CHAPTER 6

ORGANIC POISONING OF ANION EXCHANGE

RESINS

THE PROBLEM
Until recent years, water treatment by ion exchange was considered
as a largely inorganic process and resins were judged solely by their
ability to exchange or remove mineral ions. The total organic
content of natural water supplies is considerably lower than the
inorganic content, and when the standard of water purity required
for industrial purposes was 1-2 Jl-S/cm (S = mhos), organic con-
tamination of the input water was rarely sufficient to cause notice-
able changes in the quality of the treated water, or the apparent
working capacity of the resins. The greater purity requirements of
modern boilers, necessitating conductivities of down to 0·1 Jl-S/cm,
have converted into severe problems phenomena which were
virtually unrecognized until the 1950's. The nature of the problem
is best illustrated in Figure 37 which shows typical results of mixed
bed operation under a number of conditions. If the input water
contains no organic matter, then results such as Curve Bare
readily obtainable, and under ideal conditions, the treated water
quality can be as good as in Curve A. Typical results for London
water, a mixed and variable supply with about 300 p.p.m. T.D.S,
and 1-2 p.p.m. O.A, are given in Curve C, while old mixed beds
which have become poisoned will do no better than Curve D.
Indeed, some of the thin, organic-Ioaded waters occurring for
example in Wales will rapidly foul a conventional mixed bed to a
much greater extent, the conductivities being around 2-3 Jl-S/cm.
In the four cases shown, the total capacities of the resins are
proportional to the area (allowing for the logarithmic vertical
scale) between curves A-D and the horizontalline representing the
input water. The difference between the best and worst case is only
0·2 per cent. The operating capacity, on the other hand, is entirely
dependent on the quality of the water required. If a conductivity
below 1·0 is adequate, the total usable operating run for Curve A
is 9·5 kgr /ft 3 (21'8 g/l) while that of Curve D is 7·5 kgr /ft 3 (17'2 g/l),
90
T. V. Arden, Water Purification by Ion Exchange
© Butterworth and Co. (Publishers) Ltd. 1968
 THE PROBLEM

an operating capacity loss of 21 per cent. In 1950, therefore, this

unit would have been considered to be in quite serviceable condition.
With the quality standard of 0·5 fl-S /cm, the unit represented by
Curve D is giving no usable water at all, and since 1960, when this
standard has been commonly applied, its capacity would have been
regarded as zero.

500
_~~_~~e~~~~u.:.t~ iY_ _ __ _ ___ _____

100

E
u
I
~ 10'0

)
:1.
Z;o
~u
J
-go ' -0 I~ 0
U
0-5
X
I'=:: - - - - - -c- - - -X' - - - - - - - - - ?)Y
0'3 - - -
0 '1 8
A
0 2-01 4-0 1 6-0 I B-O 11(-Okgr/ftl
5 10 15 20 g/l.
Total an ions absor bed as CaC0 3

Figure 37. The effect of organic poisoning

By the same standard, Curve C shows an operating run XY of

9 kgr/ft3 (20,6 g/l), which is only 5 per cent less than the ideal case,
but a tightening of the conductivlty specification to 0·3 fl-S/cm
immediately reduces the operating capacity by 50 per cent to
X'Y', and the slightest deterioration in the results would result in
the whole of the water being rejected. Finally, no conditions except
the absence of organic matter (Curves A and B) give any treated
water, to the modern standard ofO·l fl-S /cm.
It follows that symptoms of organic poisoning can be discussed
only in relation to the standards of water purity required, and
91
 ORGANIC POISONING OF ANION EXCHANGE RESINS

statements in the ion exchange literature of the 1955-1960 period,

before this fact was realized, are accordingly often misleading.

ORGANIC MATTER IN WATER

Water originating from deep wells or springs is normally almost free
from organic matter, and can be treated by ion exchange without
special precautions. Problems arise in the case of surface waters,
such as rivers, lakes, or reservoirs, which contain organic matter
resulting from the decomposition of grass, ferns and leaves on the
land from which the water is collected. The most difficult cases are
the soft, thin coloured waters found in peaty regions such as Scotland
and parts of Wales, which contain both dissolved and colloidal
organic matter. There is no absolute method of measuring the
total quantity of organic matter present, and it is customary to use
one of a number of generally accepted empirical tests. A commonly
used method is the O.A. test, in which the oxygen absorbed from
potassium permanganate solution in 4 h at 27°C is measured. All
results quoted in this chapter refer to this method. Another widely
used test is similar, but carried out for 30 min in a thermostat
bath at 100°C. Finally, in the Kubel test, which is favoured in
Europe, the sampie is boiled für 10 min. The high temperature
tests tend to be slightly less reprüducible than the 27°C method, and
if the chloride content of the water is high, re action between chlo-
ride and permanganate yields indicated O.A. figures which are
slightly too high, and require a correction. There is no absolute
relationship between the results of the three methods, but for most
surface waters, a reasonably reliable general relations hip is

