugs, where they can give extractions of one hundred stages for less investment
= than conventional equipment containing five stages. Applications to gas
absorption are less certain. Flue gases are the best defined short-term target,
but aeration and chlorinated hydrocarbon stripping may also prove attractive long-
term targets.
Table 4-5.Future Research Directions for Facilitated Transport
Prospects for
Research Topic Importance Realization Comments
Solid facilitated oxygen

selective membranes 10 GOOd Air separations of higher
selectivity are a target common
to all types of membranes.
Optimal design of
membrane contactors a Excellent As membranes get better, module
design maximizing mass transfer
per dollar becomes key.
Solid facilitated olefin-

selective membranes Fair Membrane life is the key
question, especially with
sulfide contaminants.
Membrane contactors for

copper and uranium 5 Excellent Dramatic successes for drugs can
be repeated with metals and
possibly with toxic waste.
Membrane contactors for

flue gases and for aeration 4 Good Success in the field is less
certain than success in the lab.
Mechanisms of solid
facilitated transport 4 Fair Chemical reactions for
faciiitated transport promise
improved selectivity.
Enhanced liquid membrane

stability Fair Even if successfully made,
liquid membranes have not proved
superior to extraction.
Liquid membranes for

biotechnology I Excellent Systems work, but applications
are very small scale; chances of
acceptance are remote.
The last two topics in Table 4-5, both of which involve conventional liquid
membranes, are of lower priority. That on membrane stability, which is the
more important, should include setting targets for membrane stability, and
measuring how closely existing membranes approach these targets. At present,
although everyone agrees that stability is crucial, no one has considered whether
a stability measured in weeks, months or years would be technically or
economically acceptable. Finally. although liquid membranes will give good
separations of biochemicals. they are unlikely to be used commercially for the
reasons detailed in Section 4.8.3.
A possible time line describing these efforts is shown in Figure 4-6. The
clear region in this figure represents fundamental research; the shaded region is
development work; and the stippled region indicates practice. The width of the
bars represents the amount of effort merited.
While this line is speculative, its implications seem reasonable. Liquid

membranes represent a small, almost entirely fundamental effort which will get
smaller as other forms of facilitated diffusion become more attractive.
Membrane contactors already have their fundamentals in place; they should see
rapid development and commercialization. Facilitated transport in solid
membranes both has the greatest promise and the greatest problems. It will
require the most fundamental work before it becomes practical.
Six areas where little research appears to be warranted are:
1. Alcohol-water separations;
2. Emulsion liquid membranes;
3. Facilitated diffusion of proteins;
4. Hydrocarbons, excluding olefins;
5. Mathematical models of liquid membranes; and
6. Mathematical models of staged membranes.
These topics are unlikely to produce significant advances in our use or knowledge
of facilitated transport.
Time 1990 2000 2010
Liquid I
Membranes I I
Membrane
Contactors
Solid
Membranes
Figure 4-6. A Proposed Time Line for Facilitated Transport. The clear region
represents fundamental research; the shaded region is development
work; and the stippled region indicates practice. The width of the
bars represents the amount of effort merited.
REFERENCES
1. Cussler. E. L., Diffusion, Cambridge, London (1984).
2. Izatt, R. M., R. L. Bruening. J. D. Lamb, and J. S. Bradshaw, “Design of

Macrocycles to Perform Cation Separations in Liquid Membrane Systems,”
paper at Second Annual National Meetinn of the North American Membrane
Society. Syracuse, N.Y., June, 1988.
3. N.N. Li, U.S. Patent 3,410,794 (1968).
4. Ward, W. J., “Analytical and Experimental Studies of Facilitated Transport,”

AIChE J,, 16 405 (1970).
5. Cussler, E. L., “Membranes Which Pump,” AIChE J. 17, 1300-1303 (1971).
6. P.R. Danesi, L.Reichley-Yinger and P.G. Richer& “Life-Time of Supported

Membranes,” J. Memb. Sci. 31, 117-145 (1987).
7. A.M. Neplenbroek, D. Bargeman and C.A. Smolders, “Stability of Supported

Liquid Membranes,” paper presented at 7th Eurooean Summer School in
Membranes Science, Twente University, June, 1989.
8. Noble, R. D., C. A. Koval, and J. J. Pellegrino. “Facilitated Transport

Membrane Systems,” Chem. Enar. Pro&, 85(3) 58-70 (1989).
9. Draxler, J. W. Furst, and R. Marr, “Separation of Metal Species by Emulsion

Liquid Membranes,” J. Memb. S& 38, 281 (1989).
10. Sharma, A., A. N. Goswani, and R. Kuishma, “Use of Additives to Enhance

the Selectivity of Liquid Surfactant Membranes,” J. Memb. Sci,. 40, 329
(1989).
I I. Nakashio, F., M. Gato, M. Matsumoto, J. Irie, and K. Kondo, “Role of

Surfactants in the Behavior of Emulsion Liquid Membranes - Development of
New Surfactants,” J. Memb, Sci., 38. 249 (1988).
12. Yang, M.C., and E. L. Cussler, “Artificial Gills,” J. Memb. Sci., 42. 273
(1989).
13. Prasad, R., and K. K. Sirkar. “Dispersion-Free Solvent Extraction with

Microporous Hollow-Fiber Modules,” AIChE J, 33. 1057 (1987).
14. Stein, W. D., Transoort and Diffusion Across Cell Membranes, Academic
Press, Orlando, 1986.
15. Li, N. N., R. P. Cahn, and A. L. Shier, “Water in Oil Emulsions Useful in
Liquid Membrane,” U.S. Patent #4,360,448, Nov. 23. 1982.
16. N.N. Li, R.P. Cahn. D. Naden and R.W.M. Lai, “Liquid Membrane Processes
for Copper Extraction,” Hvdrometall., 277 (1983).
17. E.S. Matulevicius and N.N. Li, “Facilitated Transport Through Liquid
Membranes,” a and Pur. Methods 4:L, 73 (1975).
18. Baker, R. W., “Extraction of Metal Ions from Aqueous Solution,” U.S. Patent
#4,437,994, Mar. 20, 1984.
19. R.J. Bassett and J.S. Schultz, “Nonequilibrium Facilitated Diffusion of Oxygen
Through Membranes of Aqueous Cobaltodihistidine.” Bipchim. Bioohvs. Acp
u 194 (1970).
20. I.C. Roman and R.W. Baker, “Method and Apparatus for Producing Oxygen
and Nitrogen and Membrane Therefor,” U.S. Patent 4.542.010 (September 17,
1985).
21. R.W. Baker, I.C. Roman and H.K. Lonsdale, “Liquid Membranes for the
Production of Oxygen-Enriched Air. I. Introduction and Passive Liquid
Membranes,” j. Memb. Sci. 31 15-29 (1987).
22. B.M. Johnson, R.W. Baker, S.L. Matson. K.L. Smith, I.C. Roman, M.E. Tuttle
and H.K. Lonsdale, “Liquid Membranes for the Production of Oxygen-Enriched
Air. II. Facilitated-Transport Membranes,” 1, Memb. Sci. 31 31-67 (1987).
23. Nishide, H., M. Ohyanagi, 0. Okada, and E. Tsuchida, “Dual Mode Transport
of Molecular Oxygen in a Membrane Containing a Cobalt Porphyrin Complex
as a Fixed Carrier,” Macromolecu&, 20. 417-422, (1987).
24.. Nishide, H., and E. Tsuchida. “Facilitated Transport of Oxygen Through the
Membrane of Metalloporphyrin Polymers,” paper at &co d Annual National
Me,etina of the North American Membrane Society. SyraIuse, N.Y., June,
1988.
25. Kimura, S. G.. W. J. Ward, III. and S. L. Matson, “Facilitated Separation of

a Select Gas Through and Ion Exchange Membrane,” U.S. Patent #4,318,7 Id,
Mar. 9, 1982.
26. Matson, S. L.. K. Eric, L. Lee, D. T. Friesen, and D. J. Kelly, “Acid Gas
Scrubbing by Composite Solvent-Swollen Membranes,” U.S. Patent e4.737.166,
Apr. 12. 1988.
27. Hughes, R. D.. J. A. Mahoney, and E. F. Steigelman, “Olefin Separation by

Facilitated Transport Membranes,” Recent Devel. Seo. $&, 9. I73 (1986).
28. Teramoto, M.. H. Matsuyama. T. Yamashiro, and S. Okamoto, “Separation of

Ethylene from Ethane by a Flowing Liquid Membrane Using Silver Nitrate
as a Carrier,” J. Memb. S.&, 45. 1I5 (1989).
29. Pez. G. P., et al., U.S. Patent 4.761.164, 1988.

30. Pez, G. P., and L. V. Laciak, “Ammonia Separation Using Semipermeable

Membranes,” U.S. Patent #4,762,635, Aug. 9, 1988.
31. Pez, G. P., and R. T. Carlin. “Method for Gas Separation,” U.S. Patent
lr4,617,029, Oct. 14, 1986.
32. Dane& P. R., “Separations by Supported Liquid Membrane Cascades.” U.S.

Patent #4,617,125, Oct. 14, 1986.
5. Reverse Osmosis
by R. L. Riley. Separation Systems Technology, Inc.
5.1 .l The Basic Process
Reverse osmosis (RO), the first membrane-brued separation process to be

widely commercialized, is a liquid/liquid separation process that uses a dense
semipermeable membrane, highly permeable to water and highly impermeable to
microorganisms, colloids, dissolved salts and organia. Figure 5-1 is a schematic
representation of the process.
A pressurized feed solution is passed over one surface of the membrane. As

long as the applied pressure is greater than the osmotic pressure of the feed
solution, “pure” water will flow from the more concentrated solution to the more
dilute through the membrane. If other variables are kept constant. the water
flow rate is proportional to the net pressure. The osmotic pressures of some
typical solutions are listed in Table 5-1.
Table S-l. Typical Osmotic Pressures at 25’C (77-F)
Concentration Concentration Osmotic Pressure

Compound (mg/L) (moles/L) (psi)
NaCl 35,000 0.60 398

Seawater 32,000 340
NaCl 2,000 0.0342 22.8
Brackish water 2-5.000 15-40
NaHCOs 1,000 0.0119 12.8
Na SO, 1,000 0.00705 6.0
Mg SO, 1,000 0.0083 1 3.6
MgCI, 1,000 0.0105 9.7
CaCI, 1,000 0.009 8.3
Sucrose 1,000 0.00292 1.05
Dextrose 1,000 0.0055 2.0
Reverse osmosis is widely used to remove salts from seawater or brackish

water. An ideal desalination membrane would allow unhindered wafer passage,
with complete rejection of dissolved salts. In practice, both water and salts
cross the membrane barrier, but at differing rates. There are two principal
models for the transport process: the solution-diffusion model, first described by
Lonsdale, Merten and Riley’, and the capillary pore model, described by
Sourirajan,’ both of which describe the reverse osmosis process. However, the
solution-diffusion model is more widely accepted.
276
Reverse Osmosis 277
SCHEMATIC OF THE REVERSE OSMOSIS PROCESS
Pressurized
feed Concentrate
(200-1.000 PSI)
Permeate
Figure 5-I. Schematic of the reverse osmosis process.

The performance characteristics that describe a reverse osmosis system are

the product water flux, the salt flux, the salt rejection and the water recovery.
The equations that define these terms are given below. The water flux,
F, (g/cmfsec), and the salt flux, F, (g/cm2sec), are linked to the pressure and
concentration gradients across the membrane. From the solution-diffusion model,
these terms are described by the equations:
F, = A(Ap-Ax) (I)
where A is the water permeability constant (g/cm2secatm), Ap is the pressure

differential across the membrane (atm), and Ax is the osmotic pressure
differential across the membrane. and
F, = B(c,-c2)
where B is the salt permeability constant (cm/set), and Ct-Cs is the salt
concentration difference across the membrane (g/cmS). Thus, the water flux is
proportional to the applied pressure but the salt flux is independent of the
applied pressure. This means the membrane appears to become more selective as
the pressure is increased. In reverse osmosis, the membrane selectivity is
normally measured by a term called the salt rejection, R, which is proportional to
the fractional depletion of salt in the product water compared to the feed. The
salt rejection is defined as
x 100% ( (3)
where C, is the concentration of the product water and Cr is the concentration

of the feed water. Finally. the fraction of the feed water that is produced as a
purified product is called the % recovery of a reverse osmosis system, and is
defined as
% recovery = ~xXOO% , (4)

Nf
where N, is the product flow rate from the reverse osmosis system and Nr is the
feed solution flow rate to the reverse osmosis system.
Figure 5-2 shows the effects of pressure, temperature, and water recovery
on the membrane flux and product water quality of reverse osmosis membranes.
These are generalized curves and deviations can be expected depending on the
nature and concentration of the feed solution. Figures 5-2a and 5-2b show the
effect of applied pressure on membrane flux and product water quality. Figures
5-2~ and 5-2d show the effect of feed temperature on membrane flux and product
water quality. Finally. Figures 5-2e and 5-2f show the effect of water recovery
on membrane flux and product water quality.
Reverse Osmosis 279
EFFECT of APPLIED PRESSURE,

FEED TEMPERATURE and WATER RECOVERY on the
MEMBRANE FLUX and PRODUCT WATER QUALITY
of REVERSE OSMOSIS MEMBRANES
lb)
El
64
Membrene Product
llux quality
Prerrure Pressure
(4 I I (4
El
Membrsne Product
I/i
flux quality
Temperature Temperature
(4 (9
Membrane Product
flUX
Water recovery Water recovery
Figure 5-2. Schematic curves showing the effect of pressure, temperature,

and water recovery on the performance of reverse osmosis
systems.
An efficient reverse osmosis process, that is a process that achieves a high

water flow at low energy expenditure, should meet the following requirements:
a) High intrinsic water permeability with low intrinsic salt permeability.
b) Large membrane surface area.
c) Low pressure drop across the membrane.
d) Very thin, defect-free membrane.
5.1.2 Membranes
Loeb and Sourirajan developed the first asymmetric, cellulose diacetate

membranes.‘*’ A major research and development effort, using their work as a
basis, took place through the 1960s and 197Os, with substantial sponsorship by the
Department of the interior, Office of Saline Water. Figure 5-3 summarizes the
key developments that have occurred over the past thirty years in membrane
development for desalination. Today, asymmetric, flat-sheet membranes made from
cellulose triacetate/cellulose diacetate blends are widely used in the industry for
brackish water treatment. Cellulose triacetate hollow fibers are in limited use for
seawater treatment, but most seawater plants use hollow-fiber or more commonly
thin-film composite membranes made from aromatic polyamides. Thin-film
composites made from aryl-alkyl polyetherurea and other polymers are also used.
Cellulose acetate membranes are rugged, resistant to moderate levels of

chlorine and inexpensive. They are susceptible to bacteriological attack, and to
hydrolysis when exposed to strongly basic solutions. The newer materials offer
better performance than cellulose acetate membranes in some circumstances.
Some have negatively charged surfaces, which enhance their rejection of sulfate,
carbonate or phosphate ions, for example. However, they tend to be less
chlorine-resistant and more susceptible to oxidation. Table 5-2 summarizes the
properties of the two main types of commercial polyamide and polysulfone
composite membranes.
5.1.3 Modules
Reverse osmosis membranes are prepared in the form of hollow fibers or flar
sheets. The membranes are then packaged into modules designed to enable a
large membrane area to be contained in a small volume. Hollow fibers are made
into bundles, several of which are contained within one housing, to form a single
module, or permeator. The feed solution flows on the outside of the fibers, in
contact with the dense skin layer. DuPont is the principal company
manufacturing hollow-fiber modules. Figure 5-4 shows a schematic drawing of a
DuPont module. This module is designed for single-stage seawater desalination,
and it contains 2.3 million individual fibers enclosed in an S-in-diameter, 5-ft-long
pressure vessel. Hollow fiber modules offer a very high membrane area at a
relatively low cost. However, these modules are susceptible to fouling by
suspended matter in the feed solution and this is a major operating problem.
FluId SySl_
I
IWO
I
I- I-
, ,
WrfnIcm
I I I,
F
I I ,I
1
1970
Fk
-
1, , , , ( , ,
I980
,
I
( * , , , ) , , , ,
I
1990
, , ( , , , , , 1
2000
Figure 5-3. Key developments in reverse osmosis membrane development.

Table 5-2. Comparative Properties of Commercial Thin-Film Composite Membranes
Comparative Properties of Commercial Thin-film
Composite Membranes
Crosslinked
Aryl-Alkyl Fully Aromatic
Polyetherurea Polyamide
Seawater Capability Yes Yes
Brackish Water Capability Yes Yes
Low Pressure Capability Marginal Yes
Thin-film Thickness (A)*
Seawater 500 2000
Brackish 250 2000
Low Pressure ~250 (Marginal) 2000
Intermediate Layer Between
Thin Film and Support Yes No
Silica Rejection Moderate High
Organic Rejection Moderate High
Chlorine Tolerance None Limited
Oxidant Tolerance None Limited
Temperature Stability Process Dependent Stable to 45 C
Membrane Surface Charge Positive Negative
Cleaning Agents (anionic) Incompatible Compatible
Cleaning Agents (cationic) Compatible Incompatible
Acid Cleanrng Limits pH 3.5 PH ’
Alkaline Cleaning Limits pH 12 pH 12
Angstrom Units: 1 micron = 10,000 angstroms

Reverse Osmosis 283
Figure 5-4. Schematic drawing of a DuPont hollow-fiber module for single-

stage seawater desalination.
Spiral-wound modules are the more widely used configuration throughout the
industry, principally because they are less susceptible to fouling. In these
modules, flat-sheet membranes are made in long rolls, 40 or 60 inches wide.
Between two and six spiral-wound modules are then placed in a single pressure
vessel. A schematic of a spiral-wound module and pressure vessel assembly are
shown in Figure 5-5. The features and advantages of spiral-wound modules for
use in reverse osmosis are:
Features Advantw
Large membrane surface area Compact membrane plants
Compact packaging Low initial capital costs
Optimum feedwater passages High surface flow velocity

Reduced polarization effects
Higher permeate output
Increased salt rejection
Simple, rugged, disposable design Low replacement costs
Reverse osmosis spiral-wound modules were originally designed to operate

optimally at 12-17 gailon/ft2-day. Higher flow rates would be desirable, but lead
to major fouling problems. The nature of the spacer materials, especially the
feed channel spacer material, is also an important aspect of reverse osmosis
technology.
5. I :4 Systems
The factor which has the greatest influence on reverse osmosis membrane
system design is fouling, caused by particulate and colloidal matter that become
concentrated at the membrane surface. Pretreatment is used to remove
particulate matter from the feed water, but is seldom completely effective. The
concentration of foulants at the membrane surface increases with increasing
permeate flux and product recovery rate. A system designed to operate at a high
permeate flux is likely to experience high fouling rates and will require frequent
chemical cleaning.
A measure commonly used to determine the clarity of the feed water and its
membrane fouling potential is the Silt Density Index (SDI). Determining the feed
water’s SD1 involves measuring the necessary time to filter a fixed volume
(usually 500 mL) of feed water through a clean filter pad. The test is repeated
with the same pad 15 minutes after the first test. The SDI is calculated from the
change in time required to filter the feed water sample through the clean versus
the fouled filter pad. To make SD1 measurements comparable, a constant applied
pressure of 30 psig and a Millipore 0.45-pm filter pad are commonly used for this
test.
the product water flows through the
porous matertal I” a rptral path until
I, contacts and flows through the holes
I” the product water tube.
9
3
,o
-.
v)
Figure 5-5. A schematic drawing of a spiral-wound module and pressure vessel

assembly.
A very low SDI is particularly important for hollow-fiber modules, and a

feed water SD1 of 3.0 is considered the maximum value to which these modules
can be exposed. Spiral-wound elements are more tolerant to particulate matter
fouling and a feed water SD1 of 3.0-5.0 is normally specified. Obviously, a lower
value for the feed water is beneficial to both types of membrane systems.
Typical pretreatment processes are media filtration (sand, anthracite, etc.)

and cartridge filtration. The media filters may be preceded by coagulation and
sedimentation for highly turbid waters.
Chemical dosing is also required for scale control and for dechlorination if
the raw seawater is chlorinated. Typical chemicals are:
l Sulfuric acid for pH control and calcium carbonate scale prevention.
l Antiscalants/scale inhibitors for prevention of calcium carbonate and
sulfate scale.
l Sodium bisulfite for dechlorination and inhibition of fouling.
Typical design guidelines for spiral-wound polyamide type membrane elements in a
seawater desalination system are shown in Table 5-3.
Table 5-3. Typical Design Guidelines for Spiral-wound Modules using

Polyamide-type Thin-film Composite Membranes
Silt density index 3-5 (or less)

Maximum permeate flux 20 gal/ft2-day
Maximum water recovery for 40-inch elements 15%
Maximum permeate flow rate per 40-inch element
8 x 40-in. 6,000 gal/day
4 x 40-in. I.500 gal/day
2.5 x 40-in. 500 gal/day
Maximum feed flow rate for 40-inch elements
8 x 40-in. 60 gal/min
4 x 40-in. I6 gal/min
2.5 x 40-in. 6 gal/min
Standard salt rejection (avg.) 99.5%

Standard salt rejection (min.) 99.3%
Figure 5-6 shows a diagram for a typical reverse osmosis desalination

plant, which could apply to either seawater or brackish water desalination. In
this flow diagram, the pressure vessel does not necessarily indicate a single
vessel, but generally is an array of pressure vessels, as shown in Figure 5-7.
For brackish water applications, each six-element pressure vessel is designed to
operate at 50% water recovery. Thus, assuming a feed flow of 100 gal/mitt, the
module array in Figure 5-7a represent a 75% water recovery system, while the
array in Figure 5-7b represents 87.5% recovery. Obviously, a system operating at
the highest possible water recovery is desirable, since the overall operating costs
will be reduced. The level of water recovery that can be attained is dictated by
the feed water composition. In the absence of high concentrations of salts,
particularly sulfates, carbonates, etc. that can cause scaling, high water recovery
Reverse Osmosis 287
LOW
luctlon
Pump discharge
Figure 5-6. Flow diagram of a typical reverse osmosis desalination plant.

TWO POSSIBLE SPIRAL-WOUND MODULE ARRAYS for RO
Concentrate
Foed-
Product
Figure 5-7. Two possible spiral-wound module arrays for reverse osmosis.
Assuming a feed flow of 100 gal/min, a) represents a 75% water
recovery system and b) represents 87.5% recovery.
Reverse Osmosis 289
levels generally can be attained. For seawater desalination systems, however, the
overall water recdvery is limited to about 45%, due to the high osmotic pressure
of the seawater brine in the final array of the system. At this point the osmotic
pressure approaches the applied pressure and. consequently, the net driving force
becomes so small that the permeate flow becomes uneconomically small. This
occurrence will be discussed later, along with the effect on plant operating costs.
5.2 THE REVERSE OSMOSIS INDUSTRY
5.2.1 Current Desalination Plant Inventory
Worldwide, more than 6,235 desalting plants were in operation at the

beginning of 1988, with a total capacity of 3,178.500,000 gallons per day.
Figure 5-8 shows a breakdown of this total with respect to desalination methods,
worldwide location, and type of feedwater.
Presently, there are about 750 desalination plants operating in the United
States with individual capacities in excess of 25,000 gallons per day. This
translates into a combined capacity of about 200 million gallons per day. Reverse
osmosis accounts for about 75% of this capacity, with about 70% of the plants
used for industrial purposes. There are many smaller reverse osmosis plants in
operation, but their combined capacity is relatively low. These applications
include point-of-use home RO systems, mobile military units, pleasure boats and
merchant ships, small industries, off-shore drilling rigs, etc.6
5.2. I. 1 Membrane sales
Recent figures indicate that sales of desalination membranes by U.S. industry

are approximately 585 million/year. Even with an increasing share of the world’s
desalination market, U.S. manufacturers have been hit hard by stiff competition
and declining profits. The domestic membrane industry has experienced great
change with an overall trend toward fewer, but generally larger, companies
during the last few years. Some firms have been acquired by larger chemical
corporations involved in water treatment and/or process separation. Price cutting,
particularly in spiral elements, has been severe, thereby lowering gross margins
and making membrane products commodity items. Declining profits have forced
some firms to go out of business.
The U.S. desalination membrane market continues to be dominated by sales

of small to moderate-size RO plants. replacement membrane elements, and plants
for military uses. Sales of home water RO units continue to climb rapidly,
encouraged by public concerns about water pollution. Potential international
competition in the U.S. market, especially from the Japanese, is a concern. Some
efforts have been made to introduce membrane systems into domestic industrial
applications, primarily aimed at process waste-water treatment and recovery of
valuable materials. Since these waste streams are so diverse, major investments
are often required to develop such markets in industries where other technologies
are entrenched. However, many of these older technologies are energy intensive.
e.g., distillation or other evaporative methods, so the benefits of reverse osmosis
will become increasingly important as energy prices rise. More stringent
application of discharge limits under the EPA’s National Pollutant Discharge
Elimination System will also help the market to grow. As the cost of membrane
CURRENT WORLDWIDE DESALINATION PROCESS CAPACITIES
75
DESALINATION
CAPACITY
50
(%)
25
METHOD LOCATION FEEDWATER
Figure 5-8. Worldwide desalination process capacity, location and feedwater

type.’ (MSF: multistage flash, RO: reverse osmosis).
Reverse Osmosis 291
systems decrease, accompanied by energy reductions in low-pressure operations,

desalination of industrial waste streams will be a more attractive alternative to
established treatment systems, especially in areas where water supplies are limited
and water reuse becomes a necessity.0
Estimated worldwide 1988 sales of reverse osmosis membrane are shown in

Table 5-4. The figures have been adjusted for recent fluctuations in the U.S.
dollar exchange rate.
Table 5-4. The Reverse Osmosis Membrane Industry
Estimated Sales of Company Manufactured

Membrane Products (S millions)
Hollow Spiral
Company Location Fiber Elements Total
DuPont USA 24, 5 29

Dow/Filmtec USA __ 26 26
Fluid Systems/
UOP USA __ 13 13
Hydranauticsl USA/
Nitto-Denko Japan __ 12 12
Desalination
Systems USA __ 5 5
Millipore USA __ 3 3
Osmonics USA __ 3 3
Toray Japan __ I2 12
Toyobo Japan 4* __ 4
Others World __ II II
Total 28 90 118
l Membrane permeator includes pressure vessel
Table 5-5 presents a breakdown of the total estimated membrane sales by

membrane type. Pressure vessel sales are not included, except for hollow-fiber
membrane permeators, which include the pressure vessel as a complete package.
5.2.2 Marketing of Membrane Products
The previous discussion has focused on membrane sales only, ignoring plant
design and operation. At the present time, only one major membrane company,
Nitto-Denko/Hydranautics. supplies systems other than small pilot units. All other
domestic companies have abandoned plant and system sales, due to lack of
profitability and the necessity of maintaining a large engineering and equipment
fabrication staff. Plant design and construction is handled by original equipment
manufacturers (OEMs).
Table 5-5. Estimated 1988 Reverse Osmosis Membrane Market Share by

Membrane Type
% of Total $ (millions) Trend
Membrane ConfiauratlQn
Hollow fibers 26.2 2s Decreasing

Spiral-wound 73.8 90 Increasing
Membrane TVD~
Spiral polyamide composite 45.4 52.7 Increasing

Cellulose acetate blend 28.3 37.3 Stable
Replacement Market
Spiral-wound (all types) 20.0 44.5 Increasing
Table 5-6 lists representative OEMs and engineering companies engaged in

reverse osmosis plant design.
Table 5-6. OEMs and Engineering Firms Involved in Reverse Osmosis Plant
Design and Construction
OEMs Enaineerina Comoan&
Aqua Design Bechtel

Arrowhead Black and Veach
Culligan Burns and Roe
Gaco Systems, Inc. CHZM Hill, Inc.
Graver Water Systems Stone and Webster
lonics VBB-SWECO
L.A. Water Treatment
Polymetrics
The OEM designs and builds the plant according to the guidelines for
elements, vessel arrays and operating conditions set by the membrane
manufacturer. Most membrane elements are standardized as far as size and flow
rates are concerned, and are thus interchangeable, so OEMs buy freely from all
membrane manufacturers. Customers typically purchase from the OEM tendering
the lowest bid. In these competitive conditions, it is hard for both membrane
makers and OEMs to ensure that recommended operating standards are adhered to.
This is an industry weakness.
Reverse Osmosis 293
5.2.3 Future Direction of the Reverse Osmosis Membrane Industry
In the next decade, the large membrane-producing companies will increase

market share by acquiring technology from smaller. more innovative companies.
Thus, the number of membrane companies will decrease. At present, in spite of
the size and softness of the membrane market, a number of large corporations
with little, if any, membrane experience are anticipating entry into the business.
Since the development of the crosslinked, fully aromatic polyamide thin-film

composite membrane by Cadotte in 1977, emphasis has focused on this membrane
type throughout the industry. Several dominant membrane companies have
directed their attention to methods of circumventing the patents protecting this
technology, in lieu of research and development on new membranes with superior
properties, principally chlorine resistance.
The ultimately successful companies will be characterized by:
l Ability to make high-quality products, consistently, reliably, efficiently.

l Strong proprietary and patent positions.
l Strong, effective R&D effort.
. Ability to offer oxidation-resistant, high-performance membranes.
. Technically trained, customer-oriented marketing and support operation.
l Development of strong OEMs with product loyalty.
l International marketing and manufacturing capability.
These attributes will be achieved by effective, dedicated management rather than

by acquisition.
5.3 REVERSE OSMOSIS APPLICATIONS
Figure 5-9 shows estimated reverse osmosis membrane sales worldwide for
1988, by application area. Projections for 1998 are also given.
The largest single application area at present is desalination of seawater and

brackish waters, which accounts for about 50% of total sales. The remaining
sales are directed to diverse applications, ranging from industrial process water,
to water for medical use and for food processing. Applications showing the
highest projected growth rate percentages are purification of relatively dilute
streams such as industrial process water, ultrapure/medical, and small package
sales. Lowest projected increases are in seawater, brackish water, and food
applications.
The sales of home “under the sink” reverse osmosis units is a rapidly rising
segment of the reverse osmosis market in the United States and Europe.
However, these systems are very wasteful of water and use 5-10 gallons of tap
water for each gallon of water processed, thereby dumping what is generally
considered potable water. Sales of these units could, therefore, be regulated in
the future.
GROWTH RATE (%I

7.5 13 12 9 15
!
I7
I
11
I
XX
! I I !
loo- ’
MEMBRANE 75
SALES
L
i”
($, millions) t
50
25 S118M
0 /
E Total
._
Ii
f
m
Figure 5.9. Principal applications of reverse osmosis and estimated worldwide

sales of membrane products (projected annual growth rate, 1988-
1998).
Reverse Osmosis 295
Although reverse osmosis membrane sales for food processing applications

represent only a small percentage of the total sales, with a modest projected
growth rate, the food industry is very energy intensive. Thus adoption of
alternate low-energy separations such as reverse osmosis could result in
considerable energy savings. Many of the food materials requiring separation are
dilute suspensions and solutions, and are ideal candidates for concentration by
reverse osmosis. Reclamation of wastewater and high-temperature process waters
are good applications.e
To date, three U.S. companies have obtained FDA approval for use of their
membrane modules in food processing applications. These are:
. The DuPont B-15 linear aromatic polyamide hollow fiber permeator.

l The FilmTec/Dow Ff-30 crosslinked fully aromatic polyamide and NF-40
polypiperazineamide thin-film composite spiral-wound membrane elements.
l The Fluid Systems/UOP thin-film composite spiral-wound aryl-alkyl
polyetherurea membrane elements.
U.S. industry consumes about 8 billion gallons of fresh water daily; industrial
and commercial facilities discharge about I8 billion gallons of wastewater daily,
mostly into open water supplies or sewage systems! Both supply of high quality
water and treatment of wastewater offer opportunities for reverse osmosis
technology. Table 5-7 lists industries and reverse osmosis markets within that
industry, grouped by their maturity status. There is significant potential for
applying desalination technology to reclamation and reuse of wastewaters. If
drought conditions continue in the Western United States, desalination of brackish
waters or wastewaters may become increasingly attractive, but relatively high
costs will make the installation of seawater desalination plants unlikely.
5.4 REVERSE OSMOSIS CAPITAL AND OPERATING COSTS
Four techniques can be used to remove salts from water:
l Distillation - Multiple-effect (ME), Multi-stage flash (MSF). Vapor

compression (VC) and Solar.
l Reverse Osmosis (RO).
l Electrodialysis (ED).
. Ion Exchange (IX).
Selection of the most appropriate technology depends on many factors

including the water analysis, the desired quality of the treated water, the level of
pretreatment required, the availability of energy, and waste disposal. Both
reverse osmosis and electrodialysis membranes can be tailored to the application,
based on the feed water composition. Table 5-8 shows the various methods
available for desalination based on the nature of the feed. Ion exchange and
electrodialysis are generally preferred for very dilute streams while distillation is
preferred for very concentrated streams. Reverse osmosis is the best technique
for streams containing 3,000 to 10,000 ppm salt but is still widely used for
streams more dilute than 3,000 ppm and streams as concentrated as seawater at
35,000 ppm salt.
Table S-7. Reverse Osmosis Market Applications
Status Industry Applications
Mature Desalination Potable Water Production

Seawater
Brackish Water
Municipal Wastewater Reclamation
Ultrapure Water Semiconductor manufacturing

Pharmaceuticals
Medical Uses
Growing Utilities and Power Boiler Feedwater

Generation Cooling Tower Blowdown Recycle
Point of Use Home Reverse Osmosis
Emerging Chemical Process Process Water Production and Reuse

Industries Effluent Disposal and Water Reuse
Water/Organic Liquid Separation
Organic Liquid Mixtures Separation
Metals and Metal Mining EffIuent Treatment

Finishing Plating Rinse Water Reuse and
Recovery of Metals
Food Processing Dairy Processing

Sweeteners Concentration
Juice and Beverage Processing
Production of Light Beer and Wine
Waste Stream Processing
Textiles Dyeing and Finishing

Chemical Recovery
Water Reuse
Pulp and Paper Effluent Disposal and Water Reuse
Biotechnology/ Fermentation Products Recovery and

Medical Purification
Analytical Isolation, Concentration, and

Identification of Solutes and
Particles
Hazardous Substance Removal of Environmental Pollutants

Removal from Surface and Groundwaters
Reverse Osmosis 297
Table 5-8. Methods of Desalination
Technique Typical Applications
Brackish Water Seawater Higher Salinity

Brines
3,000 3,000- 10.000 35,000

ppm ppm ppm
Distillation t s P P
Electrodialysis P S
Reverse osmosis P P :, s
Ion exchange P
Key: P = Preferred application

s = Secondary application
t = Technically possible, but not economical
Source: Office of Technology Assessment, 1987
Although desalination costs have decreased significantly in the last 20 years, cost
remains a primary factor in selecting a particular desalination technique for water
treatment. Figure S-10 shows a comparison of estimated desalination operating costs
based on plant size. These costs include capital and operating costs for plants
producing I to 5 million gallons per day of potable water (in 1985 dollars). costs
include plant construction (amortized over 20 years), pretreatment, desalination, brine
disposal, and maintenance. As Figure 5-10 shows, there is a significant decrease in cost
with increasing plant size for seawater desalination, but only a slight decrease in cost
with increasing plant size for brackish water plants.
Table 5-9 shows the current capital and operating costs for seawater and brackish
water desalination plants. These costs are strongly influenced by several factors, as
previously indicated. Consequently, the range of the total costs does not closely agree
with the costs shown in Figure 5-10.
Table 5-9. Capital and Operating Costs for Seawater and Brackish Water
Desalination by Reverse Osmosis
Capital Costs Operating Costs

