United States
Environmental Protection
Agency
Municipal Environmental Research
Laboratory
Cincinnati OH 45268
EPA-600/2-78-052
June 1978
Research and Development
c/EPA
Nitrate Removal
From Water
Supplies by
Ion Exchange
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research arid 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-
vironmental technology,1 Elimination 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 PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-78-052
June 1978
NITRATE REMOVAL FROM WATER SUPPLIES BY ION EXCHANGE
by
Dennis A. Clifford
Walter J. Weber, Jr.
The University of Michigan
Ann Arbor, Michigan 48109
Grant No. R-803898
Project Officer
Thomas J. Sorg
Water Supply Research Division
Municipal Environmental Research Laboratory
Cincinnati, Ohio 45268
MUNICIPAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
This report has been reviewed by the Municipal Environmental
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.
Environmental Protection Agency nor does mention of trade names
or commercial products constitute endorsement or recommendation
for use.
11
-------
FOREWORD
The Environmental Protection Agency was created because of
increasing public and government concern about the dangers of
pollution to the health and welfare of the American people.
Noxious air, foul water, and spoiled land are tragic testimony
to the deterioration of our natural environment. The complexity
of that environment and the interplay between its components re-
quire a concentrated and integrated attack on the problem.
Research and development is that necessary first step in
problem solution and it involves defining the problem, measuring
its impact, and searching for solutions. The Municipal Environ-
mental Research Laboratory develops new and improved technology
and systems for the prevention, treatment, and management of
wastewater and solid and hazardous waste pollutant discharges
from municipal and community sources, for the preservation and
treatment of public drinking water supplies, and to minimize
tne adverse economic, social, health, and aesthetic effects of
pollution. This publication is one of the products of that
research; a most vital communications link between the researcher
and the user community.
Serious and occasionally fatal poisonings in infants have
occurred following the ingestion of water containing concentra-
tions of nitrate. This report presents the results of an
investigation on the removal of nitrate from water supplies by
two-bed (strong-acid, weak-base) ion-exchange treatment systems,
and by single-bed (chloride form) ion-exchange systems. Detailed
information is given on nitrate selectivity, rates and capacities
for nitrate and competing ions, and regeneration requirements
for various commercially available weak-base ion-exchange resins.
Also, an economic comparison is made between the single-bed and
the two-bed ion-exchange systems.
Francis T. Mayo
Director
Municipal Environmental Research
Laboratory
111
-------
ABSTRACT
Single-bed strong-base anion exchange with NaCl regenera-
tion is currently the method of choice for removal of nitrate
from water supplies. In non-arid non-coastal locations/ disposal
of regenerant brine from such a system is a definite problem.
An alternative ion-exchange process comprising a strong-acid
cation exchanger followed by a weak-base anion exchanger with
bypass blending of raw water and regeneration with HN03 and
NH.OH has been proposed. In addition to nitrate reduction, the
process would yield low hardness water and produce a regenerant
easily disposed of as a fertilizer. The process would be opera-
ted to nitrate breakthrough with chromatographic elution of less-
preferred ions. A two-phase study was undertaken; Phase I to
determine the anion resin characteristics associated with high
nitrate selectivity in the presence of sulfate, chloride, and
bicarbonate, and Phase II to establish the column elution be-
havior of these anions as a function of the process variables:
resin type, fluid detention time, and raw water composition.
Thirty-two commercially available anion resins, thirteen
weak-base and nineteen strong-base, with various polymer
matrices, amine functionalities, capacities, degrees of cross-
linking, and pKa's were evaluated for sulfate/nitrate, chloride/
nitrate and bicarbonate/nitrate selectivity in .005 N acid
solution. Binary isotherms,and H2SO., HHCU, and HC1 titration
curves were developed. Average separation factors were deter-
mined and related to resin properties.
IV
-------
The sulfate/nitrate separation factor (o) had an extreme
range of variability (1.7 - 137) with matrix and functionality
being primary determinants. A hypothesis relating distance of
charge separation to selectivity is proposed; when the exchange
sites are incorporated into the resin polymer backbone at
a guaranteed-close distance, the resin is very diralent ion
selective. The effect of functionality was verified; the
sulfate selectivity sequence is polyamines > tertiary > quater-
nary, out it is argued that the effect is due more to size
than to the previously reported basicity.
The nitrate/chloride selectivity (acl) exhibited a much
narrower range of variability (1.85 - 4.33) with matrix and
degree of cross linking (porosity) primarily determining its
magnitude. High nitrate/chloride selectivity is associated
with hydrophobic resins: polystyrene > non-polystyrene and
macroporous > gel for non-polystyrene resins.
Carbonic acid was not significantly taken up under the
experimental conditions, so the predicted, and verified, resin
selectivity sequence is sulfate > nitrate > chloride » bicarbon-
ate.
Statistical techniques were used to develop predictive eq-
C 1VT
uations for a^ and acl as functions of matrix, functionality,
and porosity; five such equations are given.
In Phase II, eleven column runs were made with five resins,
two different nitrate concentrations (14 and 21 ppm) and two bed
depths (31 and 61 cm). Four-component effluent profiles are
given for all the runs and nitrate throughput comparisons are
plotted to illustrate the effects of the variables.
The important factors influencing the process efficiency
v
-------
N
are acl, equivalent fraction of nitrate in the water (XN), and
detention time. Surprisingly, highly sulfate-selective resins
actually increase the relative fraction (yN> of nitrate on the
resin at breakthrough. Some explanations are proposed for this
and a tentative method for calculation of y based on multicom-
ponent chromatography theory is presented.
Even with operation to nitrate breakthrough the overall
chemical efficiency, as meq nitrate removed per meq regenerant,
was low (13%) for the representative artificial groundwaters
tested.
A regeneration cost comparison between the single and two-
bed processes revealed that, with HC1-NH4OH regeneration, the
two-bed costs were triple those of the single-bed NaCl system.
However, the two-bed regenerant was estimated to be land dispos-
able whereas the single bed regenerant was not.
One-percent solutions of the resins equilibrated overnight
were found to contain 3 to 100 ppm organic carbon. If not
eliminated, these extractable organics may cause serious pro-
blems in water supply applications.
This report was submitted in fulfillment of Research Grant
No. R-803898 by The University of Michigan under the sponsorship
of the U.S. Environmental Protection Agency. This report covers
the period July 21, 1975, to December 31, 1976.
VI
-------
CONTENTS
Foreword
Abstract iy
Figures viii
Tables x
Abbreviations and Symbols xii
Acknowledgments xvi
1. Introduction 1
2. Conclusions 3
3. Recommendations 11
4. Theoretical Considerations 13
Nitrate Problem 13
Process Proposed for Study 22
Structure of Ion-Exchange Resins .... 27
Ion Exchange Selectivity Theory 32
Multicomponent Equilibrium Theory ... 41
5. Phase I: Anion Resin Selectivity Study . . 48
Objectives 48
Procedural Outline 48
Visual Interpretation of Isotherms ... .51
Statistical Analysis 55
Phase I Results Summary 106
6. Phase II: Multicomponent Column Studies. . 127
Objectives 127
Procedural Outline ..... 128
Experimental Methods 136
Data Evaluation Methods 140
Visual Interpretation of Profiles . . . 143
Discussion of Column Results 144
Phase II Results Summary 164
References 173
Appendices 184
A. Equilibrium Isotherms 184
B. Titration Curves 218
C. Ion-Exchange Column Runs 230
D. Experimental Apparatus and Procedures . . . 241
E. Calculations and Derivations 257
F. Statistical Results ."" 272
G. Glossary 288
vn
-------
FIGURES
Number Page
1 Proposed Two-Bed Ion-Exchange Process 24
2 Single-Bed Ion-Exchange Process 25
3 Example Isotherm 37
4 Isotherm Areas 37
5 Expected Resin-Phase Concentration Profile 45
6 Chromatographic Enrichment of Ions in a Column .... 46
7 Scatter Plot: In a vs. Nitrogen Position,
All Resins 86
N
8 Scatter Plot: In ap, vs. Nitrogen Position,
All Resins . . . . 87
c
9 Scatter Plot: In aN vs. Size of Functional
Group, All Resins 89
N
10 Scatter Plot: In ou, vs. Rel. Degree of
Xlinking, Non-polys ty 93
N
11 Scatter Plot: In ap, vs. Rel. Degree of
Xlinking, Polysty7 94
12 Composite Isotherms: STY-DVB, Tert.
Amine, WBA 99
13 Composite Isotherms: Phenolic, Polyamine,
WBA 100
14 Composite Isotherms: Epoxy-Amine, Polyamine,
WBA 101
15 Composite Isotherms: Type I, Quat. Amine,
SBA 102
16 Composite Isotherms: Type I, ISO, Quat.
Amine, SBA 103
17 Composite Isotherms, Type II, Quat. Amine,
WBA 104
18 Hysteresis Isotherms 112
19 Variable Total Concentration Isotherms 113
20 Selectivity vs. Matrix, Tertiary Amine
Resins . 115
21 Selectivity vs. Matrix, Polyamine Resins 116
Vlll
-------
FIGURES
Number Page
22 Selectivity vs. Functionality, Polystyrene
Resins 119
23 Selectivity vs. Functionality, Acrylic
Resins 120
24 Schematic: Experimental Column Set-Up 129
C TVT
25 Throughput: Effects of o£ and acl on y"N 148
26 Throughput: Effects of a on yN/ Neutral 149
27 Throughput: Effect of a on yN, Acidic 150
28 Throughput: Effect of Bed Depth 152
29 Throughput: Effect of Nitrate Cone, on yN .... 166
APPENDIX
Al-32 SuIfate/Nitrate and Chloride/Nitrate
Isotherms 184
A33 Isotherm Curve Fitting for Resin No. 3 216
A34 Isotherm Curve Fitting for Resin No. 8 217
Bl-13 HN03, HC1 and H2S04 Titration Curves 218
Cl-11 Column Run Effluent Concentration Profiles .... 230
Dl Glass Ion-Exchange Column Details 251
D2 Bicarbonate Selectivity Apparatus 252
D3 Isotherm Tumbler 253
D4 Bicarbonate Selectivity Apparatus 253
D5 Column Run Flow System 254
D6-7 Plexiglas Ion-Exchange Column Details 255
El Example Isotherm 261
E2 Resin Phase Concentration Profile 266
Fl-7 Scatter Plots with Regression Equations 272
IX
-------
TABLES
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
pKa's of Various Functional Groups
Anion Resins Obtained for Study
Phase I Data-Summary: Anion Resin
Characteristics. .
Sizes of Functional Groups
Meaningful Correlations: Weak -Base Resins . .
Meaningful Correlations: Weak and Strong-Base
Resins
Porosity and Relative Degree of Crosslinking .
c
Effects of Porosity and Type on CL.
ANOVA: Variables Explaining OL,
ANOVA: Variables Explaining a ,
pKa's of Alkylamines in Water.
Effective Ionic Radii in Aqueous Solution. . .
Individual Ionic Activity Coefficients ....
Limiting Equivalent Ionic Conductance
g
Predicted vs. Measures Values CLT
N
N
Predicted vs. Measures Values of a_,
Varialbes Influencing ac
N
Variables Influencing a_,
Na Test Water Composition for Column Run 1 . .
Na Test Water Composition for Column Runs
2-8
Ca-Mg-Fe Test Water Composition for Runs
9-11
Phase II Data Summary Column Performance
Characteristics
Calculated Column Performance of WBA Resins . .
Calculated Chemical Regenerant Costs
Pagi
. . 32
. . 49
. . 56
. . 69
. . 72
. . 73
. . 78
. . 79
80
. . 81
. . 90
. . 95
. . 96
. . 96
. . 107
. . 108
. . 125
126
. . 131
. . 132
. . 132
. . 145
. . 154
. . 157
-------
TABLES
Number Page
25 Comparisons Between Two-Bed and One-Bed
Processes 159
26 Organics Leached From Conditioned Anion
Resins ..... 166
27 Ranking of Resins for Use in Nitrate Removal
Service 172
APPENDIX
Dl US Ion-Exchange Resin Manufacturers 241
D2 Chemical Makeup of Na Test Water 242
D3 Chemical Makeup of Ca-Mg-Fe Test Water 242
Fl Data Set for Statistical Analysis by MIDAS . . . 272
F2 Correlation Matrix: Weak-Base Resins 273
F3 Correlation Matrix: Strong-Base Resins 273
F4 Correlation Matrix: All Resins 274
F5 Correlation Matrix: Polystyrene Resins 275
F6 Correlation Matrix: Non-Polystyrene Resins . . . 275
XI
-------
LIST OF SYMBOLS AND ABBREVIATIONS
SYMBOLS
* Denotes multiplication when used between variables
o
a Debye-Hiickel ion-size parameter
o
A Angstroms
a"c- Activity of ion i in the resin phase, eq/1
ac. Activity of ion i in the liquid phase, eq/1
a,^ Separation factor for ions i and j , dimensionless
c
OL. Sulfate/nitrate separation factor, dimensionless
ou. Nitrate/chloride separation factor, dimensionless
CL Nitrate/bicarbonate separation factor, dimensionless
D
Cn Total, initial liquid phase concentration meq/1
C Liquid phase concentration, meq/1
C/ Initial liquid phase concentration of i, meq/1
*, u
C. Liquid phase concentration of i, meq/1
Y— Activity coefficient of i in resin, dimensionless
y. Activity coefficient of i in liquid phase, dimensionless
e Resin bed void fraction, dimensionless
F The F statistic in analysis of variance, dimensionless
EM Maximum possible chemical efficiency for nitrate removal,
dimen s i onle s s
E_ Overall nitrate removal efficiency, dimensionless
E Regeneration efficiency, dimensionless
,K
xii
-------
SYMBOLS
K Protolysis constant for amines
cl
pKa Negative logarithm of protolysis constant
K^ Selectivity coefficient
K^ Solution phase, corrected selectivity coefficient
Q Resin exchange capacity, meq/ml, meq/gm
R Relative crystal ionic radius dimensionless
p Density of resin gms/ml
q Resin phase concentration, meq/ml, meq/gm
r Correlation coefficient, dimensionless
2
r Coefficient of determination, dimensionless
TEC Total Equivalent Capacity of resin, meq/ml
T Superficial detention time, minutes
T Throughput, eq. solution/eq. exchanger
x. Equivalent fraction of ion i in liquid phase,
dimensionless
x Equivalent fraction of bicarbonate in liquid phase,
dimensionless
x ^ Equivalent fraction of chloride in liquid phase,
dimensionless
x Equivalent fraction of nitrate in liquid phase,
dimensionless
xg Equivalent fraction of sulfate in liquid phase,
dimensionless
y^ Equivalent fraction of ion i in solid phase
y. Average equivalent fraction of ion i on the resin at
the end of the run, dimensionless.
y Average equivalent fraction of nitrate on the resin at
•N
the end of the run, dimensionless
xxii
-------
SYMBOLS
Ve Bed volumes of feed to nitrate breakthrough,
dimensionless
V Total bed volumes of feed solution
v Volume of resin bed, ml, 1
Z Valence of ion, dimensionless
ABBREVIATIONS
ANOVA Analysis of Variance
DETA Diethylenetriamine
HCHO Formaldehyde
ISO Isoporous
S
LOG S/N Ln a (used in statistical tables)
6 IN
N
LOG N/C1 Ln cu, (used in statistical tables)
NO~-N Nitrate concentration measured as nitrogen,
mg/1
N2POSITN Dummy Variable indicating whether nitrogen is
in the matrix (N2POSITN = 1.0) or out of the
matrix (N2POSITN = 0.0)
NAS National Academy of Science
PA Polyamine functionality
POLY Polyamine functionality
Q-l Quaternary amine, type 1
Q-2 Quaternary amine, type 2
Quat. Quaternary amine
RSIZE Dummy variable indicating relative size of amine
functional group; polyamines = 2.0, tertiary = 2.19,
quaternary =2.36
"R" Amine functional group
xiv
-------
ABBREVIATIONS
R
R
Ra
SIGNIF
S1
SBA
STY-DVB
TDS
TETA
X
WBA
Overbar denotes resin phase
Organic radical, -CH-, -C^H.OH etc.
Ratio of area below isotherm to area above isotherm
Statistical level of significance
Denotes abrupt transition zone in resin concentration
profile where species i is absent downstream
Strong-base anion
Styrene-divinylbenzene
Total dissolved solids
Tetraethylenetriamine
Liquid phase equivalent fraction
Weak-base anion
XLINKING Dummy variable indicating relative degree of
crosslinking: isoporous = 0.5, microporous =1.0
and macroporous = 2.0
xv
-------
ACKNOWLEDGMENTS
The following persons are acknowledged for their assistance
in the accomplishment of this work:
Tom Sorg EPA project officer for his interest and coopera-
tion during the length of the study.
R. Anderson and Dr. I. Abrams of the Diamond Shamrock
Chemical Company and D. Harrington of the Dow Chemical Company
for their very informative discussions on the composition and
properties of ion-exchange resins, and Dr. Judd Posner for his
helpful theoretical discussions.
Professor W.A. Ericson Director of the U-M Statistical
Research Lab for his assistance with the statistical analysis
and the interpretation of statistical results.
Steve Reiber, Ann Farrell, Jeff Meyers, Linda Burns and
Bill Hodgins for their work in the laboratory and in the prepara-
tion of the computer data plots.