Kubel : 30 min 100°C: 4 h 27°C = 3 : 1·5 : 1

Typical values for the 4 h O.A. for various waters are as follows.
Deep well waters 0·1-0·4 p.p.m. as O 2
Normal surface waters 0,8-2,0
Peaty waters 5·0-10·0
The exact nature ofthe organic matter is not yet fully established,
and a considerable volume of research is being carried out by
authorities in many parts of the world. The methods in general are
based on removing the organic matter from the water by a suitable
anion exchange res in, eluting it from the resin with brine, and
dialysing the concentrated eluate to remove the salto Chromato-
graphie separations of the organic constituents are then carried out
92
 ORGANIC MATTER IN WATER

by ion exchange or gel filtration, the fractions being examined by

the standard techniques of organic analysis. It has been established
that natural organic matter consists almost entirely of a mixture of
polycarboxylic acids, of fairly high molecular weight, and of acid
strength similar to that of formic acid. The acids are conveniently
grouped into two main types, humic acids which are insoluble in
HCI at pH 1,0, and fulvic acids which are soluble under these
conditions. Humic acids have less carboxyl groups than fulvic, and
in general have a higher molecular weight. They are thus less
hydrophylic, and are not so readily absorbed by ion exchange

r··
2'0

1·5 ....,. /'--. \ / '.

E 1\
•• /'-., /
;.. \\ ... .
, \ • \.
1\
Cl. 1.0 .....,. •• '. _.... ,_ ...... I- •
Cl. r--'. .' .....- - -'. • ...
r-- '.'
"~ 0'5~J~M~M~~J~S~N~I~J~M~MLLJ~~S~N~I~J~M~MLLJ~S~~N~IJ~M~M~J~
~ FAJAODFAJAODFAJAODFAJA
~ 1956 I 1957 I 1958 I 1959

'~" 2'0
Cl
>.
)(
o 1·5

0·5 A 0 D F A J A 0 D F A J A 0 F A J A 0 0 F
5 NIJM MJ 5 NIJM MJ S NIJM MJ 5 NIJ
1959 1960 1961 19q2 1963
Figure 38. Seasonal variation in organic content of London water

resins. Once absorbed, however, they are very firmly held by normal
resins, and are difficult to remove. Both types are found in a range
of molecular weights, removability becoming more difficult as the
size of the molecule increases, with the larger humic molecules
presenting the most serious problem.
Studies of a large number of different surface waters in Great
Britain have shown that the organic content varies seasonally in a
regular fashion. A typical set of results is shown in Figure 38 for
London water over several years. The variation is due principaHy
93
 ORGANIC POISONING OF ANION EXCHANGE RESINS

to the fact that as vegetation decays in the autumn, its decomposition

products are leached by rain and find their way into rivers and
lakes, whose organic content therefore rises in October and is at a
maximum in the winter. With the new growth of vegetation in
spring the organic content falls and is at a minimum in summer.
There is some correlation between the heights of the peaks and the
type of weather in autumn. A warm wet autumn, giving rapid
wet degradation of the vegetation, gives rise to early high peaks,
while a dry autumn gives lower and more widespread maxima.
The seasonal effect is slightly enhanced by the fact that the bio-
logical degradation of the organic matter present in rivers and
lakes is higher in the warm weather than during the winter.
The nature of the organic matter itself varies according to the
time of the year and it has been shown in the author's laboratories
that water taken from a river in midsummer is much less poisoning
to ion exchange resins than the same water taken in midwinter.
The effect is greater than can be accounted for by the difference
in total organic content. It seems clear that there is a variation in
the ratio of humic and fulvic acids according to the seasons. In
winter time the type of organic matter present can cause severe
poisoning of ion exchange resins, while in summer the same water
may have no poisoning properties at all. The water supply at
Pontyclun, in South Wales contains organic matter which in
summer time totals less than 0·5 p.p.m., appears to have neg-
ligible ion exchange properties and is almost unabsorbed by any
type of resin. The same supply in winter may have an O.A. of 1·0
p.p.m. ofwhich 90 per cent is absorbed by anion exchange resins.

POISONING OF RESINS
In considering the poisoning of anion exchange resins by organic
matter, it is important to distinguish between the chemical effects
of soluble organic ions and the purely mechanical effect of the
presence of insoluble matter in the water, causing blocking of the
resin beds and consequently faulty operation. These two quite
different phenomena are often confused, since they give the same
symptoms of long rinse and poor water quality, and it is not un-
common to find water treatment plant users undertaking expensive
programmes of resin cleaning, or even replacement, when the
real requirement is efficient filtration of the incoming water,
together with a programme of vigorous agitation and backwashing
of the resin to remove physical blocking of the bed, which has
already occurred.
94
 POISONING OF RES INS