Type of Water (S/gal-day) (S/l000 gal of product)
Seawater 4.00 - 10.00 2.50 - 4.00
Brackish 0.60 - 1.60 1.00 - 1.25

10
:L
9
L
0
DESALINATION 7 MSF
Seawater
COST 6
R
($/l,OOO gal)
5
Brackish
\ RO 6 ED water
1 I I I I
5 10 15 20 25 30
PLANT SIZE (MG/day)
Figure S-10. Comparison of estimated desalination operating costs based on Plant size.
(MSF: multistage flash, RO: reverse osmosis, ME: multiple-effect
evaporation, ED: electrodialysis).
Reverse Osmosis 299
The estimated energy requirements associated with the various types of

desalination processes are shown in Table 5-10. These energy costs are a function of
feed solution salt concentration. The effect of the salt concentration in brackish water
on energy usage is less marked with reverse osmosis than with other technologies. The
principal effect as far as RO is concerned is an increase in the osmotic pressure of the
feed solution. Higher pressures (400 psi) are used for desalination of brackish feeds, in
the 3500-8000 mg/L range. Lower pressure operations are used with low feedwater
concentrations.
Table 5-10. Estimated Energy Requirements for Desalination Processes
Energy Requirements
Process Type kWh/ms kWh/ 1,000 gal
Distillation I5 56
Electrodialysis
Seawater >50+ >,100*
Brackish Water 4 I5
Reverse Osmosis
Seawater 7 26
Brackish Water (400 net psi) 2 7.6
(200 net psi) 0.8-1.5 3.0 - 5.7
* Electrodialysis is not economical for seawater desalination. Energy consumption is

approximately 5 kWh for each 1000 mg/L salt reduction per each 1,000 gal. of purified
water.
The development of energy saving, low-pressure, thin-film composite membranes

has made reverse osmosis superior to low pressure distillation/evaporation processes for
seawater desalination. Energy recovery systems would enable further reductions in
energy consumption to be made.
5.5 IDENTIFICATION OF REVERSE OSMOSIS PROCESS NEEDS
5.5.1 Membrane Fouling
Membrane fouling is a major factor determining the operating cost and membrane
lifetime in reverse osmosis plants. If operated properly the life of reverse osmosis
membranes is excellent. Cellulose acetate blend membranes and polyamide thin-film
composite membranes have been operated without significant deterioration on brackish
water feeds for up to I2 years.
Several types of fouling can occur in reverse osmosis systems depending on the
feed water. The most common type of fouling is due to suspended particles. Reverse
osmosis membranes are designed to remove dissolved solids, not suspended solids. Thus,
for successful long-term operation, suspended solids must be removed from the feed
stream before the stream enters the reverse osmosis plant. Failure to remove
particulates results in these solids being deposited as a cake on the membrane surface.
This type of fouling is a severe problem in hollow-fiber modules.7
A second type of fouling is membrane scaling. Scaling is most commonly produced

when the solubility limits of either silica, barium sulfate, calcium sulfate, strontium
sulfate, calcium carbonate, and/or calcium fluoride are exceeded. Exceeding the
solubility limit causes the insoluble salt to precipitate on the membrane surface.
Chemical analysis of the feed stream is required to establish the correct reverse osmosis
plant design and operating parameters. It is essential that a plant operate at the design
recovery chosen to avoid precipitation and subsequent scaling on the membrane surface.
A third cause of membrane fouling is bacteria. Reverse osmosis membrane

surfaces are particularly susceptible to microbial colonization and biofilm formation.’
The development of a microbial biofilm on the feedwater membrane surface has been
shown to result in a decline in membrane flux and lower the overall energy efficiency
of the system. Bacterial attachment and colonization in the permeate channel has the
potential to rapidly deteriorate permeate water quality. In recent years, Ridgway and
coworkers have systematically explored the fundamental mechanism and kinetics of
bacterial attachment to membrane surfaces.g-12 The cumulative effects of membrane
biofouling are increased cleaning and maintenance costs, deterioration of product water
flow and quality, and significantly reduced membrane life.
Other types of fouling which can occur are the precipitation of metal oxides, iron
fouling, colloidal sulfur, silica, silt, and organics.
All of these membrane fouling problems are controlled by using a correct plant
design including adequate pretreatment of the feed and proper plant operation and
maintenance. It is now well accepted that scaling and fouling were the major problems
inhibiting the initial acceptance of reverse osmosis.
The type of pretreatment required for a particular feed is quite variable and will
depend on the feed water and the membrane module used. Pretreatment may include
coagulation and settling and filtration to remove suspended solids, and disinfection with
chlorine is commonly used to kill microorganisms. When chlorine-sensitive, thin-film
membranes are used, dechlorination of the feed will then be required, a factor that
tends to inhibit the use of these membranes. Scaling is usually controlled by addition
of acid. polyphosphates, or polymer-based additives. These pretreatment steps are all
standard water treatment techniques, but taken together, their cost can be very
significant. Typically. the cost of pretreatment may account for up to 30% of the total
operating cost of a desalination plant.
Irrespective of the level of pretreatment, some degree of fouling will occur in a

reverse osmosis system over a period of time. Periodic cleaning and flushing of the
system is then required to maintain high productivity and long membrane life. With
good pretreatment, periodic cleaning will maintain a high level of productivity, as
illustrated in Figure 5-I I.ls When pretreatment is inadequate, frequent cleaning is
required. In addition, cleaning is then less effective in restoring membrane productivity,
leading to reduced membrane life. A membrane system should be cleaned when:
Reverse Osmosis 301
PRODUCT
FLOW
RATE
/ I 1;;;;: pretreated feed 1

--- Marginal pretreatment, periodic cleaning
lnade,quate pretreatment, frequent
OPERATING TIME
Figure 5-I I. Membrane productivity as a function of operating time.

l Productivity is reduced by more than 10%.

l Element pressure drop increases by 15% or more.
l Salt passage increases noticeably.
l Feed pressure requirements increase by more than 10%.
l Fouling/scaling situation requires preventative cleaning.
l Before extended shutdowns.
Chemical cleaning solutions have been developed empirically that are quite
effective, provided the system is cleaned on a proper schedule. Cleaning chemicals,
formulations, and procedures must be selected that are compatible with the membrane
and effective for removing the scale or foulant present. Due to the complexity of
most feedwaters, where both inorganic and organic foulants are present, the cleaning
operation generally is a two-step process. Inorganic mineral salts deposition usually
requires an acidic cleaning solution formulation, whereas organic foulants such as algae
and silt require alkaline cleaning. The selection of chemical cleaners is hindered by the
sensitivity of most reverse osmosis membranes to such factors as temperature, pH,
oxidation, membrane surface charge, etc. The development of more durable membrane
polymers would allow a more widespread use of strong cleaners.
5.5.2 Seawater Desalination
Currently, seawater plants are designed and operated with spiral-wound polyamide
type thin-film composite membranes in a single-stage mode at operating pressures of 800
to 1000 psi, producing 300 mg/L product water quality at rates of 9 gal/ft2-day and
water recovery levels of 4596.“’
Five classes of membranes are commercially used for seawater desalination today.
They are:
. Cellulose triacetate hollow-fine fibers.

. Fully aromatic linear polyamide hollow-fine fibers.
. Spiral-wound crosslinked fully polyamide type thin-film composite membranes.
. Spiral-wound aryl-alkyl polyetherurea type thin-film composite membranes.
. Spiral-wound crosslinked polyether thin-film composite membranes.
The performance characteristics of the dominant seawater desalination membranes

are shown in Figure 5-12.16
With the exception of cellulose triacetate hollow-fine fibers, the other membranes
are rapidly degraded in an oxidizing environment. This limitation creates major
problems in designing for effective pretreatment, cost of pretreatment chemicals, fouling
by microorganisms, decreased membrane performance, and reduced membrane life.
Reverse Osmosis 303
99.99 I I IIJ I I III I , I,
99.98 -
Crosslinked
Cellulose polyether
99.95 - triacetate
NaCl Minimum rejection

Rejection Crosslinked fully for single-stage
(%) aromatic polyamide seawater operation
99.3 --__ .-
Other thin-film
polyamide
95 -
90 1 I I III I I III , , ,,
0.01 0.1 1.0 10
Flux (m3 /m2 - day)
Figure S- 12. Performance characteristics of membranes operating on seawater at

56 kg/cm2, 2S”C.
To date, few seawater desalination plants have operated without difficulty. The
common factors in those plants that have operated efficiently are:
A reducing (redox potential) seawater feed.

An oxygen-free (anaerobic) seawater feed taken from a seawater well that has
been filtered through the surrounding strata.
High levels of sodium bisulfite in the seawater feed (40 to 50 me/L) to
remove oxygen and/or chlorine if the feed has been chlorinated.
Most difficulties have been encountered with plants operating from surface
seawater intakes where chlorination is required to control growth of both algae and
microorganisms. In these cases, chlorination must be followed by dechlorination to
protect the polyamide type membranes. Dechlorination is generally carried out by
adding sodium bisulfite at an amount slightly in excess (6 to 8 mg/L) of that
stoichiometrically required to remove the 1 to 2 mg/L residual chlorine that is present.
This amount of sodium bisulfite is insufficient to remove chlorine since oxygen, which is
present in seawater at approximately 7 to 8 mg/L, competes with chlorine for sodium
bisulfite. As a result, very small amounts of chlorine or oxidation potential remain, as
determined by redox measurements. Approximately 50 mg/L of sodium bisulfite is
required to totally remove the dissolved oxygen present in seawater. It is now known
that the presence of heavy metals, commonly deposited on the surface of these
membranes in a seawater desalination environment, accelerates degradation of polyamide
type membranes by chlorine. As a result, even small traces of residual chlorine can
rapidly degrade the membrane.
Attack of polyamide type membranes by dissolved oxygen in seawater in the

presence of heavy metals is also a concern. Japanese researchers have shown this type
of attack does occur with some membranes. They have concluded that heavy metals
catalyze the dissolved oxygen to an active state that is very aggressive toward the
membrane. With these membranes, both chlorine and dissolved oxygen are totally
removed with a combination of vacuum deoxygenation and sodium bisulfite.16
Because of these residual chlorine problems, the most successful polyamide

membrane seawater desalination plants employ large quantities of sodium bisulfite to
remove chlorine and dissolved oxygen and create a reducing environment to protect the
membrane. The addition of sodium bisulfite at the 50 mg/L levels is equivalent to
approximately 900 pounds of sodium bisulfate per million gallons of seawater processed.
This adds approximately $0.20 per thousand gallons to the cost of the product water.
This cost is clearly a disadvantage for the process, not to mention the logistics burden
of transportation and storage of large quantities of the chemical.
Low levels of copper sulfate have been used effectively to treat seawater feeds to
reverse osmosis plants to control algae. Its effectiveness as a biocide, however, has not
been well documented. Furthermore, the discharge of copper sulfate into the seawater
environment is undesirable.
Reverse Osmosis 305
Currently, seawater desalination by reverse osmosis is viewed with considerable

conservatism. However, as increasing energy costs make distillation processes more
costly, the inherent energy advantage of reverse osmosis will give the process an
increasing share of the growing market. Toward that goal, membrane manufacturers
must:
. Develop an improved understanding of seawater pretreatment chemistry.

. Develop a better understanding of the limitations of their products.
. Develop high performance membranes from polymer materials that do not
degrade in oxidative seawater environments.
. Develop membranes that are not dependent on periodic surface treatments
with chemicals to restore and maintain membrane performance.
. Develop membranes from polymers exhibiting low levels of bacteria attachment
to minimize fouling.
. Employ efficient energy recovery systems to reduce the operating costs of
seawater desalination.
Single-stage seawater desalination systems are generally limited to a maximum of

45% water recovery when operating on seawater feeds with total dissolved solids in the
35,000 to 38.000 mg/L range. The limiting factor in plant operation and water recovery
is the high osmotic pressure observed in the final elements in a membrane pressure
vessel, severely reducing the net pressure, or driving force, to produce sufficient
permeate flow and permeate concentration.
Figure S-13 shows the effect on the osmotic pressure of the feed and the net
driving force as the seawater traverses through each element in the pressure vessel. It
is the net driving pressure that dictates both flow rate and membrane rejection in all
membrane systems. Considerable savings in both capital and operating costs would be
realized by a thinner membrane barrier, because plant operation could be carried out at
a lower pressure.
5.5.3 Energy Recovery for Large Seawater Desalination Systems
To lower the energy requirement and the high cost per gallon of fresh water from
large seawater desalination plants, energy recovery systems to recover the energy
contained in the pressurized concentrated brine streams are essential. An energy
recovery system allows a smaller, less costly motor to be used and results in a
significant savings in the cost per gallon of product water produced.” This effect is
illustrated in Figure 5-14. Energy savings of 35% are typical when energy recovery
systems are used. Although energy recovery systems are best suited for high-pressure
desalination systems, energy savings are also available for low-pressure reverse osmosis
membrane systems.
There are three types of energy recovery devices used on reverse osmosis plants
today. All are well developed and operate at high recovery efficiencies.
. Pelton wheel, supplied by Calder of England and Haywood Tyler in the U.S.:
recovers 85-90% of hydraulic energy of the brine.
. Multistage reverse running centrifugal pump, supplied by Pompes-Ginard of
France and Johnson Pumps, Goulds Pumps, and Worthington Pumps in the U.S:
80-90% hydraulic energy recovery.
. Hydraulic work exchanger across piston or bladder: 90-95% hydraulic energy
recovery when used on small systems.
Concentrate Membrane PreSSUre

seal elements vessel
Concentrate
Feed - B
Product
800.
600
PRESSURE
(psi)
400
POSITION
Figure 5-13. Effect of water recovery on the seawater feed osmotic pressure and net
driving pressure.
Reverse Osmosis 307
10 I I I
Water recovery (%I:
8-
ENERGY
COST
(kWh/mS)
2, I I I
800 900 1,000 1,100 1,200
OPERATING PRESSURE (psi)
Figure S- 14. Energy cost vs. operating pressure for seawater desalination systems
employing energy recovery devices.
55.4 Low-Pressure Reverse Osmosis Desalination
In the past, brackish water reverse osmosis desalination plants operated at applied
pressures of 400 to 600 psi. Today, with the development of higher flux membranes,
both capital and operating costs of the reverse osmosis process can be reduced by
operating brackish water plants at pressures between 200 and 250 psi. Low-pressure
operation also means these high-flux membrane systems operate within the constraints
of the spiral element design (12-17 gallons/ftz-day), thus minimizing membrane fouling
that would occur if higher pressure were used. The significant energy savings achieved
are illustrated in Table S- 11, which shows the comparative energy requirements of a
membrane system operated at 400 and 200 psi.
Table 5-11. Energy Requirements of Reverse Osmosis Brackish Water Membranes
Energy Requirements
Process Type kWh/m3 kWh/ 1,000 gal
Brackish Water - 400 psi 2 7.6
Brackish Water - 200 psi 0.8-1.5 3.0-5.7
In addition to the increased productivity exhibited by the new low-pressure reverse

osmosis membranes, they show enhanced selectivity to both inorganic and organic
solutes and, in some cases, increased tolerance to oxidizing agents. These properties
make the membrane particularly attractive for a wide variety of applications, including
brackish water desalination. These applications include:
. High purity water processing for the electronic, power, and pharmaceutical
industries.
Industrial feed and process water treatment.
Industrial wastewater treatment.
Municipal wastewater treatment.
Potable water production.
Hazardous waste processing.
Point of use/point of entry.
Desalting irrigation water.
A comparison of the desalination performance of several commercially available

low-pressure membranes is shown in Figure 5-15. I6 These low-pressure reverse osmosis
membranes have significantly lower operating costs, improved wafer quality, and reduced
capital costs. It is anticipated, therefore, that these membranes will soon become the
state-of-the-art design for new and replacement plants and will have a higher than
normal growth rate. This growth should also allow membrane manufacturers to increase
gross margins by producing an improved product which will help stabilize the industry in
this period of price instability.
Reverse Osmosis 309
99.9 - I I I 1
su-700
..:.
99.8 - Toray
99.5 - BW-30
FilmTec
0_
NaCl 99.3 --- CT3
rejection gg -
(%) A-l 5
DuPont
98 -
NTR-739HF
95 - Nitto-Denko 4 I
90 I I I I
0.6 0.8 1.0 1.2 1.4 1.6
Flux(m3 /m2 * day)
Figure 5- 15. Desalination performance of several commercially available low-pressure

membranes. Feed: 1,500 mg/L NaCl at IS kg/cm2, 25°C.
5.5.5. Ultra-Low-Pressure Reverse Osmosis Desalination
The ultra-low-pressure reverse osmosis process is commonly referred to as

nanofiltration or “loose reverse osmosis”. A nanofiltration membrane permeates
monovalent ions and rejects divalent and multivalent ions, as well as organic compounds
having molecular weights greater than ZOO. As the name implies, nanofiltration
membranes reject molecules sized on the order of one nanometer. Nanofiltration is a
low-energy alternative to reverse osmosis when only partial desalination is required.
Nanofiltration membranes are thin-film composite membranes, either charged or

non-charged. They are currently being commercialized by several membrane
manufacturers. A comparison of desalination performance of these commercial
membranes is shown in Figure 5-16.23
These membranes typically exhibit rejections of 20 to 80% for salts of monovalent

ions. For salts with divalent ions and organics having molecular weights about 200,
the rejections are 90 to 99%. Thus, these membranes are well suited for:
. Removal of color and total organic carbon (TOC).

. Removal of trihalomethane precursors (humic and fulvic acids) from potable
water supplies prior to chlorination.
Removal of hardness and overall reduction of total dissolved solids.
Partial desalination of water (water softening) and food applications.
Concentration of valuable chemicals in the food and pharmaceutical industries.
Concentration of enzyme preparations.
Nitrate, selenium, and radium removal from groundwater in potable water
production.
. Industrial process and waste separations.
For most water with dissolved solids below 2.000 mg/L, nanofiltration membranes
can produce potable water at pressures of 70 to 100 psi. However, a low-pressure
reverse osmosis plant operating at 200 psi can produce a higher quality permeate.
Taylor et al.,17 in a survey of ten operating nanofiltration and reverse osmosis

membrane plants concluded that the capital cost for a nanofiltration plant is equivalent
to a reverse osmosis plant. Although pumps, valves, piping, etc. are less expensive for
a nanofiltration plant, the same membrane area will be required because the fouling
potential is the same for both membrane types. The lower cost for construction is
offset by the higher cost of the nanofiltration membranes. The membrane cost
represents only about II% of the total cost of the permeate produced. Thus, the
initial membrane price is not as important as the membrane and plant reliability.
Estimated cost data for a 38,000 ms/day (10 MGD) low-pressure reverse osmosis
plant operating at 240 psi and 75% recovery give a projected water cost of S0.38/m3
permeate. An equivalent nanofiltration plant operating at 100 psi is about 7% cheaper,
due to the lower pumping costs. The relative cost advantage for nanofiltration
membranes compared with reverse osmosis membranes is, therefore, slight, but will
increase as energy costs rise. These ultra-low-pressure membranes are currently finding
acceptance in Florida for softening groundwater. The future for these reverse osmosis
membranes appears to be bright.
Reverse Osmosis 311
95 - I I I I
MT&729HF
Nitto-Denko
+?. UTC-4OHF _
90 - Q '..
'f To ray
::.e;.
N F-70
Q * Fi,mTec
NaCl 60 -
Rejection Nitto-Denko
(W UTC-2OHF -
70 - NF-40
Toray
FilmTec
Q
40 -
NF-40HF NF-50
FilmTec FilmTec
I I I I
0
0 0.5 1.0 1.5 2.0 2.5
Flux (r$ /m2 l day)
Figure 5- 16. A comparison of the desalination performance of several commercial

ultra-low-pressure membranes operating on a 500 mg/L NaCl feed at 7.5
kg/cmt, 2S’C.
5.6.1 Projected Reverse Osmosis Market: 1989-1994
A significant number of large desalination plants, both multistage flash (MSF) and
reverse osmosis, are slated to be built between now and the early 1990s. primarily in
the Middle East, Spain, and Florida in the United States. New plant construction for
1988 is estimated at 440 mgd; estimates for 1989 and 1990 are for 220 mgd. Projections
through 1994 include 267 mgd of new construction for that year. If projections are
realized. this would increase the present world capacity of 3 billion gpd by more than
3396.5
About 50% of this new capacity would be MSF. with 41% reverse osmosis. This
includes 100 mgd of brackish water reverse osmosis in South Florida over the
intermediate time range 1993- 1994. These plants will utilize low-pressure and ultra-
low-pressure energy-efficient membranes for brackish groundwater desalting, water
softening, and trihalomethane precursor removal.
Presently, Spain with 12.3% of desalination plant sales is the fastest growing
worldwide market. This market is focused on seawater desalination in the Canary
Islands using reverse osmosis.
The use of desalination as a water treatment process will continue to increase

worldwide. In the United States, reverse osmosis will be the most rapidly accepted
process. This process will be used primarily for:
. Desalting brackish groundwater for potable purposes.

. Treating municipal waste water.
. Industrial process water.
Some bias against desalination technologies still exists, particularly in the area of
municipal water treatment. Conventional processes are still preferred by some
consulting engineering firms who design the plants, by water utilities that build and
operate the plants, and by public health agencies. This bias is expected to decrease
significantly with the introduction of new and improved low- and ultra-low-pressure
membranes. These membranes, because of their inherent lower energy requirements,
will be dominant factors in expanding the reverse osmosis market.
5.6.2 Research and Development: Past and Present
Between 1952 and 1982, Federal funding for desalination research, development, and
demonstration averaged about 511.5 million per year (as appropriated) - about $30
million per year in 1985 dollars. This program, a portion of which was directed to
membrane processes, was primarily responsible for the development of reverse osmosis.
As shown in Figure 5-17, the program peaked in 1967 and was virtually nonexistent in
1974.6 By that time, the United States had established a technological leadership role
for reverse osmosis desalination throughout the world. The technology developed under
this program was made freely available throughout the world through workshops and the
wide distribution of published papers. The western drought of 1976-77 renewed interest
in membrane improvement for reverse osmosis and additional funding was provided at a
rate of about SIO million per year until 1981 when the program was officially
terminated.
Reverse Osmosis 313
AMOUNT
($, millions)
1950 1960 1970 1980 1990
YEAR
Figure S- 17. Annual Federal funding for desalination research and development.“7
Between one-half and two-thirds of the money was spent on membrane
research.
Since the termination of the Federally-funded desalting program, most U.S.

companies have committed little, if any, funding for reverse osmosis membrane research
and development. Further, with worldwide membrane product sales of $118 million per
year and due to the low or negative profit margins associated with the competitiveness
of the industry, it is unlikely that this situation will change. Much of the present
research and development effort is applied research, rather than basic research, directed
toward the development of specific products or improving plant efficiencies. Most of
this work is done within private companies. There are no industry coordinated research
efforts being conducted at this time. In addition, there is little, if any, research being
conducted in U.S. universities. As a result, many of the dominant patents upon which
the U.S. membrane industry was built have expired, leaving most membrane
manufacturers with a weak or non-existent patent position. In the private sector, the
level and focus of research and development is controlled largely by the marketplace.
In this case, development costs are indirectly passed on to the end user.
The reverse osmosis industry is presently unable to justify sponsoring significant

amounts of research and development to maintain world leadership. For the most part,
the industry believes that a Federal program should assist on a cost sharing basis where
individual companies could retain a proprietary position. Such a program would improve
the competitive position of the U.S. membrane industry in foreign markets, as well as
benefit municipal and industrial users of this technology in the U.S.
5.6.3 Research and Development: Energy Reduction
Reverse osmosis is a low-energy desalination process because, unlike distillation, no

phase change occurs. Significant advances have been made over the past decade in
membrane polymers, structures, and configurations that have reduced energy
consumption. However, reverse osmosis is still far from the theoretical minimum energy
requirements of the process. In the following section, areas of research directed toward
reducing the energy requirements of the reverse osmosis process are described.
5.6.4 Thin-Film Comoosite Membrane Research
5.6.4. I Increasing water production efficiency
Thin-film composite membranes for desalination were created by the reverse

osmosis industry primarily because asymmetric membranes were unable to provide the
transport properties required for seawater desalination. The work that led to the
development of these membranes was empirical and limited to polyamide-type membrane
systems. The water production efficiency of these membranes, as described below, is
only about 30% of the theoretical value. Thus, energy consumption is high even with
the new low-pressure membrane systems.
The development of the thin-film composite membrane was undertaken soon after it
was realized that the Loeb-Sourirajan method was not a general membrane preparation
method that could be applied to a wide variety of polymer materials. Thin-film
composite membranes, for the most part, are made by an interfacial process that is
carried out continuously on the surface of a porous supporting membrane.” The thin
semipermeable film consists of a polyamide-type polymer, while the porous supporting
membrane is polysulfone. This method of membrane processing provides the opportunity
to optimize each specific component of the membrane structure to attain maximum
Reverse Osmosis 315
theoretical water productivity and/or minimum theoretical energy consumption. The

discussion that follows describes how the energy consumption required by state-of-the-
art thin-film composite membranes can be reduced significantly.
. Theoretically, the thin, dense film can be made on the order of 200 A thick,
five to ten times thinner than either asymmetric or current commercial thin-
film composite membranes. Attainment of such film thicknesses has been
demonstrated. Reduction of the thickness of these films can significantly
reduce the energy consumption of the process, since the water throughput
through the thin polymer film is inversely proportional to thickness for a
given set of operating conditions.
. Both the thin barrier film and the porous supporting membrane can be made
from different polymer materials, each optimized for its own specific
function.
0 The surface pore size, pore size distribution, and surface porosity of the
porous supporting membrane must be such that flow through the thin-film is
not hindered. Commercial thin-film composite membranes operate at only
about 30% efficiency because of flow restrictions.”
. The charge on the surface of the thin-film barrier can be varied and
controlled. To minimize energy consumption, interactions between the
membrane surface and feed components that will increase resistance to flow
must be avoided. Ideally, membrane surfaces should be neutral.
Research opportunities abound in the area of membrane optimization -- which leads

to reduced energy consumption. Other research opportunities in this area are
suggested below:
. Present thin-film composite membranes are limited to operating temperatures

of 40-WC. For example, composite membranes are required for the food
industry that are capable of operating up to IOO’C for energy savings and to
control the growth of microorganisms by sterilization.20
. Attachment of biocides, surfactants, anti-foulants. anti-scalants, and enzymes
at the thin-film interface with the feed stream is of interest to minimize
fouling, thereby reducing the pressure drop across the membrane element and
energy consumption. Experimental investigation of surface attachment is
needed.21
. Develop new types of interfacially formed thin-film systems that are capable
of withstanding a constant level of chlorine (an effective biocide) at 0.5 to
2.0 mg/L for periods of three to five years.
5.6.4.2 Seawater reverse osmosis membranes
There is a need to improve the stability and reliability of the present state-of-the-
art polyamide-type reverse osmosis seawater membranes. This is particularly needed for
plants operating on open seawater intakes where disinfection by chlorination is required
to control growth of microorganisms. The sensitivity of these membranes to chlorine
and oxidizing agents is such that chlorination/dechlorination is required when chlorine
is used as a disinfectant in pretreatment. Today, there is a lack of understanding by
both membrane manufacturers and OEMs of how to successfully dechlorinate a seawater
feed to polyamide-type thin-film composite membranes at a reasonable cost. A number
of large seawater desalination plants have been lost due to insufficient dechlorination.
To improve the reliability of the reverse osmosis seawater desalination process, the
following areas of research should be considered:
Development of seawater membranes from oxidation-resistant polymeric

materials.
Development of effective methods for disinfecting seawater feeds at
competitive costs that do not damage polyamide type membranes.
Development of a comprehensive understanding of the process of
dechlorinating seawater with sodium bisulfite and/or sulfur dioxide.
Determine the influences of dissolved, heavy metals, etc. on the process.
Analysis of the effects of dissolved oxygen in seawater, if any, on the
stability of polyamide type membranes.
Comprehensive evaluation of the oxidation-reduction (redox potential)
characteristics of pretreated seawater and the influence, if any, it has on
membrane degradation.
As previously discussed, research should be directed toward increasing the water

production efficiency of polyamide type thin-film composite membranes, because
significant energy savings could be attained by operating reverse osmosis plants at lower
applied pressures.
5.6.4.3 Low-pressure membranes
The economic advantages of low pressure (200-240 psig) membranes are lower
energy consumption and operating costs. For these reasons, older reverse osmosis
plants are being retrofitted and converted to low-pressure operation.”
The membranes used in these processes, for the most part, are not new
developments, but material and morphological optimixations of current technology.
These long-overdue improvements have been demonstrated with both asymmetric cellulose
acetate and thin-film composite membranes. Low-pressure asymmetric cellulose acetate
membranes are now operating at the 5 mgd Orange County Water District’s reverse
osmosis plant on a municipal waste water feed at significantly reduced energy costs.
Even though great strides have been made, these membranes retain the material
limitations of the original system with respect to temperature, bacterial adhesion, lack
of oxidation resistance, etc. Research and development efforts are required to develop
new membrane materials that overcome these limitations. Thin-film composite membrane
systems offer the greatest promise to attain these objectives?
5.6.4.4 Ultra-low-pressure membranes
The ultra-low-pressure membrane process (<LOO psig) is not strictly defined.

Unlike normal reverse osmosis, the process operates at lower pressures and allows
selective permeation of ionic salts and small solutes. The transport characteristics of
these membranes fall into the area between reverse osmosis and ultrafiltration and they
are commonly referred to as nanofiltration membranes.
Reverse Osmosis 317
The development of nanofiltration membranes is relatively new, with only a limited

number of products in the marketplace today. The potential for this process looks
particularly promising since the energy consumption is significantly less than for low-
pressure membranes operating on the same application. This savings has been shown to
be as much as 15%.”
Most of the ultra-low-pressure membranes developed to date are thin-film

composite membrane structures, whose performance characteristics are determined
primarily by the chemistry, structure, and thickness of the thin-film barrier surface.
Directed research efforts in this area should result in the development of

membranes capable of operating at still lower energy consumption. The focus of this
future research, however, should be on improving the selective permeation between
various inorganic and organic solutes at ultra-low applied pressures. In addition, it is
desirable for the membranes to be resistant to oxidation, stable at high temperatures,
and solvent-resistant.
5.6.5 Membrane Fouling: Bacterial Adhesion to Membrane Surfaces
Chlorine is perhaps the most commonly employed disinfectant in the pretreatment

of water to control microorganisms. Unfortunately, many types of bacteria exhibit
resistance to this biocide, and those microorganisms comprising an attached biofilm in a
reverse osmosis membrane element may be entirely resistant to the biocidal effects of
such halogen disinfectants. Furthermore, chlorine cannot be used with polyamide type
membranes, since this oxidant rapidly deteriorates these membranes with a dramatic loss
in salt rejection. Given the limitations of chlorine, and the current lack of alternative
biocides. it is no surprise that bacteria and other microbes have been demonstrated to
rapidly adhere to and colonize both the feedwater and permeate channel surfaces in
membrane elements. The major symptoms associated with microbial fouling of membrane
surfaces are:
. A gradual decline in water flow per unit membrane area.

. An associated increase in transmembrane operating pressure of the system,
which may eventually exceed the manufacturer’s specifications.
. A gradual increase in salt transport through the membrane.
. A gradual deterioration of permeate quality with increasing bacterial count in
the permeate.
The cumulative effects of membrane biofouling are greatly increased cleaning

frequency and significantly reduced membrane life. These factors result in increased
operating and maintenance costs which adversely affect the overall efficiency and
economics of the reverse osmosis process. Increased energy consumption is associated
with both cleaning operations and reduced water productivity.
Fundamental research is needed in most aspects of membrane biofouling.