Jill Schultz for her work in typing this and earlier drafts,
and Tom Hadfield and Diane Rumps for their help in typing the
initial drafts of this difficult manuscript.
xvi
-------
SECTION 1
INTRODUCTION
It is anticipated that the provisions of the U.S. Safe
Drinking Water Act of 1974 (Public Law 93-523) will apply to all
public water systems in mid 1977. Incorporated into that act
is a provision which, when it takes effect, will legally limit
the concentration of nitrate as nitrogen to 10 mg/£. This level
is equivalent to the long-standing, recommended limit established
by the U.S. Public Health Service for the prevention of methe-
moglobinemia in infants. Public and private water supplies in
nearly all of the fifty states and in many foreign countries
nave been found to be polluted with nitrates in amounts regularly
exceeding this 10 mg/A limit. Nitrate removal by ion exchange
with synthetic, organic, anion-exchange resins is the treatment
method which appears to offer the most readily available, proven
technology at a cost which is not unreasonable. However, dis-
posal of the spent nitrate-containing, regenerant-brine solution
is an unsolved problem and, previous to the time of this research,
tnere was a lack of technical information in the literature re-
garding the selectivity of the various anion exchange resins for
nitrate with respect to the important ground-water anions:
chloride, sulfate and bicarbonate. Neither was there sufficient,
useful information available for the prediction of multicomponent
effluent concentration profiles from ion-exchange columns econo-
mically operated by chromatographically eluting the ions not in-
tended to be removed.
-------
The research described here was undertaken to provide the
missing data and to propose hypotheses concerning the prediction
and control of anion exchange selectivity in general. A further
objective was to provide a means of describing the multicomponent
chromatogrphic column behavior of anion-exchange resins, espec-
ially weak-base resins, in nitrate removal service. A final
objective was to perform technical and economic evaluations com-
paring a conventional, single-bed, strong-base, nitrate removal
process to a two-bed, strong-acid, weak-base, nitrate removal
process which would produce a spent ammonium nitrate regenerant
amenaole to disposal as a fertilizer.
-------
SECTION 2
CONCLUSIONS
PHASE I: RESIN SELECTIVITY STUDY
Nineteen strong-base and thirteen weak-base resins were
tested for sulfate, nitrate, chloride and bicarbonate selectiv-
ities. These selectivities were then related to the following
resin properties: matrix, functionality, porosity, capactiy,
pKa and type.
Sulfate was always preferred over nitrate by all the strong
and weak-base resins tested. These synthetic polymers exhibited
an extremely wide range of selectivities . For strong-base
s s
resins a = 1.71 to 3.37, and for weak base resins afl = 2.67 to
N N
137. It is expected that the sulfate preference will hold true
for any of the resins tested here with ground waters having
total dissolved solids concentrations up to at least 3000 ppm,
i.e., 0.06 N as CaC03.
Nitrate was always preferred over chloride by all the anion
resins tested although the range of preferences was relatively
N
narrow. For stong base resins: ou^ = 2.85 - 3.64 and for weak-
base resins: acl =1.70-4.86. As expected this separation
factor was independent of total solution concentration.
Bicarbonate and carbonic acid were not significantly taken
up by the ion-exchange resins in binary equilibrium with dilute
-------
HNO_. The expected selectivity sequence has been verified as
sulfate > nitrate > chloride » bicarbonate.
The average separation factor, ou, as measured by the ratio-
of-areas technique proposed here, provided an adequate sulfate/
nitrate isotherm description and an excellent chloride/nitrate
isotherm description.
g
Resins with relatively low sulfate selectivity (aN = 2-4)
had modestly "S" shaped isotherms explained by their tendancy
to have sterically constrained sites of unequal preference for
g
divalent sulfate. Resins with high sulfate selectivity (aN =
13-137) had smooth-shaped isotherms, and titration curves with
inflection points for divalent H-SO, but not for monovalent HC1
or HNO.,; it is hypothesized that these latter resins have a
preponderence of pairs of appropriately spaced sites available
for divalent-ion interactions.
Matrix was the single most important factor in the determin-
ation of both oc^ and a^ and consequently of nitrate selectivity
in general. If the electrostatically active nitrogen atoms are
in the continuous polymer structure, as they are with all but the
polystyrene resins where the active nitrogen is pendant on the
polymer structure, then the resin is highly sulfate selective.
This, it is hypothesized, is due to the almost-guaranteed pro-
ximity of two active nitrogen atoms which are expected to be
separated by about 4.48 A in the polymer backbone. This distance
derives from the nitrogen separation distance of one ethylene
group in the amine monomers: diethylenetriamee—BETA, and
triethylenetetraamine—TETA, commonly used to provide function-
ality and crosslinking in anion exchange resins. For both
entropic and electrostatic reasons, these properly spaced, pro-
tonated amines much prefer multivalent ions to univalent ions.
-------
High nitrate/chloride selectivity tended to be associated
with polystyrene resins and highly crosslinked (macroporous) non-
polystyrene resins. These categories of resins are more hydro-
phobic than are the microporous non-polystyrene resins which
demonstrated lower nitrate/chloride selectivities.
Functionality was nearly as important as nitrogen-nitrogen
site proximity in determining sulfate selectivity but had no ap-
parent effect on the nitrate/chloride preferences of resins.
The size of the nitrogen functional group seems to be the
determining factor; larger functional groups tend to prevent the
required proximity of a pair of nitrogen atoms. Furthermore,
these large groups hinder the approach of the mobile counterions
to the positively charged nitrogen centers.
Porosity was a mc>jor determinant of sulfate selectivity,
among Type I strong-base anion resins where isoporous resins
with a relatively low degree of crosslinking were considerably
more sulfate selective (OL, = 2.98) than were the more-cross-
s
linked gel and macroporous resins (a = 1.82).
Type II, strong-base anion resins had higher sulfate
selectivity (otjj = 2.99) than did the Type I resins (ajj = 1.82).
Since the major difference here is basicity/ it appears that
reducing the basicity increases sulfate selectivity.
Predictive equations developed by an optimization of the
multiple regression analysis procedure, have verified that, when
considering all possible variables and all resins, the most im-
portant determinants of sulfate/nitrate selectivity are matrix
and functionality while matrix and relative degree of crosslink-
ing are the primary determinants of the magnitude of nitrate/
chloride selectivity. Nevertheless, within particular subclasses
of resins, other factors such as type and basicity (pKa) do have
-------
significant further influences on a and
PHASE II: MULTICOMPONENT CHROMATROGRAPH1C COLUMN STUDIES
E , the maximum possible chemical efficiency in nitrate re-
moval service, has been defined as being equal to the average
equivalent fraction of nitrate on the exhausted resin (y ). This
yN will be greater than x if the resin concentrates nitrate
by eluting the lesser preferred species (H2CO^ and Cl~) in chro-
matographic fashion until nitrate breakthrough. The most impor-
tant influence on yN is, predictably, x ; when it's low, process
efficiency will be correspondingly low because the exhausted
resin will contain mostly sulfate and chloride — species not
intended to be removed. A tentative procedure based on multi-
component equilibrium theory has been developed which correctly
predicts yM given the composition of the feedwater and the
TVT C?
relevant selectivities — a , and a .
Short detention times (T < 3.0 min), shallow beds (depth
3
< 60 cm) and high exhaustion rates (> 2.5 gal/min ft ) reduced
y by causing relatively more chloride, apparently the kineti-
cally favored anion, to be on the resin at nitrate breakthrough.
Nitrate/chloride selectivity (ac,) was the most important
selectivity in determining the relative amount of nitrate on
the resin at nitrate breakthrough.
g
Sulfate/nitrate selectivity (^N) was nearly irrelevant in
determining the average equivalent fraction of nitrate on the
resin at the end of a run. Surprisingly, slight increases in the
relative amount of nitrate on the resin are possible as a result
of increasing rather than decreasing the sulfate selectivity --
c
aN. The simplified explanation offered for this is that: (1)
all the sulfate will be removed from the feedwater regardless
-------
of its actual selectivity because it is the most preferred
species and, (2) high sulfate selectivity promotes a short
sulfate-rich zone near the column entrance in which almost no
nitrate is removed thereby leaving essentially all of that
species to compete with the lesser preferred chloride in the
second equilibrium zone of the column which is where nearly all
of the nitrate is concentrated.
Nitric acid was found to be unacceptable as a regenerant
in the two-bed process even though it would have greatly en-
hanced the fertilizer value of the regenerant wastewater. It is
too costly, 46.5C/1000 gal treated water (12.3^/m ), requires
excess cation bed rinsing to reduce nitrate and allows the pos-
sibility of disastrous nitrate and acid pollution of the water
supply in the event of an operating error. Even though HCl is
more costly than H2S04/ it may be more economical where low con-
centration and large excesses of H^SO. are required due to
potential CaSO. fouling of the cation resin.
It has been determined here that a regeneration level of
300% of the theoretical HCl required must be applied to the ca-
tion bed if calcium and magnesium are the primary cations on
the resin. Levels much lower than that cause premature cation
break til rough, increasing pH and reduced anion bed capacity with
smaller values of y at breakthrough. High regeneration levels
on the other hand miximize yN but may cause unacceptably low
effluent pH forcing termination of the run. A level of 300% or
greater will also be required for NaCl regeneration of the
single-bed process.
High column capacities can improve the overall economic
efficiency of the ion-exchange process if they lead to lower
rinse volume requirements but, since high"capacity resins also
tend to be highly sulfate selective and require progressively
longer rinse volume with service time, that possible improvement
-------
in operating efficiency is not guaranteed.
Net bicarbonate removal was zero, as expected for both the
two-bed and single-bed processes. Thus, high values of bicar-
bonate in the raw water don't measurably influence yN» Bicar-
bonate and carbonic acid apparently have a catalytic effect in
columnar ion-exchange processes so it is tentatively concluded
that a system degasifier should be located downstream from the
anion bed rather than preceding it.
A summary of the advantages ( + ) and disadvantages (-) of
the single-bed and two-bed processes follows:
Single-bed, strong base anion with NaCl regeneration
(+) Simple, no balancing of beds and regenerants
(+) Low cost regeneration
{-) Very difficult and costly to dispose of regenerants
in non-coastal locations where natural evaporation is
impossible
(-) Iron must be removed to prevent resin fouling
(-) Continuous nitrate analysis required for process
control
Two-bed, strong-acid, weak-base, NH3 & HC1 regenerants
(+) Partial softening in addition to nitrate removal
(+) No problem with iron fouling. Precipitated iron is
removed from the cation bed during each regeneration
(+) Regenerants wastewaters expected to be easy to dispose
of^by land application as fertilizer
(-) Complex system: bed sizes and regenerants must be
balanced
(-) Degasifier for C02 control required
(-) Continuous pH and nitrate analysis required for pro-
cess control
(-) High regenerant costs
-------
Wherever it can be used, the single-bed process will be
cheaper than the two-bed process. A comparative process economic
evaluation reveals that the two-bed process with NH3 and HC1 as
regenerants has chemical plus disposal costs which are approxi-
mately 50% higher than the single-bed process. Those costs are
27.84/1000 gal (7.34e chorinated to produce carcinogens.
The acceptable anion resins for nitrate removal service
are as follows considering that high nitrate/chloride selectiv-
ity, high capacity and moderate sulfate/nitrate selectivity are
the desirable characteristics:
Acceptable
STY-DVB, Polyamine, Gel Resins
STY-DVB, Tertiary Amine, MR Resins
STY-DVB, Quat. (I&II) Amines, Gel, MR & ISO
Acrylic-Amine, Polyamine, MR Resins
Phenol-HCHO/ Polyamine, MR Resins
Unacceptable
Epoxy-Amine, Polyamine Gel Resins
Acrylic-Amine, Tertiary Amine, Gel Resins
Aliphatic-Amine, Polyamine Gel Resins
-------
The overall chemical efficiency (E») can be expected to be
about 13.3% for both the single-bed and two-bed processes. This
is based on tne observed average equivalent fraction of nitrate
on the resin at the end of the runs (yN = .40) with a feedwater
containing the same equivalent concentration of nitrate, chloride
and sulfate and an irrelevant amount of bicarbonate which under-
goes no net removal in either process. This overall chemical
efficiency has been defined as the equivalents of nitrate re-
moved per equivalent of regenerant supplied and is the product
of VN and E_, the regeneration efficiency, which has been
determined to be 0.33 based on a regeneration level of 300%.
10
-------
SECTION 3
RECOMMENDATIONS
The degree to which various cation and anion resins yield
leachable organic compounds from breakdown of their polymer
structures should be quantified and the compounds identified.
These unwanted hydrocarbons are likely to be chlorinated during^
the traditional, water supply disinfection practices with the
subsequent formation of possible carcinogenic compounds. The
problem has some degree of urgency as the legal provisions of
the 1974 Safe Drinking Water Act will necessitate the more wide-
spread use of synthetic organic ion exchangers as the best
available water treatment technology for removal of trace ionic
contaminants, viz.: toxic metals, fluoride and nitrate.
The mathematical model*; available for the description of
multicomponent ion exchange should be modified to accommodate
the apparent acid-adsorption behavior of the kinetically-slow,
weak-base resins. These modified models should be amenable to
computer solution using numerical methods of analysis. There
is sufficient experimental data in this report to initially
verify a mathematical model of the multicomponent behavior of
ground waters containing nitrate, chloride, sulfate and bicar-
bonate fed to and eluted from weak or strong-base/ anion-exchange
columns.
Polymer research should be undertaken with the objective
of designing ion exchange polymers which will be selective for
monovalent-nitrate ions over divalent-sulfate ions based on the
11
-------
findings that the distance of nitrogen functional group separa-
tion is the most significant factor in univalent/polyvalent
ion separations. This proposed research work differs markedly
from the many previous attempts at designing nitrate-selective
resins by incorporating nitrate-specific organic radicals into
the resins; resins designed on that basis had severe polymer
stability problems and were nearly impossible to regenerate
because of their nitrate specificity. Such is not expected to
be the case with the proposed polymers although low ion-exchange
capacity might be a problem.
Combination water treatment systems incorporating reverse
osmosis for TDS reduction and ion exchange for nitrate removal
should be investigated.
Pilot plant studies of the single-bed and two-bed ion-ex-
change systems described here should be undertaken to verify the
cost estimates, assess the relative design and operational
complexities, and evaluate the alternative means of regenerant
disposal especially land application as a fertilizer. These
studies might best be done in a geographic location where
fluoride is also a problem. Fluoride is expected to be the
least-preferred ion, and control of the process by monitoring
the effluent for fluoride by ion-selective electrode would seem
to be practical.
The anticipated, beneficial, catalytic effect of carbonic
acid and bicarbonate ions on the nitrate removal efficiency of
anion exchangers in multicomponent ion-exchange service should
be investigated. Results of such an investigation should resolve
the question of whether to place the system degasifier upstream
or downstream of the anion exchanger.
12
-------
SECTION 4
THEORETICAL CONSIDERATION
THE NITRATE PROBLEM — BACKGROUND
Sources of Nitrate
Nitrate nitrogen (NO--N) is occasionally found in ground
water supplies at concentrations significantly above the long-
standing, recommended limit [125] and the interim legal limit of
10 mg/Jl [127]. Common sources of this nitrate pollution are:
(1) nitrified percolation from septic-tank tile fields, (2)
drainage and infiltration of fertilizer and feed-lot nitrogen
and, (3) ground water recharge operations using high-rate in-
filtration of secondary sewage effluents. The first two sources
generally give rise to NO3-N in the range of 20-50 mg/& [84, 109,
130] while the third produces ground waters in the 10-30 mg/&
range due to oxidation of the NH. in the recharge water by the
aerobic soil bacteria [16, 17, 28, 58]. Increasingly higher
food production and semi-rural population density throughout
the U.S., coupled with the pressing need to recharge ground
water supplies in semi-arid and coastal locations will serve
to worsen the N03~N problem in the near future.
Nitrate appears to be accumulating in many of our ground
waters at an alarming rate. Geographically, the ground water
nitrate problem is very widespread in rural, surburban and even
urban areas. Recently documented problems have been reported
13
-------
in Fresno, California [109, 130] and Long Island, New York [130],
Kentucky [96, 129] , Missouri [110], North Carolina [24], Texas
[50, 117], Wisconsin [33], Georgia, Iowa, Minnesota, Kansas,
Oklahoma and Illinois [123] , and in Europe [29] and specifically
in England [49], A recent bibliography with abstracts by
Lehman [79] and a bibliography by Summers [116] list numerous
references to this ubiquitous problem.
The seriousness of the health problem continues to be
debated because of the difficulty and expense of removing
nitrate from water and because the nitrate pollution is largely
rural and diffuse. Neventheless, the National Academy of
Science Committee on Nitrate Accumulation [91] concluded in
1972 that, even though infant methemoglobinemia had been nearly
eradicated in the United States: "The Public Health Service
recommended limits for nitrate in drinking water [10 ppm
NOo-N] should not be relaxed," and further that: "Equipment
for removing nitrogen from drinking water should be devised for
use in homes and on farms. Among the possible methods are
microbiological denitrification and anion-exchange resins"
[Reference 91 p. 74]. The NAS report reiterates that, since
records have been kept, 350 cases of methemoglobinemia have
been reported in the U.S., mostly in the years 1945-1950. An
estimated 2000 cases have been reported in North America and
Europe since 1945 with a 9% infant mortality rate [91].
Finally, it should be noted that the nitrate ion is much
more toxic to mammals than is the relatively nontoxic nitrate
ion, [93, 91] and it is nitrite, not nitrate that is responsi-
ble for infant deaths due £o methemoglobinemia. The water
quality standards limit the more stable and ubiquituous nitrate
ion concentration because nitrate may be reduced to nitrite in
the gastrointestinal tract, especially in infants. In the
stomachs of ruminant mammals, e.g., cattle, nitrate is readily
reduced to nitrite and death due to methemoglobinemia can occur
14
-------
within 2-5 min. after ingestion [23,91].