In between the extremes of soluble and suspended organic matter

is the twilight zone of colloidal material, which is somewhat diffi-
cult to define. Colloidal organic matter is readily removed by the
standard methods of coagulation with alum or ferric sulphate, but
a proportion of the soluble organic material is also removed by
these processes. The two cannot therefore be distinguished by this
means. As a very rough generalization, it may be said that when
the O.A. is less than 1·5 p.p.m., the organic matter is likely to be
largely soluble, and removable by purely ion exchange procedures.
As the O.A. increases above this figure, the probability of colloidal
matter increases, and waters recording O.A. values of more than
2·5-3·0 p.p.m. nearly always contain sufficient colloidal material
to render coagulation necessary as aprecursor of a demineralizing
unit. Colloidal organic matter is negatively charged, and becomes
attracted to the surface ofthe anion beads, where it becomes attached
to the outer layers of the resin network, interfering with the free
transfer of normal ions. This action is not significantly modified by
changes in the resin structure, whereas, as will be shown, the nature
of the resin skeleton has a profound effect on the degree of poisoning
by soluble organic matter.
It is also important to realize that uptake of organic matter by
a resin does not necessarily imply poisoning. The two processes are
often mentioned in the literature as if they were identical, a con-
cept which was partly true of the classical polystyrene-divinyl-
benzene anion exchange resins, which were not fully reversible
to the humate and fulvate anions. As a result, there have been pro-
posed various tests for poisoning, based on titrating the resin with
oxidizing agents such as dichromate. Such experiments merely give
a measure of the oxidizable organic matter present in the resin,
and give no indication of whether or not the resin is being poisoned
by it. If, for example, a resin were used to absorb simple organic
acids as formic or acetic, which do not cause poisoning of any type
of anion exchange resin, the dichromate test would show high
values after the first use of the resin. Similarly, natural organic
acids in water, when absorbed by resin, must always lead to a
positive dichromate test, even if they are of low molecular weight,
and correspondingly harmless. The symptoms of true organic
poisoning can be classified under three main headings.

(a) Strong Base Resins in Single Columns

This is the simplest case. Ifthe resin structure is such that organic
matter is absorbed irreversibly, then there is a steady decrease in
95
 ORGANIC POISONING OF ANION EXCHANGE RES INS

capacity as the resin is used, due to the blocking of ion exchange

sites by organic molecules. In addition, poisoned resin absorbs
sodium ions from the regenerant solution, and subsequently
releases them during the rinse. This process is slow, and the result
is a long period of rinsing, before the treated water conductivity
descends to an acceptable level. New resin normally requires about
5 bed volumes of water for adequate rinsing, whereas the require-
ment for poisoned material can rise to ten times this figure. The
emergent rinse water, while not sufficiently pure to be suitable for
service, is nevertheless very largely demineralized, and has thus
consumed some of the capacity of the resin. The operating capacity
of the resin is correspondingly reduced.

(b) Weak Base Resins in Single Columns

The problem in this case is more complex. In addition to a
primary action similar to case (a) there is a rate effect, which is
considerably more important. The exchange rate of the dassical
polyamine weak base resins is considerably lower than that of the
strong base materials, and as a result the operating capacity which
can be obtained, from the start of the exhaustion run to the point
at which the water quality deteriorates, is strongly dependent on
the rate of presentation of ions to the column. This value is itself
proportional to the product of the water flow rate, and the ionic
concentration. For a fixed resin capacity, the time taken für the
operating cyde is inversely proportional to this product, and it is
thus convenient to record the observed capacity against the length
in hours of the operating cyde. As has already been shown (Figure
16), the dassical polystyrene polyamine resins show capacity
reductions at short cyde times, and the effect is still more marked
with organically fouled materials.
In the earlier years of demineralizing practice, it was common
to use manually opera ted plants, which were calculated to run for
24 h or more between regenerations, in order to economize on
labour costs. Modern plants are increasingly being planned for
fully automatic operation, and to reduce capital costs, they are
designed to opera te at high flow rates, and short cyde times, down
to about 5 h. Under these conditions, the classical weak base res ins
show serious reductions in operating capacity after only a few
months service with surface waters, and the modern requirement is for
materials whose total exchange capacity may be lower than that ofthe
polyamine materials, but whose exchange rate is rapid, and unaffected
by the presence of organic matter in the water under treatment.
96
 POISONING OF RES INS

(c) Mixed Beds

The symptoms of mixed bed poisoning are similar to those of
case (a), but are more immediately recognizable, because of the
high quality of water demanded from mixed bed operation. In
addition, the anion resin of a mixed bed can show signs of severe
poisoning, and yet yield perfectly normal results if it is removed
from the unit, and operated in a two-bed system.
This effect is due largely to the method of operation of a mixed
bed, and is not directly connected with the structure of the resin
itself. Anion exchange resin which has taken up natural organic
matter, but is not necessarily poisoned by it, is at each regeneration
cyde brought into contact with the mineral acid solution used for
regenerating the cation exchange resin. The humic acids present
become precipitated in and around the resin beads. After re-mixing
the res in, a fraction of this anion material will have found its way
to the bottom of the bed and during the subsequent rinse and
service cydes the humic acids slowly re-dissolve and are re-absorbed
by the anion exchange resin with emission of a corresponding
concentration of inorganic acid. This is the reason for the charac-
teristic phenomenon of high rinse and conductivity associated with
organically loaded mixed beds.
The correctness of the mechanism described above can be readily
demonstrated by taking a sam pie of old, organically loaded,
strong base resin and placing it in a laboratory mixed bed regener-
ated in the normal fashion. The bed will show all the usual signs of
poisoning. On the next cyde the resins are separated and great care
is taken to prevent any anion material being left mixed with the
cation resin. The two resins are then regenerated separately and
re-mixed. During the next cyde the rinse is rapid and the quality
is excellent throughout the run. Finally, the reverse experiment is
carried out in which, during the res in separation stage, a pro-
portion of the anion exchange resin is deliberately left with the
cation resin. After normal regeneration and re-mixing, the operat-
results are extremely bad and the conductivity may never fall
below 5·0.
It is thus seen that if organic ions are present in the raw water
reaching a mixed bed of dassical design, then part of the humic
fraction must become precipitated, and the bed must in due course
show the symptoms of poisoning. This action occurs, even if the
anion exchange resin is one of the non-poisoning types, freely
reversible to organic matter which are described below. It follows,
that, while non-poisoning resins are preferable for mixed beds, as
97
 ORGANIC POISONING OF ANION EXCHANGE RES INS

for every other duty, their use is in itself insufficient to ensure good
results. While good design of the mixed bed unit will minimize the
degree of acid contact, it is impossible to eliminate it. Consequently
it is essential to remove the organic matter from the water, before
it reaches the mixed bed unit.

