EPA-600/1-77-035
June 1977
Environmental Health Effects Research Series
ELECTRO-REGENERATED ION-EXCHANGE
DEIONIZATION OF DRINKING WATER
Health Effects Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vi ronmental technology. Eli mi nation of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. “Special” Reports
9. Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL HEALTH EFFECTS RE-
SEARCH series. This series describes projects and studies relating to the toler-
ances of man for unhealthful substances or conditions. This work is generally
assessed from a medical viewpoint, including physiological or psychological
studies. In addition to toxicology and other medical specialities, study areas in-
clude biomedical instrumentation and health research techniques utilizing ani-
mals — but always with intended application to human health measures.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/1-77-035
June 1977
ELECTRO-REGENERATED ION-EXCHANGE
DEIONIZATION OF DRINKJNG WATER
by
Thomas A. Davis
Southern Research Institute
Birmingham, Alabama 35205
Contract No. 68-03-2209
Project Officer
Frederick C. Kbpfler
Health Effects Research Laboratory
Cincinnati, Ohio 45268
HEALTH EFFECTS RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
This report has been reviewed by the Health Effects Research
Laboratory, U.S. Environmental Protection Agency, and approved
for publication. Approval does not signify that the contents
necessarily reflect the views and policies of the U.S. Environ-
mental Protection Agency, nor does mention of trade names or
commercial products constitute endorsement or recommendation
for use.
ii
-------
FOREWORD
Man and his environment must be protected from the adverse
effects of pesticides, radiation, noise and other forms of
pollution, and the unwise management of solid waste. Efforts
to protect the environment require a focus that recognizes
the interplay between the components of our physical environ-
ment-—air, water, and land. In Cincinnati, the Environmental
Research Center possesses this multidisciplinary focus through
programs engaged in
• studies on the effects of environmental contaminants
on man and the biosphere, and
• a search for ways to prevent contamination and to recycle
valuable resources.
The Health Effects Research Laboratory conducts studies
to identify environmental contaminants singly or in combination,
discern their relationships, and to detect, define, and quantify
their health and economic effects utilizing appropriate clinical,
epidemiological, toxicological, and socio—econolniC assessment
methodologies.
Trace quantities of organic chemicals are present in most
drinking waters. This study was conducted as part of a broad
research effort to determine the potential health effects of
these substances. The objective of the study was to develop
a technique for removing inorganic ions from drinking water
samples without affecting the organic species. Water thus
treated could be subjected to reverse osmosis to produce con-
centrated solutions of organics with salt content low enough
to be acceptable for animal feeding experiments.
R.J. Garner, Director
Health Effects Research Laboratory
iii
-------
ABSTRACT
This report presents the development of a device for re-
moval of inorganic salts from drinking water to facilitate
the subsequent concentration of organic solutes for bioassay.
Prior attempts to concentrate the organic solutes by reverse
osmosis CR0) resulted in precipitation of the inorganic salts.
To prevent this precipitation, the drinking water is pretreated
by Electro—Regenerated Ion—exchange Deionization (ERID). The
ERID device developed for this purpose is essentially an elec—
trodialyzer with thick depleting compartments packed with a
mixture of anion— and cation—exchange resin beads. The resins
provide a conductive medium for electrical transport of ions
out of the demineralized water, through ion—exchange membranes,
arid into a concentrated waste stream.
Several types of ion-exchange resins were evaluated to
determine their electro—regenerability, their affinity for
organic solutes, and their release of contaminants. The elec-
trical conductivity of the resins equilibrated with solutions
of hydrochloric acid and sodium hydroxide provided an indica-
tion of the electro—regenerability of the resins. The uptake
of test organic solutes and the release of contaminants were
measured in batch and continuous experiments with aqueous media.
The best resin combination, a 2:1 mixture of Amberlite IRA—
68 and Duolite C—433, could be continuously regenerated while
deionizing tap water from Birmingham, Alabama, which contains
appreciable amounts of calcium, sulfate, and bicarbonate.
Precipitation of calcium carbonate in the waste compartments
was prevented by acidification of the waste feed.
Experiments with combined ERID—RO treatment demonstrated
that, while greater than 90% demineralizatiOn was achieved
with ERID, uncharged organic solutes tended to pass through
the ERID device with little change in concentration. They
could be concentrated by the RO unit if their molecular weights
were sufficiently high to permit rejection by the RO membranes.
Organic acids and bases tended to be removed from solution
by the ion—exchange resins, but they could be recovered by
rinsing the resins with acidic or basic solutions.
iv
-------
I.
II .
III.
IV.
V.
VI.
CONTENTS
• . . . . . 1
• . . • . . 6
• . . S • • 7
• . . . • . 8
• . . . . • 30
• S • • • • 55
References . . .. . • . • • • • • • • • • • • • S S
Glossary . . . . . . . • . . . . . • . . . . . . . . .
Appendices
A. Recovery of Bis(2—chlorOethYl) Ether by Sparging.
B. Reverse Osmosis Treatment of Water containing
Test Organic SoluteS. . . . . . . . . . . . •
C. Procedures for Operation and Maintenance of the
Prototype ERID Device . . . . . . . . . .
D. Sources of Materials for ERID Devices . . . • .
Foreword . . . • • . . . . .
Abstract . • . . . • . . . . • • • . . •
Figures. . . . • . . . . • • . • • • . .
Tables.. • . . . . . • . • . • • . . . •
Acknowledgments. . • . . . . . • • • • •
Introduction. • • • . . . . . . .
Conclusions • • . . . * . . . . .
Recommendations • . . • . . . . . .
Evaluation of Candidate Resins. . .
Design and Evaluation of ERID Devices
Performance of Combined ERID—RO System.
S • • • • S •
S • • S • S
S • • • S S
. • . S • •
. S • • • •
iii
iv
vi
vii
viii
59
60
62
65
67
70
.
S
S
S
V
-------
FIGURES
Number Page
1 Mechanism of ERID 3
2 Exploded view of plate—and—frame ERID device 32
3 Cylindrical ERID device 34
4 Fiber ERID device 35
5 Design of teflon resin compartments 43
6 Gasket spacer for waste compartments 45
7 Flow scheme for feedwater proportioning system. . . 47
8 Effect of feed conductivity on operating parameters
of ERID 50
9 Effect of applied voltage on operating parameters
of ERID 51
10 Reverse osmosis treatment of a solution containing
2.05 jig/liter of dimethylnitrosamine 57
v i.
-------
TABLES
Number Page
1 CharacteriStics and ConductivitieS of Candidate
Resins . . . . . . . . 10
2 Affinity of Resins for Test Organic SoluteS 15
3 Results of Dynamic Sorption Studies . 17
4 Data from Affir ity Studies with Humic Acid Fraction
and CC Resin • 20
5 Humic Acid Removed by IRA—68 Anion-Exchange Resin. . 22
6 Carbon Analyses of Extracts from Ion-Exchange
Resins 26
7 Organic Carbon Content of Aqueous Extracts from
Resins Washed with Organic Solvents. . . . . . . . 28
8 Comparative Data from ERID Devices with NaC1
Feed Solution. • . . 42
9 performance of ERID Device with 8 Teflon Resin
Compartments . . . . . . . . . 48
io performance of Prototype ERID Device with Tap
Water as Feed. • • • . • . . . 52
11 Carbon Content of Water before and after Treatment
with Prototype ERID Device containing 2:1
Mixture of IRA—68 and c—433. . • 54
12 Recovery of Bis(2—chlorOethYl) Ether by Sparging • . 64
vii
-------
ACKNOWLEDGMENTS
The project was carried out in the Engineering and Applied
Sciences Research Department under the general direction of
Dr. Donald R. Cowsar, Head, Biosystems Division, and under
the direct supervision of Dr. Thomas A. Davis, Principal Inves—
tigator. The laboratory ERID and RO experiments were carried
out by Mr. Joseph J. Forbes, Assistant Chemical Engineer, with
the assistance of Mr. R. David Couch, Engineering Technician,
and Mr. Raphael J. Thornton, Chemical Technician. Mrs. Ruby
James, Research Chemist, and Ms. Christina Gwin, Associate
Chemist, carried out analytical studies with test organic solutes.
TOC analyses were performed by Mr. Heinz J. Kollig of
Environmental Research Laboratory of the Environmental Protec-
tion Agency, Athens, Georgia.
Dr. Robert Kunin of Rohm and Haas Company, Dr. I.M. Abrams
of Diamond Shamrock Company, and Mr. Michael Gottlieb of lonac
Chemical Company graciously provided samples of resins, technical
data, and advice on selection of candidate resins.
viii
-------
SECTION I
INTRODUCT ION
The Environmental Protection Agency has an ongoing research
program to determine the toxicity of the organic contaminants
present in drinking water. To accomplish this objective, these
substances must be obtained in a state suitable for bioassay.
The total—organic-carbon content of drinking water from a sur-
face source is about 2 mg/liter. Thus, more than 1500 liters
(396 gal) of drinking water must be processed to obtain at
least 3 g of organics for animal toxicity tests.
Other contractors have conducted research on the use of
reverse osmosis to concentrate the organic chemical contaminants
of drinking water. With cellulose acetate reverse-osmosis
membranes, they have achieved 10 to 20—fold reductions of the
volume while retaining 85% of the total organic carbon present
in the original sample. Further concentration of the samples
is accomplished by lyophilizatiOrl, but precipitation of large
quantities of salt occurs. The salt content of the aqueous
concentrate is too high for bioassay, and furthermore, some
organics are co—precipitated with the salts. Extraction of
these salts and concentrates with penetane and methylerie chloride
has not proved efficient; consequently, overall recovery of
the organics is poor.
The purpose of the research program described herein is
to devise a practical method for deionizing aqueous solutions
of organics derived from drinking water. Such removal would
allow the aqueous solutions of organics to be concentrated
to a small volume for bioassay. The process for removing the
inorganic salts must not change the content of organics in
the sample by introducing contaminants, irreversibly absorbing
the organic solutes, or changing the chemical nature of the
organic solutes. Such a restraint greatly limits the applica-
bility of conventional dernineraliZatiOn processes.
Of the available deionization processes, ion exchange
is the most effective for complete removal of salts from solu-
tion. However, the cyclic nature of the process and the poten-
tial loss of organics during chemical regeneration and subsequent
rinsing of the resins makes conventional ion—exchange processes
unsuitable for this application. Instead, a process was chosen
1
-------
in which resins would be continuously regenerated electrically.
This process is called Electro—Regenerated Ion-exchange Deioniza-
tion (ERID).
In the ERID process, ion—exchange resins are used to remove
dissolved salts from solution, and the ion—exchange capacity
of the resins is regenerated in situ electrically. The essential
components of an ERID device, as shown in Figure 1, are the
anode, the anode—rinse—solution (anolyte) compartment, the
anion—exchange membrane, the resin bed containing a mixture
of anion— and cation—exchange resin beads, the cation-exchange
membrane, the cathode-rinse—solution (catholyte) compartment,
and the cathode. The feed solution containing the salt to
be removed flows through the resin bed and the rinse compartments.
The deionization process is illustrated in the bottom
portion of Figure 1. As the feed solution comes in contact
with ion—exchange resin beads, the Na+ ions exchange with H+
ions on the cation-exchange resin beads, and the C1 ions ex-
change with OH ions on the anion—exchange resin beads. The
H+ and 0H ions released in the process combine to form water.
Thus the resins are converted from the regenerated H+ and 0H
forms to the exhausted Na+ and C1 forms.
The electro—regeneration process is illustrated in the
top portion of Figure 1. When an electric potential is applied
to the electrodes, the cations, Na+ and H+, migrate toward
the cathode, and the anions, C1 and 0H, migrate toward the
anode. The ion—exchange beads are much more conductive for
their respective ions than the solution surrounding them; there-
fore, whenever possible, the ions migrate from bead to bead
in the direction of the appropriate electrode. When Na+ ions
are removed from exchange sites on cation-exchange resin beads,
they are replaced by the most readily available cations, which
may be Na+ ions from nearby exchange sites, Na+ ions from the
surrounding solution, or H ions formed by dissociation of
water. Similar considerations apply to the anions.
The ion—exchange membranes that form the barriers between
the resin bed and the rinse streams permit ions of the proper
electrical charge to leave the resin bed but they prevent op-
positely charged ions from entering the resin bed. For example,
the cation—exchange membrane (C in Figure 1) which is essentially
a sheet of cation—exchange resin, is permeable to Na+ and H+
ions but impermeable to C1 and 0H ions. The ions that pass
through the membranes are carried out of the ERID unit with
the electrode rinse solutions.
The ERID device can be operated continuously at steady
state with the electrodes being energized while the water is
being deionized, or it can be operated cyclicly. In cyclic
operation, the water is demineralized by simple ion exchange;
2
-------
CAT HO L V T E
ANOLYTE
PRODUCT
IA I C
02
H
+
ANODE
CATHODE
A ANION-EXCHANGE MEMBRANE
FEED
C = CATION—EXCHANGE MEMBRANE
Figure 1.
(NeC1)
Mechanism of ERID
3
-------
then the electrodes are energized when product water is not
being used to regenerate the resin bed. If the resins are
selected properly, the conductivity of the resin bed is very
low when the resins are in the regenerated form. Therefore,
the power consumption is minimal if the electrodes remain ener-
gized during periods of little or no demand for deionized water.
The ERID unit can deionize water at rates in excess of its
continuous capacity for short periods because the exchange
capacity of the resins is available to meet surge demands.
Selection of ion—exchange resins for use in the ERID unit
is very important because some resins can be regenerated more
efficiently than others. Early investigators, including Walters,
Wieser and Marek’, Glueckauf 2 , and Sammon and Watts 3 , demon-
strated that ion—exchange resins could be regenerated electri-
cally but the coulomb efficiencies they achieved were usually
below 50% and were often below 10% when dilute feed solutions
were used. These workers observed that the degree to which
the resins could be regenerated was quite low, and efforts
to regenerate them further resulted in progressively lower
coulomb efficiencies. Walters, Weiser and Marek’ obtained
6.2% regeneration with 52% coulomb efficiency, but by the time
the resin was 14% regenerated, the coulomb efficiency for the
entire time period had dropped to 17%, and the actual coulomb
efficiency at the end of the regeneration period was only 5%.
Prober and Myers measured the electrical conductivities
of several types of ion-exchange beads. Our evaluation of
their data revealed that the strong-acid and strong—base ion-
exchange resins used by previous investigators were much more
conductive in the regenerated form than in the exhausted form.
Thus, when a portion of the resin bed had been regenerated
electrically, most of the current passed through that regenerated
portion and was not effective in regenerating the remaining
portion of the resin. It follows that more efficient regenera-
tion would be achieved if the resins were more conductive in
the exhausted form than in the regenerated form. On the basis
of this hypothesis, a research program was conducted under
sponsorship of the Artificial Kidney—Chronic Uremia Program
of the National Institute of Arthritis, Metabolism, and Digestive
Diseases (NIAMDD) to develop an efficient system of electro—
regeneration to deionize water for use in artificial kidneys.
The research for NIAMDD involved selection of resins with
suitable electrical conductivity. In that situation it was
sufficient that the resins be chemically stable and electrically
regenerable. However, for the present application it is neces-
sary to know what and how much organic matter is picked up
or released by candidate resins or other components in the
system. If the organic content of the water does change, then
we must determine whether the resins can be pretreated to pre-
vent release of organics, and whether organics removed by the
4
-------
resins can be recovered in usable form. The other tasks involved
were the design of a prototype ERID device to demineralize
water at a flow rate of 600 mi/mm (9.5 gph) and the evaluation
of the prototype alone and in conjunction with reverse osmosis.
5
-------
SECTION II
CONCLUS IONS
ERID is an effective process for continuous deionization
of drinking water prior to treatment with reverse osmosis to
recover trace organic solutes.
When sodium chloride is the primary salt in the water,
the ERID device can be operated for an extended period under
a variety of conditions without danger of malfunction. When
calcium salts constitute a large fraction of the total inorganic
content of the water, it is necessary to operate the ERID de-
vice under conditions that do not allow precipitation of calcium
salts. Addition of hydrochloric acid to the solution entering
the cathode and intermediate—waste compartments effectively
prevents precipitation of calcium salts in the ERID device.
Ionic organic species such as amines and carboxylic acids
are removed from solution by ion—exchange resins. Most of
them, humic acid being the major exception, can be recovered
from the resins by rinsing with acid or base.