Nitrate Removal Processes
Conventional water treatment processes including coagula-
tion, filtration and chlorination have little effect on the
tt03 concentration of raw water. The high solubility of all
the common nitrate salts and the lack of co-precipitation and
adsorption of the nitrate anion are primarily responsible for
its perseverance in these processes.
Advanced water treatment processes appear to be limited
to ion exchange with synthetic, anion resins. Nitrate rejection
by cellulose-acetate reverse-somosis membranes is low (50-85%)
compared to HCO3" (80-98%), SO4= (99-100%), and Cl~ (86-97%)
[133]. Distillation and electrodialysis can usually be elimina-
ted from consideration because of their high cost for treating
low (less than 3000 ppm) TDS waters [133] , a consideration also
applying to reverse osmosis where it has been estimated to be
10 times as costly as ion exchange at 10 ppm NO--N in a 450
ppm Tl5s water (See discussion in Ref. 40). Biological denitri-
fication, [114,115] the preferred method for treating waste-
waters, has been demonstrated to be technically feasible but
its high cost, long detention times (1-2 hours), increased
use of chlorine to eliminate bacteria and nitrite, and the re-
quirement for post aeration and filtration would seem to pre-
clude its use for treatment of public and private water supplies.
Previous studies using anion exchange for nitrate removal
from secondary sewage effluent and agricultural drainage have
met with considerable success. Eliassen found, on a pilot scale
that, nitrified secondary effluent containing 18 ppm NCU-N,
65 ppm SO4~ and 200 ppm Cl" could be successfully treated for
nitrate removal using diatomaceous earth filtration and anion
exchange for about 21C/1000 gallons (1965 dollars, not including
15
-------
brine disposal) [44, 45, 46]. In these tests the anion selec-
tivity was SO4= > N03~ > N02~ > Cl~ > HC03~ for the type II,
a strong-base anion resin Duolite A-102-D. The findings of
Midkiff and Weber [87] which were derived from type I, strong-
base, anion column experiments are in accord with this selectiv-
ity sequence. Additional data regarding selectivity with strong-
base resins can be found in a review by Sabadell [107] and in
N
Table I of reference 56 where a.-,, is in the range of 2 — 3.
The Search for Nitrate Selective Resin
The search for a highly nitrate-selective resin has domina-
ted several investigations. Grinstead and Jones [55] described
the development of a very nitrate-selective (N0_ /Cl = 20/1
and N03~/SO.~ » 20/1) adsorbent for wastewater treatment.
Their system comprised a macroreticular, polystyrene resin with-
out functional groups into which was adsorbed a nitrate-selective
amidine reagent dissolved in an organic solvent. This they
have termed the "extractant-in-bead-approach". Its one advan-
tage was high nitrate selectivity with respect to sulfate,
chloride, and bicarbonate. Its disadvantages were low capacity
(in meq/ml) compared to conventional ion-exchange resins, and
the continuous loss of organic solvent and extractant into the
treated water.
"Entirely inadequate capacity" was a disadvantage of the
nitrate selective resins prepared by Meloan and Gran [85] who
chemically"incorporated nitrate-selective, organic reagents
into commercially avaialbe, weak-base, anion exchangers. While
not specifically investigated in that work, the problem of
extractable organics leaching into the water supply is also
expected to be a significant problem with this approach.
Wallit and Jones [129] succeeded in developing a true,
anion-exchanger resin with salinogen functional groups attached
16
-------
to a polystyrene matrix. Working only with nitrates and
chlorides, they obtained nitrate selectivities in the range of
8/1 to 14/1. In fact, the resins were so selective for nitrates
that they couldn't be regenerated by ordinary methods, thereby
rendering the process commercially unfeasible. Here again
extractable organics are expected to cause problems.
The dilemma of high selectivity producing efficient ion
excnange with inefficient regeneration due to the very signifi-
cant stoichiometric excess of regenerant required has been
pointed out previously [10, 129]. The desirable selectivity
sequence for nitrate removal is regenerant anion » NO ~ > SO.",
Cl , HCO.. with a very low regenerant anion concentration
in the raw water. This situation is approached with weak-base,
anion resins where hydroxide is the much preferred anion and
is essentially absent from the acidic effluent of a strong-acid,
cation unit. Finding or producing a nitrate-selective, weak-base
resin would help resolve the dilemma, although that is not expec-
ted to be an easy task. The nitrate selectivities of commer-
cially available resins were determined in this study.
The generally accepted anion selectivity sequence for both
weak and strong base resins is SO," > NO ~ > Cl" > HC03~ [36,
60, 78, 86, 133]. Verification of the sequence at anion con-
centrations normally found in groundwaters and wastewaters has
been demonstrated by many investigations [10, 44, 49, 56, 86,
87] for strong-base resins. Nevertheless, significant differ-
ences in the actual selectivity values among various, strong-
base anion resins do exist as demonstrated by Gregory and Dhond
who experimented with ten different, strong-base resins and the
anions: S04~, HPO4= and Cl~ [54],
Beulow et. al. [5] clarified the statement by Chemical
Separations Corporation that Dowex 21K, strong-base resin was
nitrate selective by showing that it was true only at concentra-
17
-------
tion levels near and above 50 meq/£ (2400 ppm SO or 3100 ppm
NO ~), concentrations which are of little interest in water
supplies. This reversal of selectivity is due to the activity
coefficient-concentration relationship and has been previously
reported for the SO ~/Cl"~ selectivity which inverts in favor
of chloride at concentrations above 63 meq/1 (2240 ppm Cl }
[78, 134]. Divalent ions are nearly always preferred over mono-
valent ions by synthetic organic ion exchangers in the total
concentration levels usually found in water supplies [13, 18,
36, 40, 60, 82]. This preference has been termed "electro-
selectivity" [60] .
Iron Fouling of Nitrate Removal Resin
The ferrous iron commonly occurring in ground waters may be
converted to ferric iron by dissolved oxygen either in the water
prior to contact with the ion-exchange resins or, in the case of
cation exchange resins, within the actual resin pores where the
resulting, insoluble, ferric iron oxides precipitate and foul
the beads [3], Iron fouling of anion resins is primarily con-
fined to the surface of the beads where it tends to prevent the
exchangeable counterions from gaining access to the interior of
the beads. Using X-ray scanning techniques on the cross section
of a weak-base anion resin used in the desulfatization of sea
water, Aveni et. al. [8] verified that the iron fouling was
serious but limited to the anion resin bead surface. With
simple backwashing and NaCl regeneration, these iron deposits
are only partially removed and eventually reduce the exchange
capacity to intolerably low levels as experiences by Beulow [10]
in a nitrate removal study using strong-base anion resins.
Where the fouling is severe, the resins must be removed from the
columns (if the columns aren't acid resistant) and washed with
acid to dissolve the precipitated iron oxides. Such cleaning is
not expected to be required in the case of cation resin beds re-
18
-------
generated with HC1, H-SO. or HNO., which should remove the ad-
hering iron oxides on every regeneration cycle. This can be
considered an advantage offsetting the capital and operating
cost disadvantages of two-bed (cation-anion) nitrate removal
processes when strong acids are used to regenerate the cation
resin. Any large-scale, nitrate removal process for ground
water supplies must successfully deal with the iron fouling pro-
blem or it cannot be considered widely applicable.
The Problem of Regenerant Brine Disposal
The regenerant NaCl and its disposal have been shown or
estimated to be the most significant costs when removing nitrates
with a single, strong-base, ion-exchange column [10, 66]. Un-
fortunately, both costs can only increase for alternative systems
since NaCl is the lowest cost regenerant available on a $/lb-
equivalent basis. Furthermore, trucking the NaCl-NaNO^-Na,,S04
brine to a nearby stream for "dilution" disposal, the low-cost
method suggested by Holzmacher [66], cannot be recommended here
oecause of its detrimental, nutrient effect on the receiving
stream.
The alternatives for regenerant brine disposal are assumed
to be limited to the following:
(1) ocean outfalls in coastal locations
(2) evaporation ponds in semi-arid regions
(3) sanitary sewers where permitted, but only recommended
where sewage denitrification facilities exist and
where the brine doesn't seriously dilute the sewage
(4) deep-well injection where permitted, very costly for
large volumes of brine
(5) sale as fertilizer, the most desirable.
For reuse as a fertilizer, the nitrogen content of the brine
should be maximized and electrolytes like sodium ions should be
19
-------
minimized because of their detrimental effects on soils [84, 92,
118]. This effectively eliminates NaCl (and KC1) from considera-
tion.
Bingnam [11] has described a two-bed, strong-acid, weak-
base, continuous, ion-exchange process for nitrate removal from
fertilizer plant effluent. HNO^ and NH.OH are used as regener-
ants in that process to produce a NH.NO., brine which is recycled
to the NH.NO- fertilizer production plant. The process is not
directly applicable to water supply because their nitrate levels
were "extremely" high, and no competing anions were mentioned.
However, this basic system appears very attractive with respect
to regenerant disposal as a soluble, fertilizer by-product which
would have relatively low concentrations of the persistent ions:
Ca , Mg , Na , Cl~ and SO,". In addition, a material balance
performed on isolated geographic areas where ground water nitrate
is a problem would demonstrate that recycling the "old" nitrate
as a local fertilizer would lessen the accumulation of nitrate
in the local and surrounding ground water and surface water due
to a reduction in the input of new fertilizer nitrogen required
from outside the area.
The Sirotnerm Process; Thermal Regeneration
It has been suggested in a review by Sabadel [61] that the
waste disposal problem in nitrate ion exchange might be elimina-
ted by use of thermal rather than chemical regeneration. A
tnermal regeneration process trade-named Sirotherm has been dev-
eloped by Weiss et. a. [15, 134] comprising a single-bed ex-
changer of mixed weak-acid, weak-base resins operated at low
(20°C) temperature during the ion-exchange step and at high
(80°C) temperature during the regeneration, acid-base elution,
step. Since "low-grade heat sources" (under 90°C) are used, the
costs of regeneration, and hence the operating costs of the
process are said to be nil. However, in 1969, Bregman and
20
-------
Schackelford [22] pointed out several significant disadvantages
of the process:
(1) The ion-exchange kinetics are very slow since the op-
eration takes place at neutral (5 to 7) pH.
(2) The extremely fine particles (5 to 20 microns) which
must be used to obtain reasonable ion-exchange rates lead to
enormous bed surfaces and very-low flow rates which in turn
promote flow distribution problems and prevent rapid heat trans-
fer during regeneration.
(3) Because resin capacity is limited to that arising from
the differences in resin pKa's between 20 °C and 80°C, capacities
of less than 1 meq/gm result as compared to 5-9 meq/gm for these
same resins regenerated chemically after being operated at
basic or acidic pH's in two-bed systems.
(4) High wastewater-to-product-water flow rates are char-
acteristic of the Sirotherm process due to these low resin
capacities and the need for frequent regeneration.
According to Bolto (personal communication and Ref. 15) by
1975 the disadvantages arising from the well-known, slow kinetics
for weak resins [6, 14, 61, 62, 76, 134, 135] and very low resin
capacities had been largely overcome and several successful
12,000 gpd (45 m /day) pilot plants (both fixed bed and contin-
uous) had demonstrated the usefulness of the process in partially
desalting ground waters containing 1000—2000 ppm TDS. Product
water yields were in the 67—91% range with typical TDS reduc-
tions of 50—60% while wasting 9—33% of the feed as an 80°C
wastewater with 3000 to 5500 ppm TDS. Typical product-to-waste-
water flows were 4/1 to 9/1 with more complicated, continuous
ion-exchange designs and staged operations being required for
the 9/1 ratio. The reported capacities were still quite low
(0.12 to 0.20 meq/ml) compared to conventional, chemically re-
generated resins (1.0 to 3.0 meq/ml) in the same type of service.
In coastal locations and in semi-arid regions where low-cost
21
-------
land is available the disposal of large volumes of saline, 80°C
water may be accomplished by discharging into the ocean or by
evaporation.
For non-coastal, non-arid, nitrate-removal applications,
wastewater disposal will be a serious problem. Furthermore,
particulates and oxygen must be completely removed prior to
Sirotherm desalination. Lastly, the inevitable, accelerated
resin deterioration upon repeated cycling to 80°C may cause a
serious organics problem in the product water.
No published cost figures are available on the Sirotherm
process, but ICI Australia Ltd. a partner in the process devel-
opment is planning to build a 165,000 gpd (625 m /day) Sirotherm
commercial desalting process at one of their plants [15].
Strong-Base Anion Exchangers; Summary
The previous discussion pertaining to single-bed strong-
base anion systems for nitrate removal from water supplies can
be summarized as follows:
(1) Resin selectivity for nitrate is a serious problem
because sulfate is preferred with a selectivity ratio of over
2/1 at low TDS.
(2) Ferrous iron, when present, oxidizes, precipitates,
and seriously fouls the resin.
(3) Regeneration and brine disposal are the major economic
and environmental problems yet to be solved even with low-cost
NaCl regeneration.
THE PROCESS PROPOSED FOR STUDY
Process Description
Because the strong-acid, weak-base process appeared to have
22
-------
certain advantages with respect to regeneration efficiency, iron
removal, regenerant disposal and possibly nitrate selectivity,
the little-studied, weak-base resin part of the system shown in
Figure 1, following, was studied in detail. The thermodynamic
and kinetic results of the work were compared and contrasted
to those of a single, strong-base, anion resin in similar
nitrate removal service; see Figure 2.
Advantages and Disadvantages of Proposed Process^
Evans [47] reported on similar nitrate removal process but
with HC1 and lime as the regenerants. He pointed out that,
even after the cation bed was exhausted, and sodium was being
eluted, the system continued to provide softening and nitrate
removal thereby delivering greater than stoichiometric efficiency
due to tne weak-base anion resin's apparent selectivity for
nitrate over all the other anions present. Interpreting his
published results, this author calculated the following selec-
tivity sequence: HN03 » H2C03 > H2S04 > HCl which is in ob-
vious contrast to the previously reported sequence of H SO. >
HN03 > HCl » H2C03 [36, 60, 78] for both weak and strong base
resins. The high selectivity for H_C03 is very weakly held on
strong-base resins and is always the first ion to break through
in column studies [10, 44, 86]. Further, several sources of
published, ion-exchange design information [36, 104] state un-
equivocally that carbonic acid is not significantly removed by
weak-base, anion resins. This very unusual selectivity sequence
indicated by Evans' data is most likely due to one or more of
the following: (1) true thermodynamic selectivity (2) kinetic
selectivity due to non-equilibrium, mass transfer or (3) analy-
tical errors. An explanation based primarily on (1) and (2)
above was favored since a weak-base resin with pK = 8 would
certainly have reacted with (adsorbed) H2C03 at low pH (<4) as
CO2 has been shown to be readily stripped from air by weak-base
resins [132]. Further, weak-base resins are known to be kine-
23
-------
Raw Water Influent
Flow = Q
Nitrate-N = 20mg/l
TDS= 380 mg/l
Hardness =225 mg/l
NaHC03
Ca(N03)2
Mg S04
UOUI2
Fe S04
1
i
0.25 Q
, Regenerant
| NaCI
i ( lOW COSt)
1
Strong
Base
Anion
Exchanger
Chloride
Form
1 ;
i
Bypass k ^
L
[ 0.75 Q
— Ion Exchange Column
Effluent
Ca CI2
MgCI2
NaCI
Fe CI2
1
Raw r'
Water
|
Spent Regenerant
NaCI -NaN03 Brine
(Disposal Problem)
LBlended Product Water
Nitrate-N =5-10 mg/l
IDS = 296-380 mg/l
Hardness = 225 mg/l
Chloride = 53-195 mg/l
Figure I Conventional Single Bed Ion Exchange Process
24
-------
-Row Water (Typical)
Flow s Q
Nitrate-N= 20ppm
TDS =380ppm
Hardness =225ppm
NaHC03
Ca(N03)2 I HMOs Regenerant
MqSCU ([Alternatively,]
CaCI2 ![HCI or H2S04J
FeS04
NH4OH|
Regeneranti
^
Bypass Rai
^r t
Strong
Acid
Cation
Exchanger
Form
1
1
1
1
1
r-Cation Effluent
H2C03
HN03
H2S04
HCI
r "
Spent Acid | Spent
1 Amonia
w Water J^
t t
Weak
Base
Anion
Exchanger
Free
Base
Form
_ ?
\
Flow*. 25 Q !
\
\
\
\
\
\
\
\
•\ i
-Ion -Exchange
Column Flow*. 75 Q
Combined Regenerants
NH4N03 Solution
(Fertilizer)
Blended
Product Water
Nitrate-N=5-10ppm
IDS'95-380ppm
Hardness = 56-225
ppm
Figure 2 Proposed Two-Bed. Ion-Exchange Process
25
-------
tically much slower than tne strong base variety, a fact which
potentially leads to mass transfer limitations of the separa-
tions. During the course of the proposed research, the rate and
extent of caroonic acid adsorption on weak-base resins was
examined to determine which of the above reasons accounted for
the anomalies reported.
With regard to the proposed process (Fig. 1), note that,
prior to breakthrough of the cation bed, the influent to the
anion bed will be quite acidic (pH * 2.4 for a ground water
with 250 ppm CaCCU hardness) and the total anion capacity will
depend on the quaternary equilibria with HN03, H2S04, HCl, and
H2CO^ assuming that OH will be negligible. With a neutral pH
influent to the anion bed, i.e. after cation bed exhaustion,
the resin capacity will depend on the 5-component, ion-exchange
equilibria of OH~, N03 , S04~, Cl" and HCO ~; consequently, the
equilibria and column kinetics of both these situations were
studied.