CLEANING POISONED RESINS

Two methods of removing organic matter from poisoned res ins
have been regularly practised during the last 10-15 years. The
first is to soak the resin in a mixture of brine and hypochlorite,
containing about 2 per cent available chlorine, for 12-24 h. This
procedure usually has a marked effect on the colour of the resin,
which in the poisoned condition is dark brown or black, but after
hypochlorite treatment is often restored almost to its original
colour. The treatment does improve the operating results, but not
as greatly as would be expected from the change in appearance of
the resin. It is clear that the organie matter present is bleached in
situ, rather than removed or destroyed. When resins of classical type
are used under severely poisoning conditions, cleaning may be
required at intervals of a few weeks, and under these circumstances,
the hypo chlorite treatment gives rise to sufficient chemical attack
on the res in to cause a marked reduction in capacity over 2-3
years life.
A more reliable method is to treat the resin with about 1·5 bed
volumes of a solution containing 10% NaCI and 1 % NaOH, which is
allowed to percolate through the bed during aperiod of 1 h or 2 h.
If the circumstances of operation permit an overnight soaking in
the solution, this procedure is advantageous. The treatment removes
a great deal of coloured organic matter from the resin, the emergent
liquor being black. I t will not normally restore the resin to new
condition, but will always effect a considerably improvement in
the results, and as it does not damage the resin, it can be carried
out as frequently as required.
I t is important, when treating poisoning waters, to carry out the
alkaline brine cleaning as a routine maintenance procedure, at
intervals of about one month, rather than to wait until the operating
results have seriously deteriorated. Organic matter becomes more
difficult to remove as the total quantity in the resin increases.
In a typical case known to the author, three mixed beds operating
in parallel on an uncoagulated water in South Wales consistently
gave results inferior to those of Curve D (Figure 37). A system was
introduced of treating one unit with alkaline brine every week, so
98
 THE EFFECT OF RESIN STRUCTURE

that each one ran for 3 weeks between treatments. The operating
results then recovered until the conductivities were consistently
about 0·5 p.Sjcm.

THE EFFECT OF RESIN STRUCTURE

The method of manufacture of the divinylbenzene crosslinked ion
exchange resins has been given in Chapter 2. The processes
described are correct in principle, and are presented in the same
way as in most textbooks on ion exchange resins. In one respect,
however, the diagrams are misleading, since they give the impression
that the final product possesses a flat, rectangular, and almost
crystalline molecular structure. This concept has been tacitly
accepted by many workers in the ion exchange field for many years,
because it gave a simplified picture of the resin structure which,
while not completely correct, has been accurate enough for the
understanding of simple industrial processes involving only small
inorganic ions. Some of the earliest work on this subject, was
carried out in 1949 by Duncan and Lister, who first demonstrated
that the selectivity coefficient for cation exchange was not constant,
but became smaller with increasing loading of the resin. The prob-
lem was studied for some years by a section of the Chemical Re-
search Laboratory, D.S.I.R., and in 1955, Reichenberg demon-
strated that the effect was associated with differences in the
environment of different sulphonic groups in the resin. He postu-
lated that groups occurring in loosely crosslinked regions, being
readily accessible would show different K values from those in
tightly knotted regions. Among much evidence produced for this
theory, there was a simple and convincing experiment, in which
different sampies of the same batch of hydrocarbon beads were
separately sulphonated so as to attach the active groups first to
the accessible parts of the structure, and secondly to whole mole-
cules. The two resulting cation exchange resins differed considerably
from each other in the way he had predicted. This inhomogeneity
of the crosslinked hydrocarbon polymer structures, which applies
equally to the cation and anion exchange resins made from them,
has subsequently been confirmed by many other workers.
Knowledge of crosslinked polymer structure in general and ion
exchange resins in particular, is less advanced than that of the
linear polymers, since many of the standard methods of investiga-
tion are carried out with solutions of the polymers. This approach
is impossible in the case of crosslinked materials, which are insol-
uble, and structure must be inferred from other measurements.
99
 ORGANIC POISONING OF ANION EXCHANGE RESINS

It is of the greatest importance in connection with the mobility

in the resin of large organic ions such as the humates and fulvates,
which can become entangled with the tight regions of the polymer
network. During the polymerization of styrene with divinylbenzene,
or indeed with any other crosslinking agent, two major forms of
inhomogeneity are introduced.