Some non—ionic organic species with low water solubility
are removed by the resins, but they may be recovered by rinsing
the resins with organic solvents.
Other organic solutes, e.g., nitrosamines, pass through
the ERID device with no appreciable change in concentration.
Commercial ion—exchange resins contain leachable organic
contaminants that can be removed by serial rinsing with acid,
base, methanol, diethyl ether, acetonitrile, and distilled
water.
Bis(2—chloroethyl) ether, and probably other volatile
solutes with low water solubility, can be effectively recovered
from water by sparging with nitrogen.
Aniines appear to be absorbed by the polyamide hollow fibers
in the Permasep® reverse osmosis module.
6
-------
SECTION III
RECOMMENDAT IONS
A 2:1 mixture of Amberlite® IRA—68 and Duolite® C—433
resins and lonac MA—3475R and MC -347O membranes should be used
in the prototype ERID device.
lonac CC, a weak—acid cation—exchange resin with low—swell-
ing characteristics, should receive further consideration.
If an effective method can be devised for removal of organic
contaminants, CC may be more useful than C-433 in the ERID
device.
Sparging of drinking water with nitrogen to recover volatile
organic solutes merits further study. A continuous sparging
process should be developed for pretreatment of drinking water
to recover volatile organics prior to treatment with ERID and
reverse osmosis.
A thorough study should be made to determine the extent
to which organic solutes, particularly amines, are absorbed
by the polyamide hollow fibers in the Permasep reverse osmosis
module.
7
-------
SECTION IV
EVALUATION OF CANDIDATE RESINS
ELECTRICAL CONDUCTIVITY OF RESINS
Since electrical conductivity in various ionic forms deter-
mined the electro—regenerability of resins, it was evident
that data concerning the electrical conductivities of various
commercially available ion-exchange resins would be needed
to make a selection of the most suitable resins for electro—
regeneration. The alternative would be to attempt electro—
regeneration of various combinations of resins with hopes of
finding a satisfactory combination. Since there are scores
of ion—exchange resins on the market, this second alternative
appeared to involve unrealistically high costs. Since the
manufacturers of ion-exchange resins do not provide data con-
cerning the electrical conductivities of their resins, it was
necessary to devise an experimental method of obtaining the
data.
The equipment for measuring electrical conductivities
of resins consists of two identical conductivity cells con-
nected hydraulically in series. One of the cells contains
the resin to be tested. The other cell contains only the
electrolyte solution with which the resin is equilibrated.
Initially a concentrated solution of the electrolyte is cir-
culated thrnucih the celic. and the conductance of the eellc
is measured precisely with a conductivity bridge (Industrial
Instruments Model RC—18). Stepwise dilutions of the electro—
lyte are made with deionized water, and the conductance values
are measured and recorded. At high solution concentrations
the conductance of the solution is greater than that of the
resin—filled cell, but in dilute solutions the resin is more
conductive than the solution. The values of conductance
measured in the two cells are plotted on logarithmic coordinates,
and the point at which the curve crosses the diagonal is deter-
mined. This equiconductance point, the point at which the resin
and solution have the same conductivity, is reported as the
conductivity of the resin in that particular electrolyte solution.
8
-------
NaOH and HC1 were the electrolytes used in the initial
screening of resins to determine their electro-regenerabilitY.
In NaOll, the ion—exchange capacity of cation—exchange resins
is completely exhausted with sodium ions, and the anion—ex-
change capacity is completely regenerated with hydroxyl ions.
In HC1, the cation—exchange resins are regenerated, and the
anion—exchange resins are exhausted. Therefore, the conduc—
tivities of resins in NaOH and HC1 would be a basis for deter-
mining the regenerability of a mixed bed of resins exhausted
with NaCl, which is the major inorganic salt present in many
drinking waters.
The identification, characteristics, and conductivity
data for all of the candidate resins are presented in Table
1. The C—20 strong-acid resin and IRA—400 strong—base resin
were not really considered candidates for electro—regefleration,
but their conductivitieS were presented for comparison with
the others. All of the weak—acid cation—exchange resins except
CS—l00 had conductivitieS in the range of 20,000 to 35,000
VS/cm in the sodium form, and the four measurements we made
in HC1 showed low values in the regenerated form. We concluded
that all of the weak—acid resins except cS—l00 would be suit-
able for electro—regeneration. Therefore, selection of the
best resin was based on other factors besides conductivity.
From the great variety of commercially available anion—
exchange resins, we selected representative samples of the
important classes of resins for our evaluation. Selection
was based on ion-exchange functionality and the chemical and
physical makeup of the resin matrix. The Type—I strong—base
resins have quaternary amine functional groups. The Type II
strong—base resins have a hydroxyethYl group attached to the
nitrogen that reduces the basicity and chemical stability of
the quaternary amine. Intermediate—base resins have a mixture
of tertiary and quaternary functional groups, and weak—base
resins have primary, secondary, and tertiary amine functional
groups in various proportions. Only the tertiary amine groups
of weak—base resins would be expected to be useful in this
application. The intermediate—base anion-exchange resins had
electrical properties that appeared most suitable for electro—
regeneration, and IRA—68 had the greatest difference in con-
ductivity between the exhausted and regenerated forms. The
two resins with combined anion- and cation—exchange functional-
ity were not sufficiently conductive for our purposes.
AFFINITY OF RESINS FOR ORGANIC SOLUTES
Some organic solutes have a high affinity for ion—exchange
materials. Organic acids and bases with low molecular weight
and high degrees of ionization behave similarly to inorganic
electrolytes in that they can enter into reversible ion ex-
change with the resins. On the other hand, it is difficult
9
-------
TPJBLS 1. CHAPAC ERISTICS ? ND (Y)NDUTIVITIES OF CANDIDATE RESINS
Eq ui cong uc ta nc e
points, US/cm
Designation Characteristics in NaOH in HC1
Weak-acid cation exchange resins
IRC 84b Crosslinked poly(acrylic acid) gel 30,000 350
XRC_50b Methacrylic acid—DVB gel 24,800 106
IRC-72 Methacrylic acid macroporous 25,000
C—433C Crosslinked polyCacrylic acid) gel 35,000 250
CS — 100C Phenolic macroporous beads 5,200 280
Bio-Rex 70 d Acrylic macroporous beads 29,000
Chelex lOOd Styrene—DVB, inuninodiacetate 20,000
cce Acrylic—DVB, low swelling 33,000
Acrylic—DVB, high capacity 34,000
Interm.diate-acid cation—exchange resin
ES_63c Phosphonic acid functional groups 8,000 870
Strong-acid cation-exchange resin
C—2O° Sulfonated styrene—DVB gel beads 23,100 77,000
Weak-base anion—exchange resins
Stratabed 93b Macroporous Btyrene—DVB beads 50 3,100
ES—374 0 Acrylic—DVB ntacroporous 49 15,500
AF P_329e Macroporous styrene-DVB beads 7,400
A_260e Aliphatic polyamine granules 34,000
WBSS b Macroporous styrene—DVB beads 18,000
IRA—47 Polyaniine gel beads
SRI—i 2:3 ratio PEI:EPON bulk gel 5,700
SRI—2 1:1 ratio PEI:EPON on string 7,000
SRI—3 3:2 ratio PEI:EPON on string 20,000
Intermediate-base anion-exchange resins
ES—34 0° Epoxy—amine gel beads 2,900 49,000
A—3OBC Gel beads 2,160 50,000
A—30G 0 Gel Granules 160 30,000
A_57c Porous beads 3,100 55,000
Bio-Rex 5 Analytical grade gel beads 2,150 39,000
A_305e Epoxy—pOlyanine gel granules, high 45,000
capacity
IRA_68b Crosslinked acrylic gel 74 30,000
A_365e Acrylic gel >1,000 44,000
Type-I l strong-base anion—exchange resins
IRA_410b Styrene—DVB gel beads 7,500 12,000
IRA 9l0b Macroporous atyrene-DVB beads 16,000 13,250
ASB_2e Styrene—DVB gel beads 9,100 13,000
A_550e Styrene, no DVB 16,000 20,000
Type-I strong-base anion-exchange resin
IRA 4OOb Quaternized atyrene-DVB gel beads 61,000 21,000
Anion— and cation-exchange groups on same resin
DS-21475C Thermally regenerable resin 420 110
AG 11A8d Ion retardation resin 6,000
a. The point at which the resin and the equilibrating solution have the
same conductivity.
b. Ainberlite trademark, manufactured by Rohm and Raas Company
c. Duo].ite trademark, manufactured by Diamond Shamrock Chemical Company.
d. Supplied by Sio—Rad Laboratories.
e. Manufactured by Ionac Chemical Company.
10
-------
to predict the behavior of organic solutes with high molecular
weights, multiple functionality, weak dissociation, and poor
water solubility. To screen the candidate resins and to gain
a better understanding of the problems we might encounter with
various classes of organic solutes in the ERID device, we Se—
lected several test organic solutes and measured their affinity
for some of the more promising candidate resins. We first
measured the equilibrium capacity of the resins for some of
the solutes; then we measured solute removal in a flow system.
Some of the solutes posed special problems that required addi-
tional study. These will be discussed at the end of this section.
Materials and Analytical Methods
Chloroform was obtained as an analytical reagent from
J.T. B k rChemica1 Co. Analysis was by GC (Hewlett Packard
5750 Research Gas Chromatograph) with electron—capture detec-
tion. Preliminary samples were determined by direct injection
of the aqueous solution. All subsequent samples were analyzed
by extraction of the chloroform into organic solvents suitable
for electron—capture analysis.
GC conditions for chloroform:
Column: 1.22 m by 6.35 mm stainless steel
Porapak Q
190°C
Detector: Ni 63, electron capture
220°C
Carrier: Helium, 30 mi/mm
Injection port: 210°C
Sample size: 1 iii
The GC response was calibrated by analyzing standard solu-
tions of chloroform under the same conditions used for analysis
of the samples.
Bis(2—chlOrOethyl) ether (boiling range 65—67°C at 2 kPa)
was obtained as iana1 Ucai reagent from Eastman Organic
Chemicals. Analysis was by GC with flame-ionization detection.
The aqueous solution was cooled and extracted with diethyl
ether (4 to 1). The ether solution was analyzed by direct
injection into the GC column. If the solute concentration
is low, the ether solution can be concentrated to extend the
limit of detection by GC; however, significant losses can occur
unless evaporation is carefully controlled.
11
-------
GC conditions for bis(2-chloroethyl) ether:
Column: 1.22 m by 6.35 mm glass
10% Apiezon L on 60/80 Gas Chrom Q
110°C
Detector: Flame ionization
190°C
Carrier: Helium, 30 mi/mm
Injection port: 140°C
Sample size: 25 p1
The GC response was calibrated by analyzing standard solu-
tions under the same conditions used for samples.
Since bis(2—chloroethyl) ether is rather volatile, we
studied the feasibility of recovering it from large quantities
of djlute aqueous solution by sparging with nitrogen. The
details of this study are presented in Appendix A. The results
indicate that sparging is an efficient method for recovering
this solute from dilute solutions.
Diethylnitrosamine (boiling range 63—64°C at 1 kPa) was
obtained from Eas€mar Organic Chemicals. Analysis was by GC
with flame—ionization detection. Direct injection of the aqueous
solution of the amine produced several peaks which interfered
with determination of low levels of the amine. A 10—mi aliquot
of the amine solution was extracted with 1 ml of chloroform,
and the chloroform extract was analyzed by GC. No interfering
peaks were observed when the chloroform solutions were analyzed.
GC conditions for diethylnitrosamine:
Column: 1.22 m by 6.35 mm glass
Chromosorb 101
145°C
Detector: Flame ionization
190°C
Carrier: Helium, 30 mi/mm
Injection port: 140°C
Sample size: 50 p1
The GC response was calibrated by analyzing standards
under the same conditions used for the samples.
12
-------
Dibenzofuran was obtained as a 98% pure compound from
Aldrich Chemical Company, Inc. Analysis was by GC with flame-
ionization detection. Aqueous solutions of dibenzofurafl were
extracted with benzene (10 to 1), and the benzene extract was
analyzed by gas chromatography. The GC response for dibenzo—
furan was calibrated by analysis of standard solutions of di—
benzofuran in benzerie under the same conditions used for the
analyses of samples.
GC conditions for dibenzofuran:
Column: 1 m x 4 mm I.D., glass,
3% OV—l on 80/100 mesh Gas Chrom Q,
120°C
Detector: Flame ionization,
190°C
Carrier: Helium
Injection port: 179°C
Sample size: 5 jil
DimethylnitroSamifle was obtained as a ‘ C—1abeled compound
from i ngIand Nuclear. The aqueous solution had a specific
activity of 5.21 Ci/mol. Concentrations were measured by liquid
scintillation counting. Concentrations as low as 1 jig/liter
(ppb) were easily measured by this technique.
Reduced Michier’S ketone , 4,4’—(methylene—’ C) bis [ N 1 N—
dimethylanilinei, wa ävai1able within the Institute as a ‘C—
labeled .compound with specific activity of 2.8 Ci/mol. Analysis
was by liquid scintillation counting.
DDT , l,l_biS(p_ChlOrOpheflyl)_2,2,2_t chboroethane, was
obtain as a ‘“C—labeled compound from New England Nuclear.
The crystalline material had a specific activity of 5.23 Ci/mol.
Analysis was by liquid scintillation counting.
A supply of 2,4,6—trimethYlaflilifle was available within
the Institute as a ‘ C—IabeIed compound with specific activity
of 2.7 Ci/mol. Analysis was by liquid scintillation counting.
Cholic acid was obtained as an 3 H-labeled compound from
New England Nuclear. The ethanol solution had a specific activ-
ity of 5.6 Ci/mmol. Analysis was by deep-well scintillation
counting.
13
-------
Humic acid , technical grade, was obtained from Aldrich
Chemiá FCompany, Inc. Analysis was by a combination of liquid
chromatography and spectrophotometry as described by Mantoura
and Riley. 6 Aliguots of humic acid in 0.2 N sodium hydroxide
were diluted to one liter with water and adjusted to pH 2.0.
An Amberljte XAD-2 column, 5 cm by 6 mm, was pretreated as
described in the reference, and the solution was passed through
the column at a drip rate of 12 ml/min. The column was back—
flushed with 4 ml of 0.2 N sodium hydroxide to elute the humic
acid. The eluate was diluted to 6 ml with 0.2 N sodium hydrox-
ide, and the absorbance was measured at 460 nm with water as
the reference. Absorbance of standard samples plotted against
concentration yielded a straight line.
Equilibrium Sorption of Solutes by Resin
During the early months of the study, we exposed several
candidate resins to solutions of five test organic solutes
to determine the extent to which the various resins differed
in their affinity for the solutes. Standard solutions of each
solute were prepared by addition of a weighed amount or mea-
sured volume of the solute to distilled water in a volumetric
flask. Aliquots of the standard solutions were diluted with
distilled water to produce working solutions in the ppm range.
Resins were pretreated by alternate rinsing in acid and base.
Then weighed samples of resin were placed in 300—mi BOD bottles
containing a Teflon® stirring bar. The bottles were filled
to capacity with the working solutions, stoppered, and shaken
on a horizontal shaker for about 16 hr. A control sample of
each working solution was treated in the same manner. After
equilibration the solutions were removed and analyzed by the
methods described above. Results are shown in Table 2.
The only resins that removed an appreciable amount of
chloroform were IRC—84 and IRA—4l0. Somewhat larger amounts
of bis(2-chloroethyl) ether and diethylnitrosamine were re-
moved by all of the resins tested. Affinity for humic acid
was measured at pH 2.2 and pH 7.0. The cation—exchange resins
appeared to have no affinity for humic acid at either value
of pH. However, all of the anion—exchange reslns had high
affinity for humic acid, and the affinity was greater at low
pH. All of the resins had considerable affinity for dibenzo-
furan. A 10—fold reduction in the amount of resin resulted
in a decrease in the degree of removal of dibenzofuran. The
proportions of 1 g of resin per 300 ml of solution is close
to the proportions expected in the final ERID device.