Invesitgation of the specific changes in biofilm chemistry and ultrastructure that affect
membrane performance is required. Knowledge is also required concerning the specific
biochemical and/or environmental signals that must regulate microbial population
fluctuations on the membrane surface. It is also important to identify and characterize
the repertoire of fouling microorganisms and subcellular components that are most
significant in terms of loss in membrane performance.
318 hllembrane Separation Systems
The bacterial adhesion process is fundamental to the often serious biofouling

problems encountered in reverse osmosis systems. Currently, there is a very limited
understanding of the adhesion process, although research methods have recently been
developed which have the potential to greatly expand our knowledge in this critical
area.” Thus, it is imperative that a detailed understanding of the subcellular and
molecular mechanisms of adhesion exhibited by fouling bacteria be acquired before truly
effective measures for controlling biofouling in reverse osmosis membrane systems can
be designed and implemented.
The current inability to effectively and reliably control microbial adhesion and
biofouling in reverse osmosis systems suggests that this is an area needing considerable
further investigation. Some worthy research goals in the area of membrane biofouling
might include:
. Additional research on the types of fouling bacteria, their specific adsorption

behavior and kinetic attributes, and the molecular basis for their attachment
to different membrane polymers operated under differing feedwater conditions.
. Exploration and development of innovative membrane polymers having
significantly reduced affinity for microbes implicated in the biofouling
process.
. Development of alternative biocidai agents that exhibit greater activity
against biofilm bacteria.
. Development of novel and cost-effective pretreatment methods which are more
capable of removing biofouling type microorganisms from feedwater streams.
. Development of reverse osmosis cleaning formulations which display increased
ability to disrupt and solubilize microbial biofilms without adversely affecting
membrane performance.
. Establishment of appropriate mathematic algorithms which can be utilized to
predictively model biofouling in reverse osmosis membrane systems. Such a
model would be extremely useful in the earliest stages of new plant design
and engineering, when membrane selection, long-term performance, and
operating and maintenance costs must be predicted with a reasonable level of
confidence and accuracy.
5.6.6. Spiral-wound Element Optimization
The design of the spiral-wound element has not changed appreciably since its
early development in the late 1960s. With the development of low- and ultra-low-
pressure membranes, the design and material selection becomes more important. For
example, the net pressure may be on the order of 70 psig for ultra-low-pressure and 180
psig for low-pressure membranes. Thus, flow resistance within the element, or
differential pressure, becomes a significant percentage of the net pressure.
The differential pressure across the element is approximately linear with feed flow
rate. A typical differential pressure drop for a new 40-inch low-pressure (214 psig)
element, as designated by the manufacturer’s specification, is <I4 psi. The differential
pressure for a vessel of six elements is ~29 psi. A large amount of this flow resistance
can be attributed to the turbulence-promoting mesh-screen type feed spacer separating
the membrane sheets within the spiral element. Thus, for a low-pressure membrane
element that operates at a net driving pressure 190 psig. a pressure drop of 29 psig
reduces the net driving pressure by 15%. Such elements are generally cleaned when the
pressure drop increases by a factor of 1.5, or 43.5 psig for a vessel of six elements.
Reverse Osmosis 319
This condition reduces the net driving pressure by 15%. If the plant operates at a
constant flow, this requires a significant increase in operating pressure and energy
consumption.
It becomes apparent that the pressure drop within a spiral element becomes a
restricting factor in setting the lower limits at which low-pressure membranes can
operate. Development work is needed to improve the efficiency of the spiral element
for low net pressure operations by:
Selecting new spacer materials that have a low affinity for bacteria adhesion
to minimize fouling.
Selecting spacer designs that are amenable to and assist cleaning. Present
type spacer materials trap suspended solids within the cells of the mesh
screen, particularly after cleaning.
Designing spacer configurations that are more effective turbulence promoters

and offer less resistance to flow.
Designing spacer materials that minimize stagnant areas within the element.
The efficiency of present elements is significantly reduced by areas of low
feed flow and stagnation near the product water tube.
Optimizing the design of anti-telescoping devices with respect to flow

resistance.
Such improvements in element materials and design would enhance the energy
efficiency of membrane processes that operate at low net driving pressures. Processes
of this type are a rapidly growing segment of the market and exhibit the greatest
potential for energy reduction.
5.6.7. Future Directions and Research Topics of Interest for Reverse Osmosis Systems
and Applications
Despite the success of membranes developed for reverse osmosis applications

during the past 25 years, considerable research efforts are warranted to achieve the full
potential of this process with significant reductions in overall cost and energy
expenditures. Tables 5-12 and 5-13 outline future directions and topics of interest for
research on reverse osmosis systems and applications, respectively. The importance of
each topic is rated on a scale of I to 10, with the higher number showing the larger
degree of importance.
Table 5.12. Future Directions for Reverse Osmosis Applications
Prospect for
Topic Realization Importance Comments/Problems
Seawater Excellent 5 The reliability of single-stage seawater

reverse osmosis process requires further
improvement to make the process more
competitive with conventional processes.
Improvements in membrane flux,
selectivity and oxidative stability are
required.
&g.ckish Ww
Surface Excellent 7 Greater selectivity at lower pressures

desirable for reducing energy and capi-
tal costs. Membrane fouling a major
problem on these complex feeds.
Well Excellent 10 Greater selectivity at lower pressures

desirable for reducing energy and capital
costs.
Softening Excellent 7 A rapidly growing market, particularly in

the Florida area for softening under-
ground water. Also promising for
seawater softening of feeds to reverse
osmosis and distillation plants. Requires
ultra-low-pressure membranes that can
discriminate between monovalent and
divalent ions.
Water Reclamation
Municipal waste Excellent 7 Bacteria fouling of the membrane surface

is the major obstacle limiting this appli-
cation. Membranes are required that ex-
hibit low levels of bacteria attachment,
oxidation resistance and are capable of
operating at low pressure to reduce
energy and capital costs.
Agricultural Fair 2 Very complex and difficult high-fouling

Drainage feedstreams that contribute to high sys-
tem costs. Requires low-pressure mem-
branes.
Industrial waste Good 5 Very complex and difficult high-fouling

feedstreams that contribute to severe
membrane fouling.
Reverse Osmosis 321
Table 5.12. continued
Prospect for
Process Water
Boiler water Excellent 8 Requires highly selective membranes

capable of producing a very high quality
product water at low pressures. The
rejection of silica must be high.
Ultrapure water Excellent 7 This application requires the highest

quality of membrane element and
manufacturing. The membrane product
must be free of any leachable materials
into the ultrapure water product stream.
Dewatering Fair 3 Dewatering of a feedstream results in a

concentrate of high solids confer&
Improvement in membrane packaging is
necessary to minimize fouling in this
application.
High temperature Fair 2 Energy savings can often be attained by

processing high-temperature feedstreams.
For this application, a high-temperature
stable membrane package is required.
Table 5.13. Research Topics of Future Interest for Reverse-Osmosis Systems

and Applications
(SW = Seawater, BW = Brackish Water, WR = Water Reclamation, PW = Process Water)
Prospect for
Topic Realization Importance Comments/Requirements
SW BW WP PW
Thin-Film Comoosite Thin-film composite membranes

Membrane Research in use today are first gener-
ation technology. As a result,
the potential for improvement
is excellent.
Increasing water Excellent 5 8 a Commercial thin-film composite

flux membranes operate at about
30% of theoretical efficiency
because of flow restrictions
within the membrane. Modest
improvement could reduce the
energy consumption of the
reverse osmosis process
significantly.
Seawater membrane Excellent 8 Single-stage seawater desali-

nation is the most cost effec-
tive mode of operation. Mem-
brane improvements in water
flux, selectivity and oxidation
resistance are necessary to
improve the reliability and
competitiveness of the process.
Low-pressure Excellent - 8 8 Thin-film composite membrane

membrane systems offer the greatest po-
tential to achieve this objec-
tive. Material and morpho-
logical optimization of current
membrane system: are required.
Ultra-low-pressure Excellent - 6 8 The need to achieve better

membrane resolution in the separation of
ions/molecules of different
types but similar sizes at ultra-
low pressure is increasing.
Reverse Osmosis 323
Prospect for
SW BW WP PW
Oxidation-resistant Good 10 10 Commercial polyamide reverse

membrane osmosis membranes rapidly de-
teriorate in the presence of
oxidizing agents such as
chlorine, hydrogen peroxide,
etc. This deficiency has
slowed the acceptance of the
process in some areas.
High-temperature Excellent I 3 Current membranes are limited

to applications that do not
exceed 35-4O’C. Many appli-
cations, particularly in the
process water areas require
high-temperature stable mem-
branes. Modification of exist-
ing thin-film composite mem-
branes can be made to achieve
this objective.
Bacterial Attachment Excellent 6 10 Bacteria fouling of membrane

U, Membrane Surfaces surfaces reduces productivity.
Affinity of microorganisms for
different membranes is mark-
edly different. Elucidation of
attachment mechanism required
to select optimal membrane
material and surface morph-
ology.
Soiral-wound Excellent 4 9 New feed spacer designs are

Element necessary to minimize fouling
Imorovement and enhance the efficiency of
membrane cleaning. Current
feed spacers trap suspended
solids within the interstices of
the feed spacer, making them
difficult to remove during
cleaning.
Prospect for
SW BW WP PW
Pretreatment/Fouling Excellent 4 4 9 6 The need for greater volumes

Cleaning processed before fouling/scaling
and subsequent cleaning of the
membrane system equates dir-
ectly to a lower cost.
Improved pretreatment Good 9 7 9 8 Improvement of classical pre-

treatment methods that will
enhance the reduction of sus-
pended solids in feedstreams to
reverse osmosis systems is de-
sired.
Chlorination/ IO 4 3 3 Process improvements are re-

Dechlorination quired to protect chlorine-
sensitive TFC polyamide mem-
branes.
Colloidal Fouling Good 6 8 6 4 The AL/FE/Si-humic acid

colloidal complex, present in
surface waters, is particularly
troublesome. Improved pre-
treatment and cleaning
processes are needed.
Other Foulants Good 7 6 6 a Adsorption of organic materials

on membrane surfaces is a
major problem with complex
feed waters containing organic
materials.
Scaling (CaCOs, Excellent I 2 2 4 Commonly used polyacrylic acid

BaSO,, etc.) anti-sealant materials are ad-
equate.
Silica Anti-Sealant Good I 6 6 7 An anti-sealant is needed to

increase the water recovery of
reverse osmosis plants opera-
ting on high silica feeds.
Reverse Osmosis 325
(SW a Seawater, BW = Brackish Water, WR = Water Reclamation, PW = Process Water)
Prospect for
SW BW WP PW
Disinfectants Good 7 4 10 4 Non-THM producing disinfect-

ants are needed to control
membrane fouling by micro-
organisms.
Cleaning Excellent 6 6 9 5 Membrane cleaning is not

Improvements always successful; it remains a
trial-and-error operation.
Enernv Recovery Poor 2 2 2 2 State of the art energy recov-

Devices ery devices are relatively ef-
ficient.
5.6.8. Summary of Potential Government-Sponsored Energy Saving Programs
The Federal Government will be one of the largest beneficiaries of energy-saving

advancemints that may result from the aforementioned research and development. With
both the military program and the Bureau of Reclamation’s 72 mgd Yuma reverse osmosis
desalination plant, the Federal Government has been the largest purchaser of reverse
osmosis elements.25 Unfortunately, membrane elements scheduled to go on line at the
Yuma Desalination Plant in 1991 are dated technology. The asymmetric cellulose
diacetate membrane elements contracted for the Yuma plant will operate at 400 psig and
above on high-salinity irrigation drainage return water to the Colorado River before
entering Mexico. Difficulties were encountered with these membranes in the pilot test
program with respect to fouling and membrane stability that have not been satisfactorily
resolved. Thus, additional membrane elements were added to the plant to compensate
for unacceptable high fouling rates.
The estimated water cost from the plant is about $0.80 per 1000 gallons. The
operating costs have increased about 50% over the original estimates, based primarily on
order of magnitude changes in energy prices. With that as emphasis, the Federal
Government must more actively pursue cost-saving potentials. Areas to consider are
low-pressure, chlorine-resistant membranes that are less susceptible to fouling, off-peak
electrical operation, particularly seasonal, and cogeneration.
Federal support of research and development for the development of energy-saving

reverse osmosis membranes and demonstration projects is in the Federal Government’s
interest. If research and development is left to the private sector, the level of effort
will be controlled largely by the market demand. This type of research and
development is usually focused on the short-term and has not proven successful for the
U.S. desalination industry. The primary issue, then, is not how much research should be
conducted, but who should fund it - the Federal Government or the users of
desalination. Since the Federal Government is one of the largest users of reverse
osmosis technology in the U.S., their participation seems apparent.
The market for membrane processes will be driven not only by energy reduction,
but also by stringent standards for drinking water, hazardous waste disposal, industrial
wastewater discharge, municipal wastewater, etc. For the most part, these processes
would utilize the same energy-efficient membranes requiring similar performance
characteristics.
Government and private industry could cooperate in the joint development of new
low-pressure, energy-efficient membrane systems for all these applications. The
development test facilities at the Yuma plant could be used both by the Federal
Government and private industry. To utilize the available resources, the Federal
program could be a cooperative one between Departments involved in supporting
desalination research. A return to high energy prices would tend to elevate the
priorities associated with such a program.
Reverse Osmosis 327
REFERENCES
1. H.K. Lonsdale, U. Merten and R.L. Riley, Transport Properties of Cellulose

Acetate Osmotic Membranes,” ,I. AopL Polv. Sci. 9, 1344 (1965).
2. S. Sourirajan, “Reverse Osmosis”, Academic Press, New York (1970.
3. S. Loeb and S. Sourirajan, Adv. Chem. Ser. 38, 117 (1962).
4. S. Loeb and S. Sourirajan. “High Flow Porous Membranes for Separating Water
from Saline Solutions”, U.S. Patent 3.133.132, May 12, (1964).
5. K. Wangnick. 1988 IDA Worldwide Desalting Plants Inventory Report No. 10,
June/July, (1988).
6. U.S. Congress, Office of Technology Assessment, Using Desalination Technologies

for Water Treatment, OTA-BP-0-46, U.S. Government Printing Office, Washington,
D.C.. March (1988).
7. J.W. Kaakinen and C.D. Moody, “Characteristics of Reverse Osmosis Membrane

Fouling at the Yuma Desalting Test Facility”, in: S. Sourirajan and T. Matsuura
(Ed.), Reverse Osmosis and Ultrafiltration, ACS Symposium Series 281, 359-382,
Washington, DC. (1985).
8. H.F. Ridgway, C. Justice, A. Kelly, and B.H. Olson, “Microbial Fouling of Reverse
Osmosis Membranes Used in Advanced Wastewater Treatment Technology:
Chemical, Bacteriological and Ultrastructural Analyses”, Applied and Environmental
Microbiology, 45, 1066-1084 (1983).
9. H’.F. Ridgway, M.G. Rigby, and D.G. Argo, “Adhesion of a Mycobacterium to

Cellulose Diacetate Membranes Used in Reverse Osmosis”, Applied and
Environmental Microbiology, 47, 61-67 (1984).
10. H.F. Ridgway, C.A. Justice, C. Whittaker, D.G. Argo, and B.H. Olson, “Biofilm
Fouling of Reverse Osmosis Membranes: Its Nature and Effect of Water for
Reuse”, J. Amer. Water Works Assn., 76, 94-102 (1984).
I I. H.F. Ridgway, D.M. Rodgers, and D.G. Argo, “Effect of Surfactants on the Adhesion
of Mycobacteria to Reverse Osmosis Membranes”, Proc. of the Semiconductor Pure
Water Conference, San Francisco, California (1986).
12. H.F. Ridgway, “Microbial Adhesion and Biofouling of Reverse Osmosis”. in B. Parekh
(Ed.), Reverse Osmosis Technology: Applications for High-Purity Water Production,
Marcel-Dekker, Inc., New York (1988).
13. C.T. Sackinger, “Seawater Reverse Osmosis System Design”, Permasep Products,
DuPont Company Technical Manual.
14. K. Frank, “Seawater Reverse Osmosis Plant Performance at Lanzarote, Canary

Islands”, Proceedings of the National Water Supply Improvement Association
Conference, San Diego, California, ‘July 31 - August 4 (1988).
15. M. Kurihara, Toray Industries, Tokyo, Japan, October (1988).
16. Calder RO Turbines Technical Product Literature, Calder, Limited, England.
17. J.S. Taylor, et al, “Applying Membrane Processes to Ground Water Sources for
Trihalomethane Control, Research and Technology”, J. Amer. Water Works Assn.,
Vol. 79, No. 8, 72-82, August (1988).
18. R.L. Riley, “Thin-Film Composite Reverse Osmosis Membranes: Development Needs
and Opportunities”, Proceedings Membrane Technology/Planning conference, Boston,
Massachusetts. November 5-7 (1986).
19. H.K. Lonsdale, et al., “Transport in Composite Reverse Osmosis Membranes”,

Chapter 6 in Membrane Processes in Industry and Biomedicine, M. Bier (Ed.),
Plenum Press, New York (1971).
20. R.L. Riley, “Reverse Osmosis Apparatus”, U.S. Patent 4.411.787, March 25 (1983).
21. J. Stefarik, J. Williams, and H.F. Ridgway, “Analysis of Biofilm from Reverse
Osmosis Membranes by Computer Programmed Polyacrylamide Gel Electrophoresis”,
presented at the 18th Meeting of the American Society for Microbiology, New
Orleans, Louisiana, May 4-18 (1989).
22. F. Crowdus, “System Economic Advantages of a Low Pressure Spiral RO System

Using Thin Composite Membranes”, Ultrapure Water, July/August (1984).
23. R.G. Sudak, et al., “Procurement of New Reverse Osmosis Membranes: The Water
Factory 21 Experience”, Proceedings of the National Water Supply Improvement
Association Conference, San Diego, California, July 31-August 4 (1988).
24. J.E. Cadotte. et al., “Nanofiltration Membranes Broaden the Use of Membrane
Separation Technology”, Desalination, 70, Nos. l-3, November (1988).
25 K.M. Trompeter, “The Yuma Desalting Plant - A Water Quality Solution”,

Proceedings of the National Water Supply Improvement Association Conference, San
Diego, California, July 31-August 4 (1988).
6. MicrofiIt rat io n
by William Eykamp, University of California, Berkeley
6.1 OVERVIEW
Of the membrane processes included in this study, microfiltration is by far

the most widely used, with total sales greater than the combined sales of all the
other membrane processes covered.
For all its economic size, microfiltration is surprisingly invisible. It is

ubiquitous, with innumerable small applications. A huge fraction of the market
for microfiltration is for disposable devices, primarily for sterile filtration in the
pharmaceutical industry, and for filtration in semiconductor fabrication processes.
The heart of the microfiltration field is sterile filtration,’ using microfilters

with pores so small that microorganisms cannot pass through them. These
disposable filters, typically in the form of pleated cartridges, are sold to a
variety of users, but the major customer is the pharmaceutical industry.
Although there is intense competition for new sales in this market, stable
relationships between suppliers and customers are the rule. The cost of switching
to a new supplier can be high and, thus, there is little incentive for substitution
of one supplier’s product for another’s.
The replacement market for sterile filtration cartridges is quite large.

Micrsfiltration cartridges used as vent air filters may last for months, but those
used to filter batches of liquid may have a useful life measured in hours. These
membranes may sell for little more than f10/ft2, an order of magnitude belew
some other membranes covered in this report. But costs to manufacture in the
volumes required by the market leave a healthy margin for selling costs, research
and development, and profit. Cash generated by the business, and the competition
within it, provide a steady stream of innovation in the industry.
A second major application for microfilters is in the electronics industry for

the fabrication of semiconductors. As semiconductor devices shrink in size, the
conductive paths on their surfaces get closer together. Dirt particles represent
potential short circuits in the semiconductor device. Therefore, filtration of
various streams throughout the manufacturing process is a vital concern. For
microfiltration companies schooled in the sterile filtration discipline, the
electronics applications seemed made-to-order. A particularly attractive
application in this industry is final filtration of the water used to rinse
semiconductors during fabrication. Most of this water is first treated by a
reverse osmosis membrane. Since this is a much finer filter than a microfilter,
this water contains only a small amount of dirt from the piping and equipment.
Thus, microfilters used in this process have long lifetimes. Another area in which
microfiltration has been applied in the semiconductor industry is in filtering the
gases and liquids used as reactants in making a chip. These chemicals are often
very aggressive. and cannot be prefiltered by reverse osmosis membranes, so these
streams are a challenge and an opportunity for microfilter manufacturers. The
electronics industry has proven to be a strong market for microfiltration, and is
now second only to sterilizing filtration.
329
In both of the major microfiltration applications, sterile filtration and

semiconductor fabrication, energy considerations are less important than other
issues such as product quality. Some of the sterile microfiltration applications
replace thermal sterilization. In these cases, there is a direct energy saving in
the process and an indirect saving through avoidance of heat exchange equipment,
which has energy-intensive fabrication requirements. Energy is a negligible
consideration in the electronics applications. Other applications for
microfiltration. particularly some of the emerging potential applications, may offer
significant energy advantages over alternative methods. Therefore, the emphasis
of this report will be less on current dominant applications of microfiltration.
where energy is not a significant issue, and more on less developed process
applications.
6.2 DEFINITIONS AND THEORY
Microfiltration is a process for separating material of colloidal size and

larger from true solutions. It is usually practiced using membranes. In this
report, only microfiltration accomplished by membranes is covered. Microfilters
are typically rated by pore size. and by convention have pore diameters in the
range 0.1-10 pm. A photomicrograph of the surface of a typical microfiltration
membrane is shown in Figure 6-1. A microfiltration membrane is generally porous
enough to pass molecules which are in true solution even if they are very large.
Thus, microfilters can be used to sterilize solutions. because they may be prepared
with pores smaller than 0.3 pm, the diameter of the smallest bacterium,
Pseudomonas diminuta.
There are several key characteristics necessary for efficient microfiltration

membranes. These are ( 1 ) pore size uniformity, (2) pore density, and (3) the
thinness of the active layer or the layer in which the pores are at their minimum
diameter. The impact of these parameters on the flow through the membrane can
be seen by examining the governing equation for flow through the pores of a
membrane, Poiseuille's law:
Q/A = [ s] C nid: (1)
where Q/A is the volumetric flow rate per unit membrane area, Ap is the pressure
drop across the membrane, p is the solution viscosity, 6 is the thickness of the
active pore layer, and d. is the diameters of the individual pores in the unit
area A.
Microfiltration 331
Figure 6- 1. A surface photomicrograph of a typical microfiltration membrane.

Membrane shown is Nylon 66 with 0.2 pm pores.
The importance of pore size uniformity is evident, since a membrane will not
reliably retain anything smaller than the largest pore, which determines its rating.
Smaller pores contribute far less to flow. According to Poiseuille's law, a pore
0.9 times as large as the rated pore size contributes only two-thirds as much
flow. Pore length may be minimized by making the active layer 6 (in which the
pores are at their minimum diameter) as thin as possible. The importance of pore
density is especially important in dead-end filtration.
The most uniform pore sizes are found in membranes made by the track-etch
process, illustrated in Figure 6-2. Track-etched membranes are made by exposing
a polymer sheet to a beam of radiation, then selectively etching away the tracks
where the polymer was damaged by the radiation. Photomicrographs of these
membranes (Figure 6-3) show a uniformity of pore size difficult to find in
membranes formed by other techniques. The pictures also show that the number
of pores per area is low. Track-etched membranes cannot be made with high
pore densities because of the probability of track intersection, which would result
in pores too large for the rating. , Polymer strength dictates a minimum film
thickness for the membrane which, in the case of the cylindrically shaped pores
found in track-etched membranes, governs the pore length.
Membrane pore size is rated by, and tested with, latex particles, bacteria,
direct microscopic examination, and bubble point. The bubble point procedure
measures the diameter of the largest pore by forcing air through the wetted
membrane until a bubble appears. This procedure is illustrated in Figure 6-4.
The bubble point is a function of pore diameter and surface tension.
Photomicrographs of membranes made by a new technique, the anodic

oxidation of aluminum, show a membrane structure with promise for producing
membranes with high densities of thin, uniform pores. An example is shown in
Figure 6-5. These membranes should be very useful for many low-solids dead-end
filtration applications.
Microfiltration membranes are made in several different forms. One of the

most common is the pore filter. As shown in Figure 6-3. a photomicrograph of
this type of membrane looks like a plate with cylindrical holes drilled in it.
These filters are usually prepared by the track-etch method, described in more
detail below. There are, however, many techniques for preparing microfiltration
membranes, and they result in physical structures quite different from the pore
filter. Like ultrafiltration and reverse osmosis membranes, some microfiltration
membranes have conically shaped pores, with the small end of the truncated cone
facing the process fluid. In this structure, any particle passing through the small
end of the pore encounters a progressively more open path as it passes through
the filter. All the filtration is done at the surface, where the "funnel" is
narrowest. This feature can significantly reduce plugging and enhance mass
transfer.
Other common ways of preparing microfiltration membranes result in

structures that resemble porous beds of spheres, slits, and fibrous structures. The
final membrane form may be flat-sheet, ceramic monolith, tube, capillary, or fiber,
and these may be further modified in preparing various forms of modules.
/I '\ /I '\
-I @TWE
I ,
I
I
I
I I ; Pores
Non-conducting ;;
material
'Tracks"
I\
\\
\\
'.J ;
Figure 6-3. Photomicrograph of a Nuclepore@ membrane made by the track-

etch method.
Microfiltration 335
@ ZERO
PR€SSURE
@
INCREASING
ruLssuuc
@
BUBBLE WIN1
W.PSSURE
Figure 6 - 4 . Procedure used in determining bubble-point.

v,
I
W
Microfiltration 337
Microfiltration membranes can be operated in two ways: I ) as a straight-

through filter, known as dead-end filtration, or 2) in crossflow mode. In dead-
end filtration, all of the feed solution is forced through the membrane by an
applied pressure. This is illustrated in Figure 6-63. Retained particles are
collected on or in the membrane. Dead-end filtration requires only the energy
necessary to force the fluid through the filter. In the simplest applications. a
laboratory vacuum or simple pump provide enough motive force to drive the
application at an acceptable rate. The ideal energy requirement. if rate is not
critical. is negligibly low. The dead-end microfiltration membrane may be i n one
of many different forms (flat-sheet. pleated cartridge, capillary, tube. etc.)
The second wav to operate microfiltration membranes is in crossflow. I ?

this operational mode, shown in Figure 6-6b, the fluid to be filtered is pumped
across the membrane parallel to its surface. Crossflow microfiltration produces
two solutions; a clear filtrate and a retentate containing most of the retained
particles in the solution. By maintaining a high velocity across the membrane.
the retained material is swept off the membrane surface. Thus. crossflow is used
when significant quantities of material will be retained by the membrane. resulting
in plugging and fouling.
A principal difference in the operation of these two schemes is conversion

per pass. or the amount of qolution that passes through the membrane. In dead-
end filtration. essentially all of the fluid entering the filter emerges as permeate.
so the conversion is roughly 100%. all occurring in the first pass. For a
crossflow filter, far more of the feed passes by the membrane than passes
through it, and conversion per pass is often less than 20%. Recycle permits the
ultimate conversion to be much higher, however.
Another difference between dead-end and crossflow operation is the energy

required. The energy requirements of the crossflow method of operation are many
times higher than those of dead-end flow, because energy is required to pump the
fluid across the membrane surface. However, for high solids applications, and for
those where the solids would normally plug the filter when i t is operating as a
dead-end filter, crossflow is the method of choice.
6.3 DESIGN CONSIDERATIONS
The optimum design of a microfiltration membrane system depends on a

number of parameters and on the characteristics of the feed stream to be treated.
Two important design considerations are I ) the choice of operational mode, either
dead-end or crossflow, and 2) module design. Both of these are discussed below.
6.3.1 Dead-end vs. Crossflow Operation
One important characteristic of a feed stream is the level of solids that

must be retained by the microfilter. The higher the level of solids, the higher
the likelihood that crossflow filtration will be used.
Typically, streams containing high loadings of solids ( > O S % ) are processed by

membrane filters operating in crossflow. The operation of microfilters in
crossflow is similar to the operation of ultrafilters. The major difference is in
Dead-end filtration
Feed
ao oo
I aqaaooDO
Particle
/build-up on
membrane surface
-A
*
e *a6%-@$
b?!&
4-4 %%
I Particle-free permeate
a ) Dead-end filtration
Cross-flow filtration
Feed mo
I
b o Qa s oo e
$Q-s*3Gpsq&gq@&~
c3 Ba
I
Retentate
&Z%&sa@A!3B*9-.
8 Particle-free permeate
b) Crossflow filtration
FIgure 6-6 Schematic representations of a ) dead-end and b) crossflow

operation of microfiltration membranes
Microfiltration 339
the behavior of the polarized layer near the membrane. The limit to the rate at
which a crossflow device produces permeate is paradoxically the rate at which
solids retained by the membrane can redisperse into the bulk feed flowing past
the surface. Were it otherwise, the crossflow filter would be acting as a dead-
end filter, where the solids simply build up at the filter face. Using the theory
and concepts developed for reverse osmosis and ultrafiltration, the molecular
diffusivity of the retained material is one direct determinate of how fast it
diffuses away from the surface. The colloidal material retained by a microfilter
has even a lower diffusivity than the macrosolutes in ultrafiltration. The
redispersion rate of retained material is thus calculated to be very low. In fact,
microfiltration rates are often quite high compared to ultrafiltration, even at
lower crossflow velocity. The explanation seems to lie in a shear enhanced
particle diffusivity which results in dramatic increases in flux.2 The final state
of the solids retained in crossflow filtration differs from conventional dead-end
filtration. Crossflow devices produce a concentrated liquid retentate, not a dry
cake. As this retentate is recycled, it becomes more concentrated in retained
solids, the driving force for redispersion of material retained by the membrane
declines, and filtration rates decline. Ultimately, there comes a point where it is
not economical to concentrate further. Depending on the nature and value of the
permeate and the retentate, techniques exist to achieve high recovery of the
products.
Membrane filters operating on feeds with medium loadings of solids (<0.5%)

are generally operated in dead-end flow. Commonly, the surface of the membrane
is protected by a guard filter made of packed glass or asbestos fibers, which
entrains most of the larger solids before they reach the membrane. The structure
acts as a depth filter backed by a membrane filter. Some of these devices are
quite sophisticated, because both prefilter and membrane filter can be charged to
give superior non-plugging characteristics. In some applications, these composite
structural filters attain the same outstanding characteristics as asbestos filters.
Fluids with low solid loadings (<0.1%) are almost always filtered in dead-end
flow, where the membrane acts as an absolute filter as fluid passes directly
through it. For some time, track-etched filters dominated this market. Now,
other membranes compete successfully.
Earlier, reference was made to membranes with conical pores, and the
desirability of this shape for the prevention of plugging. In some low-solids
loading applications, these membranes are run "upside down", that is, with the
wide part of the cone towards the process stream. fn this way, the cone serves
3s a trap for particles. This configuration mimics that of a structured filter
which wraps a coarse filter outside progressively finer filters. Although the
membrane eventually plugs, its dirt holding capacity is increased. This operating
scheme is only appropriate where the load of material to be retained is low.
6.3.2 Module Design Considerations
For a membrane to be a useful device, i t must be packaged in a way that

permits the membrane to operate efficiently. Many types of membrane holders
and devices are available.
6.3.2.1 Dead-end filter housings

Disk holder$ Disk holders represent the simplest membrane filter housing,
and their design has evolved slowly since their introduction in the 1950s. The
membrane is fitted between two plates, a porous one on which the membrane
filter is supported, and a feed plate containing a cavity to permit the fluid to
contact the membrane freely. The devices are usually plastic or stainless steel,
and the membrane is usually sealed with an O-ring.
Pleated cartridees Many membranes are pleated, then formed into a

cylinder, substantially increasing the membrane area that can be fit into a given
volume. The devices resemble the familiar automotive air filter. End caps are
generally attached using curable liquid or melt sealants. Cartridges are then
fitted into housings, either singly or in groups. The housings are simple
pressure vessels, although their design may become elaborate.
Dead-end sDiral: Spiral-wound modules are popular crossflow devices, widely

used in reverse osmosis and ultrafiltration. A hybrid crossflow/dead-end filter is
being manufactured for microfiltration using the principle of running a spiral-
wound module as a dead-end filter. During initial operation, until significant
solids have built up, most of the feed passes across the membrane, becoming
dead-ended only near the outlet of the sealed spiral device. When filled with
solids, the spiral operates totally as a dead-end filter.
Air-Dulsed caDillary A novel system to handle retained solids is employed

by Memtec (Australia). Their device, illustrated in Figure 6-7, operates as a
pulse-cleaned, dead-end and crossflow filter. The feed stream passes along the
outside of microporous capillaries. It quickly builds a layer of retained material
on the surface, acting as a filter-aid formed from retained material. When the
layer has developed enough resistance to impede the filtration unacceptably, the
filtration is stopped, and air is pushed through the inside of the capillaries and
the pores to blow off the filter cake. This backwash frequency is every 10-30
minutes, with a duration of 30 seconds. For high solids loadings, Memtec is able
to operate its system in crossflow, since it can run at conversions per pass as
low as 50%.s
6.3.2.2 Crossflow devices
For the applications of greatest interest to this study, crossflow devices are
dominant. Since they are discussed in more detail in the section on
ultrafiltration, they are covered only briefly here.
When significant quantities of solids are present, crossflow operation gives

the highest output per unit membrane area. The simplest crossflow device is a
membrane formed inside a tube made from a strong, porous material. The feed
runs down the inside of the tube, under pressure. Permeate passes through the
membrane, then through the porous support.
Another commonly used device is the parallel-plate module, or cassette.
Capillaries. membranes spun so that their porous sublayer provides

mechanical support against operating pressure, are operated with bore-side feed.
By elevating the permeate pressure above the feed pressure periodically, forcing
Microfiltration 341
Operating Mode
Figure 6-7. Operating and backwash mode for the Memtec air-pulsed capillary
module.
permeate backwards through the membrane, capillary membranes may be cleaned

effectively while still running on the process stream. By so doing, solids built up
on the membrane are pushed back into the feed. This operation is fundamentally
different than the air-purge dead-end filter, since it relies on crossflow to d o
almost all of the redispersion of retained solids. The permeate back-pressure
cycle is used to remove small quantities of foulant material deposited on the
membrane.
Reverse flow of capillaries is also useful to remove a partial blockage of the

flow channels. Permeate being forced backwards into the capillary bore expands
it slightly, and also pushes the blocked material back in the direction from which
i t entered. Reverse flow is also practiced in capillary membranes on some dirty
streams. By reversing the feed direction, material that accumulates at or near
the entrance to the capillary bundle is swept away from the face.
As mentioned under dead-end flow, the air-purged capillary membranes

offered by Memtec may be set up to operate at the lower end of crossflow
velocities. Because the feed is external to the capillaries, the effectiveness of
the hydrodynamic sweeping is reduced. The use of periodic air-pulse cleaning
seems to compensate for the reduced level of flow.
6.4 STATUS OF T H E MICROFILTRATION INDUSTRY
6.4. I Background
Membrane filters can be said to have begun with Zsigmondy during The
Great War. Development was very gradual during the 1920s and 1930% and i t
occurred principally at Sartorius GmbH. At the end of World War 11, U.S.
occupation forces in Germany were assigned to evaluate German technology and to
transfer promising developments to the U.S. Membrane technology was one of the
German developments determined to be critical, particularly for its usefulness in
zssessing the level of microbial contamination in water supplies. After 3 period
of development in both academic and commercial laboratories, the company that is
the predecessor of Millipore led in the commercialization of microfiltration
membranes and supplies.
6.4.2 Suppliers
The microfiltration industry features some large companies with high growth
rates, good profitability, and healthy balance sheets. That general situation has
naturally attracted attention, and newer entrants are numerous. Table 6- 1
estimates the sales attributable to microfiltration membranes from the total
revenues of these suppliers. For the estimate, membranes and membrane-related
hardware have been combined. The definition of a membrane is fairly broad,
including polymeric, ceramic, inorganic or sintered metal dead-end or crossflow
devices that make a separation in the 0.1-5 pm range. Wound, spun-bonded, and
wire-based filters are excluded. The inclusion of sintered metal, while arbitrary,
does not influence the total numbers significantly.
Microfiltration 343
Table 6- 1. The Microfiltration Industry

Approx. Membrane
Sales
Company Location (millions S) Products/Comments
Millipore Bedford, MA 375 Dominant U.S. supplier of

laboratory microfiltration
membranes. Major producers
of cellulose membranes for
sterilizing filters, particulate
removal.
Pall Glen Cove, NY 150 Pall has expanded into the
microfiltration market in recent
years. Offers sterilization and
particulate removal filters
Sartorius Goettingen 90 First commercial producer of
West Germany microfiltration membranes.
(Hayward, CA) Principal supplier to the
European laboratory market.
Gelman Ann Arbor, MI 60 Sterilizing and particulate
removal filters. Millipore’s
principal competitor in the
laboratory market.
Fuji Filters Japan 15 Principal Japanese micro-
filtration company.
AMF Cuno Meriden, C T 10 Produces surface-charged
microfilters for use in phar-
maceutical and food industries.
Arnicon Lexington. MA 10 Laboratory microfiltration
membranes for sterilization
and other applications.
Memtec Windsor NSW, 7 Laboratory microfiltration and
A us tral ia innovative uses of MF.
Schleicher & Dassel, West 5 Originally German paper com-
Schuell Germany pany. Serves laboratory filter
market.
Nuclepore Pleasanton, CA 5 Principal product is a nuclea-
tion track membrane used in
laboratory and analytical
applications.
Table 6- 1. The Microfiltration Industry (continued)

Approx. Membrane
Sales
Company Location (millions $) Products/Commen ts
Hoech- Charlotte, NC <5 Celgard" microporous mem-

Celanese branes for various appli-
cations.
W. L. Gore Elkton. MD <5 GoreTexe microporous mem-
branes.
Other specialized 5 - 10 Ceramic, metal, metal oxide

or small companies: in total membranes and other special-
Anotec, Norton, Alcoa, ized products.
Osmonics, Mott, Brunswick,
Whatman.
6.4.3 Membrane Trends
Early in the development of commercial microfiltration membranes, cellulose

nitrate (collodion) became the material of choice. The abundance of polymeric
materials favored today by membrane fabricators had not been invented.
Collodion posed serious safety problems in manufacture, but it was a very good
polymer for laboratory membranes, and is still used for a few specialty
membranes, either by itself, or blended with cellulose acetate.
Since that time, the industry has moved toward tougher materials. both
chemically and mechanically, that can be pleated, autoclaved, washed in solvents,
acids, and bases, and still retain their original operating properties.
Polymer membranes are by far the market leaders in microfiltration. The

major firms in the worldwide microfiltration business all sell polymer membranes
in overwhelmingly greater quantities than the more trendy and more discussed
inorganics. Nylon, polysulfone, and polyvinylidene fluoride are the major polymers
used, in addition to the old workhorse cellulosics. Polypropylene is widely used
in process microfiltration.
Recently, a steady stream of innovative membranes have been introduced into

the market. The traditional polymeric membranes are made by dissolving a
polymer in a water-miscible solvent, then casting it on a surface from which the
solvent is removed by exchange with water, either from humid air, from an
aqueous solution, or from another of many variants. This process, when properly
conducted, forms a membrane. Commercial microfiltration membranes are made
from cellulose acetate, several nylons, polysulfone, polyethersulfone, polyvinyl
chloride, the copolymer of vinyl chloride and acrylonitrile, and polyvinylidene
fluoride by this traditional method.
Microfiltration 345
Polymeric microfiltration membranes can also be produced by other methods,

such as thermal inversion (Enka and Memtec polypropylene membranes), selective
leaching (Millipore), membranes made by sintering (Mott Metallurgical) or
stretching' (Goretexe, W.L.Gore and Associates, and Celgard", Hoechst-Celanese)
and polymeric membranes made as porous sheets by photopolymerization (Gelman).
Photomicrographs of the Goretex and Celgard polymeric membranes are shown in
Figure 6-8.
Some of the newer microfiltration membranes are ceramic membranes based

on alumina (Alcoa-Ceraver), membranes formed during the anodizing of aluminum
(Anotec),6 and carbon membranes (GFT). As with organic membranes, there are
several ways to prepare inorganic membranes. Ceramic membranes may be made
by slipcasting onto a porous support, then firing. Preparation of the particles for
the slip so that they are both fine enough and monodisperse is difficult. and the
sol-gel technique is favored. Organic additives to promote adhesion and to modify
viscosity are commonly used. The slip, as an aqueous dispersion, is cast onto a
dry porous support, where capillary effects draw water out of the slip and into
the support, leaving the gel particles from the slip concentrated at pore openings
in the support.6 Subsequent firing fixes the particles to the substrate, fusing
them into a permanent layer. The unconsolidated openings make up the active
membrane layer. The Alcoa alumina membrane is shown in Figure 6-9.
In addition to polymeric and ceramic membranes, there are some other less
widely used microfiltration membranes. Glass membranes are normally prepared by
the thermal separation of a glass into two phases, one of which is soluble enough
in a leachant to be extracted. Sintered metal membranes are fabricated from
stainless steel, silver. gold, platinum, and nickel, in disks and tubes.
Dynamically formed membranes have long been of interest, and some are sold
in the microfiltration field. The most prominent by far is zirconium oxide
deposited on a porous carbon tube. The membrane is formed by passing the
suspension of ZrO, across the porous support, laying down a semipermanent
precoat. The membrane seems to be stable when required, and unstable
'
(removable) when desired. Ultrafiltration and reverse osmosis membranes have

also been made by this technique. Rhone Poulenc recently bought SFEC, the
primary supplier of these types of membranes in ultrafiltration. who do some
business in microfiltration.
Aluminum oxide membranes are formed by the anodic oxidation of metallic

aluminum. If the oxidation bath is properly controlled, it is possible to make a
membrane with a very high density of uniform, thin, pores? Anotec is the
supplier of these membranes.
a) Goretexm
b) Celgard"
Figure 6-8. Photomicrographs of a) Goretex". and b) Celgard" polymeric

microfiltration membranes.
Microfiltration 347
Figure 6-9. Photomicrograph of an Alcoa alumina membrane.