(a) Chain Entanglement

The mechanism of vinyl polymerization, with peroxide catalysts,
implies that new molecular chains starts at each catalytic initiation
centre. The chains are long, and since they are crosslinked, they
must spread out in three dimensions, and cannot lie parallel. The
crosslinked structure is thus aseries of separate 3-dimensional
networks, each of which grows independently of the others from a
single starting point, and it is clear that the resulting networks
must be intertwined. The effect is illustrated in Figure 39 in which
A and B represent parts of two 8 per cent D.V.B. crosslinked
styrene molecules, formed separately, and lying alongside each
other, without entanglement. Figure 39(c) shows the same two mole-

l~~
A
1~~ B C
Figure 39. Chain entanglement

cules, which have grown through each other du ring polymerization,

and cannot now be unlocked. Resin C is still 8 per cent crosslinked,
but is a very much tighter structure than A or B, with corres-
pondingly smaller pores. In practice, for the reasons given, resins
as normally manufactured resemble C rather than A and B in
structure.
The hypo thesis has been proved by the work of Millar et al. In
addition to any natural entanglement produced during normal
polymerization, they introduced controlled degrees of extra entangle-
ment by ta king the finished beads on a polymer containing, say,
4 per cent D.V.B. imbihing into them some ofthe original monomer
100
 THE EFFECT OF RESIN STRUCTURE

mixture, and repolymerizing. The final solid polymer still contained

4 per cent D.V.B. yet when it was converted to a cation exchange
resin by sulphonation, it had the swelling characteristics of an 8 per
cent crosslinked material, as normally produced. Similarly, a
2 per cent crosslinked polymer can be given the properties of an
8 per cent material, by reswelling it several times with its own
monomer mixture, and a 1 per cent can be converted to the same
condition also by multiple reswellings. It is, in fact, quite possible
that a material could be made with all the characteristics of a
normally crosslinked res in, while containing virtually no cross-
links at all, but purely an exceptionally high degree of entangle-
ment. Although all these materials are similar in appearance,
capacity, water content, and similar readily measured character-
istics, they do differ appreciably in their equilibrium and kinetic
results. For example, the reswollen 4 per cent D.V.B. material
retains the kinetics of a simple 4 per cent D.V.B. polymer, com-
bined with the swelling characteristics of anormal 7-8 per cent
material, and the equilibrium coefficients of anormal 15 per cent
resin.
In addition to producing extra entanglement, it is equally
possible to arrange for deliberate disentanglement during manu-
facture. To the monomer mixture is added a miscible solvent such
as toluene, which is compatible with the final polymer, but which
takes no part in the reaction. The polymer chains are thus forced
to grow in the toluene solution, in an expanded, less entangled
form. On completion of the reaction the toluene is removed, and
the expanded structure collapses, while remaining disentangled. It
thus has all the measured properties of a more loosely crosslinked
resin. It is possible to prepare aseries of resins, all closely similar
to each other by increasing the proportion of toluene while simul-
taneously increasing the D.V.B. content to produce extra crosslinks
to replace the lost entanglement. In this way there can be made, for
example, a 20 per cent crosslinked material with the measured
characteristics of an 8 per cent resin.
It is thus possible to produce a complete range ofmaterials, with
almost infinitely variable divinylbenzene content, and with pro-
perties not directly dependent on the D.V.B. content. This principle
is illustrated in Figure 40 where resins along diagonal lines of the
diagram have virtually identical swelling and capacity properties
although their D.V.B. contents are widely different.
The process of disentanglement by means of toluene can be
carried to the extent that the solid polymer chains are stretched,
101
 ORGANIC POISONING OF ANION EXCHANGE RES INS

and rigidly locked so that on removal of the solvent, the structure

does not collapse, but contains genuine holes, of up to 1000 A
diameter, which form water channels. The rigid over-expanded, or
highly disentangled structures are known as 'macroporous', and as
will be shown later, are one of the available types of resins resistant
to organic poisoning. The deliberately re-entangled structures have
no industrial use at present.

Multiple
'"ova re-entangled
~e-entangled 'Normal' Disentangled'

2
1if ~
4
~ ~ 1j
8
• • ~~
16 4t
~
-=
Figure 40. Varying structural types
11
A different method of making macropores is also available. In
this process a solvent is used with which the monomers are miscible,
but with which the final polymer is incompatible. During poly-
merization, micelIes of polymer precipitate out from the monomer,
later joining together to form a 'sintered' structure. The final
result (Figure 41) is very similar to that produced by the compatible
102
 THE EFFECT OF RESIN STRUCTURE

solvent system, and gives another type of macroporous or macro-

reticular resin which is also used industrially.
There are thus four main types of resin structure, which are not
distinct, but merge into each other. Theyare
Entangled
'Normal'
Disentangled
Sintered.
The 'normal' resin has no fundamental meaning, and is merely one
containing the average degree of entanglement occurring under
the polymerization conditions used, and the structure thus contains
a considerable degree of inhomogeneity, with wide variations in
pore size in different parts of the moleeule.