In such systems a degasifier to remove CO2 is usually re-
commended as a unit process following the cation bed and prece-
ding the weak-base anion bed. Such a system was described by
Sanks and Kaufman [108] for tertiary treatment of wastewater for
recycling. With a degasifier in this position, CO» is readily
given off because a low pH is maintained by the strong acids
present. For the two-bed system being studied here, C02 removal
preceding the anion bed might prove to be a negative feature;
removing ^CO^ will prevent that acid from adsorbing on the weak-
base resin where it could later exchange HCO ~ for Cl~ or NO ~
during the softening cycle when the cation bed is spent and its
effluent is neutral. Furthermore, some beneficial kinetic
effects due to the presence of carbonic acid and bicarbonate ions
in the anion bed influent were also expected (personal communi-
cation, I.W. Abrams). For these reasons a degasifier was not
used during the experimental work.
26
-------
Regardless of the efficiency of bicarbonate removal, the
added benefits of demineralized water or soft, nitrate-free
water tend to offset the cost disadvantages of two-bed systems
with their requirements for regeneration and neutralization of
two beds rather than one.
Another feature of the system is bypass blending of the raw
water. This feature permits control of the nitrate concentration
at values approaching the permissible limit in the blended
water supplied to the distribution system. Thus, not all of the
raw water needs to be treated; typically one-forth to one-half
of the raw water will bypass the ion-exchange beds.
THE STRUCTURE OF ION-EXCHANGE RESINS
Introduction
A brief description of the chemical and physical structures
of at least one representative type of resin from each major
classification of synthetic, organic ion exchangers is included
here to facilitate the explanation of (1) differences between
strong and weak resins, (2) selectivity theory, and (3) kinetic
theory.
27
-------
A Typicaj^ Strong-Acid Cation Resin; Duolite C-20
-CH-CH2-CH-CH2-CH
S03H i S03H
CH2-CH-CH2
•••—CH2-CH
S03H S03H
Sulfonated polystyrene-divinylbenzene copolymer
Typical degree of crosslinking: 8%
Physical form: Translucent spheres
Specific gravity: 1.23, hydrogen form
Moisture retention capacity: 50%, hydrogen form
Effective size: 0.45 to 0.55 mm+ +
Swelling: -7% when going from H to Na form
Ion-exchange capacity: 4.8 meq/gm, 2.0 meq/ml
Uniformity coefficient: 1.4 to 1.8
Functional group: R-SOjH
Acidity: pK < 1, ionized at pH > 1
cl
R-S03H +
Na
RS03Na
H
28
-------
A Weak-Acid Cation Resin
CH CH
COOH COOH
Methacrylic acid-divinylbenzene copolymer
Functional group: COOH
Acidity: pKa « 4 to 6, ionized at pH > 5
Swelling: +65% going from H^to Na*form
Capacity: 10 meq/gm, 4.3 meq/gm
RCOOH + Na «- RCOON + H
cl
R denotes the resin matrix
Strong-Base Anion Resins
-CH2-CH-CH2-
R denotes a methyl on ethanol group
If all the "R" groups are methyl/ the resin is a Type 1
Quaternary ammonium resin. Type 2 resins have two methyl and one
ethanol group as shown below:
29
-------
CH* r\-
i + Cl i+ *
•CH2-N-CH3 -CH2-N-C2H4OH
CH3 CH3
Type 1 Type 2
Typical Strong-base anion resin: Duolite A-101-D, Type 1
Physical form: moist, cream-colored beads, opaque
Moisture retention capactiy: 50%, chloride form
Specific gravity: 1.07 chloride form
Capacity: 4.0 meq/gm, 1.3 meq/ml
Swelling: -12% going from OH"to Cl~ form
RN(CH3)3C1 + N03 «- RN(CH3)
Cl
Weak-Base Tfnion Resins
-CH-CH2-CH-CH2-CH-
0 O
CH3- N -CH2 I CH2-N-CH3
HCI CH2-CH-CH2
-CH2-CH CH-CH2-
CH3-N-CH2 CH2-N-CH3
HCI HCI
Styrene-divinylbenzene copolymer with tertiary-amine
functionality
Typical examples: Amberlite IRA-93, Duolite ES-368
Physical from: tan, spherical particles
Moisture retention: 50%, free base form
Capacity: 3.8 meq/gm, 1.3 meq/ml
Swelling: +23% free base to salt form
Basicity: pK - 7 to 9, ionized at pH < 8
3.
HNC
30
-------
HCI HCI HCI
Phenol-formaldehyde polyamine, condensation polymer with
secondary amine functionality
Typical example: Duolite A-7
Physical form: cream colored granules
Specific gravity: 1.12, free base form
Particle size: 0.3 to 1.2 mm
Moisture retention: 60%
Total capacity: 9.1 meq/gm, 2.4 meq/ml
Swelling: +18% going from free base to salt form
Basicity: pK_ * 7 to 9, ionized at pH < 8
cl
R NH + HCI «• R2NH-HC1
Other Common Weak-Base Resins
N-CH2-CH-CH2-N-CH2-CH2-N-CH
i z i * i * * \
CH2 OH CH2 CH2
CH2 HC-OH HC-OH
i i i
HN—- CH2 CH2
I I
--N-CH2-CH2-NH
Epoxy-polyamine condensation polymer
CH2-CH2
I
00 !
i i
NH-CH2-CH2-NH-CH2-CH2-N
C=0
•—CH2-CH —
Polyacrylic-polyamine copolymer
31
-------
Somes Significant Resin Comparisons
Strong resins shrink modestly (7 to 12%) when going from
the acid or base to the salt forms whereas weak resins swell
significantly (18 to 65%) during this same type of transition.
Shrinking denotes a thermodynamic preference for the shrunken
state in agreement with the high selectivities observed for the
hydrogen and free-base forms of weak-acid and weak-base resins
respectively.
Table 1 summarizes the pK 's associated with various func-
a
tional groups. Note that the capacity of a weak base resin is
significant only at pH's below the listed pK i.e. weak-base
a
resins won't "split neutral salts" to a significant extent. The
resins will first adsorb acids then exchange anions.
TABLE 1. SUMMARY pK 's FOR ANION RESINS [60]
Si
Resin Structure Apparent pK
a
Type 1, Strong-Base -N(CH3)3OH >13
Type 2, Strong-Base -N(C9H.OH) (CH,),OH >13
^£ f» O £»
Secondary Amine, Weak-Base -N(CHOH 7 to 9
Tertiary Amine, Weak-Base -N(CH2)2 7 to 9
Primary Amine, Weak-Base -NH? 7 to 9
Phenylamine, Weak-Base -/y—NH- 5 to 6
THEORIES OF ION-EXCHANGE SELECTIVITY
Definition of the Selectivity Coefficient; K^
Utilizing Donan membrane equilibrium theory [9], the law of
mass action [103] , or Langmuir isotherms [18], one arrives at an
32
-------
equilibrium expression which is the same for all three
sider the general ion-exchange reaction:
Con-
M
RaA
RbB
+a+b = valence of ion
R = resin
A = overbar denotes resin phase
Choosing the hypothetical state of unit activity for infinitely
dilute solution of both ions in both phases, the following
expression results at equilibrium:
B
#1? *
B
V^1
TA
ac
B
Y
B
q
c
activity of B in the resin phase
activity of B in the solution phase
activity coefficient of B in the resin
activity coefficient of B in the solution
resin phase concentration
solution phase concentration
The Selectivity coefficient, K^ has been defined as: [60]
f\.
K
B
'B
(3)
Interpreted in terms of activity coefficients:
K
.B
W JaJ Jbl
= TB TA
(4)
A B
Usually, in dilute solutionsVso that the selectivity coefficient
is determined by the activities of the respective ions in the
33
-------
resin phase only:
Jbl R B solution-phase
K & 'A = K^ K" = corrected selectivity
A ~ja]~ coefficient
YB (5)
g
It is important to note that K is a coefficient and is not
necessarily constant as the activities of the ions in the very
concentrated resin phase tend to depend on the ratio of the
n
concentrations present. Generally/ K decreases as yB (the
equivalent fraction of B in the resin) increased [103].
•p
The Separation Factor; a
-• -- - - ... . . - . ^^
The widely accepted definition for the parameter describing
partitioning of solutes between two phases is:
B _ qB/C ,6)
aA - B/ B (6)
For monovalent ion exchange then:
B A = B A (7)
A
C a x
•Q "a •"•n
O A D
where:
y. = Equivalent fraction of A in resin phase
x = Equivalent fraction of A in solution phase
B B
For univalent-divalent exchange a ^ K . Since the separation
£\ £\
factor doesn't include the stoichiometric coefficients as ex-
ponents, it's a mathematically and physically more satisfying
description of solute distribution even though it is also not
usually a constant. See Appendix 5, Justification of a..
The Concept of Electroselectivity [60, 103]
Consider the case of S04~/N03~ ion exchange in dilute
34
-------
(0.010 N.) aqueous solution where CN = Cg = 0.005 N.
SO" -i- 2RN03 «- R2S04 + 2N03
K? = KM =
A N
P — 1
qs
_LS _
1 — —1
CN
_qN_
2
1 ~1
co
Q
1 — — i
YS
Lxs_
,— —,
XN
YN
(8)
(9)
CQ = Total solution concentration, meq/ml
Q = Resin exchange capacity, meq/gm
Theoretically -(and acutally), this selectivity coefficient is a
function of the total solution concentration CQ (which is not
the case in univalent-univalent or divalent-divalent exchange).
Now, if we further assume that the resin has no "selectivity"
i.e., K^ = 1, this does not imply an inability to separate SO4~
from NO3~. For example, if Q = 8 meq/gm (a typical value), the
calculated separation factor - 50, i.e., the resin phase contains
50 times as many equivalents of sulfate as nitrate. This
theoretical ability to separate multivalent ions has been
termed "electroselectivity" [60] , and has been found to be a
fair approximation for cation exchange. For strong-base anion
c
exchange in this range of concentration however, a has been
found to be more like 2.5 not 50. So, here, Donan membrane
equilibrium and mass-action derivations which assume nearly
equal resin-phase, activity coefficients are poor approximations.
This was thought to be fortunate since we intuitively desired
g
that a be as low as possible. To that end, the equilibrium of
sulfate and nitrate with a large number of strong and weak-base
anion resins was studied. It will be shown later that the
intuition about the sulfate/nitrate selectivity being the most
important selectivity was incorrect.
As a final comment, it can be shown that this electroselec-
tivity preference for the multiply-charged ion becomes greater
with increasing dilution of the external solution. Conversely,
35
-------
at high solution concentration, the electroselectivity dimini-
shes, and in some cases, e.g. SO ~/Cl at C ^> 0.063 N and
SO ~/NO ~ at C >_ 0.050 N, inverts in favor of the monovalent
ion. See Refs. 77 and 10 respectively.
Binary Isotherms
Having chosen the separation factor for description of
T3
anion equilibria, it should be noted that even if a is a
£\
constant, linearity of the isotherm plot of a y vs. x is not
A A
implied. In fact, the Langmuir, multicomponent equilibrium
P
treatment leads to a = constant; for example;
, _ (10)
(11)
aa = B A = B = constant (12)
Q = Langmuir ultimate solid-phase adsorption (or ion ex-
change) capacity
b. = Langmuir constant related to adsorption (or ion ex-
change) energy
Q
1
Q
%
- b7V C,
A TV
i~X *"*
+ bA CA H
" bB CB
+ t»A CA H
A A
h bB CB
•bB CB
q C b
/» Xj A
36
-------
Figure 3 below illustrates constant and variable separation
factor isotherms:
0
B
/Theoretical, non-constant a , curves
Jfor exchange of ions of dissimilar
tvalence, i.e. a and KB = f(Cn)
FIGURE 3
EXAMPLE ISOTHERMS
7}
f Theoretical, constant cc. curve for ideal
exchange of ions of equal valence or non-ideal
exchange in narrow concentration ranges for
ions with dissimilar valence on resins of
low selectivity.
0
1
1
0
Figure 4- below demonstrates that the binary separation factor
aA is equivalent to the ratio of the area 1 and 2.
£
(1-yA)(xA)
Area 1
Area2
FIGURE 4
ISOTHERM AREAS
37
-------
General Considerations Regarding Selectivity
As a result of much theoretical speculation and some ex-
perimental verification [21, 18, 20, 26, 39, 40, 43, 60, 86, 103
113] the following factors have been found to influence the
preference which any resin exhibits for a given ion (or ions).
The cation or anion exchanger is reported to prefer:
(1) The counterion with the highest valence.
(2) The counterion with the smallest, hydrated-ionic
raduis.
(3) The counterion which interacts most strongly with the
fixed ionic groups on the resin (especially true for weak-acid
or weak-base resins).
(4) The counterion with the greatest polarizability.
(5) The counterion causing the least swelling of the
resin.
(6) The counterion with the lowest free energy of hydra-
tion in aqueous solution.
Rules (1) and (2) can be applied without exception to the
alkali and alkaline earth cations, in fact it is from experi-
ments with these cations that the rules were derived. However,
anion exchange is not exactly analogous to strong-acid cation
exchange, the following important differences being relevant
to the research done here.
(1) The charge on the counterion in anion exchange has
much less effect on selectivity than with cation exchange [39,
113] .
(2) The nature of the functional group, especially its
size and charge density have a significant effect on anion
selectivity [13].
(3) When going from strong-base -N(CH_)_ , quarernary
+
ammonium groups to weak-base -N(CH3)2H groups, the selectivity
sequence for the halide ions: I»Br~»F~ (1000»150»5) remains
unchanged but the magnitude of the differences is reduced mark-
38
-------
edly (100 40 10). See figures 4-2 and 4-3 from reference [39].
On this basis it was expected that the selectivity sequence
SO4= > NO-~ > Cl~ would be the same with weak-base resins but
with smaller absolute differences in selectivity values. Gener-
ally this was not found to be true.
(4) The hydrated ionic radius is not necessarily the most
important factor in anion exchange selectivities. In fact,
Reichenberg [103] argues that this "apparent correlation" be-
tween selectivity and hydrated ionic radius is an "unfortunate
accident" and that the true causal relationship is due to the
free energy of hydration i.e., selectivity is inversely propor-
tional to this energy. (See also Eisemen) [43]. As an example
of this he cites the well-known selectivity sequence of
CIO4 > I~ > Br~ > Cl~ on strong-base resins with CIO." being
preferred to Cl~ by more than 100/1. Based on hydrated ionic
radius the sequence should be Br~ > I~ > ci~ > CIO.". However,
based on anionic-hydration enthalpy the correct sequence is
predicted. It is interesting to note that HCIO. is also much
preferred to HC1 during adsorption from aqueous solution onto
activated, coconut-shell carbon [112] probably for the same
reason as with synthetic organic resins.
In this same vein, Midkiff [50] in what appears to be a
rather bold departure from the accepted hydrated ionic radius
theories, had a good degree of success in correlating selecti-
vity to ionic valence and crystal ionic radius. The basic equa-
tion which he applied to polyatomic-anion exchange on strong-
base resin in dilute aqueous solution is:
(13) K^ a =^ Z = valence
A *^T
R... = crystal ionic radius
a = indicates proportionality
The selectivity sequence predicted by the above equation was
calculated as:
39
-------
P04 > CO3 > SO4 > HP04 > N03 > HC03 > H
His experimentally observed selectivity sequence was nearly
as predicted with the position of CO.," being the only notable
exception, i.e.,
S04 > HP04 >
Careful examination of the actual relationship between hydrated
ionic radii and his calculated, crystal ionic radii (using the
accepted criterion of ionic conductance being inversely pro-
portional to hydrated ionic radius) discloses that the inverse
relationship expected, based on observations of the alkali
and alkaline earth metal cations completely fails with the
polyatomic anions studied. This is to say that hydrated ionic
radius is not inversely proportional to crystal ionic radius as
it is with cations, but directly proportional to it, and that
explains why the prediction was so good. So, the rule-of-thumb
stating that selectivity is inversely proportional to hydrated
ionic radius still applies (except for CO ~) in the specific
system described.
Summary of Selectivity Considerations
Based on published data [10, 36, 60, 78, 113] the selec-
tivity (and separation factor) sequence expected for strong-
base ion exchange with the anions of interest is:
S04" > N03" > N02~ > Cl~ > HCOj » OH"
No single criterion such as limiting ionic conductance, free
energy of hydration, valence or combination of valence and radius
can be used to correctly predict the entire sequence even in the
simplest of systems. Part of the intended research effort was
aimed at verifying and quantifying the above sequence for ground
waters in equilibrium with the most nitrate selective strong-base
40
-------
resins.
For weak-base resins, the sequence based on published data
[18, 60, 78] is essentially the same as above:
H2S04 > HN03 > HC1 » H2C03
However, it must be observed that prior to the research reported
here no useful systematic treatment of weak-base resin equili-
bria could be found in the literature , thus the above sequence
was originally viewed only as a guide although it was later
verified. It is to be noted that HN02 is absent completely due
to lack of any published data. Finally, recalling the earlier
discussion of Evans1 experimental results where the calculated
sequence was (surprisingly)
HNO3 » H2CO3 > H2SO4 > HC1
it was concluded that a systematic experimental treatment of
weak-base equilibria needed to be undertaken to resolve the
problem as the actual sequence would determine the nitrate ion
exchange capacity in chromatographic elution service.
MULTICOMPONENT EQUILIBRIUM THEORY
Batch Equilibrium Studies
Based on the preceeding discussion, it may reasonably be
concluded that there is no point in dealing with predictive
equations based on thermodynamic considerations for multicompon-
ent equilibria when binary selectivities can't even be correctly
predicted.