Carbon membranes are prepared by the controlled pyrolysis of microporous

polymeric membranes. Carbon composite membranes are still under development,
but these membranes have the potential for toughness, pore-size tailoring, and
extremely thin active skins. These properties would make them important
industrial membranes if applications develop in sufficient volume to make their
manufacture economical.8 GFT is the leading supplier of carbon microfiltration
mem branes.
In some applications, membrane microfiltration competes with depth

filtration. Unstructured depth filtration employs diatomaceous earth. pearlite.
asbestos, or other finely divided filter aids. Structured depth filtration uses
paper or other fiber structures, including glass and polypropylene.
6.4.4 Module Trends

Microfiltration modules began as flat-sheet filters for use in the laboratory.
and were soon incorporated into plate-and-frame devices. A growing diversity of
applications has led to the development of numerous devices such as the spirai-
wound module, copied from other membrane applications, the pleated cartridge.
referred to above, the stack-filter module, designed for certain pharmaceuticai
applications. and capillary modules. Continued development of module types is
likely. but the low manufacturing cost of spirals and capillaries makes them the
product to beat for volume applications in crossflow. Pleated cartridges enjoy a
similar advantage in dead-end applications.
Some of the more innovative membranes developed recently are amenable to

compact. economical fabrication. Carbon membranes may be pyrolized from
fibers, and are described as being formed already sealed to an end plate." Most
ceramics are available in tubular form, but one firm has pioneered a low-cost
monolith,' and another makes ceramic capillary tubules which can be wound up
into a cartridge.
6.4.5 Process Trends

Microfiltration, a relatively mature industry, has had its most profitable
growth in relatively small filters operating in high-value applications. The
industry trend is to build on this base, but to expand into lower value, higher
volume applications. Most of this growth will be with membrane devices
operating in a manner that handles retained material efficiently, meaning either
crossflow or backwash.
6.5 APPLICATIONS FOR MICROFILTRATION TECHNOLOGY

6.5.1 Current Applications
Most applications for microfiltration membranes are relatively small,
specialized uses that add up to a large market. These applications, while very
important to industry, health. and research, do not have a significant impact on
energy use.
The existing market for microfiltration membranes and equipment is on the

order of SI billion. This market is served by large companies, commanding
generous R&D budgets and possessing excellent market research, marketing, and
management. Within the areas they have chosen to pursue, sterile filtration,
Microfiltration 349
medical and biotechnology applications, and fluid purification, it is hard to

imagine a real research need that has not been identified or that could not be
supported by internal funds.
In addition to those well-established applications, there is a major effort to

introduce microfiltration into a wide variety of process applications. These
applications, while presently small in number, use large quantities of membrane.
Their potential for growth is great, and in terms of membrane area installed. they
may grow fast enough to catch the other applications in the 20-year timespan of
this report. It is unlikely that they will match the conventional microfiltration
applications in dollar value, however. Table 6-2 lists a number of these
applications, their primary markets and competing processes, and points out somz
of the problems of the technology for each.
Table 6-2. Current Process Microfiltration Applications
Application Customers Equip. Type Competing Processes Problems
Haze removal Food Spiral-wound; Diatomaceous High viscosity;

from gelatin companies plate-and- earth very high protein
frame filtration passage required
Dextrose Corn Spiral-wound; Diatomaceous High viscosity
clarification refiners plate-and- permeate
frame
Wine Wineries Spiral-wound; Asbetos Fouling; yield;
capillary flavor
plate-and-
frame
Beer bottoms Breweries Spiral-wound; Centrifuge Foam stability;

recovery capillary flavor
plate-and-
frame
Bright beer Breweries Pasteurization Reliability;
sterilization Huge market
potential
Pharma- Biotech Dead-end Market huge,
ceutical/ and Phar- filtration but usually
Biological maceutical done on a smaller
companies scale than others
6.5.2 Future Applications
Although microfiltration is a mature technology, there are several new

applications in which it could become the filtration method of choice. These are
summarized in Table 6-3 and discussed below. Their importance for the future
growth of microfiltration has been rated between 1 and 10, I being the lowest.
Table 6-3. Future Applications for Microfiltration

Application Prospect for
Area Realization Importance Comments/Problems
Drinking Good 8 Economics potentially superior to sand fil-

water t e n if market acceptance is good. Large
scale use would complement displacement of
chlorine disinfection. Modular construction
could change economics of waterworks.
Municipal Fair 5 Competes with other technologies. Superior

sewage to ultrafiltration in cost, inferior for
treatment containing fermentation reactions. Seems
capable of virus retention. Impact of suc-
cess would be large. Economic and techni-
cal impediments. Long-range potential for
distributed processing of sewage.
Diatomaceous Excellent 9 Membrane does superior job; economics

earth dictate pace.
displacement
Hydrocarbon Poor 9 Removing waxes, asphaltenes. Economic

separations hurdle high, working conditions extreme
Milk-fat Fair 5 Membrane could replace centrifuge in re-

separation covery of butterfat. Huge volume.
Abattoirs Fair 3 Membrane could remove cellular material

and debris prior to use of ultrafiltration to
recover proteins.
Food and Excellent 7 Many applications in broad food areas based

beverage on microfiltration ability to retain micro-
organisms without affecting desirable prop-
erties. Applications to be large, but to
grow slowly.
Yon-sewage Good 7 Microfiltration looks good for removing

waste intractable particles in oily fluids,
treatment aqueous wastes containing particulate
toxics, and stack gas.
Coal liquids Fair 6 Particulates a tough problem. Membranes

ought to succeed, if tough enough and
cheap enough.
Paint Good 5 Separation of solvents from pigments
Biotech Excellent 8 Concentration of biomass; separation of

soluble products.
Microfiltration 351
6.5.2.1 Water treatment
The largest emerging opportunity for microfiltration is for the treatment of

municipal water, permitting it to be sterilized without chlorine. This would take
microfiltration back to its World War I1 roots. There is no doubt that micro-
filtration membranes have the ability to remove bacteria from water. A recent
Australian studylo showed that microfiltration membranes can also remove viruses
from contaminated surface water. Since viruses are much smaller than the pores
in an microfiltration membrane, the finding has been attributed to the viruses
being adsorbed on clay particles, which are large enough to be caught by a
microfilter. There are, however, concerns about subsequent contamination in the
water distribution system, since chlorine has a residual effect that protects water
against contamination after treatment. Recent Federal regulations probably mean
that chlorination will continue to be required for drinking water supplies.
For communities whose water is hard, it has also been proposed to

incorporate lime softening into the microfiltration system. This ancient
technology relies on the fact that most calcium hardness is in the form of the
bicarbonate. By adding calcium hydroxide to the water, the reaction
C ~ I ( H C O ~t ) Ca(OH),
~ = 2 CaCO, t2H,O
reduces the calcium level in the water to the solubility limit of calcium carbonate.
The newly precipitated calcium carbonate acts as a precoat on the very open
microfiltration mernbrane.l' Fresh calcium hydroxide may be generated by
heating some of the calcium carbonate to the calcining temperature. Either
approach to the microfiltration of water would require a major change in the
economics of microfiltration. Public health regulations are an additional limiting
factor, as those responsible will need to be shown that the new technology is
safe.
6.5.2.2 Sewage treatment
The other potentially very large market for which microfiltration might be a
candidate is the treatment of municipal sewage. A scheme proposed by Memtec
(Australia) would shift the treatment of sewage to distributed processing, a plan
that envisions many small sewage treatment facilities. centrally monitored. Should
this plan ever become reality, the market for microfiltration membranes would be
immense.
6.5.2.3 Clarification: diatomaceous earth replacement
The economics of diatomaceous earth purchase and disposal make i t an

attractive target for displacement by microfiltration. Microfiltration membranes
3re usually capable of doing the same job as diatomaceous earth filters, only
better, with higher clarity products and higher yield. These advantages are
currently marginal in the biggest applications, with not enough economic incentive
to achieve the displacement of installed diatomaceous earth filters. As
microfiltration starts to chip away at the more attractive applications, however,
costs will drop. There is a likelihood that microfiltration will become preferred
over diatomaceous earth filtration in new installations within five years, and that
it will become attractive enough to displace existing applications at some time
within 15 years.
6.5.2.4 Fuels
Fuel-oriented hydrocarbon separations by means of microfiltration represent

a high-risk, high-reward opportunity. Presently, most fuel applications appear to
be in high-temperature, physically aggressive environments. Economics will be a
severe test. To become the process of choice, microfiltration will need to be
cheap. Based on what we know today, ceramic or inorganic membranes are the
likeliest candidates, but the economic viability of processes based on these
membranes is a major uncertainty.
6.5.3 Industry Directions
For any of the major new process opportunities to come to fruition, low-cost
process equipment will be required. There is every indication that low-capital
designs are viable in process microfiltration. Historically, microfiltration has
worked from a massive and diffuse base in which low capital was desirable but
not really necessary. Economics we:'e dominated by the cost of membrane
replacement, daily in some cases, but rarely less frequently than monthly. AS
process engineers have started to attack applications such as glucose, beer
bottoms, and gelatin, the picture has changed. In these applications, membrane
life is dramatically longer. Classic microfiltration firms were competing with
ultrafiltration firms, for whom long membrane life is normal. Since this trend
means there will be less revenue from membrane replacement, there is a greater
incentive to make money on the original capital equipment sale. However, this
requires a reversal of the historical trend of declining equipment costs.
As the microfiltration industry targets water and sewage treatment, another

major cost decrease must be achieved. In time, and with sufficient volume, this
should be possible. However, this will require a well-coordinated effort, designed
to deal with the public policy issues of health, sanitation, and regulation for new
technologies for water and sewage treatment. Such an effort would cut years off
the implementation time for new technologies such as microfiltration. Past efforts
to improve equipment design and process economics in this market suggest that
dealing with the shape of a publicly acceptable solution at the front end might
produce a more cost-effective solution. Firms that are focused on this potential
are Memtec (Australia) and Allied Signal.
In the area of food processing, beverages, milk fat removal, and general
displacement of diatomaceous earth, the concerns are so dominantly in the
private sector, that given the absence of any overriding public issue, neither need
nor opportunity for a government program are apparent. Many ultrafiltration
firms are active in this area, particularly DDS, Koch Membrane Systems, Dorr
Oliver/Amicon (Grace), Romicon (Rohm & Haas), and Enka.
Large microfiltration firms continue their quest for products to fit their
vital medical, biological, and pharmaceutical markets. This huge, ongoing effort is
totally outside the scope of this report, even though it represents the major
activity in the microfiltration field.
As biotechnology begins to increase in scale. the size of the microfiltration

equipment will grow to process size. These markets should be intensely
scrutinized and hotly contested, if recent history is any guide. Membrane makers
Microfiltration 353
will need to address issues of fouling, non-selective protein adsorption, and

lifetime, in addition to specificity and compatibility.
There is every indication that market forces will satisfy the demand, and
that U.S. firms will continue to lead even though they will be continue to be
challenged from Japan and Europe.
Some of the trends that will dominate the microfiltration industry are
summarized in Table 6-4.
Table 6-4. Future Industry Trends for Microfiltration
Topic Prospect Importance Comments/Problems

For Realization
cost Excellent 10 Huge potential applications will

require commodity pricing, far from
today’s reality. Bright prospects
for success based on hemodialysis
experience.
Continuous Good 8 Applications where biological

integrity integrity is required need evidence
testing of continued compliance, especially
if operation is remote and
automatic.
Nonfouling, Good 7 Critical for abattoirs, dairies, beer,

cleanable, wine. Must be tolerant of the
durable industry approved sanitizer.
membranes
Cheap, Fair 5 Current prospects for cheap

trash- membranes do not lend themselves
tolerant to incorporation into trash-tolerant
designs modules. Problem could be solved
less desirably by pretreatment.
Foul in g Good 9 Critical to improving rates. See

also ultrafiltration.
High- Good 9 Inorganics best bet, but refractory

temperature polymers good long-shot.
solvent resistant
membranes
6.6. PROCESS ECONOMICS
The economics of small microfiltration plants are so sensitive to the

application that it is difficult to generalize costs meaningfully. In this report, a
moderate-sized water plant is used to illustrate the current economics of one of
the least costly applications. The capital and operating costs are given in Tables
6-5 and 6-6, respectively. In fact, all other known applications will have higher
capital and operating costs.
A good approximation of the costs of operating a large-scale industrial

microfilter in crossflow mode can be found in the ultrafiltration section (Chapter
7) of this report, since the equipment in large-scale applications is similar.
Membrane life for a prototype microfiltration application would be 4 years on
average, somewhat longer than for ultrafiltration.
Table 6-5. Capital Costs for a Surface Water Microfiltration Plant
Basis: A microfiltration surface water treatment plant processing 500

m3/day (130.000 gpd), using capillary modules with air backwash.
Item Installed % of Total

cost ($1
Replaceable Membranes 60,000 37
Pumps 29,000 18
Pipes and Valves 26,000 16
Tanks and Frame 24,000 15
Instruments and Controls 22.000 -
I4
TOTAL 16 1,000 IO0
The total installed cost of this base-case plant is S320/mS water treated per
day, or S1.30/gpd. Note that the fraction of the capital in the replaceable
membranes is relatively high in this plant because the housing and membranes are
integral. This is normal for capillary membranes. Costs for the replaceable
elements for a spiral plant would be lower. The total operating costs are
estimated to be %0.23/m3 treated (SO.O6/kgal).
Table 6-6. Operating Costs of a Surface Water Microfiltration Plant
Item S/Yr % of Total

Membrane replacement
(2-yr guarantee) 30,000 57
Power (0.6 kWh/m3,

S0.07/kWh, 500x365 m3/yr) 1,700 15
Maintenance. P & I @ 6% of
nonmembrane capital 6,000 I1
Labor, 10 hr/wk m -
17
TOTAL 52.700 100

Microfiltration 355
6.7. ENERGY CONSIDERATIONS
Microfiltration uses mechanical energy to drive fluids through and past the
membrane. The ideal energy required to move a fluid through a microporous
membrane is negligible. At an operating pressure difference across the membrane
of 5 psi in a dead-end filter, energy requirements are only 0.01 kWh/ms of
permeate passing through the membrane.
Crossflow devices consume energy to keep the membrane surface clean, as

well as to push the permeate through the membrane. Practical crossflow devices
consume about 5 kWh/ms permeate, essentially all of which is used to minimize
polarization, increase rate, and thus lower the membrane area requirement and
capital cost. Membrane microfilters compete with centrifuges, clarifiers,
coagulation, and with nonmembrane filtration devices such as precoat filters.
Microfilters running in dead-end configuration are comparable in energy
requirements to competitive processes. Crossflow microfiltration consumes
somewhat more energy than processes that compete with it, but it is not an
energy-intensive process.
There are numerous energy trade-offs to be considered in determining

whether membrane microfiltration saves energy on balance compared to competing
processes. For instance, there is a minor benefit from considering energy
required to mine and dispose of diatomaceous earth. Balanced against the
materials needed to construct a microfilter. it is not clear which requires more
energy. It is far easier to make an argument for microfilters as a pollution
reduction device than as an energy reduction device.
There are several potential uses of microfiltration, primarily related to fuel

processing, that would provide energy savings, if successfully developed. These
fuel-related applications are discussed briefly below. In spite of the massive
alteration in the microfiltration market that these changes would create, the
overall reduction in direct energy consumed would be small. Microfiltration plants
do not differ greatly in energy consumption from conventional filters, no matter
how large. There will be savings indirectly, from mining, transport, and disposal
of diatomaceous earth, but these are small.
Microfiltration membranes have the potential for significant impact on the

processing of fuels. Given the fact that most liquid fuels are viscous, and that
viscosity declines with temperature, processes in fuel-related applications are
likely to be at temperatures above the operating limits of all but the most exotic
polymer materials. Potential gas-phase applications for microfiltration also require
stable operation at high temperature. Therefore, these applications will require
membranes different from those with which we are familiar, such as advanced
ceramic, mineral, carbon, metallic, or exotic polymeric membranes.
Oil refining has several potential applications for microfiltration and

ultrafiltration membranes. Separation of asphaltenes is the application most
frequently mentioned, and if it can be achieved, it would reduce the energy
necessary to accomplish solvent de-asphalting. Asphaltene removal can be an
ultrafiltration or a microfiltration application, depending on the process. The
microfiltration process uses a 0.03-pm membrane to catch catalyst particles in a
residual hydrotreater blowdown. Colloidal asphaltenes and some of the metal
content are captured as well, and recycled to the hydrotreater with the catalyst.
The application uses a membrane at 450'C.12 Heavy crudes, shale oils and tar
sands all have particulate problems for which high-temperature microfiltration
membranes might be a good solution.
Coal liquefaction produces liquids with submicron ash difficult to remove by

conventional means. Microfiltration is a promising candidate for this separation.
Microfiltration of gas streams in fuel related applications will be at high
temperature. Two applications with a strong energy angle are the removal of
soot from diesel exhaust, and the removal of particulates from boiler stack gases.
In both these applications, there is existing or emerging technology which
competes with a membrane solution. A membrane may be superior, due to
compactness, selectivity, and high operating temperature, perhaps higher than is
available with competing technology.
A third application with a strong relationship to energy is the removal of

particulates from coal gasification plants. In addition to the ability to operate at
very high temperature, these membranes would need to be very resistant to
erosive and chemical attack.
Treatment of used lubricating oil, particularly automotive crankcase oil is

another possible use for high-temperature microfiltration membranes. Many of the
contaminants are particulates, and it is possible that microfiltration technology
could help this spent oil find a higher use, with less environmental impact than
the present technology for treating this waste.
6.8 OPPORTUNITIES IN THE INDUSTRY
6.8. I Commercially-Funded Opportunities
Many current trends in the microfiltration industry will proceed regardless of

outside support. Membrane producers will continue to work on continuous
cleaning, pulse cleaning. backflushing, reduced fouling. and exotic membrane
materials. Progress will be slow, because the markets for these materials are too
small and the costs too high to assure high growth.
6.8.2 Opportunities for Governmental Research Participation
In reviewing the possibilities for the future of this robust and healthy
technical field, where so much money has been spent by private companies with
numerous dramatic successes, the question arises as to whether outside support is
needed at all. Millipore was founded in 1954; Pall entered the microfiltration
field just twenty years ago. Those two firms alone have membrane microfiltration
sales exceeding 50.5 billion, with sales doubling every 5 years.
Nonetheless, shifts in membrane materials and markets have been slow.

Major capital and commitment is required to launch a new sub-technology. Some
of the more innovative, but as yet unproven, technologies are the result of the
acquisition of assets developed for other purposes at a price below the cost of
development.
Microfiltration 357
Ceramic membranes are an example of this type of technology acquisition.

Alcoa bought Ceraver. a firm which had built ceramic barriers for the French
uranium isotope separation enterprise. During its early history, Ceraver was
heavily funded by an outside institution to develop a product similar to a
microfiltration membrane. Therefore, Ceraver easily mastered the microfiltration
technology. In their purchase of Ceraver, Alcoa acquired a small but successful
marketing organization, large capacity, and a presumably competitive Cost of
manufacture, all of which permitted rapid advancement down the learning curve.
In contrast, in an attempt to enter the same business, Norton relied on

internal funding and internal know-how. Attempts to capitalize on one-of-a-kind
government procurements, and then to force the technology into markets where
there was adequate, lower cost technology, produced too little cash flow to
sustain the business, and it was offered for sale. Limited demand for the unique
traits of Norton's technology forced them to attack markets served adequately by
other membranes, in an attempt to build volume, improve the product, and thus
lower costs.
When there are societal or strategic interests at issue, outside Support is

essential to get past the high early technical and economic hurdles. With enough
lead-time, the existing manufacturers will expand into new markets. However, if
those applications are to appear suddenly in the future, outside Support may be
necessary to achieve quick implementation.
Where public agencies are the major market, the public procurement process
discourages expensive proprietary innovation. The innovator is often precluded
from harvesting the fruits of his innovation by the nature of civil procurement.
If innovation that primarily benefits the public sector is to come about, it will
almost certainly require support from outside.
A second area ripe for governmental research participation is in adapting

existing technology for high-visibility applications. An example is in petroleum
refining. For microfiltration techniques to find industrial process acceptance,
they must have some attributes absent in current offerings. They must be very
reliable at very demanding environmental conditions, cheap. and easily tailored to
fit dynamic needs. Oil refiners usually have ample assets. and a cost-saving new
technology will get attention from capable engineers. The development time and
cost, however, makes that future market less attractive than the less demanding
applications which have a far lower importance insofar as energy saving is
concerned.
Building on the huge technical advantage possessed by the U.S., and

extending the technology in ways that make it attractive to energy intensive
applications, is a prudent policy option. For the technology to be useful, it must
have excellent reliability under the most demanding conditions of temperature and
chemical stress. It must be very cheap to use, because most operations in the
fuels industry are quite low cost. Projects that promote invention and early
development of devices that fit potential needs in coal, shale. heavy crudes, and
other future fuel sources would be prudent additions to our technological base.
In summary, there are two major areas in which microfiltration could make a
significant change in between 1995 and 2010, and which are not likely to occur as
the result of a private, ongoing initiative. The first is the general handling of
water and waste water. As the report indicates, the energy consequences of
changing this are not very large. The second general area is fuel and combustion
related applications. Since the rewards are more speculative, only firms with
significant long-term resources, or those supported by an agency with a visionary
mission, will be able to discover whether microfiltration membranes will have a
future role in this area. The fuel field is energy-related. The separations in this
field tend to be energy intensive, and microfiltration is a low-energy alternative.
There are three companies that are positioned for some of these high-risk
potential opportunities. The major player with potential in this area is Alcoa,
through its Ceraver subsidiary. If any firm has the potential to innovate in this
field without assistance, it is Alcoa. Minor players are Westinghouse, who has
worked in the area of high-temperature gas cleanup for some time under
contract from the Office of Fossil Energy, and CeraMem.
CeraMem is a small startup firm whose major accomplishment has been the
adaptation of a cheap, high-temperature porous device into a cheap support for
microfiltration membranes. The Department of Energy has supported CeraMem's
development through several SBIR projects. The effort is directly focused on how
to make a cheap module, a critical necessity if microfiltration is to have any role
in the fuels field.
Memtec's basic interest is microfiltration, and its product is a capillary

polypropylene membrane with 0.2 micrometer pores. It is operated with shellside
feed, and operates in crossflow with relatively high (50%) typical conversion.
Modules operate at low pressure, 1.5-2 bar being common. Memtec's unique
feature is a periodic gas backwash, occurring 4-6 times per hour, typically.
Membrane fouling is well controlled by this technique, and it lends itself to
periodic leak checking, to insure that the pores have not grown. Memtec claims
the ability to detect one large pore in 1O12 normal pores. Their energy
requirement is modest, in the range of 0.2-0.5 kWh/ms depending on flux. The
air backwash requires about 10 L (STP)/m2, adding only about 10% to the energy
requirement.
Microfiltration 359
REFERENCES
I. R.H. Meltzer, Filtration in the Pharmaceutical sr , Marcel Dekker, New

York (1987).
2. A.L. Zydney, and C.K. Colton, "A Concentration Polarization Model for the
Filtrate Flux in Cross-Flow Microfiltration of Particulate Suspensions,"
Chem. Enn. Commun. 47, 1-21 (1986).
3. Memtec/Memcor product bulletin, PB88/02, Memtec, Locked Mail Bag 1,

Windsor NSW 2756, Australia.
4. R.W. Gore, "Porous Products and Process Therefor," US. Patent 4,187,390
(February 5, 1980).
5. R.C. Furneaux, et al.. U.S. Patent 4,687.551 (August 18, 1987).
6. H.P. Hsieh, "Inorganic Membranes," in New Membrane Materials and

Processes for Seoaration, K.K. Sirkar and D.R. Lloyd (Ed.), A.1.Ch.E.
Symposium Series 84, No. 261, 1 (1988).
7. R.C. Furneaux, et al., "The Formation of Controlled-Porosity Membranes from

Anodically Oxidized Aluminium," Nature 337, No. 6203, 147-9 (1989).
8. H.L. Fleming, "Carbon Composites: A New Family of Inorganic Membranes,"

1988 Sixth Annual Membrane Planning Conference, Cambridge, MA (1988).
9. R.L. Goldsmith, U.S. Patent 4,781,831 (November I , 1988).
IO. M. Kolega, R.B. Kaye, R.F. Chiew and G.S. Grohmann, "Disinfection of
Secondary Sewage Effluent by Advanced Membrane Treatment Technology,"
Memtec Ltd., Windsor NSW, Australia.
11. N. Li, private communication (1987).
12. R.L. Goldsmith, private communication (1989).

7. Ultrafiltration
by W. Eykamp. University of California, Berkeley
Ultrafiltration is a membrane process with the ability to separate molecules

in solution on the basis of size. An ultrafiltration membrane acts as a selective
barrier. It retains species with molecular weights higher than a few thousand
Daltons (macrosolutes). while freely passing small molecules (microsolutes and
solvents). The separation is achieved by concentrating the large molecules
present in the feed on one side of the membrane, while the solvent and
microsolutes are depleted as they pass through the membrane.
For example, an ultrafiltration membrane process will separate a protein

(macrosolute) from an aqueous saline solution. As the water and salts pass
through the membrane, the protein is held back. The protein concentration
increases and the salts, whose concentration relative to the solvent is unchanged,
are depleted relative to the protein. The protein is, therefore, both concentrated
and purified by the ultrafiltration. Figure 7-l illustrates the ultrafiltration
process.
Ultrafiltration may be distinguished from two related processes, reverse

osmosis and microfiltration. Given the same example of a solution of protein,
wafer and salt, the reverse osmosis membrane will pass only the water,
concentrating both salt and protein. The protein is concentrated, but not
purified. The microfiltration membrane will pass water, salt and protein. In this
case, the protein will be neither concentrated nor purified, unless there is another
larger component present, such as a bacterium.
Ultrafiltration membranes are typically rated by molecular weight cutoff, a

convenient but fictitious value giving the molecular weight of a hypothetical
macrosolute that the membrane will just retain. Microfiltration membranes,
however, are rated by pore size, specifically the pore diameter. In practice, the
distinction between the two membrane processes is blurred. There is some
overlap in size between the largest polymer molecules and the smallest colloids.
Microfiltration membranes with the smallest pore sizes sometimes retain large
macrosolutes. An ultrafiltration membrane is sometimes used for what appears to
be a microfiltration application because in that use it has a greater throughput.
Most ultrafiltration processes operate in crossflow mode, although a few

laboratory devices and industrial applications operate in dead-end flow. The use
of check filters or guard filters at the point-of-use in ultrapure water
applications is an example of dead-end ultrafiltration.
Most ultrafiltration membranes are made from polymers, by the phase

inversion process. Ultrafiltration membranes may also be formed dynamically, for
example by the deposition of hydrous zirconium oxide on a porous support under
controlled conditions of flow and pressure. A few ultrafiltration membranes
prepared from ceramic materials are available.
360
Ultrafiltration 361
Retentate
(liquid and
Feed (liquid and t macrosolutes)
macrosolutes) s, 3
~IYSSXSX\\\V- Membrane
t Permeate
I (liquid)
Figure 7- I. A schematic diagram of the ultrafiltration process.

Ultrafiltration membranes may be characterized in terms of pore size and

porosity, even though there is little direct evidence for the kinds of pores that
the terminology suggests.l-s A frequently used model characterizes the
membrane as a flat film with conical pores originating at its surface, as seen in
Figure 7-2. The surface pores are large enough to permit passage of solvent and
microsolute molecules, but are too small for effective penetration of the larger
macrosolute. The conical shape is desirable, in that any entity that makes it
through the opening at the membrane surface can continue unimpeded; there is no
danger of pore-plugging.
Membranes may be made with differing pore sizes, pore densities, and pore-
size distributions. These attributes are determined by measuring the flux of pure
water through the membrane. The membrane is tested with dilute solutions of
well characterized macromolecules, such as proteins, polysaccharides, and
surfactants of known molecular weight and size, to determine the molecular
weight cutoff. The above description implies the possibility of separating large
macrosolutes from small macrosolutes by using an appropriate pore size. In
practice, however, due to concentration polarization, only limited changes in the
relative concentration of macrosolutes of different size are feasible.
7.1.1 The Gel Model
Ultrafiltration is a pressure-driven operation. The flow of solvent through

the membrane increases linearly with pressure, according to Darcy’s law.
However, experimental data fit this law only in the case of very clean feeds, or
at very low pressures as can be seen in Figure 7-3, a typical experimental result
for an ultrafiltration membrane.
The flow through the membrane increases linearly with pressure at low
pressure values. As pressure is increased to higher values, the membrane
throughput levels off and becomes constant. This unexpected behavior has
attracted considerable attention,‘-s starting with a seminal paper by Blatt, who
pointed out that the concentration of macrosolute at the face of the membrane
must increase to a higher level than its concentration in the bulk of the fluid,
even in the presence of vigorous stirring. The macrosolute is carried to the
membrane by the bulk flow of solvent, but it cannot pass through the membrane.
It must then “swim upstream” back into the bulk of the solution whence it came.
Because the effects of stirring are insignificant very near the surface of the
membrane, diffusion becomes the only means by which the macrosolute may
redistribute itself. Fick’s law describes the flux of the macrosolute, J,, away
from the membrane. D is the molecular diffusivity of the macrosolute, and AC is
its concentration gradient
J, = D VC (1)
Ultrafiltration 363
43 ..
Feed
flow
.
- : -
Residue
enrlched
6 .. .
in macrosolutes
l
.
2 .
.
Permeate: solvent and microsolutes
Figure 7-2. A model of the ultrafiltration of a solution containing

macrosolutes (e.g., proteins) and microsolutes (e.g., salts).
Unfouled water,
flow independent
Process flux
Flux
Process flux
Transmembrane AP
Figure 7-3. The effect of transmembrane pressure on the ultrafiltration flux.