Figure 41. Sintered structure

(b) Differential Polymerization

A second effect, completely independent of the entanglement
process, is due to the fact that in a mixture of divinylbenzene and
styrene, three different polymerizations are possible. In practice,
there are more, since ethylstyrene is also present, but for simplicity
this monomer is ignored.
Styrene-styrene
Styrene-divinylbenzene
Divinylbenzene-divinylbenzene
The reaction rates of these three processes increase in the order
given above, so that as polymerization starts round the initial
catalytic centres, the first polymers chains formed are excessively
rich in D.V.B. and the remaining monomer mixture is correspond-
ingly poorer. The chains which grow later are in consequence
progressively more lightly crosslinked. The resultant polymer
103
 ORG ANIC POISONING OF ANION EXCHANGE RESINS

therefore is not uniform but has aseries of tightly linked centres

separated by more loosely linked regions as shown in Figure 42.
It follows that the pore sizes differ 'very widely from the average,
and a material which is nominally 8 per cent crosslinked, for
example, may have in the moleeule regions whose crosslinking
varies from as little as 3 per cent to as much as 30 per cent, with a
gaussian variation between. This structure is fundamental to all
copolymer networks which can be used in ion exchange resin
manufacture, and gives a second source of inhomogeneity.
From a knowledge ofthe resin structure, it is possible to build up a
picture of the mechanism of organic poisoning of the copolymer
anion exchange resins which, while difficult to demonstrate In

Early Final
Initiation stages copolymer

Figure 42. Progress of co-polymerization

detail, has nevertheless proved sufficiently accurate to explain the

observed effects, and provide at least two different means of over-
coming them. As the large organic anions travel through the resin
structure to exchange sites in the interior of the beads, they must
inevitably pass through both lightly and excessively crosslinked
regions. The knotted regions form bottlenecks in which the large
organic ions tend to collect. They can become entangled with each
other, and are unable to leave the structure to be replaced by OH-
ions during the normal process of regeneration. The quantity of
organic matter present in the water is normally insufficient to fill
more than a very small proportion of the highly crosslinked sites
in any one operating cyde, but as the resin continues to be used,
these sites become increasingly blocked by a concentration of
organic matter, thus making them unavailable for normal in-
organic ion exchange. The total capacity of the res in thus falls
slowly to a level which may be 80-90 per cent of the value für a
104
facing jwge IO.f
Plate 2. Fully automatic demineralizing plant at a large paper mill in Kent
 THE EFFECT OF RESIN1STRUCTURE

new resin, according to the proportion of knotted regions in the

particular resin sam pie. The process of poisoning then stops, and
the resin continues to operate at its reduced total capacity for very
long periods. The poisoned regions of the resin molecule have,
however, completely changed their character. The anion exchange
properties have been lost, since the positively charged exchange sites
are blocked. These regions now contain fixed carboxylic groups,
contained in the organic molecules which are now anchored
immovably in the resin structure. Parts of the resin have thus been
converted from anion to cation exchangers (Figure 43).

+
(CHJl3
ow

OW
(a)
Figure 43. Conversion 0/ anion to cation exchange sites (a) New resin. All quaternary sites
available (b) Poisoned resin. Quaternary sites in tightly crosslinked regions replaced by
carboxylic sites

1t is the presence of these cation exchange regions which causes

the phenomenon of long rinse times, with correspondingly high
rinse water consumption, which is characteristic of resin poisoning.
The fixed carboxylate groups are converted to the sodium form
during regeneration of the resin, and their subsequent hydrolysis
causes unacceptable effiuent water conductivities, until the sodium
ions have all been lost.

R - COO-Na+ +H 2 0 ~ R COOH + Na+OH- .... (37)

During this long rinse period, other, unblocked ion exchange
sites are becoming exhausted by inorganic anions in the influent
105
H
 ORGANIC POISONING OF ANION EXCHANGE RESINS

water, thus reducing the operating capacity to 50-60 per cent of

the original value, a much greater extent than would be calculated
from the simple loss of active groups by poisoning.

NON-POISONING RESINS

1. Macroporous (Macroreticular)
The problems of organic poisoning manifested themselves
increasingly during the years 1950-1960, due to the more stringent
standard of water quality which were applied during that period.
Towards the end of the decade, three different resin manufacturers
had developed independently, within a few months of each other,
the first method of combating the problem. It was reasoned that, as
differential crosslinking could not be avoided, its effects must be
reduced by producing a spongelike structure, containing a complex
network of water channels through which the large ions could
travel unhindered. The webs of solid material between the channels
would be extremely thin, as compared with the diameter of a
normal bead. A high proportion of the knotted resin regions would
thus be elose to the surface of the internal water channels, thus
giving the possibility that the organie matter collecting in these
regions could be freed during regeneration, to be removed through
the channels, which become known as macropores. The two
methods of producing macropores already described were invented,
and were highly successful in producing resins essentially reversible
to organic ions, with consequent freedom from poisoning. The
first demineralizing plants using resins of this type were installed
during 1959, and have since given many years of excellent operating
service with waters known to cause intolerably rapid poisoning of
the elassical D.V.B. gel resins.
Unfortunately, these resins, which became known as macro-
porous, or macroreticular, possessed a defect which partially
counterbalanced their obvious advantages. Because of the presence
of a considerable proportion of holes in the resin, their total content
ofactive matter, and therefore their capacities, were lower than those
ofthe corresponding D.V.B. gel materials. In addition, the method
of production, whether by the over-expansion or the sintering
procedure, resulted in a structure which held the simple inorganic
ions more firmly than had been the case with the earlier resins.
As a result, it became more difficult to remove these ions during
regeneration, and a greater quantity of caustic soda was required
to achieve the same degree of removal, that is, to free the same
106
 NON-POISONING RESINS

number of aetive sites for use during the next operating eyde.
Consequently, in addition to a loss of total eapacity, there was an
increase in the quantity of regenerant required to achieve a given
operating capacity. Plants containing the maeroreticular resins
thus suffered from higher capital and operating costs than those
using the earlier D.V.B. gel materials. However, the virtual
elimination of poisoning implied that the initial operating figures
were maintained over long periods, so that after several years in
service with bad waters, the macroretieular resins often retained
residual operating capacities than the dassical gel resins.

