Dechow - Separation and Purification Techniques in Biotechnology

SEPARATION AND
PURIFICATION
TECHNIQUES
IN BIOTECHNOLOGY
Frederick J. Dechow
Reed & Carnrick Pharmaceuticals
Piscataway, New Jersey
I-J
nP
NOYES PUBLICATIONS
Park Ridge, New Jersey, U.S.A.
Copyright @I 1989 by Frederick J. Dechow
No part of this book may be reproduced or utilized in
any form or by any means, electronic or mechanical,
including photocopying, recording or by any informa-
tion storage and retrieval system, without permission
in writing from the Publisher.
Library of Congress Catalog Card Number: 88-34502
KEN: o-8155-1197-3
Printed in the United States
Published in the United States of America by

Noyes Publications
Mill Road, Park Ridge, New Jersey 07656
10987654321
Library of Congress Cataloging-in-Publication Data
Dechow, Frederick J.
Separation and purification techniques in
biotechnology.
Includes bibliographies and index.

1. Biomolecules--Separation. 2. Biomolecules--
Purification. 3. Biotechnology--Technique. I. Title.
TP248.25.S47D43 1989 660.2’842 88.34502
ISBN 0-8155-l 197-3
Preface
Sorptive separation techniques are found in almost every product separation

or purification scheme treating fermentation or biochemical feedstreams. These
sorptive techniques include adsorption, ion exchange and liquid chromatography
on solid supports. The major objective of this book is to place these different
methods in perspective, relative to each other, so that selection of the appropri-
ate technique or combination of techniques may be readily made. While the
emphasis has been placed on laboratory evaluation techniques, the scale-up of
these techniques and their industrial applications; it is hoped that sufficient
theory has been provided so that the interested reader may use the references
to pursue the selectivity and kinetic considerations for sorptive procedures.
The first chapter provides a brief sketch of the nature of the biochemical
feedstream and all the processes which might be involved in isolating the de-
sired products from that feedstream.
Chapter 2 covers adsorptive separation, which is the oldest of the sorptive

techniques. While many people now regard adsorption as strictly for the removal
of unwanted impurities or color bodies, this chapter shows that there are many
applications where adsorption is useful in isolating biochemical products. The
theory of column processes developed in this chapter is also applicable to the
column operations for the sorptive processes described in the other chapters.
Ion exchange procedures are described in Chapter 3. This chapter builds

upon what was presented in Chapter 2 by demonstrating the effects of the
additional sorptive specificity associated with the exchange of ions. Operating
parameters and equipment developed for water treatment and metal recovery
applications have also been included since fermentation broths have character-
istics which may benefit from the use of resin-in-pulp, fluidized bed and the
other procedures presented.
V
vi Preface
Chapter 4, Column Chromatography Processes, covers the use of sorptive

materials to create an environment that allows the separate recovery of two or
more solutes. The biospecific recognition of a solute for a ligand attached to
the column material is covered in the last chapter on Affinity Chromatography.
It is my hope that this book will serve as a useful guide to the solution of
the practical problems associated with separating and purifying fermentation
and biotechnology products.
I would like to acknowledge the helpful suggestions of Henry C. Vogel,

the patience of George Narita and the support and understanding of my wife,
Joan Dechow. I also wish to recognize the assistance of Audrey Wildeck for
secretarial support and of Bart Alazio for preparing the illustrations.
April 1989 Frederick J. Dechow
NOTICE
To the best of the Publisher’s knowledge the informa-

tion contained in this publication is accurate; however,
the Publisher assumes no liability for errors or any con-
sequences arising from the use of the information con-
tained herein. Final determination of the suitability of
any information, procedure, or product for use con-
templated by any user, and the manner of that use, is
the sole responsibility of the user.
Mention of trade names or commercial products does

not constitute endorsement or recommendation for use
by the Publisher.
The book is intended for information only. The

reader is warned that caution must always be exercised
when dealing with hazardous, or potentially hazardous,
materials, and expert advice should be obtained at all
times when implementation is being considered.
Contents
1. INTRODUCTION. ..................................... .l
1.1 Fermentation Broth. ................................. .l
1.2 Recovery Unit Operations. ............................. .4
1.2.1 Mechanical Operations ............................ .6
1.2.1.1 Filtration. .............................. .6
1.2.1.2 Centrifugation ........................... .9
1.2.1.3 Evaporation. ........................... .16
1.2.1.4 Crystallization .......................... .22
1.2.1.5 Drying. ............................... .29
1.2.2 Membrane Processes. ............................ .34
1.2.3 Solvent Extraction, ............................. .42
1.2.4 Electrophoresis and Electrodialysis ................... .47
1.3 Recovery Processes ................................. .54
1.4 References. ...................................... .60
2. ADSORPTION. ...................................... .65

2.1 Introduction. ..................................... .65
2.2 Adsorption Theory ................................. .66
2.2.1 Adsorption Isotherm ............................ .68
2.2.2 Adsorption Kinetics. ............................ .77
2.3 Types of Adsorbents ................................ .95
2.3.1 Carbon and Activated Charcoal ..................... .99
2.3.2 Silica Gels, AIuminas and Zeolites. ................... 102
2.3.3 Organic Polymer Adsorbents ....................... 105
2.4 Laboratory Evaluation of Adsorbents ..................... 109
2.5 Adsorbent Operations ............................... 114
2.5.1 Pretreatment Operations. ......................... 114
2.5.2 Batch Operations. .............................. 116
2.5.3 Column Operations ............................ .117
vii
viii Contents
2.5.4 Regeneration Operations. ......................... 125

2.6 Process Considerations. .............................. 131
2.6.1 Design Factors ................................ 131
2.6.2 Pressure Drop ................................ .134
2.6.3 Computer Design. ............................. .136
2.7 Adsorption Applications. ............................ .136
2.7.1 Decolorization ................................ 136
2.7.2 Antibiotics. .................................. 141
2.7.3 Enzymes ................................... .144
2.7.4 Microorganism Adsorption ........................ 146
2.7.5 Organic Acids and Amines. ........................ 147
2.7.6 Vitamin Adsorption. ............................ 150
2.7.7 Miscellaneous Product Purifications .................. 152
2.7.8 Ethanol Adsorption. ............................ 153
2.7.9 Adsorption as a Fermentation Aid ................... 155
2.8 References. ..................................... .156
3. ION EXCHANGE .................................... .163

3.1 Introduction. .................................... .163
3.2 Theory ........................................ .164
3.2.1 Selectivity .................................. .166
3.2.2 Kinetics .................................... .179
3.3 Ion Exchange Materials and Their Properties. ................ 193
3.3.1 Ion Exchange Matrix ............................ 194
3.3.2 Functional Groups. ............................. 198
3.3.3 Porosity and Surface Area. ........................ 201
3.3.4 Particle Density. ............................... 209
3.3.5 Particle Size .................................. 210
3.4 Laboratory Evaluation Resin. .......................... 211
3.5 Ion Exchange Processes .............................. 220
3.5.1 Process Categories .............................. 220
3.5.1.1 Demineraiization. ........................ 220
3.5.1.2 Conversion. ........................... .220
3.5.1.3 Purification ............................ 221
3.5.1.4 Concentration. .......................... 221
3.5.2 Purification Procedures. .......................... 221
3.5.2.1 Purification of Strong Bases. ................. 221
3.5.2.2 Purification of Strong Acids ................. 222
3.5.2.3 Purification of Weak Bases .................. 223
3.5.2.4 Purification of Weak Acids .................. 224
3.5.3 Biopolymer-Resin Interactions. ..................... 225
3.6 Ion Exchange Operations ............................ .228
3.6.1 Pretreatment ................................ .229
3.6.2 Batch Operations. ............................. .229
3.6.3 Column Operations ............................. 230
3,6,3.1 Fixed Bed Columns ....................... 230
3.6.3.2 Continuous Column Operations ............... 234
3.6.3.3 Fiuidized Column Operations ................ 236
Contents ix
3.6.4 Elution/Regeneration. ........................... 244

3.6.5 Backwashing. ................................ .246
3.7 Process Considerations. .............................. 248
3.7.1 Design Factors ................................ 248
3.7.1.1 Scaling-Up Fixed Bed Operations .............. 248
3.7.1.2 Comparison of Packed and Fluidized Beds ........ 253
3.7.1.3 Pressure Drop. .......................... 255
3.7.2 Computer Calculations. .......................... 256
3.7.3 Economic Analysis ............................. 260
3.7.4 Ion Exchange Resin Limitations. .................... 262
3.8 Biotechnology Applications. ........................... 267
3.8.1 Amino Acid Purification. ......................... 268
3.8.2 Antibiotics. .................................. 274
3.8.3 Pharmaceuticals and Organic Acids. .................. 286
3.8.4 Protein Purification ............................. 288
3.8.5 Sugar Stream Processing .......................... 295
3.8.5.1 Decolorization .......................... 295
3.8.5.2 Demineralization. ........................ 298
3.8.6 Vitamins ................................... .303
3.8.7 Wine Treatment .............................. .305
3.8.8 Enzyme Immobilization .......................... 307
3.8.9 Catalytic Applications ........................... 310
Appendix. ......................................... .312
3.9 References. ..................................... .320
4. COLUMN CHROMATOGRAPHY PROCESSES ................ .333

4.1 Introduction. .................................... .333
4.2 Classifications of Chromatography ....................... 335
4.2.1 Chromatographic Methods ....................... .335
4.2.2 Types of Chromatographic Separations ................ 337
4.2.3 Chromatographic Processing Techniques ............... 342
4.3 Chromatographic Materials ............................ 343
4.4 Theory ........................................ .348
4.4.1 Theoretical Plate Height. ......................... 350
4.4.2 Zone Spreading. .............................. .352
4.4.3 Resolution. .................................. 353
4.5 Laboratory Techniques. ............................. .366
4.6 Scale-Up Procedures. ................................ 370
4.7 Industrial Systems. ................................ .376
4.8 Chromatography Applications. ........................ .385
4.8.1 Amino Acid Separation .......................... 386
4.8.2 Antibiotics. .................................. 388
4.8.3 Glucose-Fructose Separation ....................... 388
4.8.4 Glycerol Purification ............................ 390
4.8.5 Oligosaccharide Removal ......................... 391
4.8.6 Peptides and Proteins. ........................... 392
4.8.7 Polyhydric Alcohol Separation. ..................... 400
4.8.8 Regenerant Recovery. .......................... .400
x Contents
4.8.9 Sugar from Molasses. ............................ 402

4.9 References. ..................................... .405
5. AFFINITY CHROMATOGRAPHY. ........................ .416

5.1 Introduction. .................................... .416
5.2 Theory of Affinity Chromatography. ..................... 418
5.2.1 Equilibrium and Binding Selectivity ................. .419
5.2.2 Kinetics of Affinity Purification. ................... .424
5.3 Affinity Chromatography Materials. ..................... .430
5.3.1 Supports. .................................. .430
5.3.1.1 Agarose. ............................. .431
5.3.1.2 Polyacrylamide Gels. ...................... 435
5.3.1.3 Controlled Pore Glass. ..................... 436
5.3.2 Spacers. ................................... .438
5.3.3 Llgands .................................... .439
5.4 Laboratory Practices ............................... .443
5.4.1 Selecting a Ligand ............................. .443
5.4.2 Preparing the Support-Ligand ..................... .446
5.4.3 Ligate Adsorption ............................. .449
5.4.4 Ligate Elution. ............................... .450
5.5 Scale-Up of Laboratory Procedures. ..................... ,453
5.6 Affinity Chromatography Operations .................... .456
5.6.1 Batch Operations. ............................. .456
5.6.2 Column Operations ............................ .458
5.6.3 Affinity Purification with Membrane Operations. ........ .462
5.6.4 Ligand Costs. ................................ .466
5.7 Affinity Chromatography Applications .................... 467
5.7.1 Protein Purification ............................ .467
5.7.2 Interferon Purification. ......................... .472
5.7.3 Receptor-Binding Affinity Chromatography. ........... .473
5.7.4 Immuno Affinity Chromatography. ................. .476
5.7.5 Cell Separation by Affinity Chromatography ........... .476
5.8 References. ..................................... .478
INDEX..............................................485
1
Introduction
The recovery of products from fermentation or

biochemical processes has been cited (1) as the last hurdle to
be overcome in bringing biotechnology from the laboratory to
commercial status. This text will attempt to describe how
adsorptive materials, ion exchange resins, column
chromatography and affinity chromatography can be utilized
in these recovery and purification operations. This chapter
will examine the nature of the fermentation broth and will
serve to put these recovery operations in perspective with
other purification techniques not covered by this text. It is
essential to understand the relative advantages of each and
their interrelationships since most purifications will require
combinations of different techniques.
1 .l FERMENTATION BROTH
The fermentation broth is the combination of insoluble,

gelatinous biomass, the nutrient fluid, and the soluble
metabolites resulting from the fermentation operation. When
the fermentation is carried out without any form of inert
support for the biomass, the limit of fluidity for stirring or
aeration is approximately 3 to 7% WV dry weight of biomass.
Physically, biomass is a compressible, gelatinous solid with
surface layers of polysaccharide material which make it
cohere and adhere. Downstream processing, therefore, has to
deal with a viscous, highly non-Newtonian slurry as its
feedstock.
1
2 Separation and Purification Techniques in Biotechnology
For example, a bacterial fermentation for single cell

protein will produce a broth of 3% WV in which the slurry is
about 60% (by volume) wet biomass and 40% interparticle
fluid. When biomass supports are used, somewhat higher
operating biomass concentrations are possible and, in waste
disposal fermentations, the concentration has been raised
from 2 to 5 g/liter to 10 to 40 g/liter.
Compared with the feed streams to recovery processes

in conventional chemical processing, fermentation broths are
very dilute aqueous systems (see Table 1.1). Therefore, it
will be particularly important to avoid energy intensive
thermal operations and to select processes which give large
concentration increases in the first stage or stages.
Table 1.1: Typical Product Concentrations Leaving Fermenters
Product Grams per Liter

Antibiotics (e.g., Penicillin G) 10-30
Enzyme protein (serum protease) 2-5
Ethanol 70-120
Lipids 10-30
Organic acids (citric, lactic) 40-100
Riboflavin 10-15
Vitamin B12 0.02
The fluid volume in microbiological processes must be

reduced by at least an order of magnitude between the broth
and the final fluid stages of the recovery processes and, in
some cases, by very much more. For vitamin B- 12 the
reduction ratio is over 1OOO:l. Consequently, the plant
design and the range of acceptable unit operations changes
significantly as the fluid progresses from the broth handling
stages to the final isolation stage.
Many fermentation broths are unstable. Once any broth

leaves the controlled, aseptic environment of the fermenter,
it is exposed to a drastic change of conditions. An actively
growing biomass from an aerated culture will be suddenly
deprived of oxygen and will experience a rapidly falling
concentration of nutrients. This frequently produces rapid
changes in physical properties, leading to destruction of
desired product. Lipids may be consumed as an alternative
energy source for continuing metabolic activity. Enzymes
may be destroyed by proteases released from the
deteriorating cells. The broth also becomes susceptible to
Introduction 3
contamination from foreign organisms which can have the

same effects.
Similar problems can occur if the recovery operations of

a batch fermentation are delayed. The problems can be
reduced by chilling to around 5OC. This is commonly done
for enzymes and other relatively small output processes.
However, this is to be avoided, if possible, with larger fluid
volumes since chilling from a typical fermentation
temperature of 35OC to 5OC requires refrigeration energy of
about 40 kWh/m3 of broth and considerable capital
expenditure. The time for appreciable product loss to occur
can be as little as 20 minutes at fermentation temperatures.
The necessary and practical recovery operations

employed and the order in which they are used in
downstream processing can be reduced to deceptively simple
looking recovery sequences. Figure 1.1 (2) shows a schematic
of the recovery sequence and the techniques associated with
each process.
Producl Product in broth
in CIIII. or .queous phase
Sedimentation
Solvent extraction
Filtration from whole cdls Ion exchanpe
solvsnf/d~lrrgent Adsorption
rend&-q membranes, Gel filfrstion
rtc, “leaky’! Affinity method9
Distillation
Membranes
Electrophoresir
Differential freezing
Evaporation
Membrsner
Pmcipitalion
Adsorption. ion exchange, clf%ily. erc
Frocra.ThDwing-cryslalliralion
Figure 1.1. Schematic of the processes which may be

involved in the separation and fractionation of fermentation
products (Reference 2).
If the desired product is extracellular, it is only

necessary to filter the biomass from the broth and isolate
the product from the fluid. If a product is intracellular, a
cell disruption step must first be employed. If the product is
also water soluble, this disruption should be performed while
the biomass is still in a slurry form. The chemical stability
and solubility of the product will dictate the most suitable
recovery techniques. Most microbiological products have
limited chemical stability. This puts severe restrictions on
the temperatures, the reactants, and pH levels which can be
used.
An important consideration in determining the

appropriateness of a recovery technique is the actual purity
requirement for the product. Many products, such as
enzymes and vaccines, will not need to be isolated as pure
compounds. An appropriate product in these cases would be
a complex mixture having the desired properties. However,
removal of specific materials, such as pyrogens in injectable
medicinals, would be necessary.
The major recovery difficulties arise when it is

necessary to separate specific compounds from other
chemically and physically similar materials. The isolation of
enhanced purity enzyme protein from other protein requires
highly specific physico-chemical effects to be used. Today it
is the exception rather than the norm for the separated
compounds to be that similar. Normally there is a range of
alternative recovery techniques and a selection can be made
on the basis of cost, familiarity, and reliability criteria.
1.2 RECOVERY UNIT OPERATIONS
The various methods for treating the fermentation broth

can be divided into mechanical or chemical unit operations.
Table 1.2 lists the processes normally included in each
category.
Filtration and centrifugation are unit operations in

which the suspended solids are separated from the fluid
phase. Drying is the removal of moisture or solvent from
solid particles, while evaporation is the removal of moisture
or solvent from a solution. In crystallization, the conditions
of a solution are adjusted to change the solubility of one of
Introduction 5
Table 1.2: Separation and Recovery Techniques
Mechanical Chemical
Filtration Adsorption
Centrifugation Ion exchange
Evaporation Column chromatography
Crystallization Affinity chromatography
Drying Solvent extraction
Reverse osmosis Electrophoresis
Ultrafiltration Electrodialysis
the dissolved compounds so that it leaves the solution as a

solid.
Microfiltration, ultrafiltration, and reverse osmosis are

membrane processes in which separation is based on
differences in ability to flow through a thin barrier that
separates two fluids. Microfiltration is a hydraulically driven
process using a membrane with a pore size in the 100 to
3000 A range. For ultrafiltration, the pore size is from 10
to 125 A, while for reverse osmosis, the pore size is from 3
to lo A.
Adsorption, ion exchange, column chromatography, and

affinity chromatography can be grouped as recovery
techniques in which the removed compound or solute
establishes an equilibrium between sites on a solid phase
material and the solution. In adsorption, the removed species
is bonded to the solid phase material by polarity or weak
chemical bonds. Ion exchange recovers material by the
interchange of ions between the liquid and solid phases.
Column chromatography may use adsorptive, ion exchange or
molecular sieve materials to separate solutes which are first
loaded onto a column of the separation material and then
eluted in such a manner that the individual solutes are
collected in separate fractions. In affinity chromatography,
the removed species is bound with a high level of selectivity
to ligands covalently attached to a solid matrix.
In solvent extraction, the removed compound establishes

an equilibrium distribution between immiscible solvents,
usually water and an organic liquid.
Electrophoresis and electrodialysis are separation

techniques that separate charged molecules or ions using an
electric field. Electrophoresis separates charged components

by accentuating small differences in ionic mobility in an
electric field using a moving carrier fluid. Electrodialysis
concentrates components on the basis of electromigration
through ionic membranes.
1.2.1 Mechanical Operations

1.2.1.1 Filtration: Filtration is typically the first step
in the isolation of any product from the fermentation broth.
This process separates the biomass cells, the cell debris, and
any precipitates from the broth fluid. The mathematical
representation of the incremental time dt to filter an
additional incremental volume dV after a volume V has been
filtered is given by:
dt/dV = a t bV (1.1)
The right side of the equation contains two components,
a and b. The a term is r)r,/AP and the b term, which
depends on V, is equal to r]r,/A2PW, where n is the liquid
viscosity, rs is the specific cake resistance of the filter
material, A is the filter area, P is the constant applied
pressure difference, rC is the specific cake resistance, and W
is the cake dry weight per unit volume of filtrate. Normally
the resistance of the filter material term includes the
resistance contribution of any filtration aid, pipes, and
valves.
According to this equation, the resistance to filtration

is due initially to only the constant term a. In theory, as
filtration proceeds and the biomass cake becomes thicker, the
resistance would be expected to increase linearly according to
the b term with this dependence on V. Although many
practical considerations must be taken into account when
applying this equation, its simplicity, the complexity of a
more exact description, and the uniqueness of many industrial
applications result in Equation 1.l being the most useful
filtration representation.
The specific practical limitations that must be

considered are the blockage or blinding of the filter, the
compressibility of the biomass cake, and the variable pore
structure of the cake. Blinding of the filter may be
prevented by starting the filtration at a low hydraulic
pressure by partially by-passing the pump. This will avoid
Introduction 7
driving the first solids support.

into the The biomass
filter
cake’s compressibility will usually be proportional to the
applied hydraulic pressure up to a certain pressure. Beyond
that pressure, the cake will collapse to a new compressed
form so that throughput is reduced with the incremental
pressure increase.
Table 1.3 shows filtration design and operation for

different fermentation broths. In two of the cases noted in
Table 1.3 a precoat was used. The filter will often be
precoated with a filtration aid such as diatomaceous earth to
reduce blinding of the filter and to increase filtration rates.
The filtration aid might be added to the broth but then the
quantity of filtration aid required is more than double the
precoat amount (3).
Table 1.3: Representative Design and Operating Results for

Fermentation Broth (Vacuum 0.68-0.85 Bar)
Bacillus Streptomyces Penicillium
licheniformis ery threus chrysogenum
filter type Vacuum precoat Vacuum drum Vacuum drum or
precoat
Design filtration
rate (9_/hr-m2) 160-320 400 1,400-1,800
Solids in slurry
(wt %) 8 25 2-8
Cycle time
(min/rev) 0.5 3 3
Filter medium Precoat Nylon Polypropylene
Cake discharge
mechanism Precoat String String or precoat
The equipment in this operation can be as simple as an

enlarged laboratory vacuum filter to more elaborate rotary
vacuum filters. These latter filters essentially consist of a
hollow segmented drum covered with a filter cloth. The
drum rotates in a bath of the broth to be filtered while a
vacuum inside the drum sucks liquid through the filter cloth,
forming a coating of solids on the outside of the drum.
Provisions are usually made to wash the filter cake during
filtration followed by removing the solid from the cloth.
Instead of vacuum, pressure can be used to drive the fluid
through a filter cake. Plate and frame pressure filters
consist of wire or perforated metal frames which act as the
mechanical support for the filter medium which can be fine
wire mesh, woven cloth or cellulosic pads. With deep frame

patterns, these filters can be used with diatomaceous pre-
coats.
Example 1 .l
Figure 1.2 shows the influence of pH on the filtration

rate of S. griseus broth, 1.5 centipoise, using 100 cm2 cotton
cloth filter, diatomaceous earth filter aid and a constant
pressure difference of 2 bar (4). What size of filter is
needed to filter 1000 L of fermentation broth at pH 3.8 in 15
minutes?
Figure 1.2. The filtration time (t) divided by the volume of

filtrate collected is plotted against the volume of filtrate
collected for different pH values (Reference 4).
For the sake of simplicity, it is assumed that the filter

:ake is incompressible so that Equation 1.1 can be integrated
10 give:
t nrS 1 arc V 1
-- + (1.2)
v = P ( A) 2PW ( x2, 1
Introduction 9
The slope and intercept for the pH 3.8 curve can be used to
obtain the values for rs and r,/W from the information
given. Using the data points (300, 0.7) and (600, 1.82), rs =
-55.73 and r,/W = 99.55.
Equation 1.2 is a quadratic equation in (l/A), so that:
1 _(y t &gyy
(A ) = nrcV
(1.3)
PW
With V = 1000 L and t = 900 set, this equations gives l/A =

5.5 x 10w6, or A = 18.18 m2,
The slight curvature of the experimental lines in Figure

1.2 indicates that the cake has some degree of
compressibility. This is to be expected for cakes formed
from fermentation cells. As the volumes, and therefore
required times, for the same filter area increase, the
curvature will become more pronounced. While decreasing
the pH would speed up the filtration, one must be careful so
that the product is not adversely affected.
Several reviews and texts (3, 5-10) exist which should

be consulted for additional information on this integral part
of any fermentation broth recovery system. A list of
industrial filtration equipment suppliers is given in Table 1.4.
1.2.1.2 Centrifugation: Centrifugation, although widely

used for cell recovery, is not nearly as ubiquitous as
filtration. This process enhances the gravitational settling of
the suspended solids in the fermentation broth. The
mathematical representation for the centrifugation process is
given by:
18n . St2
cp = - (1.4)
2
d* (cc - PL) g w rV
In this equation Cp is the time required for complete

particle removal, d is the particle diameter, ps is the particle
density, pL is the fluid density, r7 is the fluid viscosity, s is
Table 1.4: Industrial Filtration Equipment Suppliers
Company Location
Allied Filtration Co. Kingsley, PA
Avery Filter Co. Westwood, NJ
Bird Machine Company South Walpole, MA
Carl C. Brimmekamp & Co. Stamford, CT
Denver Equipment Company Colorado Springs, CO
Dorr-Oliver, Inc. Stamford, CT
EIMCO (Div. of Envirotech) Salt Lake City, UT
B C Hoesch Industries Wharton, NJ
Inlay, Inc. Califon, NJ
Komline-Sanderson, Inc. Peapack, NJ
Lenser America, Inc. Lakewood, NJ
Peterson Filters, Inc. Salt Lake City, UT
R & R Filtration Systems Marietta, GA
Serfilco,‘Ltd. Glenview, IL
Sparkler Filters, Inc. Conroe, TX
D.R. Sperry & Co. North Aurora, IL
the thickness of the liquid layer in the centrifuge, w is the

angular velocity, r is the rotation radius, V is the total
volume of liquid in the centrifuge and g is the gravitational
constant.
The right-hand side of the equation is shown divided

into two terms. The first corresponds to the terminal
velocity of the particle with diameter d under the force of
gravity. This term is solely concerned with properties of the
fluid and suspended particles. The second component is
concerned only with fixed characteristics of the centrifuge
operating at a fixed speed, w. This term is normally written
as sigma, C, the equivalent area of sedimentation centrifuge.
If it can be assumed that the liquid layer is thin and that
there are no interactions between solid particles, then C
provides a basis for comparing different sizes and types of
centrifuges (11).
The equation points out that, with respect to the feed

material, centrifugation is enhanced by large particle
diameter, a large difference in density between particles and
the fluid, and low viscosity. Likewise, separation is favored
by centrifugation equipment which is operated at a high
angular speed, has a large centrifuge radius, and can hold a
large volume of liquid while at the same time has a thin
liquid sedimentation layer.
Introduction 11
Example 1.2
For a fermentation broth that is a dilute suspension of

cells in an essentially aqueous solution, the diameter of the
cells is 5 microns. They have a density of 1.001 g/cc. The
viscosity of the broth is 2 centipoise. If 50 ml of the broth
is placed in a centrifuge tube whose bottom in 10 cm from
the rotation axis, what is the time required for complete
separation of the cells from the liquid supernatant at 3000
rpm?
If one assumes that the fermentation broth contains 5%

(by volume) suspended cells and that the cross-sectional area
of the centrifuge is 5 cm2, the liquid layer through which
the cells must travel is a maximum of 9.5 cm. Therefore,
using Equation 1.4, the time for complete removal of the
suspended cells is 304 sec.
The difficulties associated with centrifugation become

apparent when the typical fermentation broth characteristics
are considered in the light of these centrifugation
enhancement factors. The biomass particles, typically, are of
very small size and of low density. The fluid can be highly
viscous and, depending on the amount of dissolved solids, of
density comparable to the biomass. There are physical
limitations on the equipment as well. Attempts to increase
capacity by increasing the centrifuge radius soon reach the
point where mechanical stress and safe operation place a
limit. The increase in bowl capacity and the continuous
flowing through of fluids restrict the safe angular velocity.
When solids are discharged during operation, the inherent
imbalance this causes will further restrict the maximum safe
angular velocity.
Within these limitations, several centrifuge equipment

designs have been developed or adapted for the fermentation
industry. The centrifuges utilized for fermentation product
recovery fall into two basic types, the perforated bowl or
basket type and the solid bowl high speed types. The basket
centrifuge (Figure 1.3a) normally is used with a filter bag of
nylon, terylene, or cotton. The feed is added continually.
When the bowl is filled with the biomass cake, fresh washing
liquid can be added to displace the residual broth fluid
retained in the cake. This has many of the characteristics
of filtration but has considerably faster throughput.
FEED
POROUS LINER
FILTER CAKE
PERFORATED FRAME
>
rl
EFFLUENT m EFFLUENT
FILTER CAKE
FEED
b
Figure 1.3. Centrifuges. (a) Perforated bowl or basket type
centrifuge. (b) Tubular bowl centrifuge. (c) Multi chamber
solid bowl centrifuge. (d) Nozzle disk centrifuge. (e) De-
sludger or intermittent discharge centrifuge. (f) Scroll
centrifuges.
Introduction 13
FEED
EFFLUENT CG
SOL IDS SOLIDS
Figure 1.3. (continued)

INTERMITTENT
FEED
EFFL

Introduction 15
Solid bowl centrifuges are available in five basic types.

The first type is called tubular bowl (Figure 1.3b) and
operates like a batch laboratory centrifuge. A continuous
feed of material passes through the machine. As soon as the
bowl is full of the biomass cake, the centrifuge is stopped,
stripped down, the cake removed, the machine cleaned and
the sequence repeated. Typical batch size is about 4 kg wet
weight. A variation on this type is the multichamber
centrifuge (Figure 1.3~) which allows three or more time the
capacity of the tubular bowl. However, the disassembly and
cleaning of this centrifuge is more difficult.
The third and fourth types operate continuously since

the solids are automatically removed from the bowl. The one
is called a nozzle disk centrifuge (Figure 1.3d). This
centrifuge discharges a continuous stream of concentrated
solid-slurry. The other disk centrifuge is termed a de-
sludger (Figure 1.3e). This centrifuge has an intermittent
discharge of solids with lower water content than the nozzle
centrifuge.
The final type is called a scroll centrifuge (Figure 1.3f).

This centrifuge uses a screw within a rotating bowl to allow
continuous removal of the solid and the liquid portions. This
type of centrifuge is recommended when the solids are so
coarse that they would blind the discharge systems of the
disk type of continuous centrifuges.
Table 1.5 shows the characteristics of the different

types of centrifuges. There is a specific particle size range,
maximum centrifugal force and fluid capacity for each type
of centrifuge. Reviews and texts on centrifugation (1 l-15)
can be used for additional information. A list of industrial
centrifugation equipment suppliers is given in Table 1.6.
Table 1.5 : Characteristics of Different Types of Centrifuges
Particle Size Centrifugal Force Capacity

Type (pm) (a’r/g) (Q/min)
Basket lo-8,000 1,800 300”
Tubular bowl 0.01-100 16,000 80”
Multichamber 0.01-100 6,000 250*
Disk, nozzle 0.1-100 9,000 4,000
De-sludger 0.1-100 6,000 1,000
Scroll 2-3,000 3,000 1,150
*With interruptions for removal of solids.
Table 1.6: Industrial Centrifuge Manufacturers
Company Location
Alfa Lava& Inc. Fort Lee, NJ
Amtek, Inc. El Cajon, CA
Baker Perkins, Inc. Saginaw, MI
Barrett Centrifugals Worcester, MA
Bird Machine Company South Walpole, MA
Centrico, Inc. Northvale, NJ
Clinton Centrifuge, Inc. Hatboro, PA
Commercial Filters Lebanon, IN
Heinkel Filtering Systems South Norwalk, CT
Krauss Filtering Systems, Inc. Charlotte, NC
IEC Division, Damon Corp. Needham, MA
Pennwalt Corp., Stokes Division Warminster, PA
Quality Solids Separation Co. Houston, TX
Robatel, Inc. Pittsfield, MA
Sanborn Associates, Inc. Wrentham, MA
Tema Systems, Inc. Cincinnati, OH
Western States Machine Company Hamilton, OH
1.2.1.3 Evaporation: Evaporation is the concentration of

a feed stream by the removal of the solvent through
vaporization of the solvent. This process is distinguished
from distillation in that the vapor components are not
separately collected in evaporation.
Many fermentation fluids are dilute aqueous streams

which require the evaporation of large amounts of water.
This requirement is complicated by many fermentation
products being damaged or destroyed if exposed to too great
a temperature over an extended period of time. Highly
efficient evaporators with short residence times are required
for evaporation of these heat sensitive products.
The efficiency of an evaporator is determined by its

ability to transfer heat, to separate liquid from vapor and its
energy consumption per unit of solvent evaporated.
Heat transfer rates can be treated with conventional

heat transfer calculations (16) if the evaporator has an
outside heating element where the solution or slurry is
pumped through the tubes. Under steady state conditions of
flow rate, temperature, pressure and composition, the energy
transferred across a heat exchanger or evaporator surface is
given by:
Introduction 17
Q = UAAT. (1.5)
In this equation, Q is the overall rate of heat transfer, U is
the overall heat transfer coefficient for the system, A is the
evaporator surface area and AT is the difference between the
steam temperature and the liquid temperature leaving the
vapor head for natural circulation equipment. For forced
circulation equipment, AT is replaced with the log mean
temperature difference to correct for the fact that the
temperature difference is not constant along the evaporator
surface. The overall heat transfer coefficient, U, is obtained
from pilot plant data or calculated from the individual heat
transfer coefficients, hi
l/U = C l/hi . (1.6)
These theoretical rates are greatly reduced by the
presence of air in other configurations such as multiple
effect evaporators. These venting problems make it difficult
to translate small scale evaporation equipment data to full
scale equipment.
The vapor-liquid separation portion of the evaporator is

designed to prevent entrainment or liquid droplets from being
carried through the evaporator system. Entrainment is a
function of the vapor velocity, temperature difference,
mechanical design and the physical properties (viscosity and
surface tension) of the fluid being concentrated.
The vapor head or flash chamber is that area of the

evaporator that provides a region for the liquid phase and
the vapor phase to separate. In natural circulation
evaporators, the boiling process causes liquid to move across
the heating surface. The two phase mixture of liquid and
vapor is less dense than the solution which pushes it forward
and upward to the flash chamber. More viscous liquids and
those with a high solids content require a centrifugal pump
to circulate them through a loop around the heating unit.
The energy consumed per unit of solvent evaporated is

best obtained by forming an energy balance and a mass
balance around each stage or effect of an evaporator (Figure
1.4). For a single stage evaporator, the material balance is
simply the splitting of the feed entering the evaporator into
the portion discharged as vapors (the distillate) and that
portion remaining (the concentrate). Entrainment is normally
ignored in material balance calculations. The energy balance
uses the temperature of the evaporator as the reference

temperature. The steam entering the heat exchanger supplies
the heat to increase the temperature of the feed stream, to
form crystals (if any) and to vaporize the evaporated solvent.
Radiation may require an additional 1 to 5% heat.
CONCENTFAIE. IP
l
UC)
FEED CONCENTRATE DISTILLATE
I:^
BOUNDARY CONDITIONS: 0.0 c mF
0.0 ( fC & I.0
Figure 1 A. Energy and material balance of a single

evaporator stage (Reference 18).
An evaporator can be something as simple as a jacketed

vessel, but is usually more complex such as horizontal tube
evaporators, thin film evaporators, falling-firm evaporators,
long- or short-tube vertical evaporators, propeller calandrias,
forced circulation evaporators, plate evaporators, flash
evaporators and multiple effect evaporators.
Long-tube vertical evaporators are the most widely used

evaporators today. This evaporator (Figure 1.5a) has a
vertical heat exchanger topped by a vapor head. Just above
the tubes is a deflector to act as a primary liquid-vapor
separator. These evaporators are the most economical large
scale evaporators provided the product neither is heat
sensitive nor forms a precipitate during the evaporation.
Introduction 19
FEED
I
VAPOR
-CONDENSATE
FE)ED PROliUCT
a b
STEAM
PRODUCT
Figure 1 S. Evaporators. (a) Long-tube vertical evaporator.

(b) Falling-film evaporator. (c) Propeller calandria
evaporator. (d) Plate evaporator (e) Thin film evaporator.
STEAM STEAM
PACERS
INT GASKETS
VAPOR AND -
CONCENTRATE DISCHARGE
TO SEPARATOR
BELT DRIVE
CARTRIDGE-TYPE MAIN BEARING

DOUBLE MECHANICAL S
MOTOR
VAPOR NOZZLE
FEED NOZZLE
ROTOR
HEATING JACKET
HEATING JACKET
A: HEATING NOZZLE
B: HEATING NOZZLE
HEATING JACKET
LOWER ROTOR
BOTTOM CONE
GUIDE BUSHING
DISCHARGE NOZZLE

Introduction 21
The falling-film evaporator (Figure 1Sb) is similar to

long tube vertical evaporators except in the location of the
heat exchanger. To operate properly a significant amount of
liquid must be circulated through the tubes. This design is
very useful for fermentation fluids that are heat sensitive
because of its very short liquid retention time.
A propeller calandria evaporator (Figure 1.5c) is

basically a short-tube vertical evaporator with a propeller in
the center of the column. The propeller keeps any solids
suspended in the liquid phase even if boiling stops. These
evaporators are useful for evaporating small volumes of liquid
from solutions either with or without suspended solids.
Plate evaporators (Figure 1.5d) use flat or corrugated

plates as the heat transfer surface. Plate units may be used
for systems that have potential scale problems. In some
designs, the flat surfaces serve alternately as the
fermentation fluid side and the steam side. Scale deposited
when in contact with the feed solution can be removed when
in contact with the steam condensate.
In thin film evaporators (Figure 15e), the treated fluid

is moved at high speed over a heated surface by a
mechanical wiper arm. The particular advantage of this type
of equipment is its ability to handle high viscosity fluids and
it requires only short retention times.
Multiple effect evaporators are based on the principle

that the heat given up by condensation in one stage can be
used to provide the reboiler heat for a different stage.
These large, complex devices are only justified when large
amounts of dilute aqueous fluids must be evaporated.
Example 1.3
Figure 1.6 (21) shows a comparison of heat transfer

coefficients for aqueous solutions and slightly more viscous
solutions. What are the relative surface areas required for
long tube vs. short tube evaporation at 80°C boiling
temperature for the aqueous solution when the AT is lO”C?
How does this change as the viscosity of the solution is
increased?
U&W JUICE EVAPORATORS

NDER CWARAGLE CONDIT
-38 60 a2 116
BOILING TEMPERATURE (‘0
Figure 1.6. Comparison of heat transfer coefficients

(Reference 2 1).
The graph shows that the heat transfer coefficient is

0.05877 and 0.04950 cal/(sec-cm2-“C) for long tube and short
tube vertical evaporators, respectively, in aqueous solutions.
The long tube requires 18.7% less surface area. When the
viscosity of the solution is increased, the heat transfer
coefficient for both types of evaporators is reduced.
However, the advantage in surface area for the long tube
vertical evaporator is reduced to 17.0%.
Review articles (18-21) about evaporators can be

consulted for additional information. Manufacturers are
listed in Table 1.7.
1.2.1.4 Crystallization: Crystallization separates a

material from a supersaturated solution by creating crystal
nuclei and growing these nuclei to the desired size. This
separation technique is applicable to those fermentation
Introduction 23
Table 1.7: Industrial Equipment Manufacturers
Company Location
Alfa-Lava1 Fort Lee, NJ
APV Company, Inc. Tonawanda, NY
Dedert Corp. Olympia Fields, IL
Distillation Engineering Co, Livingston, NJ
Goslin-Birmingham Birmingham, AL
HPD, Inc. Naperville, IL
Industrial Filter and Pump Cicero, IL
The Kontro Co., Inc. Orange, MA
Luwa Corporation Charlotte, NC
Paul Mueller Co. Springfield, MO
Niro Atomizer, Inc. Columbia, MD
Pfaudler Co. Rochester, NY
Swenson Process Equipment Harvey, IL
Unitech, Div. of Graver Co. Union, NJ
Henry Vogt Machine Company Louisville, KY
products that have a low solubility in the solvent utilized.

The separation is usually accomplished at low temperatures so
that there is the advantage of minimal product loss for
thermally sensitive materials.
The solution’s supersaturation, which precedes the actual

crystallization, is usually accomplished by the removal of
solvent or by lowering the solution temperature. The driving
force for the formation and growth of crystals is related to
the extent of supersaturation.
Phase equilibrium data are essential in the determination

of the most efficient method of creating the supersaturated
solution and the optimum degree of supersaturation. The
three most common methods of forming a supersaturated
solution are shown in Figure 1.7. Cooling may be
accomplished indirectly with build-in cooling surfaces if there
is no possibility of incrustation formation. When there is the
possibility of incrustation formation, vacuum cooling may be
used. This approach has other requirements on the solution,
such as a minimum amount of dissolved gases and a solution
boiling temperature near that of the solvent. Evaporation to
reach supersaturation differs from vacuum cooling in the
larger amount of solvent removed by the evaporator and in
the steady supply of heat to the evaporator. Should either
of these techniques prove inadequate, the crystalline product
may be salted out by the addition of chemicals.
NUCLEATION
TEMPERATURE (T)
Figure 1.7. Supersaturation to obtain crystal nucleation may

be obtained by: 1. cooling; 2. adiabatic evaporation; or 3.
isothermal evaporation.
The amount and particle size distribution of crystals

produced from a solution with a given crystallizer is a
function of the nucleation and crystal growth process. These
processes depend on a great number of factors, including the
hydrodynamics of the system, the presence of trace
impurities, temperature gradients and concentration gradients
in the equipment. Many mathematical models (22-24) have
been developed to describe the functioning of a crystallizer
for a supersaturated solution. A typical approach uses a
combination of equations describing the rate of formation and
the rate of growth of crystals which are then removed from
the system and the steady state supersaturation with the
introduction of fresh solution (25). A general review is
given by Nyvlt (26).
Introduction 25
Crystallizer equipment is usually grouped according to

the method of suspending the growing crystals. There are
four groups of crystallizers: circulating magma (Figure 1.8a),
circulating liquor (Figure 1.9), scraped surface (Figure 1.lO)
and tank crystallizer (Figure 1.l 1). It should be noted that
there have been lists of as many as sixty-eight different
types of crystallizers using other classification techniques
(26).
In the circulating magma type of crystallizer, the feed

enters the circulating pipe below the product discharge at a
point sufficiently below the free-liquid surface to prevent
flashing. The magma is the suspension of small crystals in
solvent. The crystal suspension and the feed are both
circulated through the heat exchanger where the temperature
is increased. The heated magma re-enters the crystallizer
near the liquid surface, raising the temperature locally to
cause vaporization of the solvent. The cooling which occurs
after vaporization causes the solution to become
supersaturated to the extend necessary for crystal nucleation
and growth. The product is continuously withdrawn.
A variation on this flow sequence is shown in Figure

1.8b. Here the feed enters the base of the crystallizer;
baffles and an agitator are used to cause the larger crystals
to be separated and removed continuously from the
crystallizer. The magma type of crystallizer produces quite
large (500 to 1800 microns) crystals.
The circulating liquor type of crystallizer retains the

crystals in the suspension vessel and only the solution is
circulated. The upflow solution keeps the crystals in
suspension and allows for the uniform growth of crystals.
The three distinct parts of these crystallizers are the crystal
suspension chamber where the crystal growth occurs, the
supersaturation chamber where the solution is adjusted to the
appropriate concentrations and temperature, and the
circulation pump system. These are usually the most
expensive type of crystallizer.
Scraped-surface crystallizers, such as the Swenson-

Walker crystallizer shown in Figure 1.10, are very important
in crystallizing a fermentation product from a high viscosity
solution. In this system, the supersaturation occurs by the
exchange of heat between the slurry and the coolant through
a jacket or double wall. An agitator is fitted with scrapers
to scrape the heat transfer surface to prevent the build-up

of solids which would reduce the efficiency of heat transfer
through the heat exchange surface. The main limitation on
capacity for these units is the amount of heat transfer
surface area.
NON-CONDENSABLE
GAS OUTLET
COOL I NG
WATER INLET --c(:
BAROMETRIC
CONDENSER
STEAM
JET
SWIRL BREAKER
CIRCULATING PIPE
CONDENSATE
Figure 1.8. (a) Magma forced-circulation (evaporative)

crystallizer. (b) Draft-tube-baffle crystallizer with circulating
magma.
Introduction 27
b
STEAM INJECTOR
‘VAPOR
CRYSTAL SUSPENSION
Figure 1.9. Vacuum-cooling circulating liquor crystallizer.

Figure 1.10. Swenson-Walker scraped surface crystallizer.
Tank crystallizers are used for smaller quantities of

solutions with solutes of normal solubilities. Although it is
possible to form the supersaturated solution in the tank
crystallizer without agitation, it is much more difficult to
control the crystal size distribution and the amount of
crystals produced. The agitated heat transfer coefficients
are 0.0027 to 0.027 cal/(s-cm2-“C).
FEED
- CRYSTALS
OUT
Figure 1 .ll. Tank crystallizer.

Introduction 29
Books and review articles on the subject should be

consulted for additional information, Table 1.8 shows
manufacturers of crystallization equipment who may be
helpful in providing assistance with crystallization of
fermentation products.
Table 1.8: Industrial Crystallizer Equipment Manufacturers
Company Location
APV CREPACO, Inc. Tonawanda, NY
Aqua-Chem, Inc. Milwaukee, WI
Blaw-Knox F & C E Co. Buffalo, NY
Dedert Corp. Olympia Falls, NY
Goslin-Birmingham Birmingham, AL
HPD Incorporated Nape&Be, IL
Niro Atomizer, Inc. Columbia, MD
Pfaudler Corp. Princeton, NJ
Swenson Division (Whiting Corp.) Harvey, IL
Unitech, Division of Graver Union, NJ
1.2.1.5 Drying: Drying is usually the last step in the

recovery of either the cell biomass or other fermentation
products. Drying is a difficult part of the recovery process
since high temperatures can cause inactivation of heat labile
materials while a certain amount of heat is needed to prevent
enzymatic autolysis or breakdown, The drying process can be
carried out in a non-adiabatic manner or in an adiabatic
manner. In non-adiabatic dryers, the heat of vaporization is
supplied through a heated wall to the wet solids. Adiabatic
driers pass hot gases through or across the wet solids to
vaporize the moisture from the solids.
For non-adiabatic drying, the rate of solvent removal is

given by the equation:
u.A
NV =
XV
‘TH- Ts) (1.7)
In this equation, U, is the overall heat transfer coefficient,

A is the utilized heat transfer surface area, XV is the latent
heat of vaporization, T, is the heating medium temperature
and Ts is the temperature of the heating surface in contact
with the solid being dried. This equation is only valid for an
instant in time since this is a non-steady state process. The
usefulness of this equation is that it allows comparison of Uo
for different types of dryers and, therefore, their

effectiveness for a specific product.
A similar equation represents the rate of solvent

removal for adiabatic drying:
NV = f Ko
(yw
- Yc) (1.8)
In this equation, f is the relative drying rate obtained from

psychometric charts, K, is a coefficient which depends upon
the geometric configuration of the dryer and the airflow
conditions (similar to U, in Equation 1.7), Yw is the
humidity of the air next to the wet solids and Y, is the
humidity of the bulk air. Psychometric charges are plots of
humidity versus enthalpy, with indications of wet bulb
temperature and dry bulb temperature. While typical charts
involve standard air-water vapor mixtures, separate charts
must be constructed for any other solvent-gas system being
studied.
The most common types of dryers are drum dryers,

vacuum dryers, fluidized bed dryers, spray dryers and flash
dryers. Drum dryers and vacuum dryers are non-adiabatic
types of dryers, while fluidized bed, spray and flash dryers
are adiabatic dryers.
Drum dryers (Figure 1.12a) consist of a hollow revolving

drum which is heated internally by steam and revolves slowly
in a trough containing the fluid to be dried. As the drum
rotates, a thin film of fluid adheres to the drum, is dried by
the heat of the drum, and finally is removed from the drum
by a scraper blade.
Vacuum dryers may be of either the tray type (Figure

l.l2b), the conical type (Figure 1.12~) or the rotary drum
type (Figure 1.12d). Vacuum is used to reduce the
vaporization temperature of the solvent. This method is only
useful if the low temperature does not cause denaturation of
the proteins present.
In fluidized bed dryers (Figure l.l3a), a stream of

heated air is passed through material resting on a screen of
wire gauze or a sintered plate. The flow of air is adjusted
to fluidize the particles. A filter bag is used to prevent
escape of the dried solids.
Introduction 31
FEED SCRAPING
MEDIUM TEMPERATURE
LOW TEMPERATURE
VACUUM
CONNECTION
DOOR
\ TRAYS
CONNECTIONS
Figure 1.12. Dryers. (a) Drum dryer. (b) Tray-type vacuum

dryers. (c) Conical vacuum dryers. (d) Rotary drum vacuum
dryer.
DISCHARGE VALVE
VACUUM
FEED \ PORT CONNECTION
/
J DISCHARGE VALVE
Figure 1 .12. (continued)

Introduction 33
- EFFECTIVE BED HEltHT
DILUTE REGloN
I
___
TRANSITION
PHASE BOUNDARY
ZONE_ _
I-
DENSE REGION
FEED STREAN -
- DISTRIBUTOR
HOT AIR
a
FEED FROM HIGH @ESSURE PUMP
l.
+ EXHAUST AIR AND

FINES TO
COLLECTOR
DRlED
TPRODUCT
b
Figure 1.13. Dryers. (a) Fluidized bed dryer. (b) Spray
dryer. (c) Flash dryer.
RECYCLE
WET
DRY FEED
PRODUCT
0
AIR
0
” k?cq
HEATER
Spray dryers (Figure 1.13b) have the fluid material to

be dried added to a stream of hot gas. The inlet and outlet
temperatures are controlled to dry the material
instantaneously without any product decomposition or
denaturization.
Flash drying equipment (Figure 1.13~) requires that the

particulate to be dried be introduced into a hot gas stream
in a duct. The particulate is both dried and conveyed by the
hot gas during the very short transit time in the duct.
Separation of the dried material and the spent gas occurs at
the discharge of the duct.
Manufacturers of drying equipment are given in Table

1.9. Reviews and texts (27-33) should be consulted for
additional information.
1.2.2 Membrane Processes
Ultrafiltration and reverse osmosis, along with

Introduction 35
Table 1.9: Industrial Drying Equipment Manufacturers
Company Location
Aeromatic Towaco, NJ
Al Jet Equipment Company Plumsteadville, PA
APV Anhydro, Inc. AtUeboro Falls, MA
C.E. Raymond, Combustion Engineering, Inc. Chicago, IL
Dedert Corp. Olympia Falls, IL
Glatt Air Techniques, Inc. Ramsey, NJ
Komline-Sanderson Engineering Corporation Peapack, NJ
Krauss Maffei Corp. Charlotte, NC
Luwa Corp., Process Div. Charlotte, NC
Mikropul Corp., Micron Products Division Summit, NJ
Niro Atomizer Columbia, MD
Procedyne Corp. New Brunswick, NJ
Renneburg Div., Hey1 and Patterson Pittsburgh, PA
Swenson Process Equipment Inc. Harvey, IL
Wyssmont Corporation Fort Lee, NJ
microfiltration, are membrane separation processes. The

membrane is a barrier between two fluids. Its barrier
properties are its permeability or rate of transfer for a
component through the membrane and its permselectivity or
relative permeability flux for two components under the same
operating conditions.
In traditional filtration and microfiltration, the feed

solution flows directly onto the filter (membrane) and
particles larger than the pores accumulate on the membrane
surface (Figure 1.14a). The limitations of filtration are
realized as this boundary layer builds and gradually cuts off
the flow of solution through the membrane.
Ultrafiltration and reverse osmosis both separate a

solute from a solution by forcing the solvent to flow through
a membrane using a hydraulic pressure gradient (Figure
1.14b). The permselectivity of the membrane depends
strongly on molecular size: small molecules pass through the
membrane while large molecules are retained. When the
dimensions of the solute are within an order of magnitude of
the solvent dimensions, the process is called reverse osmosis.
When the dimensions of the solute are from ten times the
solvent dimension to less than 0.5 microns, the process is
called ultrafiltration.
FEED _ RET~TATE
MEbi8h4~E
------_---
JI
PERNEATE
b
Figure 1.14. (a) Schematic of a simple filtration system. (b)
[dealized membrane filtration system.
In ordinary filtration, the applied hydrostatic pressure

ranges from a fraction of an atmosphere to an atmosphere.
In ultrafiltration one to ten atmospheres and in reverse
osmosis 10 to 100 atmospheres of hydrostatic pressure are
applied as the driving force.
Introduction 37
For ultrafiltration, the flux of the solute, J,, is a

function of the pressure driving force and the concentration
driving force. This is represented mathematically as:
dCm
JS = Jv Cf (1 - a) - D; + (1.9)
In this equation, J,, the solvent flux, is equal to:
2
E r --AP
Jv = (1.10)
g n rx
where 6 is the membrane porosity, r is the effective pore

radius, n is the solution viscosity, AP is the pressure
difference across the membrane, r is the tortuosity factor for
the membrane and X is the membrane thickness. C, is the
concentration of the solute in the feed solution, u is the
fraction of the pure solvent flux which passes through pores
smaller than those retaining the solute molecules, Dp is the
diffusion coefficient and dCF/dx is the concentration
gradient of solute across the membrane.
For most ultrafiltration systems, the pressure term of

equation 1.9 contributes much more than the diffusion term.
Therefore, the solvent flux will be directly proportional to
the applied pressure.
The percent rejection rate, R, of a given ultrafiltration

membrane for a solute is given by:
R =
Js 100
‘- JC (1.11)
v f
If one can ignore the diffusion driving force, the rejection

rate is seen to be independent of the applied force.
In reverse osmosis, molecular diffusion through the

membrane is the rate limiting condition. A solution in which
the solute has a molecular weight of 500 or less will have a
significant osmotic pressure, even as high as 100
atmospheres, which the hydrostatic pressure must first
overcome before transport will occur through the membrane.
38 Separation and F’urification Techniques in Biotechnology
Example 1.4
Figure 1.15 (34) shows the flux through an
ultrafiltration membrane in a stirred cell as a function of
transmembrane pressure. Why does the stirrer rpm affect the
3’
UM-IO MEMBRANE
M-50 CELL
0065% PROTEIN (1830 RPM)
B-
3,9!tPROTEIN (1830RPH)
6-
6,5X PROTEIN (1830 RPd
p 6.5% PROTEIN ( 880 RPH)
I I I I 1
,344 .G89 1.034 1,378 1.722 2 I67
TRANSMEWIRANE PRESSURE (BAR)
Figure 1.15. Flux pressure relationships for bovine serum

albumin solutions in a stirred batch cell (Reference 34).
When protein solutions are passed by an ultrafiltration

membrane, a protein gel layer builds up on the membrane.
The faster stirrer speed reduces the thickness of this layer
and allows greater flux.
For reverse osmosis, the flux of the solute, Js, and of

the solvent, Jv, are conveniently represented by:
K1
Js = - (Cf - Cp) (1.12)
x
and
Introduction 39
K2
Jv = - (LIP - An> (1.13)
x
In these equations, K, and K, are the transport coefficients

of the membrane for the solute and the solvent, respectively;
X, Cf and AP are as defined for the ultrafiltration system; Cp
is the solute concentration in the product stream; and An is
the osmotic pressure difference between the two solutions.
Using Equations 1.12 and 1.13 in Equation 1.11, the

rejection rate is given by:
K&(LIP - Aa)
R = X 100 (1.14)
1 + Kl/K$W - An)
Thus it can be seen that for reverse osmosis, both the

rejection and the solvent flux increase with increases in the
hydrostatic pressure.
Membrane separation units have one of four different

configurations: tubular, planar spiral or hollow fiber (35).
Tubular modules (Figure 1.16) consist of bundles of

rigid-walled, porous tubes which are 1 to 3 cm in diameter.
The inside of the tube walls is lined with the control
membrane. The feed solution is introduced under pressure to
the inside of the tubes. The permeate is collected from the
outside surfaces of the tube while the remaining solution, the
retentate, is removed from the exit port of the tube.
POROUS TUBE
FEED
Ilk PERMEATE
Figure 1.16. Tubular membrane configuration.

Plate and frame or planar modules (Figure 1.17) consist

of a stack of grooved sheets which are covered on both sides
by control membranes. Each membrane covered sheet
alternates with a spacer sheet to form the module stack.
The membrane edges are sealed to prevent the mixing of the
permeate and the feed solution. The feed solution, under
pressure, flows tangentially along each of the porous sheets
as it is directed in a serpentine manner through the stack.
The permeate collects in the spacer regions and flows to the
permeate outlet.
STAINLESS
STEEL
TOP PLATE
ACRYLIC
TOP MANIFOL
RETENTATE
OUTLET
MEMBRANE PLATE
SEPARATOR SHEET
FILTRATE
OUTLET
FILTRATE OUTLET
BOT
TENTATE INLET
Figure 1.17. Plate and frame or planar membrane modules.
Spiral wound modules (Figure 1.18) consist of an

envelope of membrane covered sheets and separator sheets
which are wound concentrically around a hollow core and
then inserted into a canister. The pressurized feed stream is
introduced at one end of the canister and flows tangentially
along the membrane to the exit port of the canister. The
Introduction 41
permeate collects in the separator area and flows to the

hollow center where it is removed.
ROLL UP TO ASSEHRLE a’
FEED SIDE
/-
PERMEATE FLOW
THROUGH PIEHLIRANE)
PERMEATE OUT ---..)(
Figure 1.18. Spiral wound membrane system.
Hollow fiber membrane systems (Figure 1.19) consist of

bundles of fibers with an outside diameter of 15 to 250
microns. Whereas the feed solution flows inside the tubes in
the tubular modules, here it flows on the outside surface of
the fibers. The permeate flows along the inside of the fibers
to the end of the unit where the product is collected.
-) RETENTATE
PERMEATE
OUT
Figure 1.19. Hollow fiber membrane module.
Ultrafiltration is most often used for concentration, but

it can also be used to fractionate mixtures of large and small
solutes. Fairly high concentrations of suspended solids can
be processed by ultrafiltration if there is sufficient agitation
to prevent cake formation on the membrane. Reverse

osmosis has its greatest utility in concentrating such low
molecular solutes as salts and sugars.
The following review articles should be consulted for

additional information on membrane processes (36-41).
Membrane equipment suppliers are listed in Table 1.10.
Table 1.10: Industrial Membrane Equipment Manufacturers
Company Location
Amicon, Div. of W.R. Grace Danvers, MA
Culligan, Div. of Beatrice Northbrook, IL
Dow Chemical Midland, MI
Du Pont Wilmington, DE
Enka America, Inc. Ashville, NC
Gelman Sciences Ann Arbor, MI
HPD Inc. Naperville, IL
Koch Membrane Systems, Inc. Wilmington, MA
Memtek Corp. Billerica, MA
Millipore Bedford, MA
Monsanto St. Louis, MO
Osmonics Minnetonka, MN
Pall East Hills, NY
Romicon Woburn, MA
Separex Corp. Anaheim, CA
UOP Des Plaines, IL
Vaponics, Inc. Plymouth, MA
1.2.3 Solvent Extraction

Solvent extraction is a method of separation based on
the transfer of a solute from one solvent into another
solvent when the two solvents are brought into contact. It
is essential that the two solvents be immiscible. A solute
which is soluble in both phases will distribute between the
two phases in a definite proportion. The desired separation
is achieved by adjusting the chemical parameters of the
system: pH, solvent selection or ion pair formation.
The process consists of two steps: (1) intimately mixing

the two solvents until the solute has been distributed
between the liquids and (2) separating the two phases. The
distribution of the solute between the two phases is given by
the distribution coefficient, K,:
Introduction 43
2
2.3 RT log KD = vs (6s - ‘+I2 - (h2 - 6s) (1.15)
I I
-
where Vs is the molar volume of the distributing solute, 6,
..
is its solubility parameter and 6, and 6, are the solubility
parameters of the two immiscible solvents.
Most fermentation products are uncharged, although

those having acidic or basic functional groups can undergo
proton-transfer reactions that result in charged species, such
as RCOO- and RNH:. The solvent extraction of such
compounds can be carried out by means of pH control.
There is the possibility that the solute extracted into

the organic phase may become involved in a chemical
interaction. Such interactions lower the activity of the
solute and drive the overall equilibrium in the direction of
greater extractability, assuming the reaction product is more
soluble in the organic phase. As an example, the
dimerization of carboxylic acids in non-oxygen-containing
organic solvents results in higher distribution coefficients.
Ligand complexes, which are important for ion-pair formation
in metal salt extraction, can increase the distribution
coefficient by several orders of magnitude. Quaternary
ammonium salts dissolved in the organic phase have been
used to form ion-pairs with strongly acidic antibiotics so that
the resulting complex is hydrophobic. The efficiency of the
extraction depends on the lipophilicity of the quaternary
ammonium salts as shown in Table 1.l 1 (42).
Table 1.11: Comparison of Quaternary Ammonium Salts as Ion

Pair Complexing Agents for the Olivanic Acids
Quaternary Ammonium Salt Extraction Efficiency (%)

Benzyldimethyl-n-hexadecyl-
ammonium chloride 65
Trioctylmethylammonium chloride 73
Tetra-n-butylammonium chloride Trace
Dimethyldioctylammonium chloride 41
Dimethyldidecylammonium chloride 73
The fraction of solute extracted in a single stage is

given by the following equation:
cOvO
= %”
e= cv
0 0 + cvv” 1 + Kg
(1.16)
where Co and C, represent the organic and aqueous phase

concentrations, respectively; Vo and VW are the volumes of
the organic and aqueous phase, respectively; and V is the
phase volume ratio, V,/V,. From this equation it is
apparent that one can increase the extent of extraction, for
a given K, value, by increasing the phase ratio. Additional
extraction stages may be carried out on the same aqueous
solution using successive additions of organic solvent such
that V remains the same. After n such extractions, the
fraction of solute remaining in the aqueous phase is given by:
8 = (1 -eyl (1.17)
n
The relationships discussed so far are for the extraction

of a single solute from an aqueous phase to concentrate it in
an organic phase. It is also possible to use solvent
extraction to separate two solutes, A and B, which are
present in the aqueous phase.
If the initial concentration ratio of these solutes is

C&n, then the concentration ratio is reduced to
CAe,/CneB after extraction with the organic phase, w;;;
8, and 8, are the respective fractions extracted.
separation factor for the two solutes is given by the ratio
e,/e,, which is the change in the initial concentration due
to the extraction. An alternate measure would be the ratio
of the fraction of each solute remaining in the aqueous phase
after extraction, (1 - @J/(1 - e,).
Example 1.5
In a solvent extraction, what is the number of transfer

stages required to obtain 99% purity when the phase ratio
volume, VJV,, is 0.05 for the following separation factors:
(a) 1.2; (b) 1.5; (c) 2.0; (d) 2.5?
Equation 1.17 can be rearranged to give the number of

transfer stages (n) required:
log( en)
n = 1 + (1.18)
KD”
log(1 -
1 + KDV )
Introduction 45
For K, = 1.2, n = 80; for KD = 1.5, n = 65; for K, =

2.0, n = 49 and for K, = 2.5, n = 40. If the organic phase
ratio were allowed to increase so that V = 0.10, n could be
reduced to 42 for K, = 1.2. This cannot be carried too far
or the advantage of solvent extraction in increasing the
solute’s concentration is lost.
Solvent extraction may be applied at any stage of a

purification process but is usually most useful at the
beginning of an isolation procedure. High yields can be
obtained if the solute is stable and recovery from the solvent
is not difficult. In large scale manufacturing operations, the
use of more than two successive extractions with different
solvent pairs is unusual because of the large solvent volume
and expensive equipment required for each extraction.
The extraction processes can be grouped into batch,

continuous or countercurrent distribution processes.
When the extraction conditions can be adjusted so that

the fraction extracted is greater than 0.99 (VK, => loo), then
a single stage or batch extraction is all that is needed to
place the bulk of the solute in the organic extract. When
VKD is reduced to 10, it is necessary to carry out the
extraction twice to obtain 99% of the solute in the organic
phase. As Figure 1.20 shows, this equipment can be as
simple as an enlarged separatory funnel.
0
SOLVENT
AWEOUS
BROTH
1
Figure 1.20. Single stage extraction unit.

When the VK, values of the solutes are quite small,

multiple-batch extractions cannot be conveniently or
economically carried out since the large amount of organic
solvent added offers little improvement over the original
solute concentration in the aqueous phase. Continuous
extraction using volatile solvents can be carried out in
equipment in which the solvent distilled from an extract
collection vessel is condensed, contacted with the aqueous
phase and returned to the extract collection vessel in a
continuous loop. Figure 1.21 shows an example of such a
process scheme.
F.XtTaCt EXZIQCr Extract
Figure 1.21. Continuous co-current extraction scheme for use

with a volatile solvent,
Countercurrent distribution extraction is a special

multiple-contact extraction used to separate two solutes
whose K, values are very similar (Figure 1.22). The
extracted aqueous phase passes from vessels 1 to n while the
product enriched solvent phase is flowing from vessels n to
1.
Feed -
Mixev
I *eparator
1
?--
Solvent enriched
virh product
Figure 1.22. Counter-current extraction scheme.
Special centrifuges (Figure 1.23) which contain mixing

and settling sections may be used to carry out several stages
simultaneously. This Podbielniak centrifugal extractor has
played a major role in the recovery of antibiotics.
Introduction 47
:R IN
Figure 1.23. Podbielniak centrifugal extractor.
This extractor consists of a horizontal cylindrical drum

revolving at up to 5000 rpm. The heavier liquid phase is
introduced into the vessel on the shaft along the rotating
axis while the lighter liquid phase, also introduced at the
shaft, is guided internally to the periphery of the drum. As
the drum rotates, the liquid phases flow countercurrently
through the channels in the interior of the drum. The light
liquid phase moves toward the center where it is removed.
The heavy liquid phase moves to the periphery and then is
guided back to the shaft for removal. Flow rates greater
than 100,000 dm3/h are possible with this type of equipment.
The advantage of this type of extractor is the low hold-up
volume of liquid in the equipment compared to the
throughput.
Equipment manufacturers are listed in Table 1.12. For

additional information, the following texts and reviews should
be consul ted: References 43-48. Table 1.13 shows the many
enzymes which may be extracted from broths of disrupted
cells using solvent extraction (49).
1.2.4 Electrophoresis and Electrodialysis

Electrophoresis is the separation technique in which
electrically charged particles are transported in a direct-
Table 1.12: Industrial Extraction Equipment Manufacturers
Company Location
Alfa-LavaI/DeLavaI Co. Fort Lee, NJ
Baker Perkins, Inc. Saginaw, MI
Escher B.V. The Hague, Netherlands
Kuhni Switzerland
Liquid Dynamics Co. Hempstead, NC
Luwa A.G. Zurich, Switzerland
Westfalia Separators West Germany
Table 1.13: Extractive Separation of Enzymes from Disrupted

Cells Using PEG 1540/Potassium Phosphate Solutions
Cell
Concentration Partition Yield Purification
Enzyme (%I Coefficient (%) Factor
From Saccharomyces cerevkine
cY-Glucosidase 30 2.5 95 3.2
Glucose-6-phosphate dehydro-
genase 30 4.1 91 1.8
Alcohol dehydrogenase 30 8.2 96 2.5
Hexokinase 30 92 1.6
From Escherichia coli
Fumarase 25 3.2 93 3.4
Aspartase 25 5.7 96 6.6
Penicillin acylase 20 2.5 90 8.2
From Brevibacterium ammoniagenes
Fumarase 20 3.3 83 7.5
From Candida Boidinii
Formate dehydrogenase 33 4.9 90 2.0
From Leuconostoc species
Glucose-6-phosphate dehydro-
genase 35 6.2 94 1.3
current electric field. The mediums used in biochemical

applications are usually aqueous solutions, suspensions or
gels. It is the differential migration of solutes in these
mediums due to the electric force that separates the solutes.
The highest degree of resolution is obtained when an element
of discontinuity is introduced into the system, such as a pH
gradient, cellulose acetate membranes or the sieving effect of
high density gels. When the discontinuity elements are
charged membranes, the separation technique is known as
electrodialysis.
Introduction 49
It is the ionization of functional groups of the protein

or fermentation product that provides the charged species
which moves in the electric field. For large molecules, there
may be several functional groups which cause the molecule to
have a net positive charge at acid pH and a net negative
charge at basic pH. The pH where the molecule has zero
electrophoretic mobility is called the isoelectric point.
An important concept in the theory of electrophoresis is

that of the electric double layer (Figure 1.24). The electric
double layer is the structure formed by the fixed charges of
the macromolecule (or colloid) with the relatively mobile
counterions of the surrounding fluid. The total charge of
the double layer is zero. However, the spatial distribution of
charges is not random and gives rise to the electric
potential.
Figure 1.24. Schematic of electrical double layer.

The thickness of the double layer is normally given by

the inverse of the Debye-Huckel constant, k:
(1.19)
where e is the electronic charge, k is the Boltzmann

constant, no is the bulk concentration of each ionic species,
z is the valence of the electrolyte and 6 is the dielectric
constant.
The mobile counterions are bound sufficiently tightly to

the macromolecule so that they move with it during
electrophoresis, defining a surface of shear larger than that
of the macromolecule itself. The potential at this shear
boundary determines the electrophoretic mobility of the
macromolecule and is called its zeta potential.
The zeta potential cannot be measured directly. It is

calculated on the basis of theories for the electric double
layer which take into account the electrophoretic attraction,
the electrophoretic retardation, the Stokes friction and the
relaxation effect.
While many experimental adaptations of electrophoresis

have been used for analytical purposes, zone electrophoresis
and isoelectric focusing are the two variations which are also
useful for preparative separations of protein mixtures.
Zone electrophoresis is characterized by the complete

separation of charged solutes into separate zones. Many of
the earlier separations of this type were carried out using
filter paper in a Durum cell (Figure 1.25). Draped over glass
rods are strips of filter paper with their ends in contact
with wicks in the electrode buffer. The support medium
most commonly used in this type of cell now is cellulose
acetate membranes.
High density gels with specific pore sizes may be used

as the resolving medium to separate proteins with similar
dimensions. This technique is also quite simple in design and
operation and gives excellent resolution because of the
uniform porosity of the medium.
Introduction 51
LOADING SLOT
PLASTIC PEGS \
I
DRYING RACK
7\ ‘PAPER STRIPS
1 GLASS RODS
RACK SUPPORT
FEED WICK
PARTITION
BASE SECTION
BAFFLE SYSTEM
Figure 1.25. Durrum cell for electrophoresis separations.
In isoelectric focusing a pH gradient is superimposed on

a column filled with an electrically neutral medium which
exhibits a density gradient along the column. The density
gradient, to be effective, must be greater than the gradient
caused by the protein-buffer boundaries. The usual such
medium is sucrose at concentrations as high as 50%.
After the sucrose density gradient has been established,

the pH gradient is formed by the prolonged electrolysis of a
suitable mixture of amphoteric substances which have a wide
range of isoelectric points. The most acidic of the
ampholytes will be near the anode end of the column, the
most basic near the cathode end and the remaining
ampholytes covering the range between, depending on their
individual isoelectric points.
The protein mixture to be separated can be applied to

the column for electrophoretic fractionation. After such

fractionation, the column is emptied into separate fractions.
The protein fractions can be separated from the sucrose and
the ampholytes by dialysis or ultrafiltration.
Mixtures of a large number of aminocarboxylic acids

covering the most appropriate pH range (pH 3 to 10) are
commercially available in molecular weights (300 to 600)
sufficiently low for easy subsequent separation from the
isolated proteins (52).
References for additional details on the application of

this technique are the collection of papers in the books
edited by Bier (51) and the text by Shaw (53).
Electrodialysis is the diffusive transfer of an ionic

solute across a membrane due to an applied electric field. A
necessary feature for the utilization of this technique is the
solute’s potential for ionization. In electrodialysis the
membranes are usually positioned parallel to one another in a
planar configuration. This is the typical plate and frame
stack. This arrangement allows the most attractive geometry
for the efficient application of the electrical driving force.
As Figure 1.26 (54) shows, anion and cation membranes

are arranged alternately between two electrodes. Into
alternating chambers are placed an aqueous solution
containing M cations and X anions. The imposition of an
electric current causes the M cations to move toward the
cathode and the X anions toward the anode. However, each
ionic type cannot progress more than one chamber toward
the electrodes without finding its path blocked by a
membrane which is impermeable to that ionic type.
Therefore, the feed stream will be depleted of the ionic
solute and the adjacent stream will be enriched in the ionic
solute.
This technique requires that the membranes separating

the chambers have a sufficient ion exchange capacity as well
as pores of a sufficiently small size (-30 A) to repel
oppositely charged ions. The alternating series of dilute and
concentrating cells are manifolded together. The degree of
removal of the ionic solute from the feed stream is
determined by the ionization potential of the solute, the rate
of hydraulic flow and the applied current density.
Introduction 53
CATtON EXCHANGE MEMBRANE
ANION EXCHANGE MEMBRANE
SOLUTION
SPACER FRAMES,
ANOTHER END FRAME,
Figure 1.26. Electrodialysis modules (Reference 54).
The limiting factor for most products from fermentation

applications will be very low ionization potential and
concentration of the solutes. Even for an ionic compound
such as sodium chloride, the solute concentration must be
greater than 300 ppm. At concentrations less than this, the
voltage drop which occurs across the dilute compartments
will require excessive power input for economical operation.
Two parameters of importance in characterizing the ion

exchange membranes for an electrodialysis system are its
electrical resistance and its permselectivity. The membrane
resistance per unit area, measured in sea water, should be
less than 20 ohm/cm2. The perselectivity, Pa, is the measure
of the fractional increase in flux for the mobile counterion
due to the presence of a perfectly selective membrane and in
relation to the flux of the ion in the absence of the
membrane:
PS =
tll - % (1.20)
1
- %
where +m is the flux within the membrane and $B is the flux

in free solution. In practice, permselectivity is measured by
comparing the concentration gradient generated by placing
the membrane between NaCl electrolyte at two different
concentrations and comparing the actual potential with the

theoretical Nernst potential.
Most of the electrodialysis equipment in the United

States have been manufactured by Ionics (Watertown, MA).
Other manufacturers are Aqua-Chem (Waukesha, WI) and Toyo
Soda (Tokyo, Japan).
1.3 RECOVERY PROCESSES
A recovery process is the specific combination of the

recovery unit operations needed to attain the desired degree
of purity and yield for a specific solute.
The sequence of unit operations in the recovery process

is typically: (1) the removal of insoluble components, (2)
primary isolation, (3) purification, and (4) final product
isolation (55).
The removal of insoluble components is usually carried

out by filtration, centrifugation or simply decanting. This
step does not affect the concentration (0.1 to 5 g/L) or the
purity (0.1 to 2.0%) of the biochemical solute in the
fermentation broth.
The solute concentration is increased substantially (5.0

to 10.0 g/L), with a corresponding increase in purity (1 .O to
lo%), during the primary isolation. The solute is separated
from substances with widely differing polarities or size by
solvent extraction, adsorption, ion exchange, precipitation and
ultrafiltration.
Purification operations are techniques for the removal

of impurities closely related to the desired solute.
Adsorption, chromatography or fractional precipitation unit
operations are used for purification. The solute
concentration is increased to 50 to 200 g/L while the purity
is increased to 50 to 80%.
The final product isolation yields the desired solute in

the form needed for final formulation, blending or shipping.
Techniques include centrifugation, crystallization or drying.
The purity becomes 90 to 100%.
IntroducGon 55
Strategies for the development of large scale

purification systems for proteins have recently been proposed
by Wheelwright (56). Figure 1.27 (55) shows a recovery
process flowsheet for antibiotic production. The complex
process requires fifteen different separation techniques to
produce both a crude and a highly purified antibiotic product.
FERMENTER
LIIJUID-LIQUID
EXTRACTION
ANIMAL FEED
RllPPl
_I. __.._..
FMFNT .
BULK ANTIBIOTIC
Figure 1.27. Process flowsheet for recovery of an antibiotic

(Reference 55).
Flow sheets for the recovery of citric acid, sodium

clavulanate, micrococcal nuclease and three human proteins
synthesized in recombinant E. Coli are shown in Figures 1.28,
1.29, 1.30, and 1.31 to illustrate the range of techniques
which are used in the recovery of biochemical products.
Citric acid (Figure 1.28) (57) is recovered by filtration

of the biomass followed by a precipitation of calcium citrate
by the addition of calcium hydroxide. The precipitate is
treated with sulfuric acid to dissolve the calcium citrate and
attain citric acid. The calcium sulfate formed is filtered off.
The citric acid solution is decolorized, deionized and
concentrated using activated carbon and ion exchange resins.

Crystallization, centrifugation and drying operations are used
for the final recovery of pure citric acid crystals.
HARVESTED mow
4
FILTER OFF 1. NIGER HVCELlUH USING ROTARY VACUUM FILTER
4
bD tA(0,9 TO FILTRATE UNTIL pti 5.8
c
CALCIUH CITRATE
4
,bD H2S04 WHILE AT 607
c
FILTER ON ROTARY VACUUN FILTER TO REWOVE CASOQ
4
ACTIVATED CHARCOAL TO DECOLORIZE
4
CATION AND ANION EXCHANGE RESIN COLWNS
4
EVAPORATE TO POINT OF CRYSTALLIZATION AT 36’C
4
CRVSTALS OF CITRIC IWNOHVDRATE SEPARATED IN CONTINUOUS
CENTRIFUGES
c
DRIERS AT 50 - 60'C
Figure 1.28. Recovery and purification of citric acid

(Reference 57).
Sodium clavulanate (Figure 1.29) (58) belongs to the B-

lactam family and is used to inhibit the /I-lactamase bacterial
enzyme. After obtaining the supernatant from the
fermentation broth, the compound is concentrated by
extraction into butanol at acidic pH. This must be done
rapidly and at low temperatures to prevent dehydration of
clavulanic acid. The back extraction into water at pH 7.0
not only has the advantage of further purifying the sodium
clavulanic acid, but also returns it to a more stable
condition. Further purification is carried out by gradient
elution ion exchange chromatography. The salt is removed
by molecular size separation on a gel column. The final
purification is achieved by partition chromatography on
cellulose, followed by evaporation and freeze drying.
Rather than further concentrating the supernatant

containing micrococcal nuclease (Figure 1.30) (59), it is
diluted to reduce the conductivity. After acidification with
glacial acetic acid, microccal nuclease is adsorbed onto an
ion exchange resin. The enzyme-loaded resin is placed in a
column from which the nuclease can be eluted. The desired
Introduction 57
ClJLlURE SUPERNATANT
HCL TO PH 2.0; 314 VOL. OF BUTAN-I-OL

EXTRACTION AT 52
I
BUTANOL EXTRACT
1/15 VOL. OF WATER TO PH 7.0 WITH

20~(WT/VOL)
AOUEOUS NAOH
1
hEOUS BACK EXTRACT
ION EXCHANGE CHROMTOGRAPHV

PERWIT FFIP (SRA 62) RESIN CL-, NACL O-O.35 I
GRADIENT ELUTION, 5% COWlNED ACTIVE FRACTIONS
VACUUM CONCENTRATED
I
bNCENTRATED ACTIVE ELUATE
DESALTED ON BIORAD BIO~EL P2 COLUMN AT 5-C

AND VACUUH CONCENTRATED CONSINED, ACTIVE,
NACL-FREE FRACTIONS
I
bNCENTRATED DESALTED MTERIAL
COLUMN CHROWJOGRA~HV AT 5’C ON WHAT~AN CC31

CELLULOSE USlNG BUTAN-l-OL-ETHANOL-WATER,
4:1:5 (TOP PHASE); ACTIVE FRACTIONS COMGINED,
EVAPORATED TO DRYNESS, DISSOLVED IN DISTILLED WATER
AND FREEZE DRIED
I
bDlUM CLAVULANATE
Figure 1.29. Isolation procedure for sodium clavulanate

(Reference 58).
elution fraction is dialyzed to remove the unwanted salts.