Of the three cation—exchange resins evaluated, Duolite
C—433 appeared to have lowest affinity for organics. Moreover,
since its values of conductivity were satisfactory (see Table
1), C—433 appeared to be the best candidate for use in the
ERID device, but we had insufficient data for exclusion of
14
-------
TABLE 2. AFFINITY OF RESINS FOR TEST ORGANIC SOLUTES
Bis—
H
umic
Aci
d
(2—chioroethy].)
ether
Diethyl—
nitrosainine Dibenzof
uran pH 2.2
pH 7
Chloroform
Initial solute concentration, mg/i 1.5 0.91 0.94 1.0 1.0 1.0 1.0
Amount of resin used, g 10 10 10 1 10 4 4
Portion of solute removal by resin, %
IRC—84 20 24 21 70 >90 0 0
C—433 3 29 14 36 75 * 0
CC >90 >90
I-J
U i Stratabed 93 —- -— —- —- - - 50 29
ES—340 0 5 11 33 54 82 59
A—30 B 0 14 8 28 —— 71 44
A—57 24 7 28 67 79 22
Bio-Rex 5 34 67
IRA—68 —— —— —— 20 51 —— ——
IRA—41 0 60 29 15 —— 84 97 70
IRA—910 —— 92 52
*A yellow substance that apparently leached from the resin interferred with analysis.
-------
the CC resin. The data on anion—exchange resins were not suf—
ficient to select the best resin, but some could be eliminated
on the basis of data in Tables 1 and 2. IRA—410 had a high
affinity for all the test organic solutes, and its conductivity
values were only marginally suitable. IRA—910 was eliminated
on the basis of its similarity to IRA—410. Stratabed 93 was
eliminated on the basis of rather low conductivity in the
chloride form. IRA—GB, which was not on hand during the first
four tests, gave encouraging results with dibenzofuran and
was considered a good candidate for further evaluation.
Dynamic Sorption Studies
The experiments described above were useful for scieeninq
several resins and test solutes, but they did not: repr sent.
the flow conditions that would be expected in the EP.ID proce .
Therefore, an experimental procedure was devised to expose
the resins to flowing solutions of the test solutes. The ap-
paratus consisted of a l—cm—I.D. chromatographic column with
a filter disc and Teflon stopcock (Ace 5904—T—22) and glass
tubing to convey the solution into and out of the column.
The column was filled with resin to a height of about 45 cm,
nd the r in was rinsed with 0 fl2 N NaC1 .colutirrn for 8 hr
rrien 3.5 liters of u.02 N NaC1 soiution spiked with the radio—
labeled test organic solute was allowed to flow through the
column over a period of about 5 hr. Hourly samples of the
effluent and samples of the feed and composite effluent were
analyzed by scintillation counting. In some cases the column
was extracted with an appropriate solution to recover the test
solute from the resins. Results of the dynamic sorption studies
are shown in Table 3.
Our early experiments indicated that diethylnitrosainine
might be a useful test solute. Since the diethyl form of the
compound was not available as a radiolabeled compound, we
decided to use dimethylnitrosamine (DMNA) in our studies with
dilute solutions. A feed solution containing 12 iig of ‘ C—
labeled DMNA per liter of 0.02 N NaC1 was used in dynamic sorp-
tion studies with IRA—68, CC, and a 2:1 mixture of IRA—68 and
CC resins. The IRA—68 resin alone removed 2.8% of the DMNA,
the CC resin removed 0.6%, and the mixture removed 5.5%. There-
fore, we would expect only a minor change in the nitrosamine
content of tap water treated with the ERID device.
When a solution containing 2 ig/1iter of lkC_labeled re-
duced Michier’s ketone was allowed to flow through a column
of CC resin, 98% of the radioactivity was removed from the
solution. Only 8% of the radioactivity was recovered when
the column was rinsed with 1.5 liters of 0.5 N NaOH. In a
subsequent experiment, the resin removed 96% of the radioactivity
from the feed solution, and 94% of the retained radioactive
compound was recovered when the column was rinsed with 3.5 liters
16
-------
-4
TABLE 3. RESULTS OF DYNAMIC SORPTION
STUDIES
Solute Resin
Solute removed
by resin, %
Extractant
Sorbed solute
recovered, %
Dimethylnitrosainine IRA-68
2.8
-
-
CC
0.6
—
—
IRA—68
& CC
5.5
—
—
Reduced Michier’s
ketone CC
98
base
8
CC
96
acid
94
IBA—68
17
base
acid
35
23
2,4,6—triinethylaniline C—433
8
acid
87
IRA—68
0
base
0
DDT IRA—68
& C—433
.
75
—
—
IRA—68
89
benzene
100
C-433
91
benzene
44
Cholic acid IRA—68
99
base
10
C—433
1.4
—
—
-------
of 0.5 N HC1. Apparently the resin had a high affinity for
the compound in the free base form, but when the amine groups
were protonated with HC1, the resin had a stronger affinity
for the hydrogen ions than the reduced Michier’s ketone. There-
fore, we concluded that this solute, and probably other amines,
can be sorbed and desorbed reversibly with the weak—acid cation—
exchange resins.
When IRA—68 resin was exposed to the reduced Michier’s
ketone in the dynamic sorption experiment, 17% of the solute
was removed by the resin. About 35% of this was recovered
when the column was rinsed with 3.5 liters of 0.5 N NaOH, and
another 23% was recovered in a subsequent rinse with 3.5 liters
of 0.5 N HC1.
Michier’s ketone contains two tertiary amine groups that
associate with the carboxylate groups of the weak—acid cation-
exchange resins. On the other hand, 2,4,6—trimethylaniline
has only one primary amine group which would be expected to
associate less strongly with the cation—exchange resins. When
‘ C—labe1ed trimethylaniline was exposed to the C—433 weak—
acid cation-exchange resin in a dynamic sorption experiment,
only 8% of the radioactivity was retained by the resin and
87% of this was recovered by an acid rinse. No measurable
loss of radioactivity was detected when the trimethylaniline
passed through a column containing IRA—68, and only a minor
amount of radioactivity was found when the resin was rinsed
with base.
In our first experiments with DDT, a 2:1 mixture of IRA—
68 and C—433 exhausted with NaC1 removed 75% of the ‘‘ C—labe1ed
solute as it passed through the column. When these resins
in the regenerated form were exposed separately to the solution
of DDT, they removed 89% and 91%, respectively. The resins
in the column were rinsed with 100 ml of methanol to remove
water, then they were extracted with benzene. Virtually all
of the DDT was recovered from the IRA—68, but only 44% was
recovered from the C—433.
A column of C—433 resin removed only 1.4% of the 3 H-labeled
cholic acid from a solution containing 72 ng/liter. However,
when a 59—ng/liter solution of cholic acid flowed through a
column of IRA—68, 99% was removed, and only 10% could be recovered
by rinsing with 3 liters of 0.5 N NaOH solution. Evidently
the carboxilic acid group of the cholic acid was strongly bound
to the amine groups on the anion—exchange resin.
Additional Studies with Humic Acid
In our initial studies with humic acid we observed some
unusual properties that required further study. In preparing
standards of humic acid, we found that reproducible results
18
-------
could not be obtained when individual samples were weighed
for calibration or affinity studies. However, reproducible
results were obtained when a large sample was weighed and dis-
solved, and aliquots of the concentrated solution were taken
for calibration standards and affinity studies. These results
suggest that the supply of huinic acid is not homogeneous.
Losses due to adsorption on surfaces could have occurred
during transfer or handling of samples. Humic acid was ob-
served to adhere strongly to glass wool, stirring rods, cotton,
ground—glass surfaces, and many other surfaces. Therefore,
great care must be observed in handling the humic acid solution
at the ppm level. On several occasions, what appeared to be
a clear, true solution of humic acid in water produced a preci-
pitate on standing, indicating either insolubility or possible
reaction. All these factors complicate the handling and analy-
sis of trace levels of humic acid in water.
Since humic acid is extremely complex chemically and its
composition varies with source, the affinity of resins for
humic acid material could better be defined by fractionation
of the humic acid into sub—groups and measurement of the af-
finity of the resins for each sub—group. The simplest approach
appeared to be fractionation based on pH of solution. A few
preliminary experiments have been conducted to determine the
feasibility of this approach.
In the conduct of the test, it was observed that precipita-
tion did not occur when humic acid, dissolved in 0.2 N NaOH,
was neutralized at pH 9. When the pH was adjusted below 3,
a slight precipitate formed, and a characteristic sulfide—like
odor was detected above the solution.
To prepare a “fraction” of humic acid for affinity studies,
a sample of humic acid was stirred with water at pH 3 (H 2 SO
was used to adjust pH) for several hours. The solution was
decantc d. nd the residue, insoluble at p 3, was washed and
stirred repeatedly with water at pH 3. The remaining solid
was equilibrated with water at pH 9 and stirred for several
hours. A portion of the solid dissolved. The solution was
decanted, evaporated to dryness, and used in subsequent af-
finity studies as the hurnic acid fraction soluble in the pH
range 3 to 9.
The data in Table 4 were obtained with the fractionated
humic acid sample. A 100—mi solution containing 24.2 mg of
the fraction of humic acid soluble in the pH range 3 to 9 was
prepared, and the pH of this stock solution was adjusted to
7. A 4—mi sample of the stock solution, adjusted to 5 ml with
0.2 N NaOH, had an absorbance of 1.2 on the DU spectrometer
at 460 nm. Another 4—mi sample, diluted to one liter to obtain
a 1—mg/liter solution of humic acid, was adjusted to pH 2 with
19
-------
HC1 and ionic strength 0.6 N with NaC1 and was passed through
a column of Amberlite XAD—2 resin. The XAD—2 column was ex-
tracted with 4 ml of 0.2 N NaOH, the volume of the eluate was
adjusted to 5 ml with 0.2Th NaOH, and the absorbance was found
to be 0.815. Evidently, 3 % of the humic acid was lost in
the process of concentrating the sample from one liter to 5
ml.
TABLE 4. DATA FROM AFFINITY STUDIES WITH HUMIC ACID
FPACTION AND CC RESIN
Absorbance at 460 nm
Sample Experiment 1 Experiment 2
Untreated stock solution 1.20 1.22
Stock solution diluted and
recovered on XAD-2 0.815 0.940
Distilled water contacted
with cc resin 0.018 0.08
Stock diluted, contacted
with CC, and recovered
on XAD—2 0.905 0.995
4.5 ml of isoamyl alcohol
eluate from CC 0.05
4.5 ml of H 2 0:Methanol:NH3
eluate from CC 0.04
4.5 ml of 0.2 N NaOH e].uate
from CC — 0.02
A 15—cm x 8—mm glass tube was packed with 0.5 g of CC
resin. A one—liter sample of distilled water was passed through
the CC—resin column, adjusted to pH 2 and ionic strength 0.6
N, and passed through the XAD-2 column. The NaOH extract from
the XAD-2 column had an absorbance of 0.018 at 460 nm, indi-
cating very little interfering material in the system. A 4—
ml sample of the stock solution, diluted to one liter, was
passed through the CC—resin column and concentrated to 5 ml
on the XAD—2 column. The absorbance was 0.905, somewhat more
than the analytical standard, indicating that little if any
of the humic acid was removed by the resin. (The experiment
described thus far was repeated, and the results shown as
Experiment 2 in Table 4 indicate that similar amounts of humic
acid were recovered from the columns.)
20
-------
Studies by other investigators indicated that isoamyl
alcohol might be a suitable solvent for removing humic acid
from a resin column. 7 The column of CC resin that had been
in contact with one liter of 1—ppm humic acid was eluted with
4.5 miof isoamyl alcohol, the volume of the eluate was adjusted
to 5 ml with isoamyl alcohol, and the absorbance of the eluate
was found to be 0.05. The CC column was then eluted with 4.5
ml of a 1 M solution of NH 3 in a 1:1 mixture of methanol and
water. The absorbance of this eluate was 0.04. Finally, the
CC column was eluted with 4.5 ml of 0.2 N NaOH; the absorbance
of that eluate was 0.02. The significance of the absorbance
values of these eluates is questionable since the values are
in the range of those obtained when there was supposedly no
humic acid in the system.
As we have discussed previously, a brown material has
been observed in the glass—wool plugs that we used to retain
the resin beads in the glass tubes. In an experiment to eval-
uate the ability of glass wool to remove humic acid from solu-
tion, the XAD—2 resin in the recovery column was replaced
entirely with GC—grade silanized glass wool (Supelco). A 4—
ml solution of the stock humic acid fraction was diluted to
one liter, adjusted to pH 2.2 with HC1, and adjusted to ionic
strength 1.2 with NaC1. Then it was passed through the glass—
wool column. The humic acid was eluted from the column with
4 ml of 0.2 N NaOH, the volume was adjusted to 5 ml, and the
absorbance was found to be 0.965 compared to 1.23 for a 4-mi
sample that had not been subjected to the process of dilution
and recovery on glass wool. We conclude from this experiment
that glass wool is just as effective as XAD—2 in recovery of
humic acid from acidic solution. These results suggest that
glasswool might be effective as a prefilter to remove humic
acid from tap water before it enters the ERID unit.
When solutions containing humic acid were contacted with
candidate resins in our equilibrium sorption studies, we found
that humic acid had a low affinity for cation—exchange resins
and a high affinity for anion—exchange resins. These results
were in agreement with data reported by Eonsack. 8 Moreover,
he found that Lewatit MIH, a granular, gel-type, condensation
polymer, weak—base, anion—exchange resin absorbed only small
amounts of humic acid, most of which could be recovered with
aqueous NaOH. To determine whether reversible sorption was
obtainable with our resins, we carried out experiments to ex-
tract humic acid from IRA—68 anion-exchange resin. A stock
solution containing 0.165% of humic acid was prepared by addi-
tion of excess dry crude humic acid to tap water. After this
mixture equilibrated, the undissolved material was removed
by filtration. Standard solutions were prepared by dilution
of the stock solution with feed solution (deionized water with
21
-------
NaC1 added to raise the conductivity to 150 pS/cm). The stan-
dard solutions were acidified to pH 2.2 and analyzed spectro-
photometrically at 350 nm to obtain a calibration curve of
UV absorbance versus concentration of humic acid.
Pretreated IRA—68 resin in the regenerated form was placed
in a chromatographic column and rinsed with 8 liters of feed
solution. A 3.5—liter batch of 66—ppm humic acid solution,
prepared by dilution of 140 ml of stock solution with feed
solution, was allowed to flow slowly through the column. Samples
of the feed, periodic samples of the effluent, arid a composite
sample of the effluent were acidified to pH 2.2 and analyzed
spectrophotometrically. The results are shown in Table 5.
TABLE 5. HUMIC ACID REMOVED BY IRA-68 ANION-EXCHANGE RESIN
Time, hr Humic Acid Removed, %
1 46
2 41
3 38
4 41
5 36
6 36
0—6 (composite) 44
During the experiment, brown bands of varying intensity were
observed on the column of resin. There was a very dark band
at the top followed by a lighter band and another dark band
in the lower part of the column. Apparently, the soluble frac-
tion of the crude humic acid contained several components.
The column containing the resin loaded with humic acid
was rinsed with 3.5 liters of 0.5 N NaOH over a period of 5
hr in an attempt to extract the humic acid. At the end of
5 hr, the effluent appeared to contain little if any humic
acid. Spectrophotometric analysis of the composite effluent
indicated that 29% of the humic acid had been removed by the
NaOH rinse. Subsequent rinsing with 0.5 N NaC1 and 0.5 N HC1
removed no additional humic acid. —
22
-------
The above experiment confirmed our previous observations
concerning the complexity of humic acid. In the material from
Aldrich, nearly half of the water—soluble fraction had such
a high affinity for IRA—68 that it could not be removed. If
we were to use humic acid as a test organic solute in our ERID
device, we needed to obtain a fraction that could be reversibly
sorbed and desorbed. Therefore, we designed another experi-
mental procedure for fractionating humic acid. To prepare
stock solution, humic acid was stirred into 0.01 N NaOH, and
HC1 was added to adjust the pH to 7. Then the solution was
centrifuged, decanted, and filtered to remove undissolved ma-
terial. The humic acid content of the stock solution was mea-
sured gravimetrically, and concentrations of subsequent solu-
tions were measured spectrophotometrically at 350 nm.
A 1.5 x 40—cm chromatographic column was filled with 100
ml of IRA—68 resin in the regenerated form, and 3.5 liters
of a solution containing 70 mg/liter of hurnic acid in deionized
water was fed slowly through the column. The first 500 ml
of solution from the column was essentially free of humic acid.