Ultrafiltration 365
Macrosolutes have low molecular diffusivities, so the rate of redispersion

into the bulk fluid is low, especially in the first few moments when the
concentration gradient is low. At first, macrosolute will arrive at the membrane
faster than it diffuses away. This accumulation cannot continue for long,
however, and a steady state soon develops in which the rate of redispersion
balances the rate of arrival. It is more important to state the converse of this
equivalence, that the rate of arrival balances the rate of redispersion, for that is
the critical factor in controlling the rate of an ultrafiltration process.
The ultrafiltration flux declines as the concentration of macrosolutes in the

feed increases. This is indicated in Figure 7-4, a typical plot of experimental
data in which each of the data lines shown represents a constant stirring rate.
The fact that the intercept is approximately 35% macrosolute by volume for many
different materials and for differing rates of stirring, led Blatt to propose that
the macrosolute in fact forms a new phase, that of a polymer gel layer.
Increasing the transmembrane pressure difference will then result in an increased
gel layer thickness, which negates the increase driving force; i.e., the flux
remains constant. This model is a good predictor of experimental data, and has
been used widely. In recent years, it has become clear that the concentrated
boundary layer adjacent to the membrane surface does not have to be gelled in
order to reduce or limit the ultrafiltration flux. The concentrations in the
boundary layer can be so high that osmotic pressure plays a significant role, even
with macromolecules. The osmotic pressrue of macromolecular solutions increases
very rapidly with concentration and can lead to flux limitation just as severe as a
gel layer.
The existence of a polymer gel immediately raises problems about the ability
of an ultrafiltration membrane to fractionate polymers. Gels are known to be
highly entangled polymeric networks. How can a smaller polymer wiggle through
a concentrated tangle of larger polymers and find the membrane pore through
which it may theoretically fit? In fact, whether the gel hypothesis is literally
true or not, if there is a “traffic jam” at the membrane surface, the cars may be
just as stuck as the trucks. The common heuristic is that for the separation to
take place in ultrafiltration with reasonable efficiency, there needs to be a factor
of 100 in the ratio of the sizes of the materials separated.
Ultrafiltration membranes are used to separate colloidal materials from fluid

mixtures, as well as macrosolutes from true solutions. In principle,
microfiltration would be expected to be the process of choice for colloidal
suspensions. However, for one major application, electrocoat paint, and some
minor applications, ultrafiltration is the process of choice. Apparently, the
ultrafiltration membranes resist being grossly plugged by the sticky paint solids,
because their critical dimension is too small for the paint particles to penetrate.
The use of ultrafiltration membranes for separating colloids presents

problems for the gel model, since colloids have much lower diffusivities than
macromolecules in solution. This problem has been discussed in the literature and
is analyzed further in Chapter Six on microfiltration.7s
40 I
30
Flux
20
10
” 0.1 0.3 1.0

Log concentration (volume fraction)
Figure 7-4. Variation in membrane flux with feed concentration at different

stirring rates.
Ultrafiltration 367
Mass transfer in the boundary layer is the rate-controlling mechanism in

ultrafiltration. The design of ultrafiltration equipment is governed by the need to
reduce concentration polarization. Plugging and fouling are other important factors
that affect membrane system design and operation.
7.1.2 Concentration Polarization
The dynamic equilibrium established as retained material is carried towards

the membrane and is transferred backwards from the membrane is the process
that limits the output of an ultrafilter once the “knee” of the flux vs. pressure
curve, illustrated in Figure 7-3. has been reached. The retained material cannot
continue to build up at the membrane without limit. At steady state, the flow of
macrosolute carried by solvent towards the membrane must be equivalent to the
flow of macrosolute back into the bulk solution, regardless of the mechanism.
Therefore, the flow of solvent towards the membrane, and consequently the
solvent flux through the membrane, is limited by macrosolute redistribution.
The retained materials are large molecules with very low diffusivities so
unaided molecular diffusion would result in a very low redistribution rate.
Designers of ultrafiltration equipment have found that controlling the
ultrafiltration rate is primarily a matter of controlling mass transfer in the
channel next to the membrane surface. The variables determining mass transfer
rates in a channel bounded by a membrane are velocity, diffusivity, viscosity,
density, and channel height. Increasing mass transfer in the membrane channel is
possible by varying any of these, or by varying temperature, which affects most
of the variables. In practice, however, viscosity, pH. density and permissible
temperature are fixed by the properties of the feed stream. The only variables
available to the designer are channel geometry, velocity, and sometimes, within
limits, temperature.
Early in the evolution of ultrafiltration equipment, there was an emphasis

on clever devices for disrupting the polarization layer near the membrane.
Amicon worked on devices to utilize the enhanced mass transfer at the entrance
region to laminar flow channels. Abcor (presently Koch Membrane Systems)
developed various devices for disrupting the flow and increasing turbulence, in an
attempt to transfer momentum into the boundary layer. None of these devices
was successful, and the search for good design soon became a contest of
geometry.
Today, commercial ultrafiltration modules use fluid mechanics to control

polarization. Most designs are for turbulent flow, operating at Reynolds numbers
of up to IO’. Some spiral-wound modules operate in laminar flow, although this
is unusual.
7. I .3 Plugging
Plugging of the flow channels by solids present in the feed is another

major design concern. Apple juice is a good example of a solid-containing feed
solution. The juice coming from a press may contain pomace, which has a fairly
high fibre content. As the juice is concentrated, the retained solids approach a
level at which they do not flow. Since high juice yield is economically vital, the
system is operated right up to the onset of plugging. Tubular membranes
captured important early markets because their design, properly executed, was
resistant to fibers, dirt, and debris. Tubes are also very easy to clean, which
made them a good choice for edible product applications. Even though tubes are
inherently expensive to build and operate, they still dominate those applications in
which plugging is a serious concern.
Capillary membranes are inherently inferior to tubes but may be operated

safely with feed in the bore and periodic reversal of the permeate flow under
back pressure.
7.1.4 Fouling
Fouling is the term used to describe the loss of throughput of a membrane

device because it has become chemically or physically changed by the process
fluid.g Fouling is different from concentration polarization, but the effects on
output are additive. Fouling results in a loss of flux that is irreversible under
normal process conditions. Normally a decrease in flux will result from an
increase in concentration or viscosity, or a decrease in fluid velocity. This
decline is reversible by restoring concentration, velocity, etc. to prior values. In
fouling, however, restoration to prior conditions will not restore the flux.
There are several fouling effects. Prompt fouling is an adsorption

phenomenon caused by some component in the feed, most commonly protein,
adsorbing onto the surface and partially obstructing the passages through the
membrane. This effect occurs in the first seconds of an ultrafiltration operation
and may sometimes even result from dipping a membrane into a process fluid
without forcing material through it. Although this type of fouling raises the
retention, it also lowers the flux through the membrane and is a negative effect.
Prompt fouling is very co&nmon, although not always recognized as such, and most
membranes are characterized after it has occurred.rO In fact, many membranes
are not commercially useful until prompt fouling has taken place.
Cumulative fouling is the slow degradation of membrane flux during a

prolonged period of operation. It can reduce the flux to half its original value in
minutes or in months. It may be caused by minute concentrations of a poison,
but is commonly the result of the slow deposition of some material in the feed
stream onto the membrane. Usually, the deposition is followed by a
rearrangement into a more stable layer that is harder to remove. This effect is
related to prompt fouling, because the prompt fouling layer provides the foothold
for a subsequent accumulation of foulant.
Fouled membranes are frequently restorable to their prior condition by

cleaning. The vast preponderance of membranes used commercially foul, and are
cleaned to restore their output. Membrane producers devote considerable effort
to finding safe, effective, and economical means of returning the membranes to
full productivity. It is usually true, however, that the cleaning agents
themselves slowly damage membranes, making them more susceptible to future
fouling. There are totally benign cleaning agents for some types of foulant. but
the more common types, such as protein-based foulants, usually require aggressive
cleaning agents to keep the cleaning cycle short. It is said that the cleaning
requirements for membranes are the single major determinant of membrane life.”
Ultrafiltration 369
Some fouling is totally irreversible. An occasional culprit is a material

present in the feed at low concentration, especially if it is a sparingly soluble
material at or near its saturation concentration, which has affinity for the
membrane. Such a material can slowly sorb in the membrane and, in the worst
case, change the membrane’s structure irreversibly. Antifoams are the usual
culprit. With the chemically robust membranes in use today, this effect is very
unusual, unless a membrane feed has been contaminated with a damaging solvent.
The rate of fouling is influenced by system design and operation. Figure 7-3
shows the pressure-flux output curve for a normal ultrafiltration process.
Operating in the high-pressure region at the right will produce fouling more
quickly than operation around the knee of the curve.” A high-pressure, low-
flow ultrafiltration regime such as occurs in dead-end filtration represents a
worst case operating condition. Experience indicates that thick, dense, boundary
layers promote fouling.1s~14 Unfortunately, equipment that operates membranes
only in the low-fouling region, at or below the knee of the operating curve, is
very hard to design.
7.1.5 Flux Enhancement
The flux in ultrafiltration systems is, with rare exceptions, determined by

mass transfer at the membrane surface. As has been shown, the rate of passage
through an ultrafiltration membrane is determined by the rate at which retentate
can be removed from the membrane. The cleaning of fouled membranes relies on
the same principle. Foulants are removed from the membrane by a combination of
fluid flow past the membrane and chemical action to emulsify the foreign deposit
or to attack it so as to reduce its size or change its chemical form and lower its
adhesion to the membrane. Both these critical steps mandate a design which
emphasizes mass transfer at the membrane surface.
Early designs attempted to achieve mass transfer without excessive pressure

drop or energy loss. Attempts to design devices that run in laminar flow and
take advantage of enhanced mass transfer due to hydrodynamic entrance effects
were unsuccessful. Turbulence promoters were tried in a variety of membrane
devices operating in turbulent flow. The idea was to transfer some of the
momentum from the turbulent core into the boundary layer to enhance mass
transfer, thus increasing flux. None of these designs was found to be practical.
Operating at high fluid velocities in the feed channel is presently the only way of
enhancing mass transfer.
Spiral-wound modules may prove to be the first real success in improving

the flux in a steady-state device by a passive technique. Optimization of the
spacer net separating the membrane surfaces in spirals has been a serious,
ongoing effort and has led to considerable improvement in performance.
Another promising design is the capillary module developed by Mitsubishi

Rayon Engineering under the Japanese Aqua Renaissance program. These modules
consist of capillary bundles that are fixed at one end only and operate with a
shell-side feed. If it is demonstrated that the flux enhancement is due to
“flapping,” of the capillary bundles, this would represent a second successful
technique for passive mass transfer enhancement.
Flux enhancement by dynamic techniques was first commercialized by

Westinghouse in the early 1970s. Their design was a multitubular “sand log,” in
which the membranes were cast inside 25mm-diameter channels. They used a
gravity head on the permeate side of the membrane, and shut the feed pumps off
every minute or so for 5 seconds. As the permeate returned backwards through
the membrane, it lifted the polarized layer of retentate from the membrane.
Mass transfer was improved more than enough to justify the exercise. The
technique fell into disuse when Westinghouse exited the market. Permeate
backwashing is still used by Romicon (capillaries), Memtec (air backwash) and
others, although it is mostly a way to control fouling by removing nascent
foulants before they consolidate.
7.1.6 Module Designs
Early in the development of industrial ultrafiltration there was a

proliferation of module designs. Almost all of these designs are still being
marketed, although over half of the current sales are spiral-wound modules.
The most successfulearly design was the 25-mm tubular membrane (Abcor).
The module was a 2.8-m long porous pipe, with a membrane cast inside. It had
the great virtue of simplicity and reliability. The design was made practical by
connecting successive membranes together with minimum pressure loss U-bends,
carefully avoiding any changes in diameter in the flow channel. When this
arrangement was properly executed, 90% of the total process pressure drop took
place adjacent to active membrane. The absence of stagnation points proved
highly desirable when processing fluids containing fibrous contaminants. The
large tube design suffered from its large physical size, and from very low
conversion per pass, leading inevitably to high energy costs. Early models were
prone to rupture, but reliability now is almost absolute.
Other early designs which proved viable were plate-and-frame, parallel-plate

devices, and capillaries. Capillaries have been prone to problems because of
difficulties in incorporating them into modules. A series of potting-related
problems have made them difficult to produce reliably. However, for some
applications. such as ultrapure water, capillaries continue to hold a significant
share of the market.
Parallel-plate and plate-and-frame modules encountered many design

problems, principally improper flow distribution and sensitivity to fibrous debris
that led to cracks and module failures. These designs were eventually refined and
have achieved some commercial success.
The spiral-wound module dominates the ultrafiltration market. This design

did not achieve much early success due to plugging, rapid fouling and difficulty in
cleaning. New materials, better designs and manufacturing quality control
overcame all of these problems, resulting in the present favored status of spiral-
wound modules.
Ultrafiltration 371
7.1.7 Design Trends
Two factors have pushed ultrafiltration equipment manufacturers towards

compact, energy-efficient designs. In Europe and Japan, and, to a much lesser
extent, in the U.S., buyer reaction to the oil crisis in the late 1970s made it
clear that high energy-consuming ultrafiltration designs would be displaced as
lower-energy alternatives became available. The second factor was the growing
realization that ultrafilters, in growing larger, were becoming significant
consumers of valuable floorspace. Higher conversion per pass was the best
answer to both concerns.
One approach to high-conversion design is to use more membrane area at

lower energy intensity. This necessitates compact module designs, such as
capillaries and spiral-wound modules. Making more compact modules required
significant improvements in membranes and in module fabrication techniques. In
the early 1970s most membranes were made from cellulose acetate. This
material was not suitable for capillaries, and was only marginally suitable for
spiral-wound modules, because the mild cleaning agents appropriate for cellulosic
membranes could not clean the modules properly.
The search for new polymer materials led to membranes being prepared from
every available commercial polymer that exhibited any hope for success.
Commercial replacements for the early cellulosics included an acrylonitrile-vinyl
chloride copolymer, polyacrylonitrile, polyvinylidene fluoride, polyethersulfone,
and polysulfone, which has become the most important ultrafiltration membrane
polymer.
Polysulfone can be made into a variety of ultrafiltration membranes with

very retentive or fairly open properties. It is chemically tough enough to resist
aggressive cleaning, and it can be made into high-quality flat membrane with good
dimensional stability, suitable for winding into spiral-wound membrane modules. It
is one of the few polymers suitable for capillary devices, and is also used widely
in plate-and-frame and parallel-plate modules.
The effort to make membranes out of better polymers also resulted in the
production of a highly reliable capillary module by Asahi Kasei in Japan, using a
proprietary acrylic polymer. Market share for this membrane module in the
traditionally difficult electrocoat paint market rose dramatically in the early
1980s.
Since the beginning of the modern membrane era, significant effort has
been applied towards making inorganic membranes. Union Carbide developed a
dynamic membrane from hydrous zirconium oxide inside a porous carbon tube, but
this technology had a weak market acceptance. At present a number of ceramic
devices are available, but their primary use is in microfiltration. Some of the
tighter membranes are useful in the upper end of the ultrafiltration range.
The big potential advantage of ceramic membrane materials is their high

temperature capability. They are expensive, however, and there are no developed
markets where the need for temperature resistance warrants the extra expense.
Additionally, they are still too new to have demonstrated the long life needed to
justify their higher initial cost. At present. ceramic membranes are made in
multi-tube monoliths, which are inherently compact. They have not yet been
packaged into compact, high-conversion modules. There is a high operating
energy cost associated with ceramic membranes, as the membrane purchase cost
requires operation at high fluxes, to keep the conversion rate high and the
membrane area requirement low. One start-up company claims to have developed
a low-cost, high-temperature membrane, but its primary use is in microfiltration.’
7.2 APPLICATIONS
Ultrafiltration has several important applications, particularly in the food

industries, which are discussed below.
7.2.1 Recovery of Electrocoat Paint

In the coating of industrial metal, an efficient and corrosion resistant way
of applying the prime coat (for automobiles) or a one-coat finish (for appliances,
coat hangers, etc.) is the electrophoretic deposition of a colloidal paint from an
aqueous bath. This process is illustrated in Figure 7-5. After suitable
pretreatment, the metal object is immersed in a paint tank, and the paint is
plated on. During the process, the paint film undergoes electroendosmosis, and it
emerges from the tank already robust, although still wet. The “dragout” paint,
the droplets adhering to the fresh wet film, is washed off and recovered.
Ultrafilters are used throughout the world for this application. In the
automobile industry, the savings in paint is around $4 per vehicle. A large paint
recovery installation typically contains 150 m2 of membrane area, and produces
3 m3/hr of permeate.
7.2.2 Fractionation of Whey

In the production of cheese and casein, about 90% of the volume of the
milk fed to the process ends up as whey. The quantity of whey produced in the
United States, a tiny fraction of world production, is about 25 million cubic
meters per annum. About half this amount is processed in one way or another.
Ultrafiltration is used to produce high-value products, which can range from 35%
protein powder (a skim-milk replacement) to 80%+ protein products, used as high-
value high-functionality food ingredients. A large dairy ultrafilter operating on
whey will contain 1,800 m2 of membrane, and have a whey intake of 1,000 ma per
day. Figure 7-6 is a flowsheet of the whey treatment process.
7.2.3 Concentration of Textile Sizing

In the knitting and weaving of textiles, a sizing material is commonly
applied to lubricate the threads and protect them from abrasion. The sizing
material is removed before the fabric is dyed. Ultrafiltration provides an
economical means to recover and reuse the sizing solution. Recovery of the
sizing encourages the use of more expensive but more effective sizing agents,
such as polyvinyl alcohol. Since desizing baths operate at high temperatures, it is
necessary for the ultrafilter to withstand constant operation at 8SC. Some
inorganic membranes have been used in this application, but polymeric membranes
operating at low flux are more economical, especially in their much lower power
consumption. A large recovery plant has a membrane area of 10,000 m2, and a
feed rate of 60 ma per hour.
Ultrafiltration 373
Chromate Phosphate
Paint
Figure 7-5. Ultrafiltration for the recovery of electrocoat paint in the

automotive industry.
Milk 10.2 ions

$2755
1
I I -
1 ton Cheese
swoonetotcost to
convert milk to cheese
I
Whey 9.20 tons
5.5Y. total solids
Whey protein concentration

Ultratilter 0.14 tons
operating cost: 516 $183 net of drying
cost and UF cost
Pl
Evaporation
and separation
Mother liquor Lactose 0.2 ton

0.2 ton 517.51 net of drying cost
SO and UF cost
Figure 7-6. A flowsheet for the ultrafiltration of whey. The prices are as of
November 1989. The cost basis is 1 ton of cheese (2,000 lb).
Ultrafiltration 375
7.2.4 Recovery of Oily Wastewater
In the metal working industry, lubricants and coolants are used in numerous
operations, such as metal cutting, rolling and drawing. Lubricants and coolants
usually take the form of oil-in-water emulsions. Parts that have been cooled or
lubricated are generally washed, creating a dilute oily emulsion. The quantity of
spent, dilute emulsion just from washing newly formed aluminum cans is over 8
m3 per hour per can line. Ultrafiltration is the principal technology employed to
concentrate this waste into a stream of water suitable for a municipal sewer, and
an oily concentrate rich enough to support combustion, or for oil recovery.
7.2.5 Concentration of Gelatin
Gelatin coming from the extractor is a dilute solution of hydrolyzed

collagen. It can be dried economically once the total solids concentration in the
stream reaches about 30%. Ultrafiltration or evaporation may be used to remove
the bulk of the water before drying. Because the membrane passes some of the
salts along with the water, ultrafiltration reduces the extent of treatment in the
subsequent ion-exchange step. Ultrafiltration can achieve a 30% solids
concentration, but is usually used only to remove SO-90% of the water in the
feed, at a much reduced energy cost compared with evaporation.
7.2.6 Cheese Production
An emerging process for the production of cheese uses ultrafiltration before

the cheese production process, rather than behind it as in the whey application.
The process is proven for soft cheeses, and is in commercial operation for cheese
base, an intermediate in the production of several mass-consumption cheese
products. CSIRO. in Australia. has developed an ultrafiltration process applicable
to cheddar cheese. As a conservative average, the use of ultrafiltration reduces
the milk required to make cheese by 6%. This reduction is accomplished largely
by capturing soluble proteins in the cheese curd that would normally pass into the
whey.
7.2.7 Juice
A significant fraction of all clarified apple juice produced in North America

is passed through an ultrafiltration membrane. The membrane process is rapidly
displacing rotary vacuum filtration because of higher yield, better and more
reliable quality and ease of operation.
No reliable estimates have been found for the energy savings resulting from
a higher yield of juice by the ultrafiltration process. However, one 530-m3 plant,
operating seasonally, is reported to save about 400 tons/year of diatomaceous
earth.
Table 7-I summarizes the principal applications of ultrafiltration.

Table 7- 1. Principal Applications of Ultrafiltration
Leading Membrane Leading Membrane

. .
Aoolication Customer onfwtlon Suooliers
Ultrafiltration of auto, appliance tubular Koch
electropaint wastewater firms, etc. spiral Rhone-Poulenc
for paint recovery plate-&-frame Asahi
capillary Romicon
Nitto
Recovery of protein dairies spiral Koch

from cheese whey plate & frame DDS
Ultrafiltration of milk dairies spiral Koch

to increase cheese yield plate-&-frame DDS
Concentration of oily metal cutting and tubular Koch

emulsions for forming, can makers, plate-&-frame Rhone-Poulenc
pollution abatement industrial laundries capillary Romicon
High purity water chip producers capillary Asahi

spiral Nitto
Osmonics
Romicon
Pyrogen removal pharmaceutical and capillary Asahi

glucose producers Romicon
Size recovery textile firms spiral Koch
Enzyme recovery enzyme producers, plate-&-frame DDS

biotech spiral Koch
capillary Romicon
Gelatin concentration food firms spiral Koch
Juice fruit processors tubes Koch

clarification capillary Romicon
ceramics Alcoa
Pharmaceutical drug firms plate & frame DDS

spirals Koch
Rhone-Poulenc
Latex chemical firms tubes Koch

(PVC, SBR) Kalle
Gray water hotels, offices, plate-&-frame Rhone-Poulenc

(domestic waste) apartment blocks tubes,sheet,disk Nitto
Laboratory small purchases capillary Amicon

Ultrafiltration 377
7.3 ENERGY BASICS
Ultrafiltration is a pressure-driven process. Under ideal conditions, the

pressure need only exceed the osmotic pressure across the membrane, at most a
few psi. For example, if the osmotic pressure is 5 psi, the process could operate
at energy levels as low as 0.01 kWh/m3 of permeate. At this energy level, the
process would be very slow, and most membranes would not operate at high
retention.
The poor retention at low pressure is due to inevitable leaks in the

membrane. The flow of solvent and microsolutes is pressure-dependent, and small
at low pressures. The diffusion of macrosolutes through the larger openings,
however, is fairly independent of operating pressure. At low pressures the two
flows are equivalent and retention is low. Under normal operating conditions,
i.e., higher pressure, high flow, high flux, and thus high energy consumption,
leaks due to membrane imperfection are diluted to insignificance. Ideal energy is
thus an elusive concept. It is certainly low, but operation near the ideal energy
is expensive, requiring large membrane area and near perfect membranes to
achieve a good separation.
Early ultrafilters, of the tubular or plate-and-frame types, consumed over 25

kWh/m3 of permeate produced. Major improvements in all facets of equipment
and membrane design have lowered this figure to 5 kWh/m’ or less. Recent
trends indicate that further improvements are possible. The Japanese Aqua
Renaissance ‘90 project has a design goal of 0.3 kWh/m’ for the ultrafiltration
of a digester overflow stream, a target that is close to being met.”
Ultrafilters are operated at high fluxes to achieve adequate retention and

the major use for energy is to reduce concentration polarization. The energy
required to push the liquid through the membrane is trivial. The energy required
to deliver the fluid to the membrane surface is significant. The centrifugal pumps
universally employed run at 77% to 60% hydraulic efficiency. Sanitary pumps
require a greater sacrifice in efficiency in order to maintain hygienic design.
Piping and valves. manifolding in and out of the membrane array and cleaning
operations contribute to further energy losses.
The overall efficiency from electricity to fluid energy in the membrane

channel ranges from 45% for hygienic plants to 64% for the very best designed
industrial plants requiring infrequent cleaning. The energy analysis for a typical
electrocoat paint recovery plant is presented in Table 7-2. The plant operates at
an overall energy efficiency of 57%. Dairy units operate with higher cleaning
down time and lower efficiency sanitary pumps, resulting in a lower overall
energy efficiency. The weighted average energy efficiency of ultrafiltration
systems in operation today is closer to 45% energy efficiency than to 64%, due to
the large number of food and dairy installations.
Table 7-2. Power Losses in a Typical Ultrafiltration Unit
Power Loss
Item (W/M?
Power delivered to the membrane 50 (58%)
Pump losses 25 (29%)
Switchgear and motor losses 5 (6%)
Piping, valve and header losses 5 (6%)
Losses during cleaning 1 (1%)
Total 86
7.3.1 Direct Energy Use VS. Competing Processes
There are no competing processes for many ultrafiltration applications, such

as the recovery of electrocoat paint. For processes such as the separation of
proteins from microsolutes, competing processes include chromatography and
precipitation. However, chromatographic separation produces a diluted
intermediate stream that must be concentrated by evaporation. Precipitation also
requires an evaporation step to recover the precipitant. In this case,
ultrafiltration is the most energy-efficient process as both of the competing
processes require an energy-intensive evaporation step.
Gelatin is concentrated in multiple effect evaporators at an energy

consumption of 73 kWh/ms of water removed. In contrast, recovery by membrane
ultrafiltration requires only about 7 kWh/ms of water removed. In the recovery
of textile sizing, ultrafiltration typically requires only 10% of the energy used by
evaporation.
Ultrafiltration can also deliver significant energy savings by reducing the

level of downstream waste treatment required, as in the case of electrocoat paint,
or by producing a waste with a fuel value, as in the case of oily wastewater
treatment.
7.3.2 Indirect Energy Savings
The substitution of ultrafiltration for conventional unit operations can lead

to ancillary energy savings, by increasing intermediate recovery and reducing the
level of processing per volume of product. A recent DOE-sponsored study has
illustrated this effect in use of ultrafiltration in cheese manufacture.17
Ultrafiltration 379
The cumulative value of thermal energy required for animal feed, on farm
milking and refrigeration, and farm-to-plant transportation is 53 MBtu/ton of
cheese produced. The ultrafiltration of milk before cheese production increases
the yield of cheese by 6%. Since a ton of cheese can now be made with 6% less
milk, this translates to an energy savings of 3.2 MBtu/ton. The energy required
to operate the ultrafilter is 0.6 MBtu/ton, resulting in a net energy savings of
2.6 MBtu/ton of cheese produced. The annual U.S. production of cheese is 3.4
million tons, so the savings from the increased efficiency in the use of milk
could amount to 0.01 quads annually.
When cheese is made after milk has been ultrafiltered, there is a reduction
in the amount of whey produced. As a first approximation, assume that the
amount of whey produced is a function of the milk input, not the cheese output,
Since the treatment of whey consumes energy, there is the energy saving
resulting from the whey not produced. But if whey is available in excess, and
the evidence available indicates that only about half the whey produced is
manufactured into products, whey not made could be deducted from the whey not
now treated. In that case, the energy savings would be minimal, since the
energy cost of feeding or spreading whey is fairly low. If the whey not produced
results in whey products not being manufactured, the analysis becomes too
complicated for this study.
Whey is manufactured into many products of various values, but one that
bears on the energy picture is “35% whey protein concentrate,” a product sold as
a skim-milk replacement. While the picture is very complicated, a ton of cheese
produces, as an ultrafiltration byproduct, roughly a ton of a fluid with properties
similar to skim milk. It is always dried, and sold in the market as a replacement
for non-fat dry milk. Since the energy to dry milk and whey protein concentrate
is similar, that part of the energy picture will be ignored as not very different.
The energy saving produced by ultrafiltration is really the energy not needed to
produce the ton of skim milk in the first place. The power required to produce
the skim milk replacement by ultrafiltration is about 20 kWh/m3 of product.
7.4 ECONOMICS
7.4. I Typical Equipment Costs
The dairy industry offers the largest potential energy savings of any
ultrafiltration application. This section presents the economics for a large whey
ultrafilter. The general cost structure for whey and milk applications is similar.
This example considers a plant, producing 35% whey protein concentrate,

containing 1,200 m2 of membrane area and operating on a feed of 1,000 m3/day.
The feed stream contains 6 g/L true protein, 2 g/L non-nitrogen protein, 48 g/L
lactose, 5.5 g/L ash and 0.5 g/L fat. The average flux is about 35 L/m2h (21
gfd). The plant typically operates for less than 20 h/day. The installed cost is
approximately $600.000, or S500/m2 of membrane area, exclusive of buildings.
The equipment costs for the plant are presented in Table 7-3. This analysis
is based on an installed cost of $600,000, and a skid-mounted ex-factory cost of
$500,000.
Table 7-3. Equipment Costs for a 1,000 ms/day Whey Ultrafiltration Plant
Item Cost Fraction Comments
Pumps 30% Assumes a power density of 50

watts/m2 membrane at the
surface of the membrane
Replaceable Membrane Elements 20%
Housings for Membranes 10% Assumes spiral-wound mem-

branes with re-usable hous-
ings
Piping and Framework 20% Assumes stainless steel pipe

and painted steel framework
Controls 15% Automatic cleaning cycle pro-

grammable controller
Other 5% Heat exchanger, cleaning

tanks, etc.
wital c~sfs’ The residual capital cost of the ultrafilter is $480,000, which is the
actual cost, s600,000, less the cost of the replaceable membranes, f120,OOO. The
capital charges are based on residual net capital cost, plus any allocation for
building charges, which are not considered in this analysis. The residual capital
cost is spread over the capacity of the plant in terms of dry product recovered.
We assume that the plant operates 275 days/year at 20 h/day with an average
productivity of 35 L/m2h. If the desired product is present at a level of 6 g/L in
the feed stream, then the plant produces 5,000 tons/year of dry product. At a
33% capital charge rate, the capital expenses attributable to the product are
$160,00O/year, or $32/tori (SO.O32/kg, SO.OIS/lb) of total product.
If the product of interest were present in the feed at 0.6 g/L, the flux
would rise by about a factor of 3, so the plant would need to be about 3.3 times
as large, which might increase the total capital cost and capital charges per unit
of production by a factor of 2.5.
Enerav co% For a plant designed for 50 watts/ms, and an overall efficiency of
45%, with an average flux of 35 L/m2hr, the energy consumption is 3.2 kWh/ms.
At a cost of SO.O’l/kWh, the energy cost to produce permeate is $0.22/mS. For a
35% whey concentrate, the energy cost is SO.O12/kg ($O.O06/lb) of solids.
Ultrafiltration 381
Membrane reolacement Membrane life is largely a function of cleaning frequency.

Daily cleaning is the custom in the food industry and necessitates annual
membrane replacement at a price of SlOO/m 2. For the assumed operating schedule
of 300 days/year, the membrane then has a useful life of 6,000 hours. If its
average flux is 35 L/m2h, it will produce 4,200 kg/m2 of dry product over its
lifetime. This works out to a membrane replacement cost SO.O24/kg ($0.01 l/lb) of
dry product.
Labor: Membrane plants operate with a high level of automation, and they often
run unattended. A conservative estimate of the labor costs would be
$60,00O/year, or SO.OIZ/kg ($O.OOS/Ib) on the same basis as above.
Other CostS: Cleaning chemicals, heat, and other costs are estimated to add
$O.OZS,‘kg ($0.01 I/lb).
The operating costs for the plant are presented in Table 7-4, listed in order
of imoortance.
Table 7-d. Operating Costs for an Ultrafiltration Unit Producing a 35% Whey
Concentrate
Item Cost/kg % of Total
Capital Charges @ 33% $0.03’_ 31
Cleaning & miscellaneous 0.025 24
Membrane Replacement 0.024 23
Energy 0.01: I1
Labor 0.012 II
Tot31 SO.105 100

(SO.O48/lb)
7.4.2 Downstream Costs
An ultrafilter is a separation device that produces at least two products.

Usually. one is the product desired, and the other is the product one would prefer
to forget about.
Costs of disposal can be so significant that, in some cases, they may

dominate the economics of the ultrafiltration installation. An extreme, but
illustrative example is the concentration by PVC latex by ultrafiltration. The
latex is typically 40% solids by volume and contains 10” PVC particles/m’. A
membrane retaining 99.99% of this latex produces permeate containing IO’*
particles/ms, which is still a turbid fluid. The cost of treating this permeate to
remove the remaining particles is very high. Therefore ultrafiltration is not
economically feasible if the membrane is unable to retain enough particles to
produce a permeate well below the turbidity threshold.
7.4.3 Product Recovery
Many potential applications for ultrafiltration are heavily influenced by

product recovery. Getting all of the valuable product out of the feed stream and
into useful or salable form is often critical. In most applications, a more
retentive membrane will displace a less retentive rival, even at a higher price.
However. membrane retention is a function of time and increases with

fouling in most cases, especially for streams containing proteins. Most commercial
membranes rely on fouling to get the very high retention advertised. During the
first 30 minutes of batch ultrafiltration, the retention may be much lower than
the steady-state value. This is also a period when the membrane produces its
highest flux, because it is not yet fouled. Since all edible product operations run
in batches that rarely exceed 20 hours between cleaning, a small but significant
fraction of the operation occurs during the initial non-steady-state period. Most
processes requiring a product to feed concentration ratio greater than I.5 use
multiple membrane stages in series. All of the stages begin the operation filled
with unconcentrated feed and the initial unsteady period increases with the
number of stages used.
During the initial unsteady state. losses are difficult to assess, but are
thought to be higher than normal. Most membrane equipment suppliers base the
membrane specifications and guarantees only on the steady-state operating
characteristics. However, during the initial unsteady period, which may last 3s
long as I20 minutes, several percent of the entire day’s charge of desired product
may be lost.
Product losses during cleaning are also important. An edible products unit
is shut down for cleaning after a 20-hour batch operation. Any product in the
membrane modules is flushed out with water. Although the process is relatrvely
efficient, there is some product dilution and product losses are inevitable. Losses
also occur due to product that has fouled the membrane, and in product otherwise
retained in the apparatus. These losses, which may easily total several percent of
the product in the batch, are very hard to measure and are known only in the
most general terms even after decades of operating experience.
Industrial users sometimes operate their ultrafilters for very long periods
between cleanings or down-times. In fact, to avoid startup losses and instability
problems. some users even leave the ultrafilter operating on a recycled permeate
stream when the device is not needed onstream. The recovery of polyvinyl
alcohol in the textile industry is an example of an application where long runs
and high product recovery with low losses is the rule. Virtually all of the
Ultrafiltration 383
product losses are through the membrane through leaks, or in upstream

pretreatment operations.
In some cases, the economics are determined by the quality of the permeate
stream, as in the recovery of eiectrocoat paint. In normal operation, the
permeate and retentate streams are recycled to the process and a small loss of
paint into the permeate stream does not change the process economics. However,
when the ultrafilter is used as a “kidney” (some of the permeate is removed to
eliminate electrolytes or other impurities from the paint tank) minor leaks have a
significant impact on the process economics due to the need for additional waste
treatment.
Another application where the permeate dictates the process economics is in

the clarification of juice. The recovery of juice is more important than the
retention of macromolecules. The typical process operates by first concentrating
the retentate as much as possible without plugging the modules. The
concentrated retentate is then washed to recover as much of the sugar and
essence as possible. All of this water of dilution must be subsequently removed
by evaporation, resulting in a higher energy cost. The ultrafilter is rated by the
quality and yield of the undiluted juice, and the amount of recovery of diluted
juice that is required.
7.4.4 Selectivity
Ultrafiltration membranes discriminate between macrosolutes and

microsolutes. This selectivity is critical to their usefulness. When high-purity
products are needed, it is often necessary to flush the microsolute through the
membrane, a process analogous to the rinsing of juice residues.
For example, whey is fed to an ultrafilter with a protein-to-microsolute ratio

of about 1:lO. If a product with a ratio of 1:1 is needed, it is a simple matter to
concentrate the feed tenfold. The protein concentration will have increased
tenfold, but in an ideal case, the microsolute concentration will remain unchanged,
so the result is achieved. Suppose however the desired result is a protein-to-
microsolute ratio of 1O:l. A IOO-fold concentration of the feed could produce the
result, in principle. Whey, however, has a protein content of roughly 0.5%.
Increasing this IOO-fold would produce a product with 50% protein, far in excess
of the ability of any ultrafilter. In fact, 20% protein is the upper limit for
practical operation.
The industry practice is to lower the concentration of microsolutes by adding

water to the feed. This operation is called diafiltration and is expensive and
inconvenient. The cost of diafiltration can be minimized by using high selectivity
membranes, although a highly polarized protein layer forms on the membrane
surface due to fouling and some microsolute retention is inevitable.
The principal suppliers of ultrafiltration equipment are listed in Table 7-5.
Table 7-5. Major Suppliers of Ultrafiltration Membranes and Equipment

(1989 UF Sales over S3,000,000)
Estimated
Principal UF Sales
Company and Scope Location Major Products (SM)
Koch Membrane Systems Wilmington. MA 25 mm & 13 mm tubes 32

spirals
De Danske Sukker- Nakskov. Denmark plate-&-frame 15

fabriker (DDS)is2
Techsep Paris, France plate-&-frame, 15

(Rhone Poulenc)1*2 carbon tube
Romicon Woburn, MA capillary II

(Rohm & Haas)lv2
Asahi Kasei’ Tokyo, Japan capillary 6
Nitto Denko’ Osaka, Japan 12 mm tubes, capillary, 4

spiral
Minnetonka, MN spiral 9
Amicon, Dorr- Lexington, MA capillary, leaf 25

Oliver (Grace)‘12
DSI’ Escondido, CA spiral __
Allied-Signal San Diego, CA spiral __

(Fluid Systems)’
Daicel Osaka, Japan 3
Alcoa Separations Warrendale. PA ceramic tubes __

Tech (Ceraver)’
’ Membranes
2 Equipment
Ultrafiltration 385
The major source of innovation in ultrafiltration has been the supplier

industry, with minor contributions from users and academic and governmental
researchers.
7.6.1 Suppliers
The origins of the modern ultrafiltration industry grew out of the work on
reverse osmosis membranes being done at MIT in the early 1960s for the Office of
Saline Water. The formation of the Amicon Corporation, and the subsequent
decline in government funding for membrane-related work, effectively transferred
the locus of innovative activity to industry, where it has remained.
The early reduction to practice for ultrafiltration was done at Amicon

Corporation, with financial sponsorship from Dorr-Oliver. Both these companies
benefitted from their pioneering work. (The membrane portions of both of these
firms are now owned by W. R. Grace.) Shortly thereafter, Abcor (now Koch
Membrane Systems) entered the industry, quickly becoming the major competitor
of Dorr-Oliver. Many other firms began ultrafiltration efforts in the late 1060%
but most merged or exited the business in the following five years. Ce Danske
Sukkerfabriker (DDS), a major European firm, joined the field shortly after Abcor.
Its membrane activity is now owned by Dow Chemical Co. Berghof and severai
other firms were also active in the early development of ultrafiltration.
Suppliers have huge incentives for innovation in the ultrafiltrarion fiela.