Concentration is achieved by ultrafiltration and
centrifugation. That product may be used directly or further
purified on a gel filtration column followed by freeze drying.
Figure 1.31 (60) shows the recovery process steps used

to purify human insulin, human growth and human leukocyte
interferon made in recombinant E. coli. Such proteins,
synthesized from genetically engineered organisms and
intended for injection into humans, must be purified to
exacting standards. After disruption of the cells to extract
the desired protein, a series of precipitation and
chromatography steps are applied in each case to achieve the
required purity.
There is a limit, placed by economics, on the complexity

of an acceptable product recovery process for a given
biochemical. In general, the initial fermentation broth
concentration of the prepared solute will be inversely
proportional to the recovery costs. This is shown in Figure
1.32 (61) where recovery costs have been augmented by the
profit margin per kg to give the selling price. The clear
SUPERNATANT
400 LT. Pti 8.8 CONDUCTIVITY 23,000,S
1
DILUTION
ADD 400 LT. DENIN. Hz0 +ll.DOD pS
1
ACIDIFICATION
ADD GLACIAL ACETIC ACID TO PH 5.2
1
ADSORPTION
ADD SP-SEPHADEX C25 ( 750 G)
STIR 1 HR. SETTLING 2 HR.
1
COLLECTRESIN
PACK INTO AN ADJUSTABLE COLUMN, REE(OVE
UNADSDRBED PROTEIN BY PASSING 0.3 H
AMMONIUM ACETATE Pti 6.0
1
ELUTION
NUCLEASE ELUTED WITH 2 M CNH4)2S04
1
DIALYSIS
OVERNIGHT AGAINST DEMIN. H20
1
CONCENTRATION / CENTRIFUGATION
CH3 HOLLOW FIBER UNIT @#ICON), MOL.
WT. CUT OFF lO,OOO; lO,0OO G FOR 30 MIN.
GEL FILTRATION
SEPHADEX G75; O.OlZ ACETIC ACID + 0.1 N
AMMONIUM ACETATE
1
FREEZE DRYING
Figure 1.30. Purification procedure for micrococcal nuclease

(Reference 59).
trend in this figure illustrates the continuing pressure for

technological improvements in fermentation and recovery
yields.
Example 1.6
A new enzyme has been prepared by a fermentation

process that yields a concentration of 10 mg/L. Assuming a
35% return on sales (ROS), what is the limit on the process
and recovery costs to justify commercialization of this
product?
Introduction 59
HUMAN INSULIN (1979) HUMAN GROWTH HUMAN LEUKOCYTE
HORMONE (1981) INTERFERON (1981)

I
t & &
E. COLl EXTRACTION BY E. COLI EXTRACTION BY t. COLI EXTRACTION BY
MECHANICAL BREAKAGE MECHANICAL BREAKAGE MECHANICAL BREAKAGE
J 4 $
EXTRACTlON OF CENTRIFUGAL POLYETHYLENEIMINE POLYETHYLENEIMINE
PELLET WITH CHAOTROPIC PRECIPITATION PRECIPITATION
AGENT (GuHCc)
1 J
DIALYSIS
h OF CENTRIFUGAL
AMMONIUM SULFATE
PRECIPITATION OF
AMMONIUM
PRECIPITATION
SULFATE
OF
SUPERNATANT SUPERNATANT SUPERNATANT
4 & 4
CYANOGENBROMIDE ANION EXCHANGE DIALYSIS OF PELLET
CLEAVAGE OF DIALYSIS CELLULOSE CHROMA-
J
PRECIPITATE TOGRAPHY OF
IMMUNOADSORBENT
DISSOLVED PELLET
COLUMN ~~NocLONAL
EXTRACTION OF RESIDUE
\1 ANTIBODIES)
WITH GuHCL, CATION EXCHANGE
S-SULFONATION (CELLULOSE
CHROMATOGRAPHY) CATION EXCHANGE
4
DIALYSIS CELLULOSE
1
I 1
AMMONIUM SULFATE
CHROMATOGRAPHY
c PRECIPITATION
PRECIPITATE (B CHAIN)
1
J,
GEL FILTRATION
DEAE CELLULOSE
CHROMATOGRAPHY I
SEPHACRYL s-200
GEL FlLTRATlON
I .
L-~PH 5 PRECIPITATION OF
SUPERNATANT (A CHAIN)
AMINO ETHYL CELLULOSE

CHROMATOGRAPHY
JI
REVERSE PHASE HPLC
Figure 1.31. Recovery processes for human insulin, human

growth hormone and human leukocyte interferon synthesized
from E. coli (Reference 60).
From Figure 1.32 it is seen that the enzyme would need

to have a selling price of about $lO,OOO/kg, about the same
as for research and diagnostic enzymes. Therefore, with a
35% ROS, the maximum the process and recovery costs can
be for this enzyme is $6,50O/kg.
Separation and Purification Techniques in Biotechnology
CITRIC ACID, MSG
IBGERILLIC ACID
BULK ENZYMES
BGNOCLONAL
ANTIGODIES
GLYCEROPHOSPHATE
DEHYDROGENASE
8 IO-‘- LUCIFERASE
;
a
s
G
10’s-
FACTOR VI-1
8 a
10-b - UROKINASE 0
Q
THERAPEUTIC ENZVMS
IO” I I I I 1 I I I 1-I
IO” 10-l IO” IO’ IO’ IO’ IO’ IO’ rob IO7 108 IO9
SELLING PRICE (S/KG)
Figure 1.32. Relation between starting product concentration

and final selling price of prepared product (Reference 61).
1.4 REFERENCES
1. Lipinsky, E.S., Allen, B.R., Jenkins, D.M., LM Current

Research 5:102 (1984)
2. Hawtin, P., The Chemical Engineer (Jan 11, 1982)
3. Dlouhy, P.E., Dahlstrom, D.A., Chem Ena Progr 64(4):116

(1968)
4. Shirato, S., Esumi, S., J-1 41:87

(1963)
5. Bell, G.R., Hutto, F.B., Chem Enn Prog 54(6):75 (1958)
6. Boss, F.C., “Filtration”In: Fermentation and Biochemical

n in erin
Ea. (Vogel, H.C., ed.), Noyes
Publications, Park Ridge, NJ, p 163 (1983)
Introduction 61
7. Purchas, D.B., Industrial Filtration of Liauids, Leonard

Hill, London, 2nd Ed. (1971)
8. Savorsky, L., (ed.), Solid-Liauid Separation, Butterworth,

London (1977)
9. Tilles, F.M., et al., Theorv and Practice of Solid-Liauid

Separation, University of Houston Press, Houston, TX,
2nd Ed. (1975)
10. Purchas, D.B., Solid-Liauid Separation Technology,

Uplands Press Ltd., Croyden, England (198 1)
11. Ambler, C.M., “Centrifugation” In: Handbook of

Seoaration Techniaues for Chemical Engineers,
(Schweitzer, P.A., ed.), McGraw-Hill Book Company, New
York, NY, p 4-55 (1979)
12. Solomons, G.L., Materials and Methods in Fermentation,

Academic Press, London, p 299 (1969)
13. Boss, F.C., “Centrifugation” In: Fermentation and

Biochemical Enaineerina Handbook, (Vogel, H.C., ed.)
Noyes Publications, Park Ridge, NJ, p 296 (1983)
14. Aiba, S., Humphrey, A.E., Millis, N.F., Biochemical

Engineering, Academic Press, New York, 2nd Ed. (1973)
15. Gray, P.P., Dunnill, P., Lilly, M.D., Fermen Technol

Todav. Proc Int Fermen Svmp 4th, p 347 (1972)
16. Bird, R.B., Stewart, W.E., Lightfoot, E.N., Transport

Phenomena, John Wiley & Sons, New York, NY (1960)
17. Bennett, R.C., “Evaporation” In: Handbook of Separation

Techniaues for Chemical Engineers, (Schweitzer, P.A.,
ed.), McGraw-Hill Book Company, New York, NY, p 2-
131 (1979)
18. Freese, H.L., “Evaporation” In: Fermentation and

Biochemical EnPineerinP Handbook, (Vogel, H.C., ed.),
19. Newman, H.H., Chem Eng Proq, 64(7):33 (July 1968)

20. Perry, J.H., Chilton, C.H., Chemical Engineers Handbook,

McGraw-Hill New York, NY, 5th Ed., sec. 11 (1973)
21. Standiford, F.C., Chem Enq, p 157 (Dee 9, 1963)
22. Mullin, J.W., Crvstallization, Butterworth, London (1972)
23. Randolph, A.D., Larson, M.A., The Theorv of Particulate

Processes, Acad. Press, New York (1971)
24. Shirotsuka, T., Toyokura, K., Mem School Sci Enq

Waseda Univ, 3057 (1966)
25. Nyvlt, J., Industrial Crvstallization from Solutions,

Butterworth, London (197 1)
26. Nyvlt, N., Industrial Crvstallization, Verlag Chemie, New

York, p 129 (1978)
27. DeVivo, J.F., “Nonadiabatic Drying” In: Fermentation and

Biochemical Engineering Handbook, (Vogel, H.C., ed)
28. Keey, R.B., Introduction to Industrial DrvinP Operations,

Pergamon Press, Oxford (1978)
29. Cook, E.M., Chem Ennr Prog, (April 1978)
30. Quinn, J.J., Jr., “Adiabatic Drying” In: Fermentation and

Biochemical Enaineerina Handbook, (Vogel, H.C., ed.)
Noyes Publications, Park Ridge, NJ 334 (1983)
31. Keey, R.B., Drving Principles and Practice, Pergamon

Press, Oxford (1972)
32. Belcher, D.W., Smith, D.A., Cook, E.M., Chem Eng, p 112
(Jan 17, 1977)
33. Dittman, F.W., Chem Eng, p 106 (Jan 17, 1977)
34. Porter, M.C., Michaels, A.S., Chemtech, p 56 (Jan 1971)
35. Cruver, J.E., Marine Technoloav, 9:216 (1972)
36. Li, N.N., Long, R.B., Henley, E.J., Ind Ena Chem,
57(3):18 (1965)
Introduction 63
37. Michaels, A.S., In: Progress in Separation and

Purification, Vol 1, (Perry, E.S., ed.), Interscience, New
York, p 297 (1968)
38. Cheryan, M., Ultrafiltration Handbook, Technomics,

Lancaster, PA ( 1986)
39. Fane, A.G., Progress in Filtration and Separation, Vol 4

(Wakeman, R.J., ed.) (1986)
40. Trudel, M., Trepanier, P., Payment, P., Process Biochem,

18:2-4,9 (1983)
41. Olson, W.P., Process Biochem, 18:29 (1983)
42. Hood, J.D., Box, S.J., Verrall, M.S., J Antibiot, 32(4):295

(1979)
43. King, C.J., Separation Processes, McGraw-Hill, New York

(1971)
44. Brian, P.L.T., Staged Cascades in Chemical Processing,

Prentice-Hall, Englewood Cliffs, NJ (1972)
45. Belter, P.A., In: Microbial Technology (Peppler, H.J.,

Perlman, D., eds.), Academic Press, New York, Chapter
16, p 403 (1979)
46. Robbins, L.A., In: Handbood of Separation Techniaues

for Chemical Enpineers, (Schweitzer, P.A., ed.),
McGraw-Hill, New York, Section 1.9, p l-256 (1979)
47. Lo, T.C., In: Handbood of Separation Techniaues for

Chemical Engineers, (Schweitzer, P.A., ed.), McGraw-
Hill, New York, Section 1.10, p i-283 (1979)
48. Todd, D.B., “Solvent Extractions”, In: Fermentation and

Biochemical Enaineerina Handbood, (Vogel, H.C., ed.),
49. Kroner, K.H., Hustedt, H., Kula, M.R., Process Biochem,

19:170 (1984)
50. Abramson, H.A., Electrophoresis of Proteins, Hafner,

New York (1964)
51. Bier, M., ed., Electrophoresis, Academic Press, New

York, Vol I (1959); Vol II (1967)
52. One such mixture is called Ampholine, which is available

from LKB Instrument Corporation.
53. Shaw, D.J., Electrophoresis, Academic Press, New York,

(1969)
54. Lacey, R.E., “Dialysis and Electrodialysis”, In: Handbook

of Seoaration Techniaues for Chemical Engineers,
(Schweitzer, P.A., ed.), McGraw-Hill, New York, Section
1.14, p l-449 (1979)
55. Bailey, J.E., Ollis, D.F., Biochemical Engineering

Fundamentals, 2nd Ed., McGraw-Hill, New York, p 276
(1986)
56. Wheelwright, S.M., Bio/Technolonv 5:789 (1987)
57. Sodeck, G., Modl, J., Kominek, J., Salzbrunn, G., Process
Biochem, 16(6):9 (1981)
58. Reading, C., Cole, M., Antimicrob Aa Chemother,

11(5):852 (1977)
59. Darbyshire, J., In: Tonics in Enzvme and Fermentation

Biotechnoloav Vol 5, (Wiseman, A., ed.), Ellis Harwood,
Chichester, p 147 (1981)
60. McGregor, W.C., Ann N Y Acad Sci, 413:231 (1983)
61. Dwyer, J.L., Bio/Technology, 2:957 (1984)

2
Adsorption
2 .l INTRODUCTION
Adsorption is the phenomenon in which molecules in a

fluid phase concentrate on a solid surface without any
chemical change. Adsorption takes place due to unsatisfied
forces in the surface which attract and hold certain
molecules, the adsorbate, from the fluid surrounding the
surface of the solid, the adsorbent. The adsorption energy
determines the strength with which any given molecule is
adsorbed relative to other molecules in the system.
The adsorbed species are considered to form a

monolayer on the surface of the adsorbent. The method is
generally limited to molecules with molecular weights less
than 2,000 since many of the surface sites may be of limited
accessibility. Large molecules may be difficult to elute
because of the possibility of multiple sites of adsorptive
interaction. The concentration of the solute in the
adsorption technique is usually less than 1000 ppm. Above
that concentration, one should think in terms of
chromatographic methods of purification.
Adsorption, essentially a surface effect, must be

distinguished from absorption, which implies the penetration
of one component throughout the body of a second.
The utilization of adsorption processes for the

purification of drinking water, wine and oils by solid
65
adsorbents, particularly charcoal, has been done for centuries.

With the expansion of the chemical industry over the last
century and with the development of the biotechnology
applications in the last decade, the range of substances
purified by adsorption has increased enormously.
Adsorption is utilized in fermentation recovery

processes: (a) to remove unwanted molecules, such as color
bodies, from the solution of the fermentation product; (b) to
adsorb the fermentation product to increase its concentration
when the product is eluted from the adsorption medium; or
(c) to adsorb selectively the solutes in the fermentation
produce stream to separate them by chromatography. The
first two types of purification will be covered in this chapter
and the chromatographic application will considered
separately in Chapter 4.
2.2 ADSORPTION THEORY
Despite the long history of adsorption applications,

research into the fundamentals of adsorption is only a little
over sixty years old. Some of the early experiments with
solutions were reviewed by Freundlich (1). A variety of
substances, including phenol, succinic acid and simpler
aliphatic acids were adsorbed by charcoal from dilute aqueous
solution. These experiments measured the adsorption
isotherm, which for these dilute solutions can be represented
by the curve shown in Figure 2.1.
The curves can be described mathematically by what has

become known as the Freundlich equation, even though it had
been used earlier by Boedecker (2) and Kuester (3). The
equation is:
x/m = a 3” (2.1)
where x is the weight of the adsorbate taken up by a weight

m of solid adsorbent; c is the concentration of the solution
at equilibrium; a and n are constants. The form l/n was
used to emphasize that c is raised to a power less than
unity.
The Freundlich equation is based upon an approximation

which is valid only for dilute solutions. Many attempts have
Adsorption 67
BUTYRIC 4CID
PROPIONIC ACID
ACETIC ACID
FORMIC 4ClD
I I
U 1 2
CONCENTRATION OF ACID IN LIQUID PHASE, MOLES/LITER
Figure 2.1. Adsorption of organic acids onto activated

carbon following the Freundlich equation (Reference 1).
been made to derive a general equation for the adsorption

isotherm. These have been reviewed recently by Langmuir
(4) and by Jossens and coworkers (5).
Most of the theoretical models for adsorption were

developed for gases adsorbed on solids. However, these
models have also proven useful for liquid solutes with solid
adsorbents. For any specific system of solute, solvent and
adsorbent, the adsorbent’s surface sites are covered. It is
the relative affinity of this surface for the solute and the
solvent that determines the efficiency of solute adsorption.
Adsorption isotherms are used to represent the equilibrium
points of solute adsorbed on the surface as a function of the
concentration of solute in solution. At equilibrium, the rate
of solute being adsorbed on the surface equals the rate of
solute leaving the surface.
The adsorption isotherm is the starting point for any

analysis of adsorption processes. The isotherm is an
essential part of modeling adsorption kinetics and thus,
column or batch processing design, efficiency and economics.
The isotherm reveals the degree of purification that might be
achieved, the approximate amount of adsorbent required to
reach that degree of purity and the sensitivity of the
purification process to the concentration of the solute.
2.2.1 Adsorption Isotherm

The relatively simple Freundlich Adsorption isotherm has
probably been most widely used precisely because it is so
simple and because it was quite successful in correlating data
for adsorption on activated carbon over limited concentration
ranges. It is useful to examine the derivation of this
equation to appreciate its range of application.
Henry (6) has shown that the Freundlich isotherm can

be derived from the Gibbs adsorption equation combined with
a mathematical description of the free energy of the surface.
The surface free energy, Q, when a fraction (0) of it is
covered with a solute can be expressed as:
0 = a()(1 - 6) + ale = a0 - ((IO - al)0 (2.2)
where 00 is the surface free energy of the surface covered

with the pure solvent and 01 is the surface free energy of
Adsorption 69
the surface covered with a monolayer of solute. The fraction

of surface covered is the amount of sorbate per unit sorbent
(x/m) divided by the monolayer capacity of the sorbent for
the sorbate (x/m&
e = (x/ml/(x/m) M (2.3)
Equation 2.2 can then be expressed as:

(uo - ul) (x/m)
a =a - (2.4)
0 (x/m) M
The Gibbs adsorption equation forms the mathematical

description of the surface excess, I’, which is the extent by
which the surface concentration of a given component
exceeds its concentration in the bulk liquid. For dilute
solutions, the Gibbs surface excess equation is:
c da
r -_ - (2.5)
RT dc
where c is the concentration of the solute in solution.
For dilute solutions, x/m is the Gibbs surface excess, l?,

so that:
X C da C
ao - al d (x/m)
- = _-* (2.6)
m RT dc=- RT (x/m) dc
M
which can be integrated to give:
X RT(x/mlM
In -y-- = In c + In k (2.7)
ao - al
By replacing RT(X/m)M/(Ue - ol) by l/n, Equation 2.7

becomes the familiar form of the Freundlich isotherm:
X
- = k cl/” (2.8)
m
This derivation points out that the Freundlich

adsorption isotherm is limited to those cases where a dilute
solution is involved, that is, where x/m = I’. A second
limitation of the Freundlich expression is that it does not
predict a linear isotherm in the limit of zero concentration.
The adsorption isotherm derived by Langmuir has also

been useful over limited concentration ranges. This isotherm
was based upon the following assumptions: 1) all adsorption
sites are equivalent, 2) only one layer of solutes or solvent
is adsorbed, and 3) there is no interaction between adjacent
adsorbed molecules. For liquid systems, one must also assume
that all adsorption sites are occupied by either solute or
solvent molecules.
The adsorption process for such a system can be written

as:
At SY +AY + S (2.9)
with the equilibrium constant:
(2.10)
In this equation, A designates the solute molecules, S is the

solvent molecules and Y the surface sites. Although the
equation is valid only for thermodynamic activities, since the
sites are assumed to be of equal activity, the ratio of AY to
SY is the same as the ratio of the actual concentration of
AY to SY in the adsorbent. Thus:
I I
-=
AY e
(2.11)
I I
SY 1 - a
where 8 is the fraction of the adsorbent’s surface covered by

A and (1 - 0) is the fraction covered by S. Since the
concentration of A is usually small compared to the
concentration of S, the mole fraction of S in the liquid
phase, [l - A], is approximately 1. Thus, Equation 2.10 can
be written as:
= 8 (2.12)
K
(1-8)
II
A (l-8)
I1
A
At low concentrations of A and small values of 8, the

term 6A is small and the ratio 6/A becomes nearly constant.
At high concentrations of A in the liquid phase, 0 approaches
1 and the concentration of A in the adsorbent comes
constant. This is shown in Figure 2.2.
Adsorption 71
CONCENTRATION IN SOLIJTION
Figure 2.2. General Langmuir adsorption behavior. When the

behavior follows the dashed line, the adsorption is too great
for product recovery.
By replacing [A] with c and 19 with q/b, Equation 2.12

can be rearranged to give the more familiar form of the
Langmuir equation:
b K c
4
= (2.13)
l+Kc
where b is the number of moles of adsorbate adsorbed per

gram of adsorbent in forming a layer one molecule thick on
the adsorbent surface and q = x/m, the amount adsorbed per
gram of adsorbent.
The assumption concerning the equal activity of surface

sites is the most common failing of the Langmuir isotherm.
The Langmuir derivation lends itself to modification for those
cases where the assumptions are not valid.
Beverloo and coworkers (7) classified the many

adsorption equations according to the number of parameters
in the model and to the “credibility” of the relation at low
and high solute concentration. As measures of credibility,

adsorption equilibrium relations should have the following
features: (1) it should asymptotically approach a linear
equation at low concentration, (2) it should also
asymptotically approach a maximum value at high
concentrations, and (3) the level of adsorption should not
have a maximum value in the range of concentration values.
The third requirement should be seen as a limit on the
permissible values for the parameters in the adsorption
equilibrium equations.
Table 2.1 lists the equations, the number of parameters

and their applicability (credibility) at low and high
concentrations. It is interesting to note that the much used
Freundlich equation has a low “credibility” at low and high
concentrations. Beverloo (7) recommended the use of the
exponential, the Langmuir, the Volmer and the Toth
equations.
The adsorption of solutes that are completely miscible

in water, such as organic acids, could not be adequately
represented by these isotherm equations. A Chebyshev
rational approximation polynomial provided a good fit to the
adsorption data over the concentration range 10m2 to 10s4M.
The Chebyshev Equation used was:
X ac
-= (2.14)
m 1 + bc + dc2
The fitting parameters (a, b and d) are given in Table 2.2 for
the temperatures 278”K, 298°K and 323°K when the units on
x/m are mmol/g and on C are mol/L.
While most adsorption isotherms follow the type of

adsorption-concentration behavior shown in Figure 2.2, other
profiles have been observed. Giles (9) classified these
according to the system shown in Figure 2.3. The class to
which the isotherm belongs is based on the initial slope of
the isotherm. The sub-group is assigned according to the
shape at higher concentrations.
The S class is obtained if: (1) the solvent is strongly

adsorbed; (2) there is strong intermolecular attraction within
the adsorbed layer; and (3) the adsorbate has a single point
of strong attachment.
Adsorption 73
Table 2.1: Adsorption Equilibrium Equations (7)
Credibility
Equation Name Parameters Low c Highc
q = Hc Henry 1 +
q = A(c + f, linear 2
q = ; (1 - exp (-Bc)) exponential 2 t t
AC
q=1+Bc Langmuir t t
c=&jy.P(*) t t
C=+PcA 3 Myers +
(reduced)
q = Af cn Freundlich
q=( -- (Ac)” ] “* Toth t t

1 + (Bc)”
c = 9 exp ( Bfg t
A A
Volmer + t
(extended)
9= is- I1 - exp (-DC)) Langmuir

extended
t t
q= (1 + is, (1 - DC)
B.E.T. + _
q=A 0~)~ Fritz- _

(0 < n < in)
-Schluender
B (l+(Bc)“)
AC (I + DC)
(B > D > E) Langmuir 4 + t
q = (l+Bc) (l+Ec)
(extended)
Table 2.2: Chebyshev Approximations Parameters (8)

Organic
Acid T(OK) a 4 4
acetic 278 0.7100 2.2177 -2.4230
298 0.4059 -2.6185 19.0550
323 0.2761 -1.3078 10.4550
propionic 278 2.6665 -4.2880 46.592
298 1.6768 -1.2595 12.291
323 1.2356 -1.1009 9.453
butyric 278 13.4945 1.6677 222.012
298 9.8098 0.4733 144.518
323 7.5380 12.5698 -131.167
hexanoic 278 193.6 254.1 -6101.9
298 154.1 221.7 -6454.3
323 121.4 225.6 -1613.7
CLASS
S L H C
CONCENTRATION OF SOLUTE IN LIQUID PHASE
Figure 2.3. Classification system of isotherms (Reference 9).

Adsorption 75
The L class contains the Langmuir-type of curves. It is

found when there is no strong competition from the solvent
for sites on the surface. It may also be found if the
adsorbate has linear or planar molecules with the major axis
adsorbing parallel to the surface.
The H class occurs when there is high affinity between

the adsorbate and the adsorbent which is shown even in very
dilute solutions. This may result from chemisorption or from
the adsorption of polymers or ionic micelles.
The C class indicates constant partition of the

adsorbate between the solution and the adsorbent. This type
of adsorption mainly occurs with textile fibers, into which
the solute penetrates to increasing levels as its concentration
in the solution is increased.
The steps which occur for sub-groups 3 and higher are

usually interpreted to mean a phase change in the adsorbed
layer or the onset of the formation of a second molecular
layer after the completion of the first.
Example 2.1
The equilibrium adsorption of two amino acids,
phenylalanine and glycine, on a crosslinked polystyrene
adsorbent is given in Table 2.3 (10). What are the
coefficients that will fit this data to a Freundlich isotherm
and to the Langmuir isotherm?
Table 2.3: Equilibrium Adsorption of Phenylalanlne and Glycine

on Amberlite XAD-2 (10)
Phenylalanine Glycine
solution amount amount solution amount amount

concentration adsorbed adsorbed concentration adsorbed adsorbed
(moles/liter) (mg/g) (moles/g) (moles/liter) (msf 9) (moles/g)
0.0112 10.9 6.60x10-5 0.0126 0.60 7.94x10-s
0.0224 19.8 1.20x10-4 0.0251 1.06 1.41x10-5
0.0302 26.1 1.58x10-* 0.1000 4.22 5.62x10-’
0.0355 29.4 1.78x10-* 0.1995 8.41 1.12x10-4
The coefficients for the Freundlich isotherm may be

obtained by plotting the log of the amount adsorbed versus
the log of the concentration. This is shown in Figure 2.4.
-6,0
-4,8
Pl4ENYl.Al.ANlNE
0 L
0 -0.4 -0.8 -1.2 -1.6 -2.0
LO6CONCENfRAfION h.ES/LITER)
Figure 2.4. Freundich plot for the phenylalanine and glycine

data from Table 2.3.
The values for k, from the intercept of the line, are 6.4 x
10-s and 6.6 x 10Wg for phenylalanine and glycine,
respectively. The values for l/n, from the slope of the lines,
are 0.884 and 0.940 for phenylalanine and glycine,
respectively.
The coefficients for the Langmuir isotherm may be

obtained by plotting the concentration divided by the amount
adsorbed versus the concentration. This is shown in Figure
2.5. The values for b, from the reciprocal of the slope of
the lines, are 8.88 x lo-* and 6.48 x 10B5 for phenylalanine
and glycine, respectively. The intercept of the lines, l/Kb,
may then be used to give the values for K, 7.167 and 11.084
for phenylalanine and glycine, respectively. Note that the
slope for the glycine adsorption goes to 0 when the
concentration is above 0.025 moles/liter.
Adsorption 77
2000
GLYCINE
1600
PHENYLAIANINE
-0
I I I I I
0 0,04 0.08 0.12 0.16 0.20
CONCENTRATION ~OLES/L~TER)
Figure 2.5. Langmuir plot for the phenylalanine and glycine

data from Table 2.3.
2.2.2 Adsorption Kinetics

Adsorption of small molecules onto a non-porous solid
surface normally takes place rapidly as long as the solution
is not viscous. The equilibrium adsorption of butanol from
an aqueous solution onto Graphon occurs within ten minutes
(11). Larger molecules require a longer time to reach
adsorption equilibrium. Stearic acid in a benzene solution
requires four hours to reach adsorption equilibrium on iron
powder, even though 90% of the equilibrium value had been
adsorbed in the first five minutes (12).
Adsorption by porous solids may also be fast in those

cases where the liquid-adsorbent slurry is continuously
agitated. Adsorption equilibrium with aqueous solutions of
mono-, di- and trichloroacetic acid onto a silica gel was
attained within 1.5 to 2.5 minutes with continuous agitation,
but required 12 to 30 minutes with intermittent agitation.

The adsorption equilibrium with charcoal took much longer,
indicating the kinetic dependence on the pore size of the
adsorbent and on the molecular size of the adsorbate (13).
The adsorption of macromolecules proceeds more slowly.

The adsorption of dextran by collodion was still incomplete
after 50 hours. The presence of substances in the solution
with dextran, such as salts and detergents, altered the
adsorption rate (14).
Many models of the adsorption process have been

developed which may be solved to generate detailed
parametric curves using numerical techniques. These
mathematical models are useful if they assist in the design of
the pilot studies and in the interpretation of the pilot data
so that it may be used in the engineering design of the full
scale unit. A model will do this if it can predict the
effluent concentration of solutes in complex mixtures and if
it can predict the impact of changes in process parameters
on the performance of the adsorption unit (15).
Figure 2.6 illustrates the different steps in the

adsorption process which must be included or accounted for
in the model. First, the solute moves from the bulk solution
to the vicinity of the adsorbent material. If the adsorbent is
a porous material, the solute diffuses into the fluid contained
in the pores. The solute may adsorb on the exterior of the
adsorbent particle or on the surface of its pores. Once
adsorbed, the solute (now the adsorbate) may diffuse along
the surface of the pores. The concept of surface diffusion
along the surface of the pores was introduced to explain the
effective diffusivities which were determined (16,17) to be
higher than molecular diffusivities.
The adsorption process usually occurs much more rapidly

than the diffusion through the bulk solution or than the pore
or surface diffusion processes. Therefore, it is reasonable
that only the diffusion processes are included in mathematical
descriptions of the rate of adsorption.
The adsorption rate models reported in the literature

have usually been derived or solved using one or more of
these simplifying assumptions: external mass transfer
resistance has been neglected, pore diffusion has been
Adsorption 79
MODEL MECHANISMS DlFFUSlON flECHAN1SMS
DRIVING FORCE
i
!I
LIQUID PHASE ADSORBENT PHASE
Figure 2.6. Mechanisms and schematic for the homogeneous

surface diffusion model for porous adsorbents (Reference 15)
neglected, the accumulation of the solute in the liquid phase

in the pores has been neglected, or a linear isotherm was
used for mathematical expediency. Recent literature reviews
of rate models and their limitations have been prepared by
McKay (18) and by Leyva-Ramos and Geankoplis (19).
The simplest systems to be modeled are for adsorbents

without porosity (7). In this case, the rate of adsorption
will be dependent on the diffusion rate at which the solute is
transported from the bulk of the solution to the surface of
the adsorbent particles. This is normally viewed as a
concentration driven process:
NA = B1 a (c - CiA) = E d$dt (2.15)
where N, is the rate of adsorption: 61 is the mass transfer

coefficient through the liquid; a is the specific outer particle
surface; c is the bulk solution concentration of the solute;
ciA is the solute concentration near the adsorbent surface;
s,
is the average concentration of the solute adsorbed to
the adsorbent particle; t is the adsorbent fraction of the
total volume; and t is the time.
For “infinite” bath experiments in which the Langmuir

adsorption isotherm is applicable, Equation 2.15 can be
integrated to give:
a t (1 + SC) q
I
El A I
c
m
-7riir -l+lcIn I-
AC
II (2.16)
where A and B are the Langmuir parameters.
For “finite” concentrations of the solute in solution, the

analytical solution becomes more complex:
YI 4 a- 38
+I
1
02 + (0 + 6 - 112 - I
In -- +
I 20 +0-z ((a + 1)2 + 2E(O - I) + B2)4
(2.17)
I
(32 + E - 112+ II + 6 - 3 + (2 - 1) (a + lb2 + 2llto -’ 1) + !s
Ill
(30 + 8 - 112 + 0 + II - 3 - (2 - 1) -to + I)* + 28(0 - I) + n2
where
B1 a t
Y =
1 - E
(2.18)
z = C/C (2 = 1 for Y = 0) (2.19)

0
u = (II cop (2.20)
and
A E
0 = -___. (2.21)
(1 - El-
B co
A general model for porous adsorbents must include
both the mass balance of the solute in the fluid phase
surrounding the adsorbent particles and the mass balance of
the solute within the particle. The adsorption isotherm
equation is used to describe the concentrations of the
adsorbate during the adsorption process. The initial and
boundary conditions allow simplification of these equations
for the determination of several parameters from measurable
properties.
Adsorption 81
Rasmuson (20) specified the mass balance equations and

boundary conditions for batch, slurry and fixed bed
adsorption operations as follows:
Mass balance in the fluid phase surrounding the

adsorbent:
Batch:
dc - 3 No (1 - E)
-= as T-)0 (2.22a)
dt ER
Slurry:
dc 1 3 No (1 - E)
-E-
(c - c) - (2.22b)
0 E R
dt T
Fixed Bed:
ac ac a*c - 3 No (1 - E)
-+v- - DL - = (2.22c)
a2
2 E R
at a2
Mass balance inside the adsorbent particle (identical in

each case):
2
ac
s-+N = Da+ (2.23)
i EP P ar2
p at r ar
where
Ni = k (c --
‘A )
= - aqA + k q (2.24)
ads P r A
at
KA
Boundary conditions:
Fixed Bed Batch and Slurry

c(O,t) = co ------- (2.25)
cc-,t) = 0 _______ (2.26)

Fixed Bed Batch and Slurry
c(z,O) = 0 c(0) = ci (2.27)
ac
L(O,z,t) = 0 -J-(0,t) = 0 (2.28)
at ar
cp(R,z,t) = cp,r=R cp(R.t) = 'p/r=R (2.29)
given by
No =
r=R
= kf (c - c
p/r=R) (2.30)
Cp(‘.Z’O) - q*(‘.z*O) = 0 Cp(‘,O) = q*(r,O) = o (2.31)
N, which couples Equation 2.23 with the appropriate

form of Equation 2.22, is the mass flux from the exterior
fluid phase to the outer surface of the adsorbent particles.
In these equations, cp is the concentration of the solute

in the pores of the adsorbent, DL is the longitudinal
dispersion coefficient; D, is the diffusivity of the solute in
the pores of the adsorbent; K, is the adsorption equilibrium
constant; kads is the adsorption rate constant; k, is the
surface reaction of rate; r is the radial distance from the
center of the adsorbent particle with radius R; V is the fluid
flow velocity; z is the distance along the direction of flow in
the adsorption column; and cP is the void fraction of the
adsorbent particle.
The two terms on the left of Equation 2.23 represent

the accumulation of the solute in the liquid pore volume and
of the solute adsorbed on the pore surface, respectively.
The two terms on the right represent the pore volume
diffusion and the surface diffusion.
The boundary conditions state that the fixed bed is

initially free of solute in the fluid phase and on the
adsorbent. For the batch and slurry cases, the solute in the
fluid surrounding the adsorbent is initially at a concentration
Ci*
Adsorption 83
These equations were solved analytically using the

Laplace transform technique and inversion in the complex
plane to give solutions in the form of infinite integrals. The
integrals were evaluated numerically for specific values of
adsorption parameters to give the breakthrough curves shown
in Figure 2.7, 2.8 and 2.9 when kadl + oo and k,= 0; that is,
when the adsorption occurs instantaneously when the solute
reaches the adsorption site and there are no chemical
reactions taking place. These curves are plotted in terms of
the change in concentration as a function of the contact
time parameter, Y. The variables are expressed in terms of
the parameters:
Contact time:
y = ($-kfk+ (2.32)
Residence time:
3D E z (1 - E)
6 =
(2.33)
R* E V
Peclet number:
(2.34)
Distribution ratio:
KA + E (1 - E)
DR = (2.35)
E
Diffusion resistance ratio:
V = DD (2.36)
% R
The residence time, 6, has also been called the bed

length parameter since it contains the ratio of the bed
length to the adsorbent particle size, with adjustments for
&Q SePiWatiOn and Purification Techniques in Biotechnology
the void space in the bed and in the particles. The particle
diameter is always very small in comparison to the overall
bed length; typically 0.5 cm compared with 50 to 100 cm for
the packed adsorbent bed length. Therefore, values for 6
will typically range from 5 x 10e2 to 1.0 with values of V
between 0.001 and 0.21 m/s. The very sharp breakthrough
curve of Figure 2.8 was obtained assuming 6 = 100 which
would mean that V = 1.0 x lOm5, assuming one uses the same
particle diameter and bed length as in Figure 2.7.
10-l
10-Z
uo
‘u
10-3
Figure 2.7. Fixed bed breakthrough curves modeled for

different Peclet numbers with 6 = 1.0 (Reference 20).
10-l
10-2
L?
2.
10-3
10-4

different Peclet numbers with 6 = 100 (Reference 20).
Adsorption 85
The changes in the Peclet number show the influence of

longitudinal or axial dispersion at different flow rates. In
this instance, lower values of V correspond to lower Peclet
numbers which show shallower breakthrough curves. This
behavior is the result of the longitudinal molecular diffusion
causing increased spreading of the solute concentration
during the increased residence time of the lower flow rates.
Longitudinal diffusion is known to arise for Peclet numbers
less than 1.0. If longitudinal dispersion can be neglected,
then P, becomes equal to 00 .
The distribution ratio, Dn, gives a measure of the

accumulation of the solute on the adsorbent and in the void
spaces both of the bed and of the adsorbent particles. If
there is negligible accumulation in the void spaces, Dn + 00.
Figure 2.9 shows that D, = oo and D, = IO6 have nearly
identical breakthrough curves. As the value of 6 increases,
the breakthrough curves for lower values of D, will coincide
with the curve for D = 00. For example, at 6 = l-0, the
breakthrough curve of D, = IO3 coincides with that of D, =
00.
10-l
iOw4

different distribution ratios (Reference 20).
The parameter u is the ratio of the external diffusion

resistance to the internal diffusion resistance. There are few
circumstances for fixed bed operations where Y would be
greater than 10m2. Simulation results (20) have shown that
the breakthrough curves for those cases when v = 10m2 are

very similar to those when v = 0.
The experimental values for the pore diffusivity of the

solute, D,, can be determined from cPD/r, where D is the
molecular diffusivity and r is the tortuosity factor of the
adsorbent pores. A typical value of r is 3.6. A typical value
for D, in a system with activated carbon as the adsorbent
is 2.5 x lo-lo m2/s (19). For other systems, the value will
generally not vary from this by more than a factor of 2 or
3.
The mass transfer coefficient, PI, will typically have

values around 0.0001 m/s. The range of values will be from
0.0001 up to 0.0002 m/s. The mass transfer coefficient in
packed beds at low Reynolds numbers, 0.5 to 900, has been
calculated from the equation (21):
= 1.15 D;“,’ ( 2 RvP “0 )‘i’( D:o, r’3 (2.37)

$1
where u, is the solute solution velocity (m/s); CL is the

kinematic viscosity and p is the density. The molecular
diffusivity, Dmol, can be calculated from: (22)
T Mos5
D = 7.4 x 1o-8 (m3/sec) (2.38)
mol “0.6
‘D m
where V, is the molecular volume of the Solute; pD is the

dynamic viscosity; T is the temperature (“C) and M is the
molecular weight of the solute.
Example 2.2
How long does it take for the relative concentration,
c/c,, to rise to a breakthrough value of 0.90 of the solute
concentration in the feedstream, assuming 6 = 1.0 and P, =
1OO? What flow rate is needed, in this instance, when the
bed length is one meter?
With an adsorption equilibrium constant equal to 1.5 x

lo3 m3/m3, a particle radius of 0.05 cm, a particle porosity
of 0.43 and a pore diffusivity of 2.5 x lo-lo m2/s, the time
to breakthrough is 1.74 x lo4 seconds or 4.8 hours. With a
Adsorption 87
void fraction of 0.38 in the bed, the flowrate through a one

meter adsorption bed during this 4.8 hours should be 2.1 x
10s m/s.
For specific cases that are not modeled by the curves

in Figures 2.7 to 2.9, McKay (18) has provided a solution
that may be used fairly easily to calculate the concentration
distribution of a solute within the fluid and solid phase in
fixed beds of adsorbents. His derivation was based upon the
work proposed by Spahn and Schlunder (16) and by Brauch
and Schlunder (23).
The theoretical adsorption rate can be determined using

the following equation and experimental values obtained from
batch tests:
dn 3 (1 - Ch I-l) (1 - nP3
-- = (2.39)
dT 1 - (1 - l/Bi)(l - ~1”~
in which Ch is the column capacity factor; Bi is the Biot

number; q is a dimensionless solid phase concentration
parameter; and r is a dimensionless time parameter.
t-t 4;
Ch = (2.40)
“s co
B1 I(
Bi = ____ (2.41)
0
eff
II
ll =
4 / 9, (2.42)
(2.43)
where M is the mass of adsorbent; V, is the solution volume;

D eff is the effective diffusivity; and $e is the concentration
of solute adsorbed at equilibrium.
An analytical solution for r allows Dell to be calculated:

l”(pq 2B
- l’)] + l”(t.g.+~ 3/a
T = --+-[
)
(2.44)
+[ (a;)c, (&’ 2$ - tan-l 21;’ )]
where
B = 1 - l/Bi (2.45)
l/3
x = (1-n) (2.46)
a = ( l ;,“” )113 (2.47)
The effective diffusivity describes the concentration

versus time decay curve over the whole range of the
isotherm for which it applies. In the instance cited by
McKay, for an initial solute concentration of 450 mg/dm, the
equation was valid over the time interval 5 to 1440 minutes.
The column may be divided into two sections: the

section where a constant pattern of adsorption exists and the
section which is in the transitional state. The concentration
of the solute in solution as a function of position and time
can then be given by:
for the constant pattern region:
--.L=
c 1
-1In- (1 - x,)3 2B (2 + x 2 + 1) - 1 -1 2x2 + 1
2 - tan
cO 6 (1 - x,1 0 6
(2.48)
1 (1 -
3
x0) 28 (xi + xo + 1) 1 2x0 + I
-1
- - In + - tan
6 (1 - x,) 2 I/r LA-
where:
(2.49)
(2.50)
Adsorption 89
for the transition region:
1 (’ - x3)28(x2
2 2 t x2 t 1) 1 _, 2x2 t ’
EC = - In - - Ian
cO 6 (1 - x2) 2 d- fl
(2.5 1)
- 7, - 7 - --+
where
1 B
m_- (2.52)
‘1 = *
3
and 6 is a dimensionless liquid phase concentration

parameter.
Example 2.3
Using the Equation 2.48 and 2.51, the breakthrough
curves shown in Figure 2.10 were calculated. The flow rate
was 2 cmS/s; the column height was 200 cm; the particle size
was 600 microns; the equilibrium adsorbent capacity was 0.2
g/g; and the effective pore diffusion coefficient was 2.0 x
1O-1°m2/s. As the initial concentration was changed from
350 mg/dms to 450 mg/dm3 to 550 mg/dms, the shape of the
breakthrough changes and is shifted to the left, indicating
earlier breakthrough times.
0.6
CT
CO
0 40 80 120
VOLUME TREATEo (on31
Figure 2.10. Effect of initial solute concentration on the

breakthrough profile using a fixed bed height of 0.20 m
(Reference 18).
If the maximum allowable concentration of solute in the

effluent is 35 mg/dm’, what is the maximum volume of each
feedstream that can be treated under these conditions?
This would represent Q/C, values of 0.100, 0.078 and

0.064 for the three feedstreams. Therefore, the volume
treated would be 66 dm3, 49 dm3 and 30 dm3. Note that this
corresponds to different amounts (23.1 g, 21.6 and 16.5,
respectively) of solute being adsorbed with the same amount
of adsorbent, same flow rate and same bed height.
While the previous techniques used mass transfer

coefficients to model the resistance a molecule encounters as
it transfers from the bulk solution until it is eventually
adsorbed on the surface of an adsorbent particle, it is also
possible to model column behavior using the mass transfer
zone approach (24). Figure 2.11 shows a schematic
representation of the adsorbate loading on the column as a
function of time and position. Prior to the introduction of
the feedstream, the adsorbate concentration (q) on the
adsorbent is q. or zero. Once the feedstream is introduced
at time t, at a flow rate of u, the adsorption process is
underway.
The flow of the feedstream enters the column at a

constant concentration, co, and continues through the column
under steady-state (constant temperature and pressure)
conditions. By time t,, the adsorbate concentration on the
first portion of the bed has attained its equilibrium loading,
q That portion of the bed whose adsorbate loading equals
q:’ is called the Equilibrium Zone. The portion of the bed
where the adsorbate concentration remains at q. is called the
Unused Zone. Between these two zones is the Mass Transfer
Zone, where the solute is being transferred from the bulk
feedstream to the adsorbent. This normally has the “S-
shaped” curvature noted at time t + At. Under steady state
conditions, this Mass Transfer Zone moves steadily down the
column, retaining the same “S-shaped” curvature.
Eventually the leading edge of the Mass Transfer Zone

reaches the end of the column. This is termed the
breakthrough time, t = t,. Further flow results in the entire
bed reaching the equilibrium adsorbate concentration at time
t = t,.
Adsorption 91
Feed (G Y, ) Effluent (C. Y)

c
Time Analysis of Adsorbent Analysis of Effluent

(Adsorbenl toadmg VI. bed tenoh) (ComC+~~lrJteoon
of sorbrbtc
componcnl vs. tmc)
Figure 2.11. Schematic of mass transfer onto adsorbent and

solute concentration in the effluent (Reference 24).
Rather than analyze the adsorbate concentration change

in the bed, it is more usual to analyze the effluent from the
column as a function of time. The initial effluent is at a
solute concentration, cg, equal to zero, at t = t,. It is not
until t = t, that the solute concentration begins to increase.
The breakthrough concentration, however, does not

necessarily occur at tb. It occurs at a more arbitrarily set
time, t,, which may be either the maximum allowable or the
minimum detectable concentration of the solute in the
effluent.
The effluent concentration goes through the same S-

shaped increase as it reaches the equilibrium inlet solute
concentration as the adsorbate concentration went through
the S-shaped decrease. The relative length of these mass
transfer waves represents the relative strength of the
resistance to mass transfer.
This approach can be formulated into mathematical

equations for the simple estimation of adsorbent column
operation. It should be noted that the model requires that
the adsorbent bed is uniformly packed; that the feed rate,
feed temperature and the adsorbent temperature are constant;
that the feed temperature and the adsorbent temperature are
the same; that the flow is uniformly fed to the top of the
bed with no channeling in the bed; that the fluid does not
undergo a phase change; and that no chemical reactions
occur in the bed.
Under these conditions, the total amount of solute, m,

removed from the feedstream between time t and time t t At
is given by:
In = u (Ce - Co> Ag At (2.53)
where An is the cross-sectional area of the adsorbent bed.
As m kg of solute is adsorbed, the shape of the mass

transfer wave remains the same while its position changes.
The Equilibrium Zone has not been increased by a length AL:
In = (AL) AB pB (9 e - qo) (2.54)
Therefore,
AI, = (u AC At)&, Aq> (2.55)
During the steady state operation, the values of u, pu, AC

and Aq are fixed, so that uAC/(pn Aq) can be represented by
Adsorption 93
a constant U, which represents the velocity at which the

mass transfer wave moves through the bed. While the local
mass transfer rates may affect the shape of the front, the
velocity of the front under steady state conditions is only a
function of the space velocity, the bulk density of the
adsorbent and the boundary conditions.
At any time, t, the position of the breakthrough

concentration in the bed is given by:
Ls = ut (2.56)
At the breakthrough time, tt,, the equilibrium concentration

is a distance Lb from the end of the column, Lo. That is:
Lb = Ll tb (2.57)
Likewise, the equilibrium concentration reaches Le at the

time t,:
Lo = uts (2.58)
The length of the unused bed, L,, of the adsorption column

is given by:
L” = Lo - Lb = IJ (t S - tb> (2.59)
U can be eliminated to give:
L = Lo ( 1 - tb/ts) (2.60)
u
An alternate form of this equation is:
L = (2.61)
U Lo -
which allows the estimation of the unused portion of the

adsorbent when the adsorption process is stopped at tt,.
Figure 2.12 shows the change in the amount of adsorbent
used to capacity as the adsorption column is changed relative
to the length of the mass transfer zone. For processes
which are to remove impurities, the bed length should be
between 1.5 and 3 times the size of the mass transfer zone.
100
90
6-t
=’ 80
2 70
; 60
$ 50
; 40
4 30
z!
g 20
+
RELATIVE BED LENGTH (TO LENGTH OF A
STEADY-STATE MASS-TRANSFER ZONE)
Figure 2.12. Comparison of bed use as a function of

increasing bed length (Reference 24).
The particle size distribution of the adsorbent has been

shown (25) to be of importance primarily when film diffusion
contributes significantly to the rate of adsorption or for
short contact times (batch) or short column lengths (fixed
bed). Film diffusion contributions are significant mainly
when the particle size distribution is skewed toward smaller
particles. For particles with Gaussian distributions of
moderate standard deviations, the effect is quite small.
Nearly identical breakthrough curves are calculated (26)

for spherical, cylindrical and slab adsorbent particles with
the same surface-to-volume ratio when the column contact
time is either quite short or quite long. Some differences
are noted in the intermediate time domain. However, they
are of minor importance for engineering considerations of
commercial adsorption column design and operation since the
calculations always include margins for error.
Adsorption 95
Example 2.4
Figure 2.13 shows three breakthrough curves that have
different lengths in their mass transfer zones. If the
allowable concentration of the solute in the effluent is 0.2
C,, what is the degree of column utilization?
In the first case, the mass transfer zone is quite small

so that the column achieves a high degree (-90%) capacity
utilization. In the second case, the mass transfer zone is
longer and only 60% of the adsorbent bed utilized. The very
long mass transfer zone of the third case allows only 30%
utilization of the adsorbent bed before breakthrough is
reached.
BED VOLUMES TREATED
Figure 2.13. Column exhaustion curves for systems with

three different lengths of mass transfer zones.
2.3 TYPES OF ADSORBENTS
The usefulness of an adsorbent in a purification process

is a function of its composition, the presence and type of
functional groups at the surface, its porosity and surface
area, its degree of polarity and its relative
hydrophilicity/hydrophobicity. The adsorbents which have
been and will continue to be most useful in biotechnology
purifications are activated carbon, oxides of silicon and
aluminum, and crosslinked organic polymers. These materials,
their generally accepted surface functionality and surface
areas are listed in Table 2.4.
Table 2.4: Surface Structure and Area of Adsorbents
Adsorbent structure Specific Surface Area (m*/g)
/O\
Bone Charcoal
/;\ /f\ 60 - 80
Granular Charcoal
/O\ 500 - 200
/i\ /I\
Silica Gel ‘Oh

PII\ /I’\ 600
Zeolites 400
Polystyrene
Poly(methylmethacrylate) yH3 73 100 - 500

- t-CH, - F-CH2 -
COOCH3 COOCH,,
Bone chars, natural and synthetic, are sized granules

used almost exclusively for decolorizing sugar. Activated
carbons are sized granules derived from the pyrolysis of
wood or coal. Activated alumina is a synthetic, porous
crystalline gel formed into spherical pellets or granules.
Adsorption 97
Silica gel is a synthetic, porous non-crystalline gel formed

into spherical pellets or granules. Zeolite- type alumino-
silicates (molecular sieves) are unique in that their synthesis
produces a pore volume of almost 100% uniform size. The
most common synthetic macroporous polymer beads used in
adsorption are copolymers of styrene-divinylbenzene,
methacrylate-divinylbenzene or phenol-formaldehyde.
Although it is possible to analyze the chemical

constituents of these adsorbents, it is difficult to
characterize the chemical nature of their surfaces. In
general, polar adsorbents contain surface groups capable of
hydrogen bonding with adsorbed materials. This is probably
the single most important property of this type of adsorbent.
Non-polar adsorbents have no hydrogen bonding properties
and London dispersion forces become the most important in
the adsorption mechanism. Synthetic polymers may have
functional groups to enhance their adsorptive strength.
Since adsorption occurs on the surface of the adsorbent,

the total surface area is an important parameter. If the
surface of the adsorbent were completely uniform, the
amount of surface covered would be a function of the shape,
but not the size, of the solute. However, the capacity of
activated carbon to adsorb solutes decreases as the solute
size increases, indicating the fine pore structure inside the
carbon particles.
If the adsorbent did not have pores throughout its

structure to provide additional adsorption sites, the maximum
amount of solute that could be adsorbed would increase as
the particle size decreased. This has the practical limitation
caused by excessive pressure drop across a packed column of
the adsorbent when the particle size becomes too fine. The
surface area of a non-porous adsorbent is shown as a
function of particle size in Figure 2.14. Most adsorbents
used in industrial purification processes have surface areas
greater than 100 m2/g and particles ranging from 150 to
1500 microns in size. Therefore, a large portion of the
adsorbent’s surface area must consist of a pore network
inside each adsorbent particle. The size and uniformity of
the pores affect the adsorption of large molecules. The
effective surface area available for adsorption should be
expected to decrease as the molecular size of the adsorbed
solute increases.
a,01 _ I I I111111 I I I111111

0,001 O*Ol 081 1,O
SPECIFIC SURFACE AREA ( t?/c)
Figure 2.14. The surface area of a non-porous adsorbent as

a function of particle size at three different densities.
The surface area of adsorbents is usually measured by

degassing the adsorbent under high vacuum at elevated
temperatures, followed by the addition of a known quantity
of gas, such as nitrogen or helium. The final pressure of the
chamber containing the adsorbent is related to the surface
area of the adsorbent (S) using the Brunauer, Emmett and
Teller (BET) equation (27):
S(m2/s)= [ v 1’; - ‘) (‘O +,d’ - .“‘)]A aN (IO-” m/Aj2 (2.62)
where the first term is equal to V,, the volume of gas

adsorbed when the entire adsorbent surface is covered with a
complete monomolecular layer; v is the volume of gas
adsorbed; p is the measured pressured; p” is the vapor
pressure of the gas at experimental temperature; c is a
constant; A is Avogadro’s number; and aN is the area
occupied by a single gas molecule (16.2 A2 for nitrogen).
Abrams (28) has pointed out the danger in relying upon

high surface area (as measured by BET) in comparing resins
Adsorption 99
with inorganic or charcoal adsorbents for use in purification

processes. As Table 2.5 shows, a gel resin with a surface
area of only 1.4 m2/g had the highest relative adsorption.
The gel resin swells in water to form an open-lattice which
readily allows diffusion of the organic impurities from the
pond water into the particle for adsorption. BET experiments
on the dry particle would not have shown these lattice
openings. The relatively poor adsorption with the carbon is
most likely due to its fine pore size not allowing the organic
impurities to the interior adsorption sites. The adsorption
performance of the macroporous adsorbent resin is limited by
its lack of anion functional groups.
Table 2.5 : Influence of Surface Area on Relative Adsorptivity(28)

Surface Relative
Adsorbent Composition area (m*/g) adsorptivity
Nuchar CEE Carbon 740 1.0
Duolite S-30 Phenol-formaldehyde 128 1.3
ES-140 Styrene-divinylbenzene 110 2.4
Duolite A-7D Phenol-formaldehyde 24 3.4
Duolite A-30B Epoxy-amine 1.4 5.0
ES-111 Styrene-divinylbenzene 1.0 2.8
It is generally held that one should use the adsorbent

with the highest surface area having a suitable polarity, with
consideration for the size of the molecule to be adsorbed.
Since the average pore diameter of the adsorbent decreases
as its surface area increases, the adsorption of large
molecules in a reasonably rapid process necessitates the use
of adsorbents with larger pores and thus less surface areas.
2.3.1 Carbon and Activated Charcoal

Commercially available carbons are characterized by
having high surface areas which are covered with combined
hydrogen and oxygen. These surfaces are likely to consist of
reactive polar groups. It was suggested (29) in 1959 that the
surface of these carbons may have carbonyl, hydroxyl,
carboxyl and lactone groups. Measurements (30) on
commercial activated carbons showed that they had carboxylic
acid capacities ranging from 0.30 to 0.90 meq/g and phenolic
hydroxyl capacities from 0.05 to 0.07 meq/g.
The raw material selected for the preparation of

activated charcoal has a definite effect on the adsorbent’s
properties. Wood, coconut, lignite, bituminous coal,
subbituminous coal, petroleum acid and sludge activated coal
have been used as raw materials. Table 2.6 shows the test
parameter values for activated charcoal prepared from these
different materials.
Table 2.6: Characteristics of Activated Carbon Prepared from

Different Raw Materials(34)
Iodine Molasses cc14 Pore Volume

Raw Material Number Number Number (cc/g)
Wood 1230 470 76 0.57
coconut 1350 185 63 0.49
Lignite 550 490 34 0.23
Bituminous
Coal 900-1000 200-250 60 0.45
Subbituminous
Coal 1050 230 67 0.48
Petroleum Acid
Sludge Coke 1150 180 59 0.46
The tests used to arrive at these numbers are the

industry standards for carbons and activated charcoal. The
iodine number (31) can be roughly correlated with the
surface area of the charcoal, which is an indication of its
capacity to adsorb small to medium molecules. The molasses
number (32) can be correlated with the rate of adsorption of
large molecules such as occurs in decolorization processes.
The carbon tetrachloride number (33) is primarily of interest
for those using activated charcoal for vapor phase adsorption.
The higher the number, the greater the amount of an organic
vapor which can be expected to be adsorbed per unit weight
of charcoal. The determination of the pore volume as a
function of the pore size of the adsorbent is a useful
indication of the molecular size which may be adsorbed by
the charcoal. Figure 2.15 shows such a plot for charcoals
prepared from different raw materials with butane as the
penetrant.
Adsorption 101
.**..
: ..
: '.,- COCONUT
‘-// \+- BlTuMlN0~s
WOOD
4 5 6 7 8 9 10 20
PORE DIAMETER (ANGSTROMS)
Figure 2.15. Pore volume as a function of pore size for

carbons from different sources (Reference 30).
Higher molecular weight dissolved organic molecules

adsorb on activated charcoal in preference to lower molecular
weight molecules. Also, adsorption is favored for molecules
which are non-polar. Organic compounds such as methanol,
which are both polar and possess low molecular weight,
adsorb very poorly and are easily displaced by other organic
compounds. Aromatic, non-polar compounds such as toluene

are effectively adsorbed and held by the activated charcoal
until it is regenerated.
The pH of the water and the solute’s solubility in the
solvent are also important in its adsorbability on activated
charcoal. The presence of additional solutes will also affect
the adsorption characteristics for the removal of a specific
contaminant. In general, however, the molecular weight of
the solute is the primary factor which controls the feasibility
for its adsorption. Molecules with three or more carbon
atoms may usually be removed by adsorption on activated
charcoal.
There is usually an upper limit of molecular weight

above which adsorption decreases with further increases in
molecular weight. This happens as the solute approaches
polymer size (molecular weight > 600). At this point, the
molecule is so large that its diffusion into the internal pore
structure of activated charcoal is inhibited.
In general, bituminous coal based granules have a high

surface area and small pores which make them suitable for
adsorbing lower molecular weight non-polar solutes. Lignite
based granules have a much lower surface area but larger
pores which make them better for larger molecular weight
solutes. Table 2.7 lists several of the common carbon and
activated charcoal adsorbents.
New developments continue to be made in the

processing of raw materials to obtain modification in the
structure of the charcoal so that it will be more suitable for
a specific application. Table 2.8 (34) shows the range of
values coal-based adsorbents can have. It should be noted
that the primary difference between the charcoals used in
liquid or vapor applications is in the diffusion rates of the
solutes through the charcoals which are possible. In
biotechnology purifications, activated charcoal will be used
almost exclusively in liquid media.
2.3.2 Silica Gels, Aluminas and Zeolites

Silica gels can be considered to be an amorphous,
inorganic condensation polymer of silicic acid. Silica gel, a
Adsorption 103
Table 2.7 : Commercial Activated Carbons
Surface Area
Company Brand Name Source (m'/g)
American Norit
Company Norit (various) Wood 700 - 1400
Atlas Chemical
Industries Darco
s 51 Lignite 500 - 550
G 60 Lignite 750 - 800
KB Wood 950 - 1000
Hydro Lignite 550 - 600
Calgon Corporation Pittsburgh Bituminous Coal

PCC SGL 1000 - 1200
PCC BPL 1050 - 1150
PCC RB 1200 - 1400
PCC GW 800 - 1000
Union Carbide Columbia Coconut Shell

CXA/SKA 1100 - 1300
AC 1200 - 1400
G 1100 - 1150
Westvaco Nuchar Pulp flill Residue

Aqua 550 - 650
C 1050 - 1100
(various) 300 - 1400
Witco Witco Petroleum Acid

Sludge Coke
Table 2.8: Comparison of Activated Carbon Used in Different

Applications (34)
Iodine Molasses ccl, Pore Volume

Application Number Number Number (cc/g)
Liquid Phase 1080 251 65.6 .472
Vapor Phase 1065 209 63.4 .447
Evaporative Loss 1200 400 90.1 .62
typical polar adsorbent, contains four main types of

functionality at its surface:
The formation of hydrogen bonds between the surface

and the adsorbed compound contributes substantially to the
adsorption process. Some surface groups are weakly acidic
and therefore can enter into ion exchange reactions at high
PH. At high activation temperatures, silica surfaces sinter,
becoming less active. This is most likely due to chemical
changes in the surface groups.
Most commercial silica gels are either of the narrow

pore type (average pore diameter about 2.5 nm, specific
surface area about 700 m2/g and pore volume about 0.4 cc/g
or the wide pore type (average pore diameter about 14 nm,
specific surface area about 300 m2/g and pore volume about
1.1 cc/g). Sorbents capable of hydrogen bonding are good
eluents for regenerating silica gels.
The surface of the alumina adsorbents contains hydroxyl

groups of similar structure to the silica gels. However, the
potential for ion exchange reactions is greater with alumina
than with silica gel. Calcined alumina is strongly basic and
must be acid washed to produce a neutral or slightly acidic
surface. The strength of the adsorption on alumina for
aliphatic compounds containing various functional groups has
been shown (35) to increase in the following order: -S-, -0-,
-C=N, -CO 2-, -CO-, -OH, -N=, -NH,, -SO, -CONH,.
Zeolites are characterized by ordered crystal

structures. They are hydrated crystalline metal alumino-
silicates and possess uniformly small-sized pores leading. from
the exterior surface to an internal, three dimensional
cagework formed of interconnecting silica and alumina
tetrahedra. This formulation combines the surface
functionality of the silica gel and of the alumina adsorbents,
resulting in an adsorbent that has a strong affinity for polar
or polarizable molecules.
Since the zeolites have the uniform, small pore size

Adsorption 105
throughout each particle, most of their surface area is inside

the particle. Zeolites are often called molecular sieves
because the pores may be small enough to allow certain
molecules in solution to pass through the pores while larger
molecules are excluded, thus separating one compound from
others.
The affinity of zeolites for polar or polarizable

molecules lead to their use for the removal of water, carbon
dioxide and hydrogen sulfide from liquids (36). Zeolites have
been used for the dehydration of chlorinated solvents,
refrigerants, acetylene and aromatics. They have been used
for the removal of COs from ethylene and liquid propane gas.
Butane and liquid propane gas have been desulfurized with
zeolites.
Zeolites, along with silica gels and alumina, have been

used extensively in the immobilization of enzymes for
biotechnology applications (37-39). However, the bulk of the
purification applications with zeolites are in the treatment of
gas streams. The purification of biological fluid streams
themselves is usually best accomplished with an adsorbent
with a larger pore size and one which is more stable over a
wider pH range.
Table 2.9 shows the properties of several of the

commercial silica gels, aluminas and zeolites. New families of
zeolites have been developed in the mid-1970’s which have a
hydrophobic surface. These new materials prefer lipophilic
over a more polar compounds. In 1984 Union Carbide (36)
announced a new family of microcrystalline porous structures
based on aluminum and phosphorus. It is possible that these
new materials may find applications in the purification of
biotechnology products.
2.3.3 Organic Polymer Adsorbents

Ion exchange resin manufacturers have also produced
macroporous resins which may be used as adsorbents in
purification applications. Although these resins may possess
very weakly acidic or basic groups, in this chapter we will
only consider their use as adsorbents, not as ion exchange
resins. They are generally used in neutral or slightly acidic
solutions where little or no ion exchange activity is
displayed.
Table 2.9: Properties of Some Commercial Aluminas, Silica Gels

and Zeolites
Tradename Specific Surface Mean Pore
Area (ma/R) Diameter (A )
ICN Alumina A 4.5 155-200 ICN Biomedicals

ICN Alumina B 10.0 155-200 ____ ICN Biomedicals
ICN Alumina N 7.5 155-200 ____ ICN Biomedicals
A-305 CS Alumina _-- 325 700 Kaiser Chemicals
Celite Silica 7 __- 200 Manville
Matrex Silica 500-600 60 Amicon
Matrex Silica 7 100 Amicon
Matrex Silica 7 _-- 275 Amicon
Nova-Pak Silica 120 60-100 Waters
Resolve Silica 200 60-175 Waters
p Porasil Silica 7 300 SO-300 Warers
Nucleosil 50 500 50 Macherey-Nags1
Nucleosi 1 100 7 300 100 Hacherey-Nagel
Zeolite ZSM-5 ___ ___ 5.4 x 5.6 Mobil
Zeolite A 5 Union Carbide
Zeolite X ___ 7-6 Union Carbide
The surface active groups were shown in Table 2.4.

The more complete chemical structures are shown in Figure
2.16. The non-functionalized polystyrene adsorbents, made
from styrene and divinylbenzene, are extremely hydrophobic,
with a dipole moment of only 0.3 Debyes. The
polymethacrylate structures are more hydrophilic, with dipole
moments of approximately 1.4 Debyes. The phenol-
formaldehyde resins and the functionalized versions of the
polystyrene and polymethacrylate resins are very hydrophilic,
with dipole moments greater than 2.0 Debyes. Table 2.10
shows the physical properties of commercial adsorption resins.
Most of the adsorbent resins with functionality have

primary, secondary, tertiary or quaternary amine groups since
these groups are useful in removing the organic
contaminants, most of which are acidic, from sugar extracts
and aqueous feed streams. However, the affinity of aromatic
acids or salts or aromatic acids for quaternary amines is so
great that their adsorption is practically irreversible (40).
In general, the non-polar resins are very effective in

adsorbing the non-polar solutes form polar solvents.
Likewise, highly polar resins are effective in removing polar
Adsorption 107
-CH-CH,-
.
STYRENE-DIVINYLBENZENE
-CH-CH,-
~~ETHYLHETHACRYUTE-
DIVINYLLENZENE
-CH-CH,-
PHENOL-FORMALDEHYDE
,+27542 ~5~~2
CH2
I
Figure 2.16. Chemical structure of non-functionalized

synthetic organic resin adsorbents.
solutes from non-polar solvents. The binding energies of the

non-functionalized resin adsorbents are lower than those of
activated charcoal for the same organic molecules (4 1).
Enthalpies of adsorption are only -2 to -6 kcal/mole for
compounds like butyric acid, sodium naphthalene sulfonate or
sodium anthraquinone sulfonate on large pore, non-polar
adsorbent resins. Although this decreases the efficiency of
adsorption, it also means that the regeneration of the resins
is easier to accomplish.
Table 2.10: Physical Properties of Synthetic Organic Resin Adsorbents
L?
3.
Typical Surface 3
Physical Capacity to Moisture Area g
Trade Name tlatrix Functional Group FOtlll HCl (equiv/l) Retention (pz/B) P
8
Duolite S-761 Phenol-Formaldehyde Phenolic hydroxyl Granules 40 150 - 300
c3
8
Duolite ES-33 Phenol-Formaldehyde Tertiary Amine C~~llUleS 0.7 58 100 - 200
&
Duolite S-37 Phenol-Formaldehyde Secondary and Granllles 1.4 55 120 9
Tertiary Amine
E
Duolite ES-111 Styrene-DVB Quaternary Amine Beads 1.0 67 1.0 3
Styrene-DVB -_
E!
Amberlite XAD-2 Beads 300 0
B
Amberlite XAD-4 Styrene-DVB __ Beads __ 780 &
Amberlite XAD-7 Carboxylic Ester __ Beads __ 450 g
Amberlite XAD-8 Carboxylic Ester __ Beads __ 140
XFS-4022 Styrcne DVB Beads 80 - 120
XX-4257 Styrene DVB Beads _- 400 - 600

Adsorption 109
2.4 LABORATORY EVALUATION OF ADSORBENTS
Several tests (iodine number, molasses number, surface

area and pore distribution) were mentioned in the previous
section as useful in the characterization of adsorbent
materials. While these tests are valuable in assessing the
probability that an adsorbent is suitable for a specific
purification operation, laboratory evaluations of the selected
adsorbents with the fluid to be purified are necessary.
Three main types of experiments are possible for

measuring adsorption of solutes from liquids by solid
adsorbents. The essence of the first, which is by far the
most frequently used, is to bring together known weights of
solid and liquid, allow equilibrium to be established at a
given temperature, and determine the resulting change in
composition of the liquid.
The simplest experimental technique is to seal the liquid

and adsorbent into clean, dry test tubes. Freezing the
mixture in a carbon dioxide slush bath immediately before the
tube is sealed will minimize the loss of volatiles. The tubes
are then shaken in a constant temperature bath until
equilibrium is reached. The time required for this stage
varies considerably depending on the nature of the
components of the system being studied.
During the adsorption measurement, the liquid must be

in constant motion and the adsorbent must be continuously
washed by the liquid. A reciprocating shaker may be used
provided the frequency is sufficiently low so that mechanical
breakdown of the adsorbent does not occur. For dense
solids, tumbling is often more effective. The tumbling rate
should be adjusted so that the adsorbent will fall completely
through the fluid each half cycle. If the adsorbent is in the
powder form, it will be necessary to centrifuge the liquid
suspension to separate the adsorbent from the liquid.
A second method (42) of measuring adsorption from a

liquid solution on a solid adsorbent is shown in Figure 2.17.
In this technique, the liquid containing the solute is
circulated over the adsorbent particles while changes in
concentration are monitored by either refractometry or
spectroscopic changes. The experiment may be carried out in
a closed system, with solutions that have been purified and
degassed. It also allows a single adsorbent sample to be

treated in a reproducible manner with successive experiments.
The experimental temperature can be changed over a wide
range in a simple experiment with this technique.
PI
wc-____l I
Figure 2.17. Circulation method for determining adsorption
from solution. Solution is circulated by pump Pl over
adsorbent in cell A in the temperature controlled unit TS and
through one arm of a differential refractometer R. Solution
of the initial composition is circulated through the reference
cell W of the refractometer (Reference 42).
Somewhat related to that technique is a

chromatographic method in which a solute at a given mole
fraction is passed through a column of the adsorbent
particles and the concentration of the effluent is monitored.
The concentration of the effluent plotted as a function of
time normally gives a curve like the one shown in Figure
2.18. The experiment is complete * when the effluent
concentration of the solute returns to its concentration in
the feed stream. If, beyond a sharp rise, the concentration
of the solute in the effluent becomes constant, the system is
said to have reached equilibrium. The adsorption isotherm
can be plotted from a series of such curves. The method
can also be extended to deal with the simultaneous
adsorption of two or more solutes. For precise results, it is
necessary to correct for the dead volume in the column and
for any variation in the molar volume of the solution as its
concentration is changed (43).
Adsorption 111
Co - c
ONCENTRATION
-I_
OF I”= EEDSTREA
z l
0
2
& CYCLE (COLUMN EXHAUSTED)
5
s
CONCENTRATION OF IMPURIUL_
I,N EFFLUENT
y
VOLUME OF FLUID TREATED PER UNIT ~/EIGHT OF ADSORBENT
Figure 2.18. Concentration of solute in effluent as a

function of time.
Example 2.5
Figure 2.19 shows the concentration of the effluent
from an adsorption column using the same amount of
adsorbent and the same flow rate but with changes in solute
concentration in the feedstream. While the column
dimensions (1 cm x 10 cm) were too small for gathering
scale-up information, the data can be used to construct the
adsorption isotherm shown in Figure 2.20.
Figure 2.21 shows an adsorption measurement technique

which is useful for modeling the behavior in a stirred
suspension of the adsorbent in the liquid solution with the
solute. The relative surface excess is measured and
represented by the equation:
(2.63)
Lx;
where dn, is the amount of solute 2 which has to be added

together with a mass dm of solid adsorbent to maintain a
constant bulk liquid concentration. Powdered or particulate
adsorbent material can be added in successive amounts along
with the solute to maintain a constant mole fraction.
Normally the isotherms obtained should be independent

of the ratio of the weights of adsorbent and solution used.
If this is not so, the presence of small quantities of
impurities should be suspected. There may be other reasons
as well. As an example, in the adsorption from aqueous
solutions into anodized aluminum, the slight dissociation of
BED VOLUMES OF EFFLUENT
Figure 2.19. Concentration of sodium cholate in effluent as a

function of time for different concentrations of the solute in
the feedstream.
-5.0 _
-4,o -
p
‘;; -3,o -
d
6
F
&
3 -1.0
0 I 1
0 -1.0 -2.0 -3.0 -4.0 -5.0

LOG CONCENTRATION ~LES/LITER)
Figure 2.20. Adsorption isotherm for sodium cholate on

porous polystyrene-DVB adsorbent, based on data from Figure
2.19.
Adsorption 113
A SLURRY
B,B’ THERMOSTATTED WATER
.3 I
”
C MAGNETIC STIRRER E
D FRIT F fi
E PH ELECTRODE 1
F SEPTUM
G :;;;fB’N’ 43 tit
P Cp:;CpULATING
S UVhISIBLE
DETECTOR
Figure 2.21. Alternate circulation adsorption method in

which the adsorbent is used as a suspension rather than in a
packed column (Reference 98).
the alumina film in the water causes this type of behavior

(44).
In the adsorption of solutes from the liquid phase onto

solid adsorbents, substances present in low concentration are
often adsorbed preferentially. Therefore, the presence of
impurities in the liquid phase may have an effect on the
measurement quite out of proportion to the quantity present.
A practical check on the presence and effect of all the
impurities in the fluid to be treated is valuable in avoiding
spurious results. Quantities of adsorbent and, separately, of
each impurity should be shaken together for the length of
contact time to be used in the adsorption experiments. Then
examine the fluid by the analytical technique chosen for the
experiment to insure that any change is within the
experimental error of the technique.
Many efforts are now focused on ways to correlate and

predict the performance of adsorbents in multicomponent
systems. The models are often heavily dependent upon
experimental data derived using the actual fluid to be
treated. Industry has usually simulated full scale column
performance by using columns of the same height but with
reduced diameters. The time required to carry out a test

using such a column is of about the same duration as the full
scale operation. Rosene (45,46) has reported column
simulation techniques using much smaller columns which
permit engineering design in as little as one twentieth of the
testing time, with corresponding reductions in the feedstream
and adsorbent required for the experiment.
2.5 ADSORBENT OPERATIONS
2.5.1 Pretreatment Operations

Solid adsorbents require some pretreatment before use.
The usual treatments are for removal of water vapor or
surface impurities or for improving the wetting of the
adsorbent surface by the solvent which contains the solute.
These pretreatment operations are to insure that a
reproducible behavior will be exerted by the surface of the
adsorbent during the adsorption process.
When adsorbents must be dried before use, the drying is

usually carried out by heating for about two hours at 110 to
120°C. It has been claimed (47) that this temperature is not
high enough for complete drying of all adsorbents. However,
the use of a higher temperature may result in irreversible
changes in the adsorbent, such as sintering, change in
crystalline form or change in the functional groups on the
surface. The last effect is especially seen for polymer resin
adsorbents with strong base functionality. Polymer resin
adsorbents are usually only dried before surface or pore
characterization experiments and not before use in adsorbent
operations.
Many adsorbents, particularly activated carbons, contain

impurities which may be leached from the adsorbent to
contaminate the solution being purified. The pretreatment
needed to overcome this problem must use a solvent at least
as powerful in leaching ability as the solution being purified.
Activated carbons, which may contain several percent of
inorganic oxides or carbonates, are usually washed with a
mineral acid to remove these impurities. A volume of 1N
sulfuric acid equal to the volume of adsorbent (one bed
volume) is passed through the adsorbent at a temperature
between 50-60°C. Next, two bed volumes of water, at the
same temperature, are passed through the adsorption column.
Adsorption 115
Finally, ambient temperature water is run through the column

until the effluent pH is above 4.0. Figure 2.22 (48) shows
the difference between the adsorption which occurs with
treated and untreated adsorbents.
1.3732 -
1.3716
0 5 10
ACETIC ACID : CHARCOAL (w/w)
Figure 2.22. Effect of treated vs. untreated charcoals on the

refractive index of acetic acid. The refractive index for
pure acid is 1.3716 (Reference 48).
The non-polar synthetic resin adsorbents are not easily

wetted by aqueous solutions because of their hydrophobic
nature. The pretreatment of these resins consists in passing
5 to 7 bed volumes of methanol through the beds of resin,
followed by rinsing with 4 to 5 bed volumes of deionized
water.
It may also be necessary to pretreat the solution

containing the solute before putting it in contact with the
adsorbent. This pretreatment is primarily concerned with the
removal of excess amounts of suspended solids, oils and
greases. The presence of these materials in excess amounts
would coat the adsorbent particles, thereby dramatically
reducing their adsorptive effectiveness. Suspended solids,
including bacteria and yeasts, in amounts exceeding
approximately 50 mg/L should be removed prior to applying
the fluid to the adsorption column. Oil and greases in

concentrations above 10 mg/L should not be applied to
adsorbents in either batch or column operations.
Synthetic organic polymer adsorbents are subject to de-

crosslinking if oxidative materials are present in the feed
solution or the eluant. These should be removed by
deaeration or by passing the solution through a sacrificial
carbon bed first.
Filtration, oil flotation and chemical clarification are

the pretreatment operations commonly employed. The
adjustment of the solution pH may also be used to enhance
adsorption efficiency. Dissolved organic substances adsorb
best at the pH which imparts the least polarity to the
molecule. Weak acids, such as phenol, show better adsorptive
properties at lower pH values, while amines usually adsorb
better at higher pH values.
2.5.2 Batch Operations

In the batch operation, measured quantities of the
adsorbent are added to the fermentation broth or filtrate in
one or several stages. The adsorbent is left in contact with
the fluid for the appropriate length of time, usually with
agitation. Then the suspension is screened to isolate the
adsorbate-adsorbent complexes. If the object is to recover
the adsorbate, the adsorbent is usually transferred to a
column through which an eluting solvent is passed. The
fractions of the eluate containing the desired product are
collected for further purification and/or concentration. If
the adsorbate is an unwanted impurity, after being isolated
by screening, the adsorbate-adsorbent complex is discarded or
regenerated, depending on the economics of the operation.
Example 2.6
An example of this method of purification was given by
Ores and Rauber (49). To a fermentation broth of 4.5 liters
with 510 units of the antibiotic bacitracin per cc were added
1.5 liters of a porous polystyrene resin. The suspension was
stirred together for six hours and then screened to separate
the resin. The resin was placed in a column and eluted with
3.9 L of methanol. Water (0.55 L) was added to the eluate
which was then distilled at 20°C under vacuum. The result
was an 86% recovery of the bacitracin (1.15 L contained
1,740 units/cc).
Adsorption 117
An important advantage of the agitated slurry systems

in adsorptive operations is that the resistance to mass
transfer of the solute through the slurry to the adsorbent
narticles can be lowered by increasing the agitation power
input. The mass transfer coefficient in such solutions
correlates well with agitation power according to the
equation (50):
0.25
33
(2.64)
where Sh is the Sherwood number; SC is the Schmidt number;

d is the particle diameter; i- is the power input per unit
mass; Y is the kinematic viscosity; and C, is a constant.
Since T- is proportional to the third power of the

agitator speed for turbulent flow, the mass transfer
coefficient is seen to vary as the 3/4 power of the agitator
speed. By knowing the mass transfer at one agitator speed,
the mass transfer at a second is given by:
k2 = kl (n2/n1)0’75 (2.65)
where k, and k, are the mass transfer coefficients when the

agitator speeds are n1 and n2 rpm’s, respectively.
The mass transfer coefficients were shown (51) to vary

with position and impeller characteristics. Values for the
mass transfer coefficient in the plane of the impeller can be
described by the equation 15 x 10m5 (nD2/2r)2/s and 11 x
10-s (nD2/2r)2/3 for six bladed disk turbines and curved
bladed turbines, respectively. This is shown in Figure 2.23.
2.5.3 Column Operations

Column operations are more common for industrial
adsorption than batch operations. The adsorption column
holds the adsorbent particles through which the fluid with
the solute or solutes flow. Either pressure or gravity can be
used as the driving force for fluid flow in a downflow
operation, The untreated fluid comes into the top of the
column. If the column has a large diameter, the fluid may
go through a distributor to insure a minimum amount of
unused adsorbent or “channeling“ of the fluid, Contact time
for the fluid in the bed is determined by the requirements
10
7-
CURVED BLADED
5’ TURBINE
SIX KLADED
DISK TURBINE
.05 .07 .l .2
ND2/2R
Figure 2.23. Correlation of solid-liquid mass transfer

coefficient in the plane of the impeller. The different
symbols represent types of contactor and impeller speeds
(Reference 51).
for the system. Flow rates or space velocities for

fermentation filtrates with small amounts of solute to be
adsorbed are usually between 2 l/s-m3 to 16 l/s-m’. The
treated fluid is collected in a similar distributor device at
the bottom of the column and discharged to the next
adsorption column or treatment unit. Figure 2.24 (52) shows
examples of pressure and gravity flow adsorption units.
Adsorption 119
REACTIVATED BACKWASH
WATER OUTLET
CARSON BED
TlLE URFACE WASHER
FILTER
BLOCKS
TREATED CARBON REMOVAL
PIPE LATERALS
FLUlD OUTLET/
WITH
BACKWASH INLET
COLLECTORS GRADED
GRAVEL
SPENT CARBON OUTLET
Figure 2.24. Pressure-type and gravity type fixed bed

adsorber units (Reference 52).
The shape and length of the adsorption zone in a

column determine the number of packed bed adsorbers which
should be used to attain the desired degree of solute
removal. This is shown in Figure 2.25 (53) for one, two and
four columns operating in series. The shape of Curve A
indicates that the mass transfer zone is short, that is,
saturation of the adsorbent occurs shortly after the allowable
concentration of the solute in the effluent is reached. When
the effluent has reached the allowable concentration with
respect to the solute, it is said to have reached
“breakthrough.” Curves B and C in this figure show
increasingly long mass transfer zones. A second or more
adsorption columns are needed to purify adequately the
effluent while utilizing more fully the adsorptive capacity of
the initial column.
An alternate technique for the purification of a system

with an intermediate mass transfer zone using a single
packed column is to collect the effluent in a holding tank.
Then the overpurified effluent from the earlier part of the
cycle can be mixed with the underpurified effluent from the
later part to achieve an effluent with the desired purity.
This technique is appropriate when impurities are being
removed but should not be used when the adsorbate is the
desired product.
Another way of mixing underpurified and overpurified

effluents without the use of holding tanks is to connect two
I -1
I I
cO I I
L-1
t
c.5
ii .
":
ii! A
I!2
C
BED VOLUMES TREATED
Figure 2.25. Effect of shape and length of the adsorption

zone on the number of columns needed to obtain both a high
degree of adsorbent utilization and the desired solute removal
(Reference 53).
Or more adsorption columns in parallel and then begin

feeding the untreated fluid to them at evenly spaced
intervals. Then, when one column is discharging
underpurified effluent late in its exhaustion cycle, a different
column would be discharging overpurified effluent since it
would be early in its exhaustion cycle. Discharge manifolds
need to be arranged such that appropriate effluents are
Adsorption 121
connected to achieve the necessary purity. This packed bed

arrangement is shown in Figure 2.26.
IN
Figure 2.26. Column operation in a down-flow, parallel mode

(Reference 54).
There are a number of other column equipment

configurations. The three most common types of column
adsorber systems are shown in Figure 2.27.
’ IN ’
BS:‘.‘. . ::
.~‘.::“’
,..f.::
:‘.... ::...
IIOVINGBED DOWN FLOW IN SERIES UPFLOW-EXPINDEDIN SERIES
Figure 2.27. Additional adsorption column configurations

(Reference 54).
In the packed moving bed adsorber, the untreated fluid

flows upward through the bed and is collected and removed
at the top of the bed. Periodically, a portion of the
adsorbent is removed from the bottom and fresh adsorbent is
added to the top of the column. The volume of adsorbent

removed at one time is between 2 and 10% of the total
adsorbent volume. It is necessary to have a column height
to diameter ratio of 3 to 1 in order to facilitate plug flow
during the adsorbent removal portion of the cycle. The
bottom and the top of these columns are 67” and 35” cones,
respectively. Peripheral ring headers are located in these
cones for feeding the untreated fluid and for removing the
effluent. Well screens, or septs, are submerged in the top of
the adsorbent to retain the adsorbent particles while passing
the treated fluid. However, the movement of the adsorbent
in these systems can add to operational problems and will
often result in the loss of friable adsorbents due to attrition.
As was mentioned during the section on pretreatment,

packed column operations require that the amount of
suspended solids in the feedstream should not exceed
50 mg/L. Packed columns also require that the adsorbent
particle size be greater than 300 microns so that the
pressure drop through the column is not excessive. Fluidized
or semi-fluidized beds of adsorbents may be used to
circumvent these limitations.
The feedstream in fluidized systems enters at the

bottom of the column at a flow rate 10% to 100% greater
than in packed bed systems. The adsorbent bed expands in
proportion to the flow rate and in inverse proportion to the
particle size of the adsorbent. Fluidized bed operations take
advantage of small adsorbent particle size while avoiding the
problem of excessive pressure drop which would occur in
packed beds with similar sized particles.
Fluidized beds also provide good contact between the

fluid and the adsorbent because of the high degree of mixing
and high mass transfer coefficients which the small particles
allow. This is shown in Table 2.11 (55) where the effect of
particle size on the mass transfer rate is illustrated.
However, because of the high degree of mixing, plug flow
does not occur with this technique. The lack of plug flow
and the short residence time mean that fluidized adsorbent
operations do not remove all of the solute. This limits the
use of fluidized beds to applications which require some
removal of impurities but not the complete removal of
solutes.
Adsorption 123
Table 2.11: Effect of Carbon Adsorbent Particle Size on

Mass Transfer Rate with a 200 ml/mm Flowrate(55)
Particle Size Mass Transfer Rate
Solute (micron) (min-‘)
Astrazone Blue 150 - 250 0.17
Teflon Blue 150 - 250 1.06
Teflon Blue 250 - 355 0.98
Teflon Blue 355 - 500 0.58
Teflon Blue 500 - 710 0.40
Semi-fluidized beds combine the features of both

fluidized beds and packed beds in a single column. Hsu and
Fan (56) have evaluated the usefulness of semi-fluidized beds
in biochemical processing. In this technique, the fluid bed is
compressed by placing a screen or sieve in the column to
form two distinct regions of packed and fluidized particles.
The time before breakthrough is comparable for packed beds
and semi-fluidized beds; it is much longer than for fluidized
beds. However, the flow rate that is possible for the same
pressure drop is much greater for the semi-fluidized bed than
for the packed bed. This is shown in Table 2.12. (57)
Design of the feedstream inlet systems for semi-fluidized and
fluidized adsorption columns is critical since the distributors
must be able -to pass suspended solids, but must retain the
adsorbent particles.
Table 2.12: Effect of Fluidization on Flow Rate for

Specific Pressure Drops (57)
Hydraulic Loading Rate (m/min)
Degree of Fluidization
Adsorbent Pressure Drop Packed Bed 25’6 so% 75x
F400
Carbon 0.3
0.6 .24 .39 .55 1.05
APC
Carbon 0.2 .12 .15 .23 .35
0.4 .23 .27 .40 .57
0.6 .38 .43 .60 .92

1% Separation and l’tuification Techniques in Biotechnology
Tavares and Kunin (58) have reported on the use of a

very shallow bed of finely divided adsorbent particles to
remove impurities from fluids. This technique combines
filtration and purification in a single unit operation. A
powder is applied to a retaining screen so that the adsorbent
thickness is between 0.5 and 1 cm. The feedstream is passed
through several of these adsorbent layers in series to achieve
the desired purity. Once the adsorptive capacity has been
completely utilized, the layer is removed and discarded. This
technique relies upon the increased efficiency of the fine
size of the adsorbent, the savings in regeneration costs and
the lower capital investment associated with pre-coat filter
equipment to compete with typical adsorbent column systems.
Example 2.7
Figure 2.28 shows the effect of changes in bed height
for fixed column adsorption. In this case, the particle size
was 600 microns; the concentration of the solute in the
feedstream was 500 mg/dm3; the flow rate was 2 cm3/s; the
equilibrium adsorption was 0.2 g/g and the effective pore
diffusivity was 2 x 10-10m2/s .
0 40 80 l20 160
VOLUME TREATED (DII~)
Figure 2.28. Effect of changes in bed height for fixed

column adsorption (Reference 18).
Adsorption 125
If it were required that the solute concentration in the

effluent remain below 0.1 Co and there was 50 dms of
fermentation broth to be treated, the column bed height
required would be 25 cm.
2.5.4 Regeneration Operations

The adsorbent may be used once and then discarded.
More often, the used adsorbent is regenerated for reuse.
When adsorbent is regenerated, two or more beds of the
adsorbent are used with a manifold valving arrangement
which allows continuous processing of the treated fluid.
As was seen with the adsorption isotherm in the earlier

section, adsorption loading increases as the concentration
increases and decreases with increases in temperature. The
regeneration or desorption portion of the operating cycle
utilizes some variation of this behavior. Regeneration may
be accomplished by a thermal process, a displacement
adsorption process, an inert stripping process or a
combination of these processes.
Thermal regeneration is a very common practice with

adsorbents, particularly with granular activated carbon, which
have been used for the removal of organic species.
Temperatures between 870-980°C are normally used in rotary
kilns or multiple-hearth furnaces. Typically, the adsorbent
spends 5 minutes in a drying zone, 10 minutes in a charring
zone and 15 minutes in a reactivation zone. The oxygen
content of the atmosphere is held at a low level to allow
oxidation of the adsorbed organic impurity rather than of the
activated carbon. Carbon losses per cycle are normally
between 2 to lo%, depending on the size of the regeneration
system. Large systems are designed to operate continuously
which allows more efficient control for lower losses.
Schematics of a multiple-hearth furnace and a rotary

kiln are shown in Figure 2.29 and Figure 2.30, respectively
(59). The multiple-hearth furnace is said to have better
control of temperature and residence time, better control of
heat input, more uniform regeneration, lower temperature
gradient and less space requirements than the rotary kiln
units. The primary disadvantage of the multiple-hearth
furnaces are their higher capital costs and, in some cases,
higher carbon losses compared to the rotary kiln furnaces.
CARBON IN I 1
HEARTH
RABBLE ARM
RABBLE TEETH
’ BURNERS AND
STEAM INLETS
. .
CARBON OUT
Figure 2.29. Multiple hearth furnace (Reference 59).