The effluent contained 42 mg/liter after 1000 ml of flow and
56 mg/liter after 3000 ml. The composite effluent contained
46 mg/liter or 66% of the initial humic acid. The column was
rinsed with 3.5 liters of 0.5 N NaOH. After 100 ml had flowed
through the column, the huinic acid content of the effluent
was 134 mg/liter, and it decreased thereafter. The composite
effluent contained 18 mg/liter or 64% of the hurnic acid that
had been removed from the original feed solution by the resin.
The solution containing the recovered humic acid was neu-
tralized to pH 7 with HC1 and fed through a fresh column of
regenerated IRA—68 resin. Essentially all of the humic acid
was retained in the column. However, only 61% of the retained
humic acid was recovered when the column was rinsed with 3.5
liters of 0.5 N NaOH. Evidently the quaternary—amine func-
tional groups of the resin held the humic acid so tenaceously
that it could not be recovered. Although the amount of humic
acid present in tap water would probably not be enough to alter
the performance of an ERID device, it appeared unlikely that
humic acid could be recovered quantitativety once it had been
removed from the water by the ERID device.
EXTRACTION OF ORGANICS FROM RESINS
Most ion-exchange resin beads are prepared by a pearl
polymerization process in which the monomers and polymeriza-
tion catalyst are suspended as droplets in an aqueous medium.
Impurities such as unreacted monomers, catalysts, and suspen-
sion stabilizers may remain within or on the surface of the
resin beads, and the routine washing of commercial resins does
not remove these organic impurities completely. When we ex-
tracted l—g samples of eight candidate resins with distilled
23
-------
water and obtained a tJV spectrum of the extracts, we found
evidence of organic contaminants released from six of the resins,
but we had no information about the quantity or identity of
the extracted contaminants.
A standard procedure (a modification of the procedure
recommended in Duo1ite Tech Sheet 105) was developed for pre—
treating resins to remove organic contaminants:
Step 1. The resin was placed in a beaker and rinsed with
deionized water.
Step 2. The water was drained from the resin, and two
bed volumes of 1.5 N NaOH were added. After
the mixture was stirred for 20 mm, the basic
solution was drained from the resin, and the
resin bed was rinsed with five bed volumes of
deionized water.
Step 3. The water was drained from the resin, and two
bed volumes of 2 N NC1 were added. After the
mixture was stirred for 20 mm, the acidic solu-
tion was drained from the resin, and the resin
was rinsed with five bed volumes of deionized
water.
Step 4. Steps 2 and 3 were repeated for 5 cycles, and
the resin was left in the regenerated form (i.e.,
Step 2 was the last treatment for anion—exchange
resins)
A 50—g sample of each pretreated resin was placed in a
5 00—ml Erlenmeyer flask and was rinsed 28 times with a full
flask of distilled water. (A 50—g batch of resin in an ERID
device would contact 14 liters of water in 48 hr.) Then 500
ml of distilled water was added to each flask, and the flasks
were shaken for 30 mm. The extracting water was decanted
from the resin through a glass—fiber filter that had been thor-
oughly washed with distilled water, and the filtrate was re-
served for carbon analysis. The 14—liter rinse followed by
the 500—mi extraction was carried out four times. In the fourth
extraction the resin was permitted to remain in contact with
the water for 48 hr before filtration. The extracts were placed
in clean glass—stoppered bottles and delivered to the Environ-
mental Research Laboratory of the Environmental Protection
Agency, Athens, Georgia, where they were analyzed for total
carbon and inorganic carbon. Organic carbon was determined
by difference. Analyses were performed with a Beckman Model
915 Total Organic Carbon Analyzer, to which a Model 215B infrared
detector had been added to permit simultaneous determination
of total and inorganic carbon.
24
-------
The results of the carbon analyses are shown in Table
6. The first extract from each resin contained a considerable
amount of organic carbon, second and third extracts had progres-
sively smaller amounts, and the fourth extract which had been
in contact with the resin for 48 hr contained more organic
carbon. The IRC—84 resin contained the largest amount of ex-
tractable organic material. Although extensive rinsing con-
siderably reduced the organic content of the extracts, the
high level in the 48—hr extract indicated ineffectiveness of
rinsing as a technique for cleaning the IRC—84 resin. The
other resins showed only moderate increases in the organic
content of the 48—hr extract.
A high concentration of inorganic carbon was found in
the first extract of the cation-exchange resins. The low levels
in the blanks indicated that the inorganic carbon did not come
from the distilled water. Evidently, a considerable amount
of CO 2 was released by the resins, possibly as a result of
decomposition of carboxylate groups. However, the low level
of inorganic carbon in the 48—hr extracts suggests that further
decomposition of carboxylate groups did not occur during the
experiment. The very low level of inorganic carbon in the
extracts from the anion—exchange resins is explainable. Since
the anion-exchange resins in their regenerated form have a
high capacity for bicarbonate ions, the dissolved C02 in the
distilled water was readily removed by ion exchange.
Several conclusions can be drawn from this experiment.
First, since IRC—84 released considerably more organic material
than the other two weak—acid cation—exchange resins, it should
be eliminated as a prime candidate for use in ERID. Second,
the standard procedure of cycling resins between acid and base
is inadequate as a pretreatment for removal of organic contami-
nants from resins used in ERID devices. This pretreatment
probably removes a large fraction of the total extractable
organic material from the resins, but the removal is not com-
plete. Third, rinsing with large quantities of distilled water
would probably remove most of the extractable organic material
from the resins, but the amount of water required would appear
to make the treatment impractical.
Since the standard resin pretreatment techniques were
inadequate, we developed and evaluated the following procedure
for serial rinsing of resins with organic solvents after the
usual pretreatment with acid and base:
Step 1. The resin was placed in a chromatography column,
washed with distilled water, and drained.
25
-------
TABLE 6. CARBON ANALYSES OF EXTRACTS FROM ION -EXCHANGER RESINS
Resin
Extraction
Total
Carbon,
Inorganic
Carbon,
Organic
Carbon,
Corrected
Organic
Carbon,
Designation
Number
mg/i
mg/i
mg/i
mg/i
cc
CC
CC
cc
1
2
3
4
250
14
ii
20
210
5.0
3.9
7.5
40
9.0
7.1
12.5
37
5.6
3.7
8.3
C—433
C—433
C—433
C—433
1
2
3
4
270
13
8.9
20
160
5.9
4.2
8.6
110
7.1
4.7
11.4
107
3.7
1.3
7.2
IRC—84
IRC—84
IRC—84
IRC—84
1
2
3
4
930
27
14
167
67
5.0
4.0
6.3
863
22
10
161
860
18.6
6.6
157
ES—340
ES—340
ES—340
ES—340
1
2
3
4
350
13
5.0
14
0.3
0.3
0.3
0.3
350
12.7
4.7
13.7
347
9.3
1.3
9.5
A30—13
A30—B
A30—B
A30—B
1
2
3
4
154
5.9
4.8
11
0.3
0.2
0.2
0.3
154
5.7
4.6
10.7
151
2.3
1.2
6.5
IRA—68
IRA—68
IRA—68
IRA—6$
1
2
3
4
260
4.3
3.3
8.0
0.3
0.1
0.3
0.3
260
4.2
3.0
7.7
257
0.8
—0.4
3.5
Blank
1—3
6.3
2.9
3.4
0
Blank
4
7.2
3.0
4.2
0
26
-------
Step 2. Two bed volumes of 1.5 N NaOH (2 N HC1 for cation)
were allowed to flow through the column at a
rate that provided 10 mm of contact.
Step 3. The resin bed was drained and rinsed with 5 bed
volumes of distilled water.
Step 4. Two bed volumes of 2 N HC1 (1.5 N NaOH for cation)
were introduced with 10 mm of contact.
Step 5. The resin bed was drained and rinsed with 5 bed
volumes of distilled water with 30 mm of contact.
Step 6. Steps 2 through 5 were repeated for a total of
5 cycles, and the resins were retained in their
exhausted form for further processing.
Step 7. The resin was then rinsed sequentially with two
bed volumes of methanol, diethyl ether, and
acetonitrile, each with a residence time of 20
mm. Finally, the resins were rinsed again with
5 bed volumes of distilled water.
The wet resins were placed in an oven and dried at 60°C
overnight. Each resin was then held under vacuum overnight
without heating. Then 50-g samples were placed in a chromatog-
raphy column and extracted with distilled water. The extrac-
tion procedure and sequence of sampling were as follows:
In an initial wash cycle, 5 liters of distilled water
flowed through the column over 60 mm. The first
sample was taken at the beginning of the 5-liter wash
cycle. The column was then filled and allowed to stand
for 30 mm. A second sample was taken, and the resins
were washed with another 5 liters of distilled water.
The column was refilled, and a third sample was taken
after a 30—mm waiting period. Finally, 30 ml of resin
was placed in the 60-mi sample bottle, and the bottle
was filled with distilled water and stoppered. The
water remained in contact with the resin for about
four days while the sample was transported to the EPA
laboratory in Athens, Georgia for TOC analysis.
The results of the carbon analyses of these aqueous extracts
are shown in Table 7. The blanks were samples of the distilled
water we used for the extractions, and they were inserted in
the order they were taken during the experiment. We used a
new batch of distilled water near the end of the experiment
beginning with Blank Sample G. In every case the larger value
of TOC in the fourth extract than in the third indicated that
either the 30—mm waiting period was insufficient for full
27
-------
A
B
1
2
3
4
1
2
3
4
C
D
1
2
3
4
1
2
3
4
E
F
G
H
1
2
3
4
I
J
1.5
1.5
8.3
2.2
1.7
3.7
12.4
3.3
3.9
34.0
1.5
1.6
1.7
4.5
2.8
9.6
5.1
3.9
2.0
2.6
1.4
1.3
1.8
2.0
8.2
5.0
3.1
4.3
1.8
1.5
0.2
0.2
0.2
0.3
0.3
0.4
0.3
0.3
0.2
0.4
0.4
0.3
0.2
0.4
0.4
1 . 0
0.1
0.].
0.1
0.2
0.1
0.2
0.4
0.5
0.3
0.4
0.4
0.6
0.5
0.4
1.3
1.3
8.1
1.9
1.4
3.3
12.1
3.0
3.7
33.6
1.1
1.3
1.5
4.].
2.4
8.6
5.0
3.8
1.9
2.4
1.3
1.].
1.4
1.5
7.9
4.6
2.7
3.7
1.3
1.1
TABLE 7. ORGANIC CARBON CONTENT OF AQUEOUS EXTRACTS
FROM RESINS WASHED WITH ORGANIC SOLVENTS
Total
Resin Sample Carbon, mg/].
Inorganic
Carbon, mg/i
Organic
Carbon, mg/i
Blank
I PA —68
Cc
Blank
C—433
ES—340
Blank
Blank
A—30B
Blank
28
-------
equilibration or some decomposition occurred during the 4—day
storage. The highest levels of organic carbon were found in
the extracts from lonac CC resin. This was unexpected because
CC was the cleanest resin in the experiment reported in Table
6. The results with Duolite C—433 were about the same as be-
fore. All of the anion—exchange resins released smaller amounts
of organic carbon than we measured in the previous study.
Therefore, we concluded that leaching with the organic solvents
followed by rinsing with 10 liters of water cleaned the anion—
exchange resins more effectively than batch rinsing with 56
liters of water. The sequential rinsing procedure described
above became our standard procedure for pretreating resins
for use in the ERID device.
SELECTION OF BEST RESINS FOR ERID
The conductivity data in Table 1, the data on affinity
of resins for test organic solutes in Tables 2 and 3, and the
data on extraction of resins in Tables 6 and 7 were all con-
sidered in our selection of the best resins for ERID. The
acrylic cation—exchange resins had the highest conductivities
in the sodium form, so we carried out further evaluations with
IRC—84, CC, and C—433. IRC—84 was eliminated on the basis
of the first extraction study, because it released considerable
organic material after extensive washing with water. The CC
resin was eliminated on the basis of high organic content of
aqueous extracts after the resin had been subjected to sequen-
tial washing with organic solvents. However, the results of
this single test should not discourage further evaluation of
the CC resin. All of the other tests indicated that CC was
an excellent candidate, and indeed may be the best candidate
if a suitable technique were developed for removing organic
contaminants. Duolite C—433 was selected as the resin that
would be used in the final prototype device because it had
good conductivity properties, released the smallest amounts
of organic contaminants, and had lower affinity for test organic
solute than the other two candidates.
The intermediate-base anion—exchange resins had electrical—
conductivity properties that made them appear most suitable
for electro—regeneration. These were avilable in bead and
granular forms with gel or macroporous matrixes. The macro—
porous matrix was eliminated because it was considered more
likely to adsorb organic solutes. Granular resins tended to
release debris. Therefore, gel beads were considered the best
resin form. Acrylic anion—exchange resins were particularly
appealing because they had the same matrix as the cation—exchange
resins, but they were obtained after most of the studies with
test organic solutes had been completed. The extraction studies
indicated that ES—340, A—30B, and IRA—68 were relatively free
of organic contaminants. Of these IRA—68 was selected for
use in the prototype device because of its acrylic matrix.
29
-------
SECTION V
DESIGN AND EVALUATION OF ERID DEVICES
DESIGN CONSIDERATIONS
The overall system for recovery of organic solutes must
have the capacity for treatment of 1514 to 1893 liters (400
to 500 gal) of water in a 48—hr period. Therefore, we selected
a design capacity of 600 mi/mm (228 gal/day) as the basis
for the development of a prototype device. Since ERID is capa-
ble of removing electrolytes from rather dilute solutions,
we chose to use ERID as a pretreatment for reverse osmosis.
The alternative would be to continuously demineralize the re-
circulating reject solution during the reverse osmosis treat-
ment. We set a goal of at least 90% demineralization of tap
water in a single pass through the ERID device. Our previous
studies demonstrated that more than 99% of the electrolyte
content of the feed water could be removed with low throughput
rates; however, larger amounts of the organic solutes would
likely be removed under such conditions.
In the Introduction, the principles of ERID were explained
in Figure 1. In that figure there is a single resin bed and
a single product stream, and the only waste streams are the
anolyte and catholyte. It is advantageous, but not essential,
that there be multiple resin beds between a pair of electrodes.
With multiple resin beds, the dimensions of expensive electrode
materials can be reduced, the ERID device can be more compact,
and a smaller amount of electric current is consumed. It is
important that there be no mixing of the product and waste
streams.
A recent analysis of tap water in Birmingham, Alabama,
showed that it contained 168 mg/liter of total dissolved solids,
22 mg/liter of calcium, 4 mg/liter of magnesium, 9 mg/liter
of sodium, 36 mg/liter of sulfate, 68 mg/liter of bicarbonate,
and only 1 mg/liter of chloride. Since the solubility of cal-
cium sulfate is 2 g/liter, a 5— to 10—fold increase in its
concentration in the waste stream would not be expected to
cause precipitation unless stagnation occurs at some point
in the waste compartments. Therefore, we attempted to maintain
a waste flow rate of about 60 to 120 mi/mm or 10 to 20% of
the product flow rate. The solubility of calcium carbonate
30
-------
is only about 15 mg/liter, and this salt could easily precipitate
in the ERID device if a high pH exists. Fortunately, calcium
carbonate is soluble in acid, so precipitation could be pre-
vented by acidification of the solutions that flow into the
rinse compartments, especially the cathode compartment.
Weak—acid and weak—base resins do not have sufficient
salt—splitting capacity to demineralize water unless they are
intimately mixed. If the two types of resin become stratified,
the demineralization performance is greatly reduced. Perform-
ance also deteriorates if channels develop in the resin bed
which allow uneven distribution of flow through the device.
In conventional ion—exchange processes, channeling and separa-
tion of resins can be avoided by downward flow of water through
a mixed ion—exchange bed. However, entrapment of air in the
resin bed is a serious problem in ERID devices because air
spaces between resin beds offer extremely high resistance to
electric current. Therefore, upward flow through the resin
beds is highly advantageous.