Not only dc they have the mosf to gain from successful innovation. but most of
the more difficult problems in ultrafiltration are amenable fo supplier or user
innovation. Membrane stability can be tested in a laboratory, but field tests are
the only reliable measure. Fouling can be modeled only with difficulty, and
modeling results are not always transmutable into profitable designs. Lfechsnicai
innovations require customer feedback and information about manufacturing
techniques, which are very difficult to see from outside the industry.
The major supplier innovators are listed in Table 7-6, with estimates of their
research and development expenses.
Table 7-6. Major Supplier Innovators
Company Budget (SOOO) Emphasis
Koch Membrane 3,000 Food and industrial processes,

membranes
Techsep (Rhone Poulenc) 3,000 Dairy, paint, general industrial,

support of outside research
DDS 2,000 Membranes, dairy processes
Osmonics 1,000 Industrial processes, membranes,

low cost equipment
Romicon 1,000 Membranes, industrial processes
Asahi Kasei 800 Membranes
Nitto Denko Ceramic membranes
Alcoa Carbon membranes
GFT Ceramic membranes
7.6.2 Users
Few users have provided any innovation in ultrafiitration membranes. U;e:s

have occasionally forced suppliers to alter their designs by maintaining vigorous
test programs, and by insisting on certain features. Volkswagen (VW) is XI
example of a firm with a single-minded insistence on energy reduction. VW
created a major test bed at Wolfsburg, and used the data from it to intelligently
and firmly guide the direction of membrane equipment development by suppliers.
Through sheer determination and an internal experimental program, VW succeeded
in forcing improvements in line with their needs. They probably deserve credit
for the emergence of spiral-wound membrane modules in the automotive paint
field. VW may be unique in its use of technical innovation by a customer to
change the offerings of the suppliers.
Other firms, such as Kraft, have had heavy influence on ultrafiltration

applications through their competent internal engineering departments. Kraft
engineers have seen new applications develop and encouraged designs that would
allow them to be implemented economically. They and other engineers have
influenced equipment design through their purchasing practices, sometimes
rewarding higher quality, usually rewarding lower cost, trying always for both.
Ultrafiltration 387
7.6.3 Universities
Historically, the academic community has been involved at the beginning of

most membrane efforts. The invention of phase-inversion membranes at UCLA,
development of early ultrafiltration membranes at MIT and the recent
developments in the use of bacterial-cell-wall membranes at the University for
Agriculture, Vienna, are all examples of watershed inventions occurring at a
university. University researchers have a poor record of follow-up work and
development efforts. The major reasons may be the difficulty of working with
real separation streams in the context of academic research and the fact that
the model streams chosen are seldom accurate representatives of actual
commercial streams. At present the major universities involved in ultrafiltration
research are Rensselaer Polytechnic Institute, Syracuse University, the University
of Twente (Enschede, Holland) and the University of New South Wales
(Kensington, NSW, Australia).
7.6.4 Government
The U.S. government has supported research on ultrafiltration, most notably

the studies of ultrafiltration in the pulp and paper and petroleum processing
industries that were funded by the DOE’s Office of Industrial Programs. There is
also considerable support for the development of inorganic and ceramic
membranes. A more comprehensive analysis is presented in the section on
government support of research.
7.65 Foreign Activities
Several foreign governments are actively sponsoring ultrafiltration research.

Japan’s major governmental effort in ultrafiltration is the Aqua Renaissance ‘90
Project, which is developing techniques for treating wastewater from a variety of
sources. Membranes fit well into plans to build new types of waste-treatment
facilities. The state of the membrane art has been advanced significantly for
several Japanese ultrafiltration membrane firms through the Aqua Renaissance
project.
In Europe, the Brite program supports a number of membrane activities, but

there are no outstanding ultrafiltration projects among them. The UK, through
its Harwell-sponsored programs, does support some ultrafiltration-related work.
In Australia. the major research center is the Centre for Membrane and
Separation Technology, at the University of New South Wales, which conducts
research into mechanisms and control of fouling.
Potential applications for ultrafiltration that may be important from the

separations and the energy-savings viewpoints are listed in Table 7-7.
Table 7-7. Future Directions for Ultrafiltration
Application Prospect for Importance Comments/Problems

Area Realization
Ultrafiltration Good 8 Potential for economic gain great.

of milk Technical problems difficult.
for modified Regulatory/consumer acceptance
products major hurdle.
Municipal sewage Fair 5 Competes with non-membrane

treatment technologies, and may not win.
Impact of success would be large
but slow. Impediments are
economic and technical.
Industrial waste Good 7 Oily emulsions established: other

treatment excluding oily wastes to follow. Economics
pulp & paper chief barrier; membranes will
cheapen. Membrane enhanced
biotreatment looks a good bet to
treat high BOD wastes.
Food Excellent 8 Many applications in broad food

areas based on the ability to
change protein and starch/sugar,
salt and water ratios. Longer term
prospects include refining of oils.
Applications will be large, and
increase with technical progress
and customer acceptance
Pulp and paper mills Good 8 Good prospects for water recycle,
as in white water loop. Prospects
for black liquor, lignin separation,
etc. are fair.
Bioprocessing Fair 8 Separation and concentration of

biologically active components from
broths, etc. Ultrafiltration will
compete with more exotic
processes, but will certainly be a
major factor as the generic
biomaterials industry grows.
Abattoirs Fair 3 Recovery of blood fractions should

eventually be a significant market.
Industry conservatism, general
inertia, and regulatory concerns
may delay for another decade.
Technically possible.
Ultrafiltration 389
Table 7-7. Future Directions for Ultrafiltration (continued)
Application Prospect for

Area Realization Importance Comments/Problems
Small scale Good 6 Water recycle on small scale, to

water reuse pass purified gray water through
toilet flush, cutting local use 40%.
Economics now favorable in high
water cost markets, and where high
infrastructure costs passed back to
builders. Likely only in new con-
struction of 500+ person buildings.
Will require regulation changes.
Drinking water Fair 4 Under-sink units now sold in Japan.

UItrafiltration is one alternative as
concern grows over trihalo-
methanes. Will face competition
from nanofiltration and probable
restrictions if water yield problem
not solved.
Petroleum processing Poor 7 Requires revolutionary performance

and cost improvements to displace
conventional petroleum processes.
Could have big role in shale.
Would likely result in significant
energy savings.
Protein harvesting Good 5 Now useful for grass proteins, may

be useful for algal/plankton
proteins. Soy meal price now high
enough to support feed
applications.
The major dairy applications have already been discussed. Although the
benefits are indirect, the potential impact on energy consumption is large, even if
the market size for the technology is small. A recent DOE-sponsored study gives
a good summary of the economics and energy-saving potential of the on-farm
ultrafiltration of milk, estimated at 0.01 quads annually.17 On-farm ultrafiltration
is not included in Table 7-7 because the negative arguments enumerated in the
study were convincing to the expert panel reviewing this area. The technology
faces many regulatory. economic and technical obstacles, in that order of
significance, and the panel concluded that this is an application of ultrafiltration
that will never be developed.
Municipal waste treatment is viewed as a promising application, with

competition from microfiltration systems. Industrial waste treatment, with the
exception of pulp and paper wastes, has a high chance of successful
implementation, especially if the Aqua-Renaissance/Dorr-Oliver technology
generates enough interest to attract an aggressive development effort.
In the food area, there are many applications that will gain acceptance
incrementally. Many of the good ideas have been around for years but the
technology has been limited by lack of industrial acceptance.
Edible oil fractionation does have significant energy potential, perhaps 0.01
quads/year, but this application has already been under development for 15 years,
and there are good alternative technologies.1’
Pulp and paper is another industry in which the adoption of membrane

technology is expected to be slow, but important, with the emphasis primarily on
pollution control and secondary emphasis on energy savings.
Ultrafiltration is certain to be a significant participant in the bioprocessing

industry, but the energy impact will not be major.
Finally, petroleum processing remains the most speculative application.

Energy savings from deasphalting would be large, but the technology is far away
from being economical even if the technical problems could be solved.
7.8 RESEARCH NEEDS
Ultrafiltration has been an industrial process for 20 years, but existing

membranes suffer from a number of shortcomings, namely fouling, limited useful
economic life, pore size distribution, chemical instability and temperature
instability.
Fouling is influenced by many factors. The most important is the chemical

nature of the exposed membrane surface, including charge, surface free energy.
and response to the environment. Other factors are the membrane texture and
rugosity, pore shape, and pore density. The solute concentration at the membrane
surface, a function of hydrodynamics and pressure, is also important. Membranes
have been made from many different materials, primarily by the phase-inversion
process. The availability of novel techniques for preparing membranes, such as
thermal inversion, oxidation, carbonization, thin-film composite formation. dynamic
formation and methods used for ceramic membranes should be systematicaily
studied to learn how the many variables affect fouling.
A membrane has come to the end of its useful life at the point where
replacing it produces a positive discounted cash flow. Membranes are replaced
most often because they are leaking valuable product, or because their
productivity has declined. Both issues are often related to fouling, and the
periodic cleaning necessary to reverse the pernicious effects of fouling on system
productivity. Cleaning agents usually damage the membrane, at least slightly.
The search for better cleaning agents is a promising research project but more
emphasis should be placed on reducing the need for them.
Ultrafiltration 391
Membranes are also required to withstand extremes of temperature and pH,

and abuse from agents intrinsic or fortuitous to the application. The life of the
membrane module is determined by the most fragile component, which is often not
the actual retaining surface. Membrane sublayer failure and membrane substrate
failure are often to blame for shortened life. An intact membrane separation
surface is useless if the substrate fails or the module develops a leak. Research
aimed at improving membrane life must look at the whole module.
Pore-size distribution is an important concept, even if the “pores” exist only

hypothetically. The goal of all ultrafiltration membrane manufacture is to have a
uniform retaining layer with a very high pore-volume fraction. Uniformity in the
retaining layer is a necessary although not sufficient condition for a clean
separation.
Chemical stability includes resistance to physical attack by solvents, pH

resistance, and very importantly, the ability to withstand cleaning and sanitizing
by aggressive chemicals. Chemical resistance to adventitious impurities is a
general but important goal. In particular, resistance to substances with high
activity, such as antifoams, is desirable even at low concentrations. Stability also
implies resistance to any phenomenon that modifies membrane properties, such as
an attack that increases rugosity.
Although thermal stability has been ranked least important of the factors
needing improvement, it is nevertheless an important property. Organic
membranes have proven their ability to perform well at 85’C for prolonged
periods in benign aqueous media. Membrane chemists believe that polymer
membranes could be fabricated that would briefly withstand 150°C. Ceramic
membranes are operable at higher temperatures. In fact, there is no theoretical
barrier to the operation of certain inorganic membranes at temperatures in excess
of ZOO’C. The membranes ought to work, and modules to contain them ought to
be achievable. However, this has not yet been done and there is insufficient
evidence to verify that dependable operation of a membrane system can be
maintained above 150°C indefinitely.
The important research needs required to endow ultrafiltration systems with

the features discussed are listed in Table 7-8.
Fouling is at the top of the list of needs. It is a universal problem, or set

of problems. Some of the experts consulted doubt that there is a universal
solution. Others think that there is background knowledge that will help all
interested investigators in solving what may be a series of related problems.
Good fundamental work is needed to establish the direction most likely to produce
meaningful. energy-saving answers.
A comparative study of how and why membranes foul should seek to connect
effects with causes. It should reveal the mechanism of fouling, from the first
sorption or attachment through the loss of permeability and change in selectivity
of the membrane. Such a study needs to go well beyond the ability of any single
industrial or academic group.
Table 7-8. Ultrafiltration research needs
Prospect for
Fouling Good 10 Fouling is ubiquitous in Ultra-

filtration. Its elimination would
boost total throughput ~30%. and
reduce capital costs 15% further by
eliminating cleaning. Better
fractionation would also result,
expanding ultrafiltration use
significantly. The goal is too
ambitious for 2015, but very
substantial progress will be made.
Membrane cost/ Excellent 9 Real membrane costs should con-

unit permeate tinue to drop as markets expand
and economy of scale is realized.
Longer life will have decreasing
impact on economics, and will not
be an issue by 2005.
Low-energy Excellent 9 Improved, lower cost membranes

desings would permit lower energy process
designs operating away from the
fully polarized regime. Results
should be better selectivity, lower
fouling. lower energy, cleaning and
maintenance, and lower capital
cost.
High-temperature/ Fair 7 Required for petroleum applica-

solvent-resistant, tions but must also have low use
inexpensive membranes cost. Volumes would be large.
Membranes Good 5 Needed for aggressive industrial

resistant to high and recycle applications such as
pH and oxidants paper. Cost a major consideration.
Ultrafiltration 393
Reduction of membrane cost is an issue that may be solved in the

marketplace. Market expansion is at last taking ultrafiltration out of purely
niche markets and should soon start showing economies of scale. Part of the
economic burden for ultrafiltration users has been membrane replacement. As
the membrane life increases due to better membrane materials and module designs,
such as the introduction of ceramics, the economics will continue to improve.
Low-energy design has been a refractory problem for the industry. In the
early stages, dependability and a robustness were much more highly prized than
low-energy designs. However, customer demand for lower energy consumption has
been the necessary motivation for manufacturing innovation. We can propose no
better mechanism for achieving low energy designs.
High temperature, solvent resistant. oxidation resistant membranes are

possible. However, they are difficult to operate economically because of the
high manufacturing cost. There is no incentive for supplier innovation as the
market size is too small to recover development costs. If a petrochemical or fue:
stream separation application should develop, then the sales volume would be
huge and the development effort would be self-sustaining.
The primary research opportunity in ultrafiltration identified by the panel is

the studs of fouling and the development of fouling-resistant membranes. I? our
view, this research opportunity fits well with the DOE mission and goals. Fouling
is an important problem because it significantly raises the energy consumption of
every ultrafiltration. It has also retarded the growth of ultrafiltration into areas
where it would reduce energy consumption, such as the dairy industry, for which
the biggest potential energy savings were identified. Fouled membranes are less
selective, which is another impediment to the use of ultrafiltration.
The study of fouling is an important, high impact problem not likely to be

solved by industry and therefore is an ideal program for DOE sponsorship. The
effort required is likely to be too big, too long-term, and unlikely to give any
competitive benefit to the industrial sponsor. On the contrary, research in this
area is likely to benefit the producers, users and the public. Without a
government-sponsored joint effort, the industry will continue to work on the
problem with a piecemeal approach unlikely to produce a general, systematic
understanding.
The second research priority is for lower cost, longer life membranes, which
is equivalent in importance to low-energy module designs. These two can be
packaged together as the development of low-cost, low-energy membranes.
.Although regarded as an unreasonable target, Japan’s Aqua-Renaissance project in
Japan shows what can be achieved. While most ultrafiltration suppliers were
contemplating the possibility and implementation of the 3 kWh/m3 membrane
module, the Japanese found that they needed a 0.3 kWh/m3 membrane module to
meet requirements of their sewage treatment project. It now appears that they
will succeed. The project was well planned and implemented, and the cost was
modest. It is very likely that this project will put the Japanese ultrafiltration
industry on an even more competitive footing with the historically dominant U.S.
ultrafiltration industry. Some DOE sponsored competition, or cooperative
competition among and between people with module design and inexpensive,
long-life membrane expertise should propel U.S. firms back into a leadership role.
Finally, the real speculative work is in high-temperature, solvent-resistant

membranes for use in hydrocarbon separations. This could be so important, yet it
is such a difficult and long-term effort, that without a visionary sponsor, it is
unlikely to take place. Membrane separation as we know it would not be in place
today were it not for the visionary effort of the Office of Saline Water.
Hydrocarbon separations by membranes will not happen by 2010 without a similar
program in the 1990s.
REFERENCES
I. Sarbolouki, M. N., “A General Diagram for Estimating Pore Size of

Ultrafiltration and Reverse Osmosis Membranes,” sea. Sci. & Tech. 17 (2)
381-6 (1982).
Fane. A. G..“Factors affecting Flux and Rejection in Ultrafiltration.” J. Sea.

Proc. Technol. 4 (1) ~~15-23 (1983).
Capanelli, G., F. Gigo, and S. Munari, “Ultrafiltration Membranes -

Characterization methods,” J. Memb. Sci. 15, pp 289-313 (1983).
4. Jonsson. G.. “Boundary Layer Phenomena During Ultrafiltration of Dextran

and Whey Protein Solutions.” Desalination 51, pp 61-77 (1984).
5. Vilker, Vincent L., et al., “The Osmotic Pressure of Concentrated Protein and
Lipoprotein Solutions and its Significance to Ultrafiltration,” J, Memb. Sci.
0, p 63-77 (1984).
6. Wijmans, J. G., et al., “Flux Limitation in Ultrafiltration: Osmotic Pressure

Model and Gel Layer Model,” J. Memb. Sci, ZQ, p 115- I24 (1984).
7. Fane. .A. G., “Ultrafiltration of Suspensions,” ,I. Memb. Sci. 20, pp 249-259
(1984).
8. Zydney, Andrew L.. and Clark K. Colton, “A Concentration Polarization

Model for the Filtrate Flux in Cross-Flow Microfiltration of Particulate
Suspensions.” Chem. Enr?. Commun.. 47, pp I-21 (1986).
9. Belfort, Georges. and Frank W. Altena, “Toward an Inductive Understanding

of Membrane Fouling,” Desalination 47, p 105-127 (1983).
IO. Zeman, Leos J., “Adsorption effects in Rejection of Macromolecules by

Ultrafiltration Membranes,” J. Memb. Sci. 15, pp 213-230 (1983).
I I. Eykamp, W., and J. Steen, “Ultrafiltration and Reverse Osmosis,” in Handbook

of Separation Process Technoloay, R. W. Rousseau (Ed.), John Wiley & Sons.
1987.
12. Fell, C. J. D., Dianne E. Wiley, and A. G. Fane, “Optimization of module

Design for Membrane Ultrafiltration,” World Congress III of Chemical
Engineering, Tokyo, 1986.
Ultrafiltration 395
13. Reihanian, H., C. R. Robertson, and A. S. Michaels, “Mechanisms of

Polarization and Fouling of Ultrafiltration Membranes by Proteins,” J. Memb,
S&_& ~~237-258 (1983).
14. Matthiasson, E., “The Role of Macromolecular Adsorption in Fouling of

Ultrafiltration Membranes,” J. Memb. Sci. 16. p 23-36 (1983).
15. Goldsmith, R. L., personal communication, 1990.
16. Kimura, Shoji, “Japan’s Aqua Renaissance Project - Ultrafiltraton Energy,”

AIChE, San Francisco, November 1989.
17. Mohr, Charles M., et al., Membrane Aoolications and Research in Food
Processing, Noyes Data Corp., 1989.
8. Electrodialysis
by H. Strathmann, Fraunhofer IGB
8.1 INTRODUCTION
Electrodialysis is an electrochemical separation process in which electrically

charged membranes and an electrical potential difference are used to separate
ionic species from an aqueous solution and other uncharged components.
Electrodialysis is widely used for desalination of brackish water. In some areas
of the world it is the main process for the production of potable water. Although
of major importance, water desalination is by no means the only significant
application. Stimulated by the development of new ion-exchange membranes with
better selectivities, lower electrical resistance and improved thermal, chemical,
and mechanical properties, other uses of electrodialysis, especially in the food,
drug, and chemical process industry, have recently gained a broader interest.
Electrodialysis in the classical sense can be utilized to perform several

general types of separations, such as the separation and concentration of salts,
acids, and bases from aqueous solutions, or the separation of monovalent ions
from multiple charged components, or the separation of ionic compounds from
uncharged molecules.
Slightly modified electrodialysis is also used today to separate mixtures of

amino acids or even proteins. It is also used to produce acids and bases from the
corresponding salts by forced water dissociation in bipolar membranes.
In many applications electrodialysis is in direct competition with other

separation processes such as distillation. ion-exchange, reverse osmosis and
various chromatographic procedures. In other applications there are few economic
alternatives to electrodialysis. Total worldwide sales of electrodialytic equipment
exceeded $150 million in 1988. At least some of this equipment was intended for
use in the chemical process industry, in biotechnology and in water-pollution
control. Although the process has been known in principle for more the 50
years, large scale industrial utilization began about 15 years ago. Over the last IO
years the electrodialysis equipment-producing industry has enjoyed an annual
increase in sales of about 15%. Because of the inherent features of the process.
it seems likely that electrodialysis and related processes will continue to find new
applications in the chemical process, food and pharmaceutical industries. Market
growth may then exceed the present level.
8.2.1 The Principle of the Process and Definition of Terms
Electrodialysis is a process by which electrically charged membranes are used

to separate ions from an aqueous solution under the driving force of an electrical
potential difference.
396
Electrodialysis 397
8.2.1.1 The process principle
The principle of the process is illustrated in Figure 8-1, which shows a

schematic diagram of a typical electrodialysis cell arrangement. The arrangement
consists of a series of anion- and cation-exchange membranes arranged in an
alternating pattern between an anode and a cathode, to form individual ceils. If
an ionic solution, such as an aqueous salt solution, is pumped through these cells
and an electrical potential established between anode and cathode, the positively
charged cations in the solution migrate toward the cathode and the negatively
charged anions migrate toward the anode. The cations pass easily through the
negatively charged cation-exchange membrane but are retained by the positively
charged anion-exchange membrane. Likewise, the negatively charged anions pass
through the anion-exchange membrane and are retained by the cation-exchange
membrane. The overall result is an ion concentration increase in alternate
compartments, while the other compartments simultaneously become depleted of
ions. The depleted solution is generally referred to as the diluate and the
concentrated solution as the brine, or concentrate.
The efficiency of electrodialysis as a separation process is mainly determined

by the ion-exchange membranes used in the system. The operating costs are
dominated by the energy consumption and investment costs for a plant of a
desired capacity, which are a function of the membranes used in the process and
various design parameters such as ceil dimensions, feed flow velocity and pressure
drop of the feed solution in the cell.’
The energy required in an electrodialysis process is an additive of two terms:

one, the electrical energy to transfer the ionic components from one solution
through membranes into another solution2; and two, the energy required to pump
the solutions through the electrodialysis unit. Depending on various process
parameters, particularly the feed so!ution concentration and the current
utilization, either one of the terms may dominate. At high feed solution ion
concentration, the energy needed for the transfer of ions is generally the
dominating factor. The capital cost of an electrodialysis plant is proportional to
the membrane area required for a certain plant capacity, which is determined
mainly by the feed solution concentration and the limiting current density
discussed below.s
8.2.1.2 Limiting current density and current utilization
The limiting current density is the maximum current which may pass through
a given membrane area without giving rise to higher electrical resistance or lower
current utilization.’ The current that can pass through a membrane is limited
because the solution immediately adjacent to the diluate side of the membrane
becomes depleted of ions. If the limiting current density is exceeded, the process
efficiency is drastically diminished. Not only does the electrical resistance of the
solution increase, wasting energy, but the current splits water into H+ and OH-
ions, causing pH changes and various operational problems.s The limiting current
density is determined by the initial ion concentration in the diluate stream and by
the turbulence in this stream. Figure 8-2 shows the concentration gradient of
cations in the boundary layer at the surface of a cation-exchange membrane
during an electrodialysis desalting process.
Feed solution
Electrode rinse solution
C: Cation transfer membrane

A: Anion transfer membrane
Figure 8- 1. Schematic diagram of the electrodialysis process.

E!ectrodialysis 399
Cation flow
< <
cc 1
Cathode
2 c;
I
I
I
mz
-
-
I
I
I
c”b
Anode
I t
I ~~
1 c” ,
4 >
laminar boundary layer
Figure 8-2. Schematic diagram of the concentration profiles of the cations in

the laminary boundary layer at both surfaces of a cation-exchange
membrane during electrodialysis. C is the cation concentration, the
subscripts b and m refer to the bulk solution and the superscripts C
and d refer to concentrate and diluate.
Similar anion concentration profiles are obtained at the surface of an

anion-exchange membrane. The transport of charged particles to the anode or
cathode through a set of ion-exchange membranes leads to a decrease in
concentration of counter-ions (i.e. the ions that will pass the membrane) in the
laminar boundary layer at the membrane surface facing the diluate cell and an
increase at the surface facing the brine cell. The concentration increase on the
brine side does not have a severe impact on the cell performance, but the
decrease in the concentration of counter-ions on the diluate side increases the
electrical resistance of the solution in the boundary layer.’ The limiting current
density is the current density at which the ion concentration at the surfaces of
the cation- or/and anion-exchange membranes in the cells with the depleted
solution will approach zero. The limiting current density can be described by:*
him
=
Cf
”
D z, F c: ”
z, F k
(1)
YI,[‘-+~-
\
T+) J
[T+“- T+j
\ J
Here iii,,, is the limiting current density, C$ is the bulk solution concentration
in the cell with the depleted solution, D and z are the diffusion coefficient and
the electrochemical valence of the ions in the solution, F is the Faraday constant,
Y, is the boundary layer thickness, and TM and T are the ion transport numbers
in the membrane and the solution, respectively, and the subscripts + and - refer
to cations and anions. The constant k is the mass transfer coefficient, which
depends on hydrodynamics, flow channel geometry, and spacer design.
According to Equation (I), the limiting current density is proportional to the

ion concentration in the diluate and inversely proportional to the boundary layer
thickness at the membrane surface. The boundary layer thickness is determined
by the cell design and the feed solution flow velocity, which determine the mass
transfer coefficient.z The mass transfer coefficient can be related by the well-
known relations for the mass transfer by the Sherwood number, which again is a
function of the Schmidt and Reynolds number. Introducing the proper relations in
Equation (I) leads to an expression which describes the limiting current density as
a function of the feed flow velocity in the electrodialysis stack:
4im = Ctaub (2)
Here Ce is the concentration in the diluate cell. u is the linear flow velocity of
the solution through the electrodialysis cells and a and b are constants, the
value of which are determined by parameters such as cell and spacer geometry,
solution viscosity, and ion-transfer numbers in the membrane and the solution.
Electrodialysis 401
The constants a and b, and hence the limiting current density, are different
for different electrodialysis stack designs and must be determined experimentally.
Exceeding the limiting current density generally leads to an increase of the total
resistances and a decrease in the pH-value in the diiuate cell. The limiting
current density can, therefore, be determined by measuring the total resistance of
a cell pair and the pH-value in the diluate cell as a function of the current
density. When the pH-value is plotted vs. I/i, a sharp decrease in the pH-value
is noted when the limiting current density is exceeded. Likewise, when the total
resistance of a cell pair is plotted vs. I/i, a minimum is obtained at the limiting
current density. This is shown schematically in Figures 8-3a and b.
The limiting current density affects the membrane area necessary to achieve
a certain desalting effect and therefore also effects the investment costs for an
electrodialysis plant.s
Another key parameter determining the overall performance of the

electrodialysis process is the current utilization. The current utilization refers to
the portion of the total current that passes through an electrodialysis stack that
is actually used to transfer ions from a feed solution. The current utilization is
always less than 100%, and is affected by three factors:
I. The membranes are not strictly semipermeable, i.e. the co-ions that carry
the same charge as the membrane are not completely excluded, especially at
high feed solution concentrations.
2. Some water is generally transferred through the membrane by osmosis and/or

with the solvated ions.
3. A portion of the electrical current flows through the stack manifold by-
passing the membranes.
The total current utilization can therefore be expressed by the following

relation:’
Here f is the current utilization, n is the number of cells in the stack, and ‘I,,,,
n,.,, and n, are efficiency terms, all less than one. The subscripts m, w, s refer to
efficiency losses due to current flow through the stack manifold, to water
transfer and to incomplete membrane selectivities. The stack manifold factor, n,,
is determined by the cell design and in modern electrodialysis systems it can be
kept close to one. The water transfer factor, ‘I_,.. is determined by the water
transferred with the hydration shell of the ions. For feed solutions with low ion
concentrations it is also close to one, but for feed solutions with very high salt
concentrations it may be significantly smaller. The counter-ion leakage
factor, n,, is a membrane constant which depends strongly on the feed solution
salt concentration. This is due to a phenomenon referred to as the Donnan
equilibrium relationship. which can be expressed for a monovalent salt by the
following equation:
pH-value of Electrical
5.
0
3
the diluate resistance
t
1 Ii
7
Limiting current density Limiting current density
Figure 8-3. Schematic diagram illustrating the determination of the linking

current density by plotting: a) the pH-value of the diluate cell vs.
I/i and b) the total resistance of a cell pair vs. I/i.
Electrodialysis 403
(4)
where the overbar refers to the membrane phase, Ma is the concentration of

fixed charges per unit volume of water in the membrane and yt is the mean
activity coefficient of the salt in the designated phase.
Equation (4) indicates that the concentration of the co-ion in an

ion-exchange membrane, i.e. the ion that carries the same charge as the
membrane, is a function of the concentration of the fixed charges in the
membrane and the concentration of the co-ions in the feed solution. The
concentration in the membrane is proportional to the square of the concentration
in the external solution and inversely proportional to the fixed ion concentration
of the membrane. Since the transfer of ions through a membrane is proportional
to their concentration in the membrane interphase, the selectivity of
ion-exchange membranes is determined by their fixed ion concentration and the
concentration in the feed and/or concentrated brine solution.’ Thus, when the
ion concentration in the feed solution is much lower than the concentration of
the fixed charges in the membrane. an ion-exchange membrane is essentially
semipermeable, i.e. permeable to counter-ions only. When the ion concentration
in the feed solution is of the same order as that of the fixed charges in the
membrane, however, co-ions may enter the membrane, thereby reducing the
selectivity. The current efficiency as expressed in Equation (3) then becomes
smaller and the energy cost of electrodialysis higher. In most membrane
selectivity calculations an activity coefficient ratio of I is assumed. Aithough
activity coefficients generally approach unity in diluted solutions, this is not the
case within the membrane. Here the ratio of the activity coefficients may be
significantly higher than 1.7
8.2.2 Design Features and their Consequences
The performance of electrodialysis in a practical application is largely

determined by the properties of the ion-exchange membranes. However, system
design features such as stack construction, feed flow velocities, and type of
electrode are also of prime importance for the overall performance of the
process, as is the operating mode.8*g
8.2.2.1 The electrodialysis stack
An electrodialysis stack is a device to hold an array of membranes between

electrodes in such a way that the streams being processed are kept separated.”
A typical electrodialysis stack design is shown in Figure 8-4.
The membranes are separated by spacers which also contain manifolds to

distribute the process fluids in the different compartments. The supply ducts for
the diluate and the brine are formed by matching the holes in the spacers, the
membranes, and the electrode cells. The distance between the membrane sheets
should be as small as possible because water, even with salt in it, has a relatively
high electrical resistance. In industrial-size electrodialysis stacks, membrane
distances are typically between 0.5 and 2 mm.11*12 A spacer is placed between
the individual membrane sheets to support the membrane, to control the feed
solution flow distribution and, most importantly, to seal the cell. It is critically
important in any electrodialysis system to prevent leakage of liquid from the
compartment with the concentrated solution into the compartment with the
diluate.13
In a practical electrodialysis system, 200 to 1,000 cation- and anion-exchange

membranes are installed between two electrodes to form an electrodialysis stack
with 100 to 500 cell pairs, each cell pair having an anionic and a cationic
membrane.” Because the multicell stack contains only two electrodes, the
irreversible energy-consuming processes represented by the formation of hydrogen
and oxygen or chlorine at the electrodes do not affect the efficiency of the
operation.”
8.2.2.2 Concentration polarization and membrane fouling
Electrodialysis is affected by concentration polarization and membrane

fouling, as are all membrane separation processes. The magnitude of
concentration polarization is largely determined by the electrical current density,
by the cell and particularly spacer design, and by the flow velocities of the
diluate and brine solutions.z Ion transport through the membrane leads to a
depletion of ions in the laminar boundary layer at the membrane surfaces in the
cell containing the diluate flow stream and an increase of ions in the laminar
boundary layer at the membrane surfaces in the cell containing the brine solution.
This has an adverse effect on the separation efficiency of the process. One
unique consequence of concentration polarization in electrodialysis is its effect on
the limiting current density, which may lead to significantly larger membrane area
required for a given plant capacity.16*17 The salt concentration increases in the
boundary layer at the membrane surface facing the cell with the concentrated
solution. This may lead to precipitation of salts with low solubility and it will
also decrease the selectivity of the membrane due to the reduced Donnan
exclusion of co-ions described in Equation (3). These concentration polarization
effects can be controlled by providing proper spacer and cell construction and
feed flow velocities.
More difficult to control and more potentially damaging is membrane fouling

due to adsorption of polyelectrolytes, such as humic acids, surfactants or proteins.
These large molecules penetrate partially into the membrane, resulting in severely
Electrodialysis 405
Electrode cell Cation-exchange Anion-exchange

membrane
rinse solution
\ Spacer
Figure 8-4. Exploded view of components in an electrodialysis stack.

reduced ion permeability. Multivalent cations and polyelectrolytes are often very
strongly attached to the corresponding counter-ions within the membrane, so
they are very difficult to remove. Hydrodynamic procedures generally have no
effect at all. Cleaning procedures with solutions of either very high or low
pH-values are generally more effective. A very efficient cleaning technique is
short-term operation of the unit with reversed polarity, which will be described in
more detail later.‘*
8.2.2.3 Mechanical, hydrodynamic, and electrical stack design criteria
In designing an electrodialysis stack, several general criteria concerning

mechanical, hydrodynamic, and electrical properties have to be considered. Since
some criteria conflict, the final stack construction is usually a compromise.*lg
A proper electrodialysis stack design should first provide the maximum

effective membrane area per unit stack volume. Consequently, the membrane
area obscu:ed by seals or spacers should be small. Second, the stack’s solution
distribution design should ensure equal and uniform flow distribution through each
compartment and the solution distribution channel shouid not be easily blocked by
particles in the feed stream. The correct flow distribution is provided by the
spacer screen which should provide closely spaced support points for the
membranes without obscuring a large membrane area. The spacer should provide a
maximum of mixing of the solution in the cell, but at the same time should cause
a minimum in pressure loss to minimize pumping costs. The flow distribution of
the solution being processed should be equal over the entire width of each
compartment and the solution velocity should be equal at all points within a
compartment. This requires that the compartment thickness and the hydraulic
resistance of the spacer screen be uniform over the entire width and length of
the compartments. Low pressure drop across the stack is desirable to minimize the
pumping energy requirements and to avoid pressure differences between different
compartments within the stack, which could lead to bulging of the compartment.
thus changing its flow cross section and putting additional stress on the
membranes. Third, leakage between the diluate, concentrate, and the electrode
cells has to be prevented. Leakage between the individual cells and to the outside
can also be a severe problem.‘g Finally, to ensure a maximum current
utilization, electrical leakages through the solutions in the supply ducts should be
small. Therefore the resistance of the solution in the supply ducts should be high
and the cross-sectional area of the duct small. All gaskets and spacer materials
should have high electrical resistance.
Consideration of the various criteria discussed above has led to a variety of

electrodialysis stack designs. Most stack designs used in modern large-scale
electrodialysis plants are one of two basic types: tortuous path or sheet flow.
These designations refer to the type of solution flow path in the compartments of
the stack. In the tortuous-path stack, the membrane spacer and gasket have a
long serpentine cut-out which defines a long narrow channel for the fluid path.
The objective is to provide an extended residence time for the solution in each
cell, in spite of the high linear velocity that is required to limit polarization
effects. A tortuous-path spacer gasket is shown schematically in Figure 8-5. The
flow channel makes several 180” bends between the entrance and exit ports,
which are positioned in the middle of the spacer. The channels contain
cross-straps to promote turbulence of the feed solution.
Electrodialysis 407
Feed aolutlon
flaw nath
Feed solution
inlet
Product solution
outlet
Turbulence
promoters
Figure 8-5. Schematic diagram of a tortuous-path electrodiaiysis spacer gasket.