2. Isoporous
Following the diffieulties with the macroporous resins, the
opposite method of overeoming the poisoning problem was studied.
If poisoning were due to irregularities in erosslinking, it could be
prevented by making the crosslinking homogeneous, so that aIl
pores would be substantiaIly the same size, and tightly knotted
regions would no longer be present. It is impossible to achieve
this result in a crosslinked copolymer structure, by existing methods,
sinee the phenomenon of differential crosslinking is a fundamental
feature of the polymerization proeess. There is however, a weIl
known method of introducing crosslinking into anion exchange
resin struetures. It has been shown in Chapter 2 that the process
of anion resin manufacture indudes a chloromethylation stage,
which is the precursor to the addition of the active groups. During
this stage, a secondary Friedel Crafts reaction can oecur, between
a chloromethyl group and an adjaeent benzene ring

-
(38)
This reaction always takes place to some extent during chloro-
methylation, and for this reason anion exchange resins are always
107
 ORGANIC POISONING OF ANION EXCHANGE RESINS

slightly more tightly crosslinked than cation exchange materials

made from identical hydrocarbon beads. Normally, conditions are
used to minimize the side reaction, but it is possible to control it
so as to produce the whole of the crosslinking needed in the resin,
no divinylbenzene being used at all.
The methylene bridge crosslinking differs fundamentally from
copolymer crosslinking, in that its mechanism is largely random,
and does not lead to the bunching of crosslinks which is inherent in
the copolymer process. It is thus possible to produce a substantially
uniform network, free from tight regions. Resins produced in this
way have become known as 'isoporous', and they form the fifth
main type of structure. The difference between macroporous,
normal and isoporous resins is illustrated in Figure 44.

Macroporous Normal lsoporous

copolymer
Figure 44. Comparison of standard types

The isoporous resins have the same reversibility to organic

matter as the macroporous materials, removing it from the water
and releasing it on regeneration without loss of capacity or treated
water quality, and without increase in rinse requirements. They
do not, however, have the corresponding disadvantage of reduced
capacity, and in British practice at least, they have very largely
superseded the macroporous materials for demineralizing purposes.
Macroporous resins are still used for organic traps (see below) for
which their reduced demineralizing capacity is not disadvantageous.

3. Weak Base Materials

The structural considerations discussed above apply equally to
strongly and weakly basic resins ; but in the latter case, the effect
of the active group itself must also be taken into account. Any
primary or secondary amine, when reacted with chloromethylated
108
 NON-POISONING RESINS

polystyrene, will produce a weakly basic ion exchange resin, and a

large number of chemically different materials are available from
different manufacturers, in contrast to the two main types of strong
base materials, which are universally supplied.
The early weak base resins were commonly manufactured by the
use of polyamines, which, by adding more than one amino group
per benzene ring in the resin skeleton, gave resins of high total
capacity, which frequently exceeded 6·0 m.equiv.fg, or 2,500
m.equiv./l. These materials had correspondingly high operating
capacities, but were somewhat slow in action, and particularly
prone to organic poisoning, probably because of high attraction
between polyamine groups and polycarboxylic ions. When
poisoned, their exchange rate diminshed further, and they thus
became highly rate sensitive. When opera ted slowly, so as to give
eyde tim es in excess of 24 h, their operating capacities were not
markedly inferior to those of new resins. At high flow rates, and
short cyde times, the reduction in capacity was serious. The usual
phenomenon of high rinse requirement was, of course, independent
of operating flow rate. These effects were not removed by placing
the same active groups on macroporous or isoporous skeletons, and
it was necessary to use an active group which was intrinsically more
rapid in action, and more similar to a quaternary ammonium
group in its general properties. The logical step was to use the
smallest possible tertiary amine, that is to say, a resin based on
dimethylamine. A number ofresin manufacturers have now adopted
this principle, and one has gone further, by introducing a propor-
tion of strong base groups to give a further increase in reaction
rate.
The tertiary amine groups, in combination with a macroporous
or isoporous structure, are not only rapid when new, but are freely
reversible to organic ions. They thus preserve their original operat-
ing characteristics almost indefinitely. Materials of this type are
Amberlite IRA-93, De-Acidite H/IP, and Lewatit MP-60.

THE PROTECTION OF MIXED BEDS

(a) Prior Demineralizing Unit

For a large demineralizing installation the most satisfactory
method ofprotecting the final mixed bed is to use a prior demineral-
izing_ pair in which the anion res in is reversible to organic ions. If
the resin used is a non-fouling strong base type (e.g. Amberlite
lRA-900 and 910, De-Acidite FF/IP and N/IP or Lewatit MP-500
109
 ORGANIC POISONING OF ANION EXCHANGE RESINS

and 600), over 90 per cent of the organic matter is removed from
most waters, and released on regeneration of the resin. If the
corresponding weakly basic resins are employed (Amberlite IRA-93,
De-Acidite H/IP or Lewatit MP-60), the removal oforganic matter
is usually slightly less complete at 75-80 per cent. Nevertheless, the
degree of protection given to the foHowing mixed bed is almost
equally as good as that given by strong base resins, since both types
selectively absorb the most strongly held fractions of the organic
matter, which have the greatest poisoning potential. Both strong
and weak base resins carry out their full normal demineralizing
function in addition to removing organic matter, so that the mixed
beds are protected with no extra capital or running cost.