CANDIDATE DESIGN CONFIGURATIONS
All of the ERID devices evaluated in the early stages
of our research program had been designed and constructed for
previous sponsors. The ERID devices used in most of our labora-
tory experiments were of the plate—and—frame design shown in
Figure 2. They contained multiple resin beds with each repeat-
ing unit consisting of a cation—exchange membrane, a resin
compartment, an anion—exchange membrane, and a rinse compart-
ment comprising a piece of Vexar® netting with a silicone—
rubber gasket cast into its perimeter. The arrangement of
the internal components of the plate—and—frame ERID device
is illustrated in Figure 2.
The end plates served as clamping devices and connection
points for solution manifolds. Holes in the membranes and
compartments were aligned to form solution manifolds when the
unit was clamped together. One of the end plates had the
stainless—steel cathode attached to its inside surface. The
platinized—titanium anode was attached to the other end plate.
Use of the plate—and—frame design allowed considerable
latitude in design parameters. Although the cross—sectional
area of the resin beds was essentially fixed when the dimen-
sions of the electrodes were selected, the thickness of the
resin beds and the number of resin compartments were still
variables that could be increased to increase the capacity
of the ERID device. The plate-and-frame device could be made
entirely of commercially available materials. Moreover, the
device could be easily disassembled for inspection, and indivi-
dual components could be replaced if necessary before reassembly.
31
-------
ANION-EXCHANGE MEMBRANE
CATION-EXCHANGE MEMBRANE
I
Figure 2. Exploded view of plate-and-frame ERID device
10
0
10
0
0
32
-------
There were two major disadvantages to the plate—and—frame
design. First, although the components were flat, some of
the materials were not sufficiently resilient to form a water-
tight seal at their contact surfaces. Although seepage of
water to the outside did not pose any real threat of contamina-
tion of the product, it was nevertheless a nuisance. Second,
maintaining and assuring equal flow to several compartments
in parallel was a source of difficulty. Misalignment of the
components or accumulation of debris, prۖpitate, or gases
in one compartment could cause that compartment to receive
less than its full share of solution flow. Reduced flow in
a resin bed could cause a buildup of electrical resistance
and some decrease in demineralization performance, while reduced
flow in a waste compartment could result in precipitation of
calcium salts.
The cylindrical ERID device shown in Figure 3 was designed
to be operated at high pressures. The major structural compo-
nent was a 2-in. Schedule—40 pipe that served as the container
and the cathode. The cation—exchange membrane was tubular
Naf ion® (DuPont), and the anion-exchange membrane was made
by gluing a strip of lonac MA-3475R membrane in barber—pole
fashion with Pliobond® (Goodyear). This fabrication was neces-
sary because tubular anion—exchange membranes are not available
commercially. The anode was a 1.25—cm O.D. titanium tube coated
with a thin layer of platinum. The small diameter of the anode
was advantageous because platinized titanium is much more ex-
pensive than stainless steel. As a single—resin—bed device,
the cylindrical ERID device was simple and compact. However,
it lacked the versatility of the plate—and—frame design because
all of the dimensions were fixed by the dimensions of the
electrodes.
The fiber ERID device shown in Figure 4 was designed for
a commercial sponsor. It could be used with single or multiple
demineralization compartments connected hydraulically in parallel
and electrically in series. The ion—exchange fibers were cotton
yarn coated with crosslinked polyelectrolytes. This device
contained no ion—exchange beads or membranes. The ions removed
from the feed solution traveled along the ion-exchange fibers
and were discharged into the waste compartments.
Early in our research program we began con ucting experi-
ments with ERID devices to evaluate their demineralization
performance with various combinations of resins and membranes
and to determine suitable operating conditions for deminerali—
zation of drinking water. The experiments with ERID devices
were carried out concurrently with those described in Section
IV of this report. Therefore, some of the resins that were
eventually eliminated from further consideration in that study
were considered viable candidates when they were evaluated
in the ERID devices.
33
-------
STA I NLESS
STEEL P1
(CATHODE)
PLASTIC
SPACER
SCREEN
RES IN
BED
PVC CAP
Figure 3.
Cylindrical ERID device
CATION-
MEMBRANE
CATHOLYTE
INLET
PLASTIC
SUPPORT
SCREEN
FEED
SOLUTION
INLET
PLATINIZED
TITANIUM
ANODE
ANOLYTE
INLET
34
-------
Figure 4. Fiber ERID device
ANODE
35
-------
EVALUATION OF EXPERIMENTAL ERID DEVICES
For long—term unattended evaluation of ERID devices, it
is necessary to have an adequate supply of feed water of the
desired composition, a controlled power supply, and a method
of monitoring and recording operating conditions. The power
supply consisted of an autotransformer to control applied volt-
age, an incandescent light bulb to limit the current, an isola-
tion transformer to protect the operator from electrical shock,
a rectifier bridge to convert the AC to DC, a volt meter, and
an ammeter. A voltage divider and a shunt provided millivolt
signals of voltage and current for the 12—point recorder.
The conductivities of the various solution streams were moni-
tored by conductivity cells connected through a commutator
in the recorder to a single conductivity monitor. The pH of
the product stream was also monitored and recorded.
Feed solutions for our initial studies were prepared by
addition of NaC1 to deionized water in a 1893—liter (500—gal)
polyethylene tank. Unless otherwise noted, the feed conduc-
tivity was about 150 pS/cm. Product water from the ERID devices
was discharged into an identical tank for subsequent use in
feed solutions. Feed water was pumped from the tank to a
constant—head, overflow standpipe with a mROY® controlled-
volume pump Model R132A (Milton Roy). The conduits were Type—
316 stainless steel or Teflon. The feed water flowed through
an activated—carbon cartridge followed by a particulate filter
when test organic solutes were not in the feed water.
Two plate—and-frame ERID devices were used in the labora-
tory evaluations. They differed only in the materials of which
the end plates were constructed -- Micarta® (Westinghouse)
and Plexiglas® (Rohm and Haas) -- and the external manifolding
for solution flow. The membranes, spacers, and resin compart-
ments were interchangeable. Usually both of these devices
were operated with the same feed solution but with different
combinations of resins and membranes. Since differences in
their designs had no effect on their demineralization perform-
ance, no distinction will be made in the discussions that
follow.
Ion—exchange membranes from a variety of sources were
evaluated in the plate—and—frame device to determine which
was most suitable for this particular application. However,
we were also looking for an anion—exchange material that could
be manufactured in tubular form. Since RAI Research Corpora-
tion had expressed a willingness to fabricate tubular membranes,
we evaluated two of their fluorocarbon—base membranes in sheet
form. To obtain a basis for comparison.with unknown membranes
and resins, we assembled a two—bed ERID device with proven
materials: Ionac MC-3470 cation—exchange membranes, lonac
MA—3475R anion-exchange membranes, and an equal-volume mixture
36
-------
of ES—340 and IRC-84. When the anion—exchange membranes were
replaced with JL200 membranes from RAI, the electrical resis-
tance of the ERID device increased considerably, and the de—
mineralization performance was reduced. Thus, JL200 was not
suitable for use in the ERID device. An lonac MA—3148 membrane
evaluated the same way showed low electrical resistance but
poor demineralization performance.
The RAI Perimion® 4025 membrane proved to be a suitable
replacement for the lonac MA—3475R anion-exchange membrane.
There was no measurable difference in the demineralization
performance and electrical resistance when these membranes
were interchanged. An ERID device containing lonac MC-3470
and Permiori® 4075 membranes and a mixture of 3 parts of IRA—
68 to 2 parts of CC resins was operated for 38 days with NaCl
feed solution. During most of this period at least 90% de—
mineralization was achieved with an average product flow rate
of about 60 ml/min in each of the three resin compartments.
However, during shutdown for holidays, a tear developed in
one of the Permion membranes, apparently due to shrinkage when
the membranes became partially dry during shutdown. Although
the Permion 4025 and lonac MA—3475R membranes appeared to be
comparable in their ion—transport properties, the lonac membrane
had superior mechanical strength and was, in fact, the most
durable anion-exchange membrane we evaluated.
Neosepta® AV—4T anion—exchange and CL-2.5 T cation—exchange
membranes from Tokuyama Soda Company Ltd. were evaluated in
an ERID device with two compartments filled with a mixture
of 3 parts of IRA-68 and 2 parts of CC resins. The performance
of this device was excellent. We achieved 96% demineraliza—
tion with a flow rate of 65 ml/inin through each compartment
and 99% demineralization with a flow rate of 30 mi/mm; the
couionibefficiency, equivalents of salt removed per faraday
of current, was about 50%. The Neosepta membranes were con-
sidered good candidates for use in ERID, but their mechanical
strength was not as good as the lonac membranes.
In addition to the weak—acid cation—exchange resins mentioned
above (i.e., Amberlite IRC—84 and lonac CC), we also evaluated
Duolite C—433. The C—433 resin was not used in an ERID device
until our leaching studies revealed that the other two resins
released excessive amounts of organic solutes. We carried
out extensive evaluations of C—433 in the final prototype device
and found that the demineralization performance of C-433 was
comparable to that of IRC-84 and CC.
Both the 1:1 mixture of ES-340 and IRC—84 and the 3:2
mixture of IRA—68 and CC produced water that was slightly acidic.
The appearance of excess hydrogen ions in the product suggested
that the cation—exchange resins were more effective than the
37
-------
anion—exchange resins in removing ions from the water. The
closeness of the equiconductance points (see Table 1) of IRA—
68 and CC in the chloride and sodium forms, respectively, sug-
gests that a 1:1 ratio should be optimal. But, we found that
a 2:1 ratio was needed to achieve a neutral pH in the product.
We concluded that something other than the conductivity of
the resins affects the pH of the product, and we suspected
that dissolved carbon dioxide from the air may contribute to
the problem.
The importance of swelling of resins was observed when
a resin bed containing a 1:1 mixture of ES—340 and IRC—84 was
partially occluded by a microbial colony that restricted the
flow of the solutions through the compartments. When the volt-
age applied to the electrodes was lowered, the flow rate and
conductivity of the product decreased, and the coulomb effi-
ciency increased. When the voltage was raised, the reverse
occurred. Evidently the lower voltage allowed a larger portion
of the IRC—84 resin to be converted to the more highly swollen
sodium form, and the swelling further restricted the flow of
solution through the resin bed. A low—swelling resin such
as CC would have been beneficial in this situation. The CC
resin swells only 27% on full conversion from the hydrogen
to the sodium form; whereas, IRC—84 swells 65%, and C—433
swells 90%.
The most severe problem encountered in the operation of
the plate—and—frame ERID devices was the gradual accumulation
of a microbial slime in the bottom of the resin beds. Although
the slime probably had some detrimental effect on the elec-
trical conductivity of the resin beads, its major effect was
to reduce the hydraulic permeability of the resin beds. In
our previous studies we found that the slime could be controlled
by periodic rinsing of the system with Clorox® or formaldehyde.
Such a treatment was more effective in prevention of the slime
than in removal after it had accumulated. Moreover, these
treatments would not be suitable for routine use in the proposed
application of ERID, because formaldehyde would contribute
to the organic content of the water, and Clorox may change
the chemical composition of the organic solutes. We found
that adequate flow rates could be restored if the ERID device
were simply inverted, and all of the inlet and outlet connections
were switched. Although this procedure did not completely
rid the device of the microbial colonies, it did allow continued
operation until thorough cleaning could be accomplished. Ap-
proximately 4 hr were required to restore steady—state operation
after the ERID device had been inverted.
The normal operating mode of the plate—and-frame ERID
device was constant upward flow through all compartments in
parallel with constant electrical current flow through all
of the compartments in series. While the device was assembled
38
-------
with Neosepta membranes and a 3:2 mixture of IRA—68 and CC
resins, we carried out an experiment to determine its perform-
ance with solution flowing through the two resin beds in series
rather than parallel. This change resulted in a slight reduc-
tion in coulomb efficiency, a slight increase in the conduc-
tivity, and a significant increase in the hydraulic pressure
drop through the device. Although the results demonstrated
that series flow through the resin beds was feasible, there
appeared to be no advantages to such an arrangement. If longer
residence time were needed, it could be achieved more effi-
ciently with longer or thicker resin beds or with lower flow
rates.
To study the effect of increasing the thickness of the
resin beds, we constructed two resin compartments of 19-mm
(0.25—in.)-thick Plexiglas and filled them with a 3:2 mixture
of IRA—68 and CC resins. This assembly contained two pairs
of lonac MC—3470 and MA-3475 membranes. With a product flow
rate of 239 ml/min, the degree of demineralization was 91%
with 340 mA of electric current and an applied potential of
48 V. To increase the thickness of the resin compartment
further, we removed the center waste compartment and its as-
sociated pair of ion—exchange membranes. This 38—mm—thick
resin bed gave virtually the same demineralization performance
as the two 1.9—cm—thick resin beds, but it required 545 mA
of electric current. The major advantage to thick resin com-
partments is that fewer membranes are needed for a given through-
put capacity. The disadvantages arise from the increased cur-
rent densities that are required when fewer membranes are used.
For example, when eight 6.35—mm—thick resin beds were used
to treat a total of 240 mi/mm flow, the current was only 100
mA and over 98% dernineralization was achieved. Therefore,
thin resin beds were considered more applicable to an ERID
device operating in the steady—state mode.
Thick resin beds were considered to have potential advan-
tages in a cyclically operated ERID device in which the total
ion—exchange capacity could be more important than the rate
of regeneration. The 38—mm—thick resin bed was used in an
experiment to evaluate the cyclic mode of operation. The ERID
device was operated with a low feed rate and a high applied
voltage to regenerate the resins as much as possible. Then
the power supply was turned off, and the product flow rate
was maintained at 170 rnl/min for 30 hr. During this period,
306 liters of water with initial conductivity of 150 j.iS/cm
flowed through the resin bed, and the conductivity of the pro-
duct increased steadily from 1.5 to 90 jiS/cm. This device
contained about 75% of the volume of resin used in the final
prototype device, but after treatment of less than 20% of the
desired amount of feed solution, the conductivity of the pro-
duct was far above the acceptable level. Therefore, cyclic
operation of ERID did not appear appropriate for this application.
39
-------
The cylindrical ERID device shown in Figure 3 had been
constructed during a previous research program about four years
prior to the present study, and it has been stored under dry
conditions during the interim. On inspection its membranes
appeared to be intact. It was filled with a 1:1 mixture of
ES—340 and IRC-84 and operated with a feed solution of NaC1
in deionized water. Less than 70% demineralization could be
obtained, and we found evidence of intercompartmental leakage.
The device was disassembled, leaks in the anion—exchange membrane
were sealed with Silastic® (Dow Corning) RTV 731, and a ribbon
spring from a mechanical timer was used to hold the cation-
exchange membrane against the stainless—steel pipe. The device
was filled with a 1:1 mixture of IRA—68 and IRC—84 resins and
operated with the same feed solution. Initially, 90% deminer—
alization was obtained with a flow rate of 60 ml/min, but leaks
developed again. In the next assembly the timer spring was
replaced with a Type—301 stainless-steel spring, and a new
tubular anion—exchange membrane was installed.
Arrangements were made with RAI Research Corporation to
produce Permion 4025 anion—exchange membranes in tubular form
for the cylindrical device. Because of delays in procurement
of suitable Teflon tubing and difficulties in maintaining dimen-
sional stability, a suitable Permion tubular membrane was never
received. A parallel effort to prepare tubular anion—exchange
membranes was carried out in our laboratory. In another re-
search project we had prepared a durable anion-exchange membrane
by first soaking a fabric in a mixture of polyethyleneimine
and epoxy resin and then curing the material in an oven. To
prepare a tubular anion—exchange membrane to fit the cylin-
drical ERID unit, a 16—mm—diameter knitted polyester tube
(Meadox Medicals) was soaked in a mixture of 3 parts of PEI
18 (polyethyleneimine from Dow Chemical Co.) and 2 parts of
EPON® 828 (diglycidyl ether of bisphenol A from Shell Chemical
Co.) and was slipped over a 16-mm—diameter Teflon rod. The
material was cured overnight in a 60°C oven. Then the membrane
was soaked in water, removed from the rod, and installed over
a 0.58 mm Vexar spacer on the anode of the cylindrical ERID
device.