Adsorption 127
GIRT GEAR
CARBON
Figure 2.30. Rotary tube regeneration furnace (Reference

59).
Whichever thermal regeneration equipment is used, it is

necessary to transport the adsorbent between the columns
and regeneration equipment by water slurry. Various
centrifugal or diaphragm pumping arrangements can be used.
Alternately, hydraulic or pneumatic pressure could push the
slurry. The bends in the piping should have a wide radius to
reduce adsorbent attrition. About 0.5 to 3.0 kg of adsorbent
may be transported with one liter of water, with the exact
amount depending upon distance and elevation considerations.
The thermal regeneration of activated carbon also has

an effect on the surface area of the adsorbent. As Figure
2.31 (60) shows, even five regenerations cause a significant
reduction in surface area. This occurs because the ash that
is picked up during the adsorption phase of the cycle plugs
small pores and is not burned off during regeneration. The
high temperatures of regeneration are also likely to result in
the formation of inorganic metal oxide residues which build
up as ash in the carbon. An acid wash must be used to
remove this ash to reduce fouling of the adsorbent.
Thermal regeneration at more moderate temperatures by

passing a purge gas or steam through the adsorbent column
may be used in those applications where the adsorbent has
been used for solvent recovery or for drying applications.
The highest temperature reached in such regenerations is
usually less than 200°C for carbon and 300°C for molecular
sieves (6 1). This type of regeneration is more appropriately
128 Separation
1,200
\\
and Purification Techniques in Biotechnology
I
1,000
NOUS COAL CARBON

* 800
”
5 600 \
\
2 \
.
LIGNITE CARBON
200
0 I I I I I
0 2 4 6 8 10
REGENERATION CYCLES
Figure 2.31. Decrease in surface area with use and

regeneration (Reference 60).
called inert stripping. The regeneration of the hydrophobic

zeolite, ZSM-5, which had been used to adsorb ethanol from
a water-ethanol mixture, made use of this technique (62).
The ethanol was displaced from the adsorbent with the inert
gas CO,, which was later separated from the ethanol by
reducing the pressure.
In the semi-continuous adsorption system shown in

Figure 2.32 (63), a hydrophilic zeolite is used to remove
water from an ethanol-water mixture. Here, after a column
has been completely utilized for water adsorption, the column
is drained of liquid which is returned to the feedstock tank
since the solute has not been removed from this liquid.
Then hot air is blown through the column to remove the
ethanol-rich phase which clings to the adsorbent particles.
After this is removed, the temperature of the bed of
adsorbent is increased to about 260°C as the water vapor is
driven from the micropores of the adsorbent. When no more
Adsorption 129
water vapor comes from the column, cool air is passed

through the column to cool it to 80°C before the next
adsorption cycle begins.
ir
I e-l r-l
A-E ADSORPTION
FEEDTANK
COLUMNS 8
9
BLOWER
HEAT EXCHANGER
L
: FEED LINE 10 HOT AIR LINE FOR REGENERATION
3 LINE CONNECTING BEDS 12 CONDENSER
4 DRAIN TANK 13 BLOWER FOR COOLING AIR
5 DRAIN LINE 14 COOL AIR LINES
6 PRODUCT TANK 15 EXHAUST AIR LINES
7 PRODUCT Ll NE
Figure 2.32. Semicontinuous adsorption system described in

Reference 63).
When the adsorbent is a synthetic organic polymer,

regeneration of the adsorbent is usually carried out by the
elution of the adsorbate with a solvent that has a lower
solubility parameter than the adsorbate. It is also necessary
that the adsorbate be soluble in the eluting solvent so that
the adsorbed molecules may dissolve rapidly after the solvent
diffuses to the adsorption site.
Methanol is the most commonly used eluting regenerant.

It will usually remove the adsorbate from the resin in a
minimum effluent volume and is relatively inexpensive. Other
low molecular weight alcohols or ketones might also be used
alone or in combination as eluting regenerants.
The elution regeneration of granular carbon has been

examined with very little commercial success. The binding

energies of activated carbons are much higher than those of
adsorbent resins for the same organic molecules (42). While
high temperature thermal methods are necessary to overcome
the high binding energies for activated carbon adsorbents,
the smaller binding energies for resin adsorbents allow the
use of the eluting regeneration method.
Several advantages have been pointed out (64) for the

elution regeneration method for resin adsorbents compared to
the high temperature thermal regeneration required for most
carbon adsorbents. These need to be considered when
choosing an adsorbent system. Since the resin is regenerated
in the same column as the adsorptive process occurred, the
elution method does not require investment in the furnace,
quenching or conveyance equipment. Nor are there adsorbent
losses due to handling, to thermal oxidation or to irreversible
fouling by the formation of inorganic ash as occurs during
thermal regeneration.
Example 2.8
Table 2.13 shows the solubility parameters of several
solvents used in regeneration. Methanol is almost always the
eluting solvent of choice since it requires a minimum of
volume to remove the adsorbate from the adsorbent and is
relatively inexpensive,
Table 2.13: Common Regeneration Solvents for

Synthetic Adsorbents
Solvent Solubility (Ca1°‘s/cm1~5)
2-butanone 9.3
Acetone 9.9
2-propanone 10.0
I-butanol 11.4
1-propanol 11.9
Ethanol 12.7
Methanol 14.5
Water 23.2
Adsorption 131
Wallis and Bolton (65) have shown that it is also

possible to regenerate activated carbon biologically. The
Pseudomonas putida culture will regenerate carbon which has
adsorbed phenol by maintaining the solution concentration of
phenol at a low level and by converting the desorbed phenol
to less toxic substances. The rate of regeneration was on
the order of 25 mg phenol removed per gram of carbon per
hour when the carbon was originally loaded with 40 to 65 mg
phenol per gram carbon. The rate is much slower, about 1
mg phenol per gram carbon per hour, when the original
amount adsorbed is below 40 mg/g.
This approach is limited to those adsorbates which can

be converted to other less adsorbable compounds by a
biological organism. Prolonged exposure or multiple
regeneration reduces the carbon’s regenerated capacity for
adsorption. This probably occurs because of microbial fouling
of the adsorbent’s pores or adsorption of microbially
produced compounds.
2.6 PROCESS CONSIDERATIONS
2.6.1 Design Factors

The properties of both the feedstream and the adsorbent
determine the sizing and overall design of an adsorption
system. The aim is to produce a system in which the flow
during adsorption is under substantially constant temperature
and constant pressure conditions through the adsorbent bed.
The properties of the fluid that need to be known for

design purposes are (66):
(1) Chemical composition (ci, mg/U
including the content of dissolved
inorganic solids. It is important that
the pH also be known because adverse
effects may occur with some adsorbents
at the incorrect pH. The composition
will dictate any pretreatment of the
feedstream which may be necessary.
(2) Fluid density (p, g/cc) at the operating
temperature and pressure.
(3) Fluid viscosity (P, cp) at the operating

temperature and pressure.
The properties of the adsorbent that need to be known

for design are:
(1) The bulk density (pb, g/cc) of the
adsorbent and the void fraction (c) of a
packed bed of the adsorbent particles.
The way in which the adsorbent is
loaded in the bed and its particle size
can affect the void fraction in the bed.
(2) The particle size distribution and the

mean particle size (rP, mm). It is also
necessary to know if the particles
undergo volume changes during the
course of the adsorption-regeneration
cycle.
(3) The pore size distribution, which would
permit the elimination from consideration
those adsorbents which have too much
of their surface area along pores whose
diameter will not admit the solute to be
adsorbed. The pore size distribution is
usually shown as a plot of the increment
of pore volume divided by the increment
of pore radius versus the pore radius.
(4) Hardness, which indicates the attrition
that may occur during the handing of
the adsorbent.
There are a few relationships which must also be

determined before designing the adsorption unit:
(1) The equilibrium adsorption isotherm as a
function of the concentration,
temperature and pH conditions of
interest for the operation.
(2) Regeneration efficiency under the
conditions of the operation. This should
include any loss in adsorption capacity
or losses in adsorbent so that
replacement quantities and frequencies
for the adsorbent are known.
Adsorption 133
(3) The length of the mass transfer zone

should be determined in laboratory
experiments at the flow rates of
interest.
The equilibrium adsorbate concentration, qe, from the

isotherm data and the residual adsorbate concentration, qr,
on the regenerated adsorbent determine the adsorbate
gradient across the mass transfer zone. This gradient is also
the amount of adsorbate which will be loaded per unit weight
of adsorbent each cycle.
With this information, the design of the process consists

of the following steps (67):
(1) With an assumption concerning the cycle
time desired, usually 8-24 hours, the
quantity of solute, m,, in the feedstream
that is to be removed during the
adsorption phase of the operating cycle
is calculated.
(2) The amount of adsorbent material, mp, is
then calculated from the equation:
m,/(cl, - s,> (2.66)

mP =
(3) The amount of adsorbent material, m,,

that cannot be used because of the
length of the mass transfer zone is
calculated from:
m = (“iD2) pb to - Ib 1: ) rb (2.67)
ll
(4) The total amount of adsorbent material,

m,, required for the adsorbent cycle is,
therefore:
m =m+m (2.68)
C P U
(5) This amount is modified each cycle by

some loss in adsorbent capacity and by
attrition or physical loss of adsorbent.
The process requires that this quantity

be added incrementally at the beginning
of each cycle or that the adsorbent
initially in the column be augmented to
compensate for these losses for a given
number of cycles, nc. In the general
case, the column must be designed to
accommodate the following quantity of
adsorbent:
+ mu + nc Am (2.69)
mt = mP
(6) The pressure drop that this quantity of

resin would have, both during the
adsorption phase and the regeneration
phase of the operation, is then
calculated for various bed configurations.
A bed configuration that avoids
excessive pressure drop yet achieves the
desired adsorption is chosen as the
adsorption column.
(7) Auxiliary equipment is sized to meet the
requirements of the adsorption column.
Controls for insuring the constant feed
temperature and pressure, for measuring
the pressure drop across the column and
for monitoring the effluent concentration
and temperature are also included.
2.6.2 Pressure Drop

The flow velocity, the pressure drop across the column
and the bed diameter are interrelated. Setting a value or
limit for one of these parameters places limits on the other
two. Pressure drop limits usually are the basis for fixing
operational flow velocities and selecting the bed diameter.
The pressure drop, AP, through a packed column of

adsorbent is calculated using equations developed for pressure
drop in pipes (66):
APfL = 2 f G2/Dpgp (2.70)
where D, is the mean particle diameter; f is the

dimensionless friction factor; g is the gravitational constant;
Adsorption 135
G is the mass velocity; L is the depth of the adsorbent bed;

and p is the fluid density under operating conditions.
The Reynolds Number, NRe(= D G/P), can be used to

obtain the friction factor. When the Row through the bed is
.
in the lammar regron N, < lo), f = 480/N,. In the
transition flow zone (10 < Nn, ( 200), the friction factor is
given by the value in Figure 2.33. When Nne is greater than
200, flow is in the turbulent zone and the friction factor is
given by:
f = 60.3/(NRe)0.33g (2.71)
This pressure drop only describes the changes across

the bed of adsorbent. Additional considerations are the
amount of freeboard volume (liquid level above the adsorbent
bed) and the pressure drop through the bed supports and
retaining particles. These additional factors can increase the
total pressure drop by an additional 50%. Safety
requirements add another factor of 2.5 to the designed
pressure rating of the column.
100
70
TRANSITION
ZONE
&
50
IL
. 40
E
L
I;f 30
B
E
y.
g 20
10
WODIFIEDN,,
Figure 2.33. Change in friction factor, f, as a function of

the Reynolds number, for use in Equation 2.70 (Reference 66).
2.6.3 Computer Design

A computer program (68) was written for the HP-67 to
calculate the curve
breakthrough for adsorption processes
described by the Hiester and Vermeulen (69). Unfortunately
this program may result in numerical overflow problems. Tan
(70) has written a program for the HP-41 which eliminates
the overflow problems and another program which can be
used for design calculations.
To generate a breakthrough curve, the concentration of

the solute is calculated for a series of throughput values,
while holding the separation factor and the number of
transfer units constant. The program listing is shown in
Figure 2.34 and the user instructions are shown in Table
2.14, along with parameter identification.
That program forms a subroutine of the design program

which also requires the listing shown in Figure 2.35. Three
types of design problems can be solved with this program:
(1) Given the throughput ratio (T),
separation factor (R) and breakthrough
limit (X), the length parameter (N) and
the run time (NT) can be calculated.
(2) Given a run time, separation factor and
breakthrough limit, the length parameter
can be calculated.
(3) Given a length parameter, separation
factor and breakthrough limit, the run
time can be calculated.
The user instructions are shown in Table 2.15. The

parameters are the same as in Table 2.14.
2.7 ADSORPTION APPLICATIONS
2.7.1 Decolorization
The earliest applications of adsorption were the use of
bone char in the decolorization of raw sugar. The cane
sugar industry continues to use activated carbon in its
commercial purification and decolorization processes. Barton
and Knebel (71) have shown that a blended carbon enhances
carbon’s intrinsic properties to provide improved levels of
purification in the processing of sugar cane.
Adsorption 137
01 LdL -nXSm. 71 AYIW 141 m 06 211 sprr 261 m

02 LaL I 72 ‘I - . 212 sm 03
01 . w l ?’ 13 ua. 20 KEY3 213 Xa 02 :: E =
04 AVIW 74 AYIW 1UUOl 214 J9pr 26.4 xen
055IQP 73 Q 12 145 Dn 215 sm 04 15 cm ‘El’
06 sm 11 76 Am 146 a0 .Aw 286sro2
07 * 1 - 1’ 71 m 141 u ::: 1 ii!! I7 2
00 AVIW 140 sm 03 218 K? 02
09m ii iiLo?6’ 219 ClU lJ4- ixa
10 sm 12 mm00 :zY2 tzouo2 29oeo.u.
11 l I l T’ 81 ILZ 11 151 nD OI g2I;6”- 291 1
12 AVIW u8ToOo 151 1 392 sm 06
13 mw a3 sa 13 293 a 09
14 II0 13 84. ::‘z: E&X 2941
13 aa 11 es m 01 2ls 1 295mO5
16 pa 11 Yn-w ::: E OL 296mO6
17 l 87 ml 10 151 m ZJ 297u=I3.
18 no 10 m11002 15a m 05 226Ip.05 296rzOB
19 c? 12 299 n2
2om3 iEf!i :% ifi iIt= iii iJL lJ4= 3m2
21 l I l - 161 a w 231 2
22 rta 11 ii iI* Q) 1u l 232ul ELI5
23 AVIW 93 ml 01 303at/ 05
.
:: A&=12 . Mm.?01 it: _
= = iii ;a lllm.06
95 l 165 *z 07 UIlJL’mr m5sPO5
26 AVIW 166 - 3lx2
27 - I - - ii ,“v 00 161 CfX xYoO8
26 AEU 13 166 #I- 06 236 a 03 CxL:
29 Anor ET9 369 1 XI) RI 06
mrm IQ) s? 01 “,“” 310ml 06
31 LX&J :ixs 311 I
32 Em 02 :zr
103 m. 13
in u
in m 07
ii: Es 312685 2;
33 101 363 Ra# 07
34 m 01 104 ra 10 174 md 03 E E&” ::: :<5;: 364’
gym- 245 mu lw 315cm ‘P.
E in2 01 ::: != 06 246 X<OT EI ;a03
107 Ra. 11 247mO’ur ::: F w _..
la7 -
106m02 K”” . Udun 318sm llI3ctx
179mzOl 249 cm ‘P’ 319 I 389 ST+ 09
:x?;F 390 ml OR
:&tX :: E i?- x&m 391 I
::: s F. 252 Xt2 392 I c-s
43 l 113 a 13 :: zYa 253 2 323- 393 rc=Yl
44 ICL 14 114 - 254 ’ 324SO 05
:: cXu -J2. 255 sr 05 325IlM :z E ii”
:: .
230 ::: z 1% ;o, 2% M 06 326 3% IJI
47 X>Y? 117 1 217 RI 03 327CIA 397 nn
41 XOY 116 . :: 2 ii“ 256 2 326sm 05
49 6(X 119 u mpos 259 sY+ 06 329tU
120 ICL 19 260 9CL OS 330Lu ‘us. 4wRnl
5’ 121 . 261 511 06 4olLaL’EE-
%2 :,x 122 SXO 17 ::: ZY? ::: % O9 402 BCL w
51 SYU 20 123 M ::: FL O6 333al9 403 01
94 ICL 16 124 L9L ‘JFC’ :z ZY-“’ 264 1 t-6 3% ICL 03
II RCL 00 125 CT 06 195 6x0 09 265 Xhn 335ECL04 pT&Plf
16 l 126 M 01 1% 1 I-S 2% CID ‘El’
57 @CL 14 121 X9? 197 xcm 267 1 :: &cl 407 It2
58 - 126 nN 190 CID ‘JI’ 266 ICL 09 338tcL w
19 tix 129 OLS 1% LBL ‘52’ 269 Xt2 339. z iE 03
6oLw 130 ST0 05 2m tCL 05 270 CM 340PI 410 ICL 04
61 1 131 XCL 02 201 X=0? 271 EtX 341 SQRT 411 *
132 X.0? 202 CID ‘Jl’ 272 RCL 06 412 PI
:: t
l 133 SF 08 203 UC 273 * c: ;I 413 *
64 ST0 19 1% RDN 204 nm 274 2 344 - 414 WRT
65 BEEP 135 Fs? OS 205 LIL ‘53‘ 275 ’ %I Fs?02 415 2
66 NIV 136 RR1 276 PI Nb 1 E-3 416 ’
67 SF 12 II7 F-s? 01 :z zik O1 277 SQRT 347 RTN 417 w
68 FIX 138 Cl’Q ‘APJ’ 208 Rnl 210 I 348 LSL ‘PHI’ 418 -
69 ‘Y = - 139 LSL ‘Eu’ 209 LSL ‘AN’ 279 - Y9 FS? 01 419 IM
70 ARCL 19 IL0 CLX 210 KL 01 280 w 350 CIO ‘AM 410 .ENU.
Figure 2.34. HP-41C, Program “FIXBED” calculates

breakthrough curves for fixed bed sorption processes
(Reference 70).
Table 2.14: User Instructions for HP-41C Program “FIXBED”(70)
Step/Procedure Enter/Keystroke Printer Display
1. Allocate memory XEQ ALPHA SIZE ALPHA 025
2. Load program (7 card sides)
3. Set in user mode USER
4. Start GTO ALPHA FIXED ALPHA
5. To calculate .I:
By exact series expansion CF 01
By approximations SF 01
Enter u, v UIV
Calculate .I J J
6. To calculate X,Y:
Start B N=?
Enter N N RIS T=?
Enter T T R/S R=?
Enter R R R/S N, T, R
Calculate A
(Time delay; wait for beep) Y, X
Table 2.15: User Instructions for HP-41C Program “DESIGN”(70)

Step/Procedure Enter/Keystroke Printer/Display
1. Allocate memory XEQ ALPHA SIZE ALPHA 025
2. Load “Fixbed” (7 card sides)
3. Go to end GTO
4. Load “Design” (7 card sides)
5. Set in user mode USER
6. Start GTO ALPHA DESIGN ALPHA
7. Input R and X C R=?

Enter R R R/S X=?
Enter X X R/S R, X
8. To calculate N, given T:
Start F T=?
Enter T T R/S T
(Time delay; a melody is played
after each iteration) N
9. To calculate N, given NT:

Start H NT=?
Enter NT NT R/S NT
(Time delay; a melody is played
after each iteration) N
10. To calculate NT, given N:

Start G N=?
Enter N N R/S N
(Time delay, a melody is played
after each iteration) NT
Adsorption 139
01 WL ‘WSIU- 13 cu 14s sf 12 217 ACCL 11 289 Sa 23

02 LBL c 74 slo 19 1% m 2 218 AVYPILY 290-
03 SF a3 7s IaL ‘M’ 147 ‘I -- 219 aA 291 us
04 ‘I - 7‘ 76 ILZ 11 146 Ama 11 292 SKI ot
or AVIU 77 PST 04 149 NIY K?“”
78 s2vP 15OXU
ii? En 19 RQ ‘II’ EKZY?
iii z ii
08 ‘2 -7’ 8O~va- ::: z ;F- urm z ?Y?

09 rvrer 61 XEJ nor lS3 ml a0 22s 1 a97&02
IO SYW 8zm_rrs- 154 l
II m 20 83 PST 06 iii ZY” ixz
12 1n 84clocm :z _
a I4 WRO-CS~
85 &z 06 ii!;
:: ! 86 aa 07 :EF2 +2 )01aYOo6
61 l
:: E3 2l 86 sr- 11 i% s 18 zzs
17 1 zk:
18 ra 13 ;“I1 Wm.22 zyo7
::: E ii- 307 n. 10
3, 164a 17 Kol
21 m 22 ti it: 165 - 2.97 82010 k go
22m3 166 m 238IzociD=
::zPp 239lJLu- 311 I
.
t: T2 96 Is7 10 :zEs 240 IQ. 11 112 ml
2S &uL*,; 97lxO=m- 241 tad 13 313 U-YY
96 10. 11 :ii L 14 x4 cm ‘cl’
8 *z 99Ba.22 ii: p” 31s UL -a-
26=x-’ ii E 13 316 M
:: iTO 10 tt: ;
i?J iRu20 102IXO-SU- 174 . 246ul.K ::: zP
31 Inl 103 Ldld P 247 s20 10 319 sr 12
32 LBL II :z Eta m l&L ‘Cl’
. .-
33 - WI l 7’ :s ;& 7’ 177 ; UPa. :z ‘?r2
34 AVIly la SYOP ln l 250 no 19
3S SYOP 107 sm 12 I7911211 251 SlO 23 ::: eall”
106 CP 12 2s2 IL1. 11 324 m
ii Ki
.
i” 109 m 3 ;: g OS 2S3 a0 24
.
.- 110 - t - - 2% IaL ‘CI’ Es+z-
ii As 10 183 ml -Is. 25s @CL 10 327 Ip. 13
40 AVYU ::: z&l2 1% u (#
41 RCL 10 :z iv iii :. OS
42 ICL 13 ::: F * :z F l9
43 l 11s la, 12 le.7 x-Y9 :z it? ;? xx
44 sm 11 116 - 160 OPl 260 DO7 332 -
4S RCL 13 117 I 169 1 313 El1
Lb 1 116 sm 11 190 n* 19 Ex 334 w. OS
41 rm 119 KL 12 191 .005 263 XQ ‘RX
48 cm ‘Hl’ 17s ’ 264 CP 01 :E”” -
49 m 10 121 s20 10 ::: zY1*
SORcL22 122 1 x2 :‘d”” _
SI l 123 Ka. 13 :: Lt6 267 l
12 m 11 124 xc-n 1% ml 263 lm7 EziY 412 WOl

53 RCL 22 197 Aks 269 SF 07 :; It1 413 u
14 XOY ::: tit Z’ 190 xc-Y? 270 IBj ‘DWT 414 -
IS I 121 SQEY 1% SP 06 271 Xep -Tt5’ 343 : 41s St2
16 1 12.9 n 13 amum 272 PST 06 x4* 416 EL 01
57 l 129 x12 7.01 I 273 CIO -GE- 34s M 13 417 ra 02
18 I/X 130 ’ 202 sm 07 274 tClo 07 346 -
19 RCL 13 131 LBL I 203 M 10 21s xc07 341 w
60 xc-r? 132 sm II 204 tci. OS 216 Cm 00 348 m OS
61 Cm ‘Hl’ 133 RCL 12 20s I 211 acL 23 349 RN
62 RCL 10 134 ’ 2w w 278 ICL 10 ;; ylrl ‘RP’
63 HCL 13 11s SKJ 10 207 m, Ob 279 X)Y7
64 I 136 LBL ‘11’ 208 RYN 260 sm 23 312 X)Y?
61 ST+ II 137 CF 10 209 LBL c 281 CT0 01 3Sl XOY 425 DO?
66 RCL II 138 CLX 210 - N - 7’ 282 LBL 00 314 ftx 426 Ul
67 2 139 SKI I9 211 AVIW 283 RCL 24 31s RYN 421 SYO OS
68 / 140 cm ‘HI’ 212 !zOP 284 RCL 10 3S6 LBL ‘PSI’ 428 RM
69 LRL 0 I41 LBL ‘110’ 213 SKI II 285 XC-Y? 317 FS? 01 429 .mD.
I42 BEEP 214 FIX 2 286 ST0 24 318 cm ‘MS’
143 BEEP 21s CF I2 287 LBL 01 319 CLX
I44 AD” 2lb - N . - 288 RCL 24 360 SYO 06
Figure 2.35. HP-4 1C “DESIGN” performs three types of

design calculations for fixed bed sorption processes
(Reference 70).
In measuring the performance of adsorbents in sugar

decolorization operations, the transmission through a cell at
470 nm for a given dried solids concentration is compared to
the same dried solids concentration of sucrose. Asai (72) has
shown that contact time is the controlling parameter (Figure
2.36) and that several columns are necessary to obtain the
desired degree of color removal at reasonable flow rates
(Table 2.16).
.05 .07 0,l .2 .3 .4 .5 .7 .9
C/C,
Figure 2.36. The effect of contact time on the residual

colored impurities in sugar solutions (Reference 72).
Adsorption 141
Table 2.16: Effect of Multiple Column Treatment on

Effluent Quality (C/Co)(72)
C/Co of Effluent
Experiment Flow Rate Effluent Effluent Effluent Effluent
Number (cm/min) from Column 1 from Column 2 from Column 3 from Column 4
A-l 1 0.38 0.20 0.14 0.11
A-2 2 0.56 0.33 0.24 0.18
A-3 3 0.60 0.37 0.27 0.19
As an alternative to using charcoal based adsorbents to

remove color, Abrams (73) has shown that phenol-
formaldehyde resins may be used in treating dilute or
concentrated sugar solutions. Duolite S-30 removed 80-100%
of the color from decationized second carbonation juice of
beet sugar. This resin was effective in adsorbing the carmel
and melanoidin colorants when the feed solution has a pH
between 1 and 6. Regeneration of the resin adsorbent was
accomplished by eluting the colorants with 1 to 4% sodium
hydroxide. This was followed by rinsing with dilute acid
before the next adsorption cycle began.
This type of resin has also been used for the removal
of brown color from wine in Japan (74) and for decolorizing
fermentation broths. Adsorbents from phenol-formaldehyde
amine resins are particularly useful in removing flavonoid
pigments which are frequently present in the extracts of
plant materials (75). These resins decolorize these types of
feed streams in the pH 4-7 range.
The use of carbon and other adsorbents in decolorizing

fermentation broths is now so common that it is only
casually mentioned as part of the purification process.
2.7.2 Antibiotics
Macroporous polymeric adsorbents have been used to
recover and purify many antibiotics. The performance of a
polystyrene resin with a surface area of 300 ma/g and an
average pore diameter of 100 A is shown in Table 2.17 (76).
Methanol was used for the elution of the adsorbed antibiotic.
Activated carbon has been shown (77) to remove

Cephalosporin C effectively from its fermentation broth.
Elution was carried out with dilute aqueous solvents. A
Table 2.17: Performance of Amberlite XAD-2 as an

Adsorbent for Antibiotics (76)
Cont. in Saturation Elution
Holecular Feedstream Capacity Peak Elution Required
Antibiotic Wt. (epm) (mg/ml) (ppm) (Bed Volumes)
Tetracycline 444 1000 44 30,000 2
Tetracycline 444 100 1.4 16,000 2
Oxytetracycline 460 100 1.6 17,600 2
Oleandomycin 688 100 40 20,000 2
difficulty with using carbon is its lack of specificity for the

Cephalosporin C.
Voser (78) developed a scheme for purifying

Cephalosporin C using a macroporous polymeric adsorbent.
The concentration of Cephalosporin C in the filtrate from the
fermentation broth was 2.1 g. This filtrate was acidified
with sulfuric or oxalic acid to adjust the pH to 2.3. The
filtrate was fed through a column of a styrene-divinylbenzene
resin (XAD-2) at a rate of one bed volume per hour. The
height of the resin bed was 76 cm. The elution was carried
out with one-half bed volume of deionized water, followed by
two and one-half bed volumes of 10% aqueous isopropanol
solution. Regeneration of the adsorbent was accomplished
with one bed volume of 50% aqueous methanol adjusted to
1 N with NaOH, followed by a water rinse.
This technique removed 65 to 70% of the impurities

during the water wash and the first third of a bed volume of
the isopropanol elution. The rest of the isopropanol elution
contained 95% of the Cephalosporin C and 20 to 25% of the
impurities from the filtrate. The rest of the product and
impurities were removed during the regeneration. When the
flow rate was increased to 2 bed volumes per hour, the
Cephalosporin C yield dropped to 85.2% while the impurities
remaining with the product increased to 30 to 35%.
When carbon (Pittsburgh CAL) was used as the

adsorbent (79), the elution required washing with four times
as much deionized water (two bed volumes) and using almost
twice as much elution solution (4.5 bed volumes of 50%
aqueous acetone). Only the first half bed volume could be
discarded without a significant reduction in product yield.
This means that a much larger volume of eluate had to be
further treated to obtain product of the desired purity when
carbon was the adsorbent.
Adsorption 143
The distribution coefficients for different penicillins and

cephalosporins between hydrophobic adsorbents and aqueous
methanol solutions have been determined as a function of pH
(80). These are shown in Table 2.18 and illustrated in Figure
2.37.
Table 2.18: Distribution Coefficients in 10% Aqueous Methanol for

Penicillins and Cephalosporins on XAD-4 as a Function of pH
Antibiotic -1.3 -2.3 3.7 -5.5 6.0 6.8 7.6
Penicillin C, K+ - 1,150 751 80.2 81.1 76.4 62.3
Penicillin V 2,590 108 106 96.6 -
‘I-amino 12.1 12.8 15.2 - 15.3 16.2

cephalosporin
7-aminodesacetoxi- 11.1 5.1 6.9 6.25 - 5.34 4.42

cephalosporin
sodium cephazoline - 103 98.8 85.5 97.5
2500
glOO0
.
!s
; 500
E
c!j 0
6
; 360
E
p” 120
80
40
XAD-2
XAD-8
PH
Figure 2.37. The effect of pH on the distribution coefficient
for cephalosporin antibiotics on polymer adsorbents
(Reference 80).
Chu and Jabuke (81) have shown the effect of antibiotic

type and concentration on adsorption by polymeric
polystyrene adsorbents (XAD-2). When 20 g of adsorbent was
added to a 100 mL solution of streptomycin (Smg/mL), all of
the antibiotic was adsorbed on the resin, for a loading of 25
mg per gram of resin. When the solution is changed to
lincomycin (1.95 mg/mL), the loading level may be increased
to 31.4 mg per gram of resin but only 80% of the antibiotic
is adsorbed from solution. Even higher loading levels are
possible (57.5 mg per gram of resin) by increasing the
lincomycin concentration to 4 mg/mL; however, the amount
of antibiotic remaining in solution increases to almost 90%.
Neomycin was poorly adsorbed (7.5 mg per gram of resin)
from a solution with a 5 mg/mL antibiotic concentration, so
that only 15% of the antibiotic was removed from solution.
In all of these cases, the elution was carried out by

washing the loaded resin with hot water followed by three
bed volumes of 97% aqueous methyl ethyl ketone to recover
99% of the adsorbed antibiotic.
The antibiotic A-4696 has been recovered and purified

using two different grades of activated carbon (82).
Pittsburgh (12 x 40) carbon was used for the initial recovery,
while Darco G-60 was used for the further purification. The
elution used an acidified acetone-water (1:l) solution as the
eluant for each step. When sulfuric acid was used as the
acidifying agent, the sulfate ions had to be removed from the
eluate by precipitation as barium sulfate. The yield was
reduced by 25% during the purification process.
2.7.3 Enzymes
An alkaline protease (Mol. Wt. 28,000) was recovered
from a fermentation broth using a macroporous acrylate non-
functionalized adsorbent (83). The effluent concentration of
the enzyme is shown as a function of column throughput in
Figure 2.38. The resin adsorbed 73% of the enzyme, with a
loading of 776,200 AZDU units per gram of resin. After a
rinsing with water to remove non-adsorbed enzymes and
fermentation impurities, an ethanol-water (1:l) solution with
5.7% NaCl was used to eluate the protease. As the figure
shows, good concentration of the enzyme was obtained.
Adsorption 145
0 -
_Yy--Yk (
// - 200 400 600
LOAD ELUTE
COLUMN THROUGHPUT (ML)
Figure 2.38. Recovery of alkaline protease from fermentation

broth with Amberlite XAD-‘7 (Reference 83).
Volclay KWK, a western bentonite clay adsorbent, has

been used to recover lipolytic enzymes from the fermented
growth product of Mucor microorganisms (84). The
adsorption step was carried out by adjusting the growth
product to a pH of about 10.0 with 2N NaOH and thoroughly
admixing with the adsorbent at ambient temperature for
about one hour. The adsorbent material was then separated
from the non-adsorbed rennet enzyme material by filtration
using 2.0% filteraid. The desired lipolytic enzyme was then
eluted from the adsorbent by adjusting the pH to 4.0 with 2N
HCl and the solution was evaporated to one sixth the original
fluid volume. The recovery was about 100%. The sensitivity
of the recovery to the pH adjustment was demonstrated by

the yield dropping to only 22.5% when the pH was adjusted
to 11.0 rather than 10.0 during the adsorption process.
Likewise, when the first pH adjustment was to pH 5.0 instead
of 10.0, there was 0% lipolytic enzyme recovered but rather
72.7% of the rennet was recovered.
2.7.4 Microorganism Adsorption

Adsorbents may also be used for the adsorption of
microorganisms. While ion exchange resins have been used
for much of the reported work, comprehensively reviewed by
Daniels (85), charcoal, silica gel and alumino silicates have
been used for bacteria, fungi and virus adsorption. Specific
examples are listed in Table 2.19.
Table 2.19: Microorganism Adsorption on Charcoal and Sihcates(g6)

Microorganism Adsorbent Capacity (cells/g)
Bacillus subtilis Charcoal 50 x 10’0
Escherichia coli Charcoal 50 x 10’0
Escherichia coli Activated carbon 1.6 x 10”

bacteriophage T,
Escherichia coli Activated carbon 3 x 10s

bacteriophage T,
Escherichia coli Bituminous coal 4 x 10s

bacteriophage MS,
Escherichia coli Activated carbon 2-46 x 10’

bacteriophage fs
Poliovirus type 1 Activated carbon 6.5 x 10’
Poliovirus type 1 Silicates 7-9 10s
This type of adsorption of suspended cells on solid

surfaces is an additional method of control in isolating
components and purifying mixtures. Adsorbents can change
the rate of reproduction, morphology of cells, duration of
growth cycles, oxygen consumption, carbon dioxide production
and the character and amount of metabolites (86). The
specific applications of microbial adsorption are listed in
Table 2.20 from the review by Daniels (87).
Adsorption 147
Table 2.20: Applications of Microbial Adsorption(87)

Concentration of Cells
Stimulation or suppression of growth
Morphology
Growth cycles
Oxygen consumption
Carbon dioxide production
Metabolic studies of adsorbed vs unadsorbed cells
Reactants and/or products
Control of pH
2. Separation of Cells from Suspension
Nonselective removal
Selective removal of pathogens
3. Resolution of Cellular Mixtures
Grossly different microorganisms
Similar microorganisms
Distinct species or cell types
2.7.5 Organic Acids and Amines

Charcoal adsorbents have been used to remove organic
acids and amines from aqueous solutions. The bed volumes of
the feedstream are given in Table 2.21 (88). The amounts of
regenerant, l.ON NaOH, required are also listed. Since
charcoal does not have the fixed ion exchange groups which
ion exchange resins have, it does not exhibit the ion
exclusion (Donnan) effect, causing slower regeneration,
especially for the more hydrophobic solutes.
From Table 2.21 it can also be seen that within a given

homologous series, a point is reached where the higher
members are not easily removed from the charcoal. Also
within a series, the extent of adsorption is inversely
Table 2.21: Sorption of Organic Acids and Amines on Charcoal(88)

Feed Breakthrough Regeneration
Solute Concentration (%) Point (BV) Requirement (BV)
Formic acid 0.1 4.0 <2
Acetic acid 1.0 6.3 c2
Acetic acid 0.1 21 (2
Oxalic acid 0.1 22 (2
Malonic acid 0.1 36 <2
Propionic acid 0.1 42 2.5
Valerie Acid 0.1 116 13
Caprylic acid 0.1 295 >lOO
Ethylenediamine 0.1 9.3 <2
Ethylamine 0.1 11 <2
Piperazine 0.1 23 <2
Lliethylenetriamine 0.1 24 (2
n-Propylamine 0.1 34 5
n-Amylamine 0.1 95 22
Benzylamine 0.1 123 74
proportional to the water solubility of the compound. The

adsorption of monofunctional organic acids and amines is
always greater than for polyfunctional molecules of the same
molecular weight. Substitution of hydrophilic groups (-OH)
lowers the quantity of solute adsorbed and increases the
speed of regeneration. Conversely, substitution of
hydrophobic groups (-Cl) increases the quantity of solute
adsorbed and decreases the efficiency of regeneration.
Paleos (89) studies the adsorption of organic acids and

phenols on hydrophobic (styrene-based) and hydrophilic
(methacrylate-based) adsorbents. As Figure 2.39 shows, XAD-
6 with its higher surface polarity adsorbs more acetic acid
but less butyric acid than XAD-7. For both resins, as the
chain length of the acid is increased, the amount of
adsorption increases, consistent with Traube’s Rule.
The adsorption of substituted phenols increases for the

hydrophobic adsorbent XAD-4 as the substitution causes the
phenol to become more hydrophobic. This is shown in Figure
2.40.
Adsorption 149
B”T”R1C ACIDhAD-
0 0.1 0.2 0.3 0.4 0.5

SOLUTE CONCENTRATION ~OL/LITER)
Figure 2.39. Adsorption of acetic acid and butyric acid from

aqueous solution onto hydrophobic (XAD-6) and hydrophilic
(XAD-7) adsorbents (Reference 89).
2.8
2,4 -
2,4,6-TRICHLORO-
P-NITRO-
I
0 0.2 Oa4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
SOLUTE CONCENTRATION(MOL/LITER)
Figure 2.40. Adsorption of substituted phenols from aqueous

solutions on Amberlite XAD-4 at 25°C (Reference 89).
2.7.6 Vitamin Adsorption

Porous crosslinked styrene adsorbents were used to
examine the effect of porosity, crosslinking, surface area,
vitamin concentration and flow rates on the adsorption of
Vitamin B-12 (90). The degree of crosslinking, porosity,
surface area and equilibrium adsorption are shown in Table
2.22. The porosity and the surface area can be seen to be
controlling the amount of vitamin adsorption. This is
emphasized by the lack of adsorption seen with the non-
porous adsorbent (No. 7).
Table 2.22: Adsorption of Vitamin B-12 on Styrene Adsorbent

Vitamin B-12
Porosity Surface Area Adsorbed
Adsorbent Crosslinking (cm2/g) (m2/g) (mg/g)
1 15 1.89 130 3.10
2 20 1.80 174 3.25
3 25 1.77 210 3.00
4 15 0.99 65 2.45
5 20 0.89 232 3.50
6 25 1.10 342 3.85
7 8 0.05 nil 0.00
The decrease in adsorption for adsorbent No. 3, even

though it has a higher surface area than No. 2, is most likely
due to an unfavorable distribution of the pore sizes which do
not admit the vitamin molecule.
The effect of vitamin concentration is shown in Figure

2.41 and the effect of flow rate is shown in Figure 2.42.
After a certain vitamin concentration is reached (about 75
mg/dms), there is no additional increase in adsorption with
further concentration increases. As expected, increasing the
flow rate in column operations is seen to decrease the
amount of feedstream treated before breakthrough. At a
flow rate of 4.5 bed volumes per hour, 71.25 bed volumes of
vitamin B-12 solution is treated before breakthrough. This is
eluted from the adsorbent with 2 bed volumes of methanol,
for a concentration increase of 35.62 times.
Adsorption 151
32
28
0
0 50 100 150
EMJILIBRIUM CONCENTRATION hc/m3)
Figure 2.41. The effect of the concentration of the

equilibrated solution of vitamin B- 12 on its uptake by a
porous polymeric adsorbent at 30°C (Reference 90).
0 4 8 12 16 20 24
FLOW RATE (BED VOLUMES/ HR)
Figure 2.42. The effect of flow rate on the adsorption of

vitamin B- 12 by the adsorbent used in Figure 2.41 (Reference
90).
Kennedy (84) has examined the adsorption of vitamin B-

12 from a fermentation broth using commercial crosslinked
styrene adsorbents and weak acid ion exchange resins. Table
2.23 compares the adsorption and recovery with these two
types of materials. The styrene adsorbent has a much higher
capacity for vitamin B-12 and can be eluted more efficiently
than the acrylic ion exchange resin. The concentration of
vitamin B- 12 in the feedstream was 15 ppm and the flow rate
was 40 L/min-m3. Methanol was again used as the eluting
fluid.
Table 2.23: Adsorption and Elution of Vitamin B-12

Adsorption Elution
Capacity at Capacity at Elution Methanol
Breakthrough Saturation Concentration Required
(mg/ml) (mg/ml) (flp”) (Bed Volumes)
Amberlite
IRC-50 0.03 0.14 150 5
Amberlite
XAD-2 3.5 5.2 7200 2
2.7.7 Miscellaneous Product Purifications
Glucuronides and sulfates have been recovered from

aqueous solution (91) using hydrophobic polymeric adsorbents.
The loadings were typically 10 pg per gram of resin at flow
rates up to 0.4 bed volumes per minute. An important aspect
of this work was the discovery of a suitable solvent,
acetone-water, for the elution of these compounds (Table
2.24), with nearly quantitative recovery. The smaller pore
size and larger surface area of XAD-4 allowed more complete
rinsing of inorganic salt contaminants without loss of the
adsorbed compounds than did XAD-2 with its larger pores and
smaller surface area.
Table 2.24: Elution of Glucuronides and Sulfates from XAD-4

with 200 ml of Eluant (91)
Compound
_ Percent Recovered with
Methanol Acetone - Water (1: 1)
p-Nitrophenyl sulfate 2.2 + 2.6 86.8 + 1.3
p-Nitrocatechol sulfate 0 0
p-Nitrophenyl glucuronide 29.2 + 6.8 91.0 + 3.7
Androsterone glucuronide 9.5 + 3.5 78.7 + 12.9

Adsorption 153
Carcinogenic polycyclic aromatic compounds, such as

benz(a)pyrene, in concentrations as low as 15 ppb were
removed from crude normal paraffins (10 to 25 carbon atoms)
using an upflow adsorption operation through alumina/silica
adsorbents (92). The impurity concentration was reduced to
less than 1 ppb using a flow rate of 5.2 liters/hour at 50°C.
Table 2.25 shows the impurity levels obtained for various
amounts of treated fluid per unit of adsorbent. These
treated normal paraffins are acceptable for use as starting
materials in food petroleum fermentation products.
Table 2.25: Residual Benz(a)Pyrene in Normal Paraffins After

Alumina and Alumina/Silica Treatment (92)
Average Surface Particle Amount of paraffins Residual
Adsorbent Area (m2/g) size(mesh) treated/unit adsorbent benzlalpyrene(ppb)
‘AllUUlU 200 50 20-30 < 1
‘Alumina conto,n,ng
10x s111ca 400 50 14-32 < 1
SlllG3 730 10 20-30 5
AlUliln~ 200 100 20-30 7
‘Cwnples of preferred treatment
2.7.8 Ethanol Adsorption

Walsh and coworkers (93) developed a process for
separating ethanol from dilute aqueous fermentation media.
In this process, a portion of the fermentation media is passed
over activated carbon which adsorbs the ethanol and a small
portion of the water. When the adsorbent has reached its
adsorption limit, the influx of fermentation medium is halted
and a carrier gas, such as CO, from the same fermentation
process, is used to desorb the adsorbed ethanol and water.
The desorbed vapors are passed through a cellulose adsorbent
to remove the water, yielding a pure ethanol stream.
This process allows the concentration of ethanol in the

fermentor to remain at 6% for minimum inhibition of the
fermentation process. The adsorbed phase on the activated
carbon concentrates this to approximately 50% by weight
ethanol. After this stream emerges from the cellulosic
adsorbent, the ethanol content is 95%.
While this technique appears technically feasible with

the type of process shown in Figure 2.43, it has not yet been
shown to be economically feasible.
43
I
S
H
F FERMENTORS H CELLULOSE ADSORPTION BED P PUMP

G CARBON ADSORPTION BED S STEAM LINES
Figure 2.43. Proposed Fermentation-Stripping-two stage

adsorption process from Reference 93.
The zeolite adsorbent, ZSM-5, has also been reported to

allow the adsorption ethanol from a fermenting mixture SO
that the ethanol concentration remains below the limit where
it becomes toxic to the fermenting microorganism (94). This
material has an adsorption capacity of 65 to 85 mg of
ethanol per gram of zeolite. The ethanol concentration of an
aqueous ethanol solution (5 to 10% ethanol) is reduced by 57
to 86% when the solution is treated with ZSM-5.
These authors suggested the use of butane in a

“pressure swing” regeneration. That is, after the elution of
ethanol with butane, the pressure is reduced to remove the
butane from the ethanol and the butane is recovered for the
next elution of ethanol. When the zeolite loses a significant
portion of its adsorptive efficiency after several cycles of
use due to the accumulation of organic matter in its pores,
the efficiency can be recovered by burning the organic
matter away at temperatures of 450-650°C.
Feldman (95) uses crosslinked poly(4-vinylpyridine) to

remove low concentrations of ethanol from aqueous streams.
A maximum adsorption of 0.33 g ethanol per gram of resin
was obtained with solutions containing 10.1% ethanol.
However, when the concentration of ethanol is above 5%, the
time for equilibrium adsorption becomes quite long. Table
2.26 compares the percent adsorption of ethanol and other
potential fermentation products on this polymer with that on
two different carbon adsorbents. The usefulness proposed
was in keeping the ethanol level below the limit t:iat reduces
the productivity of the microorganisms by continually adding
and removing this type of adsorbent and using the CO,
Adsorption 155
generated by the fermentation reaction to strip the adsorbed

ethanol from the adsorbent.
Table 2.26: Adsorption of Fermentation Products on Poly(4-Vinyl-

Pyridine) Compared with Granular and Powdered Charcoal
Percent Adsorption at 25’C

On PVP On PCB Granular Charcoal On PCL Powder Charcoal
water 33 45 51
Methanol 92 38 48
Ethanol 80 35 49
Isopropanol 40 34 45
Tert-Rutanol 19 35 49
Acetic Acid 295 81 90
2.7.9 Adsorption as a Fermentation Aid

The removal of volonomycin from a fermentation broth
of S. paulus, UC 5142 with a polymeric adsorbent has been
shown to protect the S. paulus from its antibacterial effect
and allows its synthesis to occur at higher levels (96). This
is shown in Figure 2.44.
KU-
w-
CONCENTRATION OF ADSORBENT (G/L)
Figure 2.44. Effect of removal of volonomycin from XAD-2

by acetone treatment on the fermentation production of anti-
M. I uteus (Reference 96).
Adsorbents have also been added to fermentation broths

to enhance the microbial production on monoterpenes such as
geraniol, citronellol, nerol and linalool by adsorbing excreted
metabolites that would otherwise inhibit product
concentration (97). These effects are shown in Table 2.27.
Table 2.27: Effect of Presence of Amberlite XAD-2 on

Terpenoid Product Concentration (97)
Product Yields (mg/L) at 162 Hour Point

Terpenoid Normal Fermentation With XAD-2
Geraniol 334 1227
Citronellol 60 198
Linalool 1 46
Nero1 1 28
rota1 395 1499
2.8 REFERENCES
1. Freundlich, H., Colloid and Caoillarv Chemistry,

Methuen, London (1926)
Boedecker, C., J Landwirtschaft, 7:48 (1859)
Kuester, F.W., Annalen, 283:360 (1894)
Langmuir, J., J Am Chem Sot, 40:1361 (1981)
Jossens, L., Prausnitz, J.M., Fritz, W., et al., Chem Enq

&i, 33:1097 (1978)
6. Henry, D.C., Phil Mag, [6]44:689 (1922)
7. Beverloo, W.A., Pierik, G.M., Luyben, K., In:

Fundamentals of Adsorption (Myers, A.L., Belfort, G.,
eds.) Engineering Foundation, New York, p 95 (1984)
8. Hasanain, M.A., Hines, A.L., Ind Ena Chem Processes

Des Dev, 20:621 (1981)
9. Giles, C.H., MacEwan, T.H., Nakhwa, S.N., et al., J Chem

%, 3973 (1960)
Adsorption 157
10. Gustafson, R.L., Albright, R.L., Heisler, J., et al., Ind

Enn Chem Prod Res Develop, 7:107 (1968)
11. Young, G.J., Chessick, J.J., Healey, F.H., J Phvs Chew

60:394 (1956)
12. Greenhill, E.B., Trans Fraadav Sot, 45:625 (1949)
13. Swearingen, L.E., Dickenson, B.N., J Phvs Chem, 36:534

(1932)
14. Rothman, S., Mandel, J., McCann, F.R., Weissberg, S.G.,

J Colloid Sci, lo:338 (1955)
15. Crittenden, J.C., Hand, D.W., In: Fundamentals of

Adsorption (Myers, A.L., Belfort, G., eds.) Engineering
Foundation, New York, p 185 (1984)
16. Spahn, H., Schleunder, E.U., Chem Enn Sci, 30:529

(1975)
17. Neretrieks, I., Chem Ena Sci, 31:465 (1976)
18. McKay, G., Chem Enn Res Des, 62(7):235 (1984)
19. Leyva-Ramos, R., Geankoplis, C.J., Chem Eng Sci,

40(5):799 (1985)
20. Rasmuson, A., In: Fundamentals of Adsorption (Myers,

A.L., Belfort, G., eds.) Engineering Foundation, New
York, p 461 (1984)
21. Kataoka, T., Yoshida, H., Ueyama, K., J Chem Enq

(Japan), 5(2): 132 (1972)
22. Gorg, D.R., Rutliven, D.M., A I Ch E J, 19:852 (1973)
23. Brauch, V., Schleunder, E.U., Chem Eng Sci, 30:539

(1975)
24. Lukchis, G.M., Chem Enq, 111 (June 11, 1973)
25. Rasmuson, A., Chem Ene, Sci, 40(4):621 (1985)
26. Rasmuson, A., Chem Enn Sci, 40(7):1115 (1985)

27. Brunauer, S., Emmett, P.H., Teller, E., J Am Chem SOC,

60:309 (1938)
28. Abrams, I.M., Ind Enp Chem Prod Res Dev, 14(2):108
(1975)
29. Smith, R.N., Ouart Rev Chem Sot, 13:287 (1959)
30. Guisti, D.M., Conway, R.A., Lawson, C.T., J Water Poll

Control Fed, 46(5):947 (1974)
31. Calgon Test Method TM-4
32. Calgon Test Method TM-3
33. ASTM D-3467-76
34. Carrubba, R.V., Urbanic, J.E., Wagner, N.J., et al., A

Ch E Svmp Ser, 80(233):76 (1984)
35. Snyder, L.R., J Chromatoq, 23:388 (1966)
36. Anderson, R.A., Sherman, J.D., A I Ch E Svmp Ser,

80(233): 118 (1984)
37. Strandberg, G.W., Smiley, K.L., Biotechnol Bioenq, 14509

(1972)
38. Messing, R.A., U.S. Patent No. 3,868,304 (1975)
39. Lee, Y.Y., Fratzke, A.R., Wun, K., et al., Biotechnol

Bioeng, 18:389 (1976)
40. Abrams, I.M., Chem Eng Progr, 65(97):106 (1969)
41. Gustafson, R.L., U.S. Patent No. 3,531,463 (Sept 1979)
42. Ash, S.G., Brown, R., Everett, D.H., J Chem

Thermodynamics, 5:239 (1973)
43. Wang, H., Duda, J.L., Radke, C.J., J Colloid Interface

&i, 66:153 (1978)
44. Giles, C.H., Mehta, H.V., Rahman, S.M.K., et al., J Apnl

Chem, 9:457 (1959)
Adsorption 159
45. Rosene, M.R., Lutchko, J.R., Deithom, R.T., et al.,

Presentation at ACS National Meeting, Miami, FL,
September 198 1
46. Rosene, M.R., ACS Presentation, March 1981
47. Berthier, P., Kerlan, L., Courty, C., Compt Rend,

246:1851 (1958)
48. Blackburn, A., Kipling, J.J., J Chem Sot, 4103 (1955)
49. Ores, B., Rauber, C., U.S. Patent No. 3,741,949 (June 26,
1973)
50. Mathews, A.P., A I Ch E Svmp Set-, 79(230):18 (1983)
51. Patil, V.K., Joshi, J.B., Sharma, M.M., Chem Enn Res
&, 62:247 (1984)
52. Jordon, D., “Carbon Decolorization”, In: Fermentation

and Biochemical Ennineerinq course, The Center for
Professional Advancement, New Brunswick, NJ (1983)
53. Fornwalt, H.J., Hutchins, R.A., Chem Enq, 179 (April 11,
1966)
54. Hager, D.G., Industrial Water Engineering, 11(1):14

(1974)
55. McKay, G., Chem Enp Res Des, 61(1):29 (1983)
56. HSU, E.H., Fan, L.T., “A Novel Porous Medium Filter”,

In: Proceedings of World Filtration Congress III,
Philadelphia, PA, p 420 (September 13-17, 1982)
57. Mathews, A.P., Fan, L.T., A I Ch E Svmp Ser, 79(23):79

(1983)
58. Tavares, A., Kunin, R., Pub1 Tech Pap Proc Annu Meet
Sugar Ind Technol, 40:242 (1981)
59. Hutchins, R.A., Chem Ena Prog, 71(5):80 (1975)
60. DeJohn, P.B., Chem Enq, 113 (April 28, 1975)

160 Separation and Purification-Techniques in Biotechnology
61. LeVan, L.D., Friday, D.K., In: Fundamentals of

Adsorption (Myers, A.L., Belfort, G., eds.) Engineering
62. Dessau, R.M., Haag, W.O., U.S. Patent No. 4,442,210

(April 10, 1984)
63. McCaffrey, D.J.A., Rogers, P.E., Great Britain Patent No.

21515OlA (July 24, 1985)
64. Stevens, B.W., Kerner, J.W., Chem Eng, 84 (Feb 3, 1975)
65. Wallis, P.A., Bolton, E.E., Presented at Oak Ridge

National Lab Meeting (August 1984)
66. Johnston, W.A., Chem Enq, 87 (Nov 27, 1972)
67. Lukchis, G.M., Chem Enq, 83 (Aug 6, 1973)
68. Tan, H.K.S., Chem Enq, 117 (March 24, 1980)
69. Hiester, N.K., Vermeulen, T., In: Perrv’s Chemical

Engineers Handbook, Sec. 16, 5th Ed, McGraw-Hill, NY
(1973)
70. Tan, H.K.S., Chem Eng, 57 (Dee 24, 1984)
71. Barton, W.F., Knebel, W.J., 41st Annual Meeting, Sugar

Industry Technologists, Inc.
72. Asai, S., Proc Res Sot Japan Sugar Refin Technol, 16:l
(1965)
73. Abrams, I.M., Sugar Y Azucar, 31 (May 1971)
74. ( ), J Ferment Technol Japan, 46(3):153 (1968)
75. Abrams, I.M., Ind Eng Chem Prod Res Devel, 14(2):108
(1975)
76. Kunin, R., Amber-hi-lites Porous Polvmers as Adsorbents

- A Review of Current Practice, Rohm and Haas Co.
Brochure
77. Abraham, E.P., Newton, G.G.F., U.S. Patent No. 3,093,638

(1963)
Adsorption 161
78. Voser, W., U.S. Patent No. 3,725,400 (April 3, 1973)
79. Pines, S., US. Patent No. 3,983,108 (September 28, 1976)
80. Salto, F., Diez, M.T., Prieto, J.G., An Frac Vet Leon
Univ Oviedo, 25( 1):31 (1979)
81. Chu, D.Y., Jabuke, H.K., U.S. Patent No. 3,515,717 (June
2, 1970)
82. Hamill, R.L., Stark, W.S., DeLong, D.C., U.S. Patent No.
3,952,095 (April 20, 1976)
83. Kennedy, D.C., Ind Ena Chem Prod Res Develou, 12( 1):56
(1973)
84. Moskowitz, G.J., Como, J.J., Feldman, L.I., U.S. Patent

No. 3,899,395 (Aug 12, 1975)
85. Daniels, S.L., In: Adsorption of Microorganisms to

Surfaces (Bitton, G., Marshall, K.C., eds.) J. Wiley &
Sons, NY, p 7 (1980)
86. Zuyagintsev, D.G., Nauch Dokl Vvssh Shk. Biol Nauki,

3:97 (1967)
87. Daniels, S.L., Dev Ind Microbial, 13:211 (1972)
88. Sargent, R.N., Graham, D.L., Ind Enn Chem Proc Des
m, l(l):56 (1962)
89. Paleos, J., J Co11 Interf Sci, 31(1):7 (1969)
90. Anand, P.S., Dasare, B.D., J Chem Tech Biotechnol,

31:213 (1981)
91. White, J.D., Schwartz, D.P., J Chromatoq, 196:303 (1980)
92. Funakubo, E., Matsuo, T., Taira, T., et al., U.S. Patent
No. 3,849,298 (Nov 19, 1974)
93. Walsh, P.K., Liu, C.P., Findley, M.W., et al., In:

Fundamentals of Adsorption (Myers, A.L., Belfort, G.,
eds.) Engineering Foundation, NY, p 667 (1984)
94. Dessau, R.M., Haag, W.O., U.S. Patent No. 4,442,210

(April 10, 1984)
95. Feldman, J., U.S. Patent No. 4,450,294 (May 22, 1984)
96. Marshall, V.P., Little, M.S., Johnson, L.E., J Antibiot,

34(7):902 (1981)
97. Schindler, J., Ind Eng Chem Prod Des Dev, 21537 (1982)
98. Everett, D.H., In: Fundamentals of Adsorption (Myers,

A.L., Belfort, G., eds.) Engineering Foundation, NY, p 1
(1984)
3
Ion Exchange
3.1 INTRODUCTION
Separation and purification operations with ion exchange

resins involve the reversible interchange of ions between a
functionalized insoluble resin (the ion exchange material) and
an ionizable substance in solution. While Thompson (1)
reported in 1850 the first ion exchange applications which
used naturally occurring clays, ion exchange resins have only
been used in biochemical and fermentation product recovery
for the last few decades (2,3). In these early studies,
biochemicals such as adenosine triphosphate (4), alcohols (5),
alkaloids (6), amino acids (7), growth regulators (8), hormones
(9), nicotine (lo), penicillin (1 l), and vitamin B-12 (12) were
purified using ion exchange resins.
Ion exchange applications intensified following the work

of Moore and Stein (13), which showed how very complex
mixtures of biochemicals, in this case amino acids and amino
acid residues, could be isolated from each other using the ion
exchange resin as a column chromatographic separator.
Specific chromatographic applications of ion exchange resins
will be covered in Chapter 4. This chapter will concentrate
on the theory, characterization and process applications of
ion exchange resins.
In biotechnology applications, ion exchangers are

important in preparing water of the necessary quality to
enhance the desired microorganism activity during
163
fermentation. Downstream of the fermentation, ion exchange

resins may be used to convert, isolate, purify or concentrate
the desired product or by-products.
3.2 THEORY
The important features of ion exchange reactions are

that they are stoichiometric, reversible and possible with any
ionizable compound. The reaction that occurs in a specific
length of time depends on the selectivity of the resin for the
ions or molecules involved and the kinetics of that reaction.
The stoichiometric nature of the reaction allows resin

requirements to be predicted and equipment to be sized. The
reversible nature of the reaction, illustrated as follows:
R-H+ + Na+Cl- r-” R-Na+ + H+Cl- (3.1)

allows for the repeated reuse of the resin since there is no
substantial change in its structure.
The equilibrium constant, K, for equation 3.1 is defined

for such monovalent exchange by the equation:
(3.2)
In general, if K is a larger number,’ the reverse reaction is

much less efficient than the forward reaction and requires a
large excess of regenerant chemical, HCl in this instance, for
moderate regeneration levels.
The equilibrium constant defined in Equation 3.2

compares to Ku, the equilibrium constant defined in Chapter
2 for adsorptive processes:
s = Ci/CO (3.3)
where K, is the ratio of the solute concentration within (Ci)

and surrounding (c,) the adsorbent material.
Starobinietz and Gleim (14) distinguished between the

amount of molecular adsorption and ion exchange that
occurred with a strong base anion resin and a series of fatty
acids. The overall adsorption of carboxylic acids on the
Ion Exchange 165
resin was determined from the change in its solution phase

concentration and the amount of ion exchange from the
amount of counterions displaced from the resin into the
solution. As Figure 3.1 shows, there is a steady increase in
molecular adsorption with hydrocarbon increase in chain
length. However, the amount of ion exchange decreases from
formic through acetic to propionic acid and increases again
for the higher acids. This type of distinction is not often
measured, particularly when biomolecules with multiple modes
of interaction are under investigation.
1 .fj-
1.2
20 40 60 60 100/.---
I I I I I 1 I I I I I I I
0 20 40 60 80 100 120
Figure 3.1. Adsorption isotherms of fatty acids on Dowex

1X2 in chloride form: (a) molecular sorption; (b) ion
exchange. 1. formic; 2. acetic; 3. propionic; 4. butyric; 5.
valeric; 6. caproic; 7. heptanoic; 8. caprylic; 9. pelargonic
acids (Reference 14).
With proper processing and regenerants, the ion

exchange resins may be selectively and repeatedly converted
from one ionic form to another. The definition of the
proper processing requirements is based upon the selectivity
and kinetic theory of ion exchange reactions.
3.2.1 Selectivity
For ion exchange resins, the selectivity is developed in
terms of the Donnan potential, En. The Donnan potential is
the difference in the electric potential between the ion
exchange resin, I$~, and the solution, @,. This may be related
to the chemical potential or activity by the equation:
1
ED = Qr - +s = - ai,s - llv (3.4)
i
a.
ziF =,r
where Zi is the charge on the ion, F is the Faraday constant,

ai * and air are the activities of ion “i” in solution and the
resin, respectively, and Vi is the partial molar volume of ion
“i”. The partial molar volume is assumed to be the same in
the liquid and resin phase.
When ion “B”, which is initially in the resin, is

exchanged for ion “A”, the selectivity is represented by:
+*PB - WA) (3.5)
The selectivity which a resin has for various ions is

affected by many factors. These factors include the valence
and size of the exchange ion, the ionic form of the resin,
the total ionic strength of the solution, crosslinkage of the
resin, the type of functional group and the nature of the
non-exchanging ions. The ionic hydration theory has been
used to explain the effect of some of these factors on
selectivity (15).
According to this theory, the ions in aqueous solution

are hydrated and the degree of hydration for cations
increases with increasing charge and decreasing
crystallographic radius, as shown in Table 3.1 (16). It is the
high dielectric constant of water molecules that is
responsible for the hydration of ions in aqueous solutions.
The hydration potential of an ion depends on the intensity of
Ion Exchange 167
the charge on its surface. The degree of hydration of an ion

increases as its valence increases and decreases as its
hydrated radius increases. Therefore, it is expected that the
selectivity of a resin for an ion is inversely proportional to
the ratio of the valence/ionic radius for ions of a given
radius. In dilute solution the following selectivity series are
followed:
Li < Na < K < Rb < Cs
Mg < Ca < Sr < Ba
Al < SC < Y < Eu < Sm < Nd < Pr < Ce < La
F<Cl<Br<I
Table 3.1: Ionic Size of Cations (16)

Crystallographic Hydrated Ionization
-Ion Radius (#) Radius (I() Potential
Li 0.68 10.0 1.3
Na 0.98 7.9 1.0
K 1.33 5.3 0.75
1.43 5.31 __
NH4
Rb 1.49 5.09 0.67
CS 1.65 5.05 0.61
Mg 0.89 10.8 2.6
Ca 1.17 9.6 1.9
Sr 1.34 9.6 1.6
Ba 1.49 8.8 1.4
As Figure 3.2 (17) shows, as the valence of the

exchange ion increases and as the hydrated radius of the
exchange ion decreases, the selectivity or affinity of the ion
exchange resin for that ion increases. As should be
apparent, if the resin is in an ionic form “A” which has a
lower replacing power than an ion “B” in solution, the resin
will be converted at equilibrium to the ionic form “B”. The
selectivity of resins in the hydrogen ion or the hydroxide ion
form, however, depends on the strength of the acid or base
formed between the functional group and the hydrogen or
hydroxide ion. The stronger the acid or base formed, the
lower is the selectivity coefficient. It should be noted that
these series are not followed in non-aqueous solutions, at

high solute concentrations or at high temperatures.
8-
I,
0.6 0,7 008 0.9 1.0 1.1
EQUILIBRIUM EXCHANGE (MILLIEQUIVALENTS/GRAM)
Figure 3.2. Effect of ionic radius on ion exchange in a

carbonaceous zeolite (Reference 17).
The dependence of selectivity on the ionic strength of

the solution was related through the mean activity coefficient
to be inversely proportional to the Debye-Huckel parameter
a0 (18):
-A &-
log Y* = (3.6)
1 + Ba'fi
where r+ is the mean activity coefficient, A and B are

constants, and p is the ionic strength of the solution. The
mean activity coefficient in this instance represents the
standard free energy of formation (-AF”) for the salt formed
Ion Exchange 169
by the ion exchange resin and the exchanged ion. The

Debye-Huckel parameter is a measure of the distance of
closest approach which is also related to the ionic hydration
size. Figure 3.3 (19) shows this dependence as the ionic
concentration of the solution is changed. As the ionic
concentration of the solution increases, the differences in the
selectivity of the resin for ions of different valence
decreases, and in some cases, the affinity may be greater for
the lower valence ion.
0 0,4 0.8 1,2 1,G 2,o
MOLARITY OF SOLUTION
Figure 3.3. Dependence of the activity coefficient on the

ionic concentration of aqueous solution (Reference 19).
The selectivity of an ion exchange resin will also

depend on its crosslinking. The polymer structure of the ion
exchange resin can be thought of as collections of coiled
springs which can swell or contract during the exchange of
ions (20). The crosslinking of the polymer limits the extent

to which the resin may swell: the higher the degree of
crosslinking, the lower the extent to which the resin can be
hydrated. This limit on the extent to which the resin can be
hydrated determines the relative equivalent volumes of
hydrated ions which the crosslinked polymer network can
accommodate. This is shown in Table 3.2 (21) and in Figure
3.4 (22). As the resin’s degree of crosslinking or its fixed
ion concentration is lowered, the selectivity of the resin
decreases.
Table 3.2: Selectivity and Hydration of Cation Resins with

Different Degrees of Crosslinking (21)
4% DVB 8% DVB 16% DVB

Cation K
- -H -K----I! IT-II
Li 1.00 418 1.00 211 1.00 130
H 1.30 431 1.26 200 1.45 136
NS 1.49 372 1.88 la3 2.23 113
NH4 1.75 360 2.22 172 3.07 106
K 2.09 341 2.63 163 4.15 106
CS 2.37 342 2.91 159 4.15 102
Ag 4.00 289 7.36 163 19.4 102
Tl 5.20 229 9.66 113 22.2 a5
K = Selectivity compared to Li
H = Hydration (g H*O/eq resin)

-
The degree of crosslinking can affect the equilibrium

level obtained, particularly as the molecular weight of the
organic ion becomes large. With highly crosslinked resins
and large organic ions, the concentration of the organic ion
in the outer layers of the resin particles is’ much higher than
in the center of the particle. This is shown in Figure 3.5
for the distribution of methylene blue in carboxylic cation
exchange resins (23).
The selectivity of the resin for a given ion is also

influenced by the dissociation constants of the functional
group covalently attached to the resin (the fixed ion) and of
the counterions in solution. Since the charge per unit
volume within the resin particle is high, a significant
percentage of the functional groups may be un-ionized. This
Ion Exchange 171
0.7
0.6-- \ x
0.5 -
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
MOLE FRACTION NAR
Figure 3.4. Dependence of the selectivity coefficient for the

Na-H ionic exchange as a function of resin ionic composition
and resin crosslinking (Reference 22).
0 100 2ou
TIME (HR)
Figure 3.5. Sorption of methylene blue by carboxylic cation

exchange in the Na form. The resin for the top curve has
2% crosslinking while that for the lower curve has 10%
crosslinking (Reference 23).
is particularly true if the functional group is a weak acid or

base. For cation exchange, the degree of dissociation of the
functional group increases as the pH is increased. However,
the degree of dissociation for the ions in solution decreases
with increasing pH. Therefore, if a cation resin had weak
acid functionality, it would exhibit little affinity at any pH
for a weak base solute. Similarly, an anion resin with weak
base functionality exhibits little affinity at any pH for a
weak acid solute.
The influence of pH on the dissociation constants for

resin with a given functionality can be obtained by titration
in the presence of an electrolyte. Typical titration curves
are shown in Figure 3.6 for cation resins and in Figure 3.7
for anion resins (24). For sulfonic acid functional groups,
the hydrogen ion is a very weak replacing ion and is similar
to the lithium ion in its replacing power. However, for resin
with carboxylic acid functionality, the hydrogen ion exhibits
the highest exchanging power. Table 3.3 (25,26) summarizes
the effect different anion exchange resin functionalities have
on the equilibrium exchange constants for a wide series of
organic and inorganic anions.
PH
0 2 4 6 a 10 12
MEQ NAOH PER GRAM RESIN
Figure 3.6. Titration curves of typical cation exchange resins

(Reference 24).
Ion Exchange 173
Figure 3.7. Titration curves of typical anion exchange resins

(Reference 24).
Table 3.3: Selectivity Coefficients for Strongly Basic

Anion Resin (25, 26)
Type I Anion Type II Anion
Anion Anion
KXC1 KXCl
Salicylatt! 32.2 Salicylate 28
I- a.7 C&O 8.7
CGll50 5.2 I‘ 7.3
HSO,‘ 4.1 MO,- 6.1
NOa- 3.8 NOs- 3.3
Br‘ 2.8 nr- 2.3
CN- 1.6 CN- 1.3
HSOs- 1.3 Iwo,- 1.3
NOp- 1.2 NO*- 1.3
cl- 1.00 Cl- 1.00
HcoJ- 0.32 OH- 0.65
llZPOl 0.25 lico3- 0.53
HCOO- 0.22 112PO~ 0.34
cH3coo- 0.17 HCOO- 0.22
HzNCH2C00- 0.10 CHICDO- 0.18
Oil- 0.09 F- 0.13
F- 0.09 HYNCH,COO- 0.10

The selectivity is also influenced by the non-exchanging

ions (co-ions) in solution even though these ions are not
directly involved in the exchange reaction. An example of
this influence would be the exchange of calcium ascorbate
with an anion exchange resin in the citrate form. Although
calcium does not take part in the exchange reaction,
sequestering of citrate will provide an additional driving
force for the exchange. This effect, of course, would have
been diminished had a portion of the ascorbate been added as
the sodium ascorbate instead of the calcium ascorbate.
For non-polar organic solutes, association into

aggregates, perhaps even micelles, may depress solution
activity. These associations may be influenced by the co-
ions present.
The selectivity of synthetic ion exchange resins for

organic hydrophilic substances is usually low. The selectivity
constants for carboxylic cation resins (Na+ form) for
antibiotics of the group kanamycin, gentamycin, monomycin
and sisomycin are in the range 0.5-1.0 (27). A procedure was
developed (23) to prepare concentrated solutions of these
antibiotics. In a batch reactor, instead of merely allowing
the cation resin and the antibiotic fermentation broth filtrate
to come to equilibrium, the resin is first washed with a
complex forming agent before adding the antibiotic filtrate.
Such washing results in an additional sorption of the
antibiotic and higher elimination of both the other organic
constituents and inorganic ions from the ion exchange resin
as shown in Figure 3.8.
Many fermentation and biotechnology products are

ampholytes or zwitterions which can adsorb on ion exchange
resins not only by actual exchange but also by distribution of
the zwitterion segments and by the Donnan exclusion of the
ampholyte from the resin. A theory of ampholyte interaction
with ion exchange resins has been developed (23) which
allows a quantitative evaluation of the equilibrium state for
these systems. This evaluation can be used to select the
optimum conditions for adsorbing and desorbing the
ampholyte.
The distribution of a monoaminomonocarboxylic acid

Ion Exchange 175
1 2 3 4 5
FRACTIONS
Figure 3.8. Gentamycin desorption with a 0.5 M sulfuric

acid. Curves 1 and 4 are for gentamycin and magnesium
respectively after washing the cation resin with the complex
forming solution. Curves 2 and 3 are for gentamycin and
magnesium, respectively, without such washing (Reference 23).
between a strong acid cation resin and the solution may be

represented as:
Cation Exchange Resin
el” lk, IkD

11 (3.7)
AH+ + A+ + A- Solution
K1 K2
where AH+, A’, A-, AH+, Y@, K- are cation, zwitterion and
ampholyte anion in solution and on the ion exchange resin,
respectively.
Ki and Ki are the ampholyte ionization constants in

solution and in the ion exchange resin, respectively, KAH is
the constant of ion exchange of ampholyte cation - hyd:ogen
ion, KP is the distribution coefficient of zwitterions and K,
is the Donnan distribution coefficient of ampholyte as non-
exchangeable electrolytes.
The value of these constants for systems containing d-

phyenylglycine, 7-aminodesacetoxycephalosporanic acid,
ampicillin and cefalexin are shown in Table 3.4 when the ion
exchange resins are 4% and 8% crosslinked strong cation
resins and the anion is IRA-938. These constants can be
combined into “effective distribution coefficients”, K,,r,
which can be plotted as l/K,, versus pH to determine the
best adsorption and elution conditions. Examples of these
plots are shown in Figure 3.9.
Table 3.4: Electrochemical and Sorption Constants (23)

7-aminodesace-
Ian Exchrnge D-phenyl- toxycephalosporanic
Resin glycinr acid ampicillin cefalrxin
1.96 2.25 2.38 2.85
PKZ 9.02 4.85 7.07 7.15
KU-2-4 2.4 __ 2.7 __

PK,
Ku-2-8 2.6 2.7 __ __

PK,
IRA-938 9.1 4.8 __ 7.0

PKZ
KU-2-4 3.4 __ 6.3 __
KU-2-4 1.2 __ 2.9 -_
KU-I-B 4.6 4.0 _- __
Ku-2-8 1.0 1.4 -_ 4.6

KP
IRA-938 0.1 1.0 __ 1.4
IRA-938 0.9 1.0 __ 1.1

Ion Exchange 177
mg-equiv/g
b
1 gAi 0,4
0,2
l-6 gam g A-
1.6
DJI
a 12
&,
4 4 8 I.2
PH
Figure 3.9. Dependence of the amount of the ampholyte and

its separate electrochemical forms on the ion exchange resin
as a function of pH and mineral ion concentration. (a) d-
phenylglycine at 0.05 M, 0.1 M NaCl on an 8% crosslinked
cation resin, (b) curve (1) is 0.1 M d-phenylglycine and 0.1 M
NaCl; curve (2) is 0.1 M 7-aminodesacetoxycephalosporonic
acid and 0.1 M NaCl; curve (2) is 0.1 M 7-
aminodesacetoxycephalosporonic acid (7-A D C A) and 0.1 M
NaCl; curves (3), (4) and (5) are 0.05 M d-phenylglcyine and
0.1, 0.01 and 0.001 M NaCl, respectively. All are on an 8%
crosslinked cation resin. (c) Curves (l), (2) and (3) are 0.1
M d-phenylglycine, cefalexin and 7-A D C A, respectively,
and 0.1 M NaCl on Amberlite IRA-938 anion resin (Reference
23).
Selectivity calculations may be used to define the

operational limits of an ion exchange process in terms of
maximum exhaustion capacity, degree of regeneration or the
“leakage” from a column operation. The “leakage” is the
appearance of a low concentration of an undesired solute in
the column effluent during the operating portion of the
cycle. This may be due to the ionic form of the resin at the
bottom of the column and to the composition of the solution
moving down the column.
Example 3.1
Anderson (28) used selectivity calculations to evaluate
the removal of calcium ions from a 40% monosodium
glutamate solution. When the feedstream has 0.15% calcium,
the resin has a total exchange of 2 eq/L and a selectivity of
Cat+ over Nat of 3, and the monosodium glutamate has an
equivalent weight of 169 g/eq, the relative equivalent
concentrations of cations are:
+ 400 g/l
I I
Na =
169
m 2.37 eq/l (3.8)
gleq
1.5 g/l
Ica++ 1 = 20 x?/eq
I 0.075 eq/l (3.9)
The equivalent fraction of Ca in solution is:

0.075
X = 0.031 (3.10)
Ca++ =
2.45
The equivalent fraction of Ca in the resin is:
Tica++ ++ c x
CZa++
= Ca+ - (3.11)
KNa
(1 - XCa++) C (1 - XC_++)*
Since C = 2 eq/L and c = [Nat] t [Cat+] = 2.45 eq/L,
Xca++ 2 0.031
3--- = I-J.081 (3.12)
(1 - ZCa++) = 2.45 (0.969)2
So that
Xca++ = 0.07 (3.13)
Ion Exchange 179
This means that only 7% of the resin capacity, at best, can

be used under these circumstances. Therefore it would be
advisable to use other methods to reduce the Ca++
concentration in the feedstream.
3.2.2 Kinetics
Theoretical considerations for the selectivity of ion
exchange resins are well developed. Unfortunately, this is
not the case with the kinetics of ion exchange reactions.
The problems posed by systems more complex than the “ideal”
case of binary ion exchange of strong electrolytes with
either strong acid cation or strong base anion resins continue
to be the subject of research papers (29) and conferences
(30).
The overall exchange process may be divided into five

sequential steps: (1) the diffusion of ions through the
solution to the surface of the ion exchange particles, (2) the
diffusion of these ions through the ion exchange particle, (3)
the exchange of these ions with the ions attached to the
functional group, (4) the diffusion of these displaced ions
through the particle, and (5) the diffusion of these displaced
ions through the solution. Each step of the diffusion,
whether in the resin or solution phase, must be accompanied
by an ion of the opposite charge to satisfy the law of
electroneutrality.
Kinetics of ion exchange is usually considered to be

controlled by mass transfer in ion exchange resin particles or
in the immediately surrounding liquid phase. The theory used
to describe mass transfer in the particle is based on the
Nernst-Planck equations developed by Helfferich (31) which
accounted for the effect of the electric field generated by
ionic diffusion, but excluded convection.
The Nernst equation is (32):

RT
(I, = k+- In a (3.14)
P
ZF
where R is the ideal gas constant, F is the Faraday constant,

T is the absolute temperature, z is the charge on the
potential determining ion, ap is the activity of the potential
determining ion in the bulk solution and k is a constant.
From this, the generalized Nernst-Planck equation can

be written as:
F
Ji = -Di ( grad ci + zi ci - grad $1 (3.15)
RT
flux diffusion electric transference
Whereas the kinetics of adsorption is described using

Fick’s Law models, the Nernst-Planck equations, which take
into account the principal coupling effect of ionic fluxes,
provide a significant improvement over Fick’s Law in
describing ion exchange kinetics with only a minor increase
in complexity.
The important distinction between the kinetics of ion

exchange compared to the kinetics of adsorption is that the
forward and reverse exchange of the same two ions of
different mobilities will occur at different rates and with
different behavior of the concentration profiles. This is
shown in Figure 3.10 (33).
1.0
v A
‘V
A
V
4
V
A
V A
V
LLY
A
0 CONVERSION OF H+ TO NA” FORM