Some encouraging evidence that experimental, binary equili-
41
-------
brium data might be applied to batch systems of variable total
concentration with three or more components was presented by
Peroni and Dranoff [99], They determined that single-valued/
binary selectivity coefficients could be used to describe Cu ,
Na+, H+ equilibria with strong-acid resins in the concentration
range of 0.01 to 0.10 N. For the experiments performed in our
work, it was expected that the binary separation factors would
be reasonably constant because of the narrow range (0.002 —
0.008 N) of variation of the individual and total concentrations
Column Equilibrium Studies
If the binary separation factors are reasonably constant,
the specialized multicomponent equilibrium theories for ion ex-
change and chromatographic separations may be applied [27, 63,
64, 71, 119]. The relevant mass balances and equilibrium expres-
sions for ion-exchange columns which permit multicomponent
concentration profiles to be predicted from constant separation
factors a constant total solution concentration are:
y ,x.
»i • ^ <14)
i "i
J n = number of components
k = an arbitrary component
4 i (15) i = component number 1
j = component number 2
- = i d6)
Xi.-%-..jb±. «»»
I ajYj 1 «jYj
x aix
y = i = akxi (18)
1 * i *• ,1
ax.
42
-------
and the affinity sequence is
so that a^ > 1; aP > 1, etc.
J K
Klein, Tondeur and Vermeulen [71] have demonstrated that by
using these relationships in conjunction with integral and
differential material balances, the concentration profiles of
each component in either the resin or liquid phase can be
determined in ion-exchange columns under equilibrium conditions.
In representing the column concentration profiles and in
writing the differential balances, the dimensionless throughput
parameter "T" is utlized:
C (V-ve)
T = 1^00 vQ = Throughput (19)
T _ Total meg of ions fed to the column _
Total meq of column ion-exchange capacity
where :
CQ = Constant total solution cone. , meq/&
Q = Resin capacity, meq/ml
e = Column void fraction, dimensionless
V = Feed solution volume, A
v = Resin bed volume, H
Their mathematical development leads to the following rules
governing equilibrium column profiles:
(1) The number of plateau zones is equal to the number
of components in the system.
(2) Between each plateau zone is a transition zone which
may be either adrupt or gradual depending upon whether a bound-
ary is self -sharpening or non-self -sharpening.
(3) The "root," "alphabet," and "slope" reuls [71] may be
utilized to further define the shapes and locations of these
transition zones .
Analytical solutions are presented by Tondeur and Klein
43
-------
[119], Helfferich and Klein [64] and Helfferich [63] for the set
of integral and differential mass balances in the constant
separation factor case for any number of components. The more
general case of constant selectivity coefficients may be solved
by numerical methods [71]. Unfortunately, all of the above
solutions assume a constant total solution concentration (C0=C_)
which is a very good approximation for pure ion-exchange without
neutralization. For example, it applies to strong-base exchange
of Cl for NO3 but not to activated carbon adsorption processes
or to the second bed in a two-bed ion-exchange system where
molecular adsorption or ion exchange with neutralization occurs
causing C_ to approach zero upon continuous contact with the
solid phase. Helfferich [64] has termed this "non-stoichiometric
sorption" and suggests the creation of a dummy species whose
concentration makes up for the difference between the variable
CT and some mathematical constant, e.g., CQ, the total initial
solution concentration in the column feed. How the concept is
applied to the analytical solution of constant separation
factor column equilibria is discussed in detail in Reference
64, pp. 283-298.
If we accept the published selectivity sequence:
sulfate > nitrate > chloride > bicarbonate
as being true for the proposed process of weak-base anion ex-
change with an acidic influent to a non-presaturated bed then
Helfferich's "unique pattern" rules for column profiles [60,
pp. 163-4] may be applied if some further assumptions are made,
viz., (1) that the dummy species created has a lower affinity
than all real species, (2) that' sufficient time has passed
for coherent boundaries (i.e., stable traveling loci of constant
composition) to have developed and (3) that the resin-phase
capacity is constant. In this "unique" case the solid-phase
profile can be represented as
SS SN §C1 SB
where S denotes an abrupt boundary in the resin phase which
44
-------
separates an upstream zone containing the superscripted anion
from downstream zones in which that particular anion is absent,
The profile is, of course, read from left to right. Applying
that rule to the case in point, i.e., nitrate removal from
groundwater in the presence of the competing anions sulfate,
nitrate, and chloride results in the idealized resin phase
concentration profile shown below in Fig. 5.
1.0
a>
a:
c
o
c
a>
o
O
0.0
ysj s°4
**> •
ZONE 1
(Sulfate)
y N03
''i '
V r i
^Cl.l Cl
*N2 N°3
ZONE 2
(Nitrate)
yCI,2 CI
VC,3>
ZONE 3
(Chloride)
VS.4>
ZONE 4
(HCC£)
DISTANCE INTO BED
meq Exchanger/meq Soln. ^- I.C
FIGURE 5
HYPOTHETICAL RESIN PHASE CONCENTRATION PROFILE
y*, . =eq fraction of chloride in zone i
'Cl,i
y_, =eq. fraction of sulfate in zone I
M
N,
= eq. fraction of nitrate in zone I
^
2 = eq. fraction of nitrate in zone 2
>, =^q. fraction of bicarbonate in zone 4
4
45
-------
Figure 6 below is a more simplified presentation of Figure 5.
It illustrates how a mixture of the four typical ground water
anions is partially separated in an exhausted ion exchanger.
The first zone is enriched with the most preferred species,
the second zone with the second-most preferred species and so
on.
Zone I
Zone 2
Zone 3
Zone 4
i
inlet
Sulfate
rich
Nitrate
rich
Chloride
rich
Bicarbonate
rich
L!
outlet
ANION
EXCHANGER
FIGURE 6
Chromatographic Enrichment of Ground
Water Anions in an Exhausted Anion
Exchanger
46
-------
Knowing the shapes of the zones and assuming constant
separation factors, the y.'s of all species may be calculated
assuming that the column is run to nitrate breakthrough. It
was expected that problems would undoubtedly arise from the
non-validity of the simplifying assumptions especially the
conditions of equilibrium and constant capacity which are known
not to be true for the kinetically slow, variable capacity
weak-base resins. Nevertheless a simplified procedure was
developed wliich did closely predict yN the average equivalent
fraction of nitrate on the resin at nitrate breakthrough.
47
-------
SECTION 5
PHASE I: ANION RESIN SELECTIVITY STUDY
OBJECTIVES:
To determine the sulfate/nitrate selectivities of the
commercially available anion resins which might be used in
nitrate removal service on groundwaters with total concentrations
in the range of 0.002 to 0.008 N.
To determine the nitrate/chloride and nitrate/bicarbonate
selectivities of these resins.
To characterize the capacities of all the weak base resins
for HC1, HN03 and H2SO. as a function of pH, i.e. to determine
their titration behavior with these acids.
To establish which resin characteristics are associated
with the various selectivities and, hopefully, to determine
which physically and chemically controllable resin characteris-
tics are causative of the sulfate, nitrate, chloride and bi-
carbonate selectivities.
To provide a complete descriptive data base on the various
resins which may be used to help predict their column perfor-
mances in nitrate removal service.
PROCEDURAL OUTLINE: ANION RESIN SELECTIVITY STUDY
(1) Obtain a representative selection of strong and weak
base anion resin samples. One pint or one liter samples of the
48
-------
following resins were obtained from the four U.S. resin
manufacturers listed in Appendix D.
TABLE 2. ANION RESIN SAMPLES OBTAINED FOR STUDY
(U-M RESIN NUMBER ASSIGNED FOR THIS STUDY)
Weak Base Resins
Strong Base Resins
(1) Amberlite IRA-93
(2) Amberlite IRA-68
(3) Amberlite IR-45
(4) Dowex WGR
(5) Dowex MWA-1
(6) Duolite A-7
(7) Duolite A-340
(8) Duolite ES-368
(9) Duolite ES-561
(10) Duolite ES-374
(11) lonac A-260
(12) lonac AFP-329
(13) lonac A-305
(14) Amberlite IRA-910
(15) Amberlite IRA-400
(16) Amberlite IRA-402
(17) Amberlite IRA-900
(18) Amberlite IRA-410
(19) Dowex SBR-P
(20) Dowex SAR
(21) Dowex SBR
(22) Dowex 11
(23) Duolite A-102-D
(24) Duolite A-101-D
(25) Duolite A-104
(26) lonac A-550
(27) lonac ASB-1
(28) lonac A-641
(29) lonac ASB-2
(30) lonac ASB-1P
(31) lonac A-540
(32) AFP-100
(2) "Condition" each of the resin samples. Six 2" dia.
by 51 high glass columns were used to prepare the resins for
further testing by running them through two acid-base cycles
with backwashes and intermediate and final distilled water
rinses. See Appendix D Procedure Dl for further details.
(3) Convert resins to appropriate ionic forms at 0.002/
0.005 or 0.008 N for determination of selectivities. Samples of
49
-------
each of the resins were converted to the nitrate and chloride
forms prior to determining capacities and establishing iso-
therm behaviors. See Appendix D (Procedures D2 and D6) for
conversion and capacity determination procedures.
(4) For each isotherm point, equilibrate a predetermined,
known weight of one of the various forms (e.g. nitrate form)
of one of the resins with a measured amount of 0.005 N acid (e.
g. 100 ml of H-SOJ and analyze equilibrated supernatant for
the anions of interest (e.g. NCK and SO.~) before calculating
the relevant x. and y.. See Appendix: Procedure D-3 and
calculation E-2.
(5) Construct sulfate/nitrate and chloride/nitrate iso-
therms for the resins. See Appendix A for all the isotherms
plots.
(6) Equilibrate, in a closed columnar system, various
mixtures of HN03 and H^CO^ and analyze the column regenerants
to determine the bicarbonate/nitrate selectivities of all the
weak base resins. See Appendix: procedure D-4 and Figure D4.
(7) Construct H2SO4, HC1 and HNO3 titration curves for
each of the weak base anion resins by equilibrating a known
weight of resin with a measured amount of acid before determin-
ing aqueous phase pH. See Appendix B: Figures Bl thru B12,
and Appendix D: procedure D5.
(8) Calculate approximate pKa's of resins. pKa's were
determined by the method outlined by Helfferich (Ref. 60, pp.
84-88).
(9) Make visual observations and judgements from isotherm
plots before plotting comparison isotherms.
(10) Calculate average separation factors. See discussion
following and Appendix: Calculation E3.
(11) Plot composite isotherms for resins with similar
matrices and functional groups. See Figs. 12 thru 17.
(12) Plot comparison isotherms to illustrate effects of
matrix and functionality. See Figs. 20 thru 23.
(13) Assemble experimental and published data into a data-
50
-------
file for statistical analysis by MIDAS. See Tables 3 and Fl.
(14) Do preliminary correlations, analyses of variance,
scatter plots and regressions.
(15) Create dummy variables to convert matrix and function-
ality into analytical variables.
(16) Correlate analytical variables. See Tables 5, 6, F2
and F6.
(17) Perform analyses of variance to establish significant
t o
influences of categorical variables on selectivities: OLT and
N N
acl. See Tables 9 and 10.
(18) Do linear regression analyses and make scatter plots
of selected stratifications of the analytical variables. See
Figures 7-11 and Fl - F10.
(19) Perform multiple regression analyses and selection of
regression analyses (optimization) to establish the selectivi-
CJ vi
ties, aN and acl, as functions of the independent analytical
variables. See Appendix F for examples of selection of regres-
sion output.
(20) Attempt to explain the statistically significant re-
lationships in terms of accepted physiochemical phenomena. See
"Results of Statistical Analyses" and "Phase I Results Summary".
(21) Summarize with predictive equations and tables the
most important factors determining ot^ and a^,. See equations
37-43 and Tables 17 and 18.
Visual Interpretations of Isotherms
Graphical Representation of Selectivity
Binary sulfate/nitrate isotherms were constructed for all
32 resins and are represented as the upper curves in Figures
A1-A32. For 19 of the 32 resins, nitrate/chloride isotherms are
represented as the lower curves on the same graphs. To avoid
confusion, the reader should Leep in mind that each isotherm
is binary at a total concentration of 0.005 N and that the
exchange taking place is always between nitrate and either chlo-
ride or sulfate. Observe that all the sulfate/nitrate isotherms
51
-------
are favorable for sulf ate (convex) whereas all the chloride/nitrate
isotherms are unfavorable for chloride (concave). This generally
is the expected result and gives rise to the following
selectivity sequence for all resins:
Sulfate > Nitrate > Chloride > Bicarbonate
Although bicarbonate and carbonic acid are not represented
in any of the isotherms, it was determined, as will be discussed
in the following section on bicarbonate/nitrate isotherms, that
all resins showed negligible preference for these species.
Sulfate/Nitrate Isotherms
A rapid visual scanning of all the sulfate/nitrate isotherms
indicates that there is an extreme range of sulfate selectivity.
It appears that the styrene-DVB resins with tertiary amine
functionality (resins lf 5,8 and 12) or with quaternary amine
functionality (resins 14-32) have moderate sulfate preference.
Resins with other than Styrene-DVB matrices (Resins 2, 4, 6, 7,
9, 10, 11, and 13) have high to extremely high sulfate preference
over nitrate as evidenced by the very convex curvature of the
isotherms. One styrene-DVB resin (Resin 3) with polyamine
functionality has a high sulfate selectivity compared to the
other stytene-DVB resins with tertiary and quaternary amine
functionality (resins 14-32). Among the non-styrene-DVB resins
(Resins 2, 4, 6, 7, 9, 10, 11, and 13) one resin appears to
have significantly lower sulfate selectivity than any of the
others. That is Resin No. 2, and it differes from these others
which are polyfunctional in that it is monofunctional (tertiary)
as advertised and as verified by its experimentally determined
titration curve (Fig. B2).
52
-------
Chloride/Nitrate Isotherms
The range of chloride/nitrate selectivity differences among
resins was not nearly so apparent as was the case with sulfate
selectivity. It does appear though that the styrene-DVB resins
especially those with tertiary amine functionality (Resins 1,
5, 8 and 12) have greater preferences for nitrate as indicated
by very concave isotherms than do the others. The epoxy-amine
polyamine resins (Nos. 4 and 7) appear to have the least prefer-
ence for nitrate over chloride, i.e., their isotherms are the
least concave. This, of course, an undesirable situation for
resins in nitrate removal service.
Bicarbonate/Nitrate Isotherms
There are no bicarbonate/nitrate isotherms. Bicarbonate
wasn't measurably taken up as H2CO~. Nine weak-base resins
(Nos. 1-6 and 8-10) and three strong-base resins were chosen
for the initial bicarbonate/nitrate selectivity screening.
Considerable effort was expended developing a dynamic procedure
in which solutions containing various ratios of HCO- /NO_ sodium
salts were decationized in a large (100 ml of resin) cation
column followed by 12 small (2 ml of resin) anion columns each
containing a 1.00 meq. sample of one of the resins; see Bicar-
bonate Selectivity Determination Procedure, Appendix A.
Following exhaustion of the anion resins they were regenerated
with NH^OH or NaOH and the regenerants analyzed for HCO," and
NO .j . As one might have expected at the low solution phase pH's
existing in these studies (2.4 to 3.0), the uncharged H2C03
molecule did not appear to have participated to any significant
extent in ion exchange in either the strong base or weak base
resins. The conclusion then, which is applicable to our proposed
nitrate removal system, is that no significant, net HCO.," removal
can be expected anywhere in a weak base anion column where the
pH is 3.0 or less. That however doesn't completely preclude
53
-------
HCO- removal since the lower reaches of an unexhausted bed
will be near neutral in pH as will the entire bed during its
exposure to the near neutral effluent from an exhausted cation
bed. Furthermore, results of the bicarbonate selectivity
determination indicated that H2C03 and probably HCCU are
catalytic for the eventual removal of the more preferred
species: chloride, nitrate and sulfate in columnar, ion-exchange
processes.
Generalizations from the Observations
High sulfate selectivity is associated with non-styrene-DVB
matrices and polyamine functionality and this should be consid-
ered in choosing either a weak or strong base resin for
nitrate-ion removal in the presence of sulfate. Note that with
these particular resins the avoidance of polyamine functionality
is equivalent to the rejection of resins with mixed, secondary
and tertiary functional groups as those are the major constitu-
ents of polyamine resins. To minimize sulfate selectivity one
would choose a monofunctional styrene-DVB resin of tertiary or
quaternary amine functionality.
The chloride preferences exhibited by these resins also
appear to be much influenced by the matrix type and to a lesser
extent by the functionality. Again styrene-DVB resins, espec-
ially tertiary amines, are the preferred types for nitrate
removal in the presence of the competing ions — chloride and
sulfate.
We had yet to examine the effects of such variables as
capacity, porosity and pKa on the sulfate and chloride prefer-
ences of the resins. Since these in addition to matrix and
functionality are all controllable variables one would ideally
like to quantify their contributions to the dependent variables
aN and acl so as to be able to predict these selectivities for
54
-------
available resins and to control them when making new resins.
This objective resulted in a need to perform a comprehensive
statistical analysis on the data gathered from laboratory ex-
periments and from the resin manufacturers. That analysis is
the subject of the next section.
STATISTICAL ANALYSIS OF RESIN DATA
Objective
The overall objective of the statistical analysis of the
Phase I data was to develop a predictive equation or equations
relating to the dependent variables c*N and acl to a minimum
number of relevant independent variables from the list:
matrix type
functionality
ion-exchange capacity
porosity
pKa
quaternary type (I or II)
Of the seven variables, four are represented by interval data
(measured on a ratio scale) and the remaining three, matrix,
functionality and porosity are categorical in nature.