The cylindrical device with the new membrane and retainer
springs was filled with a 3:2 mixture of IRA—68 and CC and
was evaluated with NaC1 feed solution. The best performance
achieved was 97% demineralization at a product flow rate of
33 mi/mm. Higher flow caused considerable reduction in the
degree of demineralization. The presence of bubbles of hydrogen
gas in the product water revealed that intercompartmental leak-
age still existed. The volume of resin in the cylindrical
device was about 800 ml compared to about 280 ml in one 6.35—
mm—thick resin bed in the plate—and—frame device, and one corn-
partment in the plate—and—frame device could demineralize the
water at the rate of 60 mi/mm compared to 33 mi/mm for the
40
-------
cylindrical device. It became apparent that the plate-and-
frame configuration was superior to the cylindrical configura-
tion, and further developments of the cylindrical device were
•abandoned.
The fiber ERID device that was evaluated initially had
been assembled a year before and stored dry. The ion—exchange
fibers were cotton crochet yarn coated with a crosslinked
polyelectrolyte: PEI 18 cured with dibromopropane for the anion—
exchange fibers and Acrysol® A5 (polyacrylic acid from Rohm
and Haas Co.) cured with hexanediamine for the cation—exchange
fibers. With this device we obtained 90% demineraljzatj.on
of NaCl solution with a flow rate of 11 mi/mm.
The dimensions of the experimental ERID devices and operat-
ing data from the two best experiments with each are presented
in Table 8. The best results were obtained with the plate—
and—frame device that contained four 6.35—mm—thick resin beds
with a 2:1 mixture of IRA—68 and CC. The throughput rate of
289 mi/mm was almost half of the design flow rate for the
prototype device, and the degree of demineralization was well
above our objective of 90%. The residence time was shortest
in this device; therefore, the extent of removal of organic
solutes should be lowest in the plate—and—frame device.
DESIGN OF FULL-SIZE PROTOTYPE ERID DEVICE
The results of our studies with the three experimental
ERID devices indicated that the plate—and—frame configuration
should be used for the full—size prototype. Moreover, since
the dimensions of the resin compartments that had been used
in the experimental plate—and—frame ERID device appeared to be
suitable and convenient, those dimensions were chosen for the
prototype.
Teflon was selected as the material of construction for
the resin compartments because it is inert to attack by most
solvents, it is a very durable material, and it is easy to
machine. A one—dimensional view of the resin compartment is
shown in Figure 5. A 610 x 127 x 6.35—mm (24 x 5 x 0.25—in.)
piece of Teflon is used for each compartment. A 508 x 89—mm
opening is cut in the center, triangular portions are cut from
the corners to accommodate the clamping bolts, and several
holes are drilled in each end to allow passage of solutions
through the compartments. The 12.7—mm hole in each end is con-
nected to the opening in the center by 6.35—mm holes drilled in
the plane of the Teflon sheet. This manifold system allows
passage of water into and out of the resin beds. The two 6.35—mm
holes in each end allow access to the gasket spacers that form
the enriching compartments.
41
-------
TABLE 8.
COMPARATIVE DATA FROM ERID DEVICES
WITH NaC1 FEED SOLUTION
Plate—and frame Cylindrical Fiber
4 1 1
1
1 1
6.35
38
17.5
17.5
51 51
483
483
510
510
105 105
89
89
444 a
444
105 105
IRA—68
IRA—68
IRA—68
IRA—68
PEI & EPON
cc
cc
cc
cc
Acrysol A—5
2:1
3:2
3:2
3:2
2:1 2:1
400
600
280 b
280 b
380 380
Resin domp ±tments
Nuinbe r
Thickness, mm
Length, mm
Width, mm
Anion-exchange resin
Cation-exchange resin
Resin ratio, A:C
Void volume, ml
Operating data
Product flow, mi/mm
Waste flow, mi/mm
Feed cond., i.iS/cm
Prod cond., pS/cm
Applied potential, V
Current, mA
Calculated results
Demineralization, %
Coulomb efficiencyC,
Residence time, mm
Current flux, mA/cm 2
29
15
15
12
10
145
162
160
145
155
16
4
14
2
12
44
22
20
16
22
545
200
215
41
93
289
29
150
3
40
230
98
62
1.38
0.52
89
83
2.3
1.24
97
35
8.5
045 a
91
42
6.1
0.48
99
89
20.0
92
98
7.6
2.42
a. Based on log-mean diameter of electrodes
b. Assuming 35% void volume in 800 ml of resin beads
C. Coulomb efficiency = 0.0134 x (flow x conductivity ) roduct
nunther of resin beds x current
d. Based on area of exposed fiber ends
42
-------
Figure 5. Design of Teflon resin compartment
43
VIEW A-A
24 IN.
-------
A piece of 250—mesh Teflon screen was mounted on each
end to prevent loss of resin beads from the compartment. A
semicircular slot was cut in the inside wall of the compartment
6.35—mm from the end of each side strip, and a 100 x 38—mm
piece of Teflon screen was inserted. The screen was attached
to the surface of the Teflon sheet with a small amount of Perma—
bond® 102 (Pearl Chemical Company, Tokyo, Japan) contact cement.
The adhesive bond remained intact after overnight soaking in
water.
For the gasket spacers for the enriching compartments,
a piece of Vexar® 10 PDS 169 polyethylene netting was placed
on a sheet of polyethylene film on a flat surface, and RTV
silicone rubber (Silicone Calk and Sealer No. 8641 from Dow
Corning Corp.) was applied around the perimeter in the pattern
shown in Figure 6. Another sheet of polyethylene film and
a flat plate were placed on top, and the entire sandwich was
pressed together while the silicone rubber cured. The poly-
ethylene film was peeled off, and the gasket spacer was cut
to size. A thin layer of silicone rubber was applied to the
surface of the Teflon before the ion—exchange membranes were
put in place to improve the seal between the membranes and
the Teflon resin compartments.
The end plates for the prototype device were fabricated
from 610 x 178 x 19—mm (24 x 7 x 0.75—in.) pieces of Grade—
221 Micarta® (fabric—reinforced phenolic by Westinghouse).
Holes corresponding to those shown in Figure 5 were drilled
through the Micarta to provide access to the solution manifolds.
Threaded rods welded to the back side of each 495 x 89 x 1—
mm electrode passed through holes in the Micarta. The method
of attachment is illustrated in Figure 2. A 15—mm thick rubber
gasket was placed around the perimeter of the electrode. The
end plates with the multiple resin beds, membranes, and gasket
spacers were clamped together with bolts that passed through
holes around the perimeter of the Micarta end plates.
The necessary components of the power supply for the ERID
device are an isolation transformer to protect the operator
from electrical shock, a rectifier bridge to convert the AC
output of the transformer to a DC input to the electrodes of
the ERID device, and a light bulb which serves as a current
controller. The electrical resistance of the tungsten filament
of the bulb is several times as great when it is operated at
its rated power as it is when it is cold. Since it is in series
with the primary winding of the transformer, the light bulb
protects the power supply because it limits the amount of cur-
rent that can flow in that circuit. Therefore, even if a short
develops in the secondary circuit, only enough current would
flow to make the bulb glow brightly. During normal operation
the bulb glows faintly. The current can be changed merely
by changing the wattage of the light bulb. An autotransformer
may also be used if precise control of current is desired.
44
-------
Figure 6. Gasket spacer for waste compartments
45
-------
The prototype ERID device was designed to utilize tap
water as feed to all of the solution compartments, but it is
necessary to acidify the feed to all of the waste compartments
except the anode compartment. A proportioning system was devised
whereby a small amount of acid solution could be infused contin-
uously into the tap water to feed the waste streams. A Master—
flex® variable—speed per istaltic pump (No. 7545 from Cole Parmer)
with six heads driven by a single motor, was used to pump the
solutions in the proper proportions. The flow scheme and capaci-
ties of the pump heads are shown in Figure 7. Silicone tubing
(Cole Parmer) was used in all of the Masterfiex pump heads.
The proportioning system in Figure 7 has a product—to—waste
flow ratio of 5.83:1. A ratio of 7.9:1 could be achieved with
the largest Masterflex head that pumps 3.8 mi/rev.
OPERATION AND PERFORMANCE OF PROTOTYPE ERID DEVICE
A mixture of 2 parts of IRA—68 to 1 part of C—433 was
used in our initial evaluation of the new ERID device. Batches
of the resins were treated separately by the 7—step procedures
described in Section Iv of this report (i.e., repeated cycling
through acid and base followed by extraction with methanol,
diethyl ether, and acetonitrile) . Then the resins were mixed
and placed in the resin compartments as the device was assembled.
The device contained 8 cells comprising an lonac MA-3475R anion—
exchange membrane, a Teflon resin compartment, an lonac MC—
3470 cation—exchange membrane, and a gasket spacer.
The feed solution was deionized tap water with NaCi added
to adjust the conductivity to about 150 pS/cm. During the
first week we adjusted the flow rate and applied voltage to
various levels to determine the capabilities of the device.
Then we set the autotransforrner in the power supply at a con-
stant level of 100 volts and attempted to maintain a constant
flow rate through the solution over the range that would be
expected in tap water. Finally, we set the feed conductivity
and flow rates at constant levels and varied the applied volt-
age. The operating data are shown in Table 9, and the results
are discussed below.
After the first two days of operation in the first series
of experiments, the degree of demineraljzatjon was always greater
than 90%, and the coulomb efficiency remained above 50%. The
pH of the product water (not shown in Table 9) remained in
the range of 6.2 to 7.4 throughout this series of tests.
For the constant—voltage experiments, the conductivity
of the feed was increased each morning after the data were
recorded to allow a full 24 hr for each experiment. However,
the closeness of the morning data to those taken the afternoon
before indicated that steady state was achieved in a matter
of a few hours. The data for these experiments are presented
46
-------
MOTOR
FEED TO
RESIN BEDS
Figure 7.
Flow scheme for feedwater proportioning system
0.63 N HCL
CATHOLYTE
WASTE
COMPARTMENTS (0.36)
ANOLYTE
CAPACITIES (mi/rev) FOR
MASTERFLEX® PUMP HEADS
TAP WATER
47
-------
TABLE 9. PERFORMANCE OF ERID DEVICE WITH 8 TEFLON RESIN COMPARTMENTS
Flow
rate,
mi/mm,
of
Conductivity, imho/cm,
of
Voltage,
Current,
Coulomba
Date Product Anolyte
Catholyte Waste Feed
Product Electrode
Waste
V
mA
efficiency, %
8—26 330 21 11 49 153 25 750 2250 26 110 64
8—27 95 8 7 26 150 5 1100 3300 46 100 23
8—29 225 5 9 30 165 1 700 1370 47 98 62
8—30 240 3 7 25 165 2 700 1500 47 100 66
8—31 585 7 17 46 145 3 700 1620 35 150 93
9—1 615 5 11 22 150 5 850 3700 51 190 82
9—2 580 6 13 38 150 3 700 1940 81 220 67
9—3 600 3 13 39 150 3 950 1940 94 290 51
9—3 575 3 13 40 150 3 950 1840 81 270 54
Experiment with constant voltage and increasing feed concentration.
9—9 600 7 11 38 170 4 1700 2000 91 280 60
9—10 590 4 12 39 310 6 4200 3900 87 455 66
9—13 580 11 13 38 410 10 7000 4900 82 530 74
9—14 630 10 6 41 530 20 4800 7000 76 710 76
Experiment with constant feed concentration and increasing voltage.
9—14 607 6 37 170 17 1700 2100 30 185 84
9—14 608 40 170 9 1800 2300 40 221 74
9—15 603 - .t8 6 40 170 7 1650 2200 50 231 71
9—15 600 9 6 40 170 6 1800 2350 60 260 64
9—15 600 9 7 40 170 5 1900 2450 70 285 58
9—15 598 9 7 40 170 5 2000 2480 80 315 52
9—15 593 9 6 39 170 5 2200 2550 90 336 49
9—15 600 9 7 39 170 4 2100 2550 100 355 47
aClb efficiency = — —
-------
graphically in Figure 8. Ar the conductivity of the feed was
increased, the flow of electrical current increased almost
proportionately so that 96 to 98% of the salt was removed from
the water. The coulomb efficiency improved with increasing
feed concentration. Although a constant input voltage of 100
V was maintained, internal resistance of the power supply
caused the voltage measured across the electrodes to decrease
with increasing current.
For the constant—feed experiments, a 1514—liter (400—gal)
batch of feed solution was made up with deionized water and
NaC1. Flow rates were set at the beginning of the experiment,
and no adjustments were made. The applied voltage (measured
across the e1ectrodec ) w s adjusted in 10—V increments from
30 to .1.00 V. After c i .i..ucrease the system was allowed to
reach steady state be or:e data were recorded. The results
are shown graphically in Figure 9. Operation at 30 V removed
about 90% of the NaC1 fro ii the feed solution. The degree of
demineralization was further increased to 97% by application
of higher vitage, but this .tinprovem nt came at the expense
of considerable power. Above 60 V the incremental improvement
was slight. Therefore, 60 V is probably the maximum practical
voltage for this 8 compartment ERID device with NaC1 feed
solution.
In a final experiment of this series (not shown in Table
9) , we reduced the product flow to 300 ml/min and applied 60
V to the electrodes. The conductivity of the product dropped
to 2.4 pS/cm (i.e., 98.6% removal of salt).
In our first experiments with tap water as feed to the
prototype device, we had not yet begun to use the proportioning
system shown in Figure 7. Instead we used two separate Master—
flex pumps that allowed independent control of the flow rates
of the product and waste streams, and we experimented with
the flow rates and concentration of acid in the waste streams
to find workable operating conditions. The results of these
experiments are shown in Table 10. In the first series of
experiments the feed to the waste streams was 0.032 N HC1.
During the first three days, there was a rapid buildup of hydrau—
lic and electrical resistance of the device, and the conduc-
tivity of the product water increased to unacceptable levels.
Evidently either calcium and bicarbonate ions were accumulating
in the resins, or calcium carbonate was precipitating in the
waste compartments, or both accumulations were taking place.
The power was turned off, and the resin beds and waste compart-
ments were rinsed in place with 0.01 N HC1. Initially the
performance was improved markedly, but it soon began to deterio-
rate.
49
-------
800 I I I
600 — H
2
w -
I
I
400— -:
0
200 —
>. o—.-o--- —0
_J I
Ou. I
o W 60’ 0
100
> I-.
o 0 0 —
I—
w
I L
20— —
5
> -2
>
10—
0Q
20
01 5—
0—
>
90—
o o 0
ti -
80—
I I I o
100 200 300 400 500 600
FEED CONDUCTIVITY, pmho-cm
Figure 8. Effect of feed conductivity on operating
parameters of ERID
50
-------
I I I
.—.---.-.
.— -
..—.
a
I I
30 40
T — —V —.
50 60 70 80 90 100
APPLIED VOLTAGE, V
E
I-
2
LU
0
>.
D 0
LU
0—
- I c .)
Du.
Ou.
Ow
35U
300
250
200
150
80
70
60
50
40
20
15
10
5
I—i
u .E
0
>
00
00
20
OQ.
Figure 9, Effect of applied voltage on operating parameters of ERID
-------
TABLE 10. PERFORW NCE OF PI TIOTYPE ERID DEVICE WT H TAP WATER AS FEED
Flow rate, mi/mm, of Conductivity, iS/cm, of Voltage, Current, Coulomb
Date Product Waste Feed Product V i s A efficiency, %
9—22 510 75 220 7 79 380 48
9—23 485 72 225 22 74 225 73
9—24 430 73 225 34 73 230 60
Rinsed device with 0.01 N HC1 with power off
9—29 400 61 250 9 71 404 40
9—30 330 46 250 7 74 350 38
10—1 300 36 250 5 78 290 44
10—5 460 31 220 15 80 230 67
10—5 615 33 225 40 82 200 95
Disa8sembled device, rejuvenated resins and replaced damaged membranes
10—3 440 96 215 8 49 270 56
10—14 450 89 205 6 50 250 60
10—15 590 93 200 9 48 294 64
10—18 600 97 200 12 49 290 65
10—19 710 91 190 13 49 242 87
10—20 580 90 185 9 76 310 55
10—21 500 97 190 7 75 315 49
Reversed flow through all compartments and inverted device
10—26 680 78 195 16 74 330 62
1O—27 580 90 220 9 78 272 68
10 ?0 550 93 215 9 78 264 72
490 82 215 11 78 262 64
11 —1 410 79 240 56 0 0 —
11—2 385 82 240 28 76 284 47
11—3 560 87 250 17 76 284 77
11—4 670 68 225 17 73 380 61
11—5 630 75 230 17 74 385 58
Rinsed device in place with 2 N SC]. then 1.5 N NaOR
11—10 195 42 230 13 100 350 20
11—11 580 95 222 8 73 364 57
11—12 640 97 220 27 77 260 80
11—16 650 104 175 30 74 340 46
11—17 580 100 185 23 74 385 41
Rinsed in place with 3 cycle, of acid and base
11—30 460 80 220 4 75 360 47
12—1 660 127 225 8 71 390 62
52
-------
To determine the cause of the unsatisfactory performance,
we disassembled the ERID device and inspected the resin beds
and waste compartments. A brown slime was found in the inlet
side of all compartments, and a white precipitate was found
in the waste compartments, particularly on the surfaces of
the anion-exchange membranes. The precipitate was dissolved
with HC1 solution, and the screens were scrubbed with detergent
and rinsed with formaldehyde solution to remove the slime.