0 CONVERSIONOF NA+ TO H+ ~~~~
v CONVERSIONOF NA+ TO SR”FORM

a CONVERSIONOF SR+~ TO NA+ FORM
0 I I I I I
3 5 10 20 60 120 300 600 1000
TIME (SEC)
Figure 3.10. Comparison of forward and reverse ion

exchange rates (Reference 33).
Ion Exchange 181
The Nernst-Planck equations for ion exchange kinetics

are still the dominant theory in use today (34). It is
recognized that the theory fails to take into account the
effect of swelling and particle size changes which accompany
ion exchange or to take into account the slow relaxation of
the resin network which causes the diffusion coefficient to
vary with time. However, the approximations which the
Nernst-Planck equations provide are a reasonable starting
point and will most likely be found to be sufficient for the
biotechnology engineer. Any further refinements would
rapidly lead to diminishing returns. Therefore, the mass
transfer in the liquid phase is usually described according to
the Nernst film concept using a version (35) of the Nernst-
Planck equation or Glueckauf’s (36) simpler linear driving
force approximation.
The earliest theories of ion exchange kinetics assumed a

mechanism which was controlled by a chemical exchange
reaction coupled with diffusion. Helfferich (37) developed a
detailed mathematical theory of kinetics of ion exchange
which is accompanied with fast ionic reactions, such as
dissociation equilibrium of weak acid groups. In these cases,
the rate is controlled by diffusion as the limiting process,
but it is dramatically affected by reaction equilibrium, which
controls the concentration of diffusing species.
There are five models (38) which can be used to

represent the kinetics in ion exchange systems which involve
liquid phase mass transfer, solid phase mass transfer and
chemical reaction at the exchange group.
In Model 1, the liquid phase mass transfer with a linear

driving force is the controlling element. This model assumes
that there are no concentration gradients in the particle,
that there is a quasi-stationary state of liquid phase mass
transfer, that there is a linear driving force and that there
is a constant separation factor at a given solution
concentration.
In Model 2, the rate-controlling step is diffusion within

the ion exchange particles. This model assumes that there
are no concentration gradients in the liquid phase and that
there is no convection either through solvent uptake or
release in the solid phase.
I.82 Separation and Purification Techniques in Biotechnology
Model 3 is controlled by the exchange reaction at the

fixed ionic groups. This model assumes that the slowness of
the exchange reaction allows sufficient time for mass transfer
to establish and maintain equilibrium so that no concentration
gradient exists in either the ion exchange particles or in the
liquid phase.
Model 4 is a variation of Model 3 in which the

counterions from the solution do not permeate beyond the
portion of the particle which has been converted to the
exchanging ionic form. The boundary of the unreacted core
reduces with time, such that this is called the shrinking core
model. It is this sharp boundary between the reacted and
unreacted portion of the particles that distinguishes Model 4
from Model 3.
In Model 5, the rate controlling step is the diffusion of

the counterion across the converted portion of the particle.
Since the exchange groups undergo a fast and essentially
irreversible reaction with the counterions, their type of
reaction affects the rate of reaction and the geometry of the
diffusing zone.
Table 3.5 (38) summarizes the effect of operating

parameters (particle size, solution concentration, separation
factor, stirring rate, resin exchange capacity, and
temperature) on ion exchange kinetics described by these
different models in batch reactors.
For the cases of interest, the rate of ion exchange is

usually controlled by diffusion, either through a hydrostatic
boundary layer, called “film diffusion” control or through the
pores of the resin matrix, called “particle diffusion” control.
In the case of film diffusion control, the rate of ion

exchange is determined by the effective thickness of the film
and by the diffusivity of ions through the film. When resin
particle size is small, the feedstream is dilute or when a
batch system has mild stirring, the kinetics of exchange are
controlled by film diffusion.
In the case of particle diffusion control, the rate of ion

exchange depends on the charge, spacing and size of the
diffusing ion and on the micropore environment. When the
resin particle size is large, the feedstream is concentrated or
Ion Exchange 183
Table 3.5: Dependence of Ion Exchange Rates on

Experimental Conditions (38)
Factor Model 1 node1 2 tlodel 3 Model 4 tfodcl @
Particle size (r) (II/r al/r2 independent al/r cY’/rZ
Solution ac independent a c a.2 2

concentration (cl
Separation factor independent independent@ independent independent independent@

(a 1 up to P
specific
time when
t1;aol
when a C<l.
Stirring rate sensitive independent. independent independent independent

(rfxr)
Resin excbfnge al/c independent independent independent a L/Z

capacity (cl
Tewper~turr (T) S&%/OK -6%1°K function of function of -6%/‘K

E E
Act Act
@ Applicable to forward exchange only.
@ Provided partition coefficient is independent of solution concentration.
@ For complete conversion and constant solution composition.
when a batch system has vigorous stirring, the kinetics are

controlled by particle diffusion.
The limits at which one or the other type of diffusion

is controlling have been determined by Tsai (39). When Kk26
> 50, the rate is controlled by film diffusion. When Kk2S <
0.005, the rate is controlled by particle diffusion. In these
relationships, K is the distribution coefficient, k2 is the
diffusivity ratio (D,/Dr), 6 is the relative film thickness ((b-
a)/a), D, and D, are the pore diffusion and film diffusion
coefficients, respectively, and (b - a) is the film thickness on
a resin particle with a radius a. Between these two limits,
the kinetic description of ion exchange processes must
include both phenomena.
The characteristic Nernst parameter “S”, the thickness

of the film around the ion exchange particle, may be
converted to the mass transfer coefficient and dimensionless
numbers (Reynolds, Schmidt and Sherwood) that engineers
normally employ (40).
In terms of the solute concentration in the liquid,

between 0.1 and 0.01 mol/L, the rate limiting factor is the
transport to the ion exchange bead. Above this

concentration, the rate limiting factor is the transport inside
the resin beads (41). During the loading phase of the
operating cycle, the solute concentration is in the low range.
During regeneration, however, in which the equilibrium is
forced back by addition of a large excess of regenerant ions,
the solute is above the 0.1 mol/L limit.
One of the important factors in the kinetic modeling of

organic ions is their slow diffusion into the ion exchange
resin. The mean diffusion time is listed in Table 3.6 as a
function of resin particle size for different size
classifications of substances (42). With the larger organic
ions, the contact time for the feed solution and the resin
must be increased to have the ion exchange take place as a
well defined process such as occurs with the rapidly diffusing
ions of mineral salts.
Table 3.6: Characteristic Diffusion into Spherical Resin Particles

for Various Substances (42)
Coefficient_ of Diffusion tleau Time of
(Order of tlagnitude) Type of Sorbed Particle Intraparticle
(CUl2/srC) Substance (Ion) Radius (cm) Diffusion
10-6 Ions of mineral salts 0.5 3 qio

bearing a single charge 0.01 1 set
0.005 1.8 set
10-7 Ions of mineral salts 0.05 30 mia

bearing several charges, 0.01 1.2 min
amino acids 0.005 18 set
10-n Trtraalkylauuonium 0.05 5 hr

ions, antibiotic ions 0.01 12 min
on inacroporous resins 0.005 3 min
10-S Dyes 9alkaloids, 0.05 over 2 days

antibiotics in standard 0.01 2 hr
ion exchage resins 0.005 0.5 hr
lo-‘0 Some dyes, polypeptidrs 0.05 over 20 days

and proteux 0.01 over 20 hr
0.005 over 5 hr
A mathematical model has been developed for the

kinetics of ion exchange in batch systems (43). The
fractional attainment of equilibrium at a specific time was
developed in terms of the parameters c, CY,K, k and S.
Ion Exchange 185
The parameter E represents the ratio of the forward

chemical reaction rate to the intraparticle diffusion rate. As
< approaches infinity, the surface chemical reaction becomes
instantaneous.
The parameter CY represents the ratio of the volume of

the liquid solution to the volume of the solid ion exchange
resins. As would be expected, the exchange rate increases as
the proportion of the resin increases.
K represents the distribution coefficient of the resin

since it is the ratio of the rate constants of the forward and
reverse ion exchange reactions, each multiplied by the solute
concentration in the particle and in the solution,
respectively. The value of K decreases as the solute
concentration in the liquid phase increases. Likewise, the
reaction rate increases as K decreases. This is shown in
Figure 3.11.
1.
0.
0.
0.
0 ’
.I
0.1
0 0.1 0.2 0.3 0.4 0,5
Figure 3.11. Plots of log (1-F) vs T for various values of K

(Reference 43).
The parameter k compares the intraparticle diffusion

rate to the film diffusion rate. For small value of k, the
exchange rate is controlled by a combination of surface
chemical reaction and intraparticle diffusion. For large
values of k, the rate is controlled by a combination of
surface chemical reaction and film diffusion. This is shown
in Figure 3.12.
0 5 10 15 20 25
T/K2
Figure 3.12. Plots of log (1-F) vs T/k for various value of k

(Reference 43).
The parameter S measures the ratio of the film

thickness to the radius of the ion exchange particle. When 6
is small, the film diffusion resistance to mass transfer is
negligible so that the exchange rate is controlled by a
combination of surface chemical reaction and intraparticle
diffusion. As 6 increases, the effect of film diffusion
becomes important in controlling the exchange rate.
Example 3.2
With a particle diffusion of 1 x 10-8cm2/sec for an
amino acid such as lysine, the parameter k changes from
Ion Exchange 187
0.0383 to 0.0436 to 0.0470 as the ion exchange bead diameter

increases from 420 to 710 to 840 microns for a 0.05M
solution stirred at 300 rpm. The corresponding values of
Kk26 calculated range from 0.173 to 0.012 to 0.003,
respectively. In the first two cases, the value is between
0.005 and 50 so that the exchange rate should be expected to
be controlled by both film and intraparticle diffusion. In the
last case, the bead size has become sufficiently large that
the exchange rate should be controlled by particle diffusion.
It should be apparent that the kinetics and mass

transfer equations describing the ion exchange resin column
operation are in many respects similar to what was seen with
column operations in the chapter on adsorption.
The set of equations to be solved to model the ion

exchange process in fixed bed operations are:
a2c ac ac P a¶
D-_ = -+u-+-- (3.16)
az2 at a2 E at
aq
__ = Ka (c - c*) (3.17)
at
*
c = bWa (3.18)
K = A e-e/qmD) + D e-Ehh,) (3.19)

a
As with mass transfer in adsorption, these equations

have been solved using numerical analysis and analytical and
graphical solutions (44-47). The solution at short times is
given by (44):
c = c
cl ex
erfc ___
h (3.20)
26
ii
where dimensionless time, r, is equal to:
and the dimensionless

T =
column
c)
t--
length,
UX
”
__
X, equals:
R2
(3.21)
5X
Kd
,x = 3 (1 - a) - (3.22)
R2 v
For long times, the solution can be written in terms of

modified Bessel functions to give:
2lrJ2X(T - 0.131) _r2,sA

c = c0 e e IL(C) dS (3.23)
0 I
These two solutions can be combined to describe the mass

transfer in an ion exchange column at any value of r.
Breakthrough curves using these equations are shown in
Figure 3.13.
0 0.5 1.0 1.5 2.0
T - To (HR)
Figure 3.13. Breakthrough curves (x = 26.6 cm, v = 100 ml-

cmW2 hr-l, K = 5, 01 = 0.3 and R = 0.005 cm). Irregular
process: Cur;e (1) D = lo-” cm”/sec X = 0.004; Curve (2) D
= 10-10 cm2/sec, X = 0.04; Curve (3)’ D = 10-Q cm2/sec, X =
0.4; Regular Process: Curve (4), D = lOba cm2/sec, X = 4.0
(Reference 42).
Example 3.3
Breakthrough curves have been calculated using these
Ion Exchange 189
equations for the exchange of morphocycline onto a strong

acid cation resin. They are compared in Figure 3.14 with
experimental data. While the initial portion of the curves
are calculated to be greater than the experimental values, the
agreement becomes better as the value of r is increased.
0,a
0 0
0 I I I
0 0.2 0,4 0.6
Figure 3.14. Comparison of calculated and experimental

breakthrough curves for morphocycline sorbed onto a 4%
crosslinked cation resin at X values of 0.1, 0.25, and 0.8 for
curves (I), (2) and (3), respectively (Reference 42).
The shape of the breakthrough curve is largely

dependent on the type of exchange isotherm obtained under
equilibrium conditions (48). Three different types of
isotherms, linear, convex and concave, are possible as model
situations (Figure 3.15). Actual isotherms may have
inflection points due to variations in selectivity with resin
composition.
The simplest type, a linear isotherm, is only valid when

the exchange is between ions of the same valence or when a
second ion is present in great excess in the feedstream. A
-_____--____---_--
SOLUTE CONCENTRATION
Figure 3.15. Types of exchange isotherms: (A) convex, (B)

linear and (C) concave.
linear isotherm was used in calculating the values in Example

3.3. A closer fit may be obtained using an alternate
isotherm as shown in Figure 3.16. The more typical
isotherm, the convex type, occurs when the ion to be taken
up has a higher affinity for the resin than the original ion
in the resin. The concave isotherm represents the situation
where the ion to be taken up is attracted less strongly to
the resin than the original ion.
In principle, fluidized ion exchange beds are similar to

stirred tank chemical reactors. The general equations of
kinetics and mass transfer can be applied to the individual
fluidized units in an identical manner to those for chemical
reactors. The primary difference lies in accounting for the
behavior of suspended particles in the turbulent fluid (49).
Ion Exchange 191
0 1 2 3
Figure 3.16. Relative retention volumes based on the

infection points of breakthrough curves for oxyglucocycline (0)
and morphocycline (0) plotted as a function of the
dimensionless parameter A. The solid lines are calculated
using (1) rectangular and (2) linear exchange isotherms
(Reference 42).
The operation of these fluidized ion exchange beds is

identical to that of fixed beds, with the exception that the
resin of each stage is confined by perforated plates and
maintained in a fluidized suspension using liquid flow or
impellers.
The critical design parameter for fluidized beds is the

loss or leakage of the solute through a given stage. The
design equation for a single stage bed is given by (50):
i = 1 - [ UK+aK
a ] exp (- ---&( ,“+“,“:) (3*24)
where c is the concentration of the solute leaving the bed at

time t; co is the concentration of the solute in the
feedstream; K is the overall mass transfer coefficient; m is
the solute distribution coefficient; U is the solution flow
rate; a is the transfer surface area per stage; and V, is the
settled volume of resin in the stage. Once the flow rate,
feed concentration and desired final effluent concentration

are specified, this equation can be used to determine the
combination of m, V,, K and t which can be used,
When multiple stage operations are employed, it is

necessary to determine the loss from each prior stage to
determine the loss from stage “n”. The general equation is:
L-
c
= e2
+- .-BN’m $BNim [ +] ,i~ (3.25)
m
co
K a
where B = (3.26)
U+Na
and N, the number of bed volumes of solution, is equal to:

t Q
N = ~ (3.27)
n V
r
For the second and third stage, this equation becomes:
-=c2 * _ 6 + _f!l (3.28)

c m
0
and
c3 36%
__ = 1 - fie-BN'm - 38 + B2 + ___ - 2B2!4 +
C0 m
Equation 3.25 can then be used along with Equations

3.17 - 3.19 to model the ion exchange process in fluidized
bed operations.
Example 3.4
Experimental data for a three stage system using Dowex
1x4 resin to adsorb thorium nitrate are shown in Figure 3.17
for different experimental conditions. The three curves are
calculated using (I) a single stage bed with 900 cc of resin,
(II) a two stage bed with 450 cc of resin each and (III) a
three stage bed with 300 cc of resin each. The data should
follow curve III if there is no intercompartment mixing of
resin. It is obvious from the data that stacked,
multicompartment beds should be fluidized to the minimum
Ion Exchange 193
extent possible to allow passage of suspended particles while

minimizing interstage mixing of the resin.
1 10 30
Q
N=_
3%
Figure 3.17. Experimental data and calculated curves for

three stage multicomponent beds (Reference SO).
3.3 ION EXCHANGE MATERIALS AND THEIR PROPERTIES
Ion exchange materials are a special class of

polyelectrolytes. The chemical and physical properties of an
ion exchange material play a more important role in
1% Separation and Purification Techniques in Biotechnology
determining its suitability for a biochemical application than

for other types of applications. The chemical properties to
be considered are the matrix and the ionic functionality
attached to the matrix. The important physical properties
are the pore size, the pore volume, the surface area, the
density and the particle size.
3.3.1 Ion Exchange Matrix

These materials can be broadly categorized into those
which are totally inorganic in nature and those that are
synthetic organic resins.
Inorganic ion exchangers (5 1) include both naturally

occurring materials such as mineral zeolites (sodalite and
clinoptilolite), the greensands, and clays (the montmorillonite
group) and synthetic materials such as gel zeolites, the
hydrous oxides of polyvalent metals (hydrated zircon$m’&
oxide) and the insoluble salts of polybasic acids
polyvalent metals (zirconium phosphate).
A history of the development of synthetic ion exchange

resins and a review of current methods of synthesizing
polymeric ion exchange resins has been presented by Millar
(52).
The synthetic organic resins consist of a crosslinked

polymer matrix which is functionalized to provide their ion
exchange capacity. The matrix usually must undergo
additional reactions to provide the strong acid cation, strong
base anion, weak acid cation or weak base anion
functionality.
The most commonly used materials as the matrix for

synthetic ion exchangers are polymers based on the
copolymerization of styrene crosslinked with divinylbenzene
(Figure 3.18). These copolymers can be used to form strong
acid cation, strong base anion (Type I and Type II), and
weak base anion resins.
Weak acid cation exchangers (Figure 3.19) may be

formed by the direct polymerization of acrylic acid or
methacrylic acid with divinylbenzene as the crosslinking
agent. Similarly, the acrylate- methacrylate-
divinylbenzene copolymer may be formedOr which can be
further reacted to have strong base anion or weak base anion
functionality.
Ion Exchange 195
t +
H2SO‘l Chloromethylation
SO;H+ CH,CI
NW,), N(CH,),CH, CH,OH WH,),
I ,CH, 1 ,CH, I ,CH3

CH,N+-CH,CI- CH,N+-CH, Cl- CH,N :
\ \ \
CH, CH,CH,OH CH,
Figure 3.18. Styrene-divinylbenzene resins functionalized

with strong acid, strong base I, strong base II and weak base
groups.
7%
C=CH, +
&OOH
CH = CH,
-CH-CH,-
Figure 3.19. Methacrylic acid-divinylbenzene resins forming

weak cation resins.
1% Separation and Purification Techniques in Biotechnolo~
Epoxy-polyamine resins (Figure 3.20) are formed by the

direct polymerization of a chloroepoxide with ammonia or an
amine. The molecular weight of this type of polymer is
comparatively low with respect to the acid-adsorbing nitrogen
sites. This results in a high capacity acid-adsorbing weak
base. resin.
H z- C!L-CH 2
Cl+ NH 3
Polymerize
I
Figure 3.20. Epoxy-polyamine resin formed from chloroepoxide

and ammonia.
In the past, ion exchange matrices have been formed by

the reaction of phenol and formaldehyde (Figure 3.21). These
resins have, to a large extent, been replaced by other resins
with their inherently higher capacity, higher thermal stability
and longer life.
Figure 3.21. Phenol-formaldehyde resin matrix.
These four synthetic organic resins are the most

commonly used in industrial applications and have been used
in biochemical applications and even in protein purifications
and enzyme immobilizations. However, the hydrophobic
matrices have the disadvantages that they might denature the
desired biological material or that the high charge density
may give such strong binding that only a fraction of the
adsorbed material might be recovered.
Ion Exchange 197
Resins with cellulosic matrices (Figure 3.22) are much

more hydrophilic and these do not tend to denature proteins.
Cellulosic resins have been used extensively in the laboratory
analysis of biological material, enzyme immobilization and
small scale preparations. The low capacity and poor flow
characteristics limited the usefulness of these matrices for
larger applications.
_o+o$$&
CHZOH
R = ‘PO3H2
-S020CH2CH2NH~
-CH2COOH
-Ct12CH2NRi+ _
-CH2CH2NR3 X
-o~Oq-&o_
- -CH2-CH2-COOH
ti
Figure 3.22. Cellulosic resins.
Recently, diethylaminoethyl(DEAE)-silica gel was shown

(53) to be an improvement over typical cellulosic matrix
resins for the separation of acidic and neutral lipids from
complex ganglioside mixtures. The specific advantages
claimed were:
(1) An increase in flow rate was possible

through the DEAE-silica gel.
(2) The DEAE-silica gel was able to be

equilibrated much more rapidly with the
starter buffer.
(3) The DEAE-silica gel was more easily

regenerated.
(4) The DEAE-silica gel was less susceptible

to microbial attack.
(5) The preparation of DEAE-silica gel from

inexpensive silica gel was described as a
simple method that could be carried out
in any laboratory.
The kinetic and electrochemical properties of ion

exchange resins depend on the nature of the constituents
used in their synthesis. The changes in the initial diffusion
coefficients of organic ions inside the resin particles is
shown in Figure 3.23 for cation resins synthesized from
various monomers (23). The use of maleic acid provides a
significant increase in the exchange kinetics. However, such
cation resins are capable of dissociation at higher pH
compared to resins synthesized with acrylic acid. The pK =
7.0 and 5.5 for resins synthesized from maleic acid and
acrylic acid, respectively.
I
5 10 15 20
PERCENT DIVINYLBENZENE
Figure 3.23. Diffusion coefficients (cm2/sec) of streptomycin

in the antibiotic sorption by (1) gel and (2) modified
carboxylic cation exchangers with various crosslinking
(Reference 23).
3.3.2 Functional Groups
The strong acid cation exchange resins are made by the

sulfonation of the matrix copolymer. The resulting resin for
Ion Exchange 199
styrene-divinylbenzene copolymer has the polymer structure

shown in Figure 3.18 with the SO, attached. Associated with
that functional group is the exchangeable ion, in this case,
the hydrogen ion. Strong acid cation resins are
characterized by their ability to exchange cations or split
neutral salts. They will function throughout the entire pH
range.
The synthesis of weak acid cation resins has been

described above. The ability of this type of resin to split
neutral salts is very limited. This resin has the greatest
affinity for alkaline earth metal ions in the presence of
alkalinity. Only limited capacities for the alkali metals are
obtained when alkalinity other than hydroxide is present.
Effective use is limited to solutions above pH 4.
The anion exchange resins require the synthesis of an

active intermediate. This is usually performed in the process
called chloromethylation. The subsequent intermediate is
reactive with a wide variety of amines which form different
functional groups. Figure 3.18 illustrates the structure of
the Type I strong base anion resin, the Type II strong base
anion resin, and a tertiary amine weak base anion exchange
resin.
The Type I resin is a quaternized amine resin resulting

from the reaction of trimethylamine with the
chloromethylated copolymer. This functionalized resin has
the most strongly basic functional group available and has
the greatest affinity for weak acids. However, the efficiency
of regenerating the resin to the hydroxide form is somewhat
lower than Type II resins, particularly when the resin is
exhausted with monovalent anions.
The Type II resin results when dimethylethanolamine is

reacted with the chloromethylated copolymer. This
quaternary amine has lower basicity than that of the Type I
resin, yet it is high enough to remove the anions of weak
acids in most applications. While the caustic regeneration
efficiency is significantly greater with Type II resins, their
thermal and chemical stability is not as good as Type I
resins.
Weak base resins may be formed by reacting primary or

secondary amines or ammonia with the chloromethylated
copolymer. Dimethylamine is commonly used. The ability of

the weak base resins to adsorb acids depends on the basicity
of the resin and the pK of the acid involved. These resins
are capable of adsorbing strong acids in good capacity, but
are limited by kinetics. The kinetics may be improved by
incorporating about 10% strong base capacity. While strong
base anion resins function throughout the entire pH range,
weak base resins are limited to solutions below pH 7.
The desired functionality on the selected matrix will be

determined by the nature of the biochemical solute which is
to be removed from solution, its isoelectric point, the PH
restrictions on the separation and the ease of eventually
eluting the adsorbed species from the resin.
Some resins have been developed with functional groups

specifically to adsorb certain types of ions. The resins
shown in Table 3.7 are commercially available.
Table 3.7: Commercial Resins with Special Functional Groups
Functionality Structure
Iminodiacetate R-CH,N(CH,COOH),
Polyethylene Polyamine R-(NCzHAmH
Thiol R-SH
Aminophosphate R-CH,NHCH,POsH,
Amidoxime R-C=N-OH
I
NH2
Phosphate R-PO,H
The selectivity of these resins depends more on the

complex that is formed rather than on the size or charge of
the ions. Generally they are effective in polar and non-polar
solvents. However, the capacity for various ions is pH
sensitive so that adsorption and elution can be accomplished
by pH changes in the solution.
These chelating resins have found most of their use in

metal ion recovery processes in the chemical and waste
recovery industries. They may find use in fermentation
applications where the cultured organisms requires the use of
Ion Exchange 201
metal ion cofactors. Specific ion exchange resins have also

been used in laboratory applications that may find eventual
use in biotechnology product recovery applications (54).
A review of selective ion exchange resins has been

compiled by Warshawsky (55). A diaminotetraacetic polymer
developed by Mitsubishi (56) was developed for the
purification of amino acid feed solutions. The conversion of
chloromethylpolystyrene into thiolated derivatives for peptide
synthesis has been described by Warshawsky and coworkers
(57).
3.3.3 Porosity and Surface Area

The porosity of an ion exchange resin determines the
size of the molecules or ions that may enter an ion exchange
particles and determines their rate of diffusion and exchange.
Porosity is inversely related to the crosslinking of the resin.
However, for gel-type or microporous resins, the ion
exchange particles has no appreciable porosity until it is
swollen in a solvating medium such as water.
The pore size for microporous resins is determined by

the distances between polymer chains or crosslinking
subunits. If it is assumed that the crosslinking is uniform
throughout each ion exchange particle, the average pore
diameter of these resins can be approximated from the water
contained in the fully swollen resin. The moisture content of
cation resins as a function of the degree of crosslinking is
shown in Figure 3.24 and of anion resins in Figure 3.25. The
calculated average pore size for sulfonic cation resins ranges
from 16 to 20 A as the resin crosslinking is decreased from
20 to 2% divinylbenzene. The calculated average pore size of
the anion resins ranges from 18 to 14 A as the crosslinking
is decreased from 12 to 2%. Even at low crosslinking and
full hydration, microporous resins have average pore
diameters of less than 20 A. The dependence of pore size on
the percent of crosslinking is shown in Table 3.8 for swollen
microporous resins of styrene-divinylbenzene (58).
Sulfonic acid and carboxylic acid resins have also been

equilibrated with a series of quaternary ammonium ions of
different molecular weights to measure the average pore size
when these resins have increasing degrees of crosslinking.
These results are shown in Figures 3.26 and 3.27.
100 -
20 _-
0 . I I I I
1 2 4 8 20
Figure 3.24. Moisture content of strong acid cation resins as

a function of divinylbenzene content.
80
0
1 2 4 8 20
Figure 3.25. Moisture content of strong base anion resins as

a function of divinylbenzene content.
Ion Exchange 203
Table 3.8: Average Swollen Diameter of Crosslinked Polystyrene

Beads in Tetrahydrofuran (58)
Divinylbenzene
Concentration (Crosslinking) Swollen Pore Diameter
(%I (%>
1 77
2 54
4 37
s 14
16 13
14
12
10
LARGE ION DATA
0 5 10 15 0
Figure 3.26. Average pore diameter of sulfonic acid cation

exchange resin as a function of degree of crosslinking
(Reference 15, page 46).
34
30
26
2
0 I I I I I I I
0 4 8 12 16
Figure 3.27. Average pore diameter of carboxylic acid cation

exchange resin as a function of degree of crosslinking
(Reference 15, page 47).
Figure 3.28 shows the change in the ionic diffusion

coefficients of tetraalkylammonium ions in strong acid cation
resins as a function of mean effective pore diameter of the
resins (59). As the pore diameter is increased, the
penetrability of the resins with respect to the large ions also
increased.
Ion Exchange 205
RESIN PORE DIAMETER th
Figure 3.28. Ion ((1) tetramethyl; (2) tetraethyl; (3)

tetrabutyl ammionium ions) diffusion coefficients in
macroporous sulfonated cation resins (Reference 59).
If an inert diluent (porogen) is incorporated into the

monomer mixture before the copolymer is formed, it is
possible to form a structure containing varying degrees of
true porosity or void volume (60,61). Variations in the
amount of divinylbenzene crosslinking and diluent allow for a
range of particle strengths and porosity to be made.
Subsequent reactions with the appropriate chemicals result in
the introduction of the same functional groups as discussed
above. These are called macroporous or macroreticular
resins.
A list of the porogens can be compiled from the Ashahi

patents (62-64) for specific monomer and comonomer systems

(Table 3.9). These substances can be classified according to
their solubility coefficient, 6, their polarity and according to
their solvent capability for homopolymers. The solvent
capability is divided into three categories (65):
(1) where it is a solvent for all

homomonomers of the monomer used
(2) where it is non-solvent for all
homopolymers of the monomers used
(3) where it is a solvent for some and a
non-solvent for other homopolymers of
the polymers used.
Table 3.9: Classification of Porogens from Asahi Patents (65)

Solvent Capability Category @
Examples Solubility
SOlVellt Polarity 1 z 2 4 s Coefficient, 6
hrxanrs __ 2 2 2 2 2 17
Il~ptalleS _- 7 2 2 7 2 7.5
octanes __ 2 2 2 2 2 7 - 7.5
drcsllrs __ 2 7 2 ? 2 7.5 - a
cyclohexane __ 2 2 2 2 a
diisopropyl ketone + 3 a
ethyl benzoate + 3 1 3 a.2

ethyl propionate + 3 1 3 a.4
methyl isobutyl ketone * 3 131 3 a.4
benzooitrile (+) 1 1 a.4
butyl acetates + 3 3 a.3 - a.5
n-bury1 propionate + 3 3 3 a.8
xy1enes __ 3 1 3 1 8.a
ethylbenzenes __ 3 3 I 3 a.8
to1ue11e __ 3 1 3 1 3 0.9
ethyl acetate + 3 3 3 3 9.1

chloroform (+I 1 9.3
metbyl rtbyl ketone + 3 1 L31 3 9.3
P-nitropropane (+I 1 I 131 9.9
cyclohrxanone + 1 1 131 9.9
o-dichlorobenzene (+I 3 1 1 10
~ruyl alcohols + 3 3 2 10 - 11
octanols + 3 3 2 2 ‘*10.5
methyl brnzoate t 3 1 3 10.5
butaools + 3 3 10.5 - 11.5
IleXJllOls + 3 3 2 2 $10.1
oyclollrr.ilIol t 3 3 2 2 11.4
Ion Exchange 207
The simplest system is merely the use of a porogen

from category 3. When a porogen from category 2 is added
to a porogen from category 3, the pore diameter of the final
resin increases. Conversely, when a porogen from category 1
is added to a category 3 porogen, the pore diameter
decreases. A mixture of porogens from categories 1 and 2
results in a wide variation in pore size distribution as the
ratio of the porogens is changed. Fine control on the pore
size distribution can be obtained by using mixtures of
porogens from category 3.
Since each bead of a given external diameter that is

made by the inert diluent process will contain some void
volume, there is actually less polymer available per unit
volume for the introduction of functional groups. Therefore,
these macroporous resins are inherently of lower total
exchange capacity than gel-type resins of the same
composition.
Macroporous resins are most useful when extremely

rigorous osmotic shock conditions are encountered, when the
very high porosity is desirable from the standpoint of the
molecular weight of the material being treated or when non-
polar media are involved. The drawbacks of using
macroporous resins are poorer regeneration efficiencies, lower
total exchange capacities and higher regeneration costs.
Until the advent of macroporous resins, the synthetic

organic ion exchange resins were of such low porosity that
large proteins and other macromolecules would be adsorbed or
interact only with the exterior exchange sites on the resins.
Therefore, although the microporous resins may have higher
total exchange capacities than macroporous resins, the
effective capacity of macroporous resins for protein or
macromolecule absorption may often times be greater than
that of microporous resins.
Typical macroporous ion exchange resins may have

average pore diameters ranging from 100 A to 4000 A. Table
3.10 shows the pore sizes of several resins of different
matrices that have been used in enzyme immobilization (66).
Pore volumes for macroporous resins may range from 0.1 to
2.0 cc/g. The relationship between pore diameter and
porosimeter pressure is shown in Table 3.11 for mercury
intrusion porosimetry based upon the Kelvin equation.
Table 3.10: Physical Properties and Capacities for

Ion Exchange Resins (66)
Adsorption
Resin Pore Surface Resin Capacity
Matrix Functionality Size ADZa Capacity for Enzyme
phenolic 3O polyethylene 250 II 68.1 q=/g 4.38 meq/g 3.78 meqfg

polyamine
phenolic partially 3O 290 R 95.3 m=/g 4.24 meq/g 3.51 meq/g

polyethylene
polyamine
polystyrene polyethylene 330 II 4.6 mZ/g 4.20 meq/g 3.92 meq/g

polyamine
polystyrene polyethylene 560 8 5.1 m2/g 4.15 meq/g 4.32 meq/g

polyamine
polyvinyl polyethylene 1400 61 15.1 m’/g 4.12 meq/g 3.72 rueqlg

chloride polyamine
Table 3.11: Relationship Between Mercury Intrusion Pressure and

Pore Diameter for Polystyrene Beads
Hg PorosimeterPressure (Bar) Pore Diameter (i)

28 5000
70 2000
125 1000
275 500
690 200
1240 100
2760 50
6900 20
12400 10
Normally, as the mean pore diameter increases, the

surface area of the resin decreases. These surface areas can
be as low as 2 m2/g to as high as 300 m2/g. Table 3.10 also
points out that the total exchange capacity is not utilized in
these biochemical fluid processes. Whereas in water
treatment applications, one can expect to utilize 95% of the
total exchange capacity, in biotechnology applications it is
often possible to use only 20 to 30% of the total exchange
capacity of gel resins. Macroporous resins have increased
the utilization to close to 90% for the immobilization of
enzymes, but biochemical fluid processing applications where
Ion Exchange 209
the fluid flows through an ion exchange resin bed still are
limited to about 60% utilization even with macroporous resins.
Table 3.12 shows the molecular size of some biological

macromolecules for comparison to the mean pore size of the
resins. When selecting the pore size of a resin for the
recovery or immobilization of a specific protein, a general
rule is that the optimum resin pore diameter should be about
4 to 5 times the length of the major axis of the protein.
Increasing the pore size of the resin beyond that point will
result in decreases in the amount of protein adsorbed because
the surface area available for adsorption is being decreased
as the pore size is increased. An example of this optimum
adsorption of glucose oxidase, as defined by enzyme activity,
is shown in Figure 3.29 (67). Enzyme activity is a measure
of the amount of enzyme adsorbed and accessible to
substrate.
3.3.4 Particle Density

The typical resin densities may range from 0.6 g/cc to
1.3 g/cc for organic polymers. Silicate materials may be
more dense, up to 6 g/cc. Since the fermentation broth or
other biochemical fluid may be denser than water, the slow
flow rates that are usually involved may require resins that
have a greater density than water. A minimum flow rate
may be necessary to maintain a packed bed when a fluid
denser than water is being processed by a medium density
resin. If this is not possible, an upflow operation or batch
process may be necessary. This will be discussed in more
detail in section 3.6.3.
Table 3.12: Molecular Size of Biopolymers

Maximum Length
Biopolymer Molecular Weight of Biopolymer
Catalase 250,000 183 fi
Glucose Isomerase 250,000 - 100,000 75 - 100 61
Glucose Oxidase 15,000 84 2
Lysozyme 14,000 40 II
Papain 21,000 42 I(
80
60
ENZYME
ACTIVITY
(GOUhm)
200 400 600 800
PORE DIAMETER (I()
Figure 3.29. Effect of a resin pore diameter on the enzyme

activity of glucose oxidase (Reference 67).
The lower-density resins are usually associated with a

highly porous structure which has less mechanical strength
than the typical gel or macroporous resins. When the mean
pore diameter of a resin is greater than 2000 A, the resin
would be subject to attrition in a stirred tank or to collapse
in a tall column.
3.3.5 Particle Size

Many of the resins used in early biochemical separations
were much smaller (75-300 microns). With the development
of macroporous resins, enzyme immobilizations and protein
purifications were performed with resins of the 400-1000
micron size since the macroporous structure allowed
sufficient surface area for adsorption almost independent of
particle size.
Ion Exchange 211
3.4 LABORATORY EVALUATION RESIN
The total exchange capacity, the porosity, the operating

capacity and the efficiency of regeneration need to be
evaluated in the laboratory when comparing resins for a
given application.
The total exchange capacity is usually determined by

titrating the resin with a solution of acid or base to a
specific endpoint. This type of information is readily
available from the manufacturers of commercial ion exchange
resins.
The pore size of a microporous resin can be determined

using water soluble standards, such as those listed in Table
3.13 (68). A typical pore size distribution is shown in Figure
3.30 for a strong cation-resin.
Table 3.13: Water Soluble Standard Samples for

Pore Measurements (68)
Sample Mean Pore Diameter (8)
J’S’ 3.5
Ribose a
Xylose 9
Lactose 10.5
Raffinose 15
Stachyose 19
T-4 @ 51
T-10 140
T-40 270
T-70 415
T-500 830
T-2000 1500
8 The T-Standards are Dextrans from Pharmacia

4 6 8 10 20 ‘40
PORE DIAMETER th
Figure 3.30. Pore size distribution of the cation exchange

resin Hamilton HC-X7 (Na).
If the resin is made with an inert, extractable diluent

to generate the macroporous structure, it is easier to
determine the mean pore size and pore size distribution.
Care must be taken so that the pores are not collapsed
during the removal of the water from the resin. Martinola
and Meyer (69) have devised a method of preparing a
macroporous resin for BET surface analysis or pore size
analysis by mercury porosimetry.
(1) Convert the ion exchange resin to be

desired ionic form.
(2) Add 500 ml of water-moist resin to a
round bottom flask with an aspirator.
Add one liter of anhydrous isopropyl
Ion Exchange 213
alcohol and boil under reflux at

atmospheric pressure for one hour.
Then suck the liquid off. Repeat the
isopropyl addition, boiling and aspirating
four times. After this procedure the
resin will contain less than 0.1% water.
(3) After drying to constant weight at 10-s
torr and 5O”C, the resin sample is ready
for pore size analysis.
The design of an ion exchange unit requires knowledge

of the capacity of the resin bed and the efficiency of the
exchange process. The “theoretical” capacity of a resin is
the number of ionic groups (equivalent number of
exchangeable ions) contained per unit weight or unit volume
of resin. This capacity may be expressed as milliequivalents
(meq) per milliliter (ml) or per gram.
The amount of dissolved impurities in a feedstream is

usually expressed in terms of equivalent calcium carbonate,
abbreviated to “as CaCOs”. Calcium carbonate provides a
good reference because it has a molecular weight of 100,
which facilitates calculations. An alternative method of
expressing this is in parts per million of the ions themselves.
The interrelationships are given in Table 3.14. If the
impurities are expressed in terms of parts per million as
equivalents of the ion (meq/L), the amounts in terms of ppm
must be divided by the equivalent weight of the ions. The
equivalent weight of an ion is its molecular weight divided by
its valence.
When deciding which resin to use for a given operation,

batch testing in a small beaker or flask will allow resin
selection and an approximation of its loading capability. A
useful procedure is to measure out 1, 3, 10 and 30 milliliter
volumes of resin and add them to a specific volume of the
feedstream. These volumes were chosen to have even spacing
on a subsequent log-log plot of the data.
After the resin bed feed solution has been mixed for at
least one-half hour, the resin is separated from the liquid
phase. The solute concentration remaining in the solution is
then determined. The residual concentration is subtracted
from the original concentration and the difference is divided
by the volume of the resin. These numbers and the residual
Table 3.14: Conversion Factors to Express as Parts per Million

as CaC03
IOIlS Ionic Weight Equivalent Weight Factor
Cations
Aluminum 27.0 9.0 5.56
Ammonium 18.0 18.0 2.78
Barium 137.4 68.7 0.78
Calcium 40.1 20.0 2.49
Copper 63.6 31.8 1.57
Hydrogen 1.0 1.0 50.0
Ferrous 55.8 27.9 1.80
Ferric 55.8 18.6 2.69
Magnesium 24.3 12.2 4.10
Manganese 54.9 27.5 1.82
Potassium 39.1 39.1 1.28
Sodium 23.0 23.0 2.18
Anions
Bicarbonate 61 .O 61.0 0.82
Bisulfate 97.1 97.1 0.51
Bisulfite 81.1 81.1 0.61
Carbonate 60.0 30.0 1.67
Chloride 35.5 35.5 1.41
Fluoride 19.0 19.0 2.63
Hydroxide 17.0 17.0 2.94
Nitrate 62.0 62.0 0.81
Phosphate (primary) 97.0 97.0 0.51
Phospbate (secondary) 96.0 48.0 1.04
Phosphate (tertiary) 95.0 31.7 1.58
Sulfate 96.1 48.0 1.04
Sulfide 32.1 16.0 3.12
Sulfite 80.1 40.0 1.25
concentration are plotted on log-log paper and frequently

give a straight line.
A vertical line drawn at the feed concentration

intersects at a point extrapolated from the data points to
give an estimate of the loading of the solute on the resin.
Example 3.5
Figure 3.31 shows such a plot for glutamic acid

Ion Exchange 215
adsorbed on an 8% crosslinked strong acid cation microporous

resin in a fermentation broth with 11 mg/ml of amino acid.
The resin was placed in a beaker with 250 ml of broth. The
extrapolation of the line for the 1, 3, 10 and 30 ml resin
adsorption data indicates that the loading of glutamic acid on
this resin is expected to be 60 g/liter resin.
1,O
0.1
0,001 A
O*Ol O*l 1,O
RESIDUAL SOLUTE CONCENTRATION
Figure 3.31. Plot of glutamic acid adsorbed on an 8%

crosslinked strong cation resin.
After several resins have been tested in this manner,

the resin is selected for column evaluation which has a high
loading per ml of resin or a low residual with larger resin
quantities.
In practice, the ion exchange resin is generally operated

at a level considerably below its theoretical capacity. Since
the ion exchange reactions are equilibrium reactions, an

impractically large quantity of regenerant would be required
to drive the reaction to completion. The “operating” capacity
of a resin is the number of ionic groups actually utilized per
unit weight or volume of resin under a given set of operating
conditions.
The operating capacity of a resin is not directly

proportional to the amount of regenerant used. “Efficiency”
is the concept used to designate the degree of utilization of
the regenerant. Column efficiency is the ratio of the
operating exchange capacity of a unit to the exchange that
theoretically could be derived from a specific weight of
applied regenerant.
It is recommended that operating capacity and column

efficiency be run initially on a small, laboratory scale to
determine if the reaction desired can be made to proceed in
the desired direction and manner. The column should be at
least 2.5 cm in diameter to minimize wall effects. The
preferential flow in a resin column is along the wall of the
column. The percentage of the total flow along the wall of
the column decreases as the column diameter increases and as
the resin particle size decreases.
The bed depth should be at least 0.5 m and the flow

rate should be about 0.5 bed volumes per hour for the initial
trial. These conditions are good starting points since it is
desirable that the transition zone not exceed the length of
the column. Using much larger columns would require
quantities of the feedstream which are larger than may be
readily available. A suggested operating procedure is outlined
below.
(1) Soak the resin before adding it to the
column to allow it to reach its hydrated
volume.
(2) After the resin has been added to the
column, backwash the resin with distilled
water and allow the resin to settle.
(3) Rinse the column of resin with distilled
water for ten minutes at a flow rate of
50 mL/min.
Ion Exchange 217
(4) Start the treatment cycle. Monitor the

effluent to develop a breakthrough
curve, such as shown in Figure 3.32,
until the ion concentration in the
effluent reaches the concentration in the
feed solution.
(5) Backwash the resin with distilled water
to 50- 100% bed expansion for 5 to 10
minutes.
(6) Regenerate the resin at a flow rate that
allows at least forty-five minutes of
contact time. Measure the ion
concentration of the spent regenerant to
determine the elution curve (Figure 3.33)
and the amount of regenerant actually
used.
(7) Rinse column with distilled water until
the effluent has reached pH 7.
Feed Concentration
Bed Volumes
Figure 3.32. Concentration of adsorbed species in column

effluent during column loading.
Bed Volumes
Figure 3.33. Elution curve showing concentration of adsorbed

species eluted during resin regeneration.
Feed concentrations, flow rates, and regenerant dosages

may be varied to develop the relationship between resin
utilization and regenerant efficiency so that the optimum
operating conditions can be selected for the system.
The first portion of the breakthrough curve in Figure

3.32 shows the quality of product that can be obtained under
the processing conditions. An integration of the area up to
the breakthrough point provides an estimate for commercial
column capacities for the space velocities used in the
experiment. The velocity at which the mass transfer zone is
moving through the column is given by dividing the length of
the column by the time it takes to detect the solute in the
column effluent. The difference between that time and the
time at which the selected breakthrough concentration
appears in the effluent, when multiplied by the velocity of
the mass transfer zone, results in an approximation of the
mass transfer zone.
During the column test, the starting volume and the

final volume of the resin should be measured. If there is a
Ion Exchange 219
change of more than 5%, progressive volume changes as the

resin is operated through several cycles should be recorded.
These changes may be significant enough to affect the
placement of laterals or distributors in the design of
commercial equipment. For instance, carboxylic resins may
expand by 90% when going from the hydrogen form to the
sodium form. This type of volume change may dictate how
the resin must be regenerated to prevent the breakage of
glass columns due to the pressure from the swelling resin.
Gassing, the formation of air pockets, within the resin

bed is to be avoided. Gassing may occur because of heat
released during the exchange reaction. It will also occur if a
cold solution is placed in a warm bed or if the liquid level
falls below the resin level. Keeping the feed solution 5°C
warmer than the column temperature should prevent the
gassing due to thermal differences.
It is necessary to configure the experimental apparatus

to insure that the feedsteam moves through the column at a
steady rate to maintain a well-defined mass transfer zone.
Possible methods of maintaining constant flow are shown in
Figure 3.34.
LARGE VOLUME CONSTANT HEAD DEVICE PUMP OPERATION

Liquid
Level
Figure 3.34. Equipment for laboratory evaluation of ion

exchange resins.
Once it is determined that the reaction will proceed as

desired, subsequent optimization of the system in the
laboratory calls for setting up a packed resin column of
approximately the bed depth to be used in the final
equipment, typically one to three meters.
3.5 ION EXCHANGE PROCESSES
3.5.1 Process Categories

Processes involving ion exchange resins usually make
use of ion interchange with the resin. Examples of these
processes are demineralization, conversion, purification and
concentration. Chromatographic processes with ion exchange
resins, which merely make use of the ionic environment that
the resins provide in separating solutes, will be discussed in
Chapter 4.
3.5.1.1 Demineralization: Demineralization is the

process in which the salts in the feed stream are removed by
passing the stream through a cation exchange column in the
hydrogen ion form, followed by an anion exchange column in
the hydroxide or “free base” form. Water is the most
common feed stream in demineralization. It may also be
necessary to remove the salts from a feed stream before
fermentation.
High metallic ion concentrations and high total salt

content in the carbohydrate feed have been found to
decrease the yield in citric acid fermentation (70). These
ions can be removed by passing the carbohydrate solution
through cation and anion exchange resin beds. The salts
required for optimum microorganism activity can be added in
the desired concentration prior to fermentation.
DEXTROSE + NaCl + R-H + R-Na + HCl + DEXTROSE (3.30)
DEXTROSE + HCl + R-OH + R-Cl + Hz0 + DEXTROSE (3.31)
3.5.1.2 Conversion: Conversion or metathesis is the

process in which salts of acids are converted to the
corresponding free acids by reaction with the hydrogen form
of a strong acid cation resin. One such example would be
the conversion of calcium citrate to citric acid.
Ca Citrate + 2 R-11 -i;' R2-Ca + 2 Citric Acid (3.32)
Ion Exchange 221
The term may also be used to describe a process in

which the acid salt is converted to a different salt of that
acid by interaction with an ion exchange resin regenerated to
the desired ionic form.
3.5.1.3 Purification: Many fermentation products may

be purified by adsorbing them on ion exchange resins to
separate them from the rest of the fermentation broth. Once
the resin is loaded, the product is eluted from the column
for further purification or crystallization.
Adsorbing lysine on an ion exchange resin is probably

the most widely used industrial method of purifying lysine.
The fermented broth is adjusted to pH 2 with hydrochloric
acid and then passed through a column of strong acid cation
resin in the NH: form. Dilute aqueous ammonia may be used
to elute the lysine from the resin (71).
Gordienko (72) has reported that treating the resin with

a citrate buffer solution of pH 3.2 and rinsing with distilled
water before elution results in an 83-90% yield of lysine,
with a purity of 93-96%.
3.5.1.4 Concentration: Ion exchange can be used to

concentrate valuable or toxic products of fermentation
reactions in a manner similar to purification. The difference
between the two processes is in the lower concentration of
the desired product in the feed solution of concentration
processes.
Shirato (73) reported the concentration process for the

antibiotic tubercidin produced from fermented rice grain
using the microorganism, Streptomyces tubercidicus.
Macroporous strong acid cation resin was used to concentrate
the antibiotic from 700 pg/ml in the fermentation broth to
13M pg/ml when eluted with 0.25N HCl. The yield of the
antibiotic was about 83%.
3.5.2 Purification Procedures

A general procedure has been outlined (74) for the use
of ion exchange resins to purify mixtures containing organic
acids or bases.
3.5.2.1 Purification of Strong Bases:

(A) No Stability Limitations - Selectivity and
222 Separation and Purif’ication Techniques in Biotechnology
removal of weaker bases can be obtained

by carrying out the adsorption at high
pH (either strong or weak acid resin can
be used) or by carrying out the
adsorption at low pH and washing the
resin adsorbate with high pH buffers.
Weaker bases can be preferentially
removed from the resin. Either
procedure yields a resin adsorbate
containing only strong bases. These can
be removed from the resin by eluting
with an appropriate salt for a strong
acid resin or with a strong acid for a
weak acid resin.
@I Unstable at Alkaline pH; Stable at
Neutral and Acidic pH - Under these
conditions separation from bases of
intermediate pK may be impossible. The
adsorption should be carried out using a
weak acid resin (carboxylic) at the
highest pH compatible with the stability
limitation. The elution is by
acidification with a strong acid.
(C) Unstable at Acidic pH; Stable at Neutral
and Alkaline pH - Adsorption conditions
would be the same as (A) with no
limitations on resin type. Elution should
be carried out with a salt using a cation
preferred by the resin.
(D) Stable Only at Neutral pH - Adsorption
should be like (B) while elution must be
performed with a salt.
(El Stable Only at Acidic pH - Separation
from weak bases is not possible.
Adsorption on a strong acid resin
followed by salt elution yields a mixture
of all types of bases.
(F) Stable Only at Alkaline pH - Adsorb at
high pH and elute with salt at high pH.
3.5.2.2 Purification of Strong Acids:

(A) No Stability Limitations - Adsorb on
either strong or weak base resin at low
Ion Exchange 223
PH. Elute from a strong resin with a

salt or from a weak base resin with
alkali.
Unstable at Alkaline pH; Stable at
Neutral and Acid pH - Carry out
adsorption on a strong base resin at the
minimum pH compatible with stability.
Elute with an appropriate salt. If the
pH range is satisfactory, use a weak
base resin and elute with alkali.
Unstable at Acidic pH; Stable at Neutral
and Alkaline pH - Adsorb at low PH.
Elute with an appropriate salt.
Stable Only at Neutral pH - A
purification based on pK is not possible
but a mixture of acids can be obtained
by adsorption on a strong base resin and
elution with a salt.
Stable Only at Acidic pH - If there is
no lower pH limit on stability, the
adsorption can be carried out at low pH
on either strong or weak base resins.
The elution is carried out with strong
acid or salt.
Stable Only at Alkaline pH - A
separation based on pK is not possible.
A mixture of all acid solutes can be
isolated by adsorption on a strong base
resin followed by elution with a salt.
Elution from an intermediate strength
resin can be accomplished with alkali.
3.5.2.3 Purification of Weak Bases:

(A) No Stability Limitation - Adsorb on a
strong acid resin; elute with a weak
base. The strong base compounds will
remain on the resin.
(B) Unstable at Alkaline pH; Stable at
Neutral and Acidic pH - Unless the
compound is stable at about 2 pH units
below its pK,, ion exchange separation
is unlikely to succeed. If the adsorption
step at acidic pH can be carried out, the

weak base can be isolated by elution
with a weak base.
(0 Unstable at Acidic pH; Stable at Neutral
and Alkaline pH - Separation from
strong bases may not be possible. May
be adsorbed at acidic pH and eluted with
a salt.
(D) Stable Only at Neutral pH - Separation
from neutral compounds may not be
possible. Strong bases are removed by
adsorption on strong acid resin. Acidic
compounds are removed by adsorption on
strong base resins. Elution is with salt.
03 Stable Only at Acidic pH - Separations
from strong bases is not possible. Bases
are separated by adsorption on a strong
acid resin and elution with a salt.
03 Stable Only at Alkaline pH - Acids and
strong bases are removed by adsorption
with appropriate resins. Neutral
compounds and weak bases remain.
3.5.2.4 Purification of Weak Acids:

No Stability Limitations - Adsorb on a
strong base resin at alkaline pH; elute
with a weak acid. Strong acids remain
on the resin.
Unstable at Alkaline pH; Stable at
Neutral and Acidic pH. Separation from
strong acids may not be possible.
Adsorb acids at alkaline pH and elute
with a salt.
Unstable at Acid pH; Stable at Neutral
and Alkaline pH - Separation from
neutral compounds may not be possible.
Strong acids and all bases are removed
by appropriate ion exchange treatment.
Stable Only at Neutral pH - Strong acids
and bases removed by appropriate ion
exchange treatment. Weak bases and
neutral compounds may not be separated.
Ion Exchange 225
W Stable Only at Acidic pH - Weak acids

and neutral compounds remain after ion
exchange removal of strong acids and all
bases.
03 Stable Only at Alkaline pH - Separation
from strong acids may not be possible.
Acids are adsorbed onto a strong base
resin and eluted with a salt.
Example 3.6
The fermentation of valine using Arthrobacter
paraffineus will also result in the production of small
amounts of glutamic acid. The glutamic acid can be removed
from the fermentation broth by passing the fluid through a
strong base anion resin at a pH of 7.0. Then the eluant may
be adjusted to a pH of 2 and passed through a strong acid
cation resin to concentrate and purify the valine. Elution of
the valine from the resin occurs with 1N ammonium
hydroxide.
3.5.3 Biopolymer-Resin Interactions

As the biopolymer becomes more complex, the
mechanisms by which it adsorbs to and desorbs from resins
may involve mechanisms in addition to simple ionic exchange.
The attachment of biochemicals to the surface of the ion
exchange resin may involve a balance between London-van
der Waals forces in addition to the electrostatic forces of
electrical double layers (75).
Table 3.15 shows the types of biochemical interactions

that are possible with ion exchange resins and insoluble
polymer matrices. The degree of difficulty in forming the
specific interaction - without denaturing a protein or
biopolymer - is related to the strength of the bond which is
formed. The stronger the interaction, the more difficult it is
to be formed without adversely affecting the biopolymer.
With ionic bonds and physical adsorption, simply bringing the
resin and the biopolymer into close proximity will cause the
interaction. Likewise, changing the pH or ionic strength of
the solution will cause the interaction’s reversal or
regeneration of the resin.
Specifically for interactions with ion exchange resins,

the typical biopolymer can be thought of as a macroscopic
Table 3.15: Biopolymer Interactions with Ion Exchange Resins

and Polymer Matrices
Covalent Ionic Physical
Bond Bond Adsorption Entrapment
Formation Complicated Simple Simple Complicated
Possibility of High Low Low to High High

Biopolymer Denaturing
Bonding Strength strong Weak Weak strong
Regeneration No Possible Possible No
ion having a number of electrostatically charged surface

sites. The ionic interactions can be represented as:
R' R'
P-N+(cl13),cl- + -(lOC-~-NH2 + R-N+(Cl~&lOC-h-N"* + Cl- (3.33)
/1 t!l
for a positively charged (anion) resin and a negatively
charged biopolymer. Similarly, for a negatively charged
(cation) resin and a positively charged biopolymer, the
interaction is represented as:
R' R'
-+ +I
R-S03H + H$h-H # R-SO;H3N-7-H + H+ (3.34)
AOOH COOH
There are two other ionic interactions that are possible

with the assistance of multivalent ions or polyelectrolytes in
the feed stream. Thus, the presence of a multivalent cation
(P’2) in the solution would allow a cation resin to adsorb a
negatively charged biopolymer, represented by:
R’
Y’
K-SO;H+ + P +2 + -OOC-:-NH2 + R-SO;+P+-OOC-C-NH2 + Hf (3.35)
A A
Likewise, multivalent anions (Ne2) allow an anion resin to
adsorb a positively charged biopolymer according to:
R’ R’
+ -2
R-N (Cll,),Cl + N + “,N%” # R-N+(C”3);N-“3N%” + cl- (3.36)
LX, km
Ion Exchange 221
A single protein can, at the same time, participate in

specific ion exchange reactions with the resin, form bonds
with water-soluble polyelectrolytes and attach to other
adsorbent surfaces. As has been seen, the mathematical
representation for these complex interactions combines the
kinetics of diffusional processes, ion exchange reactions and
adsorption isotherms. Figure 3.35 (75) shows the adsorption
of bacterial cells on ion exchange resins and Figure 3.36
shows the desorption. It should be noted that there is
usually some irreversible adsorption of biopolymer on ion
exchange resins, as Figure 3.36 shows. The extent of this
irreversible bonding will define the limitations of using a
particular resin for purifying a biochemical solution. This
irreversible adsorption property can be utilized when an
enzyme is to be immobilized on an ion exchange resin.
1.00
0.70
0.40
0.20
ADSORBANCE
0.10
0.07
RATIO
0.04
0.01
0 12 3 4 5 b 7
TIME"(min)'&
Figure 3.35. Adsorption of bacterial cells onto an anion

exchange resin (Dowex 1 X 8, Chloride form), A = 0.598, pH
= 3.53 (Reference 75).
RATIO
Ao
0 1 2 3 4 5 6 7
TIME' (rn,",yp
Figure 3.36. Desorption of bacterial cells from the anion

resin in Figure 3.35, pH = 4.12, 1 M KC1 (Reference 75).
3.6 ION EXCHANGE OPERATIONS
The typical cycle of operations involving ion exchange

resins include pretreatment of the resin and possibly the feed
solution, loading the resin with the solutes to be adsorbed by
contacting the resin with the feed solution, and elution of
the desired material from the resin. The scale of operation
has ranged from analytical applications with a few milliliters
of resin and microgram quantities of material to commercial
production units containing several cubic meters of resin and
producing metric tons of material. The loading may be
applied batchwise, to a semi-continuous batch slurry, or to a
column filled with resin which may be operated in a semi-
continuous or continuous manner.
Ion Exchange 229
3.6.1 Pretreatment
The pretreatment of the feed solution described in
Section 2.5.1 for adsorption applications should also be
followed for ion exchange applications. It may be necessary
to pretreat the ion exchange resin.
Should the resin be used to purify food feedstreams,

such as sugar liquids or corn syrup, it is necessary to
pretreat the resin to assure that the extractable level of the
resin in use complies with Food Additive Regulation 21 CRF
173.25 of the Federal Food, Drug and Cosmetic Act. The
pretreatment recommended for a column of resin in the
backwashed, settled and drained condition is:
(1) Add three bed volumes of 4% NaOH at a
rate sufficient to allow 45 minutes of
contact time.
(2) Rinse with five bed volumes of potable
water at the same flow rate.
(3) Add three bed volumes of 10% HsSO, or
5% HCl at a flow rate that allows 45
minutes of contact time.
(4) Rinse with five bed volumes of potable
water.
(5) Convert the resin to the ionic form
desired for use by applying the
regenerant that will be used in
subsequent cycles.
If the column equipment has not been designed to

handle acid solutions, a 0.5% CaCl, solution or tap water may
be used in step 3 in place of H,SO, or HCl for cation resins.
Similarly, for anion resins, a 10% NaCl solution could be used
in such equipment in place of acids.

The batch contactor is essentially a single stage stirred
reactor with a strainer or filter for separating the resin from
the reaction mass once the reaction is complete. This type
of contactor has an advantage in some fermentation
operations because of its ability to handle slurries.
Additional advantages include the low capital cost and
simplicity of design.
In a batch operation, an ion exchange resin in the

desired ionic form is placed into a stirred reaction vessel
containing the solution to be treated. The mixture is stirred
until equilibrium with the exchangeable ion is reached (about
0.5 to 3 hours). Then the resin is separated from the liquid
phase by filtration. The adsorbate is removed from the resin
by rinsing with the eluting solution. An additional step may
be required to reconvert the resin to the regenerated form if
this is not done by the eluting solvent. The cycle may then
be repeated.
The batch system is basically inefficient since the

establishment of a single equilibrium will give incomplete
removal of the solute in the feed solution. When the affinity
of the resin for this solute is very high, it is possible that
the removal will be sufficiently complete in one stage. The
batch process has the advantage over fixed bed processes
that feedstreams containing suspended solids may be treated.
In these cases, the resin particles, loaded with the adsorbate
and rinsed from the suspended solids, may be placed in a
column for recovery of the adsorbate and for regeneration.
The batch contactor is limited to use with reactions

that go to completion in a single equilibrium stage or in
relatively few stages. Difficulties may also arise with batch
contactors if resin regeneration requires a greater number of
equilibrium stages than the service portion of the cycle.
3.6.3 Column Operations

3.6.3.1 Fixed Bed Column contactors allow
Columns:
multiple equilibrium stages to be obtained in a single unit.
This contactor provides for reactions to be driven to their
desired level of completion in a single pass by adjusting the
resin bed depth and the flow conditions. The main
components of a column contactor are shown in Figure 3.37.
At the end of its useful work cycle, the resin is backwashed,
regenerated and rinsed for subsequent repetition of the work
cycle. Typically, this non-productive portion of the cycle is
a small fraction of the total operating cycle.
Column contactors may be operated in co-current,

countercurrent, or fluidized bed mode of operation. The co-
current mode means that the regenerant solution flows
through the column in the same direction as the feed
solution (Figure 3.38). The countercurrent mode has the
Ion Exchange 231
regenerant flowing in the opposite direction as the feed

solution (Figure 3.39).
Inlet Distributor and

Backwash Outlet
n
Regenerent
=) Distributor
Effluent
= Distributor
Figure 3.37. Ion exchange column contactor.
R Feed
+
Kin
k9-
Resin
H+ H+
Resin
Treated
Water
Original Bed Exhausted Bed Backwash
Regenerant
& Rinse
Feed
H+ Nat
Resin
Nat i-t+
Resin
Na
@ Waste 6 Product
Regeneration Exhaustion
And Rinse
Figure 3.38. Co-current column operation with H+ form

cation resin removing Na+ ions from feedstream.
Feed
Na+
Resin Resin Resin
H+ H+
Resin
Treated Service
Water Water
Original Bed Exhausted Bed Backwash
Regeneration Exhaustion
and Rinse
Figure 3.39. Countercurrent column operation with H+ form

cation resin removing Na+ ions from feedstream.
Countercurrent operation of a column may be preferred

to reduce the ion leakage from a column. Ion leakage is
defined as the amount of the ion being removed from
solution which appears in the column effluent during the
course of the subsequent exhaustion run. Comparison of
Figure 3.38 and Figure 3.39 shows that the leakage caused by
reexchange of nonregenerated ions during the work cycle of
cocurrently regenerated resin is substantially reduced with
countercurrent regeneration.
The fixed bed column is essentially a simple pressure

vessel. Each vessel requires a complexity of ancillary
equipment. Each column in a cascade will require several
automatic control valves and associated equipment involving
process computer controls to sequence the proper flow of
different influent streams to the resin bed.
Combinations of column reactors may sometimes be

necessary to carry out subsequent exchange processes, such
as in the case of demineralization (Figure 3.40). For
demineralization a column of cation exchange resin in the
hydrogen form is followed by a column of anion exchange
resin in the hydroxide form. A mixed bed, such as shown in
Figure 3.41, may likewise be used for demineralization.
Ion Exchange 233
Na+CI’ --z--e
Solution @ +a Q+CI’
Anion Exchanger
@+CI Packed Cal.
Cation Exchanger -,..----4 H+OH-
Q +OH-
Packed Col.
I @‘H+
Figure 3.40. Demineralization ion exchange column scheme.
Anion Anion
Treated
Cation Cation
Service Backwash Simultaneous Mixing

Cycle Regeneration
Figure 3.41. Operation of a mixed bed demineralization ion

exchange column.
Mixed bed operation has the advantage of producing a

significantly higher quality effluent than the concurrently
regenerated beds of Figure 3.40, but has the added difficulty
of requiring separation of the resin prior to regeneration.
The important requirements for the successful operation

of a mixed bed include the careful separation of the strong
base anion exchange resin from the strong acid cation
exchange resin by backwash fluidization. This is followed by
contact of each of the components with their respective
regenerant in a manner to minimize the cross contamination
of the resins with the alternate regenerant. This requires
that the quantity of resin, particularly the cation exchange
resin, be precisely maintained so that the anion-cation

interface will always be at the effluent distributor level.
Typically, matched pairs of resins are used so that an ideal
separation can be repeatedly obtained during this process.
Recently, inert resins have been marketed which enhance the
distance between the anion-cation interface and allow less
cross contamination during regeneration (76).
The air mixing of the anion and cation must also be

performed in such a manner that complete mixing of the
resin and minimum air entrapment are obtained at the end of
the mixing cycle.
Ion exchange is usually in a fixed bed process.

However, a fixed bed process has the disadvantage that it is
cyclic in operation, that at any one instant only a relatively
small part of the resin in the bed is doing useful work and
that it cannot process fluids with suspended solids.
Continuous ion exchange processes and fluidized bed systems
have been designed to overcome these shortcomings.
3.6.3.2 Continuous Column Operations: When the ionic

load of the feed solution is such that the
regeneration/elution portion of the operating cycle is nearly
as great, or greater than, the working portion, continuous
contactors are used instead of column contactors.
Continuous contactors operate as intermittently moving

packed beds as illustrated by the Higgins contactor (77)
(Figure 3.42), as fluidized staged (compartmented) columns as
demonstrated by the Himsley contactor (78-80) (Figure 3.43).
In the Higgins contactor, the resin is moved

hydraulically up through the contacting zone. The movement
of resin is intermittent and opposite the direction of solution
flow except for the brief period of resin advancement when
both flows are cocurrent. This type of operation results in a
close approach to steady-state operations within the
contactor.
Elegant slide valves are used to separate the adsorption,

regeneration and resin backwash stages. The contactor
operates in predetermined cycles and is an ideal process for
feedstreams with no suspended solids.
Ion Exchange 235
Overflow -
4
1
Feed
Backwash
Product
Rinse Water
Regenerant
d
d Regenerating
Section
Figure 3.42. The Higgins contactor for continuous operation

(Reference 77).
The Higgins type of contactor is able to handle a

certain amount of slurry due to the continued introduction of
fresh resin material to act as a filter media during the
operation. A lower resin inventory should result with
continuous contactors than with column reactors handling the
same high ionic load feed stream.
The major disadvantage of the Higgins contactor is the

lifetime of the resin. Estimates range as high as 30% resin
inventory replacement per year due to attrition and breakage
of the resin as it passes through the valves (81).
In the Himsley type of contactor, the resin is moved

from one compartment or stage to the next, countercurrent
to the feed solution flow, on a timed basis that allows for
-----e-F
\
\
\
\
\
\
\
\
\
\ RINSE WATER INLET
I t
I I
I
I
I
I
I
I I
I I
I I
I I
I I ELUANT INLET
I I
I I I
I I I
4
I
I
L-- --_---__-- 2
RESIN TRANSFER LINES

SHOWN DOTTED v
FEED LIQUOR ELUATE
Figure 3.43. The Himsley Continuous Fluid-Solid Contactor

(Reference 78).
the rate of equilibrium resin loading in each stage. Thus

each compartment or tray is designed to accommodate
specific feed compositions and effluent requirements.
Equipment from different commercial suppliers differs in the
manner of resin transfer.
3.6.3.3 Fluidized Column Operations: Column

contactors, when operated with the feedstream in a downflow
mode, are poorly suited to handle fermentation slurries
because of the excellent filtration characteristics of packed
resin beds. For such slurries it is preferable to use a
fluidized bed of resin such as is shown in Figure 3.44 (82).
Ion Exchange 237
SUPPLY
PROTECTIVE LAYER
EFFLUENT OUTLET
WORKING LAYER
LOADED LAYER
REMOVAL DEVICE
FEEDSTREAM INLET
Figure 3.44. Fluid bed ion exchange column (Reference 82).
While most of the commercial applications discussed in the

literature pertain to slurries of uranium tailings and paper
mill effluent treatment, the equipment may be adapted for
use in treating fermentation broths.
The shape of the fluidized bed is important in

controlling the position of the resin in the column. The
effluent from the column should pass over a vibrating screen,
e.g., as SWECO, to retain entrained resin but allow mycelia
to pass through.
The Asahi contractor, shown in Figure 3.45 (81), uses

conventional pressure vessels as the resin column. These
vessels have a resin support grid at the base and a resin
screen at the top. The feedstream is fed in upflow through
the packed bed. Periodically the liquid contents of the
%
P
H
Ion Exchange 239
column are allowed to drain rapidly which causes the resin to

flow from the bottom of the bed to a similar vessel for
regeneration. At the same time, fresh resin is added to the
top of the active column from a resin feed hopper. This
hopper contains a ball valve which passes the resin in during
downflow operation and seals itself during the upflow portion
of the adsorption cycle.
The Cloete-Streat ion exchange equipment is a

multistage fluidized bed containing perforated distributor
plates (Figure 3.46) (83). The hole size in the plates is
greater than the maximum resin particle size. The
countercurrent movement of resin occurs due to the
controlled cycling of the feedstream. Each cycle the entire
amount of resin in one chamber is transferred to the next
chamber. Equipment with 4.5 m diameter columns and eight
stages for a total height of about 20 m are in commercial
operation.
Lernliquor Awcrrc flow
4 OUI
M::.“,,.lra,,o”o”
L-l -em
+
bbbb
: .~ . . . . . . . :.:.:.*
.:.:.:.:.:.: . .
w-w * ---
bbbb
_., : f - -. . . ..*...,...
:.:
we_
I b-b-b
A.‘... .. .. . .*
~
_W
~b”rl I
3 Reverr* flow 4 Flow
Figure 3.46. Sequence of operation of the Cloete-Streat

countercurrent ion exchange process (Reference 83).
The USBM equipment, shown in Figure 3.47 (84), is very

similar to the Cloete-Streat system. The differences are in
the plate design, the method of transferring solids and in the
method of removing the resin. This system and the Cloete-
Streat system are able to handle slurry feeds with up to 15%
by weight suspended solids.
- 30’ 0
I- 1
f !‘: 0 bolos on
4- cenloc‘8
Dirtrnbulion plota
Figure 3.47. Schematic diagram of the USBM ion exchange

column (Reference 84).
The advantage of these fluidized bed columns is the

relatively low capital cost, low operating cost, small space
requirement, simple instrumentation and control compatability
with conventional solvent extraction equipment (81). Only
those systems that can accommodate slurries with suspended
Ion Exchange 241
solids are commercially feasible for biotechnology and

fermentation operations. Otherwise the small volume of
fermentation feedstreams which need to be processed is not
on the scale of operations necessary to make continuous ion
exchange processes cost effective.
The principal disadvantage of fluidized bed columns is

the mixing of resin in various stages of utilization. This
mixing means that breakthrough occurs sooner and the degree
of resin capacity utilization is much lower than in packed
bed columns. By placing perforated plates in a column
(Figure 3.48) (89, the resin beads only mix within a
restricted area allowing more complete utilization of the ion
exchange resin’s capacity.
o ION EXCHANGE
BEAD
’ FINE SOLID
PARTICLE IN
FEEDSTREAM
Figure 3.48. The perforated plates in this fluidized bed

column allow fine solid particles to flow through the column
while the ion exchange beads are, for the most part,
confined to individual compartments (Reference 85).
Smaller continuous fluid bed systems, like the one

shown in Figure 3.49 (86), have been developed which operate
with a high concentration of ion exchange resin and
suspended solids. These units are 80% smaller than the
conventional resin-in-pulp plants of the type shown in Figure
3.50 which are used in the treatment of uranium ore slurries
(87). The pilot plant unit, which would probably be the size
needed for processing commercial fermentation broths, had
dimensions per contact chamber of 0.82 m x 0.82 m with a
0.82 m fluid bed height and an additional 0.16 m for
freeboard. The unit has been successfully operated with 25
to 50% resin and up to 45% suspended solids.
Figure 3.49. Pilot plant for resin-in-pulp contactor unit

(Reference 86).
!&Al% SUJRRY SLURRY FLOW
*
ADSORBENT FLOW
I I I I I I I I I II J
AIR TO SPARGE
PIPES AN0 AIR LIFTS
Figure 3.50. Plant design for resin-in-pulp contactor units

(Reference 87).
Ion Exchange 243
The effect of the degree of regeneration of the resin

on the degree of extraction of a solute was measured by
Slater (88) using a seven stage unit. The results are shown
in Figure 3.51 for the extraction of uranium using a fluidized
bed slurry of a strong base resin in a 10% uranium ore leach
QP(I1 W(N) 0 JE
l 0.09s 0.00s 21.6 0.5
x 0.095 000 20.4 0.5
A 0.095 0.01s 19.2 0.5
0 0.09s 0.020 18.0 0.5
I 2 3 4 5 6 7
FEED STAGE
Figure 3.51. Effect of resin feed composition on the

efficiency of a staged system (Reference 88).
However, the fluidized bed column, even with perforated

plates separating it into as many as 25 compartments, may
not be appropriate for applications in which there are strict
requirements (~1% of effluent concentration) on the effluent.
For readily exchangeable ions, the optimum utilization of this
technique occurs when an actual effluent concentration of
10% of the influent is the breakthrough point. At such
times, the ion exchange resin capacity would be 70% utilized
(85). Should it be acceptable that the breakthrough point
occur when the average effluent reaches a concentration of
10% of the influent, the utilization of the ion exchange
capacity is 90%. If the ions are not readily exchangeable
(low selectivity), the resin utilization would be significantly
less and fluidized bed operations should not be used.
3.6.4 Elution/Regeneration
Elution of proteins from ion exchangers can be achieved
with buffers containing salts, such as sodium chloride or
ammonium acetate, or by an appropriate pH change, provided
that the pH change does not result in denaturation of the
eluted protein (89). The elution may be performed with a
series of stepped changes or with a continuous gradient
change in the eluting power of the eluant. With such
changes, it is possible to separate different proteins or
protein fractions from each other based on their different
affinities for the ion exchange resin.
Elution of compounds such as penicillin with either

acids or bases will render the penicillin inactive. Although
aqueous salt solutions can elute penicillin without inactivating
it, the large volumes required make this option impractical.
Wolf and coworkers (90) developed an elution solvent
combination of organic solvent, water and salt that can elute
the penicillin with a minimum volume and no inactivation.
The mixture of organic solvent and salt is chosen so that the
salt is soluble in the resulting organic solvent-water mixture
and the organic substance eluted from the resin is soluble in
the elution mixture. Table 3.16 shows the elution volume
required to recover the indicated amount of antibiotic when
the elution solvent is 70% methanol and 5% or 7.5%
ammonium chloride in water.
Regeneration alone is not sufficient to prevent fouling

or microbial growth on ion exchange resins. If the resin is
left standing in the regenerant during non-operating times, it
Ion Exchange 245
Table 3.16: Amount of Antibiotic Recovered with Increasing

Volumes of Methanol in Aqueous Ammonium Chloride (90)
Total Volume of Amount of Antibiotic
Antibiotic Eluant Eluant (Bead Volumes) Recovered (“6)
Dihydronovobiocin 70% MeOH 1 50.0

with 5% N&Cl 2 83.3
93.3
96.6
Novobiocin 70% MeOH 3.5 99.8

with 5% N&Cl
Penicillin 70% MeOH 0.5 11.0

with 7.5% NH,Cl 1.0 69.0
1.5 94.0
2.0 97.1
2.5 99.7
is possible to suppress the microbial growth (91). The

regenerant in this instance was 10 to 20% NaCI. When the
initial microbe count was 10 per milliliter, at the end of
three weeks in 20% NaCl, the count had risen to just 800/mL
compared to 200,00O/mL for the resin stored in water. When
an alternate regenerant is used (NaOH or HCI), it is
preferable to change the storage medium to 200/6 NaCl since
extended time in an acid or base media can adversely affect
the resin matrix.
Example 3.7
When a regenerant has a 1% contaminant in a 7% HCl
solution, what will be the effect on the sodium leakage in
the effluent when a feedstream containing 34 meq/L Cat+, 2
meg/L Mgtt and 5 meq/L Nat is treated with a strong cation
resin with an +exchange capacity of 1.8 meq/ml and a
selectivity of Kp+ = 1.5?
Selectivity analysis calculations can be used as follows.
The normality of NaCl in the regenerant:

10.0 e 1,035 g 1 eq
x x = 0.18 eq/liter (3.37)
1000 g liter 58.44
The normality of HCl in the regenerant:

70.0 R 1,035 g 1 eq
____- x x ~ = 1.99 eq/liter (3.38)
1000 g liter 36.46
n. 1R
.x .~___-- = 0.08 in the regencrant (3.39)
Na+ =
(0.19 + 1.99)
%a+ -1 - XNa+ = *Na+ 0.08
1 + TiNa+
=
xp+ 1.5
1 - 0.08
= 0.13 (3.40)
After regeneration, therefore, XNa = 0.115. During the

treatment of the feedstream:
*NCi+ 1 XNa+ 1 0.115

= = = 0.087 (3.41)
-
1 - XNa+ I;Na’- 1 - XNa+ 1.5 1 - 0.115
H’
Therefore, in the effluent from the column, X,, = 0.080.

This corresponds to 331 ppm Na as CaC03.
3.6.5 Backwashing
Since most fluids contain some suspended matter, it is
necessary to backwash the resin in the fixed bed column on
a regular basis to remove any accumulation of these
substances. To carry out a backwashing operation, a flow of
water is introduced at the base of the column. The flow is
increased to a specific rate to classify the resin hydraulically
and remove the collection of suspended matter. Figures 3.52
and 3.53 show the types of flow rates which provide certain
degrees of expansion of cation and anion resins, respectively.
Since the anion resins are significantly less dense than the
cation resin shown, it would be necessary to have different
amounts of freeboard above the normal resin bed height so
that backwashing may be accomplished with only a negligible
loss of ion exchange resin. Typically, an anion resin bed
may be expanded by 100% during backwashing, while a cation
resin bed will only be expanded by 50%.
It is also necessary that the water used for the

backwashing be degassed prior to use. Otherwise, resin
particles will attach themselves to gas bubbles and be carried
out of the top of the column to give an unacceptable
increase in resin losses.
When treating fermentation broth filtrates, frequent

backwashing of the resin bed is necessary to prevent
accumulation of suspended matter. In such cases, the column
height should be designed of such a size that the bed is
regenerated at least every ten hours. Shorter columns have
Ion Exchange 247
TO DElEiWlNE FLOURATE
AT TEhlPERATURE T:
0 16.32 32.64 48.96
BACKWASH FLObi RATE hL/filN hi’)
Figure 3.52. Backwash expansion characteristics of a

macroporous strong acid cation resin, Dowex 88.
TO DETtRMlltE FLO* RATE

AT TEJWERATWE 1:
0 4.08 8.16 12.24 16.32

BACKMAW FLOW R&U hL/MlN/CR’)
Figure 3.53. Backwash expansion characteristics of

regenerated and exhausted macroporous weak base anion
resin, Dowex 66.
been designed to be regenerated at least every hour (92).