Straight-forward statistical analysis e.g. multiple linear
regression was not possible because of the combination of cate-
gorical and interval scale variables.
Data Summaries
Twenty-nine of the 32 resins evaluated are listed in Table
3 where they are characterized by particular values of the
seven variables just discussed. Three of the resins tested
were eliminated from the data analysis because they represented
single-case categories of strong-base styrene-DVB resins which
were already over represented. Styrene-DVB resins comprise 16 of
the 29 resins (cases) evaluated statistically.
55
-------
TABLE 3: ANION RESIN CHARACTERISTICS
ui
en
UM Resin Manufacturer's Matrix Functionality
Number Designation
15
17
21
27
32
16
19
22
24
28
30
14
18
20
23
29
1
5
8
12
3
2
10
6
9
11
4
7
13
AMBERLITE IRA-400 STY-DVB Q-l
AMBERLITE IRA-900 STY-DVB Q-l
DOWEX SBR STY-DVB Q-l
IONAC ASB-1 STY-DVB Q-l
IONAC AFP-100 STY-DVB Q-l
AMBERLITE IRA-400 STY-DVB Q-l
DOWEX SBR-P STY-DVB Q-l
DOWEX 11 STY-DVB Q-l
DUOLITE A-101-D STY-DVB Q-l
IONAC A-641 STY-DVB Q-l
IONAC ASB-1 P STY-DVB Q-l
AMBERLITE IRA- 910 - STY-DVB Q-2
AMBERLITE IRA-410 STY-DVB Q-2
DOWEX SAR STY-DVB Q-2
DUOLITE A-102-D STY-DVB Q-2
IONAC ASB-2 STY-DVB Q-2
AMBERLITE IRA-93 STY-DVB TERTIARY
DOWEX MWA-1 STY-DVB TERTIARY
DUOLITE ES-368 STY-DVB TERTIARY
IONAC AFP-329 STY-DVB TERTIARY
AMBERLITE IR-45 STY-DVB POLY
AMBERLITE IRA-68 ACRYLIC-AMINE TERTIARY
DUOLITE ES-374 ACRYLIC-AMINE POLY*
DUOLITE A- 7 PHENOL-HCHO-PA POLY**
DUOLITE ES-561 PHENOL-HCHO-PA POLY
IONAC A-260 ALIPHATIC-AMINE POLY
DOWEX WGR EPOXY-AMINE POLY
DUOLITE A- 340 EPOXY-AMINE POLY
IONAC A-305 EPOXY-AMINE POLY+
meq/ml
Porosity Advertised
Capacity
MICRO
MACRO
MICRO
MICRO
MACRO
ISO
ISO
ISO
ISO
FM
ISO
MACRO
MICRO
MICRO
MICRO
MICRO
MACRO
MACRO
MACRO
MACRO
MICRO
MICRO
MACRO
MACRO
MACRO
MICRO
MICRO
MICRO
MICRO
1.40
1.00
1.40
1.40
1.20
1.25
1.20
1.20
1.30
1.16
1.35
1.00
1.35
1.40
1.40
1.52
1.25
1.10
1.30
1.25
1.90
1.60
3.0
2.4
2.0
1.8
1.0
2.6
3.5
meq/ml
Measured
HC1 Capaci
1.53
1.10
1.66
1.39
1.07
1.16
1.02
1.17
1.32
1.21
1.13
1.31
-
1.50
1.48
1.33
0.98
1.15
1.43
1.26
1.76
1.42
2.59
1.67
1.22
1.81
1.53
2.54
1.51
pKa
ty
>13
>13
>1 3
>13
>13
>13
>13
>13
>13
>13
>13
>13
>13
>13
>13
>13
7.7
7.6
7.8
8.5
7.9
11.1
9.9
7.7
6.8
10.6
7.9
8.7
Average
«S
1.89
1.71
1.89
1.87
1.76
3.09
2.96
3.37
2.59
3.30
2.59
3.26
2.40
3.04
3.26
3.04
3.75
2.67
2.83
3.07
12.7
23.4
94.0
108
109
54.0
137
82.9
108
Averaqe
N
aci
3.41
2.90
-
2.97
3.11
-
-
-
3.30
-
2.85
-
-
-
3.64
4.86
4.43
3.87
4.14
3.89
1.89
3.85
3.35
2.65
2.25
1.99
1.70
-
\
POLY
Q-l
Q-2
ISO
FM
POLY*
POLY**
POLY+
= Polyamine not including quaternary amine
= Quarternary Amine - Type 1
= Quarternary Amine - Type 2
= Isoporosity or "Improved Porosity"
Fixed Macropore (MANUFACTURER'S TERMINOLOGY)
= Advertised as tertiary amine but titrates as
= Advertised as secondary amine but titrates as
= Polyamine including quaternary amine
polyamine
polyamine
-------
Table Fl (Appendix) is a listing of the computer datafile
derived from the resin data in Table 3. Missing data is coded -
0.0. Representative data are coded as follows:
Variable^ 1, VI, (dependent, interval scale, dimensionless)
Average sulfate/nitrate separation factor: aN
Range: 1.71 to 137
Variable 2, V2, (dependent, interval scale, dimensionless)
N
Average nitrate/chloride separation factor: acl
Range: 1.7 to 4.86
Variable 3, V3, (independent/ interval scale, meg/ml)
Measured HC1 capacity
Range: 0.98 to 2.59
Variable 4, V4, (independent, inteval scale, dimensionless)
pKa for HC1
Range: 6.8 to 13
All strong base resins were assumed to have pKa = 13
Variable 5, V5, (independent, categorical)
Functionality
polyamine =2; (8 cases)
tertiary amine =3; (5 cases)
quaternary amine = 4; (16 cases)
Variable 6, V6 (independent, categorical)
Matrix Type
styrene - DVB =1; (21 cases)
acrylic amine =2; (2 cases)
phenol - HCHO =5; (3 cases)
aliphatic amine = 6;' (1 case)
57
-------
Variable 7, V7 (independent/ categorical)
Porosity
Microporous or Gel = 1, (13 cases)
Macroporous or macroreticular =2, (10 cases)
Isoporous or "improved porosity" = 3, (6 cases)
Variable 8, V8 (independent, categorical)
Nitrogen in polymer backbone or out-of-backbone
nitrogen in = 1, (8 cases)
nitrogen out = 2, (21 cases)
Variable 9, V9 (dependent/ interval-scale, dimensionless)
Loge of c*S
Range = 0.57 to 4.92
Variable 10, V10 (dependent , interval-scale dimensionless)
Loge of a^
Range = 0.53 to 1.58
Dummy Variable 11 , Vll, (independent, interval-scale, di-
mensionless)
Relative crystal ionic radius of functional group
Secondary amine = 2.00
Tertiary amine = 2.19
Quaternary amine = 2.36
Dummy Variable 12 , V12, (independent/ interval-scale , di
mensionless)
Nitrogen position in resin
(related to distance of separation of charged sites)
nitrogen out of polymer backbone = 0.00 (far away)
nitrogen in polymer backbone = 1.00 (close)
Variable 13, V13 (dependent, categorical)
Quaternary functional group type
Type 1=1 (11 cases)
Type 2 = 2 (5 cases)
58
-------
Variable 14, V14 (independent, interval-scale, dimension-
less)
Relative degree of crosslinking
Isoporous resins = 0.5
Microporous resins = 1.0
Macroporous resins = 2.0
MIDAS; Michigan Interactive Data Analysis System
Actual computations of the statistics were accomplished
using the extensive UM computing facilities (Michigan Terminal
System) with the aid of the MIDAS system of data anlyses and
statistical computing programs developed by the UM Statistical
Research Lab. Documentation for the MIDAS system is presented
in Reference [48] while interpretation of statistical techniques
are given in Reference [1241.
The particularly desirable features of the system are its
ability to handle both categorical and analytical (interval-
scale) variables/ its intuitive syntax, its capability for
partitioning the dataset, and its excellent documentation.
»
The Dependent Variables of Interest; aN, a_,
Separation Factor vs Selectivity Coefficient
In an earlier discussion it was pointed out that the sep-
aration factor, cu, differes from the selectivity coefficient
K. when ions of dissimilar valence are exchanged. Although
the selectivity coefficient is theoretically more satisfying,
it's magnitude gives no simple indication of the preference
which a given resin has for the ions of interest at an estab-
lished total concentration e.g. 0.005N (250 ppm CaCG>3). The
binary separation factor, on the other hand, being simply the
ratio of the distributions of ions between phases given a clear
intuition of the preference which the resin has between the ions
of interest.
59
-------
g
For all the resins tested, the o^'s are greater than 1.0
indicating a clear preference for SO. ions over NO, ions.
Cl
Similarly, all the resins exhibited o^ ' s of less than 1.0 in-
dicating a preference of NO~ over Cl~. Additional arguments
for the choice of separation factors over selectivity coeffici-
ents are presented in Appendix E: Calculation E3.
The Calculation of Average Separation Factor
It may be observed graphically, that all of the resins having
low sulfate selectivity (tertiary and quaternary styrene-DVB
resins) also have modest inflection points in their isotherms.
Hence, the simplest mathematical model (without theoretical
basis) which could be used to describe the curve would involve
a cubic equation, again giving rise to much more complexity
and to parameters like the selectivity coefficient which give no
intuitive indication of the actual preference the resin has for
one ion over another. Having chosen to use the separation factor
to describe each isotherm the task remaining was to arrive at
a satisfactory means of determining the best, single factor
describing the curve. Using a simple averaging technique
where the mean separation factor determined at three or more
points on an isotherm at say X = 0.25, 0.50, and 0.75, was
rejected on the basis that it utilized a minimum amount of the
data available and that the selection of points would be
arbitrary.
Linear Regression Technique for Average a.
A least-squares regression technique was attempted on
several representative isotherms including the styrene-DVB
resins numbered 3 and 8. (Figures A33 and A34) The constant
separation factor description of an isotherm may be linearized
for the statistical regression analysis as follows:
60
-------
a1 = yi Xj (14)
x. = 1 - x±
Y- i x.
•* i = a. i
-x. (20)
-1 i
Using the five or six experimentally determined points (Y.Q/
x.,; y-2/ x>2; etc) plots of j^—• vs. y^— were made for
several of the isotherms. The calculated, least squares,
linear regression line was drawn through the data, and the a.
determined from its slope. This statistically determined a.
was then used to construct the "Regression" isotherm on the
usual coordinates to determine how it fit the original data
points. The fit was not at all good for the sulfate/nitrate
isotherms and only a fair approximation for the chloride/nitrate
isotherms (see Figures A33 and A34). The reason for the poor
fit is obvious. During the linearization procedure, some ex-
treme values were created which almost entirely determined the
slope of the regression line. The least squares, linear regres-
sion technique produces a "best fit" of the linearized y^— vs
x
y— equation but certainly not for the original y vs x relation-
ship. Note that the "Regression" isotherms in Figures A33 and
A34 are nearly perfect fits of the data in the range of X = 0.8
to X = 1.0 which is just what one would expect since points
in this range are responsible for the extreme values created in
the linearization process.
The Ratio of Areas Technique for Average a.
The example isotherm (Figure 4) illustrates that the sep-
aration factor can be represented as the ratio of rectangular
area 1, [equal to y(l-x)] below the isotherm, to rectangular
61
-------
area II [equal to x(l-y)] above the isotherm. Mathematically
it can be shown (see Appendix E) that the separation factor is
related to the ratio of the entire area below the isotherm to
the entire area above it by the following relationship.
entire area under isotherm
R
a entire area above isotherm (21)
(a2 - a - alna)
R = (a - I)2 (22)
^ _ (a2 -a - alna)
(a - I)2
To estimate the best fit separation factor by the ratio of areas
technique developed here, the areas were measured by planimeter,
the ratio Ra determined and a calculated by trial and error solu-
tion of equation 22. The calculated a's, referred to as the
average separation factors were then used to construct the best
fit, constant separation factor isotherms as shown in Figures
A35 and A36. Clearly, the ratio of areas technique produces
a much better fit of the original data than does the linearized,
least-squares method. Thus, average separation factors so cal-
culated were used to represent the sulfate/nitrate and nitrate/
chloride selectivities of the 29 resin analyzed statistically.
Note that in Figures A33 and A34/ the chloride/nitrate experimen-
tal data is very well represented by a constant separation factor
isotherm which is as expected for univalent-univalent ion ex-
change. The sulfate/nitrate isotherm with an inflection,
Figure A34, is only modestly well fitted by a constant a while
isotherms without inflections of the type shown in Figure A33 are
well represented by a constant a.
62
-------
Resin Characterization by Independent Variables
Resin Capacity—
This is the total wet-volume exchange capacity determined
for HC1 in 2N solution. The capacity determination procedure
is given in Appendix D. For all the strong-base resins,
measured capacities were in rather good agreement with the ad-
vertised values. Such was not the case with all the weak-base
resins especially those with polyamine functionality. One
resin (No. 4), Dowex WGR had a significantly higher capacity
than advertised and four others (No.'s 6, 9, 10 and 13),
Duolites A-7, ES-361 and ES-374 and lonac A-305 had significantly
lower capacities than advertised. With these weak-base resins,
capacity was, as expected, a function of the type of acid and
the pH of equilibration as evidenced by the titration curves
(Figs. B1-B12) where it is seen that H^SO. yields the highest
capacities, HC1 the lowest with HNO^ being intermediate be-
tween the two. This also is generally the order of preference
of the anions of those acids by the resins.
Resin pKa's—
Weak base resin pKa's were determined for HC1, HNO, and
H2SO4 bv the titration technique described in the Appendix:
procedure D5. These numbers are not particularly reliable
because of obvious difficulties encountered in determination
of the end points of the inflectionless titration curves shown
in Figures 84, 86, 87, 89, BIO and Bll. In these instances
the measured, total, wet-volume, HC1 capacity was used as the
endpoint. These pKa's should be considered as relative values
only because of the nature of the assumptions used in the
derivations. [Ref. 60, p. 84],
Strong-base resins pKa's were not determined experimentally
as they were all expected to be nearly equal at a value > 13
[Ref. 60, p. 86]. For the sole purposes of statistical analysis
63
-------
the estimated value of pKa = 13 was assigned to all strong-base
anion resins.
Amine Functionality—
A given resin was either primary, secondary, tertiary,
quarternary or a combination of these labelled polyamine which
as it turned out comprised mainly secondary amines with lesser
amounts of the primary and tertiary varieties.
Evaluations of the titration curves (Figs. B1-B12) in
combination with the manufacturers stated description of func-
tionality resulted in the observation that the only monofunc-
tional weak-base resins (having essentially a single type of
functional group) were those advertised to be tertiary amines
(resins 1, 2, 5, 8 & 12) with the exception of Duolite ES-374,
advertised to be tertiary but which titrated as a polyamine
type. Consequently, all the remaining weak-base resins were
labelled as polyamines and characterized as being basically
secondary amines with some primary and tertiary groups present.
(Resins 3, 4, 6, 7, 9, 10 and 11).
Matrix Type—
This is a description of the organic polymer backbone of
the resin. According to written and verbal information supplied
by the manufacturers,(data sheet and personal communcations),
five distinct polymers were represented in this study: polysty-
rene crosslinked with divinylbenzene (STY-DVB), polyacrylic-acid
polyamine condensation polymers (acrylic-amine), phenol-formalde-
hyde-polyamine condensation polymers (phenol-HCHO-PA), epichloro-
hydrin-polyamine condensation polymers (epoxy-amine), and an ace-
tone-formaldehyde polyamine condensation polymer (aliphatic-
amine).
It should be noted that with the exception of the STY-DVB
resins, the nitrogen-bearing functional groups are incorporated
(polymerized) into the backbone where they are probably separated
64
-------
there by no more than the distance of separation of the nitrogen
atoms in the amine monomers used in polymerization. The impor-
tance of this proximity of positively charged nitrogen atoms
is shown later when it is hypothesized as being the primary
cause of sulfate selectivity as N in the matrix always gives
g
the rise to high values of a.,.
Degree of Cross Linking—
Cross linking is the achievement of a three dimensional
polymer network by the cross as opposed to linear bridging of
polymer chains through chemical bonding. With polystyrene based,
cation resins this cross linking is easily quantified as the %
of divinybenzene (DVB) in the matrix. High degress of cross
linking (e.g. 12% DVB) produce tight structures favoring smaller
ions, are hard, mechanically and chemically stable and kineti-
cally slow. The opposite is true for low degrees of cross
linking (e.g. 4% DVB). Characterization of anion resins by
degree of cross linking is most difficult whether they contain
DVB or not (Dorfner p. 33, R. Anderson Personal Communication).
Styrene-DVB anion resins are capable of methylene bridging
between benzene rings as a result of chloromethylation prior
to the required amination step. So the % DVB doesn't truly
characterize the degree of crosslinking for these resins. The
non-styrene based resins don't even contain DVB. Their cross
linking takes place through the nitrogen in the matrix. Since
this is a study of anion resins for which the degree of cross
linking has not been well characterized, this possible variable
could not be readily included in the statistical analysis.
However, a category akin to the degree of cross linking is the
resin porosity for which data do exist, so it was included for
analysis.
65
-------
Porosity—
Unfortunately resin manufacturers use a variety of labels
to categorize resins as to porosity - a measure of the degree
of openness of the polymer matrix related to the type and degree
of crosslinking.
In this report, the distinction between gel and macroreti-
cular resins used elsewhere is adhered to [Ref. 40, p. 37; Ref.