The mixed resins were rinsed sequentially with 2 N HC1, 1.5
N NaOH, 2 N NaC1, methanol, diethyl ether, and acetonitrile.
Damaged membranes were replaced, and the device was reassembled
with two 0.58—mm-thick gasket spacers in each waste compartment.
For the next series of experiments the feed solution to
the waste compartments was 0.095 N HC1, and the feed to the
resin beds and anode compartment was tap water. The deminerali—
zation performance remained satisfactory for about two weeks
of continuous operation near the design flow rates. The ERID
device was shut down each weekend because there was inadequate
storage capacity for feed water. At the end of the second
week there was an increase in hydraulic resistance. To remedy
this apparent accumulation of debris at the entrance to the
resin beds, the device was inverted, and the inlet and outlet
connections were switched. This maneuver restored enough
permeability to the resin bed to allow operation for another
week, but the flow rate gradually dropped off again. When
the device was turned on at the beginning of the next week,
the operator failed to energize the electrodes, and tap water
flowed through the device for 4 hr during which the conduc-
tivity of the product increased to 56 pS/cm as the ion—exchange
capacity of the resins became exhausted. When the electrodes
were energized, ‘the demineralization improved, but not to the
levels achieved the previous week.
This time we attempted to clean the resins in place rather
than disassemble the ERID device. First the device was inverted,
and the flow was reversed. Then 6 liters (approximately 4
bed volumes) of 2 N HC1 were pumped through the resin beds.
There was considerable evolution of gas (presumably carbon
dioxide) at the beginning of the acid rinse, ‘but it subsided
before the full 6 liters had been used. The device was rinsed
with 3 liters of deionized water followed by 3 liters of 1.5
N NaOH and another 3 liters of deionized water. Finally it
was rinsed with 0.1 N HC1 until the effluent became acidic,
and the electrodes were energized. The demineralization per-
formance the next day was improved, but the improvement was
only temporary.
The efforts to rinse the resins in place were repeated,
but this time the resins were contacted with three cycles of
acid—water—base—water rinses to ensure adequate exposure to
the rinse solutions. The multiple rinses proved effective,
53
-------
and the performance improved. Evidently the microbial colony
and the calcium carbonate were removed. Therefore, we con-
cluded that rinsing in place was an effective method of re-
juvenating the ERID device, and we recommend such rinsing when-
ever the demineralization performance or the hydraulic perme-
ability of the ERID device begin to deteriorate. If good per—
formance cannot be restored by rinsing in place, the device
should be disassembled and inspected.
An experiment was conducted to determine the change in
total organic carbon content of water that flowed through the
ERID device. TOC analyses were carried out at the Environ-
mental Research Laboratory in Athens, Georgia. The ERID device
had been operated intermittantly for one month and had processed
over 10,000 liters of tap water. First, 15 liters of distilled
water were allowed to flow through the resin beds, and samples
of the afferent and efferent were taken. The device filled
with distilled water was allowed to sit for 30 mm; then 50
ml were drained from the bottom, and a sample was taken. The
device was put back in operation with tap water as feed for
24 hr. Then samples of feed and product were taken. The result
of the carbon analyses are shown in Table 11. The results
indicate that passage through the ERID device contributed some
organic and inorganic solutes to the distilled water and removed
some from the tap water. The values of organic carbon in the
tap water appeared to be abnormally high. A TOC value of 2.24
mg/liter was reported by the Birmingham Water Works in June,
1976.
TABLE 11. CARBON CONTENT OF WATER BEFORE AND
AFTER TREATMENT WITH PROTOTYPE ERID DEVICE
CONTAINING 2:1 MIXTURE OF IRA-68 AND C-433
Sample
Inorganic
carbon, mg/liter
Organic
carbon, mg/liter
Disti1led water
afferent
0.2
1.7
Distilled water
efferent
5.1
3.1
Distilled water
resins
drained
from
1.1
1.9
Tap water feed
21.0
22.0
Product fro
m tap
water
7.0
17.0
54
-------
SECTION VI
PERFORMANCE OF COMBINED ERID-RO SYSTEM
The system for recovering organic solutes in a concentrated
form from drinking water will utilize ERID as a pretreatment
to remove inorganic salts from the water before the organics
are concentrated by reverse osmosis (RO). This two—stage opera-
tion can be carried out as a one step operation with the pro-
duct from the ERID Device going directly to the RO unit, or
it can be carried out as two distinct steps. We chose the
two—step method because we had adequate storage capacity. More-
over, the two-step method eliminated problems of precisely
matching the throughput capacities of the two processes.
The laboratory equipment included two 1893—liter (500—
gal) tanks (one for the feed solution and the other for the
product from the ERID device), a 114—liter (30-gal) tank to
supply solution to the waste compartments of the ERID device,
a Masterfiex pump to supply solutions to the ERID device in
the proper proportions, the plate—and—frame ERID device, and
the RO unit (Continental Model 3011) . In the first experiment
the feed solution was prepared by the addition of NaCl to
deionized water, but in all subsequent experiments the feed
solution was Birmingham tap water. The ERID device was oper-
ated with constant flow and applied voltage until steady state
was achieved. Then a radiolabeled test organic solute was
added to the solution in the feed tank. Samples of the feed
and product were taken periodically for analysis by scintilla-
tion counting. The flow rates, voltage, current, and solution
conductivities were monitored and recorded.
When all of the feed solution had passed through the ERID
device into the product collection tank, the RO treatment was
begun. The feed tank became the permeate collection tank,
and the tank containing the deionized product became the reject
recirculation tank. Samples of permeate and reject were col—
lected for scintillation counting, and the conductivities of
both steams were monitored and recorded. RO treatment was
continued until the volume of the reject reached its lowest
manageable level.
The first combined ERID—RO experiment was conducted while
the experimental ERID devices were being evaluated. The ERID
device contained four 6.35—mm thick resin beds filled with
55
-------
a 2:1 mixture of IRA—68 and CC resins and bounded by lonac
MC—3470 and MA-3475R membranes. The NaC1 feed solution con-
tained 2.15 pg/liter of ‘ C—1abe1ed dimethylnitrosamine (DMNA).
The ERID treatment reduced the conductivity of the solution
from 150 pS/cm to less than 10 pS/cm, and only 1.4% of the
DMNA was lost. The data from the RO experiment are shown graphi-
cally in Figure 10. The rejection of DMNA by the Permasep®
hollow fibers was low. The average ratio of reject to permeate
concentration of DMNA was 1.9, which corresponded to a solute
rejection of only 35.7%. Greater than 98% rejection of NaC1
was achieved. The conductivity of the solution increased more
than 100 fold, while the DMNA content increased only 6 fold
as the volume was reduced from 720 liters to about 4 liters.
Therefore, substantial recovery of DMNA from drinking water
by treatment with the Permasep RO unit did not appear to be
feasible.
The final prototype ERID device was used in the second
combined ERID—RO experiment. The eight Teflon compartments
contained a 2:1 mixture of IRA-68 and C—433 resins and were
bounded by lonac MC-3470 and MA—3475R membranes. The feed
solution contained 2.3 pg of ‘ C—labeled reduced Michier’s
ketone per liter of filtered Birmingham tap water. The effluent
from the ERID device retained 12% of its initial conductivity
and 36% of its radioactivity. Only 6% of the radioactivity
was removed in the waste streams; the remainder was retained
by the resins. When the resins were rinsed with two cycles
of HC1 and NaOH solutions, 21% of the retained radioactivity
was recovered.
The 1022—liter (270—gal) batch of product from the ERID
treatment was concentrated to 151 liters (40 gal) by RO. The
final ratio of reject to permeate concentration of the radio—
labeled material was 21.5, and the rejection was 79%. Reduced
Michier’s ketone was effectively concentrated by nO, but the
two amine groups on the molecule had a strong affinity for
the resins in the ERID device.
The final combined ERID—RO experiment was conducted with
8 ng of 3 H—labeled cholic acid per liter of filtered tap water.
The eight resin beds in the ERID device contained fresh resins
and membranes. The ERID treatment reduced the conductivity
of the water from 225 pS/cm to 7 pS/cm. The retention of choljc
acid in the ERID device was 98% after 1 hr, 96% after 6 hr,
and 80% after 18 hr. Overall, 84% was retained by the resins,
and less than 0.1% was in the waste streams. When the device
was subsequently rinsed with one cycle of 2 N HC1 and 1.5 N
NaOH, 54% of the retained cholic acid was recovered.
56
-------
280
I I I I I I I I I I I
LEGEND
A CONCENTRATION iN REJECT
0 CONCENTRATION IN PERMEATE
OCOND(JCTIVITV OF REJECT
—
A
-FEED IN
TAN K
0 —o
— I— I I I I I
1 _ I I I I I I
— COMPOSITE
A REJECT
COMPOSITE PERMEATE
— 240
— 200
1 2 3 4 5 6 7 8 9 10 11 12 13
TIME, hour
Figure 10. Reverse osmosis treatment of a solution containing 2.05 jig/liter
of dimethylnitrosamine
12 —
10 —
2
0
I-
I-
2
uJ
C.,
2
0
C.) 4
2
E
U
U)
160 -
I .-
>
I—
120
0
2
0
C.)
— 80
40
-------
When the product from the ERID device was treated by RO,
96% of the cholic acid was rejected, and the final ratio of
reject to permeate concentration was 155. Cholic acid was
easily concentrated by RO, but its carboxylic acid group has
a strong affinity for the anion—exchange resins in the ERID
device. It is likely that a larger fraction of the cholic
acid could be recovered from the ERID device by more extensive
rinsing.
58
-------
REFERENCES
1. Walters, W.R., D.W. Weiser, and L.J. Marek. Concentration of
Radioactive Aqueous Wastes. md. Eng. Chem . 47, 61 (1955).
2. Glueckauf, E. Electro—Deionisation through a Packed Bed.
Brit. Chem. Eng . 4, 646 (1959).
3. Sammon, D.C., and R.E. Watts. An Experimental Study of Elec-
trodeionisation and its Application to the Treatment of
Radioactive Wastes. United Kingdom Atomic Energy Authority
Report AERE-R 3137 (l9 6O).
4. Prober, R., and C.E. Myers. Electrolytic Regeneration of
Ion—Exchange Resins, Final Report to the Office of Saline
Water on Contract 14—01—0001—1255 (1968)
5. Davis, T.A., and R.E. Lacey. Electro—Regeneration of Ion-
Exchange Resins, Final Report to Artificial Kidney—Chronic
Uremia Program of the National Institute of Arthritis and
Metabolic Diseases (1972)
6. Mantoura, R.F.C., and J.P. Riley. The Analytical Concentra-
tion of Humic Substances from Natural Water. Anal. Chim.
Acta 76, 97 (1975) .
7. Van Beneden, G., and P. Van Beneden. Chein. Abstracts 72,
136224V (1970) .
8. Bohnsack, G. Experiments on the Behavior of Anion—Exchange
Resins toward Humic Acids. Mitt. VGB . 73, 53 (1962).
59
-------
GLOSSARY
anode: A platinized—titanium plate attached to the positive
lead of the electrical power supply. Anions (e.g., chloride,
sulfate, bicarbonate, and hydroxyl ions) migrate toward the
anode.
anolyte: The rinse solution in the anode compartment. It con-
tains the hydrogen ions and the oxygen and chlorine gas
generated at the anode as well as the anions that enter
from the adjacent resin bed.
cathode: A stainless—steel plate attached to the negative
lead of the electrical power supply. Cations (e.g., sodium,
calcium, and hydrogen ions) migrate toward the cathode.
catholyte: The rinse solution in the cathode compartment. It
contains the hydroxyl ions and the hydrogen gas generated
at the cathode as well as the cations that enter from the
adjacent resin bed.
conductance: Reciprocal of resistance. Sieman = ohm’
conductivity: The conductance of a specimen 1 cm in length and
1 cm 2 in cross section expressed as pS/cm.
coulomb efficiency: The fraction of the electric current that is
utilized in removal of salt from the resin bed; specifically,
the equivalents of salt removed per faraday (96,494 A sec)
of current flowing through the resin bed.
demineralization performance: The effectiveness with which the
ERID device removes salts from the feed solution. The
degree of demineralization was measured by the percentage
change in the conductivity of the solution treated by the
ERID device.
electrodialysis: A deionization process in which solutions of
electrolytes flow through multiple compartments bounded
alternately by anion— and cation—exchange membranes.
Direct electric current transports ions through the mem-
branes and causes enrichment and depletion of solutions
in alternate compartments.
60
-------
equiconductance point: The unique value of conductivity of an
ion—exchange resin that is equal to the conductivity of the
electrolyte solution in which it is equilibrated.
ERID: Electro-Regenerated Ion—exchange Deionization, a process
by which mixed beds of ion—exchange resins are continuously
or cyclicly regenerated in the depleting compartments of an
electrodialysis apparatus.
exhaust: To convert a resin from its regenerated form to another
ionic form (e.g., chloride form for anion—exchange resins
and sodium form for cation—exchange resins).
feed: The solution that enters a solution compartment of the
ERID device.
permeate: The solution that passes through the walls of the
hollow fibers in the RO unit.
regenerate: To convert anion-exchange resins to the hydroxyl or
free base form and cation-exchange resins to the hydrogen
form.
reject: The solution that does not pass through the walls of the
hollow fibers in the RO unit.
rejection: The ability of an RO membrane to allow passage of
some molecules, such as water, while preventing the passage
of other molecules, such as organic solutes.
rejuvenate: To renew the demineralization performance of an
ERID device by rinsing away unwanted materials.
reverse osmosis, (RO) : A process in which pressure is applied
to force solvent out of a concentrated solution, through a
semipermeable membrane, into a more dilute solution.
61
-------
APPENDIX A
RECOVERY OF BIS(2-CHLOROETHYL) ETHER BY SPARGING
Our first procedure for analysis of bis(2—chloroethyl)
ether (BCE) required extraction from the aqueous solution with
diethyl ether. In this procedure the volatile BCE (b.p. 178°C)
was subject to loss by evaporation. A study was carried out
to determine the feasibility of recovering BCE from large quanti-
ties of dilute aqueous solution by sparging with nitrogen.
The all—glass experimental apparatus included a charcoal filter
to adsorb foreign organic material from the commercial bottled
nitrogen, a sparge tube in a 500—mi flask, an ice—cooled con-
denser to remove moisture from the discharge gas, and two or
three collection tubes containing 0.2 g of solid sorbent. In
a typical experiment, a 500—mi batch of solution was sparged
with nitrogen flowing at 500 mi/mm for at least one hour.
Then the contents of each collection tube, including the glass—
wool plugs, were placed in a vial, and 2 ml of ethanol were
added. Samples of the ethanol extracts of the individual tubes
were analyzed by GC.
CC conditions for BCE analysis:
Column: 2 m by 0.4—cm—i.d. glass
10% Apieson L on Gas Chrom Q
110°C
Detector: Flame ionizatojn
200°C
Carrier: Helium, 35 mi/mm
Injection port; 170°C
Sample size: 5 jil
In our early experiments, we recovered negligible amounts
of BCE by sparging at ambient temperature. When the tempera-
ture was raised to 50°C, a Porapac Q® sorbent picked up 60%
of the BCE in 4 hr and 97% in 8 hr. However, we encountered
contaminants in Porapaq Q arid used Tenax® as the solid sorbent
in subsequent experiments.