These shorter beds can then use finer resins and achieve a
high level of efficiency with lower capital costs. This may
be taken to the point of using very fine resins, as with the
PowdexR system (93) which discards the powdered resin after
a single use.
After a bed is backwashed, unless it is air-mixed as the

level of water is drained down to the surface, the beads or
particles classify according to size. The fine beads end up
on top and the large beads on the bottom of the column. In
co-current operations, the regenerant first contacts the top
of the bed. The fast kinetics of the fine particles gives a
high regeneration efficiency. However, the large beads on
the bottom will regenerate more slowly and may end up only
partially regenerated. Thus, when the feedstream is next
passed through the resin bed, leakage of undesirqhls ions may
occur from the large beads in the bottom of the column.
This may be overcome by using a countercurrent flow
arrangement described earlier or by using an air-mixing
system during the post-backwash draining (94).
3.7 PROCESS CONSIDERATIONS
3.7.1 Design Factors

The engineer designing an ion exchange column
operation usually will prefer to work with the simplest
kinetic model and linear driving force approximations. The
weakness of this approach is that any driving force law only
regards the momentary exchange rate as a function of the
solute concentration in the bulk solution and the average
concentration in the particle, neglecting the effect of
concentration profiles in the particle. Nevertheless, the
linear driving force approach provides an approximation that
is often sufficiently accurate for the engineer.
3.7.1.1 Scaling-up Fixed Bed Operations: Rodrigues (95)

has presented empirical and semi-empirical approaches which
may be used to design ion exchange columns when the solute
in the feedstream is c, and the flowrate is u,. The
breakthrough point is usually set at the point when the
effluent concentration increases to 5% of c,. The design
equations relate the total equilibrium ion exchange capacity
Ion Exchange 249
(Q) to the volume of resin required (V,) to the time of the

breakthrough (tn).
The empirical approach, the overall mass balance is

given by the equation:
vr = co 5 tS/(l +S)Q (3.42)
where ts = r(l +5) (3.43)
5 = (1 - c)QhQ, (3.44)
T =
EV/Uo (3.45)
and V is the bed volume with void space e.
It is usually necessary to modify this resin amount by a

safety factor (1.2 to 1.5) to adjust for the portion of the
total equilibrium capacity that can actually be used at flow
rate u and to adjust for any dispersive effects that might
occur during operation.
The semi-empirical approach involves the use of the

mass transfer zones as was used in the chapter on
adsorption. This approach has been described in detail
specifically for ion exchange resins by Passino (96). He
referred to the method as the operating line and regenerating
line process design and used a graphical description to solve
the mass transfer problems.
For the removal of Ca++ from a feedstream, the mass

transfer can be modeled using Figure 3.54. The upper part
shows an element of ion exchange column containing a
volume v of resin to which is added a volume V,, of the
feedstream containing Ca++. It is added at a flow rate (FL)
for an exhaustion time t,,. The concentration of Ca++ as it
passes through the column element is reduced from xeX1 to
X ex2* Therefore, the resin, which has an equilibrium ion
exchange capacity ZT, increases its concentration of Ca++ from
Lx2 to Yexl- In this model, fresh resin elements are
continuously available at a flow rate (F,) = v/t,, which is
another way of saying the mass transfer zone passes down
through the column.
The lower part of Figure 3.54 shows the operating lines

A
Exhaustion
__
._--
-.
.;..._
I I
I I
I I
I I
I 1-l
;1
fl
I
I I
XolZ XUI -
CP in solution
Figure 3.54. Schematic representation of the softening of a hard water feedstream on

an ideally continuous bed of strong acid cation resin in the sodium form (Reference 96).
Ion Exchange 251
for this process. The ion exchange equilibrium line describes

the selectivity in terms of a Freundlich, Langmuir or other
appropriate model.
The equations for the points in the lower part are given
by:
X cc=++ in the feedstream) (3.46)
exl = xcl
c v
++
Ye,2 + (Xexl - Xex2) _-LL-s- (Ca in exhausted resin) (3.47)
ye,1 =
Fv
J
V
cx
x dv ++
x = 0 (average Ca in the effluent) (3.48)
ex2
vex
Y
cx2
= 0 (Ca++ in the regenerated resin) (3.49)
and the slope of the operating line:

c c
AY Y exl-Ye,z
- =
= 0 Vex =
0 (Ft)ex (3.50)
cv c F
C AxJex Xexl - Xex2 s
The value of c,, i7 and v are known so that for any

V ex value the slope of the operating line can be calculated
from Equation 3.50. The specific points: x, is given, x,~ is
obtained by graphic integration from the breakthrough
curves. After operation and regeneration, the value of yex2
may not be zero but may be between 0.02 and 0.05 if the
regeneration is not complete. The application of this
technique has been described in terms of basic design
parameters such as number of transfer units, the height of a
transfer unit and mass transfer coefficient (97,98).
The data generated with the laboratory column may be

scaled-up to commercial size equipment. Using the same flow
rate (on a mass basis) as used in the laboratory experiments,
the appropriate increase in column size over that used in the
laboratory is a direct ratio of the volumes to be treated
compared to that treated in the laboratory equipment.
If a reasonable height-to-diameter ratio (approximately

1:l) is obtained in the scale-up using the bed dept involved
in the laboratory procedure, then that bed depth is
maintained and the cross-sectional area of the column is
increased. However, if the sizing is such that the column is

much larger in diameter than the bed depth, scale-up should
be done to maintain a height-to-diameter ratio of
approximately 1. The required resin volume is determined by
maintaining the same mass flow conditions (liters of feed
solution per minute per cubic meter of installed resin) as was
used in the laboratory operation.
Appropriate tank space must be left to accommodate the

backwash operation. This is typically 50% of bed depth for
cation exchange resins and 100% expansion in the case of
anion exchange resins.
Example 3.8
The purification of lysine-HCl from a fermentation broth
will be used to illustrate the calculations involved in scaling-
up laboratory data.
The laboratory fermentation broth, which is similar to

the commercial broth, contained 20.0 g/L lysine, much smaller
amounts of Cat+, Kt and other amino acids. The broth was
passed through 500 ml of strong acid cation resin, DOWEX
HCR-S, in the NH: form. The flow rate was 9 mL/min or
1.77 mL/min per cm2 of resin. It was determined that the
resin capacity averaged 115 g of lysine-HCl per liter of resin.
It may be noted that since the equivalent molecular weight
of lysine-HCl is 109.6 g and the “theoretical” capacity of
DOWEX HCR-S is 2.0 meq/mL, the operating capacity is 52%
of theoretical capacity.
The commercial operation must be capable of producing

9M metric tons of lysine (as lysine-dihydrochloride-H20) per
year. With a 20.0 g/L concentration of lysine in the
fermentation broth, the number of liters of broth to be
treated each year are:
9000 mton* 146.19 (MW of lysine) liter lo6 g
x--x -=
L
year 237.12 (Ml4 of 1Y-HC1420) 20.0 g
mton (3.51)
27.7 x lo7 liter/year
If the plant operates 85% of the time, the flow rate would
have to be:
I 1 year I day
27.7 x 10’ 1/yr x ---
0.85
x x . 3.13 x 104 l/hr (3.52)
365 days 24 hour8
Ion Exchange 253
At a resin capacity of 115 gm per liter of resin, the amount

of resin that must be available is:
1icer (resin) 219.12 (MU of ly-HCl) 20.0 g 3.73 x 104 1
x x-x
I15 g 146.19 (MU of 1ysine) liter hour
(3.53)
I 9.7L x lo3 l(resin)/hr
To obtain the maximum utilization of the resin in this

operation, series bed operation (Carrousel) operation is
recommended. This operation uses three beds of resin in a
method having two beds operating in series while the product
is being eluted from the third. The freshly regenerated resin
is placed in the polishing position when the totally loaded
lead bed is removed for regeneration.
The elution/regeneration step, which includes

backwashing, eluting and rinsing the resin, might take up to
four hours. Therefore, enough resin must be supplied to take
up the lysine-HCI presented during that time. The resin
requirement for the commercial scale operation would be:
9.71 x lo3 l(resin) 4 hour
x = 3.88 x lo4 l(resin)/bed (3.54)
hour bed
Thus, three beds of 39 m3 resin each is required to produce

9,000 m tons of lysine/year.
3.7.1.2 Comparison of Packed and Fluidized Beds:

Belter and coworkers (99) developed a periodic countercurrent
process for treating a fermentation broth to recover
novobiocin. They found that they were able to scale up the
laboratory results to production operations if the two systems
have similar mixing patterns and the same distribution of
residence times in the respective columns. The mixing
patterns are the same when the space velocity (F/V,) and
the volume ratio (Vn/c) are the same. This is shown in
Figure 3.55 for the effluent concentration of novobiocin from
laboratory and production columns.
The scaling-up of packed beds is subject to the

difficulties of maintaining even flow distributions. Removal
of solution through screens on side walls is not recommended
and the flow of resin from one section into another of much
greater area could distort the resin flow profile.
700 f-
l lA6ORATOflY
,,,,*,
. LABOnATORY
l PROOUCTION
0 10 60 90 120 150 180 210 240 2?0

TIME IN MINUTES
Figure 3.55 Comparison of experimental curves for

laboratory and production columns (Reference 99).
The problems of scaling-up fluidized bed operations are

more difficult in terms of design calculations, but flow
distribution is more easily designed because of the mobility
of the resin. The degree of axial mixing of the liquid and
the resin has to be taken into account when calculating the
changes necessary in the bed diameter and bed height.
Figure 3.56 shows the increases in bed height necessary when
scaling-up packed and fluidized beds with bed diameter
increases (100).
Mass transfer coefficients have been correlated for

packed and fluidized beds (101). The mass transfer
coefficients for packed beds are 50 to 100% greater than for
fluidized beds.
The volume of resin in a packed bed is about half that

in a fluidized bed but the packed bed column may be up to
eight times smaller. Despite this, a complete fluidized bed
operation may still be smaller than a fixed bed operation.
Also, fluidized bed columns do not operate at high pressures
so they can be constructed more economically.
Ion Exchange 255
WELL MIXED STAGES

EXPANDED BED
PLUG FLOW
1 , 1
L
. I .o I.5 2.0
COLUMN DIAMETER (METER)
Figure 3.56. Scale-up relationships for fluidized beds

(Reference 100).
3.7.1.3 Pressure Drop: The pressure drop across an ion

exchange bed has been represented by an equation (102)
which depends on the average particle diameter, the void
fraction in the bed, an exponent and a friction factor
dependent on the Reynolds number, a shape factor, the
density and viscosity of the fluid and the flow rate.
While that equation has internally consistent units

(English system), the variables are not normally measured in
those units. Another disadvantage is that one must check

graphs of the exponent and friction factor versus the
Reynolds number to use the equation.
For laminar flow with spherical particles, the equation

can be simplified to:
AP 0.0738 u(cp) V,(l/min) (1 - c)3
-(bar/cm) = (3.55)
3
L Di(mm*) E
For most ion exchange resins, the void volume is about 0.38,
so that (1 - e3) / c3 = 4.34 and:
AP 0.32 u(cp) Vo(limin)
-(bar/cm) = (3.56)
L Di(mm*)
Table 3.17 shows the agreement between results from

experiments and those calculated with this equation for
several ion exchange resins.
Table 3.17: Pressure Drop for Commercial Ion Exchange Resins

Flow Rate Mean Bead Pressure Drop (bar/cm)
Resin (k?/min) Diameter (mm) Calculated Measured
Dowex SBR-P 151.4 .750 5.28 5.08
Dowex WGR-2 22.1 .675 1.00 1.06
Dowex MWA-1 15.1 .675 0.67 0.71
Dowex HSA-1 37.8 .650 1.79 2.13
Dowex MSC-1 30.3 .740 1.08 1.22
Figure 3.57 shows the use of a size factor for resins

which is used to develop a pressure factor. This pressure
factor can then be used to calculate the pressure drop under
conditions of different solution viscosities, flow rates or
particle size distributions once the pressure drop is known at
one viscosity, flow rate and particle size distribution.
3.7.2 Computer Calculations

Computer methods have been used to predict the shape
of the adsorption breakthrough curves for packed bed
systems based upon data obtained from small scale batch
Ion Exchange 257
Hld3a 030 j0 Y313W Y3d YVU
..
w
2
experiments. The computed effects of changes in operating

parameters such as flow rate, bed dimensions and inlet solute
concentrations have been compared for agreement with
experimental values.
The programs developed by Tan (103), which are

discussed in Chapter 2, may also be applied to ion exchange
fixed bed systems. Liberti and Passino (98) have developed
an alternate method for calculating breakthrough curves and
a model for regeneration using graphical evaluation. The
model is one in which the solid and fluid are viewed as
contacting one another countercurrently and continuously at
their respective average concentrations. Once the operating
and equilibrium lines for a given ion exchange operation have
been determined, the graphical method allows fixed bed
performances over a wide range of operating conditions to be
predicted.
Klein (104) has developed a computer program which

computes the equilibrium composition of the fluid phase
surrounding an ion exchange resin. The program, shown in
Table 3.18 for a TI-59, is based upon an empirically modified
mass action law. The memory location of the variables are
shown in Table 3.19. The ionic composition and the
selectivity of the resin and the valences of the ions are
required as input. As many as ten ionic species may be in
the fluid phase. The ionic composition for this program can
be in terms of the concentration rather than the more usual,
and more difficult to measure, activities. The usefulness of
this program is as a subroutine for more comprehensive
programs used to predict the performance of an ion exchange
column treating a multicomponent feedstream.
Klein and coworkers (105) have, in fact, developed such

a Fortran program to predict the course of ion exchange in
fixed bed columns. The program handles an arbitrary initial
composition distribution and feed composition history. The
program calculates gradual and step compositional changes
along the resin column.
The computer models have allowed predictions to be

made on the effects and likely performance when mixtures of
two or more solutes with different adsorption parameters are
applied simultaneously to an ion exchange column. Effects
such as the displacement of a loosely bound protein by a
Ion Exchange 259
Table 3.18: Program for Calculating yi values from Known xi

Values Using a TI-59 Calculator (104)
155 13 E
053 04 4 106 04 04 160 7: L6L
05-t 00 0 107 44 SUP1 161 ,3 C’
lj55 42 ST0 108 06 Cl4 162 53 <
;?J* ‘14 04 109 45 x 163 43 RCL
110 73 RC+ 164 09 05
05; ;: +)_ 111 01 01 1rjs 65 x
OS9 42 ST0 lib 53 (
if; ;; &,
ii&] tjg, 06 147 01 1
lj& 1 00 cl l _;e 7s -
lj 6 9
_ 42 ST0 169 43 RCL
ur,3
. . 52 S2 I?0 oi 06
064 76 LBL 171 6’3 x
065 10 E’ 172 43 RCL
Oki; 2-3 IllV ! ‘3. !O 10
067 87 If-F 174 5S +
ij<.:5: rjl 01 121 il CT0 175 4.3
OiJ9 14 D 13’ 14 D I?6 52 R$r;
.‘-
070 53 ( 1% 76 LEL 177 5-t )
0: 1 s3 < 124 15 E
-_c 125 43 RCL
IJ #’.! 43 RCL
073 07 07 126 50 50
074 3S + 127 32 X:T
075 43 RCL 123 43 RCL 183 43 RCL
-...
IJ(’ k. 03 03 129 06. 06 i y.
-3 *;;
q- 30
c-
077 54 ) 50 1x1 134
;;y 77 GE
ij78 55 i 135 32 XYT
1-17Q lj2 2 196 Oij Ci
lj;:b 55 i 157 77 CE
135 3t IF6 :;; A*; D
083 54 ) 136 22 IttY 150 32 XfT

084 42 STO 137 %7 IFF 191 77 ix
03s 09 09 lQ$ 01 01 192 19 D’
iI
_‘fY;.a 76 LE:L 139 l’t;C’ 193 61 CiO
087 1-t D 14.0 00 0 194 12 B
0:‘;:s 53 ’ 141 3’ X:T 1;5 76 L9L
039 7.3 R& 142 RCL 136 1.4 [I’
050 02 02 ;;z ‘;;
45 g; 197 32 STF
091 65 x 198 01 01
09’2 43 RCL =
199 61 CTO
Q*33 09 09 :j; :; &L 200 10 E*
147 40 40 ‘01 76 LE:L
142 42 ST0 202 4F;
. yx
145 07 07
ii 4 4 0 I 0I 150 61 CT0
I:145 ij> 2 151 12 B
lj : iJ 015 ij IS2 Pi; LBL
l 5..#
3 17 lx*
Table 3.19: Storage Locations for Problems with Known x’s (104)
0 1 2 3 4 5 6 7 a 9
Index Lowest reqlster number for
0 l/q s Sillall- Larg- t1/x1
Counter Li Kil ‘I ti est tl est tl
- ------ --__ -
1
‘1 ‘2 ‘3 ‘4 ‘5 ‘6 z7 ‘8 z9 ‘10
2 1
K21 K31 K41 K51 K61 K71 Kal K31 KlO,l
3
x1 x2 X3 *4 ‘5 ‘6 x7 xa X9 x1o
4
t1 t2 t3 t4 ‘5 t6 t? ta t9 t10
5 d n ItjZi dsldtl
protein with a greater affinity for the resin correspond to

what is observed in actual practice.
3.7.3 Economic Analysis

Erskine and Schuliger (106) developed an “operating
line” method of relating particular operational performance
with relative cost effectiveness for the economic analysis of
adsorption and ion exchange systems.
Figure 3.58 shows how the operating line correlates the

“superficial liquid retention time” (the net resin volume
divided by the superficial liquid flow rate) and the
“exhaustion rate” (the loading rate as a function of the
degree of regeneration). The data points for this curve are
obtained from breakthrough determinations for different bed
depths.
The operating lines have the inherent characteristic of

exponentially approaching a finite limit at each axis. The
limit along the Y-axis, projected to the point on the ordinate
representing the minimum exhaustion rate, corresponds to the
retention time that provides saturation levels corresponding
to equilibrium with the feedstream concentration. The limit
along the X-axis, projected to the point on the abscissa
representing the minimum retention time, is the contact
period required to achieve only the treatment level specified
by the desired effluent concentration. Operating points
between these two limits are then distributed along a curve
Ion Exchange 261
whose position depends on the effluent concentration required