In stack designs employing the sheet-flow principle, a peripheral gasket

provides the outer seal and a plastic net or screen is used to prevent the
membranes touching each other. The solution flow in a sheet-flow type stack is
a straight path from the entrance to the exit ports, which are located on opposite
sides in the gasket. Figure 8-6 shows a schematic diagram of a sheet-flow
spacer. Solution flow velocities in sheet-flow stacks are typically between 5 and
10 cm/set, whereas, in tortuous-path stacks, solution flow velocities of 30 to 50
cm/set are required. Because of the higher flow velocities and longer flow paths,
pressure drops on the order of 5-6 bars occur in tortuous-path stacks. In
contrast, the pressure drop in a sheet-flow system is usually 1-2 bars. The
membranes used in tortuous-path stacks are thicker and more rigid than those
used in sheet-flow systems, because they must resist deflection between the
widely spaced support.
There are several other stack concepts described in the literature, especially
in patents. Most are not used in any practical application. Stack constructions
which provide three or four independent solution flow cycles are used in some
specific applications, for example in combination with bipolar membranes.20~21~22
8.2.3 Ion-Exchange Membranes Used in Electrodialysis
The most important components in an electrodialysis unit are the

ion-exchange membranes. They should have a high selectivity for oppositely
charged ions and a high ion permeability, i.e. a low electric resistance.
Furthermore they should have high form stability, i.e. a low degree of swelling
and good mechanical strength at ambient and elevated temperatures. Good
chemical stability over a wide pH-range and in the presence of oxidizing agents is
also required. The properties of ion-exchange membranes are closely related to
those of ion exchange resins. There are two different types of membranes:
cation-exchange membranes. which contain negatively charged groups fixed to a
matrix; and anion-exchange membranes, containing positively charged groups.
There are many possible types of ion-exchange membranes with different matrices
and different functional groups to confer the ion-exchange properties. There are
a number of inorganic ion-exchange materials, most of them based on silica,
bentonites and oxyhydrates of aluminum and zirconium. These materials are
unimportant in ion-exchange membranes, which are produced almost exclusively
from synthetic polymers.
A typical ion-exchange membrane is shown schema:ically in Figure 8-7. The

membrane is composed of a polymer matrix containing fixed negatively charged
groups, which are counterbalanced by mobile positively charged cations, usually
referred to as counter-ions. Mobile anions, usually called co-ions, are essentially
excluded, since they are carrying the same charge as the fixed negatively charged
groups. This type of exclusion is referred to as Donnan exclusion in honor of the
pioneering work of F. S. Donnan. Due to the exclusion of the co-ion in a
cation-exchange membrane. the cations carry virtually all of the electric current
through the membrane. Likewise, in anion-exchange membranes the current is
mainly carried by anions.
Electrodialysis 409
Feed SOlutiOn
Feed solution
- Spacer screen
T-1
I------
I
Product solution
outlet
Figure 8-6. Schematic diagram of a sheet-flow electrodialysis spacer gasket.

+ Matrix with Fixed Charges
OC ounter- Ion
0 Co-Ion
Figure g-7. Schematic diagram of a cation-exchange membrane.

Electrodialysis 411
Many methods of making ion-exchange membranes are described in the

literature. Some companies have more than 500 issued patents describing detailed
recipes for making membranes with special properties. Most of these patents
were issued between 1950 and 1970 and have no commercial significance today.
Ion-exchange membranes may be homogeneous or heterogeneous. Heterogeneous
membranes can be made by incorporating ion-exchange particles into film-forming
resins by various methods, including dry molding, calendering, dispersing the ion-
exchange material in a solution of the film-forming polymer, then solution
casting, and dispersing the ion-exchange material in a partially polymerized
film-forming polymer, casting films and completing the polymerization.
Heterogeneous exchange membranes have several disadvantages, the most
important of which are relatively high electrical resistance and poor mechanical
strength. Homogeneous ion-exchange membranes have significantly better
properties in this respect, because the fixed ion charges are distributed more
homogeneously over the entire polymer matrix.7 Most commonly used ion-
exchange membranes are of the homogeneous type.
The common methods of preparing homogeneous membranes are as follows:
(0 Polymerization of mixtures of reactants (e.g. phenol, phenolsulfonic acid. and

formaldehyde) that can undergo condensation polymerization. At least one of
the reactants must contain a moiety that either is or can be made anionic or
cationic.
(2) Polymerization of mixtures of reactants (e.g. styrene, vinylpyridine, and

divinylbenzene) that can polymerize. At least one of the reactants must
contain an anionic or cationic moiety, or one that can be made to do so.
(3) Introduction of anionic or cationic moieties into a polymer or preformed

films by techniques such as imbibing styrene into polyethylene films,
polymerizing the imbibed monomer, and then suffonating the styrene. Other
similar techniques, such as graft polymerization, have been used to attach
ionized groups onto the molecular chains of preformed films.27
(4) Introduction of anionic or cationic moieties into a polymer chain such as

polysulfone. followed by dissolving the polymer and casting the solution into
a film.“s
(5) Forming polymer alloys or interpolymers by mixing a finely dispersed ion-

exchange resin in a polymer matrix.
Membranes made by any of the above methods may be cast or formed around
screens or other reinforcing materials to improve their strength and dimensional
stability. Anionic or cationic moieties most commonly found in commercial
ion-exchange membranes are -SO3- or -NR3+. However. other charged groups,
such as -COO-, POsz-, and -HP02-, as well as various tertiary or quaternary
amines, are used in ion-exchange membranes. The resistance of ion-exchange
membranes used today is in the range of 1 - 2 R cm2 and the fixed charge
density is about I - 2 m equiv./g.
8.2.4 Historical Developments
Studies of electrodialysis began in Germany in the early part of this

century.2g Originally, the process was carried out in single cell arrangements
with nonselective membranes and was of no practical use. Manegold and
Kalauchsn suggested the use of selective cation- and anion-exchange membranes
and Meyer and Straussal introduced the multicell arrangement between a pair of
electrodes as indicated in Figure 1. In such a multicompartment electrodialyser.
demineralization or concentration of solutions containing ionic components was
possible with reasonable energy efficiency, since irreversible effects represented
by the water decomposition at the electrode could be distributed over many
demineralization compartments.
With the development of ion-exchange membranes with low electrical

resistance directly after the second world war, multicompartment electrodialysis
became commercially available for demineralizing or concentrating various
electrolyte solutions.3Z~33 During the 1960s the United States Office of Saline
Water directly supported research and development of electrodialysis for the
production of potable water from brackish water. Similar smaller programs in
Europe, Israel and South Africa stimulated the development of new membranes
and improved processes. At the same time, several Japanese companies developed
electrodialysis as a means of concentrating seawater for use as a brine for the
chlor-alkali industry and for producing table salt. These early electrodialysis
systems were all operated unidirectionally, i.e. the polarity of the two electrodes,
and hence the position of the dilute and concentrated cells, was permanently
fixed in an electrodialysis stack. This mode of operation often led to scale
formation and membrane fouling caused by the precipitation of low soiubility salts
on the membrane surface. Scaling affects the efficiency of electrodialysis
significantly and the materials precipitated at the membrane surface have to be
removed by flushing with cleaning solutions, the frequency depending on the
concentration of such materials in the feed solution. The control of scaling and
membrane fouling generally leads to an increase in the capital and operating
costs. Extreme cases of membrane scaling and fouling can make the process
economically unattractive. A significant advance in scaling control was the
introduction of a special operating mode referred to as electrodialysis reversal
(EDR). EDR was introduced by lonics Inc. to continuously produce demineralized
water without constant chemical addition.‘a
In EDR systems, the polarity of the electrodes is periodically reversed. This

reverses the direction of ion movement within the membrane stack, thus
controlling membrane fouling and scale formation. Typically, reversal occurs
approximately every IS minutes and is accomplished automatically. Upon reversal
the streams that formerly occupied concentrate compartments become
demineralized streams. Automatic valves switch the inlet and outlet streams, SO
that the incoming feed water flows into the new demineralizing compartments and
any concentrate stream remaining in the stack must now be desalted. This creates
a brief period of time in which the demineralized stream (product water) salinity
is higher than the specified level.
Electrodialysis 413
Because of reversal, no flow compartment in the stack is exposed to high

solution concentrations for more than 15 to 20 minutes at a time. Any build-up of
precipitated salts is quickly dissolved and carried away when the cycle reverses.
EDR effectively eliminates the major problems encountered in unidirectional
systems.
In summary, the development of electrodialysis as an efficient

demineralization and ion exchange process is characterized by three major
innovations:
(I) the use of highly selective cation and anion-exchange membranes,

(2) the use of a multi-compartment stack design,
(3) the polarity-reversal operating mode.
8.3 CURRENT APPLICATIONS OF ELECTRODIALYSIS
Electrodialysis was developed first for the desalination of saline solutions,

particularly brackish water. The production of potable water is still the most
important industrial application of electrodialysis. But other applications, such as
the treatment of industrial effluents, the production of boiler feed water,
demineralization of whey and deacidifying of fruit juices are gaining increasing
importance and are found in large-scale industrial installations. Another
application of electrodialysis, limited to Japan and Kuwait, that has gained
considerable commercial importance is the production of table sa!t from seawater.
Diffusion dialysis and the use of bipolar membranes have expanded the
application of electrodialysis significantly in recent years.52
The industrial applications of electrodialysis, their present status, market

size, expected future growth and the market leaders are summarized in Table 8-l.
In the following sections, each of these application areas is briefly reviewed.
8.3.1 Desalination of Brackish Water by Electrodialysis
Electrodialysis competes directly with reverse osmosis and multistage flash

evaporation in desalination applications. For water with relatively low salt
concentration (less than 5000 ppm) electrodialysis is generally the most economic
process, as indicated earlier. One significant feature of electrodialysis is that the
salts can be concentrated to comparatively high values (in excess of I8 to 20
wt%) without affecting the process economics significantly. Most modern
electrodialysis units operate in EDR mode, preventing scaling due to
concentration polarization effects. lonics leads the brackish water treatment
market, with more than 1,500 plants with a total capacity of more than 600,000
m3/day of product, requiring a membrane area in excess of l,OOO,OOOm2. The
largest installation produces 24,000 ms/day for a refinery in the Middle East.
Installations in Russia and China for the production of potable water are
estimated as being of the same order of magnitude. Exact data, however, are
difficult to obtain. Other manufacturers of electrodialysis equipment i? Europe and
Japan seem to be of minor importance for the production of potable water.
Table 8- 1. Industrial Applications of Eiectrodialysis, Status of the Art,

Current Problems, and Future Developments
Application Membranes & status of Key Market Market

stack design theart problems size growth lea&s
$ millions %
brackish water anion- & cation- com- costs 50 10 Ionics

desalination exchange membranes mercial
boiler feed- anion- & cation- com- scaling 30 15 Ionics

water, indusuial exchange membranes, mercial costs
process water tortuous path stack
production anion- & cation- com- costs 15 - Tokuyama

of table salt exchange membranes, mercial
sheet flow stack
industrial anion- & wtion- com- costs 15 15 Ionics

effluent exchange membmnes, mercial nontoxic
treatment sheet-flow & tortuous I-tYllOVal
stack
food and anion- & cation- com- 25 15 Ionics

pharmaceutical exchange membranes, mercial fouling,
uldusuy sheet tlow stack product loss
diffusion cation-exchange com- costs 5 - Tokuyama

dialysis membranes, mercial
of acids sheet flow stack
ultrapure lab scale process 2 hge Millipore

waler reliability
water splitting bipolar membranes corn- membrane 2 very Allied Signti

three cell stack design mercial performance
and life
Electrodialysis 415
8.3.2 Production of Table Salt
The production of table salt from seawater by using electrodialysis to

concentrate sodium chloride up to 200 g/L prior to evaporation is a technique
developed and used nearly exclusively in Japan. More than 350,000 tons of table
salt are produced annually by this technique, requiring more than 500,000 m2 of
installed ion-exchange membranes. Market leaders in this application are
Tokuyama Soda, Asahi Glass, and Asahi Chemical. The key to the success of this
technology has been low-cost, highly conductive membranes, with a preferred
permeability for monovalent ions. In Japan, however, table salt production by
electrodialysis is heavily subsidized, so the method may not be cost-effective
elsewhere.
8.3.3 Electrodialysis in Wastewater Treatment
The main application of electrodiaiysis in wastewater treatment systems is in

processing rinse waters from the electroplating industry. Complete recycling of
the water and the metal ions is achieved. Compared to reverse osmosis,
electrodialysis has the advantage of being able to utilize thermally and chemically
stable membranes, so the process can be run at elevated temperatures and in
solutions of very low or high pH values. Furthermore, the concentrations which
can be achieved in the brine can be significantly higher. The disadvantage of
electrodialysis is that only ionic components can be removed and additives usually
present in a galvanic bath cannot be recovered.
An application which has been studied in a pilot-plant stage is the

regeneration of chemical copper plating baths. In the production of printed
circuits, a chemical process is often used for copper plating. The components
which are to be plated are immersed into a bath containing, besides the copper
ions, a strong complexing agent, for example, ethylene diamine tetra acetic acid
(EDTA), and a reducing agent such as formaldehyde. Since all constituents are
used in relatively low concentrations, the copper content of the bath is soon
exhausted and copper sulfate has to be added. During the plating process,
formaldehyde is oxidized to formate. After prolonged use, the bath becomes
enriched with sodium sulfate and formate and consequently loses useful
properties. By applying electrodialysis in a continuous mode, the sodium sulfate
and formate can be selectively removed from the solution, without affecting the
concentrations of formaldehyde and the EDTA complex. Hereby, the useful life of
the plating solution is significantly extended.50 Several other successful
applications of electrodialysis in wastewater treatment systems that have been
studied on a laboratory scale are reported in the literature.52 Large,
commercially operated plants are at present, however, rare. lonics and Tokuyama
Soda are the market leaders. However, because the plant capacities needed for
this application are smaller than in desalination, there are good opportunities for
smaller companies to compete.
8.3.3.1 Concentration of Reverse Osmosis Brines
A further application of electrodialysis is concentration of reverse osmosis

brines. Because of limited membrane selectivity and the osmotic pressure of
concentrated salt solutions, the concentration of brine in reverse osmosis
desalination plants cannot exceed certain values. Often the disposal of large
volumes of brine is difficult and further concentration is desirable. This further

concentration may be achieved at reasonable cost by electrodialysis.
8.3.4 Electrodialysis in the Food and Pharmaceutical Industries
The use of electrodialysis in the food and pharmaceutical industries has been
studied extensively in recent years. Several applications have considerable
economic significance and are already well established. One is the
demineralization of cheese whey. Normal cheese whey contains between 5.5 and
6.5% of dissolved solids in water. The primary constituents in whey are lactose,
protein, minerals, fat and lactic acid. Whey provides an excellent source of
protein, lactose, vitamins, and minerals, but in its normal form it is not
considered a proper food material because of its high salt content. With the
ionized salts substantially removed, whey approaches the composition of human
milk and, therefore, provides an excellent source for the production of baby food.
The partial demineralization of whey can be carried out efficiently by
electrodialysis. The process is used extensively and is described in detail in the
literature.53
The removal of tartaric acid from wine is another application. Especially in

the production of bottled champagne, the formation of crystalline tartar in the
wine must be avoided, so the tartaric acid content must be reduced to a value
below the solubility limit. This can be done efficiently by electrodialysis.
Desalting of dextrane solutions, another application for electrodialysis, has
technical significance as a potential large-scale industrial process.
Other applications of electrodialysis in the pharmaceutical industry have

been studied on a laboratory scale. Most applications are concerned with
desalting or with treating solutions containing active agents that have to be
separated, purified, or isolated from certain substrates. Here, electrodialysis is
often in competition with other separation procedures, including dialysis and
solvent extraction. In many cases, for example the separation of amino acids
and other organic acids, electrodialysis is the superior process as far as economics
and product quality are concerned.
To date, the largest food industry application of electrodialysis is in cheese

whey processing. There is an installed capacity of more than 35,000 m” of
membrane area to produce more than 150,000 tons of desalted lactose per year.
The market leaders are Tokuyama Soda, Asahi Glass and lonics.
8.3.5 Production of Ultrapure Water
Recently, electrodialysis has been used for the production of ultrapure

water for the semiconductor industry, especially in combination with mixed bed
ion exchange resins as shown in Figure 8-8. In this process, ion exchange resin
beads are sandwiched between the electrodes of an electrodialysis stack. The
applied current continuously removes the ions trapped by the ion exchange resin.
In this way, the feed water can be almost completely deionized and no chemical
regeneration of the resin bed is required. This process was suggested many
years ago by 0. Kedem and coworkers*’ and has recently been commercialized by
Millipore.56
Electrodialysis 417
Cathode
-0
u H _r CJ. ..!
Brine
Figure 8-8. Principle of electrodialytic regeneration of mixed bed ion exchange

resins for the production of deionized water. (IX: Ion exchange
resin; A: Anion transfer membrane; C: Cation transfer membrane.)
8.3.6 Other Electrodialysis-Related Processes
In addition to conventional electrodialysis, several processes closely related

to electrodialysis have been discussed in the literature. Most of these processes
are still in the laboratory stage.
8.3.6.1 Donnan-dialysis with ion-selective membranes
In Donnan-dialysis, the ion concentration difference in two phases separated

by an ion-exchange membrane is used as the driving force for the transport of
ions with the same electrical charges in opposite directions. The principle of the
process is shown schematically in Figure 8-9. This figure shows as an example
solutions of CuSO, and HsSO,, separated by a cation-exchange membrane. Since
the H+ ion concentration in solution I is significantly higher (pH = I) than the
H+ ion concentration in solution 2 (pH = 7) there will be a constant driving force
for the flow of H+ ions from solution 1 into solution 2. Since the membrane is
permeable for cations only, there will be a build-up of electrical potential that
will balance the concentration difference driving force of the H+ ions. As a
result, ions of the same charge will be transported in the opposite direction as
indicated by the flow of Cu++ ions from solution 2 into solution 1 in Figure 8-9.
As long as the H+ concentration difference between the two phases separated by
the cation-exchange membrane is kept constant, there will be a constant transport
of Cu++ ions from solution 2 into solution 1 until the Cu++ ion concentration
difference reaches the same order of magnitude as the H+ ion concentration
difference, i.e. Cu++ ions can be transported against their concentration gradient
driving force by Donnan-dialysis.
The same process can be carried out with anions through anion-exchange
membranes. An example of anion Donnan-dialysis is the sweetening of citrus
juices. In this process, hydroxyl ions furnished by a caustic solution replace the
citrate ions in the juice.
8.3.6.2 Electrodialytic water dissociation
A process referred to as electrodialytic water dissociation or water

splittings4 to produce acids and bases from salts, has been known for a number
of years. The process is conceptually simple, as shown in the schematic diagram
in Figure S-10.
A cell system consisting of an anion-, a bipolar-, and a cation-exchange

membrane as a repeating unit is placed between two electrodes. Sodium sulfate
or other salt solution is placed in the outside phase between the cation- and
anion-exchange membranes. When a direct current is applied. water will dissociate
in the bipolar membrane to form an equivalent amount of hydrogen and hydroxyl
ions. The hydrogen ions will permeate the cation-exchange side of the bipolar
membrane and form sulfuric acid with the sulfate ions provided by the sodium
sulfate from the adjacent cell. The hydroxyl ions will permeate the
anion-exchange side of the bipolar membrane and form sodium hydroxide with the
sodium ions permeating into the cell from the sodium sulfate solution through the
adjacent cation-exchange membrane. The net result is the production of sulfuric
acid and sodium hydroxide at significantly lower cost than by conventional
Electrodialysis 419
techniques. The most important part of the cell arrangement is the bipolar
membrane, which consists of an anion- and a cation-exchange membrane
laminated together. The membrane should have good chemical stability in acid
and base solutions and low electrical resistance. Laboratory tests have
demonstrated that production costs for caustic soda by utilizing bipolar membranes
are only one-third to one-half the costs of the conventional electrolysis process.
The process has recently been commercialized by Allied Corporation’s Aquatech
Systems Division. The process is affected by limited alkaline and temperature
stability of the anion-exchange part of the bipolar membrane. Further
improvements can, however, be expected in the near future.4s~47
-
1
Phase ‘I Phase ’
/k
-
pH = 7 - pH = 1
-
H+ 0
z
4
-
-
---
b, ++
so4 cu
-
/ ,
it+
cu
++
/
Cation exchange
1 membrane
Figure 8-9. The principle of Donnan-dialysis.

HR NaOH HR
4 + 4
* + + + 4
Cathode
Anode
-I t
NaR NaR
. . . . . . . . . . . . . . . Cation transfer membrane

. . . . . . . . . . Amion transfer membrane
.___I__..._
. . . . . . . . . . . . . . Bipolar membrane
Figure 8- 10. Schematic diagram showing the electrodialytic regeneration of an

acid and sodium hydroxide from the corresponding salt employing
bipolar membranes.
Electrodialysis 421
8.4 ELECTRODIALYSIS ENERGY REQUIREMENT
8.4.1 Minimum Energy Required for the Separation of Water from a Solution
In electrodialysis, as in any other separation process, there is a minimum

energy required for the separation of various components. For the removal of salt
from a saline solution this energy is given by:
a0
W” = RT In3 (5)
as
W
Here W” is the minimum energy required to remove one mole of water from a
solution, R the gas constant and T the temperature in ‘K; a$ and c are the
water activtttes in the pure state and the solution. Expressing the water acttvtty
in the solution by the concentration of the dissolved ionic components, the
minimum energy required to remove water from a monovalent salt is given by:
W” = 2RT (co - c’)
(6)
where W” refers to the minimum required energy for the production of I L of

diluate solution. and Co, C’ and C” refer to the salt concentration in the feed
solution, the diluate and the concentrate.
8.4.2 Practical Energy Requirement in Electrodialysis Desalination
Because of irreversible effects, the energy required to remove water from a

solution by electrodialysis is significantly higher than the theoretically minimum
energy. The practical required energy may be 10 to 20 times the theoretical value.
The energy required in practical electrodialysis is an additive of three terms. The
first is the electrical energy, Eit, to transfer the ions from the feed solution
through the membrane into the brine. The second term is the energy
consumption, E+ due to the electrochemical reactions at the electrodes and the
resistance of the electrode cells. The third term is the energy required to pump
the feed solution, the diluate and the brine through the electrodialysis stack,
E pump. Thus, the total energy consumption is given by
E tot = Eit + Eclr + Epump
Depending on various process parameters, particularly the feed solution

concentration, any of these terms may be dominant, thus determining the overall
energy costs.
8.4.2.1 Energy requirements for transfer of ions from the product solution to
the brine.
The energy necessary to remove salts from a solution is directly proportional

to the total current flowing through the stack and the voltage drop between the
two electrodes in a stack. The energy consumption in a practical electrodialysis
separation procedure can be expressed as:
E = nIZRt (8)
Here E is the energy consumption, I the electric current through the stack, R the
resistance of the cell, n the number of cells in a stack, and t the time. The
electric current needed to desalt a solution is directly proportional to the number
of ions transferred through the ion-exchange membranes from the feed stream to
the concentrated brine. This is expressed as:
zFQAC
I= (9)
E
where F is the Faraday constant, z the electrochemical valence, Q the product

solution flow rate, AC is the concentration difference between the feed solution
and the diluate and [ the current utilization. The current utilization is directly
proportional to the number of cells in a stack and is governed by the current
efficiency. In practical electrodialysis, the efficiency with which ions can be
separated from the mixture by the electrical current is usually less than 100%.
A combination of Equations (8) and (9) gives the energy consumption in

electrodialysis as a function of the current applied in the process, the electrical
resistance of the stack (i.e. the resistance of the membrane and the electrolyte
solution in the cells), current utilization and the amount of salt removed from
the feed solution. Thus:
nIRtzFQAC
E =
(10)
E
Equation (IO) indicates that the electrical energy required in electrodialysis is

directly proportional to the amount of salt that has to be removed from a
certain feed volume to achieve the desired product concentration. Energy
consumption is also a function of the number of cells in a stack and the
electrical resistance in a cell. Electrical resistance is a function of individual
resistances of the membranes and of the solutions in the cells. Because the
resistance of the solution is directly proportional to its ion concentration, the
overall resistance of a ceil will in most cases be determined by the resistance of
the diluate solution. This is a factor that must be taken into account in the
design of an electrodialysis stack. A typical value for the resistance of an
electrodialysis cell pair, including the cation- and anion-exchange membrane plus
the dilute and concentrated solution, is in the range of IO to 100 II cm’.
Electrodialysis 423
8.4.2.2 Pump energy requirements
Electrodialysis systems use three pumps, to circulate the diluate (the solution
depleted of ions), the brine (the solution into which the ions are transferred), and
the electrode rinse solutions. The energy required for pumping these solutions is
determined by the volumes to be circulated and the pressure drop in the
electrodialysis unit. It can be expressed by the following relation:
E, = k, QQ APQ + k, Qg APa + ks QE APE , (11)
where E, is the pump energy, k,, k,, and k, are constants referring to the
efficiency of the pumps, Qp, Qn, and Qts are volume flows of the diluate, brine,
and electrode rinse solutions, and AP,, AP,, and APE are the pressure losses in
the diluate, the brine and the electrode cells. The pressure losses in the various
cells are determined by the solution flow velocities and the cell design. The
energy requirements for circulating the solution through the system may become
significant or even dominant when solutions with low salt concentrations, less
than 500 ppm, are processed.
W.2.3 Energy requirement for the electrochemical electrode reactions
An electrodialysis cell usually generates hydrogen at the cathode and oxygen

or chlorine at the anode. The energy consumed in electrochemical reactions at
the electrodes is given by the total current passing through the stack multiplied
by the voltage drop at the electrodes and in the electrode cells. In a multicell
arrangement the energy consumed at the electrodes is small, generally less than
1% of the total energy used for the ion transfer from a feed to a brine solution.
8.4.3 Energy Consumption in Electrodialysis Compared with Reverse Osmosis
Electrodialysis competes with other separation processes in many applications.

The minimum energy required to perform a theoretical separation is identical in
all processes, but there are significant differences as far as the energy
consumption in a practical separation problem is concerned. For desalination. for
instance, reverse osmosis, ion exchange, distillation, or electrodialysis can all be
used. All four processes require different amounts of energy depending on the
composition of the feed solution. The differences between the energy demands of
reverse osmosis and electrodialysis can be illustrated by comparing the basic
principles of the processes as shown in Figure 8-1 I.
In reverse osmosis, water passes through the membrane under a driving force
created by a hydrostatic pressure difference. Ignoring concentration polarization
effects, the irreversible energy loss is caused primarily by friction between the
individual water molecules and the polymer membrane matrix. This frictional
energy loss is independent of the salt concentration in the feed solution. In
electrodialysis, ions pass through the membrane under a driving force created by
an electrical potential difference. In this case, therefore, the irreversible
frictional energy losses are directly proportional to the salt concentration in the
feed water. Thus, for feed solutions with low salt concentration. the energy
requirements are lower in electrodialysis than in reverse osmosis, and at high

feed-solution salt concentration the energy consumption in eiectrodialysis is
higher than in reverse osmosis. This is shown in Figure S-12, where the
irreversible energy consumption is plotted versus the feed solution concentration,
assuming identical product water concentrations.
A comparison of the energy consumptions of different mass separation

processes has to take into account the different forms in which the energy may
be required. Eiectrodialysis uses electricity, a relatively expensive form, but
distillation uses heat, which is comparatively inexpensive. In ion exchange, very
little energy is required directly, but the chemicals used to regenerate the resin
require a significant amount of energy for their production.
Sdit ana bvawr Salt ana waler
Arnons Cations
AE AE
Salt Water
Reverse osmosis Electrodialysis
Figure 8-I I. Schematic diagram comparing the operating principles of reverse

osmosis and electrodialysis.
feed solution salt concentration
Figure 8- 12. Schematic diagram showing the irreversible energy losses in

electrodialysis and reverse osmosis as a function of the feed-
solution salt concentration.
Electrodialysis 425
8.5 ELECTRODIALYSIS SYSTEM DESIGN AND ECONOMICS
8.5.1 Process Flow Description
A flow diagram of a typical electrodialysis plant as used for desalination of

brackish water is shown in Figure 8-13. After proper pretreatment, which may
consist of flocculation, removal of carbon dioxide, pH control or prefiltration, the
feed solution is pumped through the electrodialysis unit. A deionized solution and
a concentrated brine are obtained. The concentrated and depleted process streams
leaving the last stack are collected in storage tanks, or are recycled if further
concentration or depletion is desired. Acid is frequently added to the
concentrated stream to prevent scaling of carbonates and hydroxides. To prevent
the formation of free chlorine by anodic oxidation, the electrode cells are
generally rinsed with a separate solution which does not contain any chloride
ions. In many cases, the feed or brine solution is also used in the electrode
cells.
An important variation to the basic operating mode of Figure 8-13 is

electrodialysis reversal (EDR). In this operating mode the polarity of the current
is changed at specific time intervals ranging from a few minutes to several hours.
When the current changes, the diluate cell becomes the brine cell, and vice versa.
The advantage of the reverse polarity operating mode is that precipitation in the
brine cells is essentially prevented. Any precipitation that does occur will be
redissolved when the brine cell becomes the diluate ceil in the reverse operating
mode. The flow scheme of a typical electrodialysis reversal plant, taken from
lonics Inc. brochures, is shown in Figure 8-14.
8.5.2 Electrodialysis Plant Components
An electrodialysis plant consists of four basic components: the membrane

stack, the power supply, the hydraulic flow system, and process control devices.
8.5.2.1 The electrodialysis stack
A typical electrodialysis stack consists of 200 to 500 cation- and

anion-exchange membranes arranged in an alternating pattern between two
electrodes, which are generally assembled in separate ceils. The membranes are
separated by suitable gaskets, with two membranes forming a cell pair. Typical
spacer-gasket configurations have been discussed earlier and are shown in
Figures 8-5 and 8-6.
The final stack design is almost always a compromise between a number of

conflicting criteria and considerations, e.g. short distances between membranes for
low electrical resistance, high feed flow velocities and high turbulence for control
of concentration polarization effects, low pressure losses in the solution pumped
through the stack and low production costs.
Electrodes rinse solution
Membrane stat
JAcld / Rinse solution \/Concentrate
Figure S-13. Flow diagram of a typical electrodialysis desalination plant.
Figure S- 14. Typical electrodialysis reverse polarity (EDR) plant.5e

Electrodialysis 427
8.5.2.2 The electric power supply
Electrodialysis requires DC electrical power for the stack and AC power for
the pumps. The DC power is usually supplied on-site by utilizing an AC-to-DC
converter. Conversion efficiencies of about 90 percent are typical. Constant
voltage regulators are utilized to maintain stable plant operation and used to
prevent stack damage. Stack resistance changes occur as a result of scale
formation, membrane deterioration and changes in the fluid concentrations within
the stack. The voltage regulator is adjusted periodically to compensate for these
changes. The voltage drop across each cell pair in an electrodialysis stack is
about 0.5-2 V, so the total voltage drop across the stack is typically about 200-
800 V. The current flowing through the stack is 400 A, which yields a current
density up to 40 mA/cm’ if the membrane aiea of the cell is 1 m2.
8.5.2.3 The hydraulic flow system
The primary considerations in designing the hydraulic flow system are to

obtain low hydraulic pressure drops, yet simultaneously achieve high volume flow
rates. The pressure drop in an electrodialysis system is on the order of 2-6
bars. Simple plastic centrifugal pumps are generally used for circulating the
different solutions through the stack.
8.5.2.4 Process control devices
The following variables are usually measured or controlled, or both:
1) DC voltage and current supplied to each electrodialysis unit,