60, p. 60]. Gel resins are microporous having an apparent
o
porosity of atomic dimensions (10-20A) whereas macroreticular
or macroporous resins, whose beads comprise aggregates of gel
resins, have internal voids with dimensions far exceeding atomic
distances of separation (up to several hundred angstroms). Still
a third type of porosity is available among the styrene-DVB
strong-base resins, i.e., isoporous resins. These are also
loosely referred to as polystyrene resins with a "higher degree
of porosity" than gel resins (lonac and Amberlite data sheets)
or as having "porous structure" (Duolite and Dow Data Sheets)
or simply as being "porose" (Boari p. 153). There are then
three types of porosity represented here: microporous, macro-
porous and isoporous. For polystyrene-based resins the degree
of cross linking is related to these classifications as follows:
Gel: Polymerization step with 6-8% DVB then chloromethylation -
little secondary crosslinking due to methylene bridging.
Product is generally transparent.
Macroporous: Polymerization with high degree of DVB cross
linking before chloromethylation and aggregate
bead formation. Product is opaque.
Isoporous: Polymerization step with very low degree (0.5 to
2.0%) of DVB crosslinking followed by chloromethy-
lation and significant degree of methylene bridging.
Product is transparent and more porous than gel
with lower degree of effective crosslinking.
66
-------
Note: During the statistical analysis of the data an interval
scale variable (XLINKING, V 14) relating porosity to the
estimated relative degree of crosslinking was created
in an attempt to improve the prediction of nitrate/
chloride selectivity.
Nitrogen Position: Nitrogen N2POSITN--
This is a straightforward classification based on the ob-
servation that the matrices: epoxy-amine, acrylic-amine, phenol-
formaldehyde-polyamine and aliphatic-amine invariably gave rise
to high sulfate selectivity regardless of functionality. The
common characteristic among these is the presence of the amines
during the polymerization step giving rise to nitrogen linking
and crosslinking in the polymer making it an inherent part of
the continuous structure. This is contrasted to the case with
polystyrene-based resins where amination takes place after
polymerization and chloromethylation yielding a matrix containing
the active nitrogen atom strictly as a part of a pendant, amine
molecule connected through a methylene group to the continuous
cross-linked polymer matrix.
Nitrogen in the matrix e.g. in a phenol-formaldehyde-poly-
amine resin is symbolically represented as:
jpCH2-NH-C2H4-NH-C2H4-N-CH2
ChU
Nitrogen in the Matrix
67
-------
whereas pendant nitrogen, out of the polystyrene matrix is
represented as:
-CH-
X^^v >^N
I CH3
CH2-N-CH3
C H2~ CH ~CHp
I t
-CH2-CH CH-CH2-
CH3(Tjl (Tj CH3
CH3-N—CH2 CH2-N-CH3
Nitrogen put of the Matrix (Pendant)
For reasons to be discussed later, these differences appeared to
greatly affect sulfate selectivity.
Size of Functional Group: RSIZE—
It has been reported that the nature of the functional group
especially its size and charge density have a significant effect
on anion selectivity [60, 21, 40, 39, 75, 78, 99, 113 and 129].
Boari, Liberti et. al.[13] have recently given special attention
to the effect of the functional group on the sulfate/chloride
selectivity of strong and weak-base anion exchangers for use in
removing slulfate from sea water prior to multi-stage, flash
c
evaporation. In their work a_,, was found to increase as the
size of the functional group decreased. The same physicochemical
c
effect is expected to have a similar influence on the OL, separa-
tion factor. 1o accomodate the expected influence of the func-
tional group in the anticipated predictive equation relating
a to the independent variables, relative values of group size
were assigned to the functional groups as follows:
68
-------
TABLE 4 [13]: SIZES OF FUNCTIONAL GROUPS
A°
Functionality Crystal Ionic Radius Relative Ionic Radius
primary
- secondary
tertiary
quaternary
-
2.97
3.27
3.49
-
2.00
2.19
2.36
Should the influence of size predominantly control the rela-
o
tionship between c*N and functionality, the newly created, in-
terval-scale variable, relative ionic size (RSIZE) should be
statistically correlated (negatively) to the aulfate/nitrate
C
selectivity (Ina ). Note that there is no need to assign a
relative size to primary amines as they are grouped with poly-
amines which are scored as secondary amines where applicable.
Quaternary-Amine Type: TYPE
It was not anticipated that the type of functional group
C TV1
would have a significant influence on either a^ or at,, .
However, examination of the respective isotherms resulted
in the tentative conclusion that the quaternary type does
measurable affect ajj but not o£ . Recall that the different
types are:
rr CH3 C|-
1+ Cl \+* <~'
•CH2-N-CH3 -CH2-N-C2H4OH
CH3 CH3
Type 1 Type 2
and that type 1 being a stronger base is harder to regenerate
(with alkali). For groundwater deionization the choice between
69
-------
types 1 and 2 would be made on the basis of ionic preferences
and ease of regeneration not on the advertised fact that type
1 is more resistant to oxidation or that type 2 offers greater
thermal resistance - these latter considerations not being
relevant.
Results of the Statistical Analysis
Overview—
A rather lengthy and complicated statistical analysis pro-
vided a hopefully unbiased look at the significant, insignifi-
cant and questionable relationships among the variables.
Quantification of these relationships followed by an assement
of level of significance was then accomplished. Given that a
test statistic had been calculated, e.g. the "F" statistic,
the level of significance which was assigned to it was very
much a function of the sample size with small samples obviously
requiring large values of the test statistics to be considered
significant at the usually accepted levels of 0.05 to 0.01.
The strength of this particular set of data appears to lie in
the sensitivity of the dependent variables especially a^ (and
S
In aN) to changes in the independent variables. Its weakness
lies in the small sizes of some of the samples used in the
various tests. It will be shown however that after having
both of these facto.rs into consideration, some very significant
relationships were developed.
Recall that the objective of statistical analysis was to
develop predictive equations relating aN and a-,, to capacity,
matrix, functionality, pKa, porosity and quaternary type.
70
-------
Statistic a1 Corre1at i ons
The Correlation Matrix: All Resins Considered—
Table F4 (Appendix) is the correlation matrix for all the
interval scale variables (1, 2, 3, 4, 9, 10, 11, 12) with all
resins (cases 1-32) considered. The matrix is a tabulation
of the Pearson, product-moment correlation coefficients (r)
each of which described the strength of the linear relationship
between a pair of variables. Pearson's "r" is dimensionless,
and ranges from -1.0 to +1.0 with these limits denoting perfect,
linear, negative or positive relationships respectively. The
2
square of the correlation coefficient (r ) known as the "co-
efficient of determination" can be interpreted as being that
fraction of the total, variability in one of the variables
which can be explained by the least squares regression line
relating it to the second variable. For this particular matrix
only 19 of the 32 resins (12 weak base and 7 strong base) were
considered i.e. only those having complete data for all the
variables considered. For a relationship to be considered
significant at the .05 and .01 levels, correlation coefficients
of +.4555 and +.5751 respectively are required. Table 6
below summarizes the non-trival correlations listed in Table F4.
Both a. and In cu were included in Table 6 to show (1) that In
i J J
a. produces higher correlations and that (2) the use of either
In a. or a"!" leads to essentially the same conclusions which are
that sulfate/nitrate selectivity is influenced by:
Matrix > Functionality > Capacity
and that chloride/nitrate selectivity is influenced by:
Matrix and Crosslinking » All other variables
Maximum nitrate selectivity then is favored by:
(1) Nitrogen out of the matrix (Polystyrene resins with a
relatively greater distance between charged sites)
(2) Quaternary and tertiary amine functionality
(3) Low capactiy
Finally, examination of the last two entries in Table 6 reveals
71
-------
that chloride selectivity is unrelated to functionality.
Clearly then matrix is the most important determinant of overall
nitrate selectivity with respect to sulfate, chloride and bi-
carbonate .
The Correlation Matrix: Weak Base Resins—
Table F2 (Appendix) is the correlation coefficient matrix
for the weak-base resins (cases 1-13) considered as a group
separate from strong-base resins. The non-trivial correlations
contained in that matrix are summarized in Table 5 below.
Compared to the correlations among all resins (Table F4), fewer
cases are represented here, consequently correlation coeffi-
cients of +0.5760 and + 0.7079 are required for significance.
At the .05 and .01 levels respectively.
TABLE 5. MEANINGFUL CORRELATIONS:
WEAK BASE RESINS ONLY
Variables Considered
r 100 r
Correlation % Variation
Coefficient Explained
g
In a with N Position
g
In a with "R" Group Size
N
In a_, with N Position
In a^.. with Relative Crosslinking
In aM with In a_,.
N Cl
"R" Group Size with N Position
Capacity with "R" Group Size
In aN with Capacity
.93*
-.87*
-.78*
.75*
-.67
-.66
-.63
.57
86
76
61
56
45
44
40
32
* = Significant at the .01 level
72
-------
TABLE 6
MEANINGFUL CORRELATIONS: WEAK AND STRONG-BASE RESINS
Variables Considered
c
In OL~T with N Position
N
c
a._ with N Position
N
In af. with "R" Group Size
N
ajj with "R" Group Size
"R" Group Size with N Position
In a*J with N Position
aJL with N Position
In ajj with Capacity
s
a with Capacity
Capacity with "R" Group Size
Capacity with N Position
C! \T
In aN with In acl
a., with a«.
N Cl
g
In a with pK
IN a
ajj with pK&
In a^ with "R" Group Size
ou- with "R" Group Size
r
Correlation
Coefficient
.95*
.88*
-.88*
-.79*
.77*
-.65*
-.62*
.63*
.53*
-.61*
.60*
-.53
-.46
-.53
-.53
.25
.20
100 r2
% Variation
Explained
90
77
77
62
59
42
38
40
28
37
36
28
21
28
28
6
4
* = Significant at the .01 level
73
-------
The Correlation Matrix: Strong Base Resins
Considering only the strong base resins as a group led to
the elimination of three possible dependent variables from
consideration: pKa, NPOSITIN and RSIZE. All have the same
pKa (13), all are quaternary (RSIZE= Const.) and all are poly-
styrene (N out of matrix). The remaining possible correlations
are among In
-------
divisions of some independent, categorical variable. Consider,
for example, the effect of matrix on sulfate/nitrate selectiv-
o
ity (In aN) where there were five categories, each correspond-
ing to one of the five matrices. The program computed the mean
g
In OL. of each category and a grand mean considering all the
s
values of In a... The variance represented by the -mean sum of
squared deviations within each category was then compared to
the variance between categories by taking the ratio:
_ _ Mean sum of squares between categories (23)
Mean sum of square within categories
For this particular example, the higher the F ratio, the more
significant was the effect of matrix on selectivity as compared
to that expected from random statistical variations.
The null hypothesis here (Ho) was that the mean In a^
was the same for all categories. For F » 1.0 we tended to
reject the null hypothesis, and the corresponding level of
significance (SIGNIF) attained (a function of the number of
cases and the number of categories) was the probability of
being wrong when making the decision. The "% variation among"
statistic is the variance explainable due to the categoriza-
tion — matrix in the example.
Tables 9 and 10, following, summarize the effects of the
important categorical variables on sulfate/nitrate and nitrate/
chloride selectivity. An unexpected result contained therein
is that porosity somehow influences the nitrate/chloride pre-
ferences of weak—base anion resins (WBA).
Effect of Matrix and Functionality on Selectivity: ANOVA
Technique
The sulfate/nitrate selectivity of all resins as a group,
and of weak-base anions resins as a group, clearly relates to
both matrix and functionality (Table 9, A-H) as indicated by the
75
-------
extreme values of the F statistic (F»1.0). Generally, the F
statistic is not as high for WBA resins compared to all resins
(e.g., compare A&B, C&D, E&F). This appears to be due to the
c
reduction in the range of In a values or to the fewer func-
tional groups considered when looking at only WBA resins com-
pared to all resins. That the matrix categorization based on
nitrogen-in-or-out of the matrix is useful is borne out by
comparison of F statistics (A&C, B&D) where the nitrogen in-or-
out classification yields higher, more significant results than
does the five-matrix categorization. It will be shown later
in the discussion of the regression equations that this dicho-
tomized, matrix variable provided a simple and direct means for
the matrix effect to be included in the prediction equations for
selectivity.
Nitrate/chloride selectivity (Table 10; a-d) is influenced
by matrix more than by any other single variable including
functionality (Table 9; e,f) which is nearly inconsequential in
N
explaining variations in In ou,. Obviously this means that
matrix is the single, most-important variable for predicting
overall nitrate selectivity with respect to both sulfate and
chloride — the problem at hand.
The combination, matrix * functionality, produces categor-
ies corresponding to all combinations of these variables, e.g.:
STY-DVB-tertiary amine, STY-DVB-polyamine, etc. Sulfate/
nitrate selectivity would appear to be almost completely ex-
plained by these combinations (Table 9; G,H) with 98.9% and
98.5% variation among categories for all resins and for WBA
resins respectively. For nitrate/chloride selectivity, such is
not quite the case; the combination, matrix * functionality, is
an improvement over matrix alone but not nearly so much as in
the former case.
These ANOVA findings based on the original categorical
76
-------
variables are completely supportive of those derived from
dummy variable analysis by correlation, regression and selec-
tion of regression. For that reason they have been included;
also, they make obvious some previously obscure relationships
between porosity and nitrate/chloride selectivity.
Effect of Porosity on Selectivity: ANOVA Technique
Categorizing all resins according to porosity and comparing
G
mean In aN's indicated no real differences due to porosity
(Table 9; I). However, from insight gained during visual in-
spection of the isotherms, the sulfate/nitrate selectivity of
Type I, strong-base anion (SBA) resins did appear to be a
function of whether a given resin was isoporous or not-isoporous
(i.e., gel or macroeticular). ANOVA L, Table 9 corroborated
this apparent relationship with an F statistic of 76.5. Type
I, isoporous SBA resins have measurably higher sulfate selec-
^
tivity (average a., = 2.97) than do Type I gel or macroporous
• §
resins (average OL, = 1.82). This sort of porosity effect was
not noted however with nitrate/chloride selectivity and Type I,
isoporous SBA resins. In fact, porosity seemed to have no
effect at all on the In acl of SBA resins in general as docu-
mented in Table 9; entries K & L.
Although it was not discernible during inspection of the
chloride/nitrate isotherms (Figures A1-A32), porosity accounted
for 37.8% of the variance in In a_,, among all resins and 66.5%
M
of the variance in In'a", for WBA resins (See Table 10; i & j).
Overall, the relationship appears to be significant only for
WBA resins since we have just seen that porosity doesn't account
for any variability among the SBA resins. At first it was
thought that this was a secondary effect - porosity being some-
how highly correlated with another relevant variable like
matrix. TO check this, porosity was converted to a dummy vari-
able (XLINKING) which is the estimated, relative degree of
cross linking.
77
-------
TABLE 7. POROSITY RELATED TO RELATIVE DEGREE OF CROSSLINKING
Relative
Estimated % Degree of
Porosity Crosslinking Crosslinking
Isoporous
Microporous
Macroporous
3
6
12
.5
1.0
2.0
II Till
In the WBA resin correlation matrix (Appendix:
Table F2) XLINKING is only modestly correlated (r = .27) with
the important variables: nitrogen position (N2POSITM), and "R
group size (RSIZE) with r's of 0.37 to -0.37 respectively.
This lack of significant correlation between XLINKING and
the other dependent variables was an encouraging sign that it's
presence would add reliability to the predictive equation for
nitrate/chloride selectivity. Also, the case for porosity
being a determinant of nitrate/chloride selectivity among WBA
resins was strengthened but difficult to explain. Close scru-
N
tiny of the average a_,, data in Table 10 indicates that the
porosity-selectivity relationship exists only for resins with
nitrogen in the matrix; no such relationship exists among the
polystyrene resins.
Effect of Quaternary Type: ANOVA Technique
As indicated by the sulfate/nitrate isotherm in Figures
15 and 17, the quaternary type does significantly influence
sulfate selectivity; this is verified by ANOVA M, Table 9
(F=67). Table 8 below summarizes these effects for strong
base resins.
78
-------
TABLE 8. EFFECTS OF POROSITY AND TYPE
ON SULFATE/NITRATE SELECTIVITY
g
Resin Average a
Type I, SBA, Gel and MR 1.82
Type II, SBA, Gel and MR 2.98
Type I, SBA, Isoporous 2.97
Quaternary type doesn't influence nitrate/chloride selec-
tivity at all as verified by ANOVA M, Table 9 (F = 0.147).
Regression Equations and Scatter Plots
Simple, Linear Regression Analysis and Plotting—
Here an attempt has been made to predict the value of
• Q M
either In a or In a .. knowing the value of one of the follow-
ing interval-scale variables: CAPACITY, pKa, RSIZE or N2POSITN.
This was accomplished using the linear least-squares regression
technique the results of which were plotted on the scatter plots
to give a feel for the degree to which the line actually fit or
didn't fit the data since correlation coefficients (r's) can be
very misleading. Usually, high correlation coefficients (e.g.,
•90 or .95) suggest mental pictures of better curve fits than
actually exist.