(Figure 3.59) and type of ion exchange column (Figure 3.60).
\
\
\
\
B
D
\
---
l
~~~---~-~
MINIMUM EXHAUSTION RATE

I
I
0
SUPERFICIAL LIOUID RETENTION TIME
Figure 3.58. Development of the operating line from

breakthrough curves at different bed heights (Reference 106).
LOW CONCENTRATION OF
ADSDRBATE IN EFFLUENT
-0
SUPERFICIAL LIQUID RETENTION TIME
Figure 3.59. Effect of effluent concentrations on the

operating line position (Reference 106).
i
I
I SINGLE FIXED BED
Ly
I
2 SERIES FIXED BED
u
I
t4
2 I
::
w
--_L---____-____
0 I
0
SUPERFICIAL LIOUID RETENTION TIME
Figure 3.60. Effect of the type of ion exchange column on

the operating line position (Reference 106).
Once the operating line is established for a specific set

of operating parameters, the economic analysis is carried out
as an operating cost optimization where a recovery factor is
specified to represent the capital associated for the
particular column system. The analysis combines the
capitalized costs obtained with the direct operating costs to
identify the design and operating parameters that are the
optimum for the specific application (107).
3.7.4 Ion Exchange Resin Limitations

When ion exchange resins are used for an extended
period of time, the exchange capacities are gradually
decreased. Possible causes for these decreases include
organic contamination due to the irreversible adsorption of
organics dissolved in the feedstream, the oxidative
decomposition due to the cleaving of the polymer crosslinks
by their contact with oxidants, the thermal decomposition of
functional groups due to the use of the resin at high
temperature and the inorganic contaminations due to the
adsorption of inorganic ions.
When a resin bed has been contaminated with organic

Ion Exchange 263
foulants, a procedure is available that can help to restore the

resin (108). Three bed volumes of a 10% NaCl - 2% NaOH
solution are passed through the resin bed at 50°C. The first
bed volume is passed through at a flow rate not greater than
1 L/s-m3. The second bed volume is introduced at the same
rate but is kept in the bed for at least 8 hours, with
occasional agitation by air lancing. The last bed volume is
then passed through at the same flow rate, followed by a
thorough rinsing and two regenerations with the standard
regenerate.
If the fouling is due to microbial contamination, the

same authors recommend backwashing the bed and then
filling the entire vessel with a dilute solution (< 0.05%) of
organic chlorine. This solution should be circulated through
the bed for 8 hours at a warm (WC) temperature. After
this treatment, the resin should be backwashed, regenerated
and rinsed before returning it to service. This procedure
may cause some oxidation of the polymeric resin, thereby
reducing its effective crosslinking and strength. Therefore,
treating the resin in this fashion should not be a part of the
normal resin maintenance program.
Physical stability of a properly made cation or anion

exchange resin is more than adequate for any of the typical
conditions of operation. These resins can be made to have a
physical crush strength in excess of 300 grams per bead.
Perhaps more important are the limitations inherent in

the structure of certain polymers of functional groups due to
thermal and chemical degradation. Thermally, styrene-based
cation exchange resins can maintain their chemical and
physical characteristics at temperatures in excess of 125°C.
At temperatures higher than this, the rate of degradation
increases. Operating temperatures as high as 150°C might be
used, depending on the required life for a particular
operation to be economically attractive.
Strong base anion exchange resins, on the other hand,

are thermally degraded at the amine functional group.
Operating in the chloride form, this is not a severe
limitation, with temperatures quite similar to those for cation
exchange resins being tolerable. However, most strong base
anion exchange resins uses involve either the hydroxide form,
the carbonate or bicarbonate form. In these ionic forms, the
amine functionality degrades to form lower amines and
alcohols. Operating temperatures in excess of 50°C should be

avoided for Type I strong base anion exchange resins in the
hydroxide form. Type II strong base resins in the hydroxide
form are more susceptible to thermal degradation and
temperatures in excess of 35°C should be avoided.
The amine functionality of weak base resins is more

stable in the free-base form than that of strong base resins.
Styrene-divinylbenzene weak base resins may be used at
temperatures up to 100°C with no adverse effects.
Chemical attack most frequently involves degradation

due to oxidation. This occurs primarily at the crosslinking
sites with cation exchange resins and primarily on the amine
sites of the anion exchange resins. From an operating
standpoint and, more importantly, from a safety standpoint,
severe oxidizing conditions are to be avoided in ion exchange
columns.
Oxidizing agents, whether peroxide or chlorine, will

degrade ion exchange resins (109). On cation resins, it is
the tertiary hydrogen attached to a carbon involved in a
double bond that is most vulnerable to oxidative degradation.
In the presence of oxygen, this tertiary hydrogen is
transformed first to the hydroperoxide and then to the
ketone, resulting in chain scission. The small chains become
soluble and are leached from the resin. This chain scission
may also be positioned such that the crosslinking of the
resin is decreased, as evidenced by the gradual increase in
water retention values.
The degradation of anion resins occurs not only by

chain scission but also at the more vulnerable nitrogen on
the amine functionality. As an example, the quaternary
nitrogen on type I strong base anion resins is progressively
transformed to tertiary, secondary, primary nitrogen and
finally to a nonbasic product.
Oxidative studies on resins with different polymer

backbones and functionalities have been performed as
accelerated tests (110). The data is shown in Table 3.20 for
polystyrene and polydiallylamine resins. Although the
polydiallylamine resins have a higher initial capacity, they
are much more susceptible to oxidative degradation. When
the polystyrene resin has a mixture of primary and secondary
amino groups or when a hydroxy-containing group is attached
Ion Exchange 265
to the amine of the functional group, the susceptibility to

oxidation is enhanced. Thus one can understand the lower
thermal limit for Type II anion resins compared to Type I
resins.
Table 3.20: Oxidation of Polystyrene and PolydiaIlylamine Resins

in a One-Week Accelerated Test at 80°C (110)
Initial Base Base Capacity Lost
Resin Backbone Functional Group Capacity (meq/g) During Test (X)
Polystyrene R-CH2N(CH3), 4.4 1
Polystyrene R-CH,N(CH,CH3), 4.5 2
Polystyrene R-CH2N(C,H40H), 3.6 17
Polydiallylamine N-H 8.3 37
Polydiallylamine N-CH,CH3 7.3 37
Polydiallylamine N-C3H7 6.7 31
The effect of thermal cycling on strong base anion

resins has been studied by Kysela and Brabec (I 11). The
average drop in strong base capacity was 2.1 x low4 mmol
(OH’)/mL over the 480 cycles between 20°C and 80°C.
Figure 3.61 shows the decrease in total exchange capacity
and in strong base (salt splitting) capacity for each of the
individual resins included in the study.
As would be expected, the combination of thermal and

osmotic shocks has been shown to have a large effect on the
osmotic strength of anion resins (112). This is illustrated in
Figure 3.62, where the decrease is shown to be over 15% in
just 150 cycles. The important processing point to remember
from this is that if the regeneration of a resin is performed
at a temperature more than 20°C different from the
temperature of the operating (loading) process, it is advisable
first to adjust the temperature of the resin with condensate
until it is at the temperature of the regenerant.
It must be noted that nitric acid and other strong

oxidizing agents can cause explosive reactions when mixed
with organic materials such as synthetic ion exchange resins.
Ion Exchange 267
0 50 100 150
OF CYCLES
NUMBER
Figure 3.62. The effect of thermal and osmotic cycling on

the osmotic strength of a strong base anion resin, AV-17.
Curve (1): effect of thermal cycling; Curve (2): effect of
osmotic cycling; Curve 3: effect of thermal and osmotic
cycling (cold alkali to hot acid); Curve (4): effect of thermal
and osmotic cycling (hot alkali to cold acid) (Reference 112).
Nitric acid should NEVER be used as the regenerating acid

for cation resins or to place anion resins in the nitrate form.
3.8 BIOTECHNOLOGY APPLICATIONS
It used to be that biotechnology was synonymous with

fermentation processes. More recently, the term has been
associated primarily with recombinant DNA technology.
Actually biotechnology applications range from fermentation,
to cell harvesting, to enzyme immobilization, to biological
sensors, to microbial manipulation. In essence, they include
all applications involving microorganisms, enzymes and
biological raw materials. Ion exchange resins have been

used, at least in the laboratory, in all of these applications.
The outgrowth of the laboratory usage of ion exchange resins
has been their use in the commercial recovery and
purification of biotechnology products. In fact, most
purification schemes for biotechnology operations are merely
changes in the scale or size for techniques developed in
laboratory biochemical processes.
The industrial purification of biochemicals will usually

involve an adsorption or chromatographic operation in
conjunction with the strictly ion exchange operation. The
applications provided will often give the complete steps
required for biochemical purification.
3.8.1 Amino Acid Purification

The amino acids may be grouped into four main types
(Figure 3.63):
(1) Basic amino acids - arginine, asparagine,
glutamine, histidine, lysine and
tryptophan
(2) Dicarboxylic amino acids - glutamic acid
and aspartic acid
(3) Neutral amino acids - alanine, cysteine,
cystine, glycine, hydroxyproline,
isoleucine, leucine, methionine, proline,
serine, threonine and valine
(4) Aromatic amino acids - phenylalanine
and tyrosine.
Tryptophan could also be categorized with the aromatic amino

acids. However, for ion exchange separations, the basic
functionality is of more significance.
All of these amino acids are adsorbed on strong acid

cation resins in the hydrogen form. This allows their
separation from non-ionic impurities, such as may be present
in protein hydrolysates. If the feed solution is first adjusted
to neutral or basic pH, only the basic amino acids are
adsorbed by the strong acid cation resin in the sodium or
ammonium form. Basic amino acids are also preferentially
adsorbed on weak acid cation exchange resins when the
Ion Exchange 269
/l&c amino aid rln~lurc Ii

I
R-T-Coo”
NH,
Clycine H- /\rpucate HO,

C-CH,
Alrnine CH,- //
0
Valine CY,
Aspanginc NY,
CH-
‘C-CH,-
C/H, //
0
Leucinc C\HI
Glutamate 0
CH-CH,- \
C-CH,-CH,-
C/H, //
0
Is&u&e CH,-CH,-CH-
Glutamine NY2
CH, C-CH,-CH,-
/
swine OH-CH,- 0
Thmoninc OH Lysinc H,N-CH,-CH,-CH,-CH,-
CH,-A-- Arginine H,N-;-NH-CH,-CH,-CH,-

H
+ NH,
Cysteine HS-CH,-
Hircidine HC=C-CH,-
Mcthionine CH,--S-CH,-CH,- I I
Phcnylalrninc , , CH,_ HY\ ,NH
o- -
Proline
Tyrosine 2
OH ’ \ CH,- H,C \ ,CooH
-u- - I C
HsC.~’ ‘”
Tryptophan Ii
Figure 3.63. Structures of amino acids commonly found in

proteins.
feedstream has been adjusted to pH 4.7. The dicarboxylic

amino acids can be preferentially adsorbed, compared to the
other amino acids, onto weak base anion exchange resins
converted to the HCl form. The neutral amino acids can be
separated from each other by stepwise elution with changes
in the pH of the eluting solution.
Whereas Stein and Moore (113) used a single strong

cation resin (Dowex 50 W) with changing elution conditions
to separate amino acids, Winters and Kunin (3) used three

resins in five columns and different elution conditions (Figure
3.64) to separate amino acids from protein hydrolyzate into
their three charge groups (acid, neutral and base) and three
individual amino acids (histidine, arginine and lysine). The
carboxylic acid resin may offer an advantage for some amino
acid purifications because of its higher buffering capacity,
compared to strong cation resins, in the critical pH region.
Proteln HydrolYZate I d.
With HCI Weak Cation IRC-50
Buffered PH 7.0
1
Weak Base Resin ' b Hlstldlne and
Arglnlne and
Lyslne Adsorbed Neutral Acids
/I Pass Through
HCI Amino Acids
Adsorbed Pass Through
1 J
Elute Wlth HCI Weak Catlon IRC-50
1 Buffered oH 4.7
Weak Base ReSlfl
1
Strong Base Resin 4/\
/\ IRA-400 Hlstldlne Neutrals
Glutamlc Acid Neutral Pass Through
Adsorbed
J
And Asoartic Acid And Basic
JL
Are Adsorbed Acids PaSS
Lyslne Arslnlne
Through
Adsorbed Passes Elute
Through Wlth HCI
Elute Wlth
i
Acetate Buffer
Elute Wlth
Acetate Buffer
Figure 3.64. Separation of amino acids from a hydrolyzed

protein feedstream (Reference 3).
The use of weak base anion exchange resins for

recovering glutamic acid and aspartic acid from protein
hydrolyzate has been reported (114) for commercial
preparations of amino acid solutions of neutral and basic
amino acids for intravenous feeding.
Another scheme (Figure 3.65) has been used

commercially for separating amino acids from protein
hydrolyzate. In these schemes, it can be seen that cystine
and cysteine have been removed prior to the ion exchange
treatment of the protein hydrolyzate.
According to the scheme in Figure 3.65, the pH of the

protein hydrolyzate solution is adjusted with HCl to a level
near the isolectric point of glutamic acid to allow the
removal of half to two-thirds of the glutamic acid by
Ion Exchange 271
jconcentratlon
GLUTAilC ACID
ARGININE, LYSINE. HISTIDINE
@-, PHENYLALANINE. TRYPTOPHAN
LEUCINE, VALINE, ALANINE.
ISOLEUCINE, GLYCINE
GLUTAMIC ACID, ASPARTIC ACID
HREONINE. SERINE, PROLINE
Figure 3.65. Separation of amino acids from a hydrolyzed

protein feedstream with controlled elution of individual amino
acids.
precipitation. This solution is then passed through a strong

base resin which adsorbs the basic amino acids. A duel
functionality resin (or a mixture of strong base and weak
cation resins) is used next to adsorb the phenylalanine and
tryptophan. The majority of the neutral amino acids are
adsorbed on a strong cation resin, while a weak base resin
adsorbs the acidic amino acids, including any residual
glutamic acid. The pH changes of the solution as it passes
through the different resins and the pH of the eluting
solutions which allow the elution separation of the adsorbed
species vary for different amino acid producers.
The recovery of amino acids from these protein

hydrolyzates has the limitation that the yield of each amino
acid is proportional to its presence in the original proteins.
While protein hydrolyzates will continue to be used as the

raw material for some commercial amino acid production, the
trend is toward more use of fermentations which produce a
specific amino acid.
Amino acid fermentation broths have characteristics

which are different from those of protein hydrolyzates. The
differences are (115):
(1) In most cases, a significant amount of a

single amino acid can be accumulated in
the fermentation broth in its free and
salt forms and other contaminant amino
acids are usually few in number and
small in quantity.
(2) Amino acids produced by fermentation

are usually the biologically active L-
form.
(3) Fermentation broths are aqueous
solutions of amino acids whose
concentrations vary from 0.1 to 10%
(W/V) depending on the individual
fermentation process.
(4) Fairly large amounts of microbial cells

and soluble protein-like biopolymers are
usually contained in fermentation broths.
(5) Some inorganic salts derived from

microbial nutrients always exist in
fermentation broths.
(6) Other organic impurities derived from

microbial nutrients and metabolites, such
as sugars and pigments, are also present
to some extent.
As an example, L-tryptophan (116) is fermented using

Arthrobacter paraffineus which requires histidine as a
nutrient. The fermentation filtrate is passed through a
strong cation resin which selectively adsorbs the L-
tryptophan. This is eluted from the resin with OSN
ammonium hydroxide solution from which it is concentrated
and crystallized. The resin used and the elution conditions
can be specified with more precision for purification of an
amino acid from a fermentation broth since the broad
Ion Exchange 273
spectrum of amino acids presented in protein hydrolyzates are

not in the fermentation broth.
Many other amino acids, a few of which are listed in

Table 3.21, have been recovered from fermentation broth
using ion exchange resins. After loading the resin to
breakthrough with the amino acid feedstream, it is necessary
to wash the resin with deionized water before eluting the
amino acid so that the glucose, protein (bacterial cell
components), salt and color impurities are separated from the
amino acid.
Table 3.21: Amino Acids Recovered from Fermentation Broths

Amino Acid Ion Exchange Resin Eluant
-_- Reference
--
N-acetylglutamine IRA-400 0.5N H,SO, 117

SK-l
L-histadine Strong Acid Cation Ammonium Hydroxide 116
Lysine Strong Acid Cation 2N Ammonium Hydroxide 119
Lysine Weak Acid Cation 0.15N Ammonium Hydroxide 120

pH 7.0
L-valine Strong Acid Cation l.ON Ammonium Hydroxide 121
L-serine Strong Acid Cation Ammonium Hydroxide 122
Glutamic acid Strong Acid Cation NaCl 123
of amino acids by ion exchange resins is

The adsorption
strongly affectedthe pH of the solution.
by Figure 3.66
shows the decrease in the adsorption of lysine by a strong
cation resin in the ammonium form as the pH is changed.
Other inorganic ions may also compete with the amino acids
for exchange sites on the resin, similar to the ion exchange
treatment of other mixtures of ions in solution.
In a different use of resins in treating amino acids, a

monosodium glutamate mother liquor, containing 4 to 6%
monosodium glutamate, has been passed through a weak base
resin to remove the color impurities in the solution of the
amino acid (124). The flow rate required is approximately 1
120
100
AMOUNT
OF
80
LYSINE
ADSORBED
b
0
l
(s/l-resi")60
40 I I I I I I 1 I 1
0 2 4 6 8
PH OF SOLUTION
Figure 3.66. The change in the adsorption of lysine on a

strong cation resin (Diaion SKI, NH: form) as a function of
the pH of the fermentation broth (Reference 191).
bed volume per hour to reduce the degree of coloration from

0.6 to 0.1. A solution of 4% NaOH is used for removing the
colored impurities from the resin. Depending on the color
quality of the mother liquor, between 20 and 60 liters of the
mother liquor may be treated by one liter of resin.
3.8.2 Antibiotics
The first commercial ion exchange operation for the
recovery and purification of biological products occurred in
the 1950’s when the antibiotic streptomycin was recovered
from a fermentation broth using a carboxylic cation exchange
resin (125). After the resin, initially in the sodium form,
became saturated with streptomycin, dilute mineral acid was
used to elute the antibiotic. After elution, the resin was
Ion Exchange 275
converted back to the sodium form using a dilute sodium

hydroxide solution.
Soon after the commercialization of that process for

streptomycin, a similar process was developed for the
recovery and purification of neomycin (126). In this case, a
strong base anion resin was utilized in the chloride form and,
after first rinsing the loaded resin with 95% methanol,
neomycin was eluted with a solution containing 16% HCI and
84% methanol. The eluate contained 80 to 95% of the
antibiotic with a tenfold reduction in fluid volume.
Belter (127) has recently presented an overview of the

various types of antibiotics which have been purified or
recovered from fermentation broths using ion exchange
resins. Currently, antibiotics are classified into four
categories: the aminoglycosides, the ,&lactams, the macrolides
and the tetracyclines. Of these, only the tetracylines do not
utilize ion exchange resins in their recovery.
The aminoglycosides are characterized by carbohydrate

structures attached to a complex ring structure. Examples of
this type of antibiotic are streptomycins, shown in Figure
3.67, neomycins, gentamicins and kanamycins. It is the
strongly basic guanidine groups that undergo the adsorption
interaction with the ion exchange resin at neutral pH
conditions.
HOC" "C-O-C"
L-C"
&OH
Figure 3.67. Chemical structure of streptomycin.
A fluidized bed can be used to treat the streptomycin-

containing fermentation broth directly with the ion exchange
resin. The broth flows up through the resin bed, expanding

it by about 25Oh. The height to diameter ratio of these
columns is about 3/4. When the resin is completely loaded
with the antibiotic, the broth in the fluidized bed is first
removed for recycling by rinsing with water and then the
antibiotic is eluted down through the packed resin bed. The
percentage antibiotic recovery using the fluidized bed
technique was about 85% while the recovery from a filtered
broth was about 73%.
When streptomycin is adsorbed from a fermentation

broth where other impurities, especially ionic ones, are
present, the amount adsorbed on the carboxylic resin can be
substantially decreased as the ionic impurity concentration is
increased. This is shown in Table 3.22 for different amounts
of sodium chloride in the streptomycin solution (128). The
interference of multivalent ions on the adsorption of
streptomycin is even greater. The interference of these ions,
however, can be overcome by adding anions to the broth
which will cause precipitation or complexation with the
multivalent ionic impurities.
Table 3.22: Effect of Sodium Chloride on Streptomycin Adsorption

on Weak Acid Cation Resin (128)
Feed Solution Streptomycin
Compositioa(mg/ml) "Ll&+~
Streptomycin NaCl Relative Adsorption During Loading
Concentration Concentration Loading (mg/ml) (% of 6)
0.88 0 1.00
0.88 20 .20
II
0.88 30 .15 ‘
0.88 50 .13 37
The degree of purity obtained after elution is only 70

to 80%. Therefore, the eluant is further purified by passing
it through a column of strong base resin, Type I, in the
chloride form to remove 90 to 95% of the color; then through
a mixed bed of strong acid cation resin and weak base anion
resin to remove the calcium and magnesium salts; and finally
through a second carboxylic acid resin bed for final
purification. The purity of the eluant is raised to over 95%
Ion Exchange 277
by these additional steps. It must be added that the cation

resin in the mixed bed needs to be highly crosslinked (20%)
to avoid adsorption of the streptomycin.
The /3-lactams are characterized by a fused four member

lactam ring and a five or six member sulfur-containing ring.
This category may be subdivided into the penicillins, when
the five member ring with sulfur is a thiazolidine and into
the cephalosporins when the sulfur containing ring has six
members. Figure 3.68 shows the structure of 6-
aminopenicillanic acid and Figure 3.69 shows the structure of
Cephalosporin C.
COOH
Figure 3.68. Chemical structure of 6-Aminopenicillanic acid.
COO"
H200CH3
Figure 3.69. Chemical structure of Cephalosporin C.
Penicillin is not usually recovered using ion exchange

resins because of the high degree of instability of penicillin
once it has been produced in fermentation (128). Penicillin
is unstable in water solution as evidenced by a 1% solution
degrading by 10% in one hour at room temperature.
Penicillin is also very sensitive to enzymatic decomposition
by the enzyme penicillinase, produced by many organisms,
which may contaminate the harvested fermentation product
during processing. More stable derivatives of penicillin and
cephalosporins have met with more success in ion exchange
recovery processes.
Cephalosporin C is hydrophilic and chemically

amphoteric. It cannot be extracted directly from the

fermentation broth with ion exchange resins. It is first
necessary to acidify, filter and perform preliminary
purification using as adsorbent resin as shown in Figure 3.70
(129-131). The eluate from the adsorbent resin contains over
80% of the antibiotic obtained by fermentation and less than
20% of the initial impurities. This eluate becomes the
feedstream for a medium base anion resin in the acetate
form. The flow rate may be as high as 5 bed volumes per
hour. Following the loading phase of the cycle,
Cephalosporin C may be eluted with a buffer solution of
pyridine acetate at a pH of 5.5 to 5.6. After elution, the
resin is regenerated with caustic soda and reconverted to the
acetate form for the next loading cycle. Over 95% of the
treated antibiotic is recovered in the eluant from the ion
exchange resin.
Broth 20% Acetone
.Eluate PyrOAc
Eluate - Ceph C
Figure 3.70. Non-ionic and anionic resin purification of

Cephalosporin C (Reference 129).
Other combinations of resin columns have been proposed

for the recovery of Cephalosporin C (132-134). A general
Ion Exchange 279
scheme for these is shown in Figure 3.71. A cation exchange

resin column in the hydrogen ion form is used to lower the
pH of the feedstream without adding additional anions. A
lead anion exchange column is used to adsorb the stronger
anions, while a second anion exchange column adsorbs the
antibiotic. The adsorbed antibiotic is eluted with potassium
acetate buffer solution. Variations on this arrangement
switch the positions of the cation and the first anion column
or use a mixed bed of cation and anion resin followed by the
bed of anion resin for cephalosporin adsorption.
Broth
Cation Exchanges H +
Exchange for Inorganic Cations
Resin
IRC!M (Lowers pH)
Anion
‘p&g, Adsorbs Strong
IRA94
Anion Impurities
Crystal
Figure 3.71. Cation and two anion resin purification of

Cephalosporin C (References 132- 134).
Since it is zwitterionic, Cephalosporin C may also be

adsorbed on a strong acid cation resin if the pH of the
feedstream is first adjusted to 2.5. However, the strong
acidity of the resin results in significant amounts of
degradation of the antibiotic to its lactone. A related
antibiotic, Cephamycin C, which contains a carbamoyl ester
in place of the acetyl ester at C-3, is more stable. This
antibiotic has been successfully purified using a low
crosslinked strong acid cation resin (135). Elution was
performed with sodium acetate buffer solutions.
An alternate approach (136), after adsorbing the

Cephalosporin C onto the cation resin, is to wash the column
with 2 bed volumes of water, one bed volume of 10% NaCl,
recirculating these effluents and then recovering the
antibiotic by eluting with cold water (O’C to S‘C) when the
effluent reaches a pH greater than 5.5. This results in a
yield of over 90%.
More recently another scheme (shown in Figure 3.72)

has been developed that uses two anion resin columns and
two adsorbent columns to recover Cephalosporin C (137).
Using this process results in a yield of 360 mg from the
original 60 liters of fermentation broth.
The macrolides are characterized by a cyclic lactone

structure to which one or more substituted sugars are
attached. Examples of this type are erythromycin, shown in
Figure 3.73, oleandomycin, rosamicin and tylosin.
Erythromycin has been recovered from a filtered

fermentation broth by passing it through a column of cation
resin (138). The effect of resin crosslinkage and acid
strength on the amount of antibiotic adsorbed is shown in
Table 3.23 for solutions that have an initial antibiotic
concentration of 1000 micrograms per milliliter. Lower
crosslinked strong acid resins had significantly higher
antibiotic adsorption levels.
The erythromycin was eluted from the low crosslinked

strong cation resin with the solvents shown in Table 3.24.
The combination of 0.25N ammonia and certain aqueous
organic solutions gave recovery of over 80% of the adsorbed
antibiotic.
Ion Exchange 281
fERMENTATlON FILTRATE
PH ADJUSTED TO 7.0
SODIUM CHLORIDE
PB 2.0 WASH
WATER YASHr PX 7.0 ELUTION
501 ACETONE
HP4D
Y
CRUDE CEPHAJIYCIN C
Figure 3.72. Alternate purification scheme for Cephamycin C

(Reference 137).
CH3 Ii
Figure 3.73. Chemical structure of Erythromycin.

Table 3.23: Effect of Resin Crosslinkage and Acid Strength on the

Amount of Erythromycin Adsorbed (138)
Type of Ion Exchange Resin Trade Name Form Amount Adsorbed (mg/ml)
Low crosslinked, Diaion PK-204 12.0

strong acid cation Diaioo PK-208 11.6
Dowex 5OWX4 11.7
Amberlite XE-100 7.5
Diaion SK-104 7.5
Duolite C-25 a.2
Lewatit SP-100 7.6
High crosslinked, Dowex SOWXS t 0.86

NH4
strong acid cation
Low crosslinked, Duolite LX-101 + 0

NH4
+
weak acid cation Amberlite IRC-50 NH4 0
Amberlite IRC-50 pH 6.0 7.4
Amberlite IRC-50 pH 6.5 5.1
Table 3.24: Elution Solvents for the Recovery of

Erythromycin (138)
Elution Solvent Erythromycin Recovered (X)
0.25 N ammonia 37.9
1.0 II NaCl 8.1
60% methanol 0
0.25 N ammonia/60% methanol 71.3
0.25 N ammonia/90% methanol 89.0
0.25 N ammonia/60% isopropanol 71.0
0.25 N ammonia/9Of isopropanol 86.0
0.25 N ammonia/60% acetone 65.0
0.25 N ammonia/90% acetone 58.2
Rakutina and coworkers (139) reported that decreasing

the particle size of microporous ion exchange resins would
greatly increase its adsorption capacity for erythromycin and
oleandomycin antibiotics when the particle size was less than
40 microns. This is shown in Figure 3.74. The adsorption
was performed under batch conditions with the antibiotic
Ion Exchange 283
initially at a concentration of 4000 units/ml. These

differences in amounts of adsorption can be interpreted as
changes in the selectivity coefficient as shown in Figure 3.75.
$ 1.6
1.4
RESIN PARTICLE SIZE (MICRON)
Figure 3.74. Dependence of the sorption capacity of the

sulfonated naphthol cation exchange resin SNK- 1OE for
erythromycin (curve 1) and oleandomycin (curve 2) on the
resin particle size (Reference 139).
RESIN PARTICLE SIZE (MICRON)
Figure 3.75. Dependence of the selectivity coefficient of

erythromycin (curve 1) and oleandomycin (curve 2) sorption
on the resin particle size of SNK-1OE (Reference 139).
Samsanov and coworkers (140) have developed a system

of finely divided ion exchange particles immobilized in a
porous, highly permeable polymer matrix to adsorb the
antibiotics erythromycin and oxytetracycline. When the ion
exchange particles were only about 7 microns in diameter,
the individual permeable polymer beads were approximately
500 to 600 microns in diameter and contained 20 to 75% of
the resin particles (by volume).
This type of system allows a higher utilization of the

ion exchange capacity and yields diffusion coefficients which
are higher by an order of magnitude. when compared to
systems with ion exchange resins of 500 to 800 micron
diameters. This is shown in Table 3.25. The effect of
solution feed rate is shown in Figure 3.76 with different
breakthrough curves for erythromycin adsorption.
The carbapenem antibiotics, thienamycin and olivanic

acid, have a structure that is a p-lactam fused to a five
member ring with a sulfur functionality attached. These
antibiotics have been adsorbed on low crosslinked strong
based anion resins in the chloride form (141). The resin
would adsorb about l,OOO,OOOunits of antibiotic activity per
kilogram of resin in an agitated slurry. After transfer of the
resin-antibiotic complex to a column, the antibiotic is eluted
using a solution of 6% KC1 with 50% acetone.
Thienamycin has also been recovered from its

fermentation broth using a column of strong base resin in
the HCOs’ form (142). A fermentation broth containing about
12 mg/mL of antibiotic can be adsorbed with about 14 mL of
resin per mg of antibiotic when the flow rate is 0.08 bed
volumes per minute. This first step only yields a product
with 5 to 10% purity. The additional steps necessary (anion
resin column, adsorption column) are shown in Figure 3.77
which raise the purity to over 90%.
In addition, thienamycin has been recovered and purified

by adjusting the fermentation broth to a pH between 3 and 5
and then passing it through a column of low crosslinked
strong cation resin in the sodium form (143). A 2% aqueous
pyridine solution can be used to elute 45% of the original
antibiotic. Further purification is performed using
chromatographic techniques.
Ion Exchange 285
Table 3.25: Influence of the Content of Cation Resin

Microdispersion on the Capacity and Diffusion Coefficients for
Erythromycin and Oxytetracycline (140)
Total Amountof
Exchange Antibiotic Capacity Diffusion
Resin Capacity Adsorbed Utilized Coefficient
Type Antibiotic (meq/g) (meqla) (%f fcm2/sec)
Whole Bead Erythromycin 1.62 0.42 26 3.4 x 10-S
Dispersions Erythromycin 0.42 0.26 62 2.5 x lo-'

0.66 0.40 61 3.0 x 10-s
0.88 0.57 69 1.3 x 10-B
Whole Bead Oxytetracyclioe 5.11 2.35 46 1.3 x 10-E
Dispersions Oxytetracycline 0.16 0.42 55 1.5 x 10-7

1.60 0.83 52 1.2 x 10-7
2.37 1.36 58 0.8 x IO-'
1.0
09
0.8
0. 7
0.6
e 0.5
CI
a. 4
0.3
0. I?
a. 1
Figure 3.76. Breakthrough curves (effluent concentration

fraction versus effluent volume) for the sorption of
erythromycin by an immobilized SNK-30D dispersion at
different solution flow rates. Curve (1): 50 ml/(cm2 -hr);
Curve (2): 300 mL/(cm2/hr); Curve (3): 500 mL/(cm2-hr);
Curve (4): 1500 mL(cm2-hr); Curve (5): 200 mL/(cm2-hr)
through a non-dispersed SNK-39 sample (Reference 140).
:l;TEF$D,B~OTH,
,
* 1 -COLD WATER,
CO, SAT’iI
IF COLD WATER
XAD-2
CON$ENTRATE
PURE THIENAMYCIN
COLD WATER
DOWEX
1 x 2
(CL)
- CONCENTRATE -
Figure 3.77. Purification sequence for thienamycin. The

first anion column may be in either the carbonate or the
acetate form (Reference 142).
The antibiotic A-4696 has been recovered from acidified,

filtered fermentation broths by passing it through a column
of strong acid cation resin in the sodium form (144,145).
After loading, the resin is washed with three bed volumes of
deionized water. The antibiotic is eluted with an aqueous
solution of NaOH (pH 10.5), which is neutralized after
elution. The eluate is further processed so that eventually
an 80% yield is obtained.
3.8.3 Pharmaceuticals And Organic Acids

In addition to use in recovering antibiotics, ion
exchange resins have been used to recover other
pharmaceuticals and valuable organic substances from
fermentation broths.
Ion Exchange 287
Khose and Dasare (146) have used the porous phenol-

formaldehyde resins they developed to recover theophylline
(Figure 3.78). The amount of adsorption was not as great as
observed with the commercial Duolite resin, S-761, on a dried
basis. However, since their resin has 20% less moisture
content, the difference is not as large as it initially appears.
Figure 3.79 shows the dependence of this adsorption on the
flow rate of the feed stream.
100
0
0 400 800 1200 1600
INITIAL CONCENTRATION (MG MI-~)
Figure 3.78. Dependence of the uptake of theophylline on

the initial concentration of the equilibrated solution for
Duolite S-761 (Curve A) and for a new synthetic resin (Curve
B) (Reference 146).
‘OS0 200
Flow rate (cm’ h-l)
Figure 3.79. Dependence of the breakthrough capacity on the

feedstream flow rate for Vitamin B-12 (Curve A) and
theophylline (Curve B) (Reference 146).
The acetate forms of strong base anion exchangers have

been used to purify acetylsalicylic acid from commonly
occurring impurities (147).
Citric acid may be produced by the fermentation of

molasses or dextrose. After fermentation, the citric acid is
precipitated by the addition of lime at a temperature of
60°C. A precipitate of calcium citrate can then be separated
from the mother solution. The calcium citrate must then be
treated with sulfuric acid which dissolves the calcium citrate
and produces a precipitate of calcium sulfate. The solution
must be filtered and then treated with ion exchange resins to
remove dissolved salts and organic impurities. The treated
fluid can then be evaporated and crystallized to obtain the
citric acid crystals.
Jacob (148) has described the recovery of xanthylic acid

from fermentation broths. The acid is initially adsorbed onto
a strong base anion resin and then eluted with an aqueous
acid solution.
Tsao and coworkers (149) extracted itaconic acid from a

molasses fermentation liquor by adsorbing the itaconic acid
onto a strong base anion resin. The best results were
obtained by first passing the fermentation liquor (diluted with
three volume equivalents of water) through a column
containing the ion exchange resin. The itaconic acid was
eluted quantitatively from the resin using 3N NaOH.
Itaconic acid has also been purified after ultrafiltration

by passing it through a cation resin in the potassium form
(150). The effluent from this column is neutralized to pH 7.0
and passed through an ion exchange column membrane
system. The dialyzate is then passed through a weak base
resin column from which the itaconic acid is eluted with a
concentrated solution of potassium bicarbonate.
3.8.4 Protein Purification

The adsorption of a protein by an ion exchange resin is
dependent on the chemical composition of the resin, on the
nature of the protein and on the other components in the
solution. Girot and Boschetti (151) showed that a maximum
protein adsorption can be obtained as one varies the pH of
the solution and that the pH of the maximum differs from
Ion Exchange 289
protein to protein. They showed that this maximum can be

predicted using the equation:
1
s = KQA (3.57)
10pH-pK, + 1
Y
where QA is the practical specific capacity, S is the amount
of protein adsorbed, p1 is the isoelectric point of the protein
and K is a constant. This equation is based on the
hypothesis that the protein sorption is proportional to the
concentration of the ionized carboxyl groups and that the net
charge of the protein is a linear function of pH in the range
studied.
The curve in Figure 3.80 shows the relationship between

the protein’s isoelectric point and the pH corresponding to
maximum protein adsorption calculated using Equation 3.57.
The agreement with experimental results is very good. Such
a curve can be used to select the pH of a buffer for
maximum adsorption of a protein on a given ion exchange
resin.
I ! I 1 I
3 4 5 6
OPTIMAL PH SORPTION
Figure 3.80. Relationship between protein isolectric point

and optimal pH for protein sorption of the ion exchange
resin. The curve is a theoretical model; the points are
experimental results (Reference 151).
Keay (152) described the general conditions for the

purification of a protein. His specific example was for the
purification of neutral and alkaline proteases from B. subtilis
by adsorption on a cation exchange resin.
The sulfonated phenol-formaldehyde resin was adjusted

to neutral pH by washing it with the solution to be used as
the enzyme solvent. This solution must have a low ionic
strength (below 0.5 molarity) to enable neutral as well as
alkaline proteases to be adsorbed. The pH should be between
pH 6 and 7.5.
The maximum concentration of protein in the solution

should be 2.5Oh with the best results when it is less than 1%.
This corresponds to an enzyme level of about 50,000 enzyme
units per milliliter.
After adsorption, the resin column is washed at neutral

pH or neutral ionic strength so that the adsorbed protein is
not removed while unadsorbed material is removed for
subsequent treatment in the next cycle. Then the adsorbed
enzymes are eluted with a 0.05 to 0.2M solution to remove
the neutral proteases, followed by elution with a 0.2 to l.OM
solution to remove the alkaline proteases. These solutions
are usually O.lM phosphate buffer followed by 1.OM sodium
chloride. The elution pH is usually slightly higher than the
pH of the feed stream to avoid contamination of the product
with any adsorbed color body impurities.
When DEAE cellulose ion exchange resins have been

used, as was the case in the purification of dextranase (153),
the process is much different. The filtrate, with its pH
adjusted to 8.5, is mixed with 2 weight percent of DEAE
cellulose which had been pre-equilibrated by being allowed to
stand in 0.02M “Tris” buffer. This suspension is stirred for a
minimum of one hour. Then the mixture is filtered and the
filter cake is washed with the “Tris” buffer.
The washed filter cake is resuspended in a solution

containing 0.3 mole of sodium chloride and 0.05 mole of
“Tris” buffer to maintain the pH at 8.5. After stirring for
one hour to assure complete elution of the dextranase
enzyme, the DEAE cellulose is removed by filtration. The
filter cake this time is washed with the same sodium
Ion Exchange 291
chloride-“Tris” buffer solution. After precipitation with

ammonium sulfate, a very pure dextranase (about 6,000 units
per mg) is recovered from the filtrate.
In the isolation of a glucopyranose amino-sugar (154),

strong acid cation resins and anion resins are added to the
fermentation broth. After mixing for about one hour, the
resin-amino sugar complexes are separated from the liquid
and mycelium using a sieve screw centrifuge. After rinsing
with deionized water, a dilute salt solution is used to elute
the glucopyranose. The eluate is deionized by passing it
through a strong cation and weak base resin. Separation
into specific derivatives of the glucopyranose is carried out
with column chromatography. These procedures result in the
recovery of over 90% of the saccharase inhibitor activity in
the original fermentation broth.
The amastatin tetrapeptides, which inhibit the activity

of aminopeptidase A, have been isolated using a combination
of adsorption, ion exchange and chromatography (155). The
process is illustrated in Figure 3.81. After each step in the
process, the eluate was concentrated under vacuum to a dry
powder. Such drying would not necessarily be employed in a
commercial process.
Kiselev (156) has described a preparative method for the

isolation of pure fractions of di- and tri-phosphoionositides
from ox brain. The petroleum ether lipid extract was passed
through a Dowex 50 W (H+) resin column to remove the
divalent metal ions. Then sodium hydroxide in methanol
solution was added to the effluent to convert the lipids to
the sodium salt. The resulting solution was added to a
DEAE-cellulose column. Gradient elution with O-0.6M
ammonium acetate in chloroform/methanol/water (20:9: 1)
allowed the separation of the lipids into fractions of di- and
triphosphoinositides. The desired salt forms of the lipids
were obtained by passing the ammonium salts through Dowex
50 W (H+) and neutralizing with the appropriate base in
methanol solution. One kg of wet ox brain tissue yields
about 0.35 mmol of diphosphoinositide and 0.63 mmol of
triphosphoinositide. Table 3.26 shows the phospholipid
concentration as the ox brain extract progresses through the
preparative purification.
FILTRATE
7 --ll
50
r
2 MEOH
r
1OWEX
XAD-4
50 x 4
(I++)
CONCENTRATE TO POWDER CONCENiRATE ACTIVE PEAK

RESUSPEND IN WATER (PR 8.2) TO POWDER
0.1 N ACETIC ACID
WATER
1214/l/l RATIO
SILICA
DOWEX
GEL
1x4
ACETATE
CONCENTRATE TO POWDER CONCENTRATE TO POWDER

RESUSPEND IN 0.5 H PYRIDINE-ACETIC ACID
1
0.3 M PYRIDINE-
ACETIC ACID
DEAE
EPHADEX
- PURE
-- CONCENTRATE AMASTATIN
TO POWDER FRACTIONS
Figure 3.81. Purification scheme for preparing pure amastatin

fractions (Reference 155).
Table 3.26: Phospholipid Yield (mmol/kg) from

Wet Ox Brain Tissue (156)
Phosphoinositides
Stage Tri- Di-
- Mono-
In Petroleum Ether Extract 0.84 0.43 1.68
After Dowex 50 Treatment 0.67 0.37 1.32
After DEAE Chromatography 0.63 0.35 1.29

Ion Exchange 293
The waste streams from food processors have been

treated successfully with ion exchange resins to recover
proteins (157). As Table 3.27 shows, the efficiency of
protein removal is very good.
Table 3.27: Protein Recovery from Waste Streams (157)

Protein Nitrogen
(g/fi) % Protein
Waste Stream Before
-- After Removed
Cheese Whey Effluent 0.60 0.02 93.3
Potato Effluent 0.78 0.035 95.5
Distillery Effluent 0.48 0 100.0
Animal By-Product 0.52 0.036 93.0
While whey has been treated by such simple operations

as spray drying to concentrate the proteins, such processes
unfortunately include the ash and lactose in the concentrate.
Other processes, shown in Figure 3.82, have been devised to
recover higher value products from cheese whey. As can be
seen, the protein concentrates from these systems have ash
and lactose contents which destroy their functional
properties. These functional properties are foaming ability,
heat gelation, water binding and fat emulsion.
A technique for the recovery of proteins from food

processing operations has been developed by Bioisolates (158)
which retains the functional properties of the proteins. As
an example, the protein in the whey from milk is adsorbed
onto a DEAE cellulosic ion exchange resin in an ion
exchange bed with a sieve bottom, shown in Figure 3.83.
The protein is eluted using O.lN HCI to yield a dilute
solution containing 1 to 5% protein (W/V). Further
concentration using ultrafiltration provides solutions with 10
to 30% proteins (W/V) which can be dried to a product that
contains less than 3% residual salts.
More recently another system has been developed (159)

which allows the separation of cheese whey into its
individual components. After removing the colloidal casein
proteins by ultrafiltration, the clarified liquid is passed
through a large pore (500 A diameter), weak base resin to
I
\L
WHEY
CHEESE
ION EXCHANGE DEEIINER.
(2-BED)
\1
MULTISTAGE EVAPORATION
CRYSTALLIZATION/ > LACTOSE,98%

CENTRWUGATION
FlULTlSTAGE
EVAPORATION
+
PROTEIN CONCENTRATE
362 PROTEIN
6X ASH
54% LACTOSE
Figure 3.82. Purification scheme for proteins in cheese whey

waste.
CHEESE WHEY
PROTEIN
CONCENTRATE
35-75X PROTEIN
ION EXCH. DEIIINER.
1 I
80%T.S.
J
COMPRESSEDYEAST CAKE
Figure 3.83. Alternate purification scheme for proteins in

cheese whey waste (Reference 158).
Ion Exchange 295
adsorb about 90% of the soluble whey proteins. These are

eluted from the resin using a dilute aqueous HCI solution. In
these systems it is essential that the molarity of the eluting
acid be kept low to prevent the irreversible binding of the
proteins onto the resin.
3.8.5 Sugar Stream Processing

The largest use, on a volume basis, of ion exchange
resins in the purification of biochemicals or organic
feedstreams is the purification of sugar-bearing juices, liquors
and syrups. The sugar purification operations may be either
decolorization or demineralization processes.
3.8.5.1 Decolorization: Although the decolorization of

liquid sugars almost always involves adsorption on some form
of carbon, ion exchange resins are widely used in conjunction
with these carbon beds. Anion exchange resins are
particularly effective since most of the color impurities are
organic substances with anionic characteristics. The resins
are normally in the chloride form initially and the colored
organic species are usually eluted from the resins using
neutral brine solutions.
Figure 3.84 (160) shows the degree of colored impurity

removal from cane and beet sugar solutions for anion resins
with different degrees of crosslinking. As the degree of
crosslinking is reduced, the amount of sugar which can be
treated increases. It is also apparent from this figure that
the efficiency of decolorization depends very much on the
source of the sugar. This is shown for cane sugar from
different locations in Figure 3.85 (161).
When strong base anion exchange resins (chloride form)

are used for decolorization, the flow rate is typically 8 to 12
kL/hr for a space velocity between 2 and 4/hr.
Approximately 80 to 85% of the color impurities are removed
with units sized to treat 60 liters of fluid per liter of resin
over a 12 or 24 hour loading cycle (162). The temperature is
normally between 70°C and 80°C to decrease the viscosity of
these 60”Brix solution. The pH must be maintained neutral
in processing sucrose solutions since basic pH will cause the
formation of undesirable colored impurities and acid pH will
cause the inversion of sucrose to glucose and fructose.
If the strong base resin is regenerated to the chloride

Relative degree of cross-linking
Figure 3.84. Relationship between the resin’s degree of

crosslinking and its decolorizing power for sugars from
different sources (Reference 160).
801
0.11 0.09 0.07 0.05
Ash content (k)
Figure 3.85. Ash content and degree of decolorization using
an anion resin in the chloride form on sugars from different
geographic regions (Reference 16 1).
Ion Exchange 297
form and then used to treat sucrose solutions, the pH of the

solution will decrease, as shown in Table 3.28 (163), which
will lead to inversion of some of the sucrose. This occurs
because of the H&O, present in the sucrose solution after
the carbonation step in the refining process, according to
equation:
R-Cl + H+ + HCO; + R-HC03 + H+ + Cl- (3.58)
The recommended method of preventing this pH drop is to

pass a 0.1% soda ash solution (1 .O to 1.5 kg CaC0,/m3 resin)
through the resin bed after the regeneration with NaCl
solution.
Table 3.28: pH Drop of Sucrose Solutions Caused by

Passing Through a Resin Bed (163)
Effluent PR PR PR
(Bed Volumes) Case 1 Case 2 Case 3
Original 6.7 8.05 5.3
10 4.9 5.4 4.0
20 4.9 5.25 3.9
30 4.9 5.25 3.85
It was shown (164) that the Type I anion resins have

greater decolorizing ability then Type II resins. This is
shown in Table 3.29. It is equally important that the
regeneration efficiency shows the same preference for Type I
resins over Type II resins even after twenty cycles, although
there is a significant reduction in the about of adsorption
after just ten cycles, as shown in Table 3.30.
Table 3.29: Decolorization with Type I and Type II Anion at

Cycle 1 (164)
TYPe Weight of Weight of Weight of Ratio of
of Color In Color In Color Decolor-
Resin Resin Inf luent Effluent Adsorbed
- -ization
Aaberlite IRA 401 Type I 65.93 6.47 59.46 90.19
Duolite A-101 Type I 65.12 8.28 56.84 87.29
Duolite Type I 63.06 5.20 57.86 91.75
Duolite Type II 62.89 a.97 53.92 85.74
Duolite A-102D Type II 64.40 12.12 54.28 84.29

Table 3.30: Decolorization with Type I and Type II Anion Resins

at Cycle 1,lO and 20 (164)
Cycle 1 Cycle 10 Cycle 20
he weight Ratio of Weight Ratio of Weight Ratio of
Of of Color Decolor- of Color D.X!CkW- of Color DCC0lC.P
Resin Resin Adsorbed w Adsorbed iration Adsorbed ization
Amberlite IRA 401 Trpe 1 59.46 90.19 686.43 87.93 1515 65.6
Duolite A-101 Type I 56.84 67.29 669.9 86.19 1453 82.9
Duolitc A-IOID TyPe I 51.06 91.75 686.8 88.46 1496 83.8
Duolite A-102 Type II 53.92 65.14 606.6 79.40 1308 75.1
Duolite A-102D Type II 54.28 84.29 616.1 80.93 1333 75.8
Furukawa and Iizuka (165) have made an economic

analysis of the cost of decolorizing sugar syrups with anion
exchange resins. Other sugar decolorization studies using
anion exchange resins have been reviewed by Parker (166)
and Pollio and McGarvey (167).
3.8.5.2 Demineralization: In the beet and corn sugar

industries, some demineralization of the sugar solution is
normally required during the refining operation. In these
cases, the sugar solutions are usually passed through a cation
exchange resin bed in the hydrogen form followed by a weak
anion exchange bed. The cation resin will exchange the
cations in solution, primarily Cat+ and Nat, with Ht, forming
acids of the mineral salts. The weak base resin functions
both to remove this acidity generated by the cation exchange
resin and to adsorb colored organic species from the sugar
solution.
For sugar beets, a thin sugar solution (~5%) is extracted

from the beets in horizontal cylindrical diffusion units in
which the beet slices circulate in an opposite direction to
the hot water used to extract the sugar. This thin extract is
treated with lime in a process known as carbonation. This
results in a sugar solution that is about 9Ooh pure sugar.
However, the 10% impurities must also be removed since they
prevent the crystallization of an equivalent amount of sugar.
The overall process is shown in Figure 3.86 for

softening and demineralization of beet sugar. In the
“softening” of sugar juice with ion exchange resins, the
“hardening” salts are removed by conversion of Cat+ and
Mgtt salts into Nat salts, thus preventing the incrustation of
Ion Exchange
r I
I
I 1
the evaporators and pipelines. In the demineralization of

sugar juices, the occurrence of molasses is reduced by
removal of salts from the clarified sugar juice, thereby
increasing the sugar yield.
The starch from corn can be converted by acid or

enzyme hydrolysis into glucose syrup (dextrose content
between 30 and 68%) and into dextrose syrup (dextrose
content over 96%). The dextrose can be isomerized with the
enzyme glucose isomerase to produce a mixture of dextrose
and fructose. The product, known as high fructose corn
syrup (HFCS), contains 42% fructose and 55% dextrose.
It is necessary to have an ionically controlled

environment for the optimum productivity of the glucose
isomerase enzyme. Therefore, strong acid cation resin
column and weak base anion resin columns are used to
demineralize the dextrose solution prior to its passage
through the enzyme reactor. This is shown in Figure 3.87.
As the dextrose is introduced to the enzyme, the necessary
magnesium sulfate cofactor, 3 millimolar, is added.
The feedstream is at a 30 to 50% dextrose

concentration. This means that the viscous (3-6 CP)
feedstream must travel at a relatively slow rate (0.5 to 1.0
L/s-m3) to have maximum utilization of the ion exchange
resin. This flow sensitivity is due, no doubt, to the slower
diffusion rate even though macroporous resins are utilized.
The operating temperature is normally about 50°C during the
passage through the ion exchange columns and 60°C during
the enzyme reaction. Using higher operating temperatures in
the resin columns, particularly once the fructose is present,
causes the formation of colored impurities.
The four bed resin system (2 cation and 2 anion) allows

more complete utilization of the ion exchange capacity of
each column. The lead cation and anion columns perform the
bulk of the purification, while the second cation and anion
act as “polishing” units to prevent the leakage of even minor
amounts of ions of color impurities.
After a column is exhausted, it is necessary to remove

the syrup solution from the column during a “sweetening-off
portion of the operating cycle. This step not only allows the
recovery of the syrup, it also allows a gradual change in the
osmotic strength of the solution. The syrup recovered during
Ion Exchange 301
the “sweetening-off’ may be used during the “sweetening-on”

portion of the next cycle, which occurs just before the full-
strength syrup feedstream is introduced.
STARCH SLURRY
L I HEWATER
a AHYLASE ENZYtiE
FEED TANK 6 - 7 pH
STEAt4
1 SACCHAR~;CATION ]
(90 - 96% DEXTROSE, 0.8.)
SALTSG$OCpH ADJUSTMENT
60 - - 8.5
40 - 50% 06YpkIk:ANCE
ISOHERIZATION
REFINING
CONCENTRATING
FRUCTQSE SYRUYP
Figure 3.87. Generalized process scheme for corn starch

liquefaction, conversion to dextrose (saccharification) and
conversion to fructose syrup (isomerization).
The cation resin is normally regenerated with HCl

solutions since it is much more efficient than H,SO, as seen
in Figure 3.88. The amount of regenerant required is
typically 0.25 kg/L of resin. The anion resin may be
regenerated with caustic, soda ash or ammonia. Table 3.31
shows how the operating capacity of a typical weak base
resin depends on the regenerant used.
-92
078
H2S04 REGENERANT
,35
,28 I I I I 1
0 0,032 0,064 0,096 0,128 0,160
REGENERATION LEVEL (KG REGENERANT/LITER RESIN)
Figure 3.88. Operating capacity of a strong acid cation resin

(Dowex 88) as a function of regeneration level for different
regenerants.
Ion Exchange 303
Table 3.31: Operating Capacity of Dowex 66 Resin
Influent FlOW Regenerant Temper- Capacity

Exhaustant Specific Rate Dosage at”Ce
& Solids Level Acids (gpm/ft’) (lb/ft3) (OF) (kgr/ft3) (meqlml)
(1,000 ppm) NaOH
Water HCl/H,SO, 4 5 78 25.2 1.15
HCl/H,SO, 6 5 78 24.1 1.10
(250 ppm) NaOH
HCl/H,SO, 4 5 100 21.4 1.26
HCl/H,SO, 4 5 120 32.2 1.47
Dextrose Hc1/H,so, 4 3 120 29.2 1.34

Fructose
(35% HCl/H,SO, 4 5 140 35.5 1.63
Dissolved
Solids) tiCI/H,SO, 2 5 120 33.8 1.54
HCl/H,SO, 6 5 120 24.0 1.10
HCl 4 5 120 31.5 1.44
H,SO, 4 5 120 32.1 1.47
(250 ppmf Mi,ott
HCl/H,SO, 4 4.35 120 26.6 1.22
(7.50 ppm) Nd,CO,

HCl/H,SO, 4 6.6 120 27.7 1.27
3.8.6 Vitamins
When vitamins are obtained from natural sources or by
fermentation, recovery processes using ion exchange resins
have been found to be commercially desirable. The most
notable example is vitamin B- 12.
The isolation of vitamin B- 12 using a carboxylic acid

ion exchange resin was first described in 1953 (168). The
adsorption proceeds best at pH 3 to 6. After the resin has
been loaded with the vitamin, the resin-vitamin complex is
washed with a O.lN HCl solution. The vitamin may then be
eluted using an aqueous organic solution. Since vitamin B- 12
is a nonionic compound, the specific adsorption interactions

with the carboxylic acid resin do not involve the exchange of
ions. It is probably for this reason that less than 50
micrograms are adsorbed per gram of resin.
Recently porous phenol-formaldehyde cation resins have

been developed (146) which have a significantly greater
adsorption capacity for vitamin B-12. This is shown in Table
3.32 as a function of flowrate and in Figure 3.89 as a
function of the initial concentration of vitamin B-12 in the
feedstream. The elution was performed at 5 bed volumes per
hour using a l.OM HCl-60% aqueous acetone solution.
Table 3.32: Effect of Flow Rate on the Uptake of

Vitamin B-12 (146)
Volume Treated
Flow Rate Until Breakthrough Net Loading
(cm3/h) (Bed Volume) (mg/p, resin) Recovery (X)
50 349 1.12 86.14
100 313 1.54 85.43
150 280 1.38 81.24
200 240 1.18 88.30
0 200 400 600 600 1000 1200
Initial concentration ( mg dmm3 1
Figure 3.89. Dependence of the uptake of Vitamin B- 12 on

the initial concentration of the equilibrated solution on
Duolite S-761 (Curve A) and on a new synthetic resin (Curve
I?) (Reference 146).
Ion Exchange 305
3.8.7 Wine Treatment

All wines contain naturally occurring potassium
bitartrate, which is soluble in the original fermentation liquid
but which precipitates as the wine develops (169).
Classically, wines were matured for long periods and carefully
decanted by the user to remove the precipitate. The greater
consumption and the development of massive wine production
facilities has led to the use of ion exchange techniques to
overcome this problem. As Figure 3.90 shows, the wine was
passed through a strong cation resin in the sodium form
which converts the potassium bitartrate into the more soluble
sodium hydrogen tartrate.
WINE MUST
t- YEAST
b/
POMACE
REMOVAL
4
J
FERMENTATION
FORTIFICATION
CATION GELATINOUS LEES

EXCHANGE RESIN
&
FINING OR ADJUSTMENT
J
BOTTLING
Figure 3.90. Generalized process scheme for the treatment of

wine.
With the concern about sodium in prepared foods,

alternate techniques have been developed for reducing the
amount of potassium while, at the same time, insuring that
the acidity of the wine is at the proper balance. Such
process schemes are shown in Figure 3.91 (170-I 72).
CORNELL RESEARU-I FOLNDATIW

u.s, #4,205,m
Cl983
E. 8 J. ‘3LU3
U.S. #3,437,491
(1969)
FREEBASE KICM WVID

U.S. #436,a?6
Cl9791
Figure 3.91. Process schemes for acid and tartrate

adjustment of wine (References 170- 172).
White wines produced during a poor growing season with

inadequate sunshine tend to have an excess of acid which
detracts from the taste of the wines (169). The acidity may
be reduced to a lower level by passing the wine through a
weak base resin in the free base form. Since the natural
color and flavor ingredients are anionic in nature, the resin
should have low porosity and be highly crosslinked to avoid
removing these ingredients.
Ion Exchange 307
3.8.8 Enzyme Immobilization

It was in the 1950’s that ion exchange resins began to
be used seriously as carriers for immobilization of enzymes
(173,174). These initial efforts resulted in immobilized
enzymes that had very low activities of no commercial use.
The results were somewhat better (175) for enzymes
immobilized on cellulose derivatives. The macroporous resins
of the 1970’s allowed them to be used in commercial
immobilized enzyme processes (176). The bonding specificity
of the affinity chromatography type of resin with an enzyme
has been employed to immobilize enzymes (177). A review of
the use of these immobilized enzymes in the food industry
has been prepared by Kilara and Shahani (178).
The advantages of immobilized enzymes over soluble

enzymes are (179):
(1) Continuous processes become practical.

(2) The stability of the enzyme may be
improved.
(3) A product of a higher purity may be
possible.
(4) The sensitivity of the enzyme to changes
in temperature or pH may be decreased.
(5) The enzyme activity may be increased.
(6) Effluent problems and material handing
problems (separations) for feed and
effluent are reduced.
Table 3.33 shows the features of a commercial process,

mentioned in the same article, for the production of L-amino
acids using an immobilized aminoacylase enzyme. More
recently (180), the aminoacylase enzyme has been immobilized
on a porous strong base resin of trimethylammonium-
introduced porous silica with a pore size of about 1000 A and
a surface area of about 25 m2/g. Additional examples of
enzymes that have been immobilized on different carriers are
given in Table 3.34 (181).
Tschang and Klefenz (182) have developed a series of

macroporous crosslinked polystyrene resins which contain
isocyanate, thioisocyanate or aldehyde functionalities for
covalently binding proteins. These resins have capacities

between 3.4 and 4.1 meq/g. The amount of enzyme bound
and its activity are shown in Table 3.35.
Table 3.33: Commercial Process for the Production of

L-Amino Acids (179)
Column capacity 1000 liter
Carrier DEAE Sephadex
Amount of acylase bound : 333 I.U./ml
Activity of bound enzyme: 157 I.U./ml
Yield of enzyme activity: 47%
Operating temperature : 5ooc
Operating pH 7.0
Activity loss 40% in 35 days
Column regeneration once in 35 days
L-Methionine yield 715 kg/day @ SV = 2.0
Table 3.34: Commercially Available Immobilized Enzyme

Technology (181)
Enzyme Carrier Level of Development
Glucose isomerase DEAE cellulose, Alumina, Industrial

Dowex MWA-1
Glucoamylase Dowex 1 x 10 Pilot
Lactase DEAF cellulose, Alumina Industrial
Invertase Silica Gel, Alumina Pilot Scale
Penicillin amidase Polyacrylamide Gel Industrial
Aminoacylase Alumina, Diaion SA-11A Industrial
Aspartase Polyacrylamide Gel Industrial
Fumarase DEAE Cellulose Pilot
Hydantoinase DEAF Cellulose Industrial

Ion Exchange 309
Table 3.35: Resin-Bound Enzymes and Activities (182)
Amount
of Bound
Binding Hydrophilic EtlZylW
Functionality GrOUp Enzyme b&g) Activity
hexane-1,6-diamine/ -SO,Na (II) p-fructosidase 142 93%

glutarodialdehyde
hydrazine/hexane- -SO,Na (H) p-fructosidase 16.5 90%

1,6-diisocyanate -
hexane-1,6-diamine/ p-fructocidase 63 80%

glutarodialdehyde
hexane-1,6-diamine/ -SO,NH(CH,)$H, p-fructosidase __ ao”b

thiophosgene
ethylene diaminel -SO,N(Propjz P-fructosidase 31.9 62%

toluylene diisocyanate
4,9-dioxadodecane- -SO,NH(But) p-fructosidase 12.2 58%

1,12-dianlinel
glutarodialdehyde
Roy and Roy (183) have synthesized protein-compatible

weak anion silica exchangers. These resins have been used
to recover over 90% of the albumin and gamma globulin
consistently with virtually no non-specific adsorption.
However, even if kept in acidic media (pH 4 to 7), the useful
life of the resin is only about three months.
Delin and coworkers (177) described an illustrative

procedure for purifying and immobilizing penicillin acylase
which could then be used to produce hypoallergenic
penicillins from other penicillins. The penicillin acylase was
purified by adsorption on a porous cation exchange resin
from a solution adjusted to pH 4.0 - 5.0. The purified
enzyme is eluted with 0.2M ammonium acetate buffer solution
(pH 6.0 to 8.0). Sephadex G200 was treated with 5N NaOH
and then reacted with cyanogen chloride at a cold
temperature (0-3°C). After the reaction was complete, the
resin was washed with ice water and a O.lM borax solution.
The wet polymer was added to the solution of purified
penicillin acylase, more borax was added and the mixture
stirred slowly for 24 hours to complete the enzyme
immobilization.
Immobilizations such as this, which are due to the

formation of a covalent bond, are more complex than the
simple adsorption or ionic bonds which may be formed by
simply passing the enzyme solution through the prepared ion
exchange columns.
3.8.9 Catalytic Applications

As an early study showed (184), ion exchange resins can
be used as catalysts for a variety of organic reactions
including esterification, acetal synthesis, ester alcoholysis,
acetal alcoholysis, alcohol dehydration, ester hydrolysis and
sucrose inversion. Sucrose inversion was examined in detail
by Bodamer and Kunin (185). They showed the effect of
temperature, resin particle size, degree of crosslinking and
resin type. The strong acid (sulfonic) functionality was much
more effective than the weak (carboxylic) functionality. The
rate of inversion increases with decreasing particle size,
increasing porosity (lower crosslinking) and increasing
temperature. These results are shown in Table 3.36.
Table 3.36: Effect of Particle Size, Porosity and Temperature

on Sucrose Inversion (185)
Parameter Value Rate of Inversion (k x IO41
Particle Size 0.24 mm 34.7
0.45 mm 26.3
0.63 mm 15.7
Degree of Crosslinking 1% 199.2

(Porosity)
4% 110.3
10% 26.3
15% 3.0
20% 0.7
Temperature 25’C 0.7
5ooc 26.3
75oc 117.0
A more recent study (186) has taken another look at

the effect of these same parameters, the contact time of the
sugar solution with the resin and the hydrogen ion
concentration of the resin on sucrose inversion. The longer
contact time and the higher hydrogen ion concentration
caused greater rates of sucrose inversion. A temperature
Ion Exchange 311
increase of only 5°C doubled the inversion rate. As the

crosslinking was increased (porosity decreased), the rate of
inversion decreased.
A similar catalytic inversion of lactose from whey into

glucose and galactose has been demonstrated (187). The
lactose solution was processed at a temperature of 95°C and
a flow rate of 0.27 bed volumes/hour to give 80 to 90%
conversion. It was necessary to remove the protein and ash
from the lactose solution before processing to insure
continued catalytic action by the cation resin. The
concentration of lactose in the feedstream was determined to
have a technical and economic optimum at 100 g/L. A
column operation has also been developed by Holmberg (188).
Takahashi (189) combined 0.4% invertase enzyme with a

strong acid cation resin to invert at 60°C a 5O”Brix molasses
solution. The contact time required was five hours. There
was a 90% recovery of the sugars with 100% conversion to
glucose and fructose, The color quality, however, was still
that of raw sugar.
A strong resin has also been used to convert fructose

into 5-hydroxymethyl-2-furancarboxaldehyde (190). This
reaction required the presence of an organic solvent such as
methyl isobutyl ketone and temperature between 80” and
90°C. The degrees of conversion for specific resins and
times of reaction are shown in Table 3.37.
Table 3.37: Synthesis of 5-Hydroxymethyl-2-Furancarboxaldehyde

from Fructose with Cation Resins (190)
Ion Exchange ReacLion Conversion @

0Iz
Resin
.-___ Time (hour) Rates (%I Yield (%I
Levatlt SCIOZ 4 34 47
Amberlitc IR118 4 3s 58
Duolite C26 4 25 54
Amberlitc AZOOC 4 13 42
Amberlysl A15 4 14 30
LcwaLit SK118 4 11 38
Lewatit SPCllB 15 56 66
Lcwat_it spclla 24 46 51
@
1h.econversion rate is the ratio of the number of moles of obtained product
to the number of moles of fructose introduced.
@ The yield is the ratio of the nwnber of moles of ANF produced to the number
of moles of fructose consumed.
APPENDIX
Ion exchange resins are prepared by a number of

manufacturers. During recent years there has been some
consolidation among the producers of synthetic ion exchange
resins. Rohm and Haas has purchased Duolite resins from
Diamond Shamrock and Dow has purchased the Montecatini-
Edison resin operation. These consolidations have caused
some changes in designations for the resins now commercially
available. The following tables include historical as well as
current designations so that past literature references to
specific commercial resins might be understood. It is
recommended that the reader contact the resin manufacturer
to learn about currently commercially available resins and
their properties.
Table 3A.l: Ion Exchange Resins from U.S. Manufacturers
. . . . . . . . . . . . . . . . . Cation Exchange Resins. . . . . . . . . . . . . . . . . .
Moisture
Content Percent
61 DVB Screen Mesh Trade Name
. . . . . . . . . . ...*..... I. Strong Acid Type . . . . . . . . . . . . . . . . . . .
a. Polystyrene, gel, -SOsH
75-85 2 20-50 Dowex 5OWX2, BioRad
AG-40WX2
50-100 Dowex 5OWX2, BioRad
AG-50WX2
AG-50WX2
65-75 4 20-50 Dowex 50X4, BioRad
AG-50WX4
40-80 Dowex 50X4, BioRad
AG-50WX4
5owx4
AG-50WX4
AG-50WX4
55-60 6 16-50 Duolite C-25D, Amberlite
XE-100
(continued)
Ion Exchange 313
Table 3A.l: (continued)
Moisture
Content Percent
(%) DVB Screen Mesh Trade Name
50-55 8 20-50 Dowex 5OWX8, Dowex
HCR AG-50WX8,
BioRex RG-50WX8
50-55 8 40-80 BioRad AG 5OWX8
AG-50WX8
loo-230 Dowex 5OWX8, BioRad
AG-50WX8
230-400 BioRad AC-50WX8
45-50 10 20-50 Dowex HGR, Dowex
5OWX10, Amberlite IR-
120, Amberlite IR-121,
Amberlite IR-122
100-500 BioRad AG-50WX10,
Ionac C-240, Ionac
C-249, Ionac C-253,
Ionac C-257, Ionac
Ci-295, Amberlite
IR-120
<325 Amberlite IR-120
40-45 12 20-50 BioRad AG-5OWX12
40-80 BioRad AG-50WX12
AG-5OWX12
AG-50WX12
230-400 BioRad AG-50WX12
35-40 16 20-50 Amberlite IR-124, Ionac
C-250, Ionac C-258,
Ionac C-255, Dowex
5OWX16
b. Polystyrene, macroporous -SOsH
46-50 16-50 Amberlite 200, Dowex 88
Amberlite 2OOC, Dowex
MSC-1
Amberlite 252
c. Sulfonated coal
14-20 16-50 Ionac C-150
d. Cellulose or dextran, -OG&SOsH
89-90 120-400 SE-Sephadex C-25
96-98 SE-Sephadex C-50,
Cellex SE
(continued)
Moisture
Content Percent
(%) DVB Screen Mesh Trade Name
e. Phenolic, -CH,SOsH
- 20-50 BioRex 40, Duolite C-l
- 50-100 BioRex 40
- 100-200 BioRex 40
- 200-400 BioRex 40
. . . . . . . . . . . . . . . . II. Intermediate Acid Strength. .. . . .. . .. . .. . . ..

a. Polystyrene, phosphoric granular
- 68-76 BioRex 63
b. Cellulose, phosphoric
- - Cellex P
Polymer Particle
Type Type Screen Mesh Trade Name
. . . . . . . . . . . . . . . . . . . .III. Weak Acid Type . .. . . . . . . . . . . . . . . . .
a. Methacrylic acid-divinylbenzene, -COOH
pK -6.0 Spherical 20-50 Amberlite IRC-50
Granular 100-500 Amberlite IRP-64
Granular <325 Amberlite IRP-64M
b. Acrylic acid-divinylbenzene, -COOH
pK-5 Spherical 20-50 Amberlite IRC-84
Dowex CCR-2, Duolite
ES-80, BioRex 70
Granular 50-100 BioRex 70
100-200 BioRex 70
200-400 BioRex 70
<400 BioRex 70
c. Miscellaneous weak acid resins
Condensation Granules 16-50 Ionac C-265
Dextran,
-OCH,COOH Spherical 120-400 Sephadex C-25
120-400 Sephadex C-50
Cellulose,
-OC&COOH Fibrous - Cellex CM, Cellulose CM-
22, Cellulose CM-23
200-400 Cellulose CM-32, Cellu-
lose CM-52
(continued)
Ion Exchange 315
Table 3A. 1: (continued)
. . . . . . . . . . . ...*.. Anion Exchange Resins. . . . . . . . . .. . . .. . . .
Cross-Link Moisture Particle

@) 6) Type Screen Mesh Trade Name
. . . . . . . . . . . . . . . . . . . . . . I. Strong Base . . . . . . . . . . . . . . . . . . . . . .
a. Polystyrene-Type I, C&N(CH,),
2 70-80 20-60 BioRad AG-1X2, Dowex
1x2
60-140 BioRad AG-1X2, Dowex
1x2
1x2
1x2
1X4, Amberlite IRA-
900 (macroporous),
Amberlite IRA-938-C
(large pore), Duolite
ES-111, Amberlite A-26
(macroporous), Amber-
lite IRA-401s
4 60-70 Spherical 40-80 BioRad AG-1X4, Dowex
1x4
1x4
loo-230 BioRad AG-1X4
230-325 BioRad AC-lx4
- 50-60 16-20 BioRad AG-21K, Dowex
21K, Amberlite IRA-
401, Ionac A-540,
Ionac A-544, Ionac A-
548, Duolite AlOlD
20-40 BioRad AG-21K, Dowex
21K, Dowex MSA-1
(macroporous)
(continued)

(%) 6) Qpe Screen Mesh Trade Name

1X8, Dowex SBR,
Dowex 11, Ionac A-546,
Ionac A-000, Amberlite
IRA-400, Amberlite
IRA-400C
1X8
1X8
- 140-325 BioRad AG-1X8, Dowex
1X8
- <230 BioRad AG-1X8, Dowex
1X8
- 40-50 Granular 100-500 Amberlite IRP-67
<325 Amberlite IRP-67M
10 35-40 Spherical 40-80 BioRad AG-1X10
60-140 BioRad AG-1X10
<230 BioRad AG-1X10
(C&k
b. Polystlrr--le-Type II, CH,-h’
‘C&CH,OH
2X4, Amberlite IRA-
910 (macroporous)
53-60 20-50 Dowex MSA-2 (macro-
porous)
47-50 16-50 Ionac A-550, Ionac A-5
7 40-45 20-50 Amberlite IRA-410
SAR
40-80 BioRad AG-2X8
10 28-36 Spherical 40-80 BioRad AG-2X10
(continued)
Ion Exchange 317

(%) (%) Type Screen Mesh Trade Name
c. Polystyrene, CH,N
3
- 46-54 Spherical 16-50 BioRex 9
100-200 BioRex 9
200-325 BioRex 9
- - Spherical 16-50 Ionac A-580
10-20 Ionac A-590
d. Dextran
- - Spherical 140-400 QAE Sephadex A-25
140-400 QAE Sephadex A-50
. . II. Weak and Intermediate Base Strength, Quaternary and Ternary. . .. .

- 60-66 Spherical 16-50 Duolite A-57
- 50-60 Spherical 20-50 BioRex 5
100-200 BioRex 5
200-400 BioRex 5
- - 16-50 Ionac A-300, Ionac A-302
. . . . . . . . . . . . . . . . . .III. Weak Base . . . . . . . . . . . . . . . . . , . . . .

a. Polystyrene
- 50-60 Spherical 20-50 Dowex 66, Dowex MWA-
1 (macroporous)
- 46-55 Spherical 20-50 Amberlite IRA-93 (mac-
roporous)
- 40-45 Spherical 20-50 Amberlite IRA-45,
Amberlite IRA-21
- 25-35 16-40 BioRad AG-3X4
b. Acrylic
- 57-63 Spherical 20-50 Amberlite IRA-68
c. Phenolic-polyamine
- - Granular 100-500 Amberlite IRP-58
<325 Amberlite IRP-58M
d. Condensation polymer
- - Granular 16-50 Amberlite IRA-401,
Dowex WGR, Dowex
WGR-2, Ionac A-260
(continued)

(%) (% Type Screen Mesh Trade Name
e. Dextran and Cellulose, DEAE-OC&C&N(G&,h

- - Spherical 160-400 DEAE Sephadex A-25
140-400 DEAE Sephadex A-50
- - Fibrous - Cellex D, Cellulose DE-
22, Cellulose DE-23
- - Spherical 100-200 Biogel DM-2
100-200 Biogel DM-30, high cap.
100-200 Biogel DM-30, low cap.
100-200 Biogel DM-100, high cap.
200-400 Cellulose DE-32, Cellu-
lose DE-52
Resin Trade Name Manufacturer
Dowex Dow Chemical Co., Midland, MI

Amberlite, Duolite Rohm and Haas Co., Philadelphia, PA
Ionac Ionac Chemical Co., Birmingham, NJ
Sephadex Pharmacia Fine Chemicals, Piscataway, NJ
Cellex, BioRad, BioRex,
Biogel BioRad Labs, Richmond, CA
Table 3A.2: Foreign Producers of Synthetic Ion Exchange Resins
Company Country Trade Name

Bayer Federal Republic of Germany Lewatit
Chemolimfex Hungary Varion
Mitsubishi Japan Diaion
Ostion Czechoslovakia Ostion
Permu tit United Kingdom Zeocarb,
Deacidite,
Zerolit
Permutit, A.G. Federal Republic of Germany Orzelith,
Permu tit
Resindion Italy Relite
Wolfen German Democratic Republic Wofatit
USSR AW-, AV-, KB-,
KU-
Ion Exchange 319
Table 3A.3: Corresponding Foreign and U.S. Ion Exchange Resins
U.S. Resin Foreign Resins

Dowex 50 W X8 Diaion SK-1B
Kationit KU 2
Lewatit S-100
Varion KS
Wofatit KPS 200
Dowex CCR-1 Diaion WK-10
Wofatit CP 300
Dowex MSCl Diaion PK-228
Lewatit SP-100
Dowex MWC-1 Diaion WK-20
Lewatit CNP
Dowex SBR Diaion 10A
Lewatit M 500
Dowex SAR Diaion 20A
Lewatit M 600
Dowex MSA-1 Diaion PA 316
Lewatit MP 500
Wofatit SZ-30
Dowex MSA-2 Diaion PA 416
Lewatit MP 600
Dowex MWA-1 Diaion WA-30
Lewatit MP-62
Table 3A.4: Ion Exchange Process Designers and Manufacturers
Company Location
Aqua Media Sunnyvale, CA
Bateman Uranium Corporation Lakewood, CO
Belco Pollution Control Corp. Parsippany, NJ
Chemical Separations Corp. Oak Ridge, TN
Chem Nuclear Systems Columbia, SC
Cochrane Division of Crane Corp. King of Prussia, PA
Downey Welding & Manufacturing Co. Downey, CA
Envirex Water Conditioning Div. Waukeska, WI
Epicor Linden, NJ
Ermco, Inc. St. Louis, MO
Graver, Division of Ecodyne Union, NJ
Hlmsley Toronto, Canada
Hittman Nuclear & Development Columbia, MD
(continued)
Table 3A.4: (continued)
Company Location
Hungerford & Terry, Inc. Clayton, NJ
Hydro-Max Corporation Milwaukee, WI
Illinois Water Treatment Rockford, IL
Industrial Filter and Pump Cicero, IL
Infilco Degremont, Inc. Richmond, VA
Intensa Mexico City, Mexico
Kinetico, Inc. Newbury, OH
Kurita Japan
L.A. Water Treatment City of Industry, CA
Liquitech, Div. of Thermotics, Inc. Houston, TX
Mannesmann Federal Republic of Germany
Mltco Water Labs, Inc. Winter Haven, FL
Morgan Company Houston, TX
*Dermutit, Division of Zum Paramus, NJ
Rock Valley Water Conditioning Rockford, IL
Smith, F.F. & Associates Houston, TX
Technichem, Inc. Belvidere, IL
United States Filter, Fluid System Corp. Whittier, CA
Wynhausen Water Softener Company Los Angeles, CA
3.9 REFERENCES
1. Thompson, H.S., J Rov APr Sot. Eng, II:68 (1850)
2. Bersin, T., Naturwissenschaften, 33: 108 (1946)
3. Winters, J.C., Kunin, R., Ind Ene, Chem, 41:460 (1949)
4. Polis, B.D., Meyerhoff, O., J Biol Chem, 169:389 (1947)
5. Carson, J.F., Maclay, W.D., J Am Chem Sot, 67:1808

(1945)
6. Nagai, S., Murakami, J Sot Chem Ind. Japan, 44:709

(1941)
7. Wieland, T., Berichte, 77:539 (1944)
8. Bergdoll, M.S., Doty, D.M., Ind Eng Chem. Arc1 Ed,

18:600 (1946)
9. Lejwa, A., Biochem Z, 256:236 (1939)

Ion Exchange 321
10. Kingsburg, A.D., Mindler, A.B., Gilwood, M.B., Chem Eng

m, 44:497 (1948)
11. Coke Cruz, E., Gonzales, F., Hulsen, W., Science, 101:340
(1945)
12. Jackson, W.G., Whitefield, G., DeVries, W., et al., J Am

Chem Sot, 73:337 (1951)
13. Moore, S., Stein, W.H., J Biol Chem, 211:893 (1954)
14. Starobinietz, G.L., Gleim, J.F., Zh Fiz Khim, 39:2189

(1965)
15. Kunin, R., Ion Exchange Resins, Robert E. Krieger Publ.

Co., Huntington, NY, p 26, (1972)
16. Wiklander, L., Ann Rov Anr Co11 Sweden, 14:l (1946)
17. Nachod, F.C., Wood, W., J Am Chem Sot, 66:1380 (1944)
18. Boyd, G.E., Schubert, J., Adamson, A.W., J Am Chem

a, 69:2818 (1947)
19. Harned, H.S., Owen, B.B., The Phvsical Chemistrv of

Electrolvtic Solutions, Reinhold Publishing Corp, (1943)
20. Gregor, H.P., J Am Chem Sot, 70:1293 (1948)
21. Argersinger, W., Davidson, A., Bonner, O., Trans Kansas

Acad Sci, 53:404 (1950)
22. Reichenberg, D., Pepper, K., McCauley, D., J Chem Sot,

p 493 (1951)
23. Savitskaya, E.M., Yakhontova, L.F., Nys, P.S., Pure &

Appl Chem, 54( 11):2169 (1982)
24. Wolf, F.J., Separation Methods in Organic Chemistrv

and Biochemistrv, Academic Press, NY, p 144 (1969)
25. Peterson, S., Ann N Y Acad Sci, 57:144 (1953)
26. Ito, T., Kino Zairvo, 2(5):8 (1983)

27. Savitskaya, E.M., Yakontova, L.F., Nys, P.S., Ion

Exchange, Nauka, Moscow, p 229 (1981)
28. Anderson, R.E., A I Ch E Svmp Ser, 71( 152):236 (1979)
29. Klein, G., A I Ch E Svmp Ser, 81(242):28 (1985)
30. NATO Advanced Study Institute on Mass Transfer and

Kinetics of Ion Exchange, Maratea, Italy (May 31-July
11, 1982)
31. Helfferich, F.G., Plesset, M.S., J Chem Phvs, 28:418

(1958)
32. Cantwell, F.F., Ion Exchange Solvent Extrac, 9:339

(1985)
33. Helfferich, F.G., J Phvs Chem, 66:39 (1985)
34. Helfferich, F.G., In: Mass Transfer and Kinetics of Ion

Exchange, (Liberti, L., Helfferich, F.G., eds) Martinus
Nijhoff Publ., The Hague, p 157 (1983)
35. Schogl, R., Helfferich, F.G., J Chem Phvs, 26:5 (1957)
36. Glueckauf, E., Coates, J.I., J Chem Sot (London), 13 15

(1947)
37. Helfferich, F.G., J Phvs Chem, 69:1178 (1965)
38. Helfferich, F.G., Liberti, L., Petruzzelli, D., et al., Israel

J of Chem, 26:3 (1985)
39. Tsai, F.N., J Phvs Chem, 86:2339 (1982)
40. Vermeulen, T., Klein, G., Hiester, N.K., In: Perry’s

Chemical Engineers Handbook, (Perry, R.H., Chilton,
C.H., eds) McGraw-Hill, NY, Section 16 (1973)
41. Helfferich, F.G., Ion Exchanae, McGraw-Hill, NY, p 255

(1962)
42. Samsonov, G.V., Elkin, G.E., Ion Exchanpe and Solv

Extract, 9:211 (1985)
Ion Exchange 323
43. Huang, T.C., Tsai, F.N., Proc Pat Chem Enn Cone. 3rd
(Seoul), 1:339 (1983)
44. Rosen, J.B., Ind Ena Chem, 46:1590 (1954)
45. Klein, G., In: Mass Transfer and Kinetics of Ion

Exchange (Liberti, L., Helfferich, F.G., eds) Martinus
46. Mathews, A.P., Weber, W.J., Jr., In: Adsorption and Ion
Exchange with Synthetic Zeolites (Flank, W.H., ed) ACS
Society, Washington, p 27 (1980)
47. Barba, D., Del Re, G., Foscolo, P.U., The Chem Ena J,
26:33 (1983)
48. Samuelson, O., Ion Exchangers in Analvtical Chemistry,

John Wiley, NY, p 52 (1953)
49. Friedlander, S.K., A I Ch E J, 3:381 (1957)
50. Marchello, J.M., Davis, M.W., Jr., I & EC Fund, 2( I):27

(1963)
51. Clearfield, A., Nanocollas, G.H., Blessing, R.H., Ion

Exchange and Solvent Extraction (Marinsky and Marcus,
eds), 5:l (1973)
52. Millar, J.R., In: Mass Transfer and Kinetics of Ion

Exchange (Liberti, L., Helfferich, F.G., eds.) Martinus
53. Daniels, S.L., In: Adsorption of Microorganisms to

Surfaces (Britton, G., Marshall, K.C., eds.), John Wiley
and Sons, NY (1980)
54. Gold, H., Calmon, C., A I Ch E Svmp Ser, 76(192):60

(1980)
55. Warshawsky, A., Die Anaewandte Makromol Chem,

109/l lo:171 (1982)
56. Yotsumoto, K., Hinoura, M., Goto, M., Ger Offen

2,718,649 (1977)
57. Warshawsky, A., Fridkin, M., Stern, M., J Polvm Sci

Chem 20(6): 1469 (1982)
-,
58. Freeman, D.H., Schram, S.B., Anal Chem, 53:1235 (1981)
59. Saldadze, KM., Brutskus, T.K., Teploenernetika, 23(9):6

(1976)
60. Millar, J.R., Smith, D.G., Merr, W.E., et al., J Chem Sot,
p 183 (1963)
61. Mindick, M., Svarz, J., U.S. Patent No 3,549,562 (Dec.

22, 1970)
62. Ikeda, A., Imamura, K., Miyaka, T., et al., U.S. Patent
No. 4,093,570 (1978)
63. Miyake, T., Kunikiko, T., A., Ikeda, et al., US. Patent
No. 4,154,917 (1979)
64. Abe, T., Ikeda, A., Sakurai, T., Jap. Patent No. 79-l 1,088
(Jan. 26, 1979)
65. Millar, J.R., In: Mass Transfer and Kinetics of Ion

Exchanne (Liberti, L., Helfferich, F.G., eds.) Martinus
66. Messing, R.A., Immobilized Enzvmes for Industrial

Reactors, Academic Press, NY (1975)
67. Okuda, T., Awataguchi, S., U.S. Patent No. 3,718,742

(Feb. 27, 1973)
68. Crispin, T., Halasz, I., J Chromatopr, 239:351 (1982)
69. Martinola, F., Meyer, A., Ion Exchange and Membranes,

2:111 (1975)
70. Puente, A., Microbial Esoan, 14:209 (1961)
71. Ishida, M., Sugita, Y., Hori, T., et al., U.S. Patent No.
3,565,951 (Feb. 23, 1971)
72. Gordienko, S.V., J Appl Chem USSR, 39:lO (1966)

Ion Exchange 325
73. Shirato, S., Miyazaki, Y., Suzuki, I., Fermentation

Industrv (Japan), 45(1):60 (1967)
74. Wolf, F.J., Separation Methods in OrPanic Chemistry and

Bicohemistrv, Academic Press, NY, p 153 (1969)
75. Nagase, T., Hirohare, H., Nabeshima, S., Brit. Patent No.
1.597,436 (Sept. 6, 1981)
76. Commerical inert resins for this application include

Dowex Buffer Beads (XFS-43179), Ambersep 359 and
Duolite S-3TR.
77. Higgins, I.R., U.S. Patent No. 3,580,842 (May 25, 1971)
78. Himsley, A., Canadian Patent No. 980,467 (Dec. 23, 1975)
79. Brown, H., U.S. Patent No. 3,549,526 (Dec. 22, 1970)
80. Cloete, F.L.D., Streat, M., U.S. Patent No. 3,551,118

(Dec. 29, 1970)
81. Streat, M., J Separ Proc Technol, l(3):lO (1980)
82. Mallon, C., Richter, M., ZfI-Mitteilunnen. Leipzig, 86:39

(1984)
83. Cloete, F.L.D., Streat, M., British Patent No. 1,070,251

(1962)
84. George, D.R., Ross, J.R., Prater, J.D., Min Enain, I:73
(1968)
85. Buijs, A., Wesselingh, J.A., Polvtech Ti idschrift

Procestechniek, 36(2):70 (1981)
86. Naden, D., Willey, G., Bicker, E., et al., Ion Exchange
Technolonv (Nader, D., Streat, M., eds.) Horwood,
Chichester, UK, p 690 (1984)
87. Naden, D., Bandy, M.R., presented at SC1 meeting,

Impact of SX and IX on Hvdrometallurav, University of
Salford, UK (March, 1978)
88. Slater, M.J., Canadian J Chem Enq, 52:43 (Feb. 1974)

89. Chase, H.A., Ion Exchange Technoloq (Nader, D.,

Streat, M., eds.) Horwood, Chichester, UK (1984)
90. Wolf, F.J., Putter, I., Downing, G.V., Jr., et al., U.S.
Patent No. 3,221,008 (Nov. 30, 1965)
91. Flemming, H.C., Vom Wasser, 56:215 (1981)
92. Spinner, I.H., Simmons, P.J., Brown, C.J., Proceed 40th

International Water Conf, Pittsburgh (1979)
93. PowdexR is a registered tradename of the Graver Water

Treatment Company.
94. Calmon, C., A I Ch E Svmp Ser, 80(233):84 (1984)
95. Rodrigues, A.E., In: Mass Transfer and Kinetics of Ion

Exchange (Liberti, L., Helfferich, F.G., eds.) Marinus
Nijhoff, The Hague, p 259 (1983)
96. Passino, R., In: Mass Transfer and Kinetics of Ion

Exchange (Liberti, L., Helfferich, F.G., eds.) Marinus
Nijhoff, The Hague, p 313 (1983)
97. Boari, G., Liberti, L., Merli, C., et al., Env Prot Enq,
6:251 (1980)
98. Liberti, L., Passino, R., Ind Enn Chem. Pros Des Dev,
21(2):197 (1982)
99. Belter, P.A., Cunningham, F.L., Chen, J.W., Biotech and

Bioeng, 15:533 (1973)
100. Slater, M.J., Effluent Water Treat J, 461 (Oct. 1981)
101. Snowdon, C., Turner, J., Proc Int Svmp “Fluidization”,

Netherlands University Press, Amsterdam, p 599 (1967)
102. Ergun, S., Chem Ene; Proar, 48:89 (1952)
103. Tan, H.K.S., Chem Enq, 91(26):57 (Dec. 24, 1984)
104. Klein, G., Computers and Chem Eng, 8(3/4):171 (1984)
105. Klein, G., Nassiri, M., Vislocky, J.M., A I Ch E Svmp

&-, 80(233):14 (1984)
Ion Exchange 327
106. Erskine, D.G., Schuliger, W.G., Chem Enn Proar, 67:11,41

(1971)
107. Weber, W.J., Jr., Thaler, J.O., In: Scale-Up Water,

Wastewater Treatment Processes (Schmidthe, N.W., et
al., eds.) Butterworth, Boston, p 233 (1983)
108. Pelosi, P., McCarthy, J., Chem Eng, p 125 (Sept. 6,

1982)
109. Wirth, L.F., Jr., Feldt, C.A., Odland, K., Ind Ena Chem,
58:639 (1961)
110. Bolto, B.A., Eldridge, R.J., Eppringer, K.H., et al.,

Reactive Polvmers, 2:5 (1984)
111. Kysela, J., Brabec, J., Jad Enerq, 27(12):445 (1981)
112. Khodyrev, B.N., Prokhorov, A.F., Teoloenernetika,

25(3):60 (1978)
113. Stein, W.H., Moore, S., Sci American, 184:35 (1.951)
114. Kunin, R., Ion Exchange Resins, Robert E. Krieger Publ.

Co., Huntington, NY, p 290 (1972)
115. Yamada, K., Kinoshita, S., Tsunda, T., et al., The

Microbial Production of Amino Acids, John Wiley and
Sons, NY, p 228 (1972)
116. Nakayama, K., Hagino, H., U.S. Patent NO. 3,594,279

(July 20, 1971)
117. Noguchi, Y., Nakajima, J., Uno, T., et al., U.S. Patent
No. 3,684,655 (Aug. 15, 1972)
118. Okumura, S., Yoshinaga, F., Kubota, K., et al., U.S.

Patent No. 3,716,453 (Feb. 13, 1973)
119. Ishida, M., Sugita, Y., Hori, T., et al., U.S. Patent No.
3,565,951 (Feb. 23, 1971)
120. Nakayama, K., Hagino, H., U.S. Patent No. 3,595,751

(July 27, 1971)
121. Abe, S., Takayama, K., U.S. Patent No. 3,700,556 (Oct.
24, 1972)
122. Nakayama, K., Kase, H., U.S. Patent No. 3,692,628 (Sept.
19, 1972)
123. Nagai, T., Yamaura, I., U.S. Patent No. 3,639,467 (Feb.
1, 1972)
124. Noguchi, L., Maruyama, H., Motoya, Y., et al., U.K.

Patent No. 2,095,232 (Sept. 29, 1982)
125. Bartels, C.R., Kleiman, G., Korzun, J.N., et al., Chem

Ena Proq, 54(8):49 (Aug. 1958)
126. Wolf, F.J., U.S. Patent No. 3,000,873(Sept. 19, 1961)
127. Belter, P.A., A I Ch E Sump Ser, 80(233):110 (1984)
128. Denkewalter, R.G., Kazal, L.A., In: Ion Exchange

Technolony (Nachod, F.C., Schubert, J., eds.) Academic
Press, NY, p 579 (1956)
129. Voser, W., U.S. Patent No. 3,725,400(April3, 1973)
130. Fujisawa, Japanese Patent No. 77 128 294 (1977)
131. Yamanouchi, Japanese Patent No. 7911298 (1979)
132. Abraham, E.P., Newton, G.G.F., Hale, C.W., U.S. Patent

No. 3,184,454 (May 1965)
133. McCormick, U.S. Patent No. 3,467,654 (Sept. 1969)
134. Stables, H.C., Briggs, K., U.S. Patent No. 4,205,165 (May
27, 1980).
135. Goegelman, R.T., Miller, T.W., U.S. patent No. 3,709,880

(Jan. 9, 1973)
136. Schubert, P.F., U.S. Patent No. 4,196,285 (April 1, 1980)
137. Kawamura, Y., Shoji, J., Matsumoto, K., US. Patent No.
4,256,835 (March 17, 1981)
Ion Exchange 329
138. Fujita, S., Takatsu, A., Shibuya, K., U.S. Patent No.
3,629,233 (Dec. 21, 1971)
139. Rakutina, N.S., Shapenyuk, T.A., El’kin, G.E., et al., &

Prikl Khim (Leningrad), 55(5): 117 1 (1982)
140. Rakutina, N.S., Borisova, V.A., Papukova, K.P., et al., Zh

Prikl Khim (Leninarad), 55(3):540 (1982)
141. Huber, G., Schindler, P., European Patent No. 80,166

(June 1, 1983)
142. Treiber, L.R., Gullo, V.P., U.S. Patent No. 4,198,338

(April 15, 1980)
143. Geogelman, R.T., Kahan, F.M., U.S. Patent No. 4,000,161

(Dec. 28, 1976)
144. Debono, M., Merkel, K.E., Weeks, R.E., et al., U.S.

Patent No. 4,375,513 (March 1, 1983)
145. Debono, M., Merkel, K.E., Weeks, R.E., et al., U.S.

Patent No. 4,322,406 (March 30, 1982)
146. Khose, N.M., Dasare, B.D., I Chem Tech Biotechnol,

31:219 (1987)
147. Olthoff, U., Pharmazie, 32(8-9):536 (1977)
148. Jacob, T.A., U.S. Patent No. 3,366,627 (Jan. 30, 1968)
149. Tsao, J.C.Y., Huang, T., Wang, P., et al., J Chinese

Chem Sot (Peking) Ser II, 2(3) (1964)
150. Kobayski, T., Nakamura, I., Nakagawa, M., U.S. Patent

No. 3,873,425 (March 25, 1975)
151. Girot, P., Boschetti, E., J Chrom, 213:389 (1981)
152. Keay, L., U.S. Patent No. 3,593,738 (July 13, 1971)
153. Sipos, T., Viebrock, F.W., U.S. Patent No. 3,627,643

(Dec. 14, 1971)
154. Rauenbusch, E., Schmidt, D., U.S. Patent No. 4,174,439

(Nov 13, 1979)
155. Umezawa, H., Aoyagi, T., Takeuchi, T., et al., U.S.

Patent No. 4,167,448 (Sept. 11, 1979)
156. Kiselev, G.V., Biochem Bioohvs Acta, 712:719 (1982)
157. Palmer, D.E., In: Food Proteins (Fox, P.F., Condon, J.J.,
eds.) Applied Science, London, p 341 (1982)
158. Phillips, D.J., Jones, D.T., Palmer, D.E., British Patent

No. 1,518,111 (Feb. 25, 1981)
159. Harmon, Z., Dechow, F.J., U.S. Patent No. 4,543,261

(Sept. 24, 1985)
160. Kunin, R., Ion Exchanne Resins, Robert E. Kreiger

Publish. Co., Huntington, NY, p 183 (1972)
161. Obara, K., Proc Res Sot Japan SuPar Ref Tech, 4(4):8
(1955)
162. Sasaki, K., Naito, T., Tsukui, K., Seito Giiutsu Kenkvu
Kaishi, 28:39 (1978)
163. Susuki, K., Seito Giiutsu Kenkvu Kaishi, 9:73 (1960)
164. Iwashina, S., Proc Res Sot Japan Sugar Ref Technol,
9:113 (1960)
165. Furukawa, N., Iizuka, T., Proc Res Sot Jaoan Sunar Ref
Technol, 11:63 (1962)
166. Parker, K.J., Chem Indus, 21:782 (Oct. 1972)
167. Pollio, F.Y., McGarvey, F.X., Suaar Y Azucar, 63 (May

1978)
168. Shive, W., U.S. Patent No., 2,628,186 (Feb 10, 1953)
169. Arden, T.V., Rowe, MC., Process Biochem, 4 (April 1977)
170. Mattick, L.r., Gogel, E.V., U.S. Patent No. 4,205,092

(May 27, 1980)
171. Peterson, R.G., Fujii, G.R., U.S. Patent No. 3,437,491

(April 8, 1969)
Ion Exchange 331
172. Gogel, E.V., U.S. Patent No. 4,156,026 (May 22, 1979)
173. Brandenberger, H., Rev Fermentation et Indus Aliment,

11:237 (1956)
174. Barnett, L.B., Bull, H.B., Biochim Biophvs Acta, 36:244

(1959)
175. Tosa, T., Mori, T., Fuse, N., et al, Enzvmologia, 3 1:214
(1966)
176. Fujita, Y., Matsumoto, A., Nishikaji, T., et al., Japan

Patent No. 50-9417 (July 26, 1975)
177. Delin, P.S., Ekstrom, B.A., Nathorst-Westfield, L.S., et

al,, U.S. Patent No., 3,736,230 (May 29, 1973)
178. Kilara, A., Shahani, K., CRC Critical Reviews in Food

Sciences and Nutrition, 12(2):161 (1979)
179. Vieth, W.R., Venkatasubramanian, K., Chem Tech, p 677

(Nov 1973)
180. Chibata, I., Tosa, T., Mori, T., et al., Brit. Patent No.
2,082,188 (March 3, 1982)
18 1. Sweigart, R.D., In: Applied Biochem Bioeng (Wingard,

L.B., Jr., Katchalski-Katzir, E., Goldstein, L.G., eds.)
Academic Press, NY, p 209 (1979)
182. Tschang, C.J., Klefenz, H., U.S. Patent No. 4,266,030

(May 5, 1981)
183. Roy, A.K., Roy, S., LC. Lia Chromatonr HPLC Maq,
1(3):182 (1983)
184. Sussman, S., Ind Eng Chem, 38:1228 (1946)
185. Bodamer, G., Kunin, R., Ind Enn Chem, 43:1082 (1951)
186. Oikawa, S., Proc Res Sot Japan Sugar Ref Technol,
16:64 (April 1965)
187. Demaimay, M., LeHeneff, Y., Printemps, P., a

Biochem, 3 (April 1978)
188. Holmberg, H., European Patent Appl No. 83 325 (July 6,

1983)
189. Takahaski, Y., Seito Giiutsu Kenkvu Kaishi, 31:76 (1982)
190. Rigal, L., Gaset, A., Gorrichon, J.P., Ind Erw Chem Proc
Res Des, 20:719 (1981)
191. Hino, T., Ito, K., Hayashi, K., Niooon NoPei-Kagaku

Kaishi, 35:773 (1961)
4
Column Chromatography Processes
4.1 INTRODUCTION
In most ion exchange operations, an ion in solution is

replaced with an ion from the resin and the former solution
ion remains with the resin. In contrast, ion exchange
chromatography uses the ion exchange resin as an adsorption
or separation media which provides an ionic environment,
allowing two or more solutes in the feedstream to be
separated. Beads of non-ionic resins, gels or molecular
sieves may also be used as the solid stationary phase in
column chromatography. The feed solution is added to the
chromatographic column filled with the separation beads and
eluted with solvent, often water for fermentation products.
The resin beads selectively slow some solutes while others
are eluted down the column (Figure 4.1). As the solutes
move down the column, they separate and their individual
purity increases. Eventually, the solutes appear at different
times at the column outlet where each can be drawn off
separately.
Tswett (1,2) was the first to identify correctly the

nature of the separation of colored vegetable pigments in
petroleum ether when the solution was passed through a
column of fine particle calcium carbonate. The process was
called a chromatographic separation because of the separation
of the pigments into bands of different colors. Following
Tswett have been numerous scientists who have developed ion
exchange chromatography into a sophisticated analytical
technique used in many scientific areas.
333
Solutes
Desorbent Added
Added
Solute
Nixture
Separation Slow
Solutes Component
Occurs Fast
Added Component Removed
To From
Removed
Column Column
From
Column
Figure 4.1. The steps of chromatographic separation are:

addition of the mixed solutes to the column, elution to effect
separations and removal of the separated solutes.
The first commercial use of ion exchange

chromatography occurred during World War II at Iowa State
University where Spedding and his coworkers (3,4) isolated
transuranium elements as their contribution to the Manhattan
Project. Several years later, the commercial preparation of
individual rare earths was performed using ion exchange
chromatography (5,6).
Despite these initial successes with chromatography in

preparing pure materials in large quantities, other commercial
applications were not developed until the mid to late 1960’s.
It was in the 1970’s before industrial separation problems
were identified which could not be solved by conventional,
well-tried methods such as distillation or crystallization. At
present, there are industrial ion exchange chromatographic
Column Chromatography Processes 335
units which separate amino acids, hydrocarbon isomers,

glucose from fructose, sucrose from molasses,
monosaccharides from di- and polysaccharides and ion
exchange regenerants from salts.
A literature search by Sitrin and coworkers (7),

covering the period from 1980 to 1983, revealed that out of
the 7,000 citations for liquid chromatography, only 100
discussed preparative work. Few of these involved
separations larger than 100 mg. That distribution remains
typical of the published activity in the chromatography area.
There are two advances that have contributed strongly

to the recent development of industrial ion exchange
chromatography: (1) improvements in chromatographic
equipment with recycle control systems and pseudo-moving
bed control systems; and (2) improvements by resin
manufacturers in developing resins with narrow, controlled
size distributions and improved osmotic resilience. General
resin considerations and commercial ion exchange resins have
been described in Chapter 3. Descriptions of
chromatographic systems will be covered in Section 7 of this
chapter.
4.2 CLASSIFICATIONS OF CHROMATOGRAPHY
4.2.1 Chromatographic Methods

The literature is replete with various column
chromatographic methods: adsorption chromatography, ion
exchange chromatography, gel filtration chromatography, gel
permeation chromatography and affinity chromatography.
These methods are differentiated on the basis of the
retentive ability of the stationary phase, the type of eluent
employed and the material used as the separation phase.
In adsorption chromatography, it is mainly physical

surface forces which are involved in retaining the solute.
This method is the oldest of the chromatographic techniques.
Although it does not have the specificity of the other
methods, adsorption chromatography is still one of the
simplest and most effective techniques of separating mixtures
of non-polar substances and compounds of low volatility.
Silica, alumina, activated carbon and macroporous non-ionic
polymers are the adsorbents used most frequently as the
separation media. Low molecular weight organic solvents are

the most common eluents.
In ion exchange chromatography, true hetero-polar

chemical bonds are formed reversibly between ionic
components in the mobile and the stationary phase. The ion
exchange beads may have either hydrophobic or hydrophilic
matrices which have been functionalized with ionizable
groups, as discussed in the last chapter. The eluent for the
chromatographic separation of fermentation products at the
preparative or higher level is usually water or an aqueous
buffer.
The stationary phase may also serve as a molecular

sieve to separate solutes on the basis of molecular size. This
method is known as gel filtration chromatography or gel
permeation chromatography.
Gel filtration chromatography originated in Sweden in

1959 (8) when columns packed with crosslinked polydextran
gels, swollen in an aqueous solution, were used to separate
water-soluble macromolecules on the basis of size differences.
These gels are still used extensively for separating water
soluble biological compounds for further characterization
studies (9). Gel permeation chromatography was developed at
Dow Chemical in 1964 (10) using crosslinked polystyrene gels
swollen in organic solvents to separate synthetic polymers on
the basis of size. Column chromatography with ion exchange
resins also can utilize size differences for separation but
more frequently relies upon adsorption for affinity
differences to effect separations.
Partitioning of a compound between a hydrophobic

stationary and a polar aqueous mobile phase is called reverse
phase liquid chromatography. Reverse phase chromatography
is usually performed on columns packed with silica gel to
which a C-8 to C-18 hydrocarbon has been covalently
attached. For such systems, the strength of an eluting
solvent increases as its polarity decreases and compounds of
similar structure can be expected to elute in order of
decreasing polarity.
Polar, charged fermentation products, such as peptides

and glycoproteins, are well suited for recovery and
purification with reverse phase chromatography. This is due
to the removal of very polar, often colored contaminants at

the solvent front, ahead of the desired material. Using this
technique early in a purification scheme can simplify the
number of steps one must employ.
Affinity chromatography, which originated in 1968 at

the National Institutes of Health (1 I), is based on a unique
and fundamental property of biological macromolecules: their
selective, high-affinity recognition of, and reversible
interaction with, other molecules. This technique is
sufficiently distinct from the other chromatographic methods
that it will be discussed separately in the next chapter.
Most literature references to column chromatography are

concerned strictly with analytical applications. While some
of the information is directly relatable to the preparative and
commercial recovery of fermentation products, one must be
cautious due to the differences in ranges of operating
parameters, throughput and the addition of components to aid
in analytical resolution.
The low crosslinked polymer beads used in analytical

chromatographic applications have mean diameters of 70 to
150 microns and can only be used at low flow rates and
pressures less than 17 bars. These beads collapse at higher
pressures which restricts the flow rate, making separations
impossible. Pilot plant and industrial chromatography use
larger mean diameter beads (200 to 450 microns) with a
higher degree of crosslinking. The necessary degree of
separation is achieved by removing a specific cut of the
chromatogram.
4.2.2 Types of Chromatographic Separations

Chromatographic separations can be classed into four
types according to the type of materials being separated:
affinity differences, ion exclusion, size exclusion and ion
retardation chromatography. These types of separations may
be described in terms of the distribution of the materials to
be separated between the phases involved.
Figure 4.2 shows a representation of the resin-solvent-

solute components of a column chromatographic system. The
column is filled with resin beads of the solid stationary phase
packed together with the voids between the beads filled with
solvent solution. The phases of interest are (1) the liquid
phase between the resin beads; (2) the liquid phase held
within the resin beads; and (3) the solid phase of the
polymeric matrix of the resin beads. When the feed solution
is placed in contact with the hydrated resin in the
chromatographic column, the solutes distribute themselves
between the liquid inside the resin beads and that between
the resin beads. The distribution for component i is defined
by the distribution coefficient, Rd.1 :
Kdi = 'rif'li (4.1)
where Cri is component i’s concentration in the liquid within

the resin bead and Cli is the component i’s concentration in
the interstitial liquid. The distribution coefficient for a
given ion or molecule will depend upon that component’s
structure and concentration, the type and ionic form of the
resin and the other components in the feed solution. The
distribution coefficients for several organic compounds in
aqueous solutions with ion exchange resin are given in Table
4.1 (12).
-Resin Bead
-Interstitial Liquid
(vl)
iquid in . Resin
,..
Figure 4.2. Representation of the three phases involved in

chromatographic separation.
Table 4.1: Distribution Coefficients (12)

SOLUTE RESIN Kd
-
Ethylene Glycol Dowex 50-X8, H+ -67
Sucrose Dowex 50-X8, H+ .24
d-Glucose Dowex 50-X8, Ht .22
Glycerine Dowex 50-X8, H+ .49
Triethylene ~lycol Dowex 50-X8, H+ .74
Phenol Dowex 50-X8, H+ 3.08
Acetic Acid Dowex 50-X8, Ht .71
Acetone Dowex 50-X8, Ht 1.20
Formaldehyde Dowex 50-X8, H+ .59
Methanol Dowex 50-X8, H+ -61
Formaldehyde Dowex l-X7.5, Cl- 1.06
Acetone Dowex l-X7.5, Cl- 1.08
Glycerine Dowex 1-x7.5, ci- 1.12
Methanol Dowex l-X7.5, Cl- .61
Phenol Dowex l-X7.5, Cl- 17.7
Formaldehyde Dowex l-X8, SO,=, 50-100 1.02
Acetone Dowex l-X8, SO,=, 50-100 .66
Xylose Dowex 50-X8, Na+ .45
Glycerine Dowex 50-X8, Na+ .56
Pentaerythritol Dowex 50-X8, Na+ .39
Ethylene Glycol Dowex 50-X8, Na+ .63
Diethylene Glycol Dowex 50-X8, Nat .67
Triethylene Glycol Dowex 50-X8, Nat .61
Ethylene Diamine Dowex 50-X8, Nat .57
Diethylene Triamine Dowex 50-X8, Na+ .57
Triethylene Tetramine Dowex 50-X8, Na+ .64
Tetraethylene Pentamine Dowex 50-X8, Na+ .66
The ratio of individual distribution coefficients is often

used as a measure of the possibility of separating two solutes
and is called the separation factor, cr.
(4.2)
Example 4.1
From Table 4.1, the separation factors for acetone-
formaldehyde separability would be 0.49, 0.98 and 1.54 for
Dowex 50WX8(H+), Dowex lX8(Cl’) and Dowex IXS(SOS_~)
resins, respectively. For comparison purposes, it may be
necessary to use the inverse of o, so that the values would
be 2.03 and 1.02 for Dowex 50WX8(H+) and Dowex lX8(Cl-),

respectively. When Q! is less than 1, the solute in the
numerator will exit the column first. When ar is greater than
1, the solute in the denominator will exit the column first.
From this limited amount of data, the resin of choice would
be Dowex 50WX8(Ht) and acetone would exit the column
first. The separation factor is sometimes called the relative
retention ratio.
The acetone-formaldehyde separation would be an

example of affinity difference chromatography in which
molecules of similar molecular weight or isomers of
compounds are separated on the basis of differing attractions
or distribution coefficients for the resin. The largest
industrial chromatography application of this type is the
separation of fructose from glucose to produce 55% or 90%
fructose corn sweetener.
Ion exclusion chromatography involves the separation of

an ionic component from a non-ionic component. The ionic
component is excluded from the resin beads by ionic
repulsion, while the non-ionic component will be distributed
into the liquid phase inside the resin beads. Since the ionic
solute travels only in the interstitial volume, it will reach
the end of the column before the non-ionic solute which
must travel a more tortuous path through the ion exchange
beads. A major industrial chromatography application of this
type is the recovery of sucrose from the ionic components of
molasses.
In size exclusion chromatography, the resin beads act as

molecular sieves, allowing the smaller molecules to enter the
beads while the larger molecules are excluded. Figure 4.3
(13) shows the effect of molecular size on the elution time
required for a given resin. An industrial chromatographic
application using size exclusion is the separation of dextrose
from di- and polysaccharides in the corn wet milling
industry.
Example 4.2
The ion exclusion technique has been used for the
separation of monosodium glutamate from other neutral amino
acids (14). The amino acid solution, neutralized to pH 7.2
with sodium hydroxide, was passed through a column
containing a sulfonic acid strong cation resin in the sodium
form. The monosodium glutamate fraction was eluted first,

as would be expected by ion exclusion, followed by the
neutral amino acids, as shown in Figure 4.4.
ETHV~ENEGLYCOL
b
J DIEMYWE GLYCOL
TRIFMYLENE GLYCOL
TETRAETHYLENE
GLYCOL
Po~yE77iy~Et+E
a_ycoL
Fw4L.a
ElH~l$t&$LYCOL
VOID %LLME
DYE
I I I I I lwml
ml 200 Sal ml
tbLECURkIQiTOF%LUlE
Fieure 4.3. Effect of molecular weight on the elution volume

reiuired for glycol compounds (Reference 13).
Effluent volume
Figure 4.4. Separation of monosodium glutamate and neutral

amino acids by ion exclusion at pH 7.2 through Dowex 50X8
(Na form) resin (Reference 14).
342 Separation and Purification Techniques in Biotechnolo~