62
-------
The data from nine sparging experiments are shown in Table
12. In Experiment 1 the low temperature and short sparging
time were apparently inadequate. In all experiments after
Experiment 1, 30 g of NaC1 were added to the solutions to lower
the volatility of the water and to increase the relative vola-
tility of the ether. A slight improvement was noted in Experi—
ment 2. The higher temperature in subsequent experiments im-
proved recovery considerably. Comparison of Experiments 4
and 5 indicates that sparging for 2 hr is adequate; longer
sparging merely forces more BCE into the second column. A
third Tenax tube in Experiment 5 contained no BCE. A shorter
sparging period in Experiments 6 and 7 again proved inadequate.
For Experiments 8 and 9 the BCE content of the solution was
reduced to 0.1 mg/i. The results of those experiments were
only semiquantitative because the low concentrations of BCE
in the ethanol extract approached the threshold level of detec-
tion by the gas chromatograph.
The data in Table 12 indicate that sparging is an efficient
method for recovering BCE from dilute solutions. Further im-
provements might be achieved by an increase in the volume of
the solution, optimization of the flow rate of nitrogen, and
a change of the design and content of the sorbent tubes.
At this stage of its development, the sparging technique
is only an analytical tool. However, a well—designed, continuous,
countercurrent sparging system might be useful for recovering
the volatile organics from large quantities of drinking water.
Moreover, the sparging process might be compatible with the
system we are now investigating. Sparging might be used to
remove a large fraction of the volatile organics before the
water is demineralized or to remove the residual organics from
the permeate of the reverse osmosis unit. However, investiga-
tion of the feasibility of continuous sparging was beyond the
scope of this contract.
63
-------
- TABLE 12. RECOVERY OF BIS(2-CHLOROETHYL) ETHER BY SPARGING
Experi- Concentra—
ment tion, Sparging Temperature, Recovery, %
number mg/liter time, hr °C total second tube
1 1.0 1 50 28 0
2 1.0 1 50 30 0
3 1.0 2 70 92 20
4 1.0 3.5 70 102 60
5 1.0 2 72 102 5
6 1.0 1.5 70 60 0
7 1.0 1.5 75 60 0
8 0.1 2 68 87 0
9 0.1 2 75 80 0
64
-------
APPENDIX B
REVERSE OSMOSIS TREATMENT OF WATER
CONTAINING TEST ORGANIC SOLUTES
In the first experiment with the combined ERID—RO system
(Section VI) we found that dimethylnitrosamine was not rejected
by the RO membrane and was, therefore, not a suitable test
organic solute for our studies. Before other organic solutes
were used in the full—scale experiments with the combined ERID—
RO system, they were evaluated in small—scale RO experiments
to determine the degree to which they would be rejected by
the RO membrane. For these experiments, the RO unit was thor-
oughly rinsed with at least 75 liters of deionized water, and
the reject rate was adjusted to 50%. Then a 38—liter batch
of deionized water spiked with a radiolabeled test organic
solute was treated by RO, and samples of the feed, permeate,
and reject were taken for scintillation counting.
In the first experiment, a solution of ‘ C—1abeled 2,4,6
trimethylaniline (TMA) was treated by RO. The radioactivity
of the effluent samples indicated that 2% of the TMA was in
the permeate and 98% was in the reject. However, a material
balance showed that 42% of the TMA had remained in the RO unit.
The permeate and reject from the first test were mixed and
treated again, the proportion of TMA in the permeate and reject
were again 2:98, but 38% of the total radioactivity was lost.
The permeate and reject were mixed again for a third test.
This time reject was recirculated into the feed tank. After
19 liters of permeate had been collected, the proportions of
TMA in the permeate and reject were 3:97, and 39% had been
lost. When this experiment was repeated another 34% was lost.
A solution containing ‘ C—labeled reduced Michier’s ketone
was treated by RO. The reject was recirculated to the feed
tank, and the permeate was collected in another tank. Samples
were taken initially and after half of the solution had permeated
the RO membrane. In the first experiment the permeate con-
tained 0.8% of the radioactivity, and 54% was lost. The permeate
was poured back into the feed tank, and the experiment was
repeated. This time the permeate contained 1.1% of the radio-
activity, and 44% was lost.
65
-------
In a subsequent experiment with a fresh batch of solution
containing 1 C—1abe1ed reduced Michler’s ketone, the RO unit
was operated with a reject rate of 50%. Negligible radioactivity
was detected in the permeate, and 63% was lost in the RO unit.
After half of the solution had permeated the RO membranes,
the reject pressure valve was opened fully to allow complete
recirculation with no permeate for a period of 10 mm. During
this period, 42% of the remaining radioactivity was lost.
The permeate and reject were mixed, and the RO experiment was
repeated with a reject rate of 75%. This time 0.3% of the
radioactivity appeared in the reject, 52% was lost during the
RO experiment, and 20% of what remained was lost during the
10—mm flush.
When a solution containing 3 H—labeled cholic acid was
treated by RU, 1.1% of the radioactivity was found in the per-
meate, and the remainder was in the reject. There was no mea-
surable loss of cholic acid in the RO unit.
The results of these experiments indicate that the Permasep
hollow fibers in the RO unit have a considerable affinity for
amines. The radiolabeled solutes were apparently absorbed
by the polyamide fibers. Since the fraction of the radioactivity
lost appeared to drop off in repeated experiments, a saturation
point may be reached if large volumes of water were treated.
However, the amount of data we collected is sufficient only
to point out the potential problem of absorption of organic
solutes by the RU unit.
The solutes of higher molecular weight; 2,4,6 trimethyl—
aniline (M.w. 135), reduced Michier’s ketone (M.W. 254), and
cholic acid (M.W. 409) ; could be effectively concentrated by
RU because they did not permeate the RO membrane to any great
extent. On the other hand dimethylnitrosamine (M.W. 74) per-
meated the RO membrane to such an extent that it could not
be concentrated by RO. Therefore, some other technique, such
as the sparging technique described in Appendix A, will be
needed to concentrate solutes of low molecular weight.
66
-------
APPENDIX C
PROCEDURES FOR OPERATION AND MAINTENANCE
OF THE PROTOTYPE ERID DEVICE
The end plates, electrodes, resin compartments, and gasket
spacers are permanent components of the ERID device that should
require only gentle cleaning when the device is disassembled.
The resin beds and membranes will require occasional cleaning
or replacement. The symptoms of malfunction and the recom-
mended remedial action are presented below.
The resin beds in the ERID device are effective filters
for particulate matter, and any material that they trap will
reduce the hydraulic permeability of the resin beds. There-
fore, it is recommended that filter with 5—urn pore size be
installed in the feed—water supply line to prevent entry of
particulate matter. Filtration of feed water will reduce,
but not completely eliminate, the need for cleaning the ERID
device. Before each experiment, ensure that the filter is
clean, pull a new section of tubing into each Masterfiex pump
head, and prepare a 40—liter batch of 0.63 N HC1 solution (52.6
ml of concentrated HC1 per liter) for the waste streams. Then
energize the feed pump and measure the flow rates and pH of
each effluent stream. The pH of the product and anolyte should
be near 7, and the catholyte and waste should be about 2.
The flow rates should be proportional to the values shown in
Figure 7. If they are not, check the flow rates from the indi-
vidual pump heads to see that they are functioning properly.
If restrictions in flow appear to be in the ERID device, invert
the device, switch inlet and outlet lines, and operate the
device with reverse flow. If proper flow cannot be achieved
disassemble and inspect the device.
After the initial check shows the proper values of pH
and flow rate, energize the electrodes and recheck the pH of
the anolyte and catholyte to ensure that both are acidic.
If the anolyte is basic, the polarity of the electrode con-
nections is reversed, and severe damage to both electrodes
will occur. Measure the conductivity of the product water
to ensure adequate demineralization. If the ERID device has
been inverted or operated with electrodes deenergized for an
extended period, several hours of operation may be required
before 90% demjneralizatjon is achieved. If demineralizatiori
67
-------
is marginal, it can be improved somewhat by increasing the
wattage of the current-control light bulb. (Do not exceed
150 w.)
Shutdown of the ERID device by the procedure described
below will facilitate startup in the next experiment. First,
deenergize the electrodes and allow all solutions to continue
flowing for 10 mm. Then invert the device, switch the inlet
and outlet lines, and pump solutions through for 10 mm with
no power to the electrodes and with tap water flowing through
the acid—feed pump. Invert the device, switch the inlet and
outlet lines, start the pump to ensure that air has been dis-
placed from the solution compartments and connecting lines,
then let the device stand with all compartments filled with
solution.
Minor accumulations of calcium carbonate in the ERID device
can be removed by rinsing in place with acid, and accumulations
of humic acid can be partially removed by rinsing with base.
Therefore, cycling the resins between acid and base is useful
for rejuvenating the resins in place. Prepare 14 liters of
2 N HC1 and 12 liters of 1.5 N NaOH. With the electrodes de—
energized, rinse the resin be successively with 4-liter batches
of HC1, tap water, NaOH, and tap water. (The acid and base
should flow upward through the device, and the tap water should
flow downward.) Repeat this cycle three times. Rinse with
HC1 until the effluent from the resin beds becomes acidic.
Rinse the excess acid from the device, energize the electrodes,
and allow the device to regenerate with normal flow rate to
all solution compartments until the pH and conductivity of
the product reach the desired range.
If the performance of the ERID device cannot be restored
by rinsing in place, there is probably a problem of poor flow
distribution that can only be solved by disassembly and inspec-
tion of the internal components. To disassemble the device,
disconnect the electrical leads and the solution lines, lay
the device on its side with blocks to support the end plate,
remove the bolts, and raise the top end plate to expose the
electrode compartment. Leaving the membranes attached to the
Teflon resin compartments, remove each resin bed as a unit,
and inspect all waste-solution compartments and membranes.
If precipitate is found, dissolve it by rinsing with 0.2 N
HC1. If the problem can be identified and solved without dis-
turbing the seal between the membranes and the Teflon compart-
ments, reassembly will be much easier. If the problem appears
to be in the resin beds, peel back the anion—exchange membrane
on each end and inspect the resins. The presence of slime
or discoloration of the resins indicates the need to recycle
the resins by the 7—step procedure described in Section IV
of this report. If new resins are needed, the anion- and cation—
68
-------
exchange resins should be pretreated separately by the 7—step
procedure. Then completely drain the water from the resins,
mix them thoroughly, and pack them tightly in the Teflon corn—
par tments.
Membranes that have precipitates of calcium carbonate
on their surfaces can be cleaned by rinsing with 0.1 N HC1.
Slime on membranes and spacers can be removed by gentle brush-
ing with a Clorox® solution. If membranes appear damaged,
they should be replaced. The anion—exchange membrane adjacent
to the anode is likely to need frequent replacement because
it is exposed to oxidizing conditions. lonac membranes, which
are shipped dry, require pretreatment before they are installed
in the ERID device. Cation—exchange membranes should be soaked
in tap water for 30 mm at room temperature, then in hot tap
water (90 to 100 C) for an additional 30 Thin. Anion—exchange
membranes should be soaked in tap water for 30 mm and in 0.2
N NaC1 solution at 75 to 90 C for 45 mm. Then all membranes
should be rinsed and stored in distilled water at room tempera-
ture until they are trimmed to size and holes are punched for
solution manifolds.
Assembly of the ERID device is facilitated if the anion—
exchange membranes are attached to the Teflon resin compartments.
Apply a thin layer of RTV silicone rubber around the entire
perimeter on one surface of the Teflon compartment. Wipe the
surface of the membrane dry, and press it in place over the
bead of adhesive. As the individual half—cells are assembled,
stack them on an end plate; then place the other end plate
on top of the stack, and bolt the end plates together. Then
add some water through the feed connection to prevent the shrink-
age that would occur if the membranes were allowed to dry.
Allow the adhesive to cure overnight, then dismantle and re-
assemble the device.
To assemble the ERID device, place the anode end plate
in a horizontal position with the electrode facing upward.
Place two electrode—gasket spacers (the ones without the entry
ports) on the end plate, place a resin compartment on top,
and fill the compartment completely level with the mixture
of resin beads. Wipe the surface of the Teflon to remove moisture
and resins, apply a bead of RTV adhesive around the perimeter,
and, after wiping its surface dry, press a cation—exchange
membrane in place. Then place two gasket spacers (see Figure
6) over the cation—exchange membrane, and add all the other
resin beds, membranes, and spacers in similar fashion being
careful that the stack is vertical and properly aligned. After
the last cation—exchange membrane is in place, insert two elec-
trode spacers, place the cathode end plate on top, and bolt
the end plates together. Connect the solution lines, and place
the ERID device in service.
69
-------
APPENDIX D
SOURCES OF MATERIALS FOR ERID DEVICES
Membranes
lonac Chemical Company
Division of Sybron Corporation
Birmingham, N.J. 08011
Plastics Department
E.I. du Pont de Nemours and Company
Wilmington, Delaware 19898
Tokuyama Soda Company, Ltd.
do Nissho-Iwai American Corporation
80 Pine Street
New York, N.Y. 10005
PAl Research Corporation
225 Marcus Boulevard
Hauppauge, L.I., N.Y. 11787
Resins
lonac Chemical Company
Division of Sybron Corporation
Birmingham, N.J. 08011
Diamond Shamrock Chemical Company
P.O. Box 829
Redwood City, Cal. 94064
Rohm and Haas Company
Independence Mall West
Philadelphia, Pa. 19105
Bio-Rad Laboratories
32nd and Griffin Avenue
Richmond, Cal. 94804
70
-------
Screens
Kressilk Products, Inc.
420 Saw Mill River Road
P.O. Box 112
Elmsford, N.Y. 10523
E.I. du Pont de Nemours & Co.
“Vexar” Sales
River Road
Buffalo, N.Y. 14207
71
-------
TECHNICAL REPORT DATA
(Please read Insijuctions on the reverse before completing)
1. REPORT NO. !2.
EPA—600/1—77—035
3. RECIPIENT’S ACCESSIO NO.
4. TITLE AND SUBTITLE
5. REPORT DATE
Electro-Regenerated Ion-Exchange Deionization of
Drinking Water
June 1977 Issuing Date
6. PERFbRMING ORGANIZATION CODE
7. AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO.
Thomas A. Davis
SORI-EAS-77.-069
9.
10. PROGRAM ELEMENT NO.
Southern Research Institute
Birmingham, AL 35205
1CC614B
11.CONTRACT/GRANTNO.
Contract No. 68-03-2209
f .SPoNSon NG AGENCY NAME AND ADDRESS
13. TYPE OF REPORT AND PERIOD COVERED
Health Effects Research Laboratory, Cm-OH
Office of Research and Deve1opr nt
U.S. Environrr ntal Protection Agency
Cincinnati.._Ohm__4 2f R
Final
14. SPONSORING AGENCY CODE
E PA/ 600 / 10
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This report presents the development of a device for removal of inorganic —
salts from drinking water to facilitate the subsequent concentration of
organic solutes for bioassay. Prior attempts to concentrate the organic
solutes by reverse osmosis (RO) resulted in precipitation of the inorganic
salts. To prevent this precipitation, the drinking water is pretreated
by Electro-Regenerated Ion-exchange Deionization (ERID). The ERID
device developed for this purpose is essentially an electrodialyzer with
thick depleting compartments packed with a mixture of anion- and cation-
exchange resin beads. The resins provide a conductive medium for
electrical transport of ions out of the demineralized water, through
ion-exchange membranes, and into a concentrated waste stream.
Experiments with combined ERID-RO treatment demonstrated that,
while greater than 90% demineralization was achieved with ERID, uncharged
organic solutes tended to pass through the ERID device with little change
in concentration. They could be concentrated by the RO unit if their
molecular weights were sufficiently high to permit rejection by the RO
membranes. Organic acids and bases tended to be removed from solution
by the ion-exchange resins, but they could be recovered by rinsing the
with -itl1’ n ’ basic solutions.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
C. COSATI Field/Group
Drinking water
Deionization
Reverse Os imosis
13 B
Ion exchanging
Electrodialysis
18. DISTRIBUTION STATEMENT
19. SECURITY CLASS (ThisReport)
21. NO. OF PAGES
Unlimited Distribution
Unclassified
80
20. SECURITY CLASS (This page)
22. PRICE
tinclassifi ed
EPA Form 2220-1 (9.73) 72 GOVERNMENT PRINTING OFFICE: 1977757056/6432 Region No. 5-Il
------- |