2) Flow rates and pressures of the depleted and concentrated streams,
and of the electrode rinse streams,
3) Electrolyte concentrations of the depleted and concentrated streams
at the inlets and outlets to the electrodialysis stacks,
4) pH of the depleted stream and the electrode rinse streams,
5) Temperature of the feed stream.
All the above variables are interrelated. Automatic control of the flows of
the depleted and concentrated streams can be achieved by the use of flow-type
conductivity cells in the effluent streams, along with a controller that compares
the conductivities of the streams with that of a preset resistance and actuates
flow-control valves in the liquid supply lines.
To prevent damage to the membranes or other components in the event of

stoppage of liquid flow to the stacks, an electrodialysis unit is normally provided
with fail-safe devices that will turn off the power to the stacks and pumps.
8.5.3 Electrodialysis Process Costs
8.5.3.1 Capital
The capital cost of an electrodialysis plant is made up of depreciable and

nondepreciable items. Nondepreciable items include land and working capital, and
are outside the scope of this outline. Depreciable items include the
electrodialysis stacks. pumps, electrical equipment, membranes, etc.
The capital costs of an electrodialysis plant will strongly depend on the

number of ionic species to be removed from a feed solution. This can easily be
demonstrated for an electrodialysis plant producing potable water from saline
water sources. The total membrane area required by a plant is given by:
zFQAcn
A = (12)
it
where A is the membrane area, z the chemical valence, Q the volume of the
produced potable water, AC the difference in the salinity of feed and product
water, n the number of cells in a stack, i the current density which should be
about 80% of the limiting current density, c the current utilization and F the
Faraday constant. The limiting current density is a function of the diluate
concentration, which changes from the concentration of the original feed to the
product solution concentration. The calculation of the minimum membrane area
required for a given desalting capacity is based on an average diluate
concentration and average limiting current density, given by:
*
llim = a.cd = a (13)
where tlim is the average limiting current density, cd is the average diluate
concentration, a is a constant, which depends on the flow cell and spacer
geometry and feed flow velocity, and c, and cd are the feed and diluate
concentrations. Substituting Equation (13) into Equation (12) yields the minimum
membrane area, A,,, required for a certain plant capacity and feed and product
solution concentrations.
A,, = a . z F Q In > (14)
For a given plant capacity and current density, the required membrane area
is directly proportional to the feed water concentration. This is illustrated in
Figure S- IS.
Electrodialysis 429
10
k
Membrane area
(m3 per
m3 product/day ’
0.1 L I I I111111 I I1 Ill,,
1 10 100
Feed solution concentration (g/l)
Figure 8- 15. Membrane area required for electrodialysis desalination as a

function of the feed water concentration at constant current
density and plant capacity.
For typical brackish water containing 3,000 ppm TDS and an average current
density of 12 mA/cm2, the required membrane area for a plant capacity of 1 ms
product per day is about 0.4 m2 of each cation- and anion-exchange membrane.
The costs of pumps, electric power supply, etc. do not depend on feed water
salinity, so the dependence of the total capital costs on the feed water salinity is
nonlinear. For desalination of brackish water with a salinity of 3,000 ppm, the
total capital costs for a plant with a capacity of 1,000 m3/day will be in the
range 5200-300 per ms/d capacity. The cost of the membranes is less than 30%
of the total capital costs. Assuming a useful life of the membranes of 5 years, of
the rest of the equipment of 10 years and a plant availability of 95%, based on a
24-hour operating day, the amortization of the investment per ms potable water,
obtained from 3,000 ppm brackish water, is in the range $0.10-0.15.
8.5.3.2 Operating costs
The largest single component of the operating cost is the required energy.
All other components are minor in comparison for large-scale plants.
The energy costs in eiectrodialysis are determined- by the electrical energy

required for the actual desalting process and the energy necessary for pumping
the solution through the stack. These factors were discussed in detail in Section
4.2. The energy requirements for the production of potable water with a TDS of
< 500 ppm of saline feed water is shown schematically as a function of the feed
water concentration in Figure 8-16.
The pumping energy is independent of the feed solution salinity. Assuming a

pressure drop in the unit of about 400 KPa (4 bar), a pump efficiency of 70%,
and 50% recovery, the total pumping energy will be about 0.4 kWh/ms product
water. This indicates that, at low feed-water salt concentration, the cost for
pumping the solution through the unit might become significant.
Other components of the operating cost include costs associated with

pretreatment, which obviously may vary from nothing to a significant portion of
the total cost, and labor costs.
8.5.3.3 Total electrodialysis process costs
The total costs of electrodialysis are shown in Figure 8-17 as a function of

the applied current density for a given feed solution calculated according to
Equations (7) and (8).
The graph in Figure 8-17 shows that capital costs decrease with increasing
current density and other costs are essentially independent of current density. For
a given feed solution, there is a current density at which the total electrodialysis
orocess costs will reach a minimum.
A comparison of the cost of desalination by various processes as a function

of the feed water salinity, is shown in Figure 8-18.
Electrodialysis 431
10
Energy requirements .
(kwhlm3)
1 I #
1 10
Feed solution concentration (g/l)
Figure 8- 16. Energy requirements for the prcduction of potable water with a
solid content of 500 ppm as a function of the feed solution
concentration (AV per cell pair = 0.8 V)
Current density
Figure 8- 17. Schematic diagram of the electrodialysis process costs as a

function of the applied current density.
r Ion-exchange
// Electrodialysis
1 .o
Costs (S/m3 )
0.1
NaCl concentration (g/l)
Figure 8- 18. Costs of desalination of saline water as a function of the feed

solution concentration for ion exchange (IE), electrodialysis (ED),
reverse osmosis (RO), and multistage flash evaporation (MSF).
Electrodialysis 433
Figure 8-18 indicates that, at very low feed-solution salt concentration, ion
exchange is the most economical process. The costs of ion-exchange processes
increase sharply with the feed solution salinity, and at about 500 ppm TDS
electrodialysis becomes the most economical process. Above 5,000 ppm TDS,
reverse osmosis is the least costly. At very high feed-solution salt
concentrations, in excess of 100,000 ppm TDS, multistage flash evaporation
becomes the most economical process. The costs of potable water produced from
brackish water sources are in the range $0.2-0.5/m3.
The electrodialysis supply industry is dominated by four large companies

that produce membranes as well as equipment. These are Ionics, Tokuyama Soda,
Asahi Chemical and Asahi Glass. The three Japanese companies sell membranes,
stacks and complete installations; Ionics sells complete installations only. There
are also several companies that produce and sell only ion-exchange membranes for
the specialized equipment manufacturing industry. In addition to these market
leaders, there are a number of small companies that specialize in certain
applications that require specially designed membranes and equipment. These
organizations often operate in regionally limited areas, unlike the larger
suppliers, who all operate on a worldwide scale. The supplier industry, broken
down by companies, with their main products and area of applications, is
summarized in Table 8-2.
The three major Japanese manufacturers of electrodialysis equipment sell

more than 80% of their products in Japan; Ionics, on the other hand, sells almost
50% of its products outside the United States. Table 8-3 lists the market
distributions of the major suppliers.
It is interesting to note that companies have specialized in applications of

importance to their home country. Table salt production is of interest only in
Japan and, not unnaturally, Japanese companies dominate this application. The
market for brackish water desalination lies predominantly in America, the Middle
East, and Europe, and here a United States based company is the market leader.
Food processing, especially deashing of whey, is of interest in almost any
industrialized nation and this market is more evenly distributed among the four
large equipment manufacturers.
Table 8-2. Manufacturers of Electrodialysis Equipment and Membranes”
Company Membrane Equipment Area of Annual electrodialysis

manufacturer producer application estimated sales
(trade names) (S millions)
Asahi Chemical Aciplex@ table salt 20
Co. Ltd. production
Asahi Glass Selemion@ table salt 20

Co. Ltd. production
Ben Gurion Neginst@

University
Ionics Inc. Aquamite brackish, waste 70

water, food
Sot. Recherches whey 4

Techniques et
Industrielles (SRTI)
Stantech GmbH wastewater, food <5
Ionac Chem. Co.

Div. of Sybron Corp.
Pall/RAI
Research Corp. PermiorF
Tokuyama Soda Neosepta@ table salt 20

Co. Ltd. production
Morinaga Milk
Ind. whey IO
Toyo Soda Manu- Scrion@

fact. Co. Ltd.
Others <20
Electrodialysis 435
Table 8-3. Market Distribution in Various Electrodialysis Applications for the

Major Electrodialysis Supplier Companies
Company Salt production Water desalination Food processing

(%I (96) @J)
Asahi Chemical 40 5 5
Asahi Glass 25 2 8
Tokuyama Soda 30 3 8
Toyo Soda Organ0 5 1 4
Ionics 0 80 65
Others 0 9 IO
8.7 SOURCES OF INNOVATION - CURRENT RESEARCH
Electrodialysis has a wide potential range of applications, requiring a

multitude of different process and stack design concepts, and/or membranes with
special properties. Applications such as brackish water desalination or table salt
production can be considered as state-of-the-art techniques. In other
appbcations, such as the recovery of bioproducts from a fermentation broth,
treatment of a special industrial effluent or the recovery of bases and acids from
the corresponding salts, electrodialysis is still a developing process being moved
forward by academic research.
Compared to other membrane processes, there is considerably less research in

electrodialysis being carried out in academic institutions and publicly funded
research centers. Most research at all levels is carried out within private
industry.
The principal problems encountered in state-of-the-art electrodialysis relate

to membrane properties and stack design criteria. Consequently much industrial
research focuses on these areas.
8.7.1 Stack Design Research
Electrodialysis efficiency, power consumption and costs are strongly affected

by the stack design. The limiting current density, feed-flow pressure losses, and
internal and external leakages impact the investment costs particularly. These
features are all determined by the stack configuration. A considerable amount of
research is being carried out in private industry to optimize spacers, gaskets and
flow ducts. Fouling and scaling, both of which are related to stack design, are
also active research topics. The best source of information concerning private
research is the patent literature. Table 8-4 lists companies active in research
related to stack design problems, with representative patent references.
Table 8-4. Companies Engaged in Electrodialysis Stack Design Research
Company Topic Representative

U.S. Patents
PPG Industries Improved electrodes 4,581,111
Ionics Cell design 4,441,918

Stack design 4,608,140
Electrode design 4,707,240
Scheicher & Schuell Cell design 4,608,147
Millipore Corp. Stack design 4,632,745
Dorr Oliver General process 4.639.380

Electrode assembly 4.670.118
Ajinomoto Co. Inc. Fouling reduction 4,711,722
Electrochem
International, Inc. Stack design 4.525.259
8.7.2 Membrane Research
Basic research is now concentrated on the mass-transport properties of

membranes, particularly bipolar membranes. Applications research in this area is
concerned with finding ion-exchange membranes with improved properties. The
most important properties required from ion-exchange membranes are low
electrical resistance, high fixed-ion concentrations, good chemical stability at high
and low pH-values, good thermal and mechanical stability, low transfer rate of
water or neutral components, low fouling tendency, and long life expectancy under
operating conditions. The fixed-ion concentration in most commercially available
membranes is on the order of l-2 m equiv./gram dry polymer. This limits the
concentration which may be obtained in the brine to a 2 N solution, because of
Donnan exclusion. A particular research goal is to obtain membranes with higher
fixed charge density, but low swelling and good mechanical stability.
Furthermore, most commercial anion-exchange membranes show poor chemical
stability in alkaline solutions at pH-values in excess of 12. Improved temperature
stability of both cation- and anion-exchange membranes is also the objective of
research activities carried out mainly in industry. The outcome of this research
work is documented mainly in the patent literature. Table 8-5 summarizes the
research as it can be established from the literature, with representative recent
references.
Electrodialysis 437
Table 8-5. Companies Engaged in Electrodialysis Membrane Research

U.S. Patents
Allied Corporation Bipolar membranes/water splitting 4,584$X7
4,592,817
4,608,141
4.629.545
4.636.289
Azkona, Inc. Higher conductivity membranes 4.652.396
BASF Bipolar membrane assembly 4,610,125

Improved ion-exchange membranes 4,585,536
Chemical-resistant membranes 4,711,907
Du Pont Improved fluorinated polymer

membranes 4,437,95 I
Eltech Systems Corporation Improved ion-exchange membranes 4.568.44 I
Celanese Corporation Improved ion-exchange membranes 4,634,530
8.7.3 Basic Studies on Process Improvements
There are many theoretical and experimental studies to improve the

electrodialysis process efficiency. The development of conductive spacers or
sealed-cell electrodialysis are typical examples.48*4P Combination of electro-
dialysis with other mass separation processes, for example, reverse osmosis or ion
exchange, has also been subject to various studies. The application of
electrodialysis in a more or less modified form to a very specific separation
problem in the chemical and pharmaceutical industries and in the treatment of
industrial effluents has also been subject to a multitude of studies.50s1 Some
representative applications research is summarized in Table 8-6.
Table 8-6. Companies Engaged in Electrodialysis Applications Research

U.S. Patents
Solco Base1 AC Bioprocessing 4576,696
4,599,176
Mitsubishi Gas Amino acid recovery 4,605,477
Shell Oil Chemical processing 4,620,9 10

4,620,9 1 I
4,620,9 I2
4,427,507
BASF Chemical processing 4,645,579
lonics Treatment of radioactive waste 4,645,625

Bioprocessing 4,678,553
4,426,323
Morton Thiokol Electroless copper plating 4.67 I,86 1
L’Air Liquide Amino alcohol production 4.678.549
Rhone-Poulenc Methionine production 4,454,012
Babcock-Hitachi Seawater desalination 4.539.088

4,539,OPl
Electrochem International Electroless copper plating 4.549,946
Allied Corporation Waste liquid treatment 4,552,635
Stamicarbon B.V. Chemical processing 4.552.636
ICI, plc Chemical processing 4,556,465

4.556,466
4.557.8 15
Laiterie Triballat Whey processing 4,559,119

Electrodialysis 439
8.8 FUTURE DEVELOPMENTS
8.8.1 Areas of New Opportunity
The electrodialysis industry has experienced a steady growth of about 15%

per year since it made its appearance as an industrial-scale separation process
about I5 years ago. This growth has been based on two major applications,
brackish water desalination and salt production. Today the markets for both
applications have very limited growth potential. To ensure further growth, new
areas of application have to be exploited. These areas will probably be in the
chemical, food and pharmaceutical industries, in the treatment of industrial and
municipal effluent, and in replacing other separation processes with high energy
consumption. In some cases the expansion of electrodialysis into new areas of
application will require only slight modifications of the conventional process. In
other cases, extensive research and development work will be needed to adapt
electrodialysis to a given separation problem. Table 8-7 lists new opportunities
for electrodialysis as a separation process as well as the developments that will
be required to realize those opportunities. The new applications of electrodialysis
are insufficiently developed to make possible any statement about the future
market. Electrodialysis processes using bipolar membranes may become
increasingly important because of their low energy consumption compared with
alternative technologies. Electrodialytic regeneration of ion-exchange membranes
may become increasingly attractive as stricter environmental discharge regulations
are enforced.
8.8.2 Impact of Present R&D Activities on the Future Use of Electrodialysis
The impact which present R&D activities will have on the future
development of electrodialysis depends on when, or if, key items such as
membranes with higher selectivities, better temperature stability or fouling
resistance, are available. A prediction of the total installed membrane area in
electrodialysis applications that are commercially available today is shown
schematically in Figure S-19. In the production of potable and industrial process
water, and in the food industries, this prediction is expected to be fairly
reliable; applications in wastewater treatment and especially the use of bipolar
membrane technology are speculative. The current and possible future
applications of electrodialysis have been ranked in terms of their technical and
commercial impact and their prospects for realization in Table 8-8. The
importance of the items is rated from I to IO, I being the lowest.
Table 8-7. New Areas of Application for Electrodialysis and Related Processes
Limiting Required Key Expected

Application factors R&D developments benefits
Electro&&& for Potable Water Production
Nitrate removal Membrane selectivity New membranes Ion-selective Low cost high
resistance membrane quality water
Seawater Membrane resistance New membranes High temperature Low cost potable
desalination ED water
Electra&&& for Wastewater Treatment
Electroplating Alkaline stability New anion- Temperature & Recycling of

wastewater of membrane exchange alkaline & acid metal ions
treatment membrane stable membranes
Pickling waste- Fixed charge New membranes Membranes with Acid recovery
water treatment density high charge
density
Paper mill waste- Fouling behavior Stack designs, New stack design Low cost effluent
water treatment new membrane nonfouling treatment
membranes
Acid and base Fixed charge density, New membranes, Membranes with Process cost
recovery from selectivity stack design high charge reduction
etching processes density,
chemically stable
Bioolar Membrane Technology
Recovery of acids Membrane efficiency, Bipolar membrane Low resistance, Energy saving
and bases from stack design stack design alkaline stable recycling of
corresponding salts bipolar membrane, acids and bases
three-compartment
stack design
Electroplating Membrane efficiency, Bipolar membrane Low resistance, Energy saving

effluent treatment stack design development, alkaline stable, recovery of toxic
stack design bipolar membrane, materials
three-compartment
stack design
Recovery of organic Bipolar membranes, Bipolar membrane, Low resistance Cost savings
acids from corre- anion-exchange stack design bipolar membrane,
sponding salts membranes, ‘open” anion-
stack cteslgn exchange membranes
Electrodialysis 441
Table 8-7. (continued)
Limiting Required Key Expected

Application factors R&D developments benefits
ponnan Dis
Acid recovery Membrane stability Membrane stack Proton selective Cost savings
design membranes
Water softening Membrane selectivity Membrane stack High permeability Cost savings
design membranes
Recovery of Membrane selectivity Membrane process Selective Cost savings

organic acids development membranes
Electrodialvsis in Food. Pharmav Industrh
Desalination of Membrane New membrane Thin membrane, Cost savings,

process water resistance low resistance low waste
emission
Ultrapure water Membrane Conductive Thin membranes, Cost savings

production resistance spacers conductive
spacers
Removal of organic Low membrane New membranes, “Open” non- Cost savings,
acids from wine and permeability stack design fouling anion- higher quality
fruit juices exchange products
membranes
Production of Low membrane Membrane “Open’ non- Cost savings

organic acid permeability stack design fouling anion-
exchange membranes
Recovery of Low membrane Membrane ‘Open” membranes, Cost savings

amino acids permeability stack design no leakage,
from fermentation stack design
Separation of Low membrane Process develop- Microporous Cost savings

proteins permeability, ment, membrane ion exchange
pH adjustment stack design membranes,
non-fouling
membranes
Desalting of Membrane fouling Membrane stack Nonfouling Cost savings

protein solutions design membranes
%
N
5
4 3
4
:
3 chemical industry 16
P
2
ste water treatment
a
3
Installed membrane area *
?
(& x 106)
z
a9 90 91 92 93 94 95
Year
Figure 8- 19. Estimated total membrane area installed in electrodialysis between

1989 and 1995 in various applications.
Electrodialysis 443
Table 8-8. Current and Future Applications for Electrodialysis, Their Relevance and Prospect of
Realization
-
Prospects for
Application Realization Importance Comments
Potable & Process Water Productios
Brackish water desalination Excellent IO Temperature (>6o”C) stable membranes
Seawater desalination Good 8 Low resistance. temperature

(>8o’C) stable membranes
Nitrate removal Good IO Ion-selective membranes
De-ionized water Very good 10 Development of conductive spacers
Brine concentration Good 5 ion-selective, low resistance membranes

(salt production)
Wastewater Treatment
Electroplating rinse water Good 8 Membranes with very high

(heavy metals removal) fixed charge density
Etch bath rinse water Excellent 8 Proton-selective, acid resistant

membranes
Pickling wastewater Good 8 Membranes with very high fixed

charge density
Desalination of special Good 6 Excellent chemical

effluent (e.g. glycerine) (alkaline) stability
Radioactive wastewater Good 6 Radiation-resistant membranes

treatment
Regeneration of ion-exchange Fair 8 Improved process design

resins
Food & Drua Industry Thermal & chemical stability

sterilizable membranes
Desalting of whey Excellent a Low lactose permeability membranes
Desalting of molasses Excellent 6 Low sugar permeability
Desalting of proteins Very good 5 Low-fouling membranes
Deacidification of fruit Good 5 Acid/sugar selective ion-exchange

juices membranes
Table 8-8. (continued)
Prospects for
Application Realization Importance Comments
Salt removal from fermentation Good 8

broths
Potassium removal from wine Good 5
Organic acids removal Good 6 Anion exchange

from fermentation broths membranes with high
acid permeability
Separation of amino acids Good 4
Donnan Dialvsis & Dialvsis
Water softening Good High permeability cation

exchange membranes
Heavy metal recovery Good
Acid recovery from etching Good

baths
Recovery of acids & bases Fair IO Chemically stable

from salts membranes;
better process design
Recovery of organic acids Very good 8
from salts
pH control w/o adding acid Very good IO

or base
8.8.3 Future Research Directions
To guarantee the future growth and expansion of electrodialysis in new

areas of application, continued research and development in membrane preparation
and process design will be necessary. In Table 8-9, desirable future research
topics, their prospects for realization and their technical and commercial
importance are listed. The prospects have been rated excellent, good and fair.
Importance for the future growth of electrodialysis has been rated between I and
IO, I being the lowest.
Electrodialysis 445
Table 8-9. Future Research Directions in Electrodialysis. Their Relevance and Prospect of
Realization
Prospects for
Research Topic Realization Importance Comments
Membranes with lower resistance GOOd 5 Membrane resistance affects energy

cost
Membranes with higher charge Good 5 Higher selectivity

density
Membranes with better Good 5 Must be thin, but withstand

mechanical properties pressure
Membranes with better Excellent 8 Temperature reduces electrical

temperature stability resistance
Better ion-selective Fair 10 Special interest for nitrate removal

membranes
Membranes with better chemical Excellent a Enable treatment of concentrated

(acid & alkali chlorine) stability streams
Fouling-resistant membranes Excellent 9 Longer useful life
Steam-sterilizable membranes Good 3 Better acceptance in food and drug

industry
Membranes with high perme- Good 5 Important for organic acid

ability for large anions separations
Membranes with low permeability Fair 6 Requirements for the drug industry
for neutral components
Better bipolar membranes Good 9
Thinner ceils Fair a Lower electrical resistance
Better flow distribution Good 10 Less concentration

(spacer design) polarization
Leak elimination Good 8 Better current utilization
Lower-cost stacks Good IO Lower investment cost

REFERENCES
I. E. Korngold, “Electrodialysis Membranes and Mass Transport,” in Svnthet&

MembraneProcesses. G. Belfort, (Ed.), Academic Press, New York (1984).
2. P. H. Prausnitz and J. Reitstotter, wohorese. Electroosmose.

Electrodialvsg. Steinkopff, Dresden (1931).
3. R. E. Lacey, (Ed.), Me brane Processes for In-, Southern Research

Institute, Birmingham, AF(l966).
4. L. H. Schaffer and M. S. Mintz, “Electrodialysis.” in Priocioles of

Desalination, K. S. Spiegler, (Ed.), Acacemic Press, New York (1966).
5. H. Strathmann, Trennu a on molekularen Mischuneen mit Hilfe svnthetischer

Membraneq, Steinkopffn-Vverlag, Darmstadt, (1979).
6. F. G. Donnan. Z. Elektrochem. 17, 572 (191 I ).
7. F. Helfferich, Ion ExchanPp, McGraw-Hill, London (1962).
8. R. E. Lacey, “Basis of Electra Membrane Processes,” in Industrial Processing

with Membranes, R. E. Lacey and S. Loeb, (Eds.), John Wiley & Sons, New
York (1972).
9. J. R. Wilson, Demineralization bv Electrodialvsiq, Butterworth, London (1960).
10. S. B. Tuwiner. Diffusion and Membrane Technolonv. Reinhold. New York

( 1962).
11. M. S. Mintz, Ind. Enn. Chem. 55, 19 (1963).
12. Product bulletin, Ionics Inc., Watertown, MA 02172, USA (1984).
13. H. Strathmann and H. Chmiei, “Die Elektrodialyse - ein Membranverfahren

mit Vielen Anwendungsmdglichkeiten,” Chem.-Inn.-Tech. 56, 214 (1984).
14. Product bulletin, Tokuyama Soda, Tokyo (1987).
15. E. W. Lang and E. L. Huffmann, U.S. Off. Saline Water Res. Dev. Reo. 439
(1969).
16. R. E. Lacey and E. W. Lang, U.S. Off. Saline Water Res. Dev. Reo. 106
( 1964).
17. R. E. Lacey and E. W. Lang, U.S. Off. Saline Water Res. Dev. Rep. 398
(1969).
18. W. E. Katz, “The Electrodialysis Reversal (“EDR”) Process,” presented at the

International Congress on Desalination and Water Re-Use, Tokyo, Japan
(1977).
Electrodialysis 447
19. R. E. Lacey, U.S. Off. Saline Water Res. Dev. Reo. 8Q (1963).
20. A. L. Goldstein et al., U.S. Patent 4,608,140, August 26 (1986).
21. H. Schmoldt, H. Strathmann, U.S. Patent 4.737.260, April 2 (1988).
22. A. Clad et al., U.S. Pat. 4,608,147, August 26 (1986).
23. N. Lakshminarayanaiah, Chem. Revs. 65, 494 (1965).
24. D. S. Flett, Ion Exchange Membrana. E. Horwood Ltd. Chichester U.K.

(1983)
25. T. Sata et al., U.S. Pat. 3,647,086, March 7 (1972).
26. T. Sata, K. Kuzumoto and Y. Mizutani, Polvmer J. 8, 225-226 (1976).
27. A. Eisenberg and H. L. Yeager, “Perfluorinated Inomer Membranes,” a

Svmposium Series I8Q, (1982).
28. P. Zschocke, D. Quellmalz, “Novel Ion-Exchange Membranes Based on an

Aromatic Polyethersulfone,” J. Membrane Sci. 22, 325-332 (1985).
29. W. Pauli. Biochem. Z. 152, 355 (1924).
30. E. Manegold and C. Kalauch, Kolloid-Z. 86, 93 (1939).
31. K. H. Meyer and W. Strauss, Helv. Chim. Acta. 21, 795 (1940).
32. J. G. Kirkwood, ,I. Chem. Phvs. 9, 878 (1941).
33. E. A. Murphy, F.V. Paton, and J. Ansell, U.S. Patent 2.331.494, October 12
(1943).
34. Th. C. Bissot et al., U.S. Patent 4,437,951, March 20 (1984).
35. R. B. Hodgdon et al.. U.S. Patent 4,504,797, March 19 (1985).
36. M. J. Covitch et al., U.S. Patent 4,568,441, February 4 (1986).
37. H. Puetter et al., U.S. Patent 4.585536, April 29 (1986).
38. H.-J. Sterzel et al., U.S. Patent 4.711.907, December 8 (1987).
39. J. E. Kuder et al., U.S. Patent 4,634,530, January 6 (1987).
40. K. B. Wagener et al., U.S. Patent 4,652,396, March 24 (1987).
41. S. Toyoshi et al., U.S. Patent 4.71 1,722, December 8 (1987).
42. H.-J. Sterzel et al.. U.S. Patent 4,71 1,907, December 8 (1987).
43. P. R. Klinkowski et al., U.S. Patent 4.668.361, May 26 (1987).
44. A. J. Giuffrida et al., U.S. Patent 4.632.745, December 30 (1986).
45. K. J. Liu, F. P. Chlanda and K. J. Nagasubramanian, “Use of Bipolar

Membranes for Generation of Acid and Base - an Engineering and Economic
Analysis,” J. Membrane Sci.. 2, 109 (1977).
46. B. Bauer, F.-J. Gerner, H. Strathmann, “Development of Bipolar Membranes,”

Desalination 68, 279-292 (1988).
47. F. P. Chlanda at al., U.S. Patent 4,116,889, September 26 (1978).
48. 0. Kedem and Y. Maoz. “Ion conductive spacers for improved electrodialysis.”
natton 19. 465-470 (1976).
49. 0. Kedem, J. Kohen, A. Warshawsky and N. Kahana, “EDS - Sealed Cell

Electrodialysis.” Desalination 46, 291 (1983).
50. E. Korngold, K. Kock and H. Strathmann. “Electrodialysis in Advanced Waste

Water Treatment,” Desalination 24, 129 (1978).
51: H. Strathmann and K. Kock, “Effluent Free Electrodialytic Regeneration of

Ion-Exchange Resins,” in Polvmer Seoaration Media”, A. R. Cooper, (Ed.),
Plenum Press, New York (1982).
52. H. Strathmann, “Electrodialysis and its Application in the Chemical Process

Industry,” -aration & Purification 14, 41-66 (1985).
53. R. M. Ahlgren. “Electra Membrane Porcesses for Recovery of Constituents

from Pulping Liquors.” in Industrial Processinn with Membranes, R. E. Lacey
and S. Loeb, (Eds.), John Wiley & Sons, New York (1972).
54. H. P. Gregor and C. D. Gregor, “Synthetic-Membrane Technology,” Scientific

American. Vol 239, No. I, Page 88, July (1978).
55. Millipore Product Bulletin, Lit. No. Cl001 (1987).
56. F. H. Miller, Ionics, Inc. Lit. (1984).
57. Science Relations Service, Market study 5 SRS-Project 068: Electrodialysis

(1985).
9. Glossary of Symbols and Abbreviations
a selectivity
A Angstrom unit (lo-” meter)
AIST Agency of Industrial Science and Technology
AN acrylonitrile
atm atmosphere
B separation factor
bbl barrel
Btu British thermal unit
C concentration
“C degrees Celsius
cf cubic foot
cm centimeter
d day
DARPA Defence Advanced Research Projects Agency
DDS De Danske Sukkerfabriker
DEA diethylamine
DOD Department of Defense
DOE Department of Energy
DuPont E.I. duPont de Nemours & Co. (Inc.)
ED electrodialysis
EDC ethylene dichloride
EDTA ethylene diamine tetra acetic acid
EEC European Economic Community
ELM emulsion liquid membranes
EOR enhanced oil recovery
EPA Environmental Protection Agency
EPO, equivalent pure oxygen
EtOH ethanol
ETP Emerging Technologies Program
“F degrees Fahrenheit
g gram
6G Gibbs free energy change
gal gallon
GE General Electric Corp.
h hour
dH enthalpy change
H/C hydrogen-to-carbon ratio
HDS hydrodesulfurization
HF hollow fiber membranes
HFPC Hexafluorinated bisphenol-A polycarbonate
HFTMPC Hexafluorinated tetramethyl bisphenol-A polycarbonate
449
ILM immobilized liquid membrane

in inch
IX ion exchange
J Joule
‘K degrees Kelvin
L liter
lb pound
LIX liquid ion-exchange
meter
ZE multiple effect evaporation/distillation
METC Morgantown Energy Technology Center
MF microfiltration
MGD million gallons per day
MIT Massachusetts Institute of Technology
MIT1 Ministry of International Trade and Industry
MSF multi-stage flash distillation
MTBE methyl tertiary-butyl ether
MTR Membrane Technology and Research, Inc.
NAMS North American Membrane Society

NASA National Aeronautical and Space Administration
NED0 New Energy Development Organization
NISI- National Institute of Standards and Technology
NSF National Science Foundation
OEM original equipment manufacturer

OER Office of Energy Research
OFE Office of Fossil Energy
OPA Office of Program Analysis
osw Office of Saline Water
P osmotic pressure
P pressure
P permeability
Pa Pascal
PC bisphenol-A polycarbonate
PC1 Patterson Candy International
PETC Pittsburgh Energy Technology Center
PF plate and frame membrane modules
ppm parts per million
PPO polyphenylene oxide
PSA pressure swing adsorption
psi pounds per square inch
psig pounds per square inch gauge
PTMSP polytrimethyi-silylpropyne
PVA polyvinyl alcohol
Glossary of Symbols and Abbreviations 451
60 enthalpy change
quad 10” Btu
R salt rejection
R&D research and development
RO reverse osmosis
s second
AS entropy change
SARA Superfund Amendments Reauthorization Act
SBIR Small Business Innovative Research Program
scf standard cubic foot
scfd standard cubic feet per day
scfm standard cubic feet per minute
SD1 silt density index
SITE Superfund Innovative Technologies Evaluation program
STP Standard temperature and pressure
SW spiral wound membrane modules
T temperature
TDS total dissolved solids
TMPC tetramethyl bisphenol-A polycarbonate
TOC total organic carbon
-l-PA tissue plasminogen activator
tpd tons per day
UF ultrafiltration
UOP Union Oil Products
vc vapor compression
VOC volatile organic chemicals
VSA vacuum swing adsorption
VW Volkswagen, Inc.
wow0
SW, AW net work
WRPC Water Re-use Promotion Center
wt weight
year
PREFIXES
k kilo ( 10s)
m milli (10-s)
MM Mega (10’)
I, micro (lo-*)
n nano ( 1OmQ)

Chemistry - Membrane Separation Systems

Uploaded by

Document Informationclick to expand document informationbook

Document Informationclick to expand document information

Copyright:

Available Formats

Chemistry - Membrane Separation Systems

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chemistry - Membrane Separation Systems

Uploaded by

Copyright:

Available Formats

MEMBRANE

R.W. Baker, E.L. Cussler, W. Eykamp,

NOYES DATA CORPORATION

Published in the United States of America by

Library of Congress Cataloging-in-Publication Data

Membrane separation systems : recent developments and future

Dr. Richard W. Baker (Membrane Technology & Research, Inc.)

The following members of the Department of Energy (DOE) made valuable

Dr. Richard Gordon (Office of Energy Research, Division of Chemical Sciences)

The following individuals, among others, contributed to the discussions and

Dr. B. Bikson (Innovative Membrane Systems/Union Carbide Corp.)

The following individuals served as peer reviewers of the final report:

Dr. J. L. Anderson (Carnegie Mellon University)

Additional information on the current Federal Government support of membrane

PJational Once Foundation

Membrane based separation technology, a relative newcomer on the separation

The book pays particular attention to identifying currently emerging innova-

The information in the book is from Membrane Separation Systems, Volume I-

The table of contents is organized in such a way as to serve as a subject index

Advanced composition and production methods developed by

The studies in this book were sponsored by the U.S.

Mention of trade names, commercial products, and

1. EXECUTIVE SUMMARY. ................................ .4

2. ASSESSMENT METHODOLOGY ........................... .9

3. INTRODUCTION. .................................... .19

4. GOVERNMENT SUPPORT OF MEMBRANE RESEARCH .......... .34

4.2.1.1 Office of Industrial Programs/Industrial

5. ANALYSIS OF RESEARCH ........................

APPENDIX A. PEER REVIEWERS’ COMMENTS ................. .86

A.1.4 Comparison with Japan ....................... .89

INTRODUCTION TO VOLUME II. ........................... .96

1. MEMBRANE AND MODULE PREPARATION. ................. 100

1.6 Membrane Modules ................................ 136

2. PERVAPORATION. .................................. .151

3. GAS SEPARATION. .................................. .I89

3.6.6 System Design and Operating Trends. .............. 206

4. FACILITATED TRANSPORT. ........................... .242

4.4.2.2 Acid Gases. ......................... 258

5.6.5 Membrane Fouling: Bacterial Adhesion to Membrane

6. MICROFILTRATION ................................. .329

7. ULTRAFILTRATION. ................................ .360

7.1.5 Flux Enhancement ......................... .369

8. ELECTRODIALYSIS. ................................. .396

8.3.1 Desalination of Brackish Water

9. GLOSSARY OF SYMBOLS AND ABBREVIATIONS ............ .449

The Office of Program Analysis in the Office of Energy Research of the

The rationale for seeking innovative, energy-saving technologies is,

Some membranes are used in circumstances where energy saving is an

saving. Related technologies, for example the membrane catalytic reactor,

This report was prepared by a group of six membrane experts representing

Three topics relating to the development of gas-separation membranes ranked

might be achieved by membrane-based gas-separation technology are exemplified

In general, facilitated-transport related topics scored low in the master

The principal problem in ultrafiltration technology is membrane fouling. The

Two high-priority topics cover research opportunities in the microfiltration

Reverse osmosis, ultrafiltration and microfiltration are all technologies with

Many of the top 10 ranked priority research topics spotlighted technology

During the course of the study, government support of membrane-related