Ion retardation chromatography involves the separation

of two ionic solutes with common counterion. Unless a
specific complexing resin is used, the resin must be placed in
the form of the common counterion. The other solute ions
are separated on the basis of differing affinities for the
resin. Ion retardation chromatography is starting to see use
in the recovery of acids from waste salts following the
regeneration of ion exchange columns.
4.2.3 Chromatographic Processing Techniques

Chromatography may also be categorized according to
the techniques used to carry out the process. The
chromatogram may be developed using frontal analysis,
elution development or displacement development (15).
In frontal analysis, the sample is continuously fed into

the column until “breakthrough” occurs. The least strongly
adsorbed compound emerges first, as shown in Figure 4.5.
,Concentrotion of somple
/ ---- ---_-_-
Solute 2
Solute 1
I
Solvent only I Solute 1
L J 1
Effluent volume
Figure 4.5. Concentration-volume relationship for frontal

analysis of two-component system (Reference 15).
Each additional plateau indicates the emergence of another

solute in the effluent. This technique only leads to the
resolution of the least strongly adsorbed solute. It may be
used to remove small amounts of a compound which is
used to remove small amounts of a compound which is

strongly adsorbed ‘from a feedstream which contains other
weakly adsorbed solutes. In such instances, the technique
may be used as a preparative technique.
In elution development, a small amount of the sample

mixture (at most a few percent of the adsorbent capacity) is
fed into the column, then the eluent, which has no affinity
for the adsorbent, is introduced. Separation is achieved in
the form of “bands” of the individual solutes on a background
of the mobile phase, as shown in Figure 4.6. This is the
most widely used chromatographic technique due to its
common occurrence in analytical methods.
Mixture injected
into column or on
Resultant curve
0 umn moter80
Figure 4.6. Schematic of the elution development of a

chromatogram (Reference 15).
Displacement development is like elution development in

that a small sample is fed to the column before elution
begins. However, in displacement development, the eluent
has a higher affinity for the adsorbent material than does
the sample. All of the solutes in the sample are forced from
the sorption sites and move ahead of the front produced by
the eluent. In turn, the individual solutes displace one
another in order of increasing sorption strength to give the
type of chromatogram shown in Figure 4.7. When used as a
preparative method, it is possible to obtain individual
concentrated components from a diluted mixture.
4.3 CHROMATOGRAPHIC MATERIALS
The adsorption materials, described in Chapter 2, and

the ion exchange resins, described in Chapter 3, are also
used as the solid phase separation media in column
chromatography. In addition, hydrophilic and hydrophobic
gels are used in size exclusion chromatography. Common
Concentration in mobile phase (effluent)
I I I
I I lb)
k Solute 4
I._____A_{
Solute 1
I
I
1
1
I
I
I
Figure 4.7. Displacement development. (a) sorption

isotherms; (b) concentration-effluent volume profile
(Reference 15).
hydrophilic gels are crosslinked dextrans, polyacrylamides and

agarose, while the hydrophobic gels are usually based on
polystyrene.
Dextran gels, under the tradename Sephadex, are

composed of linear polyglucose chains of dextran crosslinked
with epichlorohydrin. The polyacrylamide gels, under the
tradename &o-Gel, are prepared by the polymerization of

acrylamide containing N,N-methylene bisacrylamide as the
crosslinking agent. The agarose gels, under the tradenames
Sepharose and Bio-Gel A, have a more fragile gel structure
since it is due to hydrogen bonding, not chemical
crosslinking. The agarose gels must be used at temperatures
below 30°C and in a pH range of 5 to 8. Despite these
limitations, agarose gels have the large pore structure needed
to fractionate large macromolecules, viruses and subcellular
particles.
The dextran and polyacrylamide gels are crosslinked

sufficiently to render them insoluble in water. However, the
gels are very hydrophilic and swell substantially in aqueous
media. The gels are available in three size ranges of the
hydrated beads: 40 to 80/~, 80 to 150~ and 150 to 300~. The
fractionation ranges for the various hydrophilic gels are
shown in Table 4.2.
Table 4.2: Hydrophilic Type Gel Materials

Hydrated bed
Fractional range VOlIMe, Vb, in Water regain,
in nlolecular-weight ml per g ml per g dry
Gel type units dry gel gel
Bio-Gel P-Z 200 to 2,ooo 3.8 1.6
r-4 500 to 4,oocl 6.1 2.6
P-6 1,000 to 5,ooo 7.4 3.2

P-10 5,oOa to 17,000 12 5.1
P-30 20,000 to 50,000 14 6.2
P-60 30,000 to 70,000 18 6.0
P-100 40,000 to 100,ooo 22 7.5
P-l 50 50,cOo to 150,ooo 27 9
P-200 80,000 to 300,ooo 47 13.5
P-300 100,000 to 400,ooo 70 22

Sephadex C-10 up to 700 2.5 1.0
G-15 up to 1,500 3.0 1.5
G-25 100 to 5,ooo 5 2.5

c-50 500 to 10,000 10 5.0
G-75 3,000 to 70,0x 13 7.5
c-100 4,000 to 150,000 17 10
G-150 5,000 to 400,ooa 24 15
c-200 5,oOa to 800,000 30 20
A@rose 0.5 m (10%) 10,000 to 250,ooO
1.0 m (8%) 25,ooO to 700,ooO
1.5 Ill 10,000 to 1,cilo,ooo
2.0 m (6%) 50,000 to 2,ooo,KKJ
5.0 m 10,000 to 5,cOo,Ooa

15.0 m (4%) 200,000 to 15,ooo,ooa
50.0 I” 100.000 to 5o,ooo,OOo
150 I” 500,000 to 150,000,000
The hydrophobic gels, under the tradename Styragel, are

prepared by the polymerization of styrene-containing
divinylbenzene as the crosslinking agent. These beads are
much more rigid than the hydrophilic gels. Unlike the
hydrophilic gels which are normally used in an aqueous or
very polar medium, the polystyrene gels require a solvent of
low polarity, such as tetrahydrofuran, cyclohexanone or
carbon tetrachloride. The fractionation ranges for Styragel
are listed in Table 4.3
Table 4.3: Properties of Styragel Materials

Approximate
Fractionation range exclusion limit in
in molecular-weight molecular-weight units
Type units (average porosity in A)
60 Styragel 800 1,600

100 2,000 4,000
400 8,000 16,000
1 x103 20 )000 40,000
5 x 103 100,000 200,000
10 x 10 3 200,000 400,000
30 x 103 600,000 1,200,000
1 x105 2,000,000 4,000,000
3x105 6,000,OOO 12,000,000
5x105 10,000,000 20,000,000
10 x 10 5 20,000,000 40,000,000
Table 4.4 shows the operational limits of these resins,

along with additional resin suppliers.
Resins of very uniform size have been developed in

several mean size ranges which could be very useful in the
chromatographic separation of small biochemical products
such as amino acids and short peptides. New FAST-FLOW
Sepharose resins have been prepared to allow improved
through-put, easier scale-up and simpler operation for
industrial chromatography of proteins (16). Figure 4.8
illustrates the processing potential of this type of resin. The
amount of protein processed per hour is 4.5 kg, with 2.3 kg
albumin. The total processing time was only 60 minutes to
treat 75 liters of plasma feed solution.
The use of porous glass which is silanized and

functionalized (17) allows more protein-compatible resins to
Table 4.4: Packing Materials
Media PH Max temp Max flow Max pressure Makers*

range OC (cm/h) drop (cm H?O)
CompressiblePackings
Dextran,cross-linked >2 >120 <77 160-16 Ph
Dextran,bisacrylamide 3-11 1120 25-40 300-70 Ph
Polyacrylamide 2-10 >120 >2.8 100-20 BR
Agarose 4-9 40 -- 100-30 Ph, BR, LK
Agarose,polyacrylamide 3-10 40 50-18 >15 PL, Si, Se, LK
Agarose,cross-linked 3-14 >120 30-15 120-50 Ph
Polyether,cross-linked 1-14 >120 _- __ EM, T
Cellulose Me, Wh, Si
9
E
3
Rigid PackingMaterials *
Surface-modifiedglass <9 >120 Pi, EN 0
Surface-modifiedsilica <9 >120 Co, BR, Wa, Se, EN, EM F
Cross-linked n/a >120 LC, KL, Kn g
hydroxyethylmethacrylate F
See also Chapter3 Appendix
*BR = Bio-Rad;Co = Corning;EM = EM Science;EN = Electra-Nucleonics; KL = Koch-Light;Kn = Knauer AG;

LC = LaChema;LK = LKB; Me = Merck; Ph = Pharmacia;Pi = Pierce; PL = P-L Biochemicals;Se = Serva;
Si = Sigma;T = Toyo Soda; Wa = Waters;Wh = Whatman
pn 5.2. I 0 0.02
glycoprolein
/
pH 5.2,
I- 0.02
\ 1
b
0 10 20 30 40 minutes
0 100 200 300 400 lilres
Figure 4.8. High performance industrial scale ion exchange

chromatography on DEAE-Sepharose Fast Flow in a
Sephamatic column (Reference 16).
be developed. Schomberg (18) has described the chemical

modification of silica gels by polymer coating. These resins
are especially useful in reverse phase chromatography
applications.
4.4 THEORY
Mathematical theories for ion exchange chromatography

were developed in the 1940’s by Wilson (19), DeVault (20),
and Glueckauf (2 1,22). These theoretical developments were
based on adsorption considerations and are useful in
calculating adsorption isotherms from column elution data.
Of more interest for understanding preparative
chromatography is the theory of column processes, originally
proposed by Martin and Synge (23) and augmented by Mayer
and Thompkins (24), which was developed analogous to
fractional distillation so that plate theory could be applied.
One of the equations developed merely expressed

mathematically that the least adsorbed solute would be eluted
first and that if data on the resin and the column dimensions
were known, the solvent volume required to elute the peak
solute concentration could be calculated. Simpson and
Wheaton (25) expressed this equation as:
"MAX = Kd “rl + “1 (4.3)

where VMuIAxis the volume of liquid that has passed through

the column when the concentration of the solute is maximum
or the midpoint of the elution of the solute. K,, defined in
Equation 4.1, is the distribution coefficient of the solute in a
“plate” of the column, V,, is the volume of liquid solution
inside the resin and V, is the volume of interstitial liquid
solution.
The mathematical derivation of Equation 4.3 assumes

that complete equilibrium has been achieved and that no
forward mixing occurs. Glueckauf (26) pointed out that
equilibrium is practically obtained only with very small
diameter resin beads and low flow rates. Such restricting
conditions may be acceptable for analytical applications but
would severely limit industrial and preparative applications.
For industrial and preparative chromatography, column
processing conditions and solute purity requirements are
often such that any deviations from these assumptions are
slight enough that the equation can still be used as an
approximation for the solute elution profile. An example of
such an elution profile is given in Figure 4.9.
1.3420 ,
SEPARATION OF POLYHYOR/C ALCOHOLS 4 NoCf
1.3400 -
RESIN: DOWEX 50 -.X 8%,
50-100 MESH, No*
1.3300 - I-I Glycrrinr

&-A Pentorrythritol
O-0 Xylolr
1.3360 -
1.3340 -
28 36 44 52 60 68 76 84 92
VOLUME EFFLUENT (ml)
Figure 4.9. Elution chromatogram for the separation of

polyhydric alcohols and NaCl using the sodium form of a
cation resin (Reference 12).
4.4.1 Theoretical Plate Height

A second important equation for column processes is
that used for the calculation of the number of theoretical
plates, i.e., the length of column required for equilibration
between the solute in the resin liquid and the solute in the
interstitial liquid. If the elution curve approximates a
Gaussian distribution curve, the equation may be written as:
2 c (c + 1)
P = (4.4)
lJ2
where P is the number of theoretical plates, c (= K,V,,/V,)
is the equilibrium constant, W is the half-width of the
elution curve at an ordinate value of l/e of the maximum
solute concentration and e is the base of the natural
logarithm. For a Gaussian distribution, W = 40, where B is
the standard deviation of the Gaussian distribution. The
equilibrium constant is sometimes called the partition ratio.
An alternate form of this equation is:
2 v
PIAX %AX - ‘1)
P = (4.5)
W2
Here W is measured in the same units as V,,. This form

of the equation is probably the easiest to calculate from
experimental data. Once the number of theoretical plates has
been calculated, the height equivalent to one theoretical
plate (H.E.T.P.) can be obtained by dividing the resin bed
height by the value of P.
The column height required for a specific separation of

two solutes can be approximated by (27):
3.29 + 0.5 Cl + 0.5

c2
l6 = (4.6)
c2 - c1 fi2 + fil
where H is the height of the column, P is the number of

plates per unit of resin bed height, and c is the equilibrium
constant defined above. Note that the number of plates in a
column will be different for each solute. While this equation
may be used to calculate the column height needed to
separate 99.9% of solute 1 from 99.9% of solute 2, industrial
and preparative chromatography applications typically make
more efficient use of the separation resin by selectively

removing a narrow portion of the eluted solutes as illustrated
in Figure 4.10 (28).
280
1.8
0
.4 86 .8 LO la2 1A 1,G la3 2,3 2.2 2,4
kLUE OF t!LUAl’E kLLEClED (k) / kLU.E OF i&SIN BED (b)
Figure 4.10. Distribution of eluate into fractions for product,

recycle and waste for NaCl and glycol separation (Reference
28).
Table 4.5 shows how the theoretical plate number for a

chromatography system may be calculated from various
combinations of experimental data. The band variance, 4, is
calculated from the experimental data and combined with the
retention time, tn, for a given solute. Figure 4.11 shows the
different experimental values which may be used to calculate
at*
Table 4.5: Calculation of Plate Number from Chromatogram

Measurements Conversion to Variance Plate number
tR and ut ______ N = (tR/ ut)’

tR and baseline width Wb = w,/4 N = 16(tR/Wb)*
=t
tR and width at half height W,, 5 ut = WO @i-ix N = 5.54(tR/W0 +*
tR and width at inflection points = wi;2 N = 4(tR/WiJZ *
ut
(0.607 h) Wi
tR and band area A and height h at = A/he N = 2n(tRh/Alz

I I’
t Injection t
Figure 4.11. Identification of a chromatographic peak

segments for the calculation of column performance.
4.4.2 Zone Spreading

The net forward progress of each solute is an average
value with normal dispersion about the mean value. The
increased band or zone width which results from a series of
molecular diffusion and non-equilibrium factors is known as
zone spreading.
The plate height as a function of the mobile phase

velocity may be written as a linear combination of
contributions from eddy diffusion, mass transfer and a
coupling term:
B 1
II = - + Ev +
(4.7)
V S
l/A + EM/v
where A is a constant, B is the coefficient for the diffusion

term, Y is the linear flow rate, and EM and E, are the mass
transfer coefficients for the mobile and stationary phases,
respectively. The plate height is seen to depend on the
partition ratio for each species. The smaller the partition
ratio, the faster the respective zone moves and the larger is
the respective plate height.
A plot of Equation 4.7 for any type of linear elution

chromatography describes a hyperbola, as shown in Figure
4.12 (15). There is an optimum velocity of the mobile phase
for carrying out a separation at which the plate height is a
minimum, and thus, the chromatographic separation is most
efficient:
[Rt(l - Rt) d;/DS] (4.8)

voptimum =
where D, is the diffusion coefficient of the solute molecule

in the mobile phase, D, is the diffusion coefficient in the
stationary phase, d, is the diameter of the adsorbent particle
and R, = L/yt, where L is the distance the zone has
migrated in time t.
_-- - - Minimum plate height

/
’ Molecular
- J/- diffusion
‘I
Resistonce to moss tronsfer
L---- ’ I ----------------
I Eddy diffusion
L II, I 1 I I I I I I I
0 2 4 6 8 10 12 14
Corrier velocity, cm/set
Figure 4.12. Relationship between plate height and velocity

of the mobile phase (Reference 15).
4.4.3 Resolution
A variation on calculating the required column height is
to calculate the resolution or degree of separation of two
components. Resolution is the ratio of peak separation to
average peak width:
'MAX,2 - 'MAX,1
R = (4.9)
0.5(W1 + w2>
The numerator of Equation 4.9 is the separation of the

two solutes’ peak concentration and the denominator is the
average band width of the two peaks. This form of the
equation is evaluating the resolution when the peaks are
separated by four standard deviations, rr. If R = 1 and the
two solutes have the same peak concentration, this means
that the adjacent tail of each peak beyond 20 from the
V,, would overlap with the other solute peak. In this
instance, there would be a 2% contamination of each solute
in the other.
Resolution can also be represented (29) by:
R = qy-/)(l~~c2) (4.10)
Resolution can be seen to depend on the number of plates

for solute 2, the separation factor for the two solutes and
the equilibrium constant for solute 2.
In general, the larger the number of plates, the better

the resolution. There are practical limits to the column
lengths that are economically feasible in industrial and
preparative chromatography. It is possible to change P also
by altering the flow rate, the mean resin bead size or the
bead size distribution since P is determined by the rate
processes occurring during separation. As the separation
factor increases, resolution becomes greater since the peak-
to-peak separation is becoming larger. Increases in the
equilibrium constant will usually improve the resolution since
the ratio cs/( ltcs) will increase. It should be noted that
this is actually only true when c2 is small since the ratio
approaches unity asymptotically as c2 gets larger. The
separation factor and the equilibrium factor can be adjusted
by temperature changes or other changes which would alter
the equilibrium properties of the column operations.
Equation 4.10 is only applicable when the two solutes

are of equal concentration. When the solutes are of unequal
concentration, a correction factor must be used:
(A; + A; )/2A1A2
where A, and A, are the areas under the elution curve for
solutes 1 and 2, respectively. Figure 4.13 shows the
relationship between product purity (n), the separation ratio
and the number of theoretical plates. This graph can be
used to estimate the number of theoretical plates required to
attain the desired purity of the products.
1 nc 106
6 x 104
4 x 104
2 x 104
104
'6000
I4000
1000
600
400
3.0
200
4.0
100
6.0
60
10.0
40
20.0
20
II" ”
10-14 10-e 10-5 10-3 .Ol .06 .I

10-10 10-6 10-4 3x10-3 .03
qA;+A;
*Al A2
Figure 4.13. Relationship between relative retention ratio,
number of theoretical plates and product purity (Reference
26).
Example 4.3
When the product purity must be 98.0%, then r) = Am/m
= 0.01 when the amount of the two solutes is equal. If the
retention ratio, Q, is equal to 1.2, then the number of
theoretical plates from Figure 4.13 is about 650. With a

plate height of 0.1 cm, the minimum bed height would be 65
cm. In practice, a longer column is used to account for any
deviation from equilibrium conditions.
Figure 4.14 shows the effect of relative concentration

of two solutes on product purity and product recovery for a
few resolution values. When R = 1.25, the solutes are
effectively completely separated, with crosscontamination less
than 1%. Rarely would this level of purity be required in
current industrial chromatography applications. For R = 1.0,
the cross contamination is less than 3%; however, the
recovery of the more dilute component is significantly
reduced when the solute concentration ratio is less than 1:4.
When R = 0.8, the recovery and the pourity is still
acceptable in industrial chromatography for solutes of
approximately the same concentration.
biTIVE SOLUTE bNCENTRATION
EQUAL 4Tol
Figure 4.14. Effect of relative solute concentration on peak

resolution.
As was mentioned earlier, both resin properties and

process parameters will affect the numvber of theoretical
plates and, therefore, the separation resolution. The mean
size, the size uniformity and the crosslinking of the ion
exchange resin are the critical resin parameters for rate
processes.
The mean bead size is chosen so that the diffusion into

and out of the resin beads does not become the rate
controlling step and so that the pressure drop in the column
does not become excessive. Figure 4.15 (25) shows the effect
of particle size on the shape of the elution profile of
ethylene glycol. Small beads adsorb and release the solute
quickly for sharp, distinct elution profiles. Larger beads
take longer to adsorb and release the solute so that the
elution profiles are wider and less distinct. The resin with
the smallest mean bead size has the greatest number of
theoretical plates. If the separation ratio for the solutes is
large enough, a resin with coarser beads may be used, while
one with fine beads must be used for solutes with a
separation ratio near 1.0. The practical lower limit on bead
size is determined by the acceptable pressure drop limit of
the column.
Sam: ETHYLENE
kvca
“30 40 !a Go 70 30 90 130
EFFLENTVOLN (FL)
Figure 4.15. Effect of particle size on the elution profile for

ethylene glycol (Reference 25).
When there is a large distribution of bead sizes for a

resin of a given mean bead size, it becomes difficult to
obtain proper elution profiles. The beads smaller than the
mean will elute the solute more rapidly and those larger, less
rapidly, to cause additional spreading or tailing of the elution
profile. Ideally, the resin beads should all be of the same
size to amplify the attraction differences of the solutes
uniformly as they flow down the column. Separation resins
are now available for industrial chromatography that have
90% of the beads within %20% of the mean bead size. One
resin manufacturer can even supply resins that have 90% of
the beads within %10% of the mean bead size (Figure 4.16).
Table 4.6 shows the effect of size distribution on separation
for glucose-fructose separation.
The crosslinkage of the resin will affect the position of

the elution profile as is shown by Figure 4.17. This shift
occurs because an increase in the crosslinking decreases the
liquid volume inside the resin beads. As crosslinking is
increased, the exclusion factor for ionic solutes is increased.
Figure 4.16. Dowex Monosphere 99 with a mean particle

diameter of 390 microns. Dowex and Monosphere are
registered trademarks of Dow Chemical.
Table 4.6: Effect of Bead Size Distribution on Glucose-Fructose Separation
Bead Glucose-Fructose Development Length

Size Volume Peak Separation (Elution Volume in Bed Volumes From
Distribution (Mesh) _-I?_- (Bed Volumes) Initial Glucose to Final Fructose)
s
z
1. 35-40 55
0.9 0.96 5
40-50 26
c)
0.81-0.87 z
2. 40-50 80 0.10-0.12
3
3. 40-50 90 0.11-0.13 0.79-0.81 R
6
4. 40-50 95 0.12-0.15 0.76 +z
z
5. 45-50 95 0.14-0.16 0.74
z
w
v,
D
Likewise, increased crosslinking will decrease the rate of

diffusion within the resin for non-ionic solutes. The
mechanical strength and shrink/swell of organic polymeric
resins is increased by increasing the resin crosslinking. The
weight of the resin in the column combined with the pressure
effect of flow may cuase deformation of the resin beads
leading to even higher pressure drops if the resin
crosslinkage is not sufficient. Industrial chromatographic
separations of sugars have been achieved with resins of 3 to
8% crosslinking for resin bed depths of 3 to 6 meters (30).
SAUTE: ETHYLENE ha
f&SIN: hEX 9, tf ~~1

50 - 1M) tkSH
Q?
-.- I’~ I I \
.
h
0
40 50 u) 70 8)
CFFLENT Vausz h_)
Figure 4.17. Effect of resin crosslinkage on the position of

the elution profile for ethylene glycol (Reference 25).
The effect of porosity and resin crosslinking on

glucose-salt separation was studied by Martinola and Siegers
(31). Figure 4.18 shows that at the low crosslinkage
normally used in spearations, the gel resins have a lower
glucose loss and have less NaCl in the product. Macroporous
resins are not manufactured at the low crosslink levels (~6%)
needed for the glucose-salt separation.
Resin manufacturers have made significant advances in

recent years to optimize and to customize resin properties
for specific industrial applications. Often these resins are so
0 1 I I I I J
0 4 8 32 16 20
G~~~SLINKING &m)
Figure 4.18. Chromatographic separation as a function of

crosslinking and porosity (Reference 3 1).
tailored for the separation that data sheets and resin samples
are only available to the customer with the application.
The flow velocity, the solute concentrations, the cycle

frequency, the pressure drop and the column temperature are
the process parameters to be controlled to achieve the
desired separation.
The elution profile becomes sharper as the flow rate is

decreased (Figure 4.19). The flow rate should be between 0.5
and 2.0 times the critical velocity (v,) to insure that the
profile “tailing”, due to density and viscosity differences in
the solution moving through the column, is avoided or at
least minimized. The critical velocity is defined by (30):
g ( P2 - P,)
v
C
= (4.11)
k (q2 - ‘$)
where g is the gravity constant, p is the density, 0 is the

viscosity, k (= Ap/r)vL) is the permeability coefficient of the
bed, Ap is the pressure drop in the resin bed, v is the linear
flow rate of the solution and L is the height of the resin
bed. The subscripts 1 and 2 in this equaiton refer to the
low and high extremes of these values for the solution as it
cycles from feed to solvent. The permeability coefficient for
industrial chromatography of sugar and polyols has been
found to be 1 to 4 x 1010m-2.
SOLUTE: ETHYLENE GLYCOL
RESIN: DOWEX 50-X3, H FORM

46 - 50 - 100 MESH
s - .,."
i
* '*
. .
A - 0.164d/(hr-m2)
0.328 ma/(hr-m2)
0.657 nb3/(hr-m2J
1.313d/(hr-d)
.3 -
EFFLUENT VOLUME (ML)
Figure 4.19. Effect of flow rate on the elution profile of

ethylene glycol (Reference 25).
The effect of solute concentration on the elution profile

is shown in Figure 4.20. The concentration changes will
shift the point at which V,, occurs but does not change
the appearance of the initial point when the solute emerges
from the column. Therefore, the appearance of the solute in
the lowest concentration may be the limiting factor for the
total dissolved solute level. For industrial and preparative
chromatography to be effective and economically justifiable,

the solute concentration of the most dilute component should
be greater than 3% dissolved solids. With a total dissolved
concentration of 30 to 50%, this means the relative
concentration for two solutes would range from 15 to 1 to 9
to 1. Solutions with as much as 60% total dissolved solids
are being separated by industrial chromatography. Beyond
that concentration level, the solution viscosity becomes
excessive even at temperatures of 65°C.
%LUTE:EW~E GLYOOL
&IN : b&5X %-XB, H'ForCn
-**-
-
-__
--_
30 M_
2oK”
yJk.”
5 ,q_
OF
11
llx
.. . .. . .. . 2 M. 1)
Figure 4.20. Effect of solute concentration on the elution

profile of ethylene glycol (Reference 25).
As the resin beads are cycled between the feed solution

and the eluting solvent, the beads will expand and contract
in response to the relative osmotic pressure of the solution.
If the feed/eluent cycle occurs too rapidly, the rapid
expansion and contraction can cause the resin beads to “de-
crosslink”. As crosslinkage of the beads decreases, they lose
strength and become more susceptible to compression under
flow. A good indicator of the loss of crosslinkage is an
increase in the bead water retention capacity. Figure 4.2 1
shows how cycle frequency affects bead crosslinkage with an
increasing number of cycles. Typical cycle times for
industrial chromatography are between 1 and 1.5 hours, which

result in expected resin lifetimes of several years for resins
of appropriate initial crosslinkage.
10 MIN, CYCLES, 40-9 KSH

m MIN. ClClES, 30-A KSH
i?o MIN, CICLE, 3D-3 MSH
0 &Ty3 400 @IO 800 lOal 1200 14al

NWBER OF CYCLES
4.21. Effect of cycle frequency and bead size on

resin crosslinkage changes.
Although separation efficiency normally decreases with

increases in column temperature, there are two reasons for
elevating the temperature. First, elevating the temperature
may decrease the viscosity of the mixture making it possible
to separate a mixture without excessive dilution. Second,
elevated fluid temperature is often necessary to prevent
microbial growth on the ion exchange beads or in the
column. Generally, fouling due to microbial growth can be
prevented with a column temperature greater than 60°C.
On a periodic basis it will be necessary to backwash the

resin to remove small particles and to relieve compression
from ion exchange beads. For columns with organic polymer
beads, some method of allowing 50% freeboard volume must
be provided for bed expansion during backwashing. Inorganic
chromatographic materials have the advantage that they do
not compress under flow and do not have to be backwashed.
However, one must monitor the effluent to insure that
silicates are not being sloughed from the inorganic beads.
Chromatographic processes are usually not based on the

exchange of ions on the resin. At times the solutes are
loaded onto the column and then eluted chromatographically
by gradient changes in the elution medium. In this case, as
well as in chromatography with a single eluent, control of
the ionic concentration of the feed and eluent streams is
critical. Traces of ionic material in the feed or eluent may
eventually cause enough exchange of ions to necessitate the
regeneration of the resin. Figure 4.22 (32) shows the effect
even partial conversion of a chromatographic resin would
have on the resolution of molasses constituents.
I I I I I I
XCA lm 90 Eo 70 60 50
7mA 0 lo 20 30 40 50
Figure 4.22. Effect of partial conversion of resin from Ca to

Na form on the resolutions of sugar and salt in molasses
(Reference 32).
4.5 LABORATORY TECHNIQUES
Many of the same laboratory procedures described in

the Adsorpiton and the Ion Exchange chapters are useful in
preliminary chromatographic studies. In those applications
where fine adsorbent particles are being studied, it may be
necessary to cover the sintered-glass bed support with a thin
layer of sand to prevent its becoming clogged by the fine
particles. A bed height-to-diameter ratio of 10 to 1 is
recommended with a column height of approximately 20 to 30
cm for laboratory evaluations.
Prior to beginning any chromatography of biologically

active molecules, it is necessary to establish the relative
stability of the specific molecule in various eluants (33).
Table 4.7 shows such a study of the stability of bovine liver
acid phosphatase over a pH range from 4.0 to 8.5. The
optimal phosphatase activity was maintained when the pH was
approximately 5.6 The duration of such a stability test
should correspond to the length of time the protein solute
will be in the buffer solution during the proposed
chromatography operation.
Table 4.7 : Effect of Elution Buffer on Bovine Liver Homogenate

Acid Phosphatase Activity (33)
Units Activity/ml Units Activity/ml X Decrease in 3 hours
Elution Buffer (Zero Time) (3 Hour Time) at Room Temp.
20 mkl Na Acetate
0.5 N NaCl 57.35 la.87 67.1%
1 nlElEDTA, pH 4.0
20 mtil Na Acetate
0.5 N NoCl 60.02 57.68 3.9%
1 mbt EDT/i, ptl 5.6
20 mN Na Acetate
0.5 N N&l 54.35 24.39 55.1%
1 mMEDTA, pII 8.5
The adsorbent is formed into a stirred slurry using the

initial elution buffer. After a mixing time of 20 minutes for
most adsorbents, the slurry may be decanted into a tall
graduate cylinder and allowed to settle for 45 minutes. The
turbid supernatant is discarded. Additional buffer is added to
the settled resin to transfer it as a slurry into the column.
When gel-type materials are used in chromatography, it

is necesssary to let them swell in a large excess of initial
eluent buffer for the period of time recommended by the gel

manufacturer. The supernatant liquid should be replaced at
least three times with fresh buffer solution. The fully
swollen gel slurry is carefully poured into the column and
allowed to settle for at least five minutes before the excess
fluid is slowly drained away. An excessive flow rate during
the draining will lead to restrictive packing of the swollen
gel particles.
The feed solution should be applied to the column in

the same buffer solution that will be used for the initial
wash or elution. Flow rates are typically between 5 and 25
mL/cm2/hr. Such low rates are required by the rate at
which the adsorption equilibrium is attained with
macromolecules. A pump should be used to maintain this low
rate at a constant value. The column effluent should be
collected in separate vials at small time increments.
For simple molecules with large differences in

distribution coefficients, a single eluting solution may be used
to develop the chromatogram. However, more complex
materials, such as peptides and proteins, require a shift in
the ionic strength of the eluent. This can be done stepwise
or as a gradient. Semenza (34) has proposed the following
rules for the proper choice of eluent:
(1) Use cationic buffers (Tris-HCl,
piperazine-HC 1O,, etc.) with anion resins
and anionic buffers (phosphate, acetate,
etc.) with cation exchange resins.
(2) With anion resins use decreasing pH
gradients and with cation resins use
rising pH gradients.
(3) Avoid using buffers whose pH lies near
the pK of the adsorbent.
If the chromatographed solutes are to be isolated by

solvent evaporation, the use of volatile buffers, such as
carbonic acid, carbonates, acetates and formates of ammonium
should be used.
If better resolution is required, it may be obtained by

changing the type of gradient applied. A convex gradient
may be useful in improving the resolution during the last
portion of a chromatogram or to speed up separation when
the first peaks are well separated and the last few are
adequately spaced. A concave gradient can be used if it is
necessary to imporve resolution in the first part of the
chromatogram or to shorten the separation time whenpeaks in
the latter portion are more than adequately spaced.
The relationship between elution behavior and

logarithmic molecular weight for various substances has been
plotted in Figure 4.23 for several Sephadex gels, in figure
4.24 for the Bio-Gel P series of materials and in Figure 4.25
for agarose materials. The linear rising portion of each
curve defines the molecular weight range within which each
gel gives optimum separation.
Example 4.4
An elution scheme was developed for the separation of
a synthetic mixture of human serum albumin (HSA), bovine
chymotrypsinogen (BCT) and chicken lysozyme (CL) on a
column of Spheron phosphate (35). Elution was first carried
out with a series of linear gradients of buffers by
simultaneously increasing pH and ionic strength in the region
from 0.05M sodium formate (pH 3.5) to 1M sodium acetate
(PH 0 In the last gradient, 1M sodium acetate was used
with the same buffer enriched to 1M sodium chloride.
0, ’ ’ 1 1111’ I
2x10’ 104 lo5 to6
Molecular weight
Figure 4.23. Relationship between K and molecular weight of

globular proteins for Sephadex gels (Reference 15).
10.000 100.000 1000.000

Molecular weigh,
Figure 4.24. Relationship between elution volume and

molecular weight on Bio-Gel P (Reference 15).
Southern beon mosaic virus
Tobacco mosaic virus
I.000 100,000 1.000.000 10,000.000 100.000.000

Molecular weight
Figure 4.25. Relationship between molecular weight and

elution volume on agarose materials (Reference 15).
3’70 Separation and Purification Techniques in Biotechnology
Because HSA tended to be eluted with BCT under these

conditions, a series of experiments were carried out with a
gradient of ionic strength only, at each of the following pH
values: 4, 5, 5.5, 6, 6.5 and 7.5. Only two peaks (HSA and
BCT were combined) were observced in the experiments at
low pH, whereas the proteins HSA, BCT and CL were
separated at higher pH. A pH value of 6.5 was selected with
the linear ionic gradient to give the separation shown in
Figure 4.26.
80 minutes
linear gradients
A*0 B*C c*o 0

-l.
iOml l 10ml 40* 10 254 2s 30
:
I
-j6 i20
.
:
100 200
V,ml
Figure 4.26. Separation of serum albumin (S),
chymotrypsinogen (C) and lysozyme (L) on Sepheron
phosphate 1000 of capacity 3.04 meg/g (Reference 35).
4.6 SCALE-UP PROCEDURES
The aim of scaling-up a chromatographic process is to

obtain the same yield and product quality in the same time
period on laboratory, preparative and industrial scale.
Laboratory analytical purifications tend to be optimized only
for resolution of individual solutes. However, at the
preparative and production scale, it is necessary also to
maximize throughput. Table 4.8 (36) shows the effect of
chromatographic operating parameters on resulution and
throughput. While column length is a critical factor for
resolution with gel filtation and isocratic elutions, it has
little effect on resolution with gradient elution in adsorption
chromatography. The wall effects on resolution are very
noticeable with small radius columns, but decrease as the
column length is increased.
Table 4.8: Effects of Process Parameters on Resolution and
Throughput (36)
Resolution Throughput
Paramter varies with varies with
Column Length (L) L l/L
Column Radius (r) Some effect r*
Temperature (T) Positive effect T
Viscosity (q ) Negative effect 111,

Sample Volume (VI l/IV-VoptimuJ V
Flow Rate (J) l/IJ-JoptimuJ J
Voser and Walliser (37) viewed scale-up as a three step

process involving selection of a process strategy, evaluation
of the maximum and optimal bed height and finally column
design. The selection of a process strategy involves choosing
the direction of flow, frequency of backwashing, the
operation with positive head pressure or only hydrostatic
pressure, and the use of a single column or a series of
columns in batch or semi-continuous operation. The maximum
feasible bed height is determined by keeping the optimal
laboratory-scale specific volume velocity (bed volume/hour)
constant. The limiting factors will be either pressure drop
or unfavorable adsorption/desorption kinetics since the linear
velocity also increases with increasing bed height. Bed
heights from 15 cm to as high as 12 m have been reported.
The column diameter is then selected to give the required
bed volume. The column design combines these column
dimensions with the practical considerations of available
space, needed flexibility, construction difficulty and flow
distribution and dilution for the columns.
The scale-up considerations of column chromatography

for protein isolation have been described by Charm and
Matte0 (38). When several hundred liters of a protein
feedstream must be treated, the resin may be suspended in
the solution, removed after equilibration by filtration and
loaded into a column from which the desired proteins may be
eluted. Adsorption onto a previously packed column was not
recommended by them since they feared suspended particles
would clog the interstices of the column, causing reduced
flow rates and increased pressure drops across the column.
The reduced flow rate may lead to loss of enzyme activity
because of the increased time the protein is adsorbed on the
resin.
It is important that all of the resin slurry be added to

the column in one operation to obtain uniform packing and
to avoid the formation of air pockets in the column. An
acceptable alternative, described by Whatman (39), allows the
addition of the adsorbed slurry in increments. When the
resin has settled to a packed bed of approximately 5 cm, the
outlet is opened. The next increment of the slurry is added
after the liquid level in the column has dropped. It is
important that the suspended adsorbent particles do not
completely settle between each addition.
Stacey and coworkers (40) have used the relationships

shown in Table 4.9 to scale up the purification from an 8 mg
protein sample to a 400 mg sample. The adsorbent used in
both columns was a Delta-Pak wide-pore C-18 material.
When eluting the protein, the flow rate should change so
that the linear velocity of the solvent through the column
stays the same. The flow rate is proportional to the cross-
sectional area of the column. The gradient duration must be
adjusted so that the total number of column volumes
delivered during the gradient remain the same. As with size
exclusion chromatography, the mass load on the preparative
column is proportional to the ratio of the column volumes.
Figure 4.27 shows that the chromatogram from the 8 mg
separation is very similar to that obtained for the 400 mg
sample.
Many additional complicated equations have been

developed for scaling up chromatography columns. For gel
filtration, the same elution pattern is obtained with increases
in the scale of operation when the dynamic similarity is
Table 4.9: Scale-Up Calculations (40)

Small Scale Preparative Scale
Coluwn Dimensions 0.39 x 30 cm 3 x 25 cm
Flow Rate Scale Factor
(3.011 1.5 mllmin 90 mllmin

= 59
(0.39)’
Sc~~nplc Load Scale Factor
~;3.0,~ x 25
= 49 &I “IS 400 mg
(0.39) x 30
Crndlunt Duration Calculation 40 min 33 min
1.5 x 40 90 Y (Gradient Duration)=16.7

= 16.7 Colwnn
(0.195)‘x Y 30 Volumn (1.511 I x 25
Grad. Duration = 33 min.
0.8 7 7
4OOmg Sample
E
a
$8 CM- 8
05
5
8
1
K
O-
0 5 10 15 20 25
3.0 -
8mg Sample
:
4
”
:
; 1.5-
I?
I
0 CA A
0 5 x) 15 20 25 30 35 40
Minutes
Figure 4.27. Laboratory and preparative scale separation of

Cytochrome C digest (Reference 40).
maintained (4 1). This occurs when the small and scaled-up

columns have the same L/D and DVp/p ratios, where L is
the column length; V is the flow rate per unit cross section;
D is the column diameter; p is the liquid denisty; and P is
the liquid viscosity. It is also important that the volume of
protein solution be kept at the same ratio to the volume of
adsorbent.
Example 4.5 (38)

A Sephadex column, 1.5 cm in diameter and 40 cm in
length, has been shown to purify 1 mL of protein solution at
a flow rate of 100 mL/hr. The column dimensions and flow
rate necessary to fractionate 200 ml of solution can be
calculated as follows:
In the small column,

A (1.512
Volume of adsorbent = 40 = 70.5 cm2 (4.12)
4
Cross section area = 1.76 cm2 (4.13)
Volume of protein solution 1

= (4.14)
Volume of adsorbent 70.5
L 40
- = ___ = 26.6 (4.15)
D 1.5
Since p/p will be the same in both columns, it is only

necessary that the values for DV be identical for both
columns.
Therefore,
100
DV = 1.5 x = a5 (4.16)
1.76
In the large column,
Volume of protcln solution 1 200
Volume of adsorbent
=-= __ (4.17)
70.5 Volume of ndsorbcnt
Volume of adsorbent = 14,100 cm2 (4.18)

L
__ = 26.6 or L = 26.6 D (4.19)
D
Therefore:
T D2
X 26.6 D = 14,100 (4.20)
4
20.9 D3 = 14,100 (4.21)
D = 9cm, L = 239 cm (4.22)
The flow rate is:

(D) (flow rate) 9 x flow rate
DV = 85 = = (4.23)
*/4 D2 H (912
Flow rate = 600 cm3 per hour (4.24)
Even though the cross-sectional area is 28 times greater

in the scaled-up system, the flow rate only is increased by a
factor of 6 to obtain the same elution pattern. Operating at
a higher flow rate presents a risk of inferior protein
separation.
Ladisch (42) has worked with a variety of column sizes

ranging from 2 to 16 mm in diameter and 10 to 600 cm in
length. His experience is that published semi-empirical scale-
up correlations are useful in obtaining a first estimate on
large scale column performance.
When scaling-up a chromatographic process, it may be

necessary to change the order of certain steps from that
used in the laboratory. Gel filtration, though a frequent
first step at the laboratory scale, is not suitable for handling
large scale feedstream volumes (43). When gel filtration is
used to separate molecules of similar molecular weights,
sample sizes may range from 1% to 5% of the total gel
volume. Thus a 100 liter feedstream would require a gel
filtration column of 2,000 to 10,000 liters. When one is
separating a large molecule from a small molecule, as in
desalting operations, the applied sample volume may be up to
30% of the total gel volume.
On the other hand, ion exchange chromatography is a

very good first step because of its capacity of approximately
30 mg of protein per mL of resin. This capacity is relatively
independent of feed volume. For the same 100 liter
feedstream, only a 20 liter ion exchange column would be
required.
4.7 INDUSTRIAL SYSTEMS
Packed bed, process scale high performance liquid

chromatography (HPLC) equipment was first introduced by
Millipore and Elf Aquitaine in 1982 (44). Since then, several
companies, listed in Table 4.10, have entered the large scale
HPLC market. The systems use packed beds at moderate
pressures (30- 140 bar). While there are substantial time
savings in using these systems compared to other purification
techniques, the short life of the packing material and its
high cost continue to restrict this technique to applications
that warrant the $lOO/kg separation cost.
Table 4.10: Manufacturers of Process Scale HPLC Equipment
Amicon Donvers, MA
Dorr-Oliver Stamford, CT
Elf Aquitaine (Varex in U.S.) Rockville, MD
Millipore Corporation Bedford, MA
Pharmacia Piscataway, NJ
Separations Technology Wakefield, RI
YMC Morris Plains, NJ
Example 4.6
The Waters Kiloprep Chromatography pilot plant is one

example of the successful extension of an analytical
chromatography process to the process scale. The ability to
control the various operation parameters to scale up directly
from the laboratory to the pilot plant and beyond to
commercial production has been developed (45). Figure 4.28
illustrates how the performance of this larger system can be
predicted from the data generated in an equivalent laboratory
apparatus.
Voser and Walliser (37) have described the approaches

different companies have devised so that fine and soft
adsorbents may be utilized in large scale chromatography
Column Chromatography F’rocesses 377
KILOPREP Unit Separation Process System

Load: 1SQgm
Flow Rate: 4 Umin
Column: (15cm x 60cm) x 2
(14 x PrepPAK Cartridge) x 2
r I I I 8 I 1
6 5 4 3 2 1 0
Time (min.)
PrepLC SOOA
/I Load:
Flow Rate:
Column:
15 gm
500 mUmin
S7mm * 3Ocm
cilica
6;
‘If\
limo (min.)
Begin Injection
Mobile Phase Sample
KILOPREP Flow Schematic

(Dynamic Column Length Control “Recycle’)
Figure 4.28. Comparison of column performance at equivalent

loading (Reference 45).
operations. The scaling-up has usually been achieved by

increasing the column diameter and using stack columns. The
problem then becomes one of achieving uniform distribution
of the feed solution over the entire resin bed surface,
particularly when gradient elution is involved.
In the Pharmacia approach, the fluid input is split and

distributed through six ports on the column end plates (46).
At the entrance of each stream, an anti-jetting device
spreads the liquid over a fine mesh net. A coarse net
between the net and the end plate acts as both a support
and a spacer. For the soft Sephadex gels G-50 to G-200, the
maximum feasible bed height is 15 cm. Scale-up operations
have used as many as six such squat columns in series (47).
The drawbacks of this approach are the cumbersome
adsorbent filling process, the possibility of air entrapment
because of the split flow and the inability to regenerate the
adsorbent in the column.
In Amicon/Wright columns, the flow is distributed

through a carefully designed system of radial ribs cut out of
the end plate. There is a single central port with a suitable
anti-jetting device to reduce and divert the high velocity of
the entering stream. The adsorbent bed is covered with a
sintered plate. Sintered plates are claimed to be more
efficient than mesh nets to achieve an even distribution.
The larger the pore size of the sintered plate, however, the
less efficient the system. Drawbacks of sintered plates are
their tendency to adsorb substances on their very large
surfaces and the possibility of fouling. These columns may
also be stacked.
Whatman has developed a column with a new flow

distribution system specifically designed for Whatman’s
cellulosic ion exchange resins. The bed height is 18 cm.
The diameter of the first commercial unit was 40 cm and
contained 25 liters of adsorbent. The resin bed is covered
with a perforated plate with a high free surface. The
slightly conical head plate covers an empty space and has a
steep cone in the middle. The total empty head space is
about 5% of the bed volume. The feed solution enters the
steep cone tangentially in its upper part. The resulting
rotary movement efficiently mixes the suprnatant liquid and
allows gradient elution. It is claimed that filling and
equilibration take only one hour.
AMF has developed an unconventional new approach

with its ZETA-PREP cartridge. The cartridges consist of
concentric polymer screens which bear the ionic groups and
are supported by cellulosic sheets. The flow is radial from
the outer rim toward a perforated central pipe. The
available nominal cartridge lengths are from 3 cm to 72 cm

with a constant diameter of about 7 cm throughout. Scaling
up with this approach is quite straightforward. Single
cartridges, each mounted in a housing, can be combined to a
multi-cartridge system. For such a system, flow rates up to
12 L/min and bovine serum albumin capacities of 1400 g are
claimed. The present ion exchange functionalities available
are DEAE, QAE and SP.
The stepwise transition from high pressure liquid

chromatography to medium pressure chromatography, such as
described for the preparation of pectic enzyme (48), illustrate
the progression toward large scale industrial application of
the techniques developed in analytical laboratories. The
pressure in these medium pressure chromatography
applications is only 6 bar instead of the 100 to 150 bar
associated with HPLC. The lower pressure results in longer
processing times (about one hour) compared to the 5 to 20
minutes required for an analytical determination with HPLC.
Studies, such as the one by Frolik and coworkers (48),

which examine the effect and optimization of variables in
HPLC of proteins, can be expected to contribute to the
implementation of this type of protein resolution technique
into future commercial biotechnology processes.
The first chromatographic systems capable of handling

more than 100 kg/day were merely scaled-up versions of
laboratory chromatography (50,5 1). Even with some of these
systems it was necessary to recycle a portion of the overlap
region to have an economical process. A typical example of
such a system would be the Techni-Sweet System of
Technichem (52) used for the separation of fructose from
glucose. The unique distributors and recycle system are
designed to maximize the ratio of sugar volume feed solution
per unit volume of resin per cycle whiie at the same time
minimizing the ratio of volume of water requried per unit
volume of resin per cycle.
The flow through the Technichem system is 0.56 ms/(hr-

m3) with a column height of 3.05 m. The feed solution
contains 45% dissolved solids, and a feed volume equal to 22%
of the volume of the resin is added to the column each
cycle. The rinse water added per cycle is equal to 36% of
the volume of the resin. This is much less rinse water than
the 60% volume that was required by the earlier systems.
This technique is known as the stationary port

technique since the feed solution and the desorbent solution
are always added at the same port and the product streams
and the recycle stream are always removed from another
port. Technichem and Finn Sugar manufacture
chromatography systems which utilize the stationary port
technique.
One of the earlier attempts (53) at industrial

chromatography used an adaptation of the Higgins contractor
for the ion exclusion purification of sugar juices. The
physical movement of the low crosslinked resin caused
attrition as it was moved around the contactor. It was also
difficult to maintain the precise control needed on flow rates
because of the pressure drop changes and volume changes of
the resin as it cycled from the mostly water zone to the
mostly sugar solution zone.
An alternate approach (54) utilizes moving port or

pseudo-moving bed techniques. With these techniques, the
positions on the column where the feed solution is added and
where the product streams are removed are periodically
moved to simulate the countercurrent movement of the
adsorbent material. At any given time the resin column can
be segmented into four zones (Figure 4.29). Zone 1 is called
the adsorption zone and is located between the point where
the feed solution is added and the point where the fast or
less strongly adsorbed component is removed. In this zone
the slow or more strongly adsorbed component is completely
adsorbed onto the ion exchange resin. The fast component
may also be adsorbed, but to a much smaller extent. The
second zone, Zone 2, is the purification zone and is located
between the point where the fast component is removed and
the point where the desorbent solution is added. Zone 3 is
called the desorption zone and is between the point where
the desorbent is added and the point where the slow
component is removed. In this zone the slow component is
removed from the resin and exits the column. The final
zone, Zone 4, is called the buffer zone and is located
between the point where the slow component is removed and
the point where the feed solution is added. There is a
circulating pump which unites the different zones into a
continuous cycle.
Different sections of the column serve as a specific

zone during the cycle operation. Unlike the stationary port
technique, the liquid flow is not uniform throughout the

column. Because of the variations in the additions and
withdrawals of the different fluid streams, the liquid flow
rate in each of the zones will be different.
With such a system one must slowly develop the

chromatographic distribution pattern through the different
zones. It may take from 8 to 36 hours for the pattern to be
established. Another practical consideration is that the
recirculation system must represent a small (~10%) portion of
a single zone to prevent unacceptable backmixing which
would alter the established chromatographic pattern.
The flow rate and the pressure drop per unit length of
the chromatographic column are much lower for the
stationary port compared to the moving port system. Also
the stationary port method is much less caapital intensive.
The moving port technique, however, is calculated to require
only one third of the column volume and ion exchange
volume and two thirds of the desorbent volume compared to
the stationary port technique.
r-+-L,
I c \
Product “A”
I
Rinse Water
Product “B”
,
1
I
,
,
,
Feed =
J
Figure
kc 4.29. Moving port chromatographic column with four
zones for continuous chromatographic separation (Reference
54).
After the expiration of the UOP patent involving the

rotary valve, there have been several modifications to the
moving port technique by Amalgamated Sugar (55), Illinois
Water Treatment (56) and Mitsubishi (57). Each manufacturer
has its own approach for the establishment and control of
the chromatographic pattern. These are the subject of
confidentiality agreements for specific applications.
Broughton and Gembicki have used a model of

equilibrium theoretical trays with entrainment to describe
mathematically the chromatographic separation of pseudo-
moving bed operations (58). Two parameters are adjusted
with this model to reproduce the experimental concentration
profiles:
Number of theoretical trays: n = K k H/Lz (4.25)
Axial. mixing ratio: e = E/L = Kk/DL2 (4.26)
where
Z = m In (4.27)
K is the linear equilibrium constant; k is the mass transfer

coefficient; H is the bed height; L is the liquid flow rate; m
is the adsorption factor which is equal to the pore
circulation rate K/L; E is the entrainment rate and D is the
axial diffusion coefficient. Values of n and e are calculated
for each zone since each zone is treated separately according
to the component whose concentration is changing most
rapidly. It should be noted that this model views both the
adsorbent and the eluent as moving. Therefore it is
necessary to adjust the flow rate of the eluent to account
for the stationary position of the adsorbent when carrying
out experiments based on the model results.
A comparative mathematical modeling of the pseudo-

moving bed system with the conventional chromatography
system has shown that the conventional system requries 3 to
4 times more adsorbent and twice as much desorbent solution.
The reason for this is that in the pseudo-moving bed system,
every part of the bed can be shown to be performing useful
work at all times, with respect to the primary function of
each zone. In the conventional system, on the other hand,
some portion of the bed at all times is not contributing to

the separation process.
Barker and Thawait (59) have used the theoretical plate

model to show that the concentration of a component in a
given plate element is given by:
C” = c”_l 1:“*)(4*28)
J) +c::exp(“,-:
(’ - exp(“y+y,
where C, is the component concentration in the mobile phase
leaving plate n; Q is the mobile phase flow rate; At is the
time increment; VI is the mobile (fluid) phase volume of a
plate; V, is the stationary (adsorbent) phase volume of a
plate and K, is the distribution coefficient. Each component
can be considered individually assuming there is no
interaction between components. The sequential nature of
the switching valves is modeled by stepping the concentration
profile backwards by one zone at the end of the switch
period. The typical switch time is 5 to 15 minutes for
commercial units. The one difficulty with using the model is
the selection of a distribution coefficient since the measured
value changes with concentration. Figure 4.30 shows the
agreement between experimental results and those calculated
using Equation 4.27.
l Glucose
m Fructose
-- Simulated
- Experimental
0.5
Feed column
Purge column
5 t
; 0.3
F r-
g 0.2
Y
9 0.1
0.0 5
‘;; 0.4
0 1 135 201 269 337 405 473 540 607 E
Distance along column (cm)
Figure 4.30. Experimental and simulated concentration profile

for glucose-fructose separation (Reference 59).
Wankat (60) proposed a hybrid system which has some

of the characteristics of both elution chromatography and the
pseudo-moving bed system. During the feed pulse, the feed
position was moved continuously up into the column at a
velocity that lies between the two solute velocities. The
eluting solvent was continuously fed into the bottom of the
column. Elution development with solvent was used when the
feed pulse was over. This method reduces irreversible mixing
of solutes near the feed point. Wankat and Ortiz (61) have
used this system for gel permeation chromatography and
claim improved resolution, narrower bands and higher feed
throughputs compared to conventional systems. McGary and
Wankat (62) have had similar results applying it to
preparative HPLC. His technique uses less adsorbent and
produces more concentrated products compared to normal
preparative chromatography, but more adsorbent and less
concentrated products than pseudo-moving bed systems.
Wankat (63) has proposed that his system will be of most
value for intermediate size applications or when only one
product is desired.
The key items to identify when considering an industrial

chromatographic project are the capital for the equipment,
yield and purity of the product, the amount of dilution of
the product and waste stream, the degree of flexibility the
computer controls allow, the expected life of the ion
exchange material and whether the equipment allows for
periodic expansion of the resin.
New techniques are continuing to be developed which

can be expected to be used in future specialized industrial
applications. Multi-segmented columns have been
demonstrated for the preparative purification of urokinase
(64). Begovich and coworkers (65,66) have developed a
technique for continuous spiral cylinder purifications which
may allow separation on the basis of electropotential in
addition to the selective affinity of the adsorbent resin for
the components in solution. A schematic of this device is
shown in Figure 4.31.
Another new technology that offers promise for

commercial biotechnology purifications is the use of
parametric pumping with cyclic variations of pH and electric
field. This has been described by Hollein and coworkers (67).
They worked with human hemoglobin and human serum
albumin protein mixtures on a CM-Sepharose cation
ECUCYT ¶TALAYI
.Z?.S m ANWJLUS QO.
- SCCARATl?O
COMSTITUCNTS
CLULYT WASTE
COLLLCTIoM-
LXIT-
L‘
ELUATE
ANWULAA IEO
l.2? cm a A0 cm
Figure 4.31. Schematic of the pressurized continuous annular

chromatograph (Reference 65).
exchanger. The extensive equations they reported for

parametric separations allow analysis of other systems of two
or more proteins which may be candidates for this type of
separation.
4.8 CHROMATOGRAPHY APPLICATIONS
Column chromatography techniques have been

incorporated into the commercial purification scheme for
fermentation products, biomaterials and organic chemicals.
While the majority of these applications are on a small scale
(less than 500 kg/month but greater than 10 g/month),

several large industrial scale applications have arisen in the
last decade. The extraction of sugar from molasses, the
separation of glucose from fructose, the separation of
polyhydric alcohols, the separation of xylene isomers and the
separation of amino acids are carried out in industrial scale
operations preparing thousands of metric tons of purified
material each year. Examples of pilot scale and large
industrial chromatography will now be given.
4.8.1 Amino Acid Separation

Preparatory scale separation of complex amino acid
mixtures was reported by Moore and Stein (68). Today amino
acid mixtures obtained from fermentation or from protein
hydrolysis are separated by specific elution regimens after
the amino acids are adsorbed on the ion exchange resin, as
reported in Chapter 3. Although many of the commercial
operations use standard ion exchange resins, such .as Dowex
HGR or Duolite C-25, product dilution and yield could be
substantially improved by using the improved separation
resins.
The use of ligand exchange with acrylic resins has been

shown to provide advantages over conventional ion exchange
resins in separating amino acids (69). Copper(R) carboxylate
functionality provided the best kinetic performance and
separation.
The separation of d- and l-isomers of amino acids may

also be accomplished by using a derivatized copper form of
the iminodiacetate ion exchange resin (70). Contrary to
earlier assumptions, it is not necessary to have an optically
active functionality on the ion exchange resin to separate
optical isomers. With the increased importance of l-
phenylalanine and 1-tryptophan, it is expected that industrial
chromatographic separations will soon include the separation
of these amino acids from their d-isomers. Table 4.11 shows
the results at the preparatory scale.
Hamilton (71) has measured the effect of resin

crosslinking on the selectivity for amino acids. This is
shown in Figure 4.32. Since commercial resins may vary
*OS”h in crosslinking from one batch to another, this could
be an important factor in assuring reproducible separations
when different batches of the same resin are used. This
would especially be the case for the separation of alanine

from glucosamine in a system designed for use with an 8%
crosslinked resin.
Table 4.11: Separation and Recovery of d- and l-Isomers (70)
Recovered in Residual Solids in

Pure Streams Overlap Region
Amino Acid %d %l 70 d,l
dJ-Tyrosine 82.4 19.2 19.2
d,l-Phenylalanine 100.0 93.0 3.6
d,l-Hystidine 97.0 100.8 1.1
d,l-Tryptophan 98.8 93.6 3.8
d,l-C&tine 91.0 100.0 1.5
I I I I
7.0
lp Valine
6.0
5.0
K*
4.0
3.0
2.0
1.0
0 2
Taurine
Cysteic Acid
0 r.!
I I I;: I
0 2 4 6 6 10 12
Cross - Linking (Density) (%)
Figure 4.32. Selectivity coefficient (KA) for amino acid by a

sodium form cation resin as a function of the resin
crosslinking (Reference 7 1).
4.8.2 Antibiotics
One of the earliest chromatographic purifications of
antibiotics was the separation of neomycin B from neomycin
C using alumina (72). Weston and Putter (73) have used
strongly hydrophilic anion resins (Sephadex A25), weak base
resins (Amberlite IRA 68) and non-ionic crosslinked
macroporous polystyrene polymer beads (XAD 2) to purify
antibiotics in chromatography columns containing as much as
45 liters of adsorbent.
Aureomycin, an anti-tumor antibiotic, was isolated from

a fermentation broth (74) by successive application of ion
exchange chromatography with Amberlite IRA 93 and DEAE
Sephadex, gel filtration on Sephadex G-50 and hydrophobic
chromatography on Octyl-Sepharose CL-4B. The initial 64
liters of harvested broth was purified and concentrated to a
430 mg lyophilized powder.
Silica gel was used in the adsorption chromatography

purification of carminomycin (75). This is an anthracycline
antibiotic with known anti-tumor properties. The elution was
carried out using chloroform and methanol at a 17 to 3
volume ratio. The fractions collected contained less than 1%
impurities.
Cephalosporin antibiotics produced by fermentation of

Streptomyces sp. S3907C were purified by chromatography
with QAE Sephadex in the chloride form (76). A gradient
elution of lithium chloride (0 to 10%) was applied in a linear
manner. A crude fraction of 19 g from the 400 liter broth
resulted in a purified active powder of 226 mg.
Sakuma and Motomura (77) have separated saikosaponins

a, c and d in preparative quantities using ODS silica gel
chromatography. These saponins, shown in Figure 4.33, have
been used in anti-inflammatory medicinal applications. A
three step gradient elution with acetonitrile and water gave a
recovery of more than 90% with less than 1% impurity in
each of the components. From the initial 10 g of crude
saponin, 403 mg of saikosaponin c, 1210 mg of saikosaponin a
and 1604 mg of saikosaponin d were obtained in a single
preparative chromatogram.
4.8.3 Glucose-Fructose Separation

Glucose-fructose syrups may be obtained by the
&cjJ?2H
&j& @$a 2
fUCOW
31
glUCOwgglucOw
‘I
Pow
I
glucose rhamnow
lkJcow
saikosaponin a saikosaponin c saikosacmin d
Figure 4.33. Chemical structure of saikosaponins (Reference

77).
inversion of sucrose or by the enzymatic conversion of

starch. While rice and wheat starch have been used to make
glucose-fructose syrups, the most common starch source is
corn. The resulting gyrup is called high fructose corn syrup
(HFCS). The enzymes convert the starch to a syrup that
contains 55% glucose, 42% fructose and 3% oligosaccharides.
While that level of fructose has sufficient sweetness for some
applications, it is necessary to increase the fructose
concentration to at least 55% of the dissolved solids in syrup
to have a sweetness comparable to sucrose. The major
industrial chromatographic application in the United States is
this separation of HFCS into an enriched fructose stream and
a glucose stream.
The separation of fructose from glucose was first

reported by LeFevre (78,79) and Serbia (80) in 1962 using the
Ba, Sr, Ag and Ca form of 4% crosslinked cation resin. The
glucose and the higher saccharides do not form as strong a
complex with these ions as the fructose does, so the glucose
and higher saccharides appear first in the effluent, followed
by the glucose-fructose mixture and then the relatively pure
fructose. The Ca form of the resin is preferred in industrial
chromatography.
Ghim and Chang (81) have reported the effect of

different ionic forms of anions and cations in separating
fructose from glucose using the moments of the elution
curve. Table 4.12 shows the distribution ratios determined by
them. It should be noted from the distribution coefficients
that the order of elution would be fructose and then glucose
for the anion resins and glucose followed by fructose for the
cation resins. While it is possible to separate fructose from
glucose using anion resins, the industrial chromatography

applications only use cation resins. This is probably because
the cation resins generally are more durable and less
expensive than anion resins.
Table 4.12: Effect of Counterion on the Distribution Ratio (81)

(Dowex l-X8 was used for anion resin and Dowex 5OWX8 for the cation
exchanger. Resin size was 200-400 mesh)
Ionic Form Glucose Fructose
-2
so3 0.45 0.27
-2
co3 1.43 1.00
H*PO; 0.16 0.08
Ca++ 0.30 0.80
Sr++ 0.30 0.80
Zn++ 0.20 0.30
Recently, zeolite and resin adsorbents have been

compared in the chromatographic separation of glucose-
fructose mixtures (82). The resin adsorbents were seen to
have a higher equilibrium separation factor than the CaY
zeolite. However, the resins offer greater resistance to mass
transfer, resulting in little difference in overall performance
between zeolites and resins in pseudo-moving bed and pulse
chromatographic systems.
In spite of the great practical interest in the separation

of glucose and fructose, there are only a few reports (83)
covering their separation in industrial processes. Most
available references deal with one or another specific
chromatographic process (52,57,84) for carrying out the
separation. These processes were covered in Section 4.7.
4.8.4 Glycerol Purification

One of the first industrial chromatographic separations
proposed was for the purification of glycerol (85). In this
case, elevated temperatures (80°C) were found to aid the
separation of glycerol from salt since aqueous glycerol
solutions are quite viscous at the concentrations necessary

for practical operation. Figure 4.34 shows the separation
profile for a solution of 10% NaCl and 36.3% glycerine.
< WASTE
L2 DILUENT
I
I
1.0
RESIN: DOWEX 50x8
088 FEED: 0.35 VT OF
36.3% GLYCERIN
006
004
002
0
0 o-2 0.4 0.6 0,B 180 142 104 1.6
(VOLUME OF ELUATE COLLECTED)/(VOLUME OF RESIN BED)
Figure 4.34. Separation profile for a solution of 10% NaCl

and 36.3% glycerol (Reference 85).
Sargent and Rieman (86) proposed a chromatographic

method for separating non-ionic organic solutes from one
another by using aqueous electrolyte solutions as the eluent.
As Figure 4.35 shows, this would lead to several changes per
cycle in the composition of the eluent. This approach has
only been feasible for laboratory or small scale preparatory
separations. Recently, studies with ammonium carbonate
solutions as the eluent for separating fermentation products
containing amino acids showed that various strengths of the
eluent were effective in amino acid separations and that
(NH&W, could be separated economically from the
individual amino acid streams as the gaseous NH, and CO,
and recycled for the next elution cycle.
4.8.5 Oligosaccharide Removal

The presence of 3 to 5% oligosaccharides in the high
fructose syrup necessitates their removal during industrial
chromatography of glucose and fructose since the
4-A/o/or 2.4-Ah/or I
2-Ma/or O./-MO/or
L
ht.J~ so, - ; (NM4 )* JO,
0 100 200 300 400 500 600 700 800

E/T/uen f /+kdm/!
Figure 4.35. Elution profile and desorbent changes for the

separation of alcohols (Reference 86).
oligosaccharides would decrease the sweetness of fructose or

would hinder the enzymatic isomerization of the glucose. A
process for removing oligosaccharides has been described by
Hirota and Shioda (87). They used a strong cation resin in
the calcium form to obtain a fructose stream that had 94 to
97% of the dissolved solids as fructose and less than 2%
oligosaccharides. The glucose stream had a purity of 79 to
89% with an oligosaccharide concentration of 9 to 20%. The
oligosaccharide stream had a dissolved solids oligosaccharide
content of 64 to 76%. While this may not seem acceptable
from an analytical chromatography standpoint, the low level
of oligosaccharides in the feed solution (7 to 14Oh) and the
high purity of the fructose product stream make the process
acceptable for industrial chromatographic removal of a waste
component. Such a process would also utilize the same resin
and equipment that is used to separate the fructose from the
glucose.
4.8.6 Peptides and Proteins
The now classic procedure of Moore and Stein (68) has

been applied to the analytical separation of thousands of
biochemical preparations and to the isolation of preparative
quantities of biological materials (Table 4.13). Additional
reviews of these activities have been described by Weaver
(118), by Morris and Morris (119) and by Gordon (120).
Table 4.13: Proteins Purified by High Performance Ion Exchange

Chromatography (88)
PROTEIN COLUMN RESIN TYPE REFERENCES
LactaLc dchydrogenase DEAE-glycophase weak base 89,90,91,92,93

isoenzymes SynChropak AX300 weak base 91
Creatlnc kinase DEAE-qlycophase weak base 92,93.94,95
isoenzymcs Synchropak AX300 weak base 9L,96
Alkaline phosphatases DEAE-qlycophase week base 97
Hexokinase isoenzymes Synchropak AX300 weak base 98
Arylsulfatase 'I DEAE-glycophase weak base 99
Hemoglobins Synchropak AX300 weak base 100,101,102,103
DEAE-glycophase weak base 8!3,90
IEX 545 DEAE weak base 104
IEX 535 CM weak cation 104

BIORex 70 weak cation 105
Cytochrome C CM-polyamide weak cation 106

Lysozyme CM-polyamide weak cation 106
Myoglobl" CM-polyamide weak cation
106
IEX 535 CM weak cat&on 107
Soybean trypsin
inhlbltor CM-glycophase weak cation 90
Interferon Partisil SCX strong cation 108
Lipoxyqenase SynChropak AX300 weak base 109
Trypsln DEAE-glycophase weak base 90
Chymotrypslnogen SP-glycophase strong cation 90

IEX 535 CM weak cation 107
Immur~oglobulin G SynChropak AX300 weak base 110
Ovalbumln SynChropak AX300 weak base 111

Albumi" SynChropak AX300 weak base 111

DEAE-glycophase weak base 90
Apollpoprotclns SynChropak AX300 weak base 113,114

Adunylsuccinate
synthctase SynChropak AX300 weak base 115
Irlsull" Pnrtisil SCX strong cation 116

IEX 535 CM weak base 107
,f?-Lactoglobulin Partisil SCX strong cation 116
Carbonic anhydrase Partisil SCX strong cation 116
Monoamine axldase SynChropak AX300 weak base 117
DEAE and CN-glycophase are products of Pierce Chemical

Company. Rockford, Illinois.
SynChropak AX300 IS produced by SynChrom. Linden, Indiana.
IEX 535 CM and 545 DEAE are the products of Toya Soda
Company. Yamaguchi, Japan.
Blo-Rex 70 is supplied by Bio-Rad. San Francisco, California.
Partlsil SCX IS manufactured by Whatman, Clifton, New Jersey.

Kelley and coworkers (121) have examined the use of

gel chromatography for the industrial purification of proteins.
Those applicatons reported only included those with column
volumes greater than 4 liters (Table 4.14). The first seven
examples (122- 127) used soft, compressible gels, like Sephadex
G-200 and Ultrogel AcA34. Productivities varied from 0.0016
to 0.045 liters feed solution per liter of gel for a 20 hour
day (L/L/d). The average value was about 0.25 (L/L/d. The
next three examples (128- 130) used less compressible gels,
such as Sepharose 4B, Sephadex G-75 and G-50. The
productivities were on the order of 0.1 L/L/d, about 4 times
higher than for the compressible gels. Finally, the
productivity of a continuous annular chromatograph (131) was
found to be approximately 0.37 L/L/d, 15 times higher than
the compressible gels. This points out that the restricted
productivities often associated with gel chromatography are
the result of the predominant use of soft, compressible gels.
Table 4.14: Large Scale Gel Chromatography Application (121)

Protc1n Separation Ded Demensions Productivity Reference
Medium (D x L)
(cm x cm) (1 feed/l gel/
day)
E. coli
-__ Sephadex C-ZOO 21.5 x 80 0.026 122
Iso-leucyl-t-KNA
trensferase
E. coli Sephadex G-ZOO 14 x 180 0.0016 123
--.
Ncthlonyl-t-RNA
transferase
Citrobocter Sephadrx G-200 4.2 L 0.007 124
L-asparaginase
Rape Seed Proteins Sephadex C-200 45 x 05 0.025 125
‘Transferrin from Sephndcx G-200 45 x 75 0.019 126
Cohn Fraction IV
Plasmil Alkaline Ultrogel AcA 34 16 x 100 0.045 127
Phosphatilse
Plasma Ultroeel AcA 34 9.3 x 90 0.026 127
Cholinesterase
Thyrq:lobulins Sepharose 40 37 x 45 0.125 128
Insulin Sephodex C-50 37 x 90 0.071 129
Whcv Protein Sephadex C-75 37 x 15 0.128 130
Concentrate
Continuous, annular Sephadex G-75 0.37 131
Chromatograph
The chromatographic purification of recombinant protein

has been achieved with minimal effect on its tertiary
structure and no effect on its biological activity using
peptide fusion (132). Sassenfeld and Brewer used rDNA
technology to produce human urogastrone fused to a C-
terminal polyarginine. This fused polypeptide allowed a two-
step chromatographic purification using SP-Sephadex C-25.
In the first step, a substantial purification was obtained

due to the high basicity of the polyarginine fused protein.
The polyarginine was then removed using carboxypeptidase.
The second pass of the digested material through an SP-
Sephadex C-25 column gave urogastrone at 95% purity with a
yield of 44O/6 in a pilot plant (32 liter feedstream)
purification. The resolution from the previously co-
chromatographed contaminants is shown in Figure 4.36.
P
/
P
A
/
0 40 80 120 160 200

NACL Ml)
Figure 4.36. Purification of urogastrone on SP-Sephadex.

Peak A is polyarginylurogastrone and Peak B is peak A after
digestion with carboxypeptidase B and rechromatographing
(Reference 132).
The use of non-functionalized resins (133) such as XAD-

4 has been reported for the separation of amino acids and
polypeptides. The authors were able to identify specific acid
and base eluents that allowed di- and tripeptide separation at
high purity and in good yield. The separation of a mixture
of enkaphalin peptides is shown in Figure 4.37. While this
work is still at the preparatory scale, it is anticipated that
pharmaceutical companies will be utilizing this technique and
other laboratory chromatographic techniques to obtain pure
polypeptides for medicinal purposes.
D
K
& Prepcmtive Seporotion Fraction Analysis
Figure 4.37. Analysis of feed and enkaphalin peptide

fractions separated on XAD-4 (Reference 133).
Kato and coworkers (134) have examined the effect of

eluent gradient, column length and sample loading on the
retention and resolution of the proteins in commercial
ovalbumin. The study used TSK-GEL DEAE ion exchange
resins with a 10 micron diameter. While the two column
lengths studied had little influence on column performance,
the effect of eluent gradient was higher resolution as the
gradient time became longer and lower retention as the
gradient became less steep (Figure 4.38). The sample loading
needs to be less than 1 to 2 mg per unit column cross
section (cm2) in high performance ion exchange
chromatography. As the sample loading increased, the peak
width became wider, elution was faster and the peak shape
began tailing.
Dizdaroglu and coworkers (135) used a silica-based

bonded-phase weak anion exchanger to isolate peptides with
up to 30 residues. Recoveries of over 80% were obtained
because of the volatility of the eluent (gradient of increasing
O.OlM triethylammonium acetate buffer at pH 6.0 into
acetonitrile) employed. A comparison of this approach with
reverse phase HPLC showed the differences in the
selectivities for each method (136). The combined application
of these two different separation principles provides a
greater opportunity for the complete separation of a given
mixture of peptides into individual components.
z
P
’ Slope .o.b-0.7\
1
’ ’ 1 ltrtl , I ll,,l1
1 10
Gradient (mM NaCl/min)
I I I ,,I‘
= 10. 1 * 1 ’ *,a’ , t IIII

1 10
Gradient (mM NoCUmin)
Figure 4.38. Dependence of resolution (R&, peak interval

(v, - V,) and peak width (W,) on eluent concentration
gradient change rate (Reference 134).
The recently developed Mono Q and Mono S resins from

Pharmacia have been used in the recovery of several
enzymes, while maintaining the enzyme activity. This is
shown in Table 4.15 (137). That same article describes the
use of eluent selection which can facilitate the use of these
resins to purify and recovery proteins. The Mono S resin
has been reported to isolate the five separate components of
purified human bronchial proteinase inhibitor (138).
Table 4.15: Enzyme Activity Recoveries from Mono Q and Mono S

Resins (137)
Mono Q Mono S
uecovery, % Recovery, %
Protein
___- Relative Activity Relative Activity
B- Glucuronidase 106% -___
B- Clucosidase ___- 93%

Phosphodiesterase 80% ____
Creatine kinase 90% -_--
Enolasc -___ 95%

Lactate dchydrogenase _--- 102%
Aldolase _--_ 94%
Carboxylic ion exchange resins have been used in the

separation of proteinaceous materials using the resin’s
capability of dissociating only in certain pH ranges (139). At
acidic conditions below pH 4.5, ion exchange resins based on
methacrylate or acrylate polymers are undissociated. Thus,
proteins may be adsorbed on the resins at pH 4.5. Following
adsorption of the protein mixture, individual protein fractions
may be separately removed from the column by the controlled
increase of the pH of the eluent (Figure 4.39).
Adenosine, obtained by the fermentation of a Bacillus

strain, ATCC No. 21616, was purified by column
chromatography (140). The supernatant from the
fermentation broth was adjusted to pH 4.0, then passed
through a column of activated carbon for adsorption of
adenosine. The column was washed with water; then the
adenosine was eluted with solution of
methanol/isooctanol/ammonia/water (50: 1:2:4?). The eluate is
concentrated and then separated by chromatography with an
anion exchange resin (Dowex 1, chloride form) to get a pure
adenosine fraction. That fraction is concentrated and then
crystallized from ethanol.
FRACTION NUMBER
Figure 4.39. Example of the use of carboxylic acid resins in

the purification of protein fractions (Reference 139).
There are few large scale applications in protein

purification which utilize classical adsorption chromatography
with alumina, silica or charcoal as the separation media. One
example has been reported for the large scale purification
with celite of @-lactamases from Bacillus cerus culture
supernatant (141). A 20-fold purification and a IO-fold
concentration was possible with this procedure.
Marshall (142) used a column of DEAE cellulose,

followed by one of Ultragel AcA34, one of Sephadex G-100
and finally one of DEAE-Sepharose in his industrial scale
separation of a specific glucoamylase-type of enzyme to be
used in the production of low alcohol beverages. This four
column procedure allowed the isolation of cr-amylase, cr-
glucosidase, glucoamylase and a glucoamylase- type of enzyme,
exopullulanase, which has the ability to degrade pullulan at a
rapid rate.
Phosphocellulose and Sephadex C-25 have been used to

isolate micrococcal nuclease from a crude culture of
Staphylacoccus aweus after 16 to 18 hours of growth. The
cation Sephadex showed significantly higher levels of
recovery compared to the phosphocellulose as shown by the
elution results in Table 4.16 (143).
Table 4.16: Recovery of Micrococcal Nuclease from SP-Sephadex

C25 and Phosphocellulose (143)
Resin PH % Adsorption % Elution
SP-Sephadex C25 5.0 98 a2
SP-Sephadex C25 5.8 97 80
Phosphocellulose 5.0 95 30
Phosphocellulose 5.8 90 31
Large scale purifications of Trichoderma reesei strain

L27 cultures have been routinely performed by applying the
filtrate to a column of DEAE-Sepharose CL-6B and eluting
the fractions with a NaCl gradient (144). A second DEAE-
Sepharose CL-6B column was used to isolate
exocellobiohydrolase I and endoglucanase. A cation SP-
Sephadex C-SO column was used to purify P-glucosidase
further. The recovery of these proteins was 83%.
Additonal examples, mostly of laboratory studies, are

available in books specifically on the HPLC of peptides and
proteins (145,146). “LC GC” and “Chromatography” are two
periodicals with helpful HPLC operational suggestions.
4.8.7 Polyhydric Alcohol Separation

There have also been industrial chromatographic
developments in the separation and purification of
monosaccharides and polyhydric alcohols from lignocellulosic
materials such as wood (147,148). These authors were able to
separate xylitol and sorbitol from galactitol, mannitol and
arabinitol using a resin in the calcium form. The aluminum
form of the resin could be then used to separate xylitol from
sorbitol. As a further refinement of this double fractionation
process, the saccharide fraction was removed to allow the
purification of xylose from mannose, glucose and galactose.
Other variations on these process schemes may be utilizied to
recover other polyhydric alcohols or monosaccharides.
4.8.8 Regenerant Recovery

The regeneration of ion exchange resins results in
solutions which contain varying concentrations of the waste
salts stripped from the resin and of the acid or base used
for resin regeneration. Ion retardation can be used to
separate these waste salts from the regenerant acids and
bases. Special resins have been developed (149) which
facilitate this separation. Figure 4.40 shows the effect of
different resins in separating acids and salts. This method of

regenerant recovery entails such capital expenditure for
columns, controls and resin that only very large resin
installations, producing large volumes of regenerant-salt
waste stream, are economically justifiable. The high fructose
corn syrup industry and large power installations are such
applications.
0.6
pJ?jm/
~ FiETARDlON RETARDION RETARDIfffd
0.6
RE TARDION
DOWEX I (for 1)
1 ’ ’ ’ ’ I 1 ’ ’ 1 1 J
0 0.4 0.8 0 0.4 0.8 0 0.4 0.8
v,= VOLUME
VOLUME
OF ELUATE
OF RESIN
COLLECTED
BED
‘b
Figure 4.40. Effect of different resins on the elution profiles

for various acids and salts (Reference 149).
There are two other situations which would warrant ion

retardation systems: 1) when valuable salts must be separated
from the eluent which stripped the salts from a concentrating
ion exchange column or from a solvent extraction solution
and 2) when ecological coinsiderations limit the acid, base or
salt waste streams which may otherwise be discharged into
the environment.
4.8.9 Sugar From Molasses

Molasses is a by-product in normal extraction and
crystallization processing of cane sugar and beet sugar. The
sugar content in molasses is usually between 30 and 60% as a
percentage of the total dissolved solids. While it is possible
to decrease this sucrose loss to molasses by typical ion
exchange treatment to remove the salts prior to
crystallization, that process leads to substantial waste loads
(150).
When the molasses solution is added to the

chromatographic column, the highly ionized ash is excluded
from the resin beads while the non-ionic sugars are
selectively adsorbed on the active sites of the resin beads.
The distribution constants are given in Table 4.17 for K and
Ca forms of the resin for various salts and sugars of
molasses. In addition to the non-sugar inorganics, there are
several non-sugar organics in the molasses. These non-sugar
organics are different depending upon whether the source of
the molasses is cane sugar or beet sugar. The non-sugar
organics from cane molasses have a smaller amount of amino
acids and a higher content of colored components resulting
from the degradation of reducing sugars (glucose, glactose)
compared to beet molasses.
Table 4.17: Distribution Coefficients for Dowex 50 WX4,

50-100 Mesh (150)
K+ FORM ca++ FORM
KC1 0.454 0.482
CaCl* 0.466 0.530
Sucrose 0.623 0.623
Glucose 0.955 0.690
Fructose 1.07 1.26

Sargent (151) studied the variables of column shape,

density gradients and feed concentrations for ion exclusion
purifications of molasses. Stark (152) showed that both beet
and cane molasses could be purified using Dowex 5OWX4 in
the K ion form. Fifty percent of the sucrose in beet
molasses was recovered into a fraction with about 80% purity.
Over 65% of the sucrose in cane refinery molasses was
recovered with over 68% purity. At these levels of purity,
the sucrose fractions could be returned to intermediate pans
for sucrose recovery by crystallization. Takahashi and
Takikawa (153) showed that ion exclusion combined with a
recycling technique and under optimal conditions would yield
an apparent purity of 90% and a sucrose recovery of 90% for
beet molasses. Houssiau (154), using an ion exchange resin
in the K form on a second strike molasses stream, was able
to upgrade a sucrose stream at 72% purity to 95% purity at
minimum loading, but only up to 80% purity at maximum
column loading. The dissolved solids level of the recovered
sucrose fraction was reduced to 6 - 18”Brix when the
molasses feed stream was at 40”Brix. An engineering analysis
(155) of the ion exclusion process for recovering sucrose
from beet molasses showed that, under certain cost
projections for molasses and sucrose, this process would be
economically justified. Gross (50) reviewed many of these
studies along with early patents on chromatographic
separations of molasses.
Ito (156) and Hongisto (157) each have examined the

two alternatives of separating molasses into a non-sugar
fraction and a total sugar fraction (mixture of sucrose,
glucose and fructose) or of inverting the sucrose of the
molasses into glucose and fructose and then separating the
inverted molasses into a non-sugar fraction and a glucose-
fructose fraction.
In Ito’s studies, the K form of Dowex 5OWX4 was used

to separate molasses at 44”Brix with a 42% sugar
concentration into a non-sugar fraction, a sucrose fraction,
and a glucose-fructose fraction. The sucrose fraction and
the glucose-fructose fraction together recovered 77% of the
sucrose and 96% of the glucose-fructose with a purity of 82
to 89%. When the inverted molasses was separated by a K
form resin, 76.7% of the glucose-fructose was recovered at a
93.2% purity. When the Ca form of the same resin was used,
the separations were not as good.
Hongisto’s studies showed that inverting the molasses

prior to separation will increase the capacity of a
chromatographic column without a decrease in product purity
because of the sharp separation of non-sugar from
monosaccharides on a Na form resin. The purity obtained by
Hongisto’s process was between 92 and 96% with a recovery
between 85 and 92%.
Many studies have been undertaken recently to examine

the potential of isolating other non-sucrose organics from
molasses by chromatography. For a Na form resin, the
raffinose peak occurs before, while glucose and fructose
peaks occur after, the sucrose peak. Of the main non-sugar
components, betaine is eluted last.
Sayama and his coworkers (32) used a Ca form of

Dowex 5OWX4 to study the chromatographic separation of
betaine from beet molasses. The effects of solids level, flow
rate, column temperature and ionic composition of the resin
on peak resolution were examined. For this type of
separation, resolution increased with increasing temperature.
The effects of other variables on resolution were as
expected: decreasing concentration, decreasing flow rate and
increasing percentage Ca form resin increased betaine
resolution.
To separate sucrose from molasses, a recent study (158)

that examined resin crosslinkage, ionic form, flow rate and
feed load determined that optimum conditions were the Na
form of Dowex 50MW4, 50- 100 mesh, with a flow rate of 1.3
space velocity and a feed load of 0.125 liters of 30”Brix
molasses per liter of resin per cycle. The chromatographic
system could treat 2 tons of molasses per cubic meter of
resin per day with a recovery of 85% of the sucrose at a
purity of 87.5%.
Sayama and his coworkers also worked out procedures

for recovering inositol (159), raffinose (160), and adenosine
(161) from molasses. Over 70% of the inositol was recovered
using a Ca form resin. For raffinose recovery, the Na form
of the resin was superior to the Ca form. Depending on the
sucrose to raffinose ratio, from 66% to 86% of the raffinose
could be recovered at a purity of 91 to 93%. The adenosine
recovered on a pilot scale chromatographic unit was 30%
when betaine was recovered in a separate fraction. The Ca
form resin could be loaded with 0.250 m3 of 4O”Brix molasses
per m3 of resin per cycle during the recovery of betaine and

adenosine.
One of the most recent developments in the area of

chromatographic treatment of molasses is the work of Neuzil
and Fergin (162) who have applied the UOP continuous
chromatography technique to recover sucrose from molasses.
In this patent, a non-functionalized adsorbent and an alcohol
containing eluent are utilized. From a model feed stream
with 30% sucrose, 10% KC1 and 10% betaine, a sucrose
recovery of 90% with a purity of 99% was obtained.
4.9 REFERENCES
1. Tswett, M., Ber Deut Botan Ges 24:316 (1906)
2. Tswett, M., Biochem Z 56.
3. Seaborg, G.T., Science 104(2704):379 (Oct. 25, 1946)
4. Spedding, F.H., Fulmer, E.I., Butler, T.A., et al., J Am

Chem Sot 69:2812 (1947)
5. Spedding, F.H., Powell, J., J Am Chem Sot 76:2545

(1954)
6. Spedding, F.H., Powell, J., Wheelwright, E., J Am Chem

& 76:2557 (1954)
7. Sitrin, R.D., Chan, G., DePhillips, P., et al., In:

Purification of Fermentation Products, (LeRoith, D.,
Shiloach, J., Leahy, T.J., eds.), American Chemical
Society, Washington, DC, p. 71 (1985)
8. Porath, J., Flodin, P., Nature 183:1657 (1959)
9. Porath, J., Lab Pratt 16:838 (1967)
10. Moore, J.C., J Polvm Sci, Part A 2:835 (1964)
11. Cuatrecasas, P., Wilcheck, M., Anfinsen, C.B., Proc Nat

Acad Sci 61:636 (1968)
12. Wheaton, R.M., Bauman, W.C., Ann NY Acad Sci 57:159

(1953)
13. Dean, J.A., Chemical Separation Methods, D. Van

Nostrand, New York p 295 (1969)
14. Eisenbraun, A.A., U.S. Patent No. 3,045,026 (July 17,

1962)
15. Dean, J.A. Chemical Separation Methods, D. Van

Nostrand Co., New York, p 60 (1969)
16. Janson, J.C., presented at BIOTECH 84 Europe

Conference, London (1984)
17. Roy, A.K., Roy, S., Liauid Chromatonraphv l(3): 181

(1983)
18. Schomberg, G., LCGC 6(1):36 (1988)
19. Wilson, J.N., J Am Chem Sot 62:1583 (1940)
20. DeVault, D., J Am Chem Sot 65:532 (1943)
21. Glueckauf, E., J Chem Sot 69:1321 (1947)
22. Glueckauf, E., Discussions Faraday Societv 7:42 (1949)
23. Martin, A.J.P., Synge, R.L.M., Biochem J 35:1358 (1941)
24. Mayer, S.W., Thompkins, E.R., J Am Chem Sot 69:2866

(1947)
25. Simpson, D.W., Wheaton, R.M., Chem Ena Prog, 50(1):45

(1954)
26. Glueckauf, E., Trans Faradav Sot 51:34 (1955)
27. Beukenkamp, J., Reiman, W. III, Lindenbaum, S., Anal

Chem 26:505 (1954)
28. Simpson, D.W., Bauman, W.C., Ind Enp Chem 46:1958

(1954)
29. Karger, B.L., J Chem Educ 43:47 (1966)
30. Hongisto, H.J., Int SuPar J 79(940):100 (1977)

31. Martinola, F., Siegers, G., Verfahrenstechnik (Mainz),

13(1):32 (1979)
32. Sayama, K., Senba, Y., Kawamoto, T., Proc Res SOC
Japan Sugar Refin Tech 29:lO (1980)
33. Warren, W., Dwyer, M., Merion, M., J Analysis and

Purif
-, 2(2):10 (1987)
34. Semenza, G., Chimica 14:325 (1960)
35. Mikes, O., Strop, P., Hosomska, Z., et al., J Chromatog,

261:363 (1983)
36. Sofer, G., Mason, C., Biotechnolopv 5(3):239 (1987)
37. Voser, W., Walliser, H.P., in Discovery and Isolation of

Microbial Products, (Verrall, MS., ed) Ellis Horwood
Ltd., Chichester, UK, p 116 (1985)
38. Charm, S.E., Matteo, C.C., in Methods in Enzvmoloav

(Jacoby, W.B., ed) Academic Press, New York, vol 22, p
476 (1971)
39. Laboratory Methods of Column Packing, Whatman Data

Sheet, Reeve-Angel Co., Clifton, New Jersey (1968)
40. Stacey, C., Brooks, R., Merion, M., J of Analysis and

Purif
-3 2( 1):52 (1987)
41. Charm, S.E., Matteo, C.C., Carlson, R., Anal Biochem,

3O:l (1969)
42. Rudge, S.R., Ladisch, M.R., in Senaration. Recoverv and

Purification in Biotechnology (Asenjo, J.A., Hong, J.,
eds.) American Chemical Socikty, Washington, DC, p 122
(1986)
43. Sofer, G., Mason, C., Biotechnologv, 5(3):239 (1987)
44. Heckendorf, A.H., Chem Proc, p. 33 (August 1987)
45. Heckendorf, A.H., Ashare, E., Rausch, C., In:

Purification of Fermentation Products, (LeRoith, D.,
Shiloach, J., T.J., Leahy, eds.) American Chemical
Society, Washington, DC, p 91 (1985)
46. Janson, J.C., Dunnill, P., In: Industrial Aspects of

Biochemistrv, American Elsevier, New York, p 81 (1974)
47. Strobel, G.J., Chemie Technik, 11:1354 (1982)
48. Rexova-Benkova, L., Omelkova, J., Mikes, O., et al., 1

Chromatoaraphy, 238: 183 (1982)
49. Frolik, C.A., Dart, L.L., Sporn, M.B., Anal Biochem,

125:203 (1982)
50. Gross, D., Proc 14th Gen Assemblv CITS, p 445 (197 1)
51. Sutthoff, R.F., Nelson, W.J., U.S. Patent No. 4,022,637

(May 10, 1977)
52. Keller, H.W., Reents, A.C., Laraway, J.W.,

Starch/Staerke, 33:55 (1981)
53. Norman, L., Rorabaugh, G., Keller, H., J Am Sot SuPar

Beet Tech, 12(5):363 (1963)
54. Broughton, D.B., Gerhold, C.G., U.S. Patent No. 2,985,589

(May 23, 1961)
55. For more information contact K. Schoenrock,

Amalgamated Sugar, Ogden, Utah
56. Burke, D.J., presented at A.1.Ch.E. 23rd Annual

Symposium (May 12, 1983)
57. Ishikawa, H., Tanabe, H., Usui, K., Japanese Patent No

102,288 (August 11, 1979)
58. Broughton, D.B., Gembicki, S.A., In: Fundamentals of

Adsorption, (Myers, A.L., Belfort, G., eds.) Engineering
59. Barker, P., Thawait, S., Chem & Ind, 817 (Nov 7, 1983)
60. Wankat, P.C., Ind Ene. Chem Fundam, 16:468 (1977)
61. Wankat, P.C., Ortiz, P.M., Ind Ene, Chem Process Des
m, 21:416 (1982)
62. McGary, R.S., Wankat, PC., Ind Ena Chem Fund, 22:lO
(1983)
63. Wankat, P.C., Ind Ena Chem Fund, 23:256 (1984)
64. Novak, L.J., Bowdle, P.H., U.S. Patent No. 4,155,846

(May 22, 1979).
65. Begovich, J.M., Byers, C.H., Sisson, W.G., Sep Sci

Technol, 18(12 & 13):1167 (1983)
66. Begovich, J.M., Sisson, W.G., A.I.Ch.E.J., 30:705 (1984)
67. Hollein, H.C., Ma, H., Huang, C., et al., Ind Ena Chem
Fundam, 2 1:205 (1982)
68. Moore, S., Stein, W.H., J Biol Chem, 211:893 (1954)
69. Doury-Berthod, M., Poitrenaud, C., Tremillon, B., 1

Chromatopr, 179:37 (1979)
70. Szczepaniak, W., Ciszewsha, W., Chromatopraphia,

15(1):38 (1982)
71. Hamilton, P.B., Anal Chem, 35:2055 (1963)
72. Dutcher, J.D., Hosansky, N., Donin, M.H., et al., J Am

Chem Sot, 73:1384 (1951)
73. Weston, R.G., Putter, I., U.K. Patent No. 1,345,729 (Feb
6, 1974)
74. Yamashita, T., Naoi, N., Hidaka, T., et al., J Antibiotics

(Tokvo), 32(4):330 (1979)
75. Johnson, D.L., Doyle, T.W., U.S. Patent No. 4,189,568

(Feb 19, 1980)
76. Brown, D., Giles, A.F., Noble, H.M., et al., U.S. Patent
No. 4,379,920 (April 12, 1983)
77. Sakuma, S., Motomura, H., J Chromatoar, 400:293 (1987)
78. LeFevre, L.J., U.S. Patent No. 3,044,905 (July 17, 1962)
79. LeFevre, L.J., U.S. Patent No. 3,044,906 (July 17, 1962)
80. Serbia, G.R., U.S. Patent No. 3,044,904 (July 17, 1962)
81. Ghim, Y.S., Chang, H.N., Ind Eng Chem Fundam, 21:369
(1982)
82. Ho, C., Ching, C.B., Ruthven, D.M., Ind Eng Chem Res,
26:1407 (1987)
83. Bieser, H.J., deRosset, A.J., Starch/Staerke, 29:392 (1977)
84. Miyahara, S., Sakai, S., Matsuda, F., et al., Japanese

Patent No. 118,400 (Sept 11, 1980)
85. Asher, D.R., Simpson, D.W., J Phvs Chem, 60518 (1956)
86. Sargent, R.N., Rieman, C.W. III, U.S. Patent No.

3,134,814 (May 26, 1964)
87. Hirota, T., Shioda, K., Japanese Patent No. 48,400 (April
7, 1980)
88. Regnier, F.E., In: Hinh Perform Lia Chromatoar Proteins

Peut, Proc Int Svmp. 1st 1981, Academic Publications,
New York, NY (Pub. 1983)
89. Chang, S.H., Gooding, K.M., Regnier, F.E., J Chromatoar,

125103 (1976)
90. Chang, S.H., Noel, R.N., Regnier, F.E., Anal Chem,

48:1839 (1979)
91. Schlabach, T.D., Alpert, A.J., Regnier, F.E., Clin Chem

24:1351 (1978)
92. Schlabach, T.D., Fulton, J.A., Mockridge, P.B., et al.,

Clin Chem, 25:1600 (1979)
93. Schlabach, T.D., Fulton, J.A., Mockridge, P.B., et al.,

Anal Chem, 52:729 (1980)
94. Fulton, J.A., Schlabach, T.D., Kerl, J.E., et al., L

ChromatoPr, 175:269 (1979)
95. Denton, MS., Bostick, W.D., Dinsmore, S.R., et al., Clin

Chem, 24: 1408 (1978)
96. Bostick, W.D., Denton, M.S., Dinsmore, S.R., Clin Chem

26:712 (1980)
97. Schlabach, T.D., Chang, S.H., Gooding, K.M., et al., l

Chromatoar, 134:91 (1977)
98. Alpert, A.J., Ph.D., Thesis, Purdue University (1979)
99. Bostick, W.D., Dinsmore, S.R., Mrochek, J.R., et al., C&

Chem 24:1305 (1978)
-I
100. Gooding, K.M., Lu, L.C., Regnier, F.E., J Chromatogr,

164506 (1979)
101. Hanash, SM., Shapiro, D.N., Hemoalobin, 5165 (1980)
102. Hanash, SM., Kavadella, K., Amanulla, A., et al., In:

Advances in Hemoplobin Analvsis, (Hanash, S.M., Brewer,
G.J., eds) A.R. Liss, New York, NY, p 53 (1981)
103. Gardiner, M.B., Wilson, J.B., Carver, J., et al.,

International Svmosium on HPLC of Proteins and
Peptides, Paper No. 203, Washington, D.C. (1981)
104. Umino, M., Watanabe, H., Komiya, K., et al., ibid., Paper
No. 208, Washington, D.C. (1981)
105. Abraham, E.C., Cope, N.D., Braziel, N.N., et al., Biochem

Biouhvs Acta, 577:159 (1979)
106. Gupta, S.P., Pfannkoch, E., Regier, F.E., Anal Biochem

128(1):196 (1983)
107. Umino, M., Watanabe, H., Komiya, K., Ref. 103, Paper
No. 204
108. Radhakrishnan, A.N., Stein, S., Licht, A., et al., L

Chromatoar, 132:552 (1977)
109. Vanecek, G., Regnier, F.E., Anal Biochem, 121(1):156

(1982)
110. Lu, K.C., Gooding, KM., Regnier, F.E., Clin Chem,

25:1608 (1979)
111. Vanecek, G., Regnier, F.E., Anal Biochem, 109:345 (1980)
112. Kato, Y., Komiya, K., Hashimoto, T., Ref. 103, Paper No.
214
113. Alpert, A.J., Beaudet, A.L., Ref. 103, Paper No. 210
114. Ott, G.S., Shore, V.G., Ref. 103, Paper No. 201
115. Rudolph, F.B., Clark, S.W., Ref. 103, Paper No. 202
116. Frolick, C.A., Dart, L.L., Sporn, M.B., Ref 103, Paper
No. 205
117. Ansari, G.A.S., Patel, N.T., Fritz, R.R., et al., Ref. 103,
Paper No. 211.
118. Weaver, V.C., Chromapraphia, 2555 (1969)
119. Morris, C.J.O.R., Morris, P., Separation Methods In

Biochemistrv, Interscience Publ, New York, NY (1963)
120. Gordon, A.H., In: Laboratorv Techniaues in Biochemistrv

and Molecular Biology Vol 1, Part 1, (Work, T.S., Work,
E., eds.) North Holland Pub1 Co, Amsterdam - London
(1969)
121. Kelley, J.J., Wang, G.Y., Wang, H.Y., In: Separation,

Recovery and Purification in Biotechnology (Asenjo,
J.A., Hong, J., eds.) American Chemica; Society,
Washington, D.C., p 193 (1986)
122. Durekovic, A., Flosdorf, J., Kula, M.R., Eur J Biochem,

36528 (1973)
123. Bruton, C., Jakes, R., Atkinson, T., Eur J Biochem,

59:327 (1975)
124. Bascomb, S., Banks, G.T., Skorstedt, M.T., et al., J Gen

Microbial, 91:l (1975)
125. Janson, J.C., J APr Food Chem, 19:581 (1971)
126. Curling, J.M., Amer Lab, 8:26 (1976)

127. Hanford, R., Maycock, W., Vallet, L., In:

Chromatographv of Svnethic and Biological Polvmers,
Vol 2, (Epton, R., ed) Ellis Harwood, Chichester, U.K., p
111 (1978)
128. Horton, T., Amer Lab, 4:83 (1972)
129. The Large-Scale Purification of Insulin bv Gel Filtration

Chromatonraphv, Pharmacia, Uppsala (1983)
130. Donnelly, E.B., Delaney, R.A.M., J Food Technol, 12:493

(1977)
131. Nichols, R.A., Fox, J.B., J Chromatogr, 43:61 (1969)
132. Sassenfeld, H.M., Brewer, S.J., BiotechnoloPy, 2( 1):76

(1984)
133. Pietrzyk, D.J., Cahill, W.J., Jr., Stodola, J.D., J Liauid

Chrom, 5(3):443 (1982)
134. Kato, Y., Komiya, K. Hishimoto, T., J Chromatoar,

246(1):13 (1982)
135. Dizdaroglu, M., Krutzsch, H.C., Simic, M.G., 2

Chromatopr, 237:4 17 (1982)
136. Dizdaroglu, M., Krutzsch, H.C., J Chromatopr, 264:223

(1983)
137. Richey, J.R., Amer Lab, 14:104 (1982)
138. Boudier, C., Carvallo, D., Roitsch, C., et al., Arch

Biochem Biophvs, 253(2):439 (1987)
139. Sasaki, I., Gotoh, H., Yamamoto, R., et al., J Biochem

(Tokvo), 86(5): 1537 (1979)
140. Komatsu, K., Saijo, A., Haneda, K., et al., U.S. Patent
No. 3,730,836 (May 1, 1973)
141. Davis, R.B., Abraham, E.P., Melling, J., Biochem J,

143:115 (1974)
142. Marshall, J.J., U.S. Patent No. 4,318,927 (March 9, 1982)

143. Darbyshire, J., In: Topics in Enzyme and Fermentation

Biotechnoloav 5, (Wiseman, A., ed) Ellis Harwood,
Chichester, U.K., p 156 (1981)
144. Shoemaker, S., Watt, K., Tsitovsky, G., et al.,

Biotechnoloav, l(S):687 (1983)
145. Horvath, C., ed., High Performance Liauid

Chromatography, Vol 3, Academic Press, New York
(1983)
146. Hamilton, R.J., Sewell, P.A.,. Introduction to Hiah

Performance Liauid Chromatography, 2nd Ed., Chapman
and Hall, London (1982)
147. Melaja, A.J., Hamalainen, L., U.S. Patent No. 4,008,285

(Feb 15, 1977)
148. Melaja, A.J., Hamalainen, L., U.S. Patent No. 4,075,406

(Feb 21, 1978)
149. Hatch, M.J., Dillon, J.A., I&EC Proc Des Dev, 2:253
(1963)
150. Rousseau, G., Lancrenon, X., Sugar v Azucar, 27 (1984)
151. Sargent, R.N., Ind Eng Chem Proc Des Dev, 2(2):89
(1963)
152. Stark, J.B., J Am Sot Sunar Beet Technol, 13:492 (1965)
153. Takahashi, K., Takikawa, T., Proc Res Sot Jaoan Suaar
Refin Tech, 16:51 (1965)
154. Houssiau, J., Sucrerie Belne, 87:423 (1968)
155. Schultz, W.G., Stark, J.B., Lowe, E., Int Suaar J, 69:~
(1967)
156. Ito, Y., Proc Res Sot Japan Sugar Refin Tech, 22:l
(1970)
157. Hongisto, H.J., S.I.T. Paper 78, 624 (1978)
158. Sayama, K., Senba, Y., Kawamoto, T., Proc Res JaDan
SuPar Refin Tech, 29:l (1980)
159. Sayama, K., Senba, Y., Kawamoto, T., et al., Proc Res
JaPan Sugar Refin Tech, 29:20 (1980)
160. Sayama, K., Senba, Y., Kawamoto, T., et al., Proc Res
JaPan Sugar Refin Tech, 30:72 (1981)
16 1. Oikawa, S., Sayama, K., Senba, Y., et al., Proc Res

JaDan Sugar Refin Tech, 33:55 (1982)
162. Neuzil, R.W., Fergin, R.L., U.S. Patent No. 4,426,232 (Jan
17, 1984)
5
Affinity Chromatography
5.1 INTRODUCTION
Affinity chromatography was developed by Cuatrecasas,

Wilchek and Anfinsen (1) in 1968 as a laboratory protein
purification method that is based on the biospecific
recognition between a solute molecule and a ligand
immobilized onto a support material. As shown in Figure 5.1
(2), the affinity purification process can be idealized into the
selective adsorption of enzyme A from a mixture of proteins.
After the unadsorbed contaminating proteins are washed from
the column, enzyme A can be eluted from the ligand and
recovered. This technique has been used successfully to
purify enzymes, proteins, antibodies, oligonucleotides and
nucleic acids. Several recent review articles and books are
available which cover these applications (2-7).
Generalized protein purification schemes are shown in

Figure 5.2 (8) where conventional ion exchange and
chromatography techniques are compared to affinity
chromatography. The conventional techniques separate the
protein on the basis of size, electrical charge or isoelectric
point. Since these differences are often not sufficient to
allow effective separation, many additional steps are required.
This greatly reduces the overall yield of purified protein