C "NT
Effect of Capacity and Nitrogen Position on aN and acl—
Capacity is expected to influence the preference of an ion
exchanger for multivalent ions (e.g. SO,") as compared to mono-
valent ions (e.g. NOZ): "As a rule the ion-exchanger prefers
the counter ion of higher valence .... The preference increases
/
with dilution of the solution and is strongest with ion ex-
changers of high internal molality" [60]. Ames [4, 5] attempted
79
-------
TABLE 9
ANALYSIS OF VARIANCE;
VARIABLES EXPLAINING SULFATE/NITRATE SELECTIVITY; In
ANOVA
DESIG,
A
B
C
D
E
F
G
H
I
L
M
CASES
CONSIDERED
All Resins
WBA Resins
All Resins
WBA Resins
All Resins
WBA Resins
All Resins
WBA Resins
All Resins
Type I
SBA Resins
Gel & MR
SBA Resins
STRATIFICATION
(CATEGORIES EXAMINED)
Matrix
(STY-DVB) , (Acrylic) , (Phenolic)
(Epoxy) , (Aliphatic)
Matrix
(STY-DVB) . (Acrylic) , (Phenolic)
(Epoxy) , (Aliphatic)
Matrix
(Nitrogen in) , (Nitrogen out)
Matrix
(Nitrogen in) , (Nitrogen out)
Functionality
(Poly), (Tertiary), (Quat.)
Functionality
(Poly) , (Tertiary)
Matrix * Functionality
(All Combinations of Matrix
and Functionality)
Matrix * Functionality
(All Combinations of Matrix
and Functionality)
Porosity
(Micro) , (Macro) , (Iso)
Porosity
(Gel or MR) , (Iso)
Type
(I, Gel or MR), (II, Gel or MR)
F
STAT.
86.2
20.5
304
76.3
95.2
35.1
179
162
1.24
76.5
67.0
LEVEL
OF
SIGNIF.
.0000
.0003
.0000
.0000
.0000
.0001
.0000
.0000
.3036
.0000
.0000
%
VARIATION
AMONG
CATEGORIES
96.3
89.0
96.3
92.4
91.7
84.7
98.5
98.9
2.6
93.3
93.0
80
-------
TABLE 10
ANALYSIS "OF VARIANCE;
VARIABLES EXPLAINING NITRATE/CHLORIDE SELECTIVITY: In a
N
Cl
ANOVA
DESIG.
a
b
c
d
e
f
g
h
i
j
k
1
m
CASES
CONSIDERED
All Resins
WBA Resins
All Resins
WBA Resins
All Resins
WBA Resins
All Resins
WBA Resins
All Resins
WBA Resins
SBA Resins
Type I
SBA Resins
Gel & MR
SBA Resina
STRATIFICATION
(CATEGORIES EXAMINED)
Matrix
(STY-DVB) , (Acrylic) , (Phenolic)
(Epoxy) , (Aliphatic)
Matrix
(STY-DVB), (Acrylic), (Phenolic)
(Epoxy) , (Aliphatic)
Matrix
(Nitrogen in) , (Nitrogen out)
Matrix
(Nitrogen in) , (Nitrogen out)
Functionality
(Poly), (Tertiary), (Quat.)
Functionality
(Poly) , (Tertiary)
Matrix * Functionality
(All Combinations of Matrix
and Functionality)
Matrix * Functionality
(All Combinations of Matrix
and Functionality)
Porosity
(Micro) , (Macro) , (Iso)
Porosity
(Micro) , (Marco)
Porosity
(Micro), (Macro), (Iso)
Porosity
(Gel or Macro) , (Iso)
Type
(I, Gel or MR), (II, Gel or MR)
F
STAT.
5.18
6.18
12.6
15.1
1.77
2.26
18.9
17.4
4.32
12.6
.211
.362
.147
LEVEL
OF
SIGNIF.
.0089
.0188
.0025
.0030
.2000
.1600
.0000
.0033
.0316
.0053
.82
.60
.73
%
VARIATION
AMONG
CATEGORIES
60.9
70.1
56.7
70.7
11.0
17.8
89.3
91.1
37.8
i
66.5
0.0
0.0
0.0
81
-------
to explain the preference of the zeolite, clinoptilolite for
univalent ions like NH. by observing that bivalent ions could
not remain stable in ion exchangers where the fixed charge sites
were relatively far apart. This distance of separation may or
may not be related to ion exchange capacity. In synthetic,
organic, ion-exchange polymers like corsslinked polystyrene,
one would expect the distance of separation between charged
sites to be a function of capacity which would in turn be posi-
tively related to sulfate/nitrate selectivity. This distance
between charged nitrogen atoms is expected to be randomly dis-
tributed with a mean dependent on the total number of sites
per unit volume. Such is not the case with resins made from
polyamine monomers.like diethylene-triamine when the mean
distance of separation between nitrogen atoms (ion-exchange
sites) in the polymer is expected to be highly correlated with
the original separation distance in the monomer. Consder for
example, a polyacrylic polyamine resin made from acrylic acid
and diethylene-triamine [60].
NH -CH^-CH -NH-CH0-CH0-NH0
*/ * * * I
diethylenetriamine
-
.f} • • •
c=o
I
HN-CH2-CH2-NH-CH2-CH2-NH
, ,. , . c=o
polyacrylic polyamine resin ,
—CH—CH~—...
Note that, symbolically at least, this particular amine monomer
remains relatively unchanged as it provides crosslinking
between acrylic acid chains; the active-nitrogen atoms remain
separated by two methylene groups. For a similar effect, see
the structure of phenol-formaldehyde-polyamine resins in the
82
-------
section on "Structure of Ion Exchange Resins."
The experimental results did prove that sulfate selectiv-
c
ity, as measured by In a^., was predictable with modest reli-
ability from capacity data for resins (r = .54, Figure F2).
However, capacity cannot be used to predict sulfate selectivity
for strong base resins (r = .12, Figure F3). The important
conclusion to note however is that nitrogen position (in or out
g
of the matrix) is a much better prediction of In aN (r = 0.96,
Figure 7) than is capacity, or any other variable for that
matter. The relevant, simple regression equations for all
resins are:
In a?T = 2.48 CAPACITY - 1.60 (Figure Fl) (25)
N
In aj? = 3.34 N2POSITN + 1.05 (Figure 7) (26)
and for WBA resins:
In ajj = 1.77 CAPACITY + 0.369 (Figure F2) (27)
the fact that In ajj isn't related to capacity for strong base
resins is probably due to (1) the relatively narrow range of
capacities available for study — 1.02 to 1.66 meq/ml and (2)
stearic hindrance and poor polarizability of the quaternary-
amine molecule which may be large enough to prevent closer
approach of adjacent functional groups present with the highest
capacity resins compared to the lowest.
Apparently, the distances of separation and the sizes of
functional groups in resins with nitrogen in the matrix (epo-
xies, acrylics, phenolics) are such that stable, electrostatic
bonds are formed with the divalent-sulfate ion making these
resins highly sulfate selective at this level of total concen-
tration (0.005N).
83
-------
From the WBA correlation matrix Table F2 (Appendix) we have
observed that nitrate/chloride selectivity is unrelated to capa-
city but is significantly, negatively correlated with nitrogen
in the matrix. Figure 8 is the regression line/scatter plot
of this latter relationship which is significant but only modes-
tly so (r = -.65) compared to the same correlation for sulfate/
nitrate selectivity (r = .96) just discussed. Some possible
reasons for this negative correlation will be discussed in the
section on crosslinking.
CJ TvT
Effect of pKa on a and ct ,--
The pKa of a resin is a measure of its tendancy to keep
a hydrogen ion, or alternatively, to give up a hydroxide ion.
RNH3+ -*• RNH2 + H^ (28)
or
RNH2HOH + RNH3+ + OH~ (29)
The high affinity that WBA resins have for hydroxide ions
(pKa = 8) as compared to the very low affinity SBA resins have
for, hydroxide ions (pKa > 13) is the reason for the easy and
efficient regeneration of WBA resins by both strong (NaOH) and
weak bases (NH.OH) alike. The disadvantage accompanying this
ease of regeneration is that the solution to be deionized must
be sufficiently acidic to preclude the association of the pro-
tonated amine with the much preferred hydroxide ion instead
of the ion it is desired to remove, e.g., sulfate chloride
or nitrate. Insignificant concentrations of hydroxide ions
existed in the isotherm experiments performed here as acids
were utilized to provide the exchanging ions. Resulting liquid
phase pH's were near 2.3 (.005N). Furthermore, Boari [13]
showed that the sulfate/chloride separation factor was indepen-
dent of pH when the total resin capacity was constant i.e.,
at pH's several units below the resin pKa.
84
-------
In the absence of hydroxide ion interference, there would
seem to be no theoretical reason why pKa should be associated
with sulfate/nitrate or nitrate/chloride selectivity except
through some correlation of Pka with a selectivity-relevant vari-
able like "R" group size. This particular/ secondary correlation
effect does appear to have occurred. Figure F4 (Appendix) de-
picts a statistically significant negative correlation (r= -.64)
£»
between In a and pKa where it appears that the strong-base
(pKa = 13) resins determine the existance of any correlation at
all. Considering the same relationship for weak base resins only
(Figure F5, Appendix) we see an entirely different picture; here
sulfate nitrate selectivity is slightly, positively correlated
with pKa, but the relationship isn't statistically significant
i.e., pKa gives no information about the sulfate selectivity of
WBA resins in general. There is however a curious downward trend
°f the data points at the top of Figure F4. Careful examination
of the dataset revealed that all those points represented non-
o
Polystyrene resins. Figure F6 (Appendix) illustrated In ON
and pKa for non-polystyrene resins only, and that is new, useful
information for a specific subclass of resins not a secondary
correlation as is the one depicted in Figure F4 covering all the
resins but yielding no new information because pKa is highly
correlated with "R" group size (Table F2, r = .80) and we already
know theoretically and empirically that quaternary amines should
and do have relatively low sulfate (divalent-ion) selectivity.
A similar secondary correlation appears in Figure F7 where
In a cl is plotted vs pKa for polystyrene resins only; again
quaternary resins determine the relationship and no new informa-
tion is gained. The pKa of a resin appears to have no real,
causal effect on sulfate/nitrate or nitrate/chloride selectivity
escept for non-polystyrene WBA resins and would not be expected
to be included in an efficient equation predicting selectivity
from resin properties.
85
-------
SCATTER PLOT
N= 29 OUT OFr 32 ".LOG.S N VS. 12 , NPF'OSITN
LQQ,
4.9200
4,0433
3.1666
2.2899
1,4132
.53649
*
3
Nitrogen position is a dummy variable presumably
related to the distance of functional group
separation
H f-
1
,40000
.80000
.20000 .60000
1.00 = Nitrogen in Polymer Backbone
0.00 = Nitrogen Pendant
FIGURE 7
SCATTER PLOT WITH LINEAR REGRESSION EQUATION
In oC^ VS. NITROGEN POSITION
FOR
WEAK AMD STRONG BASE RESINS
1,0000
86
-------
SCATTER PLOT
N= 19 OUT OF 32 10.LOGeN CL VS. 12.N2POSITN
LOG^N/CL
1,5810 -I*
1.3710
1,1609
.95079
.74071
.53063
0.
Nitrogen position is a dummy variable
presumably related to the distance of
functional group separation
.40000
.80000
.20000 .60000
1.00 = Nitrogen in Polymer Backbone
0.00 = Nitrogen Pendant
*
-t-
N2PQSITK
1.0000
FIGURE 8
SCATTER PLOT WITH LINEAR REGRESSION EQUATION
In OC^-L vs. NITROGEN POSITION
FOR
WEAK AND STRONG BASE RESINS
87
-------
G 1VT
Effect of Functionality on a and a ,
The expected effect of functionality on sulfate selectivity
did materialize — the larger the "R" group the lower the
relative sulfate/nitrate selectivity. That relationship is
shown quite clearly in Figure 9 where r = .91. These results
are in empirical if not theoretical accord with Boari et. al
[13] who concluded that "...at every temperature and for every
bulk salinity of the solution the following selectivity towards
the SO.~ ion exists
resins with primary secondary Tertiary Quaternary
amino groups > groups > groups > groups
According to the decrease of the strength of the electric field
of the fixed charges and consequently to their basicity in-
crease." These investigations observed that matrix did have an
effect on sulfate/chloride selectivity but proposed no explana-
tion for the effect. What has been observed in this work is
that the matrix effect is at least equal to, and probably
greater, than the functionality effect in the determination of
sulfate/chloride selectivity as what we are really concerned
with is divalent/univalent selectivity effects. Note also from
the above quote that increased sulfate selectivity is being
equated with increased basicity. Recalling our prior discus-
sion of pKa, which certainly must be considered a measure
of basicity; the point is reiterated here that basicity is not
correlated with Bulfate selectivity when the 13 weak-base resins
are considered (pKa range = 6.8 thru 11.1; Figure F5). It is
only when the quaternary resins are included in or when the
polystyrene resins are excluded from the regression analysis
that statistically significant relationships are obtained
(Figures F4 & F6). This it has been pointed out is a secondary
effect, the size of the functional group being of primary im-
protance. Boari's results and ours are still in accord, but
that is due to the relative sizes of the functional groups not
their basicity, which is unrelated to size among the weak base
resins considered. Basicity is, in fact, not monotonically
88
-------
SCATTER PLOT
2,2899
1.4132
.53649
N= 29 OUT OF 32 9,LOGeS N VS. ll.RSIZE
2.0720
2.2160
2.3600
FIGURE 9
SCATTER PLOT WITH LINEAR REGRESSION EQUATION
In «C^ VS. SIZE OF FUNCTIONAL GROUP
FOR
WEAK AND STRONG BASE RESINS
89
-------
related to functionality even in simple alkylamine monomers
as the listing below (Reference 34 p. 5-15) illustrates:
TABLE 11. pKa's OF ALKYLAMINES IN WATER AT 25°C
Amine
Functionality
Amine
pKa
Primary
Secondary
Tertiary
Quaternary
Primary
Secondary
Tertiary
Quaternary
Methylamines
(CH3)3NH
NOH
Ethylamines
10.62
10.77
9.80
>13.0 (est)
10.63
10.93
10.72
>13.0 (est)
For these methylamine and ethylamine monomers there is no
clear basicity trend except that the quarternary amines
are stronger than all the others for which no simple trend
exists. The same general observation can be made regarding
the lack of correlation between functionality (or "R" size) and
the resin pKa's determined from the weak-base resin, titration
curves; r = .0155 (Appendix, Table F2).
The summary argument for the size of the functional group
being more important than its basicity in determining sulfate
selectivity is
90
-------
(1) High divalent-ion (sulfate) selectivity clearly
c
depends on functionality the order of a., being:
primary > secondary > tertiary > quaternary
(2) The size of functional groups can be ranked in the
order:
primary < secondary < tertiary < quaternary
(3) The basicity of functional groups in resins (and
aliphatic amine monomers) can only be ordered as:
quaternary > tertiary, primary, secondary
Before discussing nitrate/chloride selectivity in some detail
it is approrpaite to study the physicochemical model of selec-
tivity used by Boari [13] and attributed to Eiseman [43] the
components of which have been discussed elsewhere by Reichenberg
[103] and others [60, 38, 39, 69, and 26]. Energetically, the
ion-exchange reactions may be accounted for by two distinct
physiochemical processes: (1) the partial or total destruction
of the hydrated structure of the counterion in dilute aqueous
solution, and (2) electrostatic bonding of counterion to the
resin structure (coion). Considering the binary ion exchange
of sulfate and chloride:
2RHC1 + S0~^RS0 + 2C1 (30)
Boari calculated that the electrostatic energy term was pre-
dominant over the hydration term when the overall change in
standard free energy of the system was taken to be:
(AG'ex> S04/C1 = hydration
- (AG^, - 1/2 AGC_. ) electrostatic (31)
Cl t>u4
In fact, it is generally true, as has already been discussed,
that in dilute solution (<.06N), ion exchangers prefer the ion
with the highest valence, an observation also indicating pre-
dominance of the electrostatic term. When considering ions of
91
-------
similar valence, however, the electrostatic contribution to the
overall free energy change is not expected to outweigh the hy-
dration contribution. This appears to be the case with nitrate/
chloride exchange.
Effect of Functional Group on Nitrate/Chloride Selectivity
N
The scatter plot of In acl vs "R" group size is Figure F8
where it is seen that no significant (r = .25) linear relation-
ship exists. There do appear to be some trends however when the
polystyrene resins (circled data points) are separated from
the others: (1) the polystyrene resins have generally higher
nitrate selectivity and (2) the tertiary polystyrene resins
generally have the highest nitrate/chloride selectivity. The
former observation is a reaffirmation of the previous finding
that nitrogen out of the matrix (polystyrene) produces generally
higher nitrate preference wrt chloride (Figure 8).
Effect of Relative Degree of Crosslinking on Nitrate/Chloride
Selectivity
Recall that the categorical variable, porosity, having the
stratifications: isoporous, microporous and macroporous was
converted to the dummy variable XLIWRING i.e., the relative
degree of crosslinking with values of 0.5 = isoporous, 1.0 =
microporous and 2.0 = macroporous. For non-polystyrene resins,
nitrate/chloride selectivity is positively correlated (Figure
10, r = .89) with the relative degree of crosslinking - macro-
porous resins tending to have higher nitrate selectivity than
microporous resins. For polystyrene resins no such relationship
exists (Figure 11, r = .33). Consequently, when all resins
are considered, only a modest correlation (r = .45) results
which is barely significant at the .06 level. In summarizing
the effects of both matrix and crosslinking the following trends
are observed:
High nitrate/chloride preference is indicated by:
92
-------
SCATTER F;'LOT STRAT=N.T.TROGEN: 1
N- 7 OUT OF 8 10»L.OGeN CL VS. 14.XLINKING
LOGe.l^CL
1.3481
1.1846
1.0211
,85761
.69412
.53063
1.0000
1.4000
1.2000
1.6000
1,8000 XL.INK TNG
2.0000
FIGURE 10
SCATTER PLOT WITH LINEAR REGRESSION EQUATION