(b) Trap Columns

For small installations, in which the total demineralizing duty
can be performed by a single small mixed-bed unit, the capital
cost is increased by splitting the load between three columns. In
these cases, the mixed bed can be protected by a single column
of strong-base resin in the chloride form. Either macroporous or
isoporous resins can be used for this duty, the best results being
given by exceptionally macroporous materials having very high
affinities for organic matter, the correspondingly low capacity for
inorganic ions having in this case no importance.
Examples of res ins designed primarily for trap columns are
Amberlite IRA-904 and De-Acidite K(MP). The resins are re-
generated with brine, normally made slightly alkaline with caustic
soda to ensure that the organic matter is fuHy ionized, and therefore
in its most mobile condition. Organic removal by traps is lower than
by deionizing columns, being normally 60-70 per cent, but, as in
the case of weak base columns, the action is selective towards the
most obnoxious ions, and the mixed bed protection is far greater
than would be calculated arithmetically from the degree ofremoval.

PREVENTION OF POISONING, SUMMARY

The problem of organic poisoning has been considerably reduced
with the production of substantially non-fouling resins. While it
cannot be claimed that any one combination of processes will give
complete freedom from difficulty under all circumstances, attention
to certain basic principles can now be expected to re du ce the prob-
lem to manageable levels. The key points are as folIows:
(a) Where colloidal or insoluble matter is suspected to be present,
it must be removed prior to the demineralizing process to prevent
110
 PREVENTION

coating of the outside of resin beads, or blocking of channels in

macroporous materials. When the organic content of the water is
higher than 2-3 p. p.m. as carbon dioxide, measured by the 4 h
permanganate test, there is a high probability that part of the
organic matter is colloidal. Under these circumstance coagulation
is essential.
(h) Subject to the above, the use of modem non-fouling resins
of the isoporous or macroporous type in two-bed systems normally
gives no difficulty.
(e) Mixed-bed systems are inherently subject to poisoning, by the
extra mechanism discussed above and are more sensitive to it,
owing to the low conductivities involved. It is therefore essential
to protect the anion resin in the mixed bed, by removing ionizable
organic matter from the water before it reaches the unit. This can
be achieved either by a prior two-bed demineralizing system using
organic-reversible resins, or by means of an organic trap column.
(d) In attempting to estimate the poisoning power of a water,
note must be taken of the season of sampling. Sampies taken in
summer and early autumn are likely to present falsely optimistic
results.
(e) The poisoning power, which as a first approximation may be
considered directly proportional to the O.A. is also inversely
proportional to the total dissolved solids. The second relationship
sterns from the fact that the total quantity of water treated per
cyde is itself inversely proportional to the T.D.S. Thus at constant
organic content, if the T.D.S. is halved, the total weight of organic
matter taken up by the resin between regenerations is doubled, and
the probability of its becoming fixed in the structure is correspond-
ingly increased.
While no very precise figures can be quoted, the generalizations
in Table 7 may be taken as a reasonable guide.
In practice, waters in any of the groups tabulated may
give results characteristic of the category immediately above or
below, and the final decision on method can be made only as a
result of a pilot test, or of previous experience with the same water
supply. Unfortunately, the former is a laborious procedure, since
several hundred operating cydes are required to establish reliable
indications of the poisoning conditions. In the absence of previous
experience, it is therefore a sound general rule to treat all waters as
being in the category inferior to the one recorded in Table 7. In
particular for thin waters, the absolute value for O.A. given in
paragraph Ca), should be considered to override the ratio effect.
111
 ORGANIC POISONING OF ANION EXCHANGE RESINS
Table 7

OA/TDS* Satisfactory UnsatisftUtory

< 0·002 All systems, including None
unprotected mixed beds
with c1assical D.V.B. resins
2 0·002-0·015 Two-bed systems with all 'Unprotected mixed
resins. Mixed beds with beds with c1assical
non-fouling resins. D.V.B. resins
Protected mixed beds

3 0·005-0,015 Two-bed systems with Two-bed systems

non-fouling resins. wi th c1assical
Protected mixed beds, D.V.B. resins.
containing non-fouling All unprotected
resins mixed beds

4 0·015-0·03 Possibly as above, but As above

becoming unreliable

5 > 0·03 None All systems

unreliable.
Pre-coagulation
needed

• The figures refer to the 4 h 27°C test. If the Kubel or 30 min 100°C tests are used, the figures
should be multiplied by 3·0 and 1·5 respectively.

(f)Finally, in all cases where significant concentrations of

organic matter are present, i.e., Gases 2-5 of Table 6, regular
cleaning of the resin bed with alkaline brine is advisable as a
routing maintenance procedure, whether or not the operating
results appear to require it. In Gases 2 and 3, the interval between
treatments can be about 3 months, but in the borderline Gase 4,
a greater frequency, of up to once per month is preferable. In
Gase 5, the OAjTDS ratio after coagulation will be reduced, and
the pre-treated water can be reclassified in one of the first four
categories, the resin cleaning interval being selected accordingly.
While in many cases, the application of rules (a)-(e) may make
the cleaning procedure unnecessary, it is such a simple and inex-
pensive precaution that there is no object in omitting it.

112