compared to the affinity chromatography technique.
The “chromatography” portion of the term affinity

chromatography does not represent a true description of the
416
Affinity Chromatography 41’7
process as actually practiced. More correct descriptive terms

would be “affinity purification” or “affinity separation”.
However, for the sake of consistency with past authors on
this subject, this text will continue to use “affinity
chromatography.”
SOLID SUPPORT, SUCH AS

AN AGAROSE BEAD
cl
HIXTURE OF PROTE INS
SPACER ARH,
SUCH AS HEXAMETHYLENE ATTACHMENT OF
DIAMINE DESIRED ENZYME
WASHING BUFFER
REMOVAL OF
CONTAMINATING PROTEINS
ELUTION BUFFER
RECOVERY OF
PURIFIED ENZYME A
REUSABLE AFFINITY MATRIX
Figure 5.1. A schematic of the selective, reversible

attachment of a protein to an immobilized ligand (Reference
2).
cells
t
Disrupt
t
Extract
t
Centrifuge
A \
Affinity column Acetone precipitation
t t
Dialysis or Protamine treatment
molecular sieving
t
(NH,I$O, Fractionation
t
Dia/ysis
Hydroxyapatite
chromatography
t
Dialysis
t
Ion-exchange cellulose
chromatography
t
Molecular sieving
Figure 5.2. Comparison of conventional and affinity

chromatography techniques (Reference 8).
5.2 THEORY OF AFFINITY CHROMATOGRAPHY
When compared to the vast literature available on

empirical affinity chromatography studies, there is a paucity
of theoretical work that has been reported explaining affinity
chromatography. One reason is that the rigorous models
developed for analyzing chromatographic systems require
linear adsorption-desorption kinetics which do not occur for
the affinity purification systems. Another complicating factor
is the interference caused by non-specific adsorption which
contributes complexity and oftentimes irreproducibility to
many affinity chromatography systems.
Affinity Chromatography 419
Despite these difficulties, attempts to describe binding

selectivity, equilibrium constants and column kinetics have
provided useful insights into affinity purification processes.
5.2.1 Equilibrium and Binding Selectivity

The ideal ligand for affinity chromatography would
possess the ability to bind strongly the solute to be isolated
and the binding should be highly specific so that impurities
with similar structures are not bound. The binding should
also be reversible under readily obtained conditions so that
the isolated solute can be easily eluted from the ligand.
Thus, the binding strength, the selectivity and the elution
behavior constitute the controlling factors in a quantitative
description of affinity chromatography.
The binding strength and the selectivity are termed the

avidity of the ligand for the solute (9). The solute becomes
the ligate when it is bound to the ligand. The avidity is
described mathematically by the association constant, K,,:
K
as
=-=
k
a IEL I (5.1)
kb
for the interaction of the ligand [L] and the ligate [El:
+ L
!a._ EL
E - (5.2)
kb
The adsorption of the ligate by the ligand can be represented
by the Langmuir-type isotherm (10):
(5.3)
where
(5.4)
High values of KU? represent an extremely stable

complex. Typical values are 10 r5 M for the avidin-biotin
interaction, IO6 to 1012 for antigen-antibody interaction and
lo4 to lo6 for lectin-carbohydrate and enzyme-substrate
interactions (11).
In most cases, there are additional materials which will

compete with the desired ligate for ligand sites. In all cases,
such competition occurs when it becomes necessary to elute
the desired ligate from the ligand. These aspects of affinity
chromatography theory have been developed by Bottomley and
coworkers (12). When the competing ligate [D] is present in
the solution, the following equation can be used to describe
the relevant equilibria:
IE I =
1
+-------II
E
(5.5)
IEL I K
as ILclI ’ KED l Lo I
In this equation, Km, represents the dissociation constant for

a complex formed between the two potential ligates in
solution:
- (5.6)
The parameter r represents the ratio of the partition

coefficients cyeq and an. The partition coefficient accounts
for the possibility that there are sections of the column
which are accessible to the solvent but not to the large
protein molecule, E.
Where for [E] and [EL] are obtained from the elution
curve (Figure 5.3). The shaded area in this figure represents
the amount of enzyme E bound to the ligand L. The values
for [E] and [EL] from experiments with different
concentrations of E at constant concentrations of D can be
plotted as in Figure 5.4 to give straight lines whose abscissa
values can be used to obtain values for K,, and K,n.
Elution volume
Figure 5.3. Elution profile of enzyme B applied to columns
of immobilized A. The hatched area represents the amount
of B bound to A. This is displaced from the column by some
competing ligand C at concentration D (Reference 12).
Figure 5.4. Primary plot for the three component system.

Values of B, and B, are obtained from the chromatography
experiments of Figure 5.3 at different concentrations (D1, D,,
Da, etc.). The slopes of all the lines should be the Same
(Reference 12).
Another useful quantity is the fraction Of the tOta1

enzyme which will be bound to the ligand at equilibrium.
This is given by the equation (9):
Bound Enzyme iLo1--v

=__= 14
~0ta1 Enzyme Ki(V + v) + [Loi" + pj7 (5*7)
PI "
where Ki = l/K,,, V is the volume of enzyme solution added

and v is the volume of solution entrapped within the affinity
support particles. As is seen from Figure 5.5, the resulting
curve for any value of Ki is a rectangular hyperbola similar
to the Langmuir adsorption isotherm or the Michaelis-Menten
equation. Typical ligand concentrations used in affinity
chromatography experiments are 1OmM. At this ligand
concentration, Ki values of 10e3M or less are necessary for
effective enzyme binding; higher values would result in
ineffective retention of the enzyme.
Figure 5.5 can also be used to evaluate the fractional

saturation for a system by substituting [E]V for the quantity
[L,]v on the abscissa and using K (V + v) + [L,]v instead of
K (V t v) t [E]V as the variable parameter. The fraction
[EL]/[L,] then is simply read on the ordinate scale. These
values are given in Table 5.1 as percentages for several
representative values of Ki, [E] and w, where w is the ratio
of v to v.
0
0 I 2 3 4 5 6 7 8 9 IO II 12 13 I4 I5 I6 17 IS 19 20
TOTAL LIGANDCONCENTRATIONNJ, r)l
Figure 5.5. Fractional enzyme binding for low enzyme

concentration (Reference 9).
Table 5.1: Percentage Ligand Saturation Under Various Conditions

(For [L,] = 0.01 M)
[Eel Ki [EL]/[L.I (X)
(M) CM) w=1 w= 10 w = 100
10-4 10-3 0.8264 4.5455 a.2645
10-4 0.9709 8.2645 33.2226
10-5 0.9881 9.0009 47.5964
10-6 0.9899 9.0818 49.7488
10-5 10-3 0.0833 0.4739 0.8929
10-4 0.0979 0.8929 4.7393
10-5 0.0997 0.9794 0.3264
10-6 0.0999 0.9890 9.0082
10-6 10-3 0.0083 0.0476 0.0900
10-4 0.0098 0.0900 0.4950
10-5 0.0100 0.0988 0.9001
10-6 0.0100 0.0998 0.9803
10-7 10-3 0.0008 0.0048 0.0090

10-4 0.0010 0.0090 0.0497
10-5 0.0010 0.0099 0.0907
10-6 0.0010 0.0100 0.0989
Example 5.1
From Table 5.1, it is seen that when [E] = 10m6 M, Ki =
10m3 M and V = v, only about 83% of the enzyme in solution
will bind to the ligand even though less than 0.1% of the
ligand capacity is exhausted. The lower experimental

capacity of affinity columns compared to titratable ligand has
usually been attributed to steric exclusion or ionic repulsion.
It is seen from this example that equilibrium effects alone
may be a sufficient explanation.
Washing and elution by changing Ki to new values were

also studied by Graves and Wu (9). Ki values of 10-s M
looked much less attractive than when binding alone was
considered because much of the enzyme was lost by washing.
If Ki was 10m5 M or less, batch adsorption and elution were
as good or better than column chromatography. When Ki was
changed to a new value during the elution step, the increase
had to be at least 10m2 M in order to remove enzyme
effectively from the ligand. Achieving such a high value can
be a serious limitation if the initial Ki is very low.
The number of washing steps and the volume of each

wash have as their primary effect an influence on the
removal of contaminant. The degree of purification was
strongly influenced by the thoroughness of washing. If Ki
was less than low4 M, very little enzyme was lost even with
a large amount of washing. This is illustrated in Figure 5.6
for simulated column chromatography.
rTDi:SHES 1 - ELUTIONS
El 3 5 7 9 II I3 15 17 19 21 23 25 27 29 30
TOTAL WASHES AND ELUTIONS, (XV’+ CM I/v
Figure 5.6. Simulated column results for moderately good

binding (Ki = 10m4M) (Reference 9).
5.2.2 Kinetics of Affinity Purification

In Chapter 4, the plate theory was used to model the
column chromatography behavior. However, that theory is
only applicable where a linear isotherm can be assumed
because the plate height is a function of the fixed linear
equilibrium constant or partition ratio (13). This is not the
situation for affinity chromatography systems since
equilibrium isotherms of such systems are usually highly
curved. This points out one of the dangers of using the
term affinity chromatography rather than affinity purification
to describe these systems since it has led to the
inappropriate use of plate theory with these systems. As
Yang and Tsao (14) have pointed out, it is necessary to
choose a model with appropriate assumptions for the affinity
system studied. The porous particle, constant pattern rate
theory (15) and the statistical moment theory (16), although
more mathematically complicated, are more appropriate for
describing affinity purification systems.
The mechanism of mass transfer in affinity

chromatography is similar to what was observed with column
adsorption in Chapter 2. It can be divided into three
successive steps: the diffusion of the solute to the particle,
its diffusion inside this particle and the reaction between the
ligand and the ligate. Since the ligand-ligate reaction occurs
rapidly, the two diffusion steps are considered to control the
rate of mass transfer.
The mathematical description of the column kinetics for

affinity chromatography follows the same type of material
balance as was seen for volume elements in fixed bed
adsorption. With the assumption of uniform velocity
distribution over the cross section of the column and an
isothermal operation, the basic equation is written as (17):
dC aC dc a2c
u-- +c----- +
%
--
DZ
___2 = 0 (5.8)
dZ dt at i3Z
The axial diffusion coefficient, is essentially *z,the

eddy diffusion since the effect of molecular diffusion can be
neglected in affinity chromatography. The value of Ed u/D,
is constant when the Reynolds number is between 10” and
10. This would give:
DZ = hd u (5.9)
P
where X is a constant.
The adsorption rate can be expressed in terms of the

volumetric coefficient of the overall mass transfer based on
the fluid phase concentration, Kra:
Kf a (c - C*) (5.10)
“bt =
With the assumption of a constant pattern adsorption zone, zF_

can be set equal to (~Jc,)c so that Equation 5.10 can be
integrated to give:
* (5.11)
tE - tB =
where K,a is the average volumetric coefficient of overall

mass transfer over the range between cn and cE. If the
Freundlich adsorption isotherm, S = KcP, adequately describes
the system, an analytical integration of Equation 5.11 yields
(18):
'b Go XE B
In- * ___ In (5.12)
'E - ‘Ii = fi c 1-o
f 0 xB
Example 5.2
The breakthrough curve data for trypsin on Sepharose
4B columns with soybean trypsin inhibitor ligand (Figure 5.7)
and arginine peptide ligand (Figure 5.8) have been used with
Equation 5.12 to calculate the average values of the
volumetric coefficients of the overall trypsin transfer. These
values are shown in Table 5.2: The fact that changes in
trypsin concentration or in ligand caused little change in K>
indicates that the reaction rate of trypsin with the ligand
was not rate limiting. The finite difference method of Crank
and Nicolson (19) was used to generate the calculated
breakthrough curves. Note that the measured breakthrough
curves are somewhat steeper than the calculated curves at
the beginning of breakthrough. This has been attributed to
the variation of the intraparticle mass transfer coefficient
with the progress of adsorption.
The very tight binding of the ligate during the

adsorption step in affinity chromatography means that the
volume of etfluent , cm3
Figure 5.7. Breakthrough curve for trypsin on Sepharose 4B

columns with soybean trypsin inhibitor ligand (C, = 0.190
mg/cm3, liquid flow rate is 0.109 cm3/sec, column diameter is
2 cm) (Reference 17).
I I
0cm.-------- O
I,0 - 0
//
/
volume of effluent, cm’

Figure 5.8. Breakthrough curve for trypsin on Sepharose 4B
columns with arginine peptide ligand (C, = 0.116 mg/cm3,
liquid flow rate is 0.056 cm3/sec, column diameter is 2 cm)
(Reference 17).
Table 5.2: Volumetric Coefficients (Kfa) for Sepharose 4BSoybean

Trypsin Inhibitor and Sepharose 4B-Arginine Peptides Systems (17)
-
Flow Kate Trypsin Kfa
Concen!jration
-1
Lip,and (cnJ/sec) (w/cm ) (set )
Soybean trypsin 0.056 0.078 0.0157

inhibitor 0.056 0.176 0.0138
0.109 0.083 0.020s
0.109 0.190 0.0189
Arginine peptides 0.056 0.009 0.0162
0.056 0.011 0.0148

0.056 0.019 0.0134
0.056 0.074 0.0168
0.056 0.116 0.0150
0.135 0.041 0.0210
0.135 0.048 0.0177
0.135 0.065 0.0146
shape of the breakthrough curve depends only on the rate

limiting mass transfer (or reaction) mechanism. Arnold and
coworkers (20) have presented analytical expressions for the
breakthrough curves for four cases in terms of the number of
transfer units (N) and a dimensionless throughput parameter
(T) when the particles had a 0.01 cm diameter, the superficial
velocity was 0.01 cm/set and the solute bulk diffusivity was
7 x 10e7 cm2/sec.
The breakthrough curve for the irreversible case with

pore diffusion is:
N pore CT - 1) = 2.44 - 0.273 /m (5.13)
where X is the dimensionless liquid phase concentration

(C/CFEEDL
N :
27 (DEFF/DnULK) L (5.14)
pore
The effective diffusivity (D EFF) ’

IS usuaby l/l 00 of the value
of the bulk diffusivity (Dnu,,) (21). L is the column bed
length and
T = (V - fV)/TV (5.15)
with V the throughput volume, v the column volume, E the

void fraction and P the dimensionless distribution parameter.
The breakthrough curve for the solid homogeneous

diffusion case is:
NP(T-1) = -1.69
I
ln(l - X)2 + 0.61
1 (5.16)
where
N 2 21r (Dp/DBuLK) L (5.17)

P
with D, the pore diffusivity.
The combination of pore and film diffusion results in a

breakthrough curve given by:
N
@(X) + pore (In x + 1)
Nf
(5.18)
N
pore
+ 1
I
Nf
where
d)(X) 2 2.39 - 3.59 lfC-7 (5.19)

and
Nf z 21 L (5.20)
The combination of solid homogeneous diffusion and film

mass transfer has a breakthrough curve given by:
OCX(E (5.21)
Nf (T - 1) = -mNp(T-1) =a
m .ln (l - *) + 1 + m; 69( x 6 1 (5.22)

(1 - 8)
where
m = - Nf/N (5.23)
and
a = l/(1 + N /N ) (5.24)
f P
Example 5.3
Typical breakthrough curves for the adsorption of
bovine serum albumin conjugated to arsanilic acid are shown
in Figure 5.9, on controlled pore glass and in Figure 5.10, on
Sepharose gel. The ligand for each column was mouse

monoclonal anti-benzenarsonate IgG.
1.0
X
0.5
Figure 5.9. Experimental and calculated breakthrough curve

for bovine serum albumin on 1.5 x 16.5 cm controlled pore
glass-monoclonal antibody column (Reference 20).
Figure 5.10. Experimental and calculated breakthrough curve

for bovine serum albumin on 1.5 x 18.3 cm Sepharose 4B-
monoclonal antibody column (Reference 20).
The experimental curve for the controlled pore glass

columns was better modeled using the pore diffusion equation
(5.13) with Npore = 8. However, the breakthrough curve for
the Sepharose column showed a greater tendency to tail off
at the upper end and was better modeled using the solid
diffusion equation (5.16) with N, = 5.
5.3 AFFINITY CHROMATOGRAPHY MATERIALS
5.3.1 Supports
The earliest supports used in affinity chromatography

were agarose gels, a purified form of the naturally occurring
polysaccharide agar. Microporous glass beads, cellulose,
polyacrylamide, crosslinked dextrans and polystyrenes have
also been used as supports.
The proper selection of the support or matrix for the

ligands is of critical importance for the success of an
affinity chromatography system. An ideal support would
possess the following properties (22):
(1) It must interact very weakly with
proteins in general to minimize the non-
specific adsorption of proteins.
(2) It should be an insoluble, rigid material
which exhibits good flow properties
which are retained after coupling.
(3) It must possess chemical groups which
can be activated or modified, under
conditions innocuous to the structure of
the matrix, to allow the chemical linkage
of the ligand.
(4) These chemical groups should be
abundant in order to allow attainment of
a high effective concentration of
coupling sites, so that satisfactory
adsorption can be obtained even with
protein-ligand systems of low affinity.
(5) It must be mechanically and chemically
stable to the conditions of coupling and
to the varying conditions of pH, ionic
strength, temperature and presence of
denaturants (e.g., urea, guanidine
hydrochloride) which may be needed for

adsorption or elution. These properties
permit the repeated use of the support-
ligand material.
(6) It should form a very loose, porous
network which permits uniform and
unimpaired entry and exit of large
macromolecules throughout the entire
matrix.
May and Landgraff (23) added three additonal

requirements:
(7) The support should be resistent to
microbial attack.
(8) The support should be highly “wettable”
hydrophilic material when isolating
soluble macromolecules.
(9) It should be available at a reasonable
cost.
No support material meets all of the requirements of

this ideal. The most common matrices used are the
polysaccharide, agarose, polyacrylamide and controlled pore
glass. Other less successful materials used include
polystyrene, cellulose and dextrans. Hyrophobic supports,
such as polystyrene, cause drastic alterations in the three-
dimensional structure of proteins, resulting in substantial
decreases in the proteins’ activity (24).
Cellulose has a heterogeneous microstructure which

strongly detracts from its suitability as a support and, in
addition, has a significant amount of non-specific adsorption
(25). Cellulose particles are formed with more difficulty
compared to agarose gels. However, it has seen use as the
support material in the affinity chromatographic isolation of
immunoglobulins (26). The principal advantage of cellulose is
that it is the cheapest of the support materials.
Crosslinked dextrans are polysaccharides like agarose

and possess most of agarose’s desirable features, except for
its degree of porosity. The low porosity of dextrans make
them ineffective as adsorbents for enzyme purification (22).
5.3.1.1 Agarose: Agarose is a linear polysaccharide

consisting of alternating P-D-galactose and 3,6-anhydro-a-L-

galactose residues linked 1,3- and 1,4- respectively. Agarose
is not chemically crosslinked and the size exclusion limit
depends on the percentage of agarose in the gel. A
crosslinked agarose (Figure 5.11) has been prepared with 2,3-
dibromopropanol which substantially increases the chemical
and mechanical stability of the support.
0’
Figure 5.11. Chemical structure of crosslinked agarose

(Reference 23).
Agarose and other polysaccharides are usually activated

using the CNB method (27). This method may be carried out
in alkaline solution in a single step. Various procedures
involving addition of CNB solid or solution have been
developed. Activation is usually complete in less than 30
minutes. The activated gel must then be rapidly washed with
cold buffer and then the ligand is immediately attached. The
chemical reactions proposed for the activation process are
shown in Figure 5.12, along with the attachment of an
amino-containing ligand.
Agarose can also be activated using other chemical

reactions. The matrix can be activated by epichlorophydrins
or diepoxides which introduce reactive epoxide functionalities.
Figure 5.13 shows the epoxide formation, along with the
sulfhydryl coupling and derivatization with aminoalkyl arms
which allow spacing between the matrix and the ligand.
OH 0-CeN
t CNBr -
OH
t
\
t i?
OCNH,
C-NH
0
NH H II H
0-t-N-R
trourea Carbamrtc
Imtdocarbonace
Figure 5.12. Activation of polysaccharides with CNBr and

coupling of amino ligands (Reference 23).
Agarose continues to be one of the most popular

supports used in affinity chromatography. The porosity of
agarose results from the polymer’s ability to form a highly
hydrated open structure. Agarose is quite stable in aqueous
solutions of intermediate pH and ionic strength. The
disadvantages of agarose are its matrix compressibility, its
susceptibility to microbial attack, its instability at
temperatures above 40°C and its relatively high cost. It is
possible, however to strengthen its matrix by crosslinking it
with epichlorohydrin.
A, EPOXY ACTIVATION
a -c~,.cn -ai, o.cw,o(-CM,

t
'0'
\
Ad..Fccr*uliq
J
cy--.-- c;d/"' o.cM, CM
t I
OH-yO7
Bo DISULFIDE EXCHANGE AND SULFHYDRYL COUPLING
I” L
/"t
cr*rr-
C-W_
iti-
A’ I”
+
ALSO COUPLING OF LIGAND-SH
‘;3:I
DIRECTLY WITH: YJU
t-sH
m
C, DERIVATIZATION OF AMINOALKYL ARMS FOR FURTHER COUPLING
B
.C.O+l
REDUCTION
Figure 5.13. Formation of agarose derivatives by (A) epoxy

activation, (B) disulfide exchange and sulfhydryl coupling and
(C) derivatization of aminoalkyl arms for further coupling
(Reference 23).
5.3.1.2 Polyacrylamide Gels: Polyacrylamide gels are

synthetic, crosslinked polymers prepared by the
copolymerization of acrylamide and N,N-methylene-bis-
acrylamide (Figure 5.14). The chemistry of activation for the
polyacrylamide support is based on the exchangeability of the
carboxamide nitrogen with other amino-containing compounds,
as shown in Figure 5.15. Further modification of these
derivatives and the attachment of ligands is accomplished
using chemistry similar to that used with the agarose
activation (28).
CH2 0
II
cti- C - NH,
CH- C-NH,
II
0 I
I
CH, CH,
H Y 1
CH -C-r;- Ct$ - N-C-CH
II k
0
Figure 5.14. Chemical structure of acrylamide copolymerized

with N,N-methylene-bis-acrylamide (Reference 23).
t
C.N.CH, .CH, NH*
t k
Aminoethylated
Derivative
Hydrolysis IOH- 1
- CONH,
Carboxylated Derivative
CDNHNH,
t
Hydrazide
Figure 5.15. The chemistry of the activation of the

polyacrylamide support (Reference 23).
The advantages of polyacryamide gels include the

chemical stability due to its polyethylene backbone, a uniform
physical matrix and porosity, low non-specific adsorption of
proteins and a high concentration of carboxamide groups
which can be used as sites for ligand attachment. The major
drawback of this material is that its porosity is reduced after
the derivatization process.
A material is commercially available that combines the

three-dimensional polyacrylamide lattice with an interstitial
agarose gel (29). This material has both the hydroxyl and
carboxamide functionalities of the agarose and polyacrylamide,
respectively, which can be activated independently and
coupled to two separate ligands using independent reaction
schemes. The flow difficulties noted with agarose are
overcome by having the polyacrylamide lattice while the low
porosity of derivatized polyacrylamide supports are eliminated
by having the interstitial agarose. These combination
materials are available in varying concentrations of each
component.
5.3.1.3 Controlled Pore Glass: Ohlson (30) was the

first to use microporous glass beads as a support material in
affinity chromatography. Controlled pore glass has also seen
wide use as a support in affinity chromatography.
The chemistry of derivatizing glass supports is shown in

Figure 5.16. The glass beads must first be silanized before
they can be activated. Silanization is the coupling of the
silane to the glass bead support through the surface silanol,
or of the oxide groups to the silylalkoxy groups (31). The
reaction sequence as shown in Figure 5.16 used
triethoxysilane. Specific silane reaction procedures have been
described by Weetall for organic (32) and aqueous (33)
solvents.
The silanized controlled pore glass beads can then be

activated to form an arylamine, carboxyl or aldehyde
derivative (34). The arylamine is prepared from the silanized
support by reaction with p-nitrobenzoylchloride, followed by
reduction with sodium dithionite. The reaction is shown with
a 7-aminopropyltriethoxysilanized support in Figure 5.17. The
carboxylated derivative is prepared by reaction of the
alkylamine carrier with succinic anhydride, as shown in
Figure 5.18. The aldehyde derivative is prepared by reaction
with gluteraldehyde, which is illustrated in Figure 5.19.
:, OH
I I
0-~i-O-S,iKH2)n R
:, 0 0
I I
R (CH2), Si(OCH2CH,), + HO-ii-O- )----_* -0-Si - 0- SiKH21n R
:, :, :,
I I
HO-ii-0 -0-Si - 0-Si (CH21n R
b :, dH
I
Figure 5.16. Silanization of the surface of porous glass. R

represents an organic functional group (Reference 34).
CARRIER -0-Si (Cl-i,), NH, * NO2
:,
6 0
CARRIER -O-ii (Ctiz 13 NH: 0 NO2 SODIUM DITHIONITE
:,
CARRIER -0- ;I (CH2), Nti; f-jN&
Figure 5.17. Preparation of the arylamine derivative from

the alkylamine derivative (Reference 34).
:, :, 0
0+ O &
CARRIER -0- .di (CHZ), NH2 t f I) -O-&CH~)3NH~(CH2), COOH
b
I
Figure 5.18. Preparation of the carboxyl derivative from the

alkylamine derivative (Reference 34).
CHO
CARRIER -0-5!i(CH,), NH2 + (LH2), I) -0-Si (CH, 1, N *CH(CH *I3 CHO
bH0
Figure 5.19. Preparation of the aldehyde derivative from the

alkylamine derivative (Reference 34).
Controlled-pore glass beads are now available with

surface coatings of glycerol and glycol-like structures to
eliminate non-specific protein adsorption. The advantages of
this type of support are its mechanical strength, its well-
defined and accessible pore volume, its resistance to
microbial attack and flexibility of chemical modification
procedures for ligand coupling. The principal disadvantage is
that silica is susceptible to dissolution in alkaline media
which restricts its use to solution pH values of less than 7.5.
This support is even more expensive than the agarose gel
beads.
53.2 Spacers
Hydrocarbon spacers between ligand molecules and the
support matrix are used extensively in affinity
chromatography. The usefulness of these spacers has been
attributed to the relief of steric restrictions imposed by the
matrix backbone and to an increased flexibility and mobility
of the ligand as it protrudes farther into the solvent (35).
Shaltiel (36) has pointed out, however, that hydrocarbon
extensions by themselves may bind proteins through
mechanisms unrelated to specific recognition of the ligand.
Such binding is more likely to occur when the affinity
chromatography material is prepared by first coating the
matrix with hydrocarbon arms and then attaching the specific
ligand to those extensions. Those hydrocarbon spacers which
do not have an attached ligand may bind other proteins
through hydrophobic interactions, thereby reducing the
specificity of the system. Therefore, when spacers are
employed between the matrix and the ligand, it is advisable
first to synthesize the ligand with the hydrocarbon attached
and then to attach the elongated ligand to the support
matrix. This procedure will minimize the possibility of
hydrophobic interactions of the space interfering with the
intended affinity chromatography.
Example 5.4
Figure 5.20 (37) shows the effect of hydrocarbon
spacers on the reversible binding of glycogen phosphorylase b
to alkyl agarose supports. The columns were equilibrated at
22°C with a buffer of 50 mM sodium ,&glycerophosphate, 50
mM 2-mercaptoethanol and 1mM EDTA at pH 7.0 before
applying the protein sample to the column. Nonadsorbed
protein was washed off with the same buffer. Elution,
initiated at the arrow in the figure, was carried out with a
deforming buffer of 0.4M imidazole, 50mM 2-mercaptoethanol

adjusted to pH 7.0 with citric acid. The binding on the C,
support is quite specific, since 95% of the crude muscle
extract, containing no phosphorylase b activity, is not
adsorbed while the small amount of protein eluted by the
deforming buffer has a very high phosporylase activity and
yields over 95% of the enzyme activity. This is in contrast
to the results with C,, where phosphorylase b is not
adsorbed, with Cs, where the enzyme is retarded, and with
C,, where the enzyme is adsorbed but could not be eluted.
SEPH-Cl SEPWC3 SEPH-Cq SEPtq
FRACTION NUMBER
Figure 5.20. Preferential adsorption of glycogen

phosphorylase b on hydrocarbon-coated agaroses depends on
ihe length of their alkyl side chains (Reference 37).
5.3.3 Ligands
The ligand is usually a small molecule, covalently
attached to the solid support, that displays a special and
unique affinity for the molecule to be purified. This small
molecule must possess chemical groups that can be modified
for linkage to the solid support without destroying or
seriously altering interactions with the designated protein.
When the target protein has a high molecular weight, the
ligand groups which interact with that protein must be
sufficiently distant from the solid matrix to minimize steric
interference. The use of “spacers” to create this distance
has just been discussed.
Enzymes constitute the largest category of proteins

purified by affinity chromatography. Purification of enzymes
can be divided into two categories: (A) enzymes purified

with the aid of specific inhibitors or substrates or (1;: the
purification of a wide range of enzymes using a general
ligand.
A large number of different ligands are available for

specific interactions with enzymes; these include competitive
enzyme inhibitors, coenzymes, other substrates and cofactors.
The columns prepared from such ligands can only be used to
purify the specific enzymes for which the column was
designed. At times, specific columns may also serve for the
purification of enzymes with the identical catalytic activity
but which are derived from different organisms. Table 5.3
(38) shows some of the ligand-enzyme combinations based on
specific interactions.
Table 5.3 : Ligand-Enzyme Combinations Based on

Specific Interactions (38)
Enzyme Ligand Eluent
N - Acetylglucosamidase Thio-N-acetylglucosamide NaCl

Adenosine (phosphate) Inosine Adenosine
deaminase
D-Alanine carboxypeptidase Penicillins N&l, hydroxylamine
Carboxylpeptidase N Aminobenzoyl arginine Cuanidinoethyl-mercapto

succinate
Choline dehydrogenase Choline Dithiothreitol
Dipeptidyl peptidase 4-Phenylbutylglycylproline Ethylene glycol, NaCl
t1astase Elastin, (Ala)3 Salt, Ala
m- L - Fucosldase Fucosamine Thiofuco- Fucose

pyranoside
B - Cdlocrosidase Thiogalactoside Thiogalactoside, ethylene

glycol
Glyoxylase Glutathione PH
tiistaminase Cadaverine Heparin
38 - Hydroxysreroid oxidase Cholesterol Triton X - 100
Nyosin kinasc (I, II) Calmodulin ECl’A
Phenylalanine hydroxylase Pteridine derivative Phenylalanine, pH
Prostaglandin cyclooxygenase Flubiprofen Flufenamic acid, ethylene

glycol
Renin Pepstatin, hemoglobin PH

octapeptide
Urclte oxidase 0 - Aminoxnnthine urate

The general ligand is a ligand which may be used with a

larger group of enzymes (39). Since about 30% of the known
enzymes require a coenzyme for their activity, using this
coenzyme as the ligand will allow purification of a relatively
large range of proteins on the same column. The coenzymes
most commonly used are NAD, NADP, ATP, CoA, and other
derivatives of adenine nucleotides. Table 5.4 (38) shows an
alphabetical list of some enzymes that have been purified
with general ligands.
Table 5.4: Enzymes Purified Using General Ligands
EnZIyllle Ligand Eluent

Adenosine ATP, ANP Adenosine, pH
Alanine dehydrogenase NADP NaCl
L - Arabinose kinase ATP NaCl
Choline acetyltransferase Coenzyme A Citrate-phosphate EDTA,

dithioerythritol
Cytochrome reductase NADP ADP NADP
L - Fucose dehydrogenase AMP Pfj
Glycogen phosphorylase AMP AMP
15 - Hydroxyprostaglandin NAD NADH

dehydrogenasc
3 B - ffydroxysteroid pi
dehydrogenase
Isocitrate dehydrogenase AMP NAD
Malate thiokinase ADP CoA, ATP, malate
Nucleotide pyrophosphatase AMP N&l
Phosphoglucose isomerase ATP Glucose 6 - phosphate
Phosphorylase a AMP AMP
Pyridine dinucleotide NAD, Z’, 5’-ADP NADPfI, 2’-AMP

transhydrogenase
Ribonuclease T2 AMP 2’ (3)-AMP, NaCl
Sorbitol dehydrogenase NAD NAD
Threonine dehydratase AMP AMP
The lectin, concanavalin A, has been used as a general

ligand for polysaccharides and glycoproteins (40).
Concanavalin A will selectively bind glucose and mannose
residues in macromolecules. Therefore, when used as a
ligand, it can be used to isolate a wide variety of

biopolymers which contain one of the two sugar groups.
B-12 coenzyme and other cobalamins have been used as

general ligands by coupling to substituted agarose either
using the phosphate group or by amide formation at one of
the side chains (41). Intrinsic factor, transcobalamins and B-
12-dependent enzymes bind to these adsorbents.
The polyanionic aromatic dye chromophores mimic the

overall shape, size and charge distribution of the naturally
occurring biological heterocycles, such as the nucleotides and
coenzymes. This is seen in the use of dyes, especially the
reactive triazine-based textile dyes, as general ligands
(42,43).
Cibacron Blue F3G-A has been shown (42,44) to be an

especially effective adsorbent for the purification of pyridine
nucleotide-dependent oxidoreductases, phosphokinases,
coenzyme A-dependent enzymes, hydrolases, acetyl-,
phosphoribosyl- and aminotransferases, RNA and DNA
nucleases and polymerases, restriction endonucleases,
synthetases, hydroxylases, glycolytic enzymes,
phosphodiesterases, decarboxylases, sulfohydrolases and many
other seemingly unrelated proteins including interferon and
phytochrome.
Other triazine dyes, such as Procion Red H-E3B, Procion

Red H-8BN, Procion Green H-4G and Procion Brown MX-SBR,
have also been found (45) to be suitable ligands for the
selective purification of individual pyridine nucleotide-
dependent dehydrogenases, phosphokinases, plasminogen,
carboxypeptidase G2, alkaline phosphatase and L-aminoacyl-
tRNA synthetases.
These synthetic dyes offer five advantages as general

ligands when compared to immobilized coenzymes or other
biological ligands for use in large scale affinity
chromatography:
1. The protein-binding capacities of
immobilized dye adsorbents exceed those
of biological origin by factors of 10 to
100.
2. The synthetic dyes may be easily coupled
to the matrix materials and are resistant

to chemical and enzymatic degradation.
3. These dyes have general applicability and
allow easy elution of bound proteins
with good yields.
4. The characteristic spectral properties of
the dyes permit monitoring of ligand
concentrations and identification of
column materials (Table 5.5).
5. These dyes are readily available at a
reasonable cost.
Some of the commercially available affinity adsorbents

are shown in Table 5.6.
Table 5.5: Properties of Some Triazine Dyes (45)

Molar absorption
ReacLlve Dye (Z salt) coefficient (M-l cmml)
Cibacron Blue F3G-A 773.5 610 13,600
Procion Brown MX-5BR 588.2 530 15,000
Procion Green If-4G 1760.1 675 57,400
Procion Red H-8BN 801.2 546 21,300
Procion Red H-E3B ---__ 522 30,000
5.4 LABORATORY PRACTICES
5.4.1 Selecting a Ligand

The task of selecting a suitable ligand remains an
empirical process involving the testing of a number of
potential ligands for their ability to bind the desired protein.
As an example of a typical empirical approach, the

relative suitability of several dyes as general ligands may be
conveniently tested with a series of small columns, each
containing 500 mg of agarose-bound dye equilibrated with an
appropriate buffer in 0.8 x 4 cm disposable polypropylene
columns (43). A small sample containing the desired protein
is applied to each column and the flow is interrupted for 10
minutes to allow equilibration of the sample with the dye
ligand. Nonadsorbed protein is purged with 5 to 10 mL of
the equilibration buffer. The bound protein is then eluted
with 1M KC1 in the same buffer. Both the void and the
Table 5.6: Some Commercially Available Affinity Adsorbents
I&
Tradename Matrix Ligand Source

?
%I
Dyematrex Blue A Silica Cibacron Blue F3GA Amicon R
Dyematrex Red A Silica Procion Red HE-3B Amicon Z
0
Matrex Gel PBA Agarose Phenyl boronate Amicon
;
Affi-Gel Blue Agarose Cibacron Blue F3GA Bio-Rad m.
Affi-Gel Calmodulin Agarose Calmodulin Bio-Rad 2
g
Affi-Gel Ovalbumin Agarose Ovalbumin Bio-Rad i;
Affi-Gel Galactosamine Agarose N- acetyl galactosamine Bio-Rad :
0
J
Affi-Gel Phenothiazine Agarose Phenothiazine Bio-Rad c;l
Affi-Gel 731 Polyacrylamide Polycations Bio-Rad
'I-Methyl-GTPSepharose Agarose P- aminophenylesterof Pharmacia
7- methylguanosine5- triphosphate
2
Oligo (dT)-Cellulose Type 7 Cellulose Oligo (dT) Pharmacia
5
DNA-Agarose Agarose DNA Pharmacia E?
AG POLY (A) Type 6 Agarose Polyriboadsnylicacid Pharmacia 2
E
AG POLY (C) Type 6 Agarose Polyribocytidylicacid Pharmacia
AG POLY (I) Type 6 Agarose Polyriboinosinicacid Pharmacia
Poly (U) Sepharose 48 Agarose Polyuridylicacid Pharmacia
Progel-TSK Chelate - 5PW Synthetic polymer Iminodiaceticacid Supelco/Toyo Soda
Progel-TSK Heparin - 5PW Synthetic polymer Heparin Supelco/Toyo Soda
Progel-TSK Blue - 5PW Synthetic polymer Cibacron Blue F3GA Supelco/Toyo Soda
Progel-TSK Boronate - 5PW Synthetic polymer M-Aminophenyl boronic acid SupelcoiToyo Soda
Progel-TSK ABA - 5PW Synthetic polymer P-Aminobenzamidine SupelcolToyo Soda
eluate fractions are assayed for protein and activity to allow

evaluation of the purification, recovery and capacity of the
dye-protein combinations. Many other examples of ligand
screening procedures are to be found in the literature (46
48).
The kinetic inhibition constants or dissociation

constants of the protein-ligand interaction for a series of
ligands in free solution represents an alternate method of
screening potential ligands (49). For dye ligands, there is a
general trend of protein-dye dissociation constants (Kd)
across the color spectrum. The blue and green dyes tend to
bind more tightly to pig heart lactate dehydrogenase (Kd <
8M) than yellow dyes (Kd > 36M), while orange and red dyes
display intermediate affinities (10 M < K, < 33 M).
After these screening tests, the choice of a ligand

should be determined by two factors (50). First, it should
have a combination of chemical functionalities that allow it
to be attached to the activated support while retaining its
selective binding ability. Second, the ligand should have an
affinity for the desired product in the 10m4 to 10m8 M range
in free solution to allow maximum binding without the risk of
unacceptably difficult product elution.
Table 5.7 shows examples of many types of ligands

which have been used with Sepharose derivatives in affinity
purification applications. In general, enzymes have been used
as ligands to isolate a substrate analogue, an inhibitor or a
cofactor; an antibody may be used to isolate an antigen, a
virus or a cell type; a lectin for a polysaccharide, a
glycoprotein, a cell surface receptor or a cell type; a nucleic
acid for a complementary base sequence, a histone, a nucleic
acid polymerase or a binding protein; a hormone or a vitamin
for a receptor or a carrier protein; and a specific cell for a
cell surface specific protein or a lectin.
The capacity of affinity adsorbents has been shown to

depend on the degree of ligand substitution (51). The
greater the amount of ligand on the matrix, the lower the
specificity of the protein adsorption. The larger amounts of
ligand are likely to increase steric hindrance and non-specific
binding and to make elution difficult. Ten milligrams of
protein ligand per mL of support (2 pmols per mL) is a
useful upper limit on substitution.
Table 5.7 : Coupling Gels for Ligand Immobilization (50)
Derivative of
Type of Ligand Sepharose Comments
Proteins, peptides, CNBr -activated Method of choice for proteins and

amino acids, Sepharose 48 nucleic acids. Very well docwented.
nuclclc acids
(usin): -MI2 group)
CH-Sepharose 48 Coupling via 6-carbon spacer arm.
Or&r& solvents can be used for
water insoluble ligands. Carbo-
diimide coupling method used.
Activated CH- Activated gel for spontaneous Coup-

Sepharose 48 llng via spacer an. Especially use-
ful for small, sensitive ligands.
Epoxy-activated Activated gel for coupling amino

Sepharose 68 acids and peptides. Hydrophilic
spacer arm. Organic solvents can be
used.
Amino acids, kcto AH-Sepharose 4B Coupling via 6-carbon spacer arm.

acids, carboxylic Organic solvents can be used for
acids water insoluble ligands. Carbo-
(using -COO11group) dlimide coupling method is used.
Sugars, other Epoxy-activated Activated gel for coupling via ex-

hydroxyl compounds Sepharose 6B tremely stable ether bond. Hydrophilic
(using -011 Broup) spacer arm. Organic solvents can be
used.
Proteins Thiopropyl Sepharose 6-B For covalent chromatography of -SH

(using -St1 group) Activated Thiol containing substances. Coupling re-
Sepharose 48 actions are reversible.
Annno acids, other Epoxy-activated For a stable irreversibly coupled

low MN coIli[Jounds septlar0sO 68 product.
Low MN thiol Thiopropyl Sepharose 68 Ligand-matrix bond cleavable.

(using -31 group) (1.e. Coenzyme A)
5.4.2 Preparing the Support-Ligand

The ligand must be coupled to the solid support under
mild conditions that are well tolerated by both ligand and
support. The resulting support-ligand must be washed
exhaustively to make certain that all material which is not
covalently bound is removed. With highly aromatic
compounds, such as estradiol, complete removal of adsorbed
material is very difficult and will require many days of
continuous washing. In extreme cases, such as for Congo
red, it is necessary to wash the column with large quantities
of the protein to be purified, followed by elution to
regenerate the column.
A prerequisite to affinity chromatography experiments is

an accurate method for determining the amount of material
attached to the solid support. This is preferably determined
by measuring the amount of ligand released from the
support-ligand material after acid or alkaline hydrolysis.
Exhaustive digestion with pronase or carboxypeptidase has

been used in some cases in which oligopeptides are attached
to agarose. The degree of ligand substitution on the support
is then expressed in terms of concentration, such as
micromoles of ligand per milliliter of swollen support.
The quantity of ligand attached to the support can be

controlled by varying several parameters. Most important is
the amount of ligand added to the support. This is shown in
Table 5.8 for the attachment of 3’-(4-aminophenylphosphoryl)
deoxythymidine 5’-phosphate to agarose. When highly
substituted derivatives are desired, the amont of ligand added
should be 20-30 times higher than that which is desired in
the final product.
Table 5.8 : Efficiency of Coupling 3’-( 4-Aminophenylphosphoryl)

Deoxythymidine 5’-Phosphate to Agarose (1)
@ Moles of inhibitor/ml agarose
Experiment Added Coupled
A 4.1 2.3
B 2.5 1.5
C 1.5 1.0
D 0.5 0.3
The pH at which the attachment is performed will also

determine the degree of attachment since it is the
unprotonated form of the amino group which is reactive.
Compounds containing an o-amino group, with a pK about 8,
will react optimally at a pH of about 9.5 to 10.0. This is
shown in Table 5.9 for the attachment of alanine to argarose
(35).
Table 5.9: Effect of pH on the Coupling of Alanine to

Activated Agarose (35)
Conditions for Coupling Reaction Alanine Coupled
Buffer I?!!- ( moles per ml agarose)
Sodium citrate, 0.1 M 6.0 4.2
Sodium phosphate, 0.1 M 7.5 8.0
Sodium borate, 0.1 M 8.5 11.0
Sodium borate, 0.1 M 9.5 12.5
Sodium carbonate, 0.1 M 10.5 10.5
Sodium carbonate, 0.1 M 11.5 0.2

The time and temperature will also affect the quantity

of ligand attached. Swollen polyacrylamide beads are reacted
with 3 to 6M hydrazine hydrate in a constant temperature
bath at temperatures between 45°C and 50°C. Figure 5.21
shows the time required to achieve certain degrees of
derivatization (52). The amount of substitution varies
linearly for at least an eight hour period.
I I I I I I I I
0 1 2 3 4 5 6 7 0
Time of heating (hours)
Figure 5.21. Time course of the reaction between

polyacrylamide BIO-GEL P-60 (100-200 mesh) and aqueous
hydrazine. C’ refers to mmoles of hydrazine groups found on
the amount of derivative produced from 1 gram of the
original dry polyacrylamide (Reference 52).
5.4.3 Ligate Adsorption

The specific conditions for affinity chromatography
adsorption are dictated by the specific properties of the
ligate to be purified. When ligands with mixed electrostatic-
hydrophobic characteristics are used, the pH, ionic strength
and temperature conditions can often be adjusted to improve
adsorption of the ligate. The optimal sample size, flow rate,
buffer composition, pH, ionic strength and temperature will
have to be defined for the specific ligand-ligate combination.
Table 5.10 (45) lists a selection of conditions that have been
used to adsorb proteins to immobilized dye ligands.
Table 5.10: Operational Parameters for the Adsorption of Proteins

to Immobilized Dyes (45)
Condition Data
Equilibration buffers
Molarlty 5 - loo mM
PH 5.5 - 9.0
Conlposltlo” Tris, phosphate, acetate. triethanolamine,

tcicine, IIEPES, HOPS, bicarbonate
Additives EDTA, KCl, N&l, NH4C1. (m4)2SO4,

2-mercaptoethanol, dithiothreitol,
cysteine, thioglycerol, phenylmethylsulfonyl
fluoride, urea, sucrose, glycerol, ethylene
glycol, detergents, chloramphenlcol,
amphotericin B
Netill Ions Na, K, Mg, Ca, SC, Bo, Zm, Cu, Co, Ni,
Mn, Fe, Al, Cc.
Sample applwd per bed volume 0.03 - 950 mglml
It is not necessary that the adsorption process take

place in a column. It may be advisable to use batch
purification, for example, when small amounts of protein are
to be extracted from very crude protein mixtures with an
adsorbent of very high affinity. The flow rate of such a
mixture through a column would be very slow, decreasing
with time for those cases when particulates are present in
the protein mixture.
In some cases involving very high affinity complexes,

such as occurs with certain antibody-antigen systems, it is
preferable to adsorb the protein to the support-ligand in a
column, to wash it there extensively and then to elute the
protein by removing the adsorbent from the column and

suspending it in a vessel with an appropriate solvent. This
means of elution requires less drastic conditions and may
give higher yields because of the thoroughness of mixing,
higher dilution of the insoluble ligand and easier control of
time and temperature which are possible.
The operational capacity for specific adsorption with a

given affinity adsorbent can be determined by slowly passing
an excess amount of enzyme or protein through a sample of
the adsorbent packed in a column. After washing with buffer
until negligible protein is present in the effluent, the
adsorbed protein is eluted and quantified. The operational
capacity is the amount eluted.
Another approach to determine operational capacity is

to add successive, small samples of pure protein to the
column until protein or enzymatic activity is noticed in the
column effluent. The operational capacity is the total
amount added until breakthrough occurs.
5.4.4 Ligate Elution

In most cases, the elution of the ligate is carried out
by changing the pH, ionic strength or temperature of the
buffer. Removal of proteins from very high affinity
adsorbents may require protein denaturants, such as urea or
guanidine hydrochloride. Ideal elution of a tightly bound
ligate should utilize a solvent which causes sufficient
alteration of the conformation of the protein to decrease
significantly the affinity of the protein for the ligand but
not so severely that the protein is completely denatured or
unfolded. The eluted protein should be neutralized, diluted
or dialyzed at once to allow prompt reconstitution of the
native protein structure.
Another method for elution is to cleave the support-

ligand bond selectively. This allows the removal of the
intact ligand-protein complex. Excess ligand can then be
removed by dialysis or by gel filtration. This method can be
used for ligands attached to the support by azo linkages or
by thiol or alcohol ester bonds.
There is no single method that serves as a general

example for all affinity chromatography elutions. However,
several elution methods for both specific and general elution

are being used with success.
Elution of the adsorbed protein may also be carried out

with a solution containing a specific inhibitor or substrate.
The inhibitor can either be the one that is covalently
attached to the matrix or an alternate, stronger competitive
inhibitor. If the inhibitor is the same as attached to the
support, it must be present at higher concentrations in the
elution solution.
Specific elution is also called affinity elution

chromatography since it utilizes competition for the ligand
site by hapten, ligands or inhibitor in solution to elute the
adsorbed species. This use of the bioaffinity of the adsorbed
species for elution represents the best conditions for highly
selective purification. On the other hand, non-specific
methods can be applied when apparently specific eluants fail
to elute the desired protein.
Non-specific methods include solvent or buffer changes,

pH or ionic strength changes, temperature changes, reversible
denaturation and the chemical cleavage of the ligand from
the support (53).
A good strategy for designing an elution method is the

initial examination of several means of non-specific elution
(38). Non-specific elution can be considered as an extra step
in washing the column if the desired compound is not eluted
since it serves to remove contaminating macromolecules. The
washing should be started with a high salt concentration in
the buffer so that compounds which are bound solely due to
their ion exchange properties will be eluted. Occasionally
low salt concentrations may be used initially to elute
compounds bound through hydrophobic interaction. When
changes in ionic strength are ineffective, different buffers,
such as a change from a Tris to a borate buffer, may be
tried. The next step in this strategy is to change the pH of
the eluant, beginning with small changes followed by the
addition of O.lN acetic acid or 0.5M NH,OH. If elution of
the desired protein does not occur, treatment with ethylene
glycol, dioxane or propionic acid may result in elution.
Table 5.11 (45) lists a number of conditions which have

been used to elute bound proteins from immobilized dye
ligands. The two techniques used most often are pH and salt
elution. The ionic strength of the eluant may be increased
in steps, pulses or gradients (54).
Table 5.11: Reported Eluents of Proteins from Immobilized Dye

Affinity Adsorbents (45)
Method Data
Nonspecific
Molarity 0.025 - 6 M
Salts N&l, KCl, CaC12, NH4C1, (Nll412SO4
Polyols Glycerol, ethylene glycol
Chaotropes Urea, NaSCN, KSCN
Chelating agents EDTA, 2,2’-bypyridyl, pyridine dicarboxylic acid
Detergents 0.1% (w/v) SDS, Trlton X - 100
Specific elutants
Molarity 0.001 - 25 mM
Composition Coenzymcs, nucleotides, polynucleotides, ternary

complexes, substrates, dyes
When the elution tests just described are not effective, .

urea or guanidine may be added to remove the desired
protein through denaturation. The denaturation is time
dependent and reversible in many cases. Therefore, it is
necessary that the eluant with the denatured, desired protein
be collected in a dilution buffer immediately after leaving the
chromatography column to reverse the denaturation.
The binding of the desired protein to the affinity

column may be so tight as to preclude recovery of the
protein. Under such circumstances, the ligand should be
attached by means of a hydrolyzable bond or through a
destructible spacer arm. Esters (55) are readily hydrolyzed
with mild bases; vicinal hydroxyl groups are removed with
periodate; and diazo bonds are broken with dithionates (56).
In order to maintain specificity at the elution step, it is

advisable to perform the elution with a substrate or inhibitor
rather than other solvents. In either case, a control column
with ligand-free spacers should be used in order to evaluate
any contribution of hydrophobic interactions by the spacers
to the retention capability of affinity chromatography
columns.
In some instances, a suitable combination of substrates

and coenzymes may elute the desired protein with a much

greater yield than any of the individual components. As an
example, E. coli adenylosuccinate synthetase is eluted
efficiently from immobilized Procion Blue H-B using
quaternary complex formation with IMP, GTP and L-
aspartate, while much lower yields are obtained when a single
or pair of these substrates are employed (48).
Suzuki and Karube (57) have demonstrated the use of

light to control the affinity adsorption or elution of enzymes
to agarose modified with a spiropyran compounds, using the
isomerization shown in Figure 5.22, Trypsin from bovine
pancrease was bound in the dark on a spiropyran gel with a
soybean trypsin inhibitor ligand and was released with visible
light irradiation, as shown in Figure 5.23. Approximately 60
to 80% of bound trypsin was released with visible light
irradiation. The activity of the released trypsin was the
same as that of native trypsin. Approximately 2 1-fold
purification of trypsin was obtained using this technique. A
limitation of this method is the limited number of cycles the
spiropyran gel may be used because the spiropyran derivative
undergoes irreversible isomerization by the light irradiation.
~Q~~~N(&
( colorless form ) ( colored form )
Figure 5.22. Photoreversible isomerization of spiropyran

compounds (Reference 57).
5.5 SCALE-UP OF LABORATORY PROCEDURES
When the laboratory procedure has been carefully

optimized, it is relatively easy to scale-up the process since
it only requires extrapolation of the volumes of affinity
adsorbent, washing buffer solution and elution solutions to
quantities sufficient to process the required amount of feed
stream.
As with other adsorption processes, the feedstream

needs to be pretreated to give a clear, particle-free solution
dark visible light
((0 2s so
RETENTIONVOLUME (cc)
Figure 5.23. Elution pattern of trypsin from the column of

light-activated ligand (Reference 57).
to prevent clogging if a column is used. It is desirable to

pretreat the feedstream to reduce the amount of unwanted
proteins to minimize any nonspecific binding to the affinity
adsorbent.
Columns for process scale affinity chromatography are

short and wide, having a width to height ratio of 1 to 2-4 to
provide rapid throughput and reduced process time (Figure
5.24). When very large volumes (50 to 500 liters) of a
feedstream are to be processed, it is convenient to carry out
batch adsorption and elution in the same vessel. A very
wide column is used, fitted with a slow-speed overhead
stirrer. After batch mixing in the column, the affinity
particles are allowed to settle for washing to remove the
unbound material, followed by elution of the desired
substances. The settled bed height is typically about 10 cm

which allows high flow rates and rapid processing of large
volumes of the feedstream.
FILLER TUBE
LACK STOP
ING
OLLECT RING
IXED SEAL
ADJUSTABLE
SEAL
SUPPORT RIM
;fJL;STABLE
COLUMN TUBE
BED SUPPORT
FIXED CELL
;;JA;STABLE
FOOT UNIT
Figure 5.24. Process scale affinity column with adjustable

height, manufactured by Amicon.
The purification scheme will usually require a

subsequent step to remove the eluting agent from the
purified product, particularly when partial denaturization of
the product occurs during elution. Short connecting pipelines
and the highest possible flow rates are parameters that must
be designed into the process.
Automated instrumentation is usually employed to

control the affinity purification process. An integrated
automated system consists of a pump to maintain liquid flow,
a UV monitor and recorder connected to the column outlet to
follow the effluent composition and concentration, a
programmable fraction collector and a system of solenoid
valves to control feedstream and eluent application to the
column, collection of fractions from the column and changes
of eluents to clean and to regenerate the affinity adsorbent
(Figure 5.25) (58).
Figure 5.25. Process diagram for large-scale affinity

chromatography. S: sample; Bl to B3: buffers 1 to 3; PI to
P3: products 1 to 3; W: waste (Reference 58).
5.6 AFFINITY CHROMATOGRAPHY OPERATIONS

As with the other batch operations described in this
book, the affinity particles are simply suspended in the
sample fluid and agitated for the desired length of time,
after which they are collected by filtration. The particles
are then washed to remove non-bound substances, after
which the particles are suspended in a solution which elutes
the desired protein from the affinity particles. This scheme
is shown in Figure 5.26. It is also possible to place the

washed particles in a column for the elution of the desired
protein.
\ ,
(A) Mixing/adsorption /a) Washing
(Cl Desorption 101 Recovery
Figure 5.26. Batch processing of biopolymers on an affinity

sorbent material. The open circles represent the affinity
material, the dots are the ligate molecules and the triangles
are the contaminants.
Sundberg and coworkers (59) have described a

laboratory batch operation known as the “tea bag” method.
With this method, shown in Figure 5.27, the affinity particles
are placed inside a porous fabric bag which is immersed into
the sample fluid. Several such bags, such containing affinity
particles with different ligands, may be used to extract
several proteins simultaneously from the same sample fluid.
This simultaneous extraction offers the advantages of savings
in time, of conserving precious starting materials and of
decreasing losses of labile components.
/
0’
0
0
0
/’
k l’
Figure 5.27. Tea-bag processing of different ligate molecules

on three different affinity materials, A, B and C (Reference
59).
Wong and Charm (60) have taken the porous fabric bag
approach and converted it into a continuous belt process,
shown in Figure 5.28. The porous belt containing the
affinity particles forms a continuous loop which moves
through a series of tanks so that affinity adsorption,
washing, elution and regeneration occur sequentially. This
process requires that the ligate have a very high affinity for
the sorbent material.
56.2 Column Operations

Affinity chromatography is most commonly carried out
in a column operation. A typical arrangement of such an
operation is shown in Figure 5.29. The feedstream is pumped
onto the column, followed by a washing buffer, and finally
the elution solution. The effluent from the column may be

directed to the waste stream, to product collection or to
further processing operations. A monitor on the effluent is
needed to identify the process stage to facilitate product
recovery.
Plasma Wash Protein Regeneration

solution recovery
Figure 5.28. Continuous belt process for the affinity

adsorption of proteins (Reference 60).
A
Wash Derorbing
buffer agent
FURTHER
PROCESSING
IB
I Fraction collector I
rair 1
Figure 5.29. Affinity purification carried out in a column,

where valve A selects either sample, wash buffer or
desorbing solution. Valve B directs the column effluent to
waste or to further processing.
The industrial use of column affinity chromatography

requires the development of automatic control systems and
monitoring in order to achieve reliable operation. A suitable
control system consists of a set of valves to regulate the
application of fluid streams to the column (feedstream,
washes, eluants) and the collection of the effluent from the
bed (product and waste streams), along with a set of
instruments to provide a record of the separation. Figure
5.30 (61) shows the schematic of the microcomputer-
controlled automatic system developed by Chase. The process
includes an on-line gel filtration system for the rapid
removal of eluent from the eluted protein. This is a
desirable feature since eluents are often denaturants whose
rapid removal is required.
___--_---- __ --- - --_--_-

____---
I
-irELn
_------TV--
---I
1
r---------
+&YE. R sz ’ s3
“0
81L-4l-l
“C
Pl
Rl R3 R4
iI
’ t I
’ ’ I
I I I
I I I
I I I
IL5
D
’
P .__? r___+----
I
M,
n-l I
J
Iii
i
c’ ,-*
--4 C __
1
i
n
t--* 11
I ’ f I
I; ' I i
11;
II ! !
t ’
, LB_ ___-_~~~~_~___~__-__J I
L--------_----_-___- __________ ~~~~_.__~~-~~~~~J
Figure 5.30. Microcomputer controlled system for the

automatic control of fixed bed affinity separations. Bl:
immunosorbent bed; B2: gel filtration; C: computer; D: data
storage; F: fraction collector; P: pumps; R: reservoirs; S:
solenoid valve (Reference 61).
Several columns, each loaded with a different affinity

sorbent, may be used in series to isolate several desired
substances from a single feedstream. As an example (62), rat
liver extract was passed through two columns in series which
removed dihydrofolate reductase and guanine deaminase,
respectively.
Begovich of the Oak Ridge National Laboratory (63) has

developed a continuous column affinity chromatography
procedure using a rotating chromatograph. Instead of
changing feed, wash, elution and regeneration streams at
preset times, as occurs with conventional columns, all four
streams are fed continuously to separate segments of a
rotating column bed, as shown in Figure 5.31. These streams
are introduced at fixed points around the top of the column.
Oistributor
Anqulor
Position
Figure 5.3 I. Continuous affinity chromatography on a

rotating chromatograph. The effluent concentrations are a
function of angular position rather than time: (1)
breakthrough curve for adsorption; (2) wash profile; (3)
efution profile; (4) regeneration (Reference 63).
The fraction of the column fed with each stream is equal to

the fraction of the overall process time devoted to that
operation in a conventional column, assuming all flow rates
were equal. As the bed rotates, the adsorbent material is
contacted with one or another of these streams. Purified
product is withdrawn at a particular position as a continuous
stream. The breakthrough, wash and elution profiles for this
type of column are plots of effluent concentration at
different angular positions around the column. While this
technique offers no advantages over conventional column
operations for small scale operations, it offers easier
operation and more uniform product quality when used in
preparative and manufacturing operations.
5.6.3 Affinity Purification with Membrane Operations

Affinity chromatography has been combined in a single
step with cross-flow filtration to provide a high resolution
and high recovery technique which is capable of treating
unclarified and viscous feedstreams. This technique,
illustrated in Figure 5.32 (64), has been used to purify
concanavalin A from a crude extract of Jack beans by using
Saccharomyces cerevisiae cells as the affinity ligand. Water-
soluble macroligands are retained on one side of the
ultrafiltration membrane. The feedstream is filtered in the
presence of these ligands, the macroligand-product complex is
retained while all impurities pass through the membrane. The
product can then be eluted from the affinity ligand. A
highly purified product at a yield of 70% was obtained using
0.8M D-glucose as the eluant.
Luong and coworkers (65) synthesized an acrylamide

polymer with pendant aminobenzamidine groups to be used
with this technique to purify trypsin. The trypsin-binding
capacity of the water-soluble macroligands was close to the
theoretical value 12O:l (weight of bound trypsin to weight of
aminobenzamidine), which is considerably higher than that
obtained with ligands on crosslinked gel supports. The eluant
used was OSM arginine. This one-step procedure was able to
purify trypsin from a trypsin-chymotrypsin mixture with 90%
yield and 98% purity.
Mandaro and coworkers (66) have reported the use of

an affinity filter matrix which had either Protein A or
Protein G as ligands to purify immunoglobulins from various
animal blood serum samples. The filter matrix was a
composite of vinyl monomers grafted to a cellulose support

which contained large pores. The pores allowed the
migration of macromolecules through the matrix even though
the matrix has the physical strength to be formed into a
high-flow rate filter cartridge (Figure 5.33).
- - --- -
o**.o”. **o
08
0
l o* O
.
Figure 5.32. A: Cross-flow filtration, substances separated by

sizes. Bl: Binding and elimination in affinity cross-flow
filtration. The binder is retained by specific attachment to
macroligands. B2: Dissociation of the product from
macroligands (Reference 64).
PORTS
AFFINITY OR
FLOW
PATHS
-
INERT SUPPORT
Figure 5.33. Filtration support for affinity separations

(Reference 66).
These supports were activated with gluteraldehyde to

provide an aldehyde group at the end of an 18-atom spacer
arm. Depending on the ligand attachment process, between 4
and 5 mg of immobilized ligand is attached per gram of
matrix. Table 5.12 shows the capacity of a filter matrix with
12.4 mg of immobilized Protein A or with 10.6 mg of
immobilized Protein G. The flow rates of about 2 mL/min
resulted in pressures less than 1 bar. Larger systems with
ligand immobilization capacities of 80 mg and 3000 mg
allowed flow rates of 5 and 125 mL/min.
Table 5.12: Comparison of Protein A and Protein G Filter

Interaction with Various Polyclonal IgG’s (66)
Protein A Protein C
Species Capacity (mg) Ratio Capacity (mg) Ratio
Ihmsn I$ (puce) 43.99 3.55 46.66 4.42
lluman I& (serum) 39.03 3.15 40.05 3.79
Bovine I.$ (pure) 25.19 2.03 44.44 4.20
The Ratio is the eluted IgC/bound ligand.
Pungor and coworkers (67) developed a method of

continuous affinity-recycle extraction, shown in Figure 5.34.
The sample is fed to the adsorption stage where it contacts
the affinity adsorbent. The desired product selectively
adsorbs while contaminants are diluted with the addition of
wash buffer. The affinity adsorbent with the product then
passes to a desorbing stage where the addition of the eluting

solvent causes the product to desorb. The regenerated
affinity adsorbent is recycled to the adsorption stage while
the product is removed continuously through a membrane
filter. The filter unit may be used either internally or
externally from the two stages. Continuous operation is
achieved by keeping the adsorption and desorption vessels
well agitated and operating under steady-state conditions.
The recovery yield of j?-galactosidase from unpretreated
homogenized cells was 70% with a 35-fold purification and 4-
times concentration increase using this purification technique.
ADSORBING DESORBING
BEAD RECYCLE
ADDITIONAL PROTEIN
RECOVERY
B)
ADSORUENT
SUSPENSION
SEPARATION
MEMBRANE
Figure 5.34. Schematic of continuous affinity-recycle

extraction system (Reference 67).
A different approach to combining filtration and affinity

purification was developed by Nigam and Wang (68). They
modeled the use of small affinity adsorbent particles
imbedded in hydrogel beads to treat whole broths. The
hydrogel matrix allowed the very small particles to be used
with relatively easy handling and reduced fouling of the
adsorbent particles. The use of the small adsorbent particles
also reduced the internal mass transfer resistance which

allowed increased adsorption, as shown in Figure 5.35, for
different numbers of immobilized particles imbedded in the
hydrogel bead.
I~~MOBILIZED ~.~RBENT BEADS
0
0 02 004 006 0.8
TIME (DIMENSIONLESS)
Figure 5.35. Adsorption rate as a function of time for four

simulated cases (Reference 68).
5.6.4 Ligand Costs

When specific ligands are used to purify an enzyme, the
cost of the affinity chromatography adsorbent can be quite
expensive. Whether or not it is too expensive depends on
the end use of the purified material. Hamman and Calton
(69) have pointed out that although the cost of the
monoclonal immunosorbent used to purify urokinase is
$600/liter, this only amounts to $4.90/g of urokinase purified.
Since urokinase drug treatment charges at hospital
pharmacies are $9 1,000/g, affinity chromatography is
obviously not too expensive in this case.
5.7 AFFINITY CHROMATOGRAPHY APPLICATIONS
Affinity chromatography was originally devised for

protein isolation and purification. Its success in this role is
demonstrated by the proteins purified using the ligands shown
in Table 5.13 (2). However, the use of this technique is
certainly not limited to protein purification. Carbohydrates
and glycoporteins have been purified by the use of
immobilized lectins (Table 5.14) (3). Hormone receptors have
been used for hormone purification and monoclonal antibodies
have been used to purify antigens. Cell organelles and even
whole cells have been purified by affinity chromatography.
Some illustrative examples from different affinity purification
applications will now be presented.
5.7.1 Protein Purification

Wickerhauser and coworkers (70) used heparin-Sepharose
to perform a large-scale purification of antithrombin III from
human plasma. The purification sequence is shown in Figure
5.36. While heparin-Sepharose is very effective in extracting
antithrombin III from other plasma proteins, it will also bind
Table 5.13: Proteins Purified Using Affinity Chromatography (2)

Proteins Purified Ligand
Succinic thiokinase Coenzyme A
Phosphofructokinase Adenosine monophosphate
Transcobalamin I and II Coenzyme 817,
Methylmalonyl - CoA mutase Coenzyme B12
Luciferase Flavin
Dihydrofolate reductase Folate
Tyrosine aminotransferase Pyridoxal phosphate
Aspnrtate aminotransferase Pyri&~tsl phosphate
Glutamate oxaloacetate transaminase Pyridoxal phosphate
Aldolase Nucleotide phosphates
Citrate synthetase Nucleotide phosphates
Ribonuclease reductase Nucleotide phosphates
Sialyltransferase Nucleotide phosphates
Galactosyltransferase Nucleotide phosphates
Glycogen synthetase Nucleotide phosphates

Table 5.14: Lectin-Agarose Affinity Chromatography Purifications

of Glycoproteins (2)
Lectin Ligand Glycoprotein Purified
Concanavalin A (I - Antitrypsin
Herpes-specific membrane glycoproteins
Horseradish peroxidase
Human alkaline phosphatase
Interferon
Immunoglobulin A
Receptors for insulin
Receptors for epidermal growth factor
Porcine enteropeptidase
Rat brain glycopeptides
Thyrotropin
Ricinus communis (I- Fetoprotein
Wheat germ agglutin Erythropoietin
Glycophorin A
Receptors for insulin
Receptors for somatomedin C
Retinal glycoproteins
lipoproteins. Therefore, it was essential to remove the

lipoproteins by precipitation with 20% polyethylene glycol
prior to the affinity purification step. The pasteurization
step was included to eliminate any possibility of infectivity
remaining due to the hepatitis antigen. Table 5.15 shows the
volumes of plasma treated along with the amount of product
recovered with a 2M NaCl elution.
Several /3-lactamases, enzymes that play an important

part in antibiotic resistance, have been purified by
Cartwright and Waley (71) using boronic acid-agarose columns
with hydrophilic or hydrophdbic spacer arms. The results are
presented in Table 5.16.
For those ,&lactamases with a relatively high affinity

for m-aminophenylboronic acid (Ki < 200 PM), gels having a
hydrophilic spacer arm provide adequate binding. For those
p-lactamases with low affinity for m-aminophenylboronic
acid (Ki > 200 PM), an additional binding force is required.
This is provided by using a hydrophobic spacer arm. The
additive nature of the hydrophobic interaction and the ligand
specific interaction was illustrated for the p-lactamases that
had Ki’S in the millimolar region.
Plasma Cohn Fraction IV-1

I IO% in 0.05 M NaCI, pH 7.33
Ctyosupematant
I
1 +20% polyelhylcne glycol
+20x polyethylene glycol
(final concenlration)
I Ba~;;~~;;~;;
t
Batch adsorption with heparin-Sepharose
hcparin-Sepkarose (S-8 g Fraction IV-I IO I ml gel)
(45 ml plasma 10 I ml gel)
Wash with 12.5 liters of

0. I5 and 0.5 M NaCI
Elute wiJ 12.5 liters

of 2 M NaCl
Concen*rat!on Pellicon
cassetfe system to 500 ml
I
Pasteurize with addition of
cilratc lo 0.5 M pH 7.55. 10 h 6O’C
+
Desalt on Sephadcx C-SO
2.5 liter column
I
750 ml solution
I
Sterile filter
I
Lyophilize
Figure 5.36. Purification of antithrombin III from human

plasma or Cohn Fraction IV-l (Reference 70).
Table 5.15: Recovery of Antithrombin III from Human Plasma (70)

Volume Antithrombin III activity
Material (liters)
___ PE/ml
__ Total PE % Recovery
Plasma 113 0.949 107,237 100.0
Cryosupernatant 111 0.939 104,229 97.2
20% PEG
supernatant 115 0.655 75,325 70.2
DEAE unadsorbed
+ wash 215 0.321 69,015 64.4
HS unadsorbed 215 0.055 11,825 11.0

0.15 N NaCl wash 12.5 0.102 1,275 1.2
0.5 tl N&l wash 12.5 1.150 14,375 13.4
2.0 M NaCl eluate 12.0 4.133 49,596 46.2
Pasteurized product 0.74 46.8 34,632 32.3

Table 5.16: Purification of P-Lactamases on Phenylboronic Acid

Affinity Columns (71)
Purification K; (PM)
B - Lactamase (fold) m-aminophenylboronic kid Borate
Pseudomonas aeruginose 220 (a) 5 630
Entereobacter cloacae 12 (a) 60 5000
Pseudomonas maltophilia 2000 (a) 200 370
Bacillus cereus, I 15 (b) 2000 1000
Bacillus cereus, II _-- (c) Not inhibited
Bacillus cereus, III 60 (b) 1650 4400
Klebsiella aerogenes 40 (c) 770 600
(a) Reporter substrate was cephalosporin C.
(b) Reporter substrate was nitrocefin.
(c) Keporter substrate was penicillin G.
Fisher and Newsholme (72) developed a procedure which

directly combined the elution of adenosine kinase from a gel
filtration column with the adsorption of the enzyme onto a
5’-AMP-Sepharose 4B affinity chromatography column. The
buffer used throughout the purification procedure consisted
of 4mM Na,H,PO,, 2mM EDTA and 5% (v/v) glycerol at pH
7.0 and 4°C. Adenosine kinase was eluted from the affinity
column with 0.6 mM adenosine in the buffer. The specific
activity of adenosine kinase in the peak activity fraction
from the affinity column was about 2 pmol/min per mg of
protein, a purification of 870-fold.
Affinity chromatography on immobilized calmodulin has

been shown (73) to be a very convenient method for the
rapid purification of bull seminalplasmin to over 99% purity.
Many aspects of the complex formation between
seminalplasmin and calmodulin have been shown to be
identical with that of melittin, mastoparans and synthetic
model peptides. This similarity is shown in Figure 5.37. The
1:l Ca+2-dependent complex is fully resistant to urea and
displays an affinity constant of approximately log M-l. The
elution of seminalplasmin from the immobilized calmodulin
requires the presence of both EDTA and 4M urea in the
elution buffer.
Affinity Chromatography 4’71
0.8
0.2
-7 -6 -5
LOG ( TOTAL PEPTIDE) (14
Figure 5.37. Displacement of Melex bound [3H] calmodulin

bull seminalplasmin
(Reference 73).
(0) as compared with soluble melittin (“x4;
Malate dehydrogenase and 3-hydroxybutyrate
dehydrogenase have been purified from Rhodopseudomotlas
spheroides extract using two sequential dye ligand affinity
chromatography columns (74). The cell-free extract was
applied to a 1.5 liter column of agarose-bound Procion Red
H-3B. The 3-hydroxybutyrate dehydrogenase was eluted with
the 1M KC1 buffer wash and the malate dehydrogenase was
subsequently eluted by including 2mM NADH in the buffer.
The dialyzed fractions containing malate dehydrogenase
activity were then applied to a 1 liter column of immobilized
Procion Blue MX-4GD. The malate dehydrogenase was eluted
with a linear gradient of KC1 (0 to 700mM, 8 liters) to yield

1 g of homogeneous enzyme at a 70% overall yield.
The Procion Red H-3B column wash fractions which

contained the 3-hydroxybutyrate dehydrogenase were applied
to a 1 liter column of immobilized Procion Blue MX-4GD and
eluted with 2mM NADH to yield 300 mg of homogeneous
enzyme with an 80% overall yield. It should be noted that
the non-affinity chromatography scheme for isolating this
enzyme involves nine separate steps and has an overall yield
of only 9%.
5.7.2 Interferon Purification

Knight and Fahey (75) have used Blue Sepharose Cl-4B
in an affinity purification of homogeneous human fibroblast
interferon in yields ranging from 20 to 40%. The interferon
was produced by diploid fibroblast cells (FS-4) cultured in a
serum-free medium to give crude interferon of high specific
activity. The interferon adsorbs on the affinity material in
the presence of 1 or 2M NaCl, whereas the contaminating
proteins do not. Pure interferon is eluted from the Blue
Sepharose using 50% ethylene glycol. It is collected into a
buffered solution to make a final concentration of 37%
ethylene glycol because interferon is much more stable in
37% ethylene glycol. A second column is used for
concentration purposes, as shown in Table 5.17. The final
product has a specific activity of 5 x lo* units/mg.
Table 5.17: Purification of Human Fibroblast Interferon (75)

Volume Units, Protein Specific Recovered
Fraction (ml) Interferon (mg) activity (%)
Crude 10,000 a5 x io6 500 1.7 x 105 100
First Blue Sepharose 160 68 x 106 0.6 1 x 108 80

column, 1M N&l,
50% ethylene glycol
Second Blue 10 30 x 106 0.06 5 x 108 35

Scphnrose column,
2 i-l N&l, 40%
ethylene glycol
Staehelin and coworkers (76) at Hoffman-LaRoche have

produced a range of monoclonal antibodies for the affinity
chromatography purification of leukocyte interferon. The
process for isolating recombinant human leukocyte interferon
from cultures of E. coli is shown in Figure 5.38 and a

summary of the purification is shown in Table 5.18. A 17 ml
monoclonal antibody column was found to be able to process
the extract from 1 kg bacterial. The affinity purification
provided over l,OOO-fold purification in a single step and
yielded a homogeneous final product.
5.7.3 Receptor-Binding Affinity Chromatography

A receptor is a molecule that recognizes a specific
chemical entity, binds to it and initiates a series of
biochemical events resulting in a characteristic physiological
response (77). Because hormone receptors are present in
mammalian cells only in exceedingly small quantities (less
than 10 ppm on a dry weight basis) and because they are
relatively unstable, traditional methods for isolating and
purifying membrane-bound receptors have been of only
limited value. Only after the introduction of affinity
chromatography did such isolation become experimentally
feasible. Even with affinity chromatography, only a few cell
receptors, like those shown in Table 5.19 (2), have been
successfully purified.
The interaction between hormone, neurotransmitter or

drug and the respective receptor is selective and specific.
This selectivity causes the interaction between the ligand and
its receptor to be one of high affinity, among the highest
known among biological interactions. This means that harsh
Homogenize 1 kg E. cofi
+
Treat homogenate wilh 30% (NH.&Q
+
Treat supernatant with 65% (NH&SO4
+
Dissolve and dialyze pellet
1
Load on 17 ml column of anti-interferon-agar
1
Wash column; elute interferon at pH 2.5
1
Adjust interferon fractions to pH 4.5 (IM Tris)
and load on to CM-cellulose
c
Elute interferon with 0.5 M ammonium acetate
Figure 5.38. Process for isolating recombinant human

leukocyte interferon from E. coli cultures (Reference 76).
x
0
F;
x x
0 0
r. \D
conditions are necessary to elute receptors from columns

containing a selective affinity ligand. Monoclonal antibodies
are now being used to allow elution to be carried out under
milder conditions (78). Table 5.20 (38) shows examples of the
use of specific ligands and monoclonal antibodies for the
isolation of receptors and other binding proteins.
Table 5.19: Membrane-Bound Receptors Purified by

Affinity Chromatography
Receptor Source Ligand
Acetylcholine Electric eel Cobratoxin

Epidermal growth factor Placenta Plant lectins
Epinephrine Erythrocytes Alprenolol
(adrenaline)
Growth hormone Liver Growth hormone
Human chorionic Testes Human chorionic
gonadotrophin gonadotrophin
Immunoglobulin E Basophilic cells Immunoglobulin E
Insulin Liver, fat cells Concanavalin A
Prolactin Mammary gland Growth hormone
Thyrotropin-stimulating Thyroid gland Thyrotropin-stimulating
hormone hormone
Table 5.20: Receptor and Binding Proteins Purified by

Affinity Chromatography (38)
Receptor Ligand Eluent
B - Andrenergic Alprenolol, acebutolol Isoproterenol, alprenolol
Estradiol Estradiol derivative Estradiol, KC1

oligo (dT)
Fc IgG Urea
Insulin Insulin Guanidine, KSCN
Intrinsic facLor Intrinsic factor - Bl2 EDTA
Transcobalamin II Transcobalamin II EDTA
Binding Protein
C3b component Factor H NaCl
Elongation factors GDP GDP

Tu, Ts
Hemopexin Heme, hematin Gly-HC 1, urea
Migration inhibitory Glycolipids KSCN

factor
m RNA cap m7GDP !CCl

5.7.4 Immuno Affinity Chromatography

Affinity chromatography has been used for the
purification of antibodies by employing antigens bound to
carriers. Immobilized antibodies have also been used for the
purification of proteins, including enzymes. Under ideal
conditions, these columns allow separation of specific
antibodies from crude mixtures in a one-step procedure.
Both conventional and monoclonal antibodies have been used.
A major difficulty in the use of immobilized antibodies is the
affinity of an antibody for antigens, thereby making the
recovery of active enzymes difficult owing to the drastic
conditions required for their elution. It is not unusual to
find eluents that consist of chaotrophs (5M KSCN), low pH
(2.2) or high pH (11.5) buffers, urea (3.5-8M) or guanidine
(6M). Despite such harsh conditions, it is of interest that
antigenicity is retained in most cases. When its antigenicity
is not retained, immobilized protein A (79) or antihapten
antibody (SO) may be used to bind antibody-antigen or hapten
antibody-antigen complexes. It is a purified complex, not the
antigen, that is eluted from the support.
The use of monoclonal antibodies (81) decreases the

possibility of contamination with antibodies directed against
proteins other than that desired. Since the number of
monoclonal antibodies against a specific protein are many, it
is advantageous to select low affinity antibodies for
immobilization. The lower affinity will allow the use of
milder elution conditions and reduces the possibility of
irreversible denaturation. Some of the proteins purified by
conventional and monoclonal antibodies are listed in Table
5.21.
5.7.5 Cell Separation by Affinity Chromatography

Immunoadsorbent techniques are capable of efficient and
specific retention of cells and allow the recovery of unbound
cells in a viable, unaltered state. In these methods,
antibodies specific for a lymphocyte surface component are
bonded to solid phase matrices to provide a selective ligand
for isolating given cell subpopulations. Thus, B cells can be
conveniently isolated from a mixed population using
antiimmunoglobulin, T cells by using antibodies to Thy 1
antigens and null cells are taken as those that remain after
removal of both B and T cells. Commercial monoclonal
reagents can be used to separate T cells on the basis of Lyt
Table 5.21: Proteins Purified on Immuno Affinity Columns (38)
Protein Eluent
Adenosine deaminase Urea
a- Fetoprotein Urea
Angiotensin I converting enzyme Mg Cl2
Choriogonadotropin Mg Cl2
Estrogen receptor Na SCN
Factor VIII NH4SCN
Glucocorticoid receptor Acetic acid
Insulin receptor Mg Cl2

Ribonuclease H Mg Cl2
Serine acetyltransferase 0- Acetylserine
Somatostatin Acetic acid
Thyrotropin Guanidine
Trypsin inhibitor Propionic acid
Urokinase Glycine - II Cl
antigens, for example, into killer, helper or suppressor cells

(82).
The most important parameter determining the success

of immunosorbent chromatography of cells is the level of
antibody substitution on the matrix. For adsorption of
unwanted cells, the level of the antibody ligand must be high
enough to allow adequate retention of the bound cells. For
positive selection, the level of the antibody ligand must be
low enough so that desorption of the cells is not made
difficult.
Hubbard and coworkers (83) used the globulin fraction

of rabbit anti-mouse IgM serum for the separation of Ig+ and
Ig- cells. The cell suspension (l-5 x IO7 cells/ml; lo8 cells
maximum) was added to the column, followed by 2 bed
volumes of the buffer. This fraction contained the unbound
cells. After washing the column with 10 to 20 additional bed
volumes of buffer, the bound cells were specifically eluted
with 1 mg/mL of purified mouse IgM in the buffer. Release
of the bound cells occurs through competition between the
cell surface immunoglobulin and the free immunoglobulin for
the immobilized antibody ligand. About 80 to 90% of the Ig-

cells and no Ig+ cells were recovered in the unbound
fraction. Likewise, 80 to 90% of the eluted bound cells
possessed surface immunoglobulin, but were only about 35 to
40% of the Ig+ cells applied to the column.
The interaction between Sfaphq’lococcus ~UXUS (SPA)

and cell surface IgG has been used in this technique to
isolate lymphocytes and other cells (84). Table 5.22 shows
the decrease in the percentage of Ig-bearing mouse B
lymphocytes in the nonadherent cell population after affinity
chromatography with SPA-Sepharose 6MB. The antibody
bearing subpopulation is eluted from the column with IgG or
by mechanical treatment.
Table 5.22: Isolation of &-Bearing Mouse Lymphocytes by

Affinity Chromatography on Sp A-Sepharose 6MB (84)
Cell Fraction Ig - bearing cells (%)
Before chromatography 40.8 2 1.7
After Chromatography
Nonadherent to column 10.9 z 1.2
Adherent to column 90.3 T 1.8
Separation of well-defined cell populations by affinity

chromatography can only be achieved if antibody ligands
against specific surface antigens are available.
The subpopulation enriched in cells bearing the specific

antigen has interacted with antibody and SpA during the
affinity chromatography operation. This is an unavoidable
consequence of methods using antibody as a tool for cell
separation. The antibody present on the surface of the
eluted cells may affect their normal function, For this
reason, affinity chromatography of cells is the method of
choice for depletion of a cell population rather than for
enrichment of a population in cells bearing a specific antigen.
5.8 REFERENCES
1. Cuatrecasas, P., Wilcheck, M., Anfinsen, C.B., Proc Nat1

Acad Sci, 61:636 (1968)
2. Parikh, I., Cuatrecasas, P., Chem Enp News, 63(34):17

(1985)
3. Lowe, CR., An Introduction to Affinitv Chromatoaraphv,

Elsevier Biomedical Press, New York (1979)
4. Scouten, W.H., Affinitv Chromatonraphv: Bioselective

Adsorption on Inert Matrices, John Wiley & Sons, New
York (1981)
5. Larson, P.O., Glad, M., Hansson, L., et al., In: Advances

in Chromatonraphy, (Giddings, J.C., ed.) Marcel Dekker,
New York, Vol 21 (1983)
6. Colowick, S.P., Kaplan, N.O., In: Methods in Enzvmoloay

(Jakoby, W.B., ed.) Academic Press, Inc., Orlando, Fl,
Vol 104, Part C (1984)
7. Chaiken, I., W&heck, M., Parikh, I., eds., Affinity

Chromatographv and Biolonical Recognition, Academic
Press, New York (1984)
8. Nishikawa, A.H., Chem Tech, 5:564 (1975)
9. Graves, D.J., Wu, Y.T., In: Methods in Enzvmolony

(Jakoby, W.B., Wilchek, M., eds.) Academic Press, Inc.,
Orlando, FL, Vol 84, p 150 (1974)
10. Nishikawa, A.H., Bailon, P., Ramel, A.H., Adv Expl Med
m, 42:33 (1974)
11. Lowe, C.R., J Biotechnol, 1:3 (1984)
12. Bottomley, R.C., Storer, A.C., Trayer, I.P., Biochem J,

159:667 (1976)
13. van Deemter, J.J., Zuiderweg, F.J., Klinkenberg, A.,

Chem Ena Sci, 5:271 (1956)
14. Yang, C.M., Tsao, G.T., Adv Biochem Enq, 25:l (1982)
15. Arnold, F.H., Blanch, H.W., Wilke, C.R., The Chem Enq
.I, 30:B9 (1985)
16. Chung, I., Ph.D. Thesis, University of Tennessee, 1976

17. Katoh, S., Kambayashi, T., Deguchi, R., et al., Biotechn

and Bioeng, 20:267 (1978)
18. Hashimoto, K., Miura, K., J Chem Ena Japan, 9:388

(1976)
19. Crank, J., Nicolson, J., J Proc Cambridge Philos Sot,

43:50 (1947)
20. Arnold, F.H., Chalmers, J.J., Saunders, M.S., et al., In:

Purification of Fermentation Products (LeRoith, D.,
Shiloach, J., Leahy, T.J., eds.) American Chemical
21. Graham, E.E., Fook, C.F., AIChE J, 28:245 (1982)
22. Cuatrecasas, P., Anfinsen, C.B., In: Methods in

Enzvmologv (Jakoby, W.B., ed.) Academic Press, New
York, Vol 22, p 345 (197 1)
23. May, S.W., Landgraff, L.M., In: Recent Developments in

Separation Science (Li, N.N., Dranoff, J.S., Schultz, J.S.,
et al., eds.) CRC Press, Boca Raton, FL, Vol 5, p 227
(1979)
24. Voser, W., J Chem Tech Biotechnol, 32:109 (1982)
25. Porath, J., In: Methods in EnzvmoloPv (Jakoby, W.B.,

Wilchek, M., eds.) Academic Press, New York, Vol 34, p
13 (1974)
26. Weliky, N., Weetall, H.H., Immunochem, 2:293 (1965)
27. Axen, R., Porath, J., Ernback, S., Nature (London),

214:1302 (1967)
28. Inman, J.K., Dintzis, H.M., Biochemistrv, 8:4074 (1969)
29. Ultrogel is available from LKB, Inc., Rockville, MD
30. Ohlson, S., Hansson, L., Larsson, P.O., et al., FEBS Lett,
93:5 (1978)
31. Haller, W., Nature (London), 206:693 (1965)

32. Weetall, H.H., Hersh, L.S., Biochim Biophvs Acta, 185:464

(1969)
33. Weetall, H.H., Havewala, N.B., Biotech Bioene, Svmp,

3:241 (1972)
34. Weetall, H.H., Filbert, A.M., In: Methods in Enzvmology

(Jakoby, W.B., Wilchek, M., eds.) Academic Press, New
York, Vol 34, p 59 (1974)
35. Cuatrecasas, P., J Biol Chem, 245:3059 (1970)
36. Shaltiel, S., In: Methods in EnzvmoloPy (Jakoby, W.B.,

Wilchek, M., eds.) Academic Press, New York, Vol 34, p
126 (1974)
31. Er-el, Z., Zaidenzaig, Y., Shaltiel, S., Biochem. Biouhvs

Res Commun, 49:383 (1972)
38. Wilchek, M., Miron, T., Kohn, J., In: Methods in

Enzvmoloay (Jakoby, W.B., ed.) Academic Press, New
York, Vol 104, p 3 (1984)
39. Mosbach, K., Adv Enzvmol Relat Areas Mol Biol, 46:205
(1978)
40. Lis, H., Lotan, R., Sharon, N., Ann N Y Acad Sci,
234:232 (1974)
41. Guilford, H., Chem Sot Rev, 2:249 (1973)
42. Dean, P.D.G., Watson, D.H., J Chromatog, 165:301 (1979)
43. Lowe, C.R., Small, D.A.P., Atkinson, A., Int J Biochem,

13:33 (1981)
44. Gionazza, E., Arnaud, P., Biochem J, 201:129 (1982)
45. Lowe, C.R., Pearson, J.C., In: Methods in EnzvmoloPv

(Jakoby, W.B., ed.) Academic Press, New York, Vol 104,
p 97 (1984)
46. Bruton, C.J., Atkinson, A., Nucleic Acids Res, 7:1579

(1979)
47. Lowe, C.R., Hans, M., Spibey, N., et al., Anal Biochem,
104:23 (1980)
48. Clonis, Y.D., Lowe, C.R., Biochim Biophvs Acts, 659:86

(1981)
49. Clonis, Y.D., Lowe, C.R., Biochem J, 191:247 (1980)
50. Hill, E.A., Hirtenstein, M.D., In: Advances in

BiotechnoloPical Processes, Alan R. Liss, Inc., New York,
p 31 (1983)
51. Eveleigh, J.W., Levy, D.E., J Solid-phase Biochem, 2:47

(1977)
52. Inman, J.K., Dintzis, H.M., Biochemistrv, 8:4074 (1969)
53. Absolom, D.R., Sep Purif Methods, lo:239 (1981)
54. Byfield, P.G.H., Copping, S., Bartlett, W.A., Biochem Sot

Trans lo:104 (1982)
-,
55. Singh, P., Lewis, S.D., Shafer, J.A., Arch Biochem

Biophvs, 193:284 (1979)
56. Singh, P., Lewis, S.D., Shafer, J.A., Arch Biochem

Biophvs, 203:776 (1980)
57. Suzuki, S., Karube, I., In: Recent Developments in

Separation Science (Li, N.N., ed.) CRC Press, Boca
Raton, FL, Vol 6, p 117 (1981)
58. Eveleigh, J.W., presented at the Fourth International

Symposium on Affinity Chromatography and Related
Techniques, Veldhoven, Holland (198 1)
59. Sundberg, L., Porath, J., Aspberg, K., Biochim Biophvs

m, 221:394 (1970)
60. Wong, B.L., Charm, S.E., J Macromol Sci Chem, A IO:53

(1976)
61. Chase, H.A., In: Discoverv and Isolation of Microbial

Products (Verrall, M.S., ed.) Ellis Horwood Ltd.,
Chichester, UK, p 129 (1985)
62. Baker, B.R., Siebenieck, H.U., J Med Chem, 14:799 (1971)
63. Begovich, J.M., Rep ORNL-5915, NTIS, U.S. Dept.

Commerce, Springfield, VA (1982)
64. Mattiasson, B., Ramstorp, M., Ann N Y Acad Sci, 13:307

(1983)
65. Luong, J.H.T., Nguyen, A.L., Male, K.B., Bio/Technologv,

5:564 (1987)
66. Mandaro, R.M., Roy, S., Hou, KC., Bio/Technoloav, 5:928

(1987)
67. Pungor, E., Jr., Afeyan, N.B., Gordon, N.F., et al.,

Bio/Technology, 5:604 (1987)
68. Nigam, S.G., Wang, H.Y., In: Separation. Recoverv and

Purification in Biotechnology American Chemical
69. Hammon, J.P., Calton, G.J., In: Purification of

Fermentation Products (LeRoith, D., Shiloach, J., Leahy,
T.J., eds.) American Chemical Society, Washington, DC,
p 105 (1985)
70. Wickerhauser, M., Williams, C., Mercer, J., VOX Sang,

36:281 (1979)
71. Cartwright, S.J., Waley, S.G., Biochem J, 221:505 (1984)
72. Fisher, M.N., Newsholme, E.A., Biochem J, 221:521 (1984)
73. Comte, M., Malnoe, A., Cox, J.A., Biochem J, 240:567

(1986)
74. Scawen, M.D., Darbyshire, J., Harvey, M.J., et al.,

Biochem J, 203:699 (1982)
75. Knight, E., Fahey, D., J Biol Chem, 256:3609 (1981)
76. Staehelin, T., Hobbs, D.S., Kung, H., J Biol Chem,

256:9750 (1981)
77. Hollenberg, M.D., Cuatrecasas, P., Methods Cancer Res,

12:317 (1976)
78. Stein, B.S., Sussman, H.H., J Biol Chem, 258:2668 (1983)
79. Schneider, C., Newman, R.A., Sutherland, D.R., et al., l

Biol Chem, 257:10766 (1982)
80. Kanellopoulos, J., Rossi, G., Metzger, N., J Biol Chem,

254:7691 (1979)
81. Goding, J.W., J Immunol Methods, 39:285 (1980)
82. Matossian-Rogers, A., Rogers, P., Herzenberg, L.A., m

Immunol, 69:9 1 (1982)
83. Hubbard, R.A., Schulter, S.F., Marchalonis, J.J., In:

Methods in Enzymology (Jakoby, W.B., ed.) Academic
Press, New York, Vol 104, p 139 (1984)
84. Ghetie, V., Sjoguist, J., In: Methods in Enzvmoloav

(Jakoby, W.B., ed.) Academic Press, New York, Vol 104,
p 132 (1984)
Index
Adsorbent fermentation aid - 155

laboratory evaluation - 109 isotherm
regeneration classification - 72
biologic - 131 Freundlich - 68
stripping, solvent - 128 general - 72
thermal - 125 Langmuir - 70
surface area measurement - 98 kinetic theory - 78
type - 95 contact time - 83
activated charcoal - 99 diffusion resistance - 83
alumina - 102 distribution ratio - 83
carbon - 99 mass balance model - 79
polymer, organic - 105 mass transfer coefficient model .
silica gel - 102 87
zeolite - 102 mass transfer zone model - 90
Adsorption - 65 particle shape effect - 94
applications - 136 particle size distribution effect -
antibiotics - 141 94
decolorization - 136 Peclet Number - 83
enzymes - 144 residence time - 83
ethanol - 153 operations - 114
glucuronides - 152 batch - 116
microorganisms - 146 column - 117
organic acids - 147 fluidized bed - 122
organic amines - 147 packed bed - 119
polycyclic aromatics - 153 semi-fluidized bed - 123
sodium cholate - 111 pretreatment - 114
vitamins - 150 shallow bed - 124
computer design - 136 pressure drop - 134
design factors - 131 Affinity chromatography - 416
485
applications - 467 glutamic acid - 214, 270

cell separation - 476 glycine - 75
immuno-affinity - 476 histidine - 270, 387
interferon purification - 472 lysine - 186, 252, 270
human fibroblast inter- monosodium glutamate - 178, 340
feron - 472 phenylalanine - 75,271,387
leukocyte interferon - 472 phenylglycine - 176
protein purification - 467 serine - 271
adenosine kinase - 470 tryptophan - 271, 387
antithrombin III - 467 tyrosine - 387
calmodulin - 470 valine - 271
3-hydroxybutyrate dehy- Antibiotics - 2, 141, 176, 245
drogenase - 471 A-4696 - 144, 286
&lactamase - 468 7-aminoacetoxycephalosporanic
malate dehydrogenase - 471 acid - 176
receptor binding - 473 6-aminopenicillanic acid - 277
ligand cost - 466 ampicillin - 176
ligands - 439 Aureomycin - 388
aromatic dyes - 442 carminomycin - 388
general - 441 cefalexin - 176
specific - 440 Cephalosporin - 141, 277, 388
ligand selection - 443 Cephamycin - 280
ligand support preparation - 446 dihydronovobiocin - 245
ligate adsorption - 449 erythromycin - 280
ligate elution - 450 gentamycin - 174
operations - 456 kanamycin - 174
batch - 456 monomycin - 174
column - 458 morphocyciin - 189
continuous recycle - 464 neomycin - 388
rotating chromatograph - 461 novobiocin - 245
tea bag - 457 oleandomycin - 142, 280
with membranes - 462 olivanic acid - 284
scale-up - 453 oxytetracycline - 142, 285
adjustable height column - penicillin - 143, 244, 277
454 rosamicin - 280
spacers - 438 saikosaponins - 388
supports - 430 sisomycin - 174
agarose - 431 streptomycin - 144, 198, 274
controlled pore glass - 436 tetracycline - 142
polyacrylamide - 435 thienamycin - 284
theory - 418 tylosin - 280
binding kinetics - 424 volonomycin - 155
binding selectivity - 419 Appendix - 312
Amino acids - 75 Centrifugation - 9
acetylglutamine - 273 equipment - 11
arginine - 270 equipment suppliers - 16
aspartic acid - 270 Column chromatography - 335
cystine - 387 applications - 385
Index 487
amino acid separations - 386 effect of bead size distribu-

antibiotics - 388 tion - 358
glucose-fructose - 388 effect of concentration - 356
glycerol - 390 effect of crosslinkage - 358
molasses - 402 effect of flow rate - 361
oligosaccharides - 391 effect of ionic contaminants -
polyhydric alcohols - 400 365
proteins - 392 effect of porosity - 360
regenerant recovery - 400 osmotic effects - 363
eluant selection - 366 zone spreading - 352
industrial systems - 376 types
AMF Zeta-Prep - 378 affinity difference - 337
Amicon/Wright - 378 ion exclusion - 340
Begovich rotary system - 384 ion retardation - 342
Higgins - 380 size exclusion - 340
parametric pump system - 384 Crystallization - 22
Pharmacia - 378 equipment - 25
pseudo-moving bed - 380 equipment suppliers - 29
Amalgamated Sugar - 382 Drying - 29
IWT - 382 equipment - 30
Mitsubishi - 382 equipment suppliers - 35
UOP - 380 Electrodialysis - 48
Technichem - 379 equipment - 52
Wankat - 384 Electrophoresis - 47
Waters Kiloprep - 376 equipment - 50
Whatman - 378 Enzyme Immobilization - 307
laboratory techniques - 366 aminoacylase - 308
materials - 344 aspartase - 308
hydrophilic gels - 344 fl-fructosidase - 309
hydrophobic gels - 346 fumarase - 308
methods glucoamylase - 308
adsorption - 335 glucose isomerase - 308
gel filtration - 336 hydantoinase - 308
gel permeation - 336 invertase - 308
ion exchange - 336 lactase - 308
reverse phase - 336 penicillin acylase - 309
processing techniques penicillin amidase - 308
displacement development - 343 Enzymes - 2,48
elution development - 343 adenylsuccinate synthetase - 393
frontal analysis - 342 alcohol dehydrogenase - 48
scale-up - 370 aldolase - 398
considerations for proteins - 372 alkaline phosphatase - 393
strategy - 371 arylsulfatase - 393
theory - 348 aspartase - 48
plate number - 351 carbonic anhydrase - 393
plate height - 350 chymotrypsin - 393
resolution - 353 creatine kinase - 393, 398
effect of bead size - 357 enolase - 398
formate dehydrogenase - 48 organic acids - 288

fumarase - 48 pharmaceuticals - 287
glucoamylase - 399 protein - 288
glucose-6-phosphate dehydro- sugar - 295
genase - 48 vitamins - 303
a-glucosidase - 48 wine - 305
fi-glucosidase - 398 biopolymer interaction - 225
/3-glucuronidase - 398 commercial resins - 312
hexokinase - 48 equilibrium constant - 164
hexokinase isoenzymes - 393 isotherms - 164
isoleucyl-t-RNA trans- kinetic theory - 179
ferase - 394 models - 181
lactamase - 399 batch systems - 184
lactate dehydrogenase - 393, column - 187
398 fluidized bed - 190
lipolytic - 145 multiple stage - 192
lipoxygenase - 393 laboratory evaluation - 211
lysozyme - 368, 393 capacity, as CaCOs - 213
methionyl-t-RNA-trans- exchange capacity - 213
ferase - 394 feed apparatus - 219
micrococcal nuclease - 399 operating capacity - 213
monoamine oxidase - 393 pore measurement - 211
penicillin acylase - 48 materials - 194
phosphodiesterase - 398 functional groups - 198
protease, alkaline - 144 amine - 199
serum protease - 2 chelating - 200
trypsin - 393, 399,425 sulfonation - 198
Ethanol - 2,153 matrix - 194
Evaporation - 16 acrylic, methacrylic - 194
equipment - 18 cellulosic - 197
equipment suppliers - 23 epoxy-amine - 196
Fermentation - 1 inorganic - 194
bacterial - 2 phenol-formaldehyde - 196
broth - 2 styrenic - 194
Filtration - 6 particle density - 209
equipment - 7 particle size - 210
equipment suppliers - 10 porosity - 201
precoat - 7 surface area - 209
High fructose corn syrup - 300 operations - 229
demineralization - 301 backwashing - 246
Ion exchange - 164 batch - 229
applications - 268 column - 230
amino acid - 268 co-current - 230
antibiotics - 274 countercurrent - 232
catalytic uses - 310 fixed bed - 230
enzyme immobilization - 307 mixed bed - 232
high fructose corn syrup - continuous column -234
300 Higgins contactor - 234
Index 489
Himsley contactor - 233 reverse osmosis - 34

eiution - 244 ultrafiltration - 34
fluidized column - 236 Microorganisms - 7, 146
Asahi contactor - 237 B. subtilis - 146
CIoete-Streat contactor - Bacillus licheniformis -7
239 E, coli - 146
resin-in-pulp contactor - Penicillium chrysogenum -7
242 Poliovirus type 1 - 146
SWECO - 237 S. griseus - 8
USBM - 239 Streptomyces erythreus - 7
pretreatment - 229 Nernst equation - 179
regeneration - 244 Nernst-PIanck equation - 180
polyelectrolytes - 226 Organic acids - 2, 147
procedures - 221 acetylsahcylic acid - 288
strong acids - 222 citric acid - 288
strong bases - 221 itaconic acid - 288
weak acids - 224 xanthylic acid - 288
weak bases - 223 Organic amines - 147
processes - 220 Protein recovery - 288
concentration - 221 adenosine - 398
conversion - 220 albumin - 393
demineralization - 220 amastatin - 291
purification - 221 apolipoproteins - 393
resin limitations - 262 bovine serum albumin - 428
fouling - 262 chymotrypsinogen - 368
osmotic shock - 265 cytochrome c - 373, 393
oxidation - 264 dextranase - 290
thermal - 263 enkaphaiin - 395
resin manufacturers - 318, 347 from food stream wastes - 293
resin process designers - 319 glucopyranose amino sugar - 291
scale-up - 248 hemogiobulin - 393
computer models - 256 immunoglobulin G - 393
economic analysis - 260 insulin - 393
fixed bed - 248 interferon - 393
fluidized bed - 253 fl-lactoglobulin - 393
pressure drop - 255 myoglobuiin - 393
selectivity theory - 166 ovaibumin - 393
effect of co-ions - 174 phosphoinositides - 291
effect of crosslinking - 169 proteases - 290
effect of dissociation constant - serum albumin - 368
170 urgastone-rDNA - 394
effect of ionic size - 166 whey proteins - 394
effect of ionic strength - 168 Recovery of
effect of valence - 167 citric acid - 55
effect of zwitterions - 174 human growth interferon - 57
Membrane processes - 34 human insulin - 57
equipment - 39 human leukocyte interferon - 57
equipment suppliers - 42 micrococcal nuclease - 56
sodium clavulanate - 56 Sugar

Recovery operations - 4, 54 decolorization - 295
Solvent extraction - 42 demineralization - 298
equipment - 45 Vitamins - 2,150
equipment suppliers - 48 B-12 - 150, 287,303
Wine - 305

Dechow - Separation and Purification Techniques in Biotechnology

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

Dechow - Separation and Purification Techniques in Biotechnology

Uploaded by

Document Information

Original Title

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

Dechow - Separation and Purification Techniques in Biotechnology

Uploaded by

Copyright:

Available Formats

SEPARATION AND

Published in the United States of America by

Library of Congress Cataloging-in-Publication Data

Includes bibliographies and index.

Sorptive separation techniques are found in almost every product separation

Chapter 2 covers adsorptive separation, which is the oldest of the sorptive

Ion exchange procedures are described in Chapter 3. This chapter builds

Chapter 4, Column Chromatography Processes, covers the use of sorptive

I would like to acknowledge the helpful suggestions of Henry C. Vogel,

April 1989 Frederick J. Dechow

To the best of the Publisher’s knowledge the informa-

Mention of trade names or commercial products does

The book is intended for information only. The

2. ADSORPTION. ...................................... .65

2.5.4 Regeneration Operations. ......................... 125

3. ION EXCHANGE .................................... .163

3.6.4 Elution/Regeneration. ........................... 244

4. COLUMN CHROMATOGRAPHY PROCESSES ................ .333

4.8.9 Sugar from Molasses. ............................ 402

5. AFFINITY CHROMATOGRAPHY. ........................ .416

The recovery of products from fermentation or

The fermentation broth is the combination of insoluble,

For example, a bacterial fermentation for single cell

Compared with the feed streams to recovery processes

Table 1.1: Typical Product Concentrations Leaving Fermenters

Product Grams per Liter

The fluid volume in microbiological processes must be

Many fermentation broths are unstable. Once any broth

contamination from foreign organisms which can have the

Similar problems can occur if the recovery operations of

The necessary and practical recovery operations

Producl Product in broth

in CIIII. or .queous phase

Figure 1.1. Schematic of the processes which may be

If the desired product is extracellular, it is only

An important consideration in determining the

The major recovery difficulties arise when it is

1.2 RECOVERY UNIT OPERATIONS

The various methods for treating the fermentation broth

Filtration and centrifugation are unit operations in

Table 1.2: Separation and Recovery Techniques

the dissolved compounds so that it leaves the solution as a

Microfiltration, ultrafiltration, and reverse osmosis are

Adsorption, ion exchange, column chromatography, and

In solvent extraction, the removed compound establishes

Electrophoresis and electrodialysis are separation

electric field. Electrophoresis separates charged components

1.2.1 Mechanical Operations

According to this equation, the resistance to filtration

The specific practical limitations that must be

driving the first solids support.

Table 1.3 shows filtration design and operation for

Table 1.3: Representative Design and Operating Results for

The equipment in this operation can be as simple as an

wire mesh, woven cloth or cellulosic pads. With deep frame

Figure 1.2 shows the influence of pH on the filtration

Figure 1.2. The filtration time (t) divided by the volume of

For the sake of simplicity, it is assumed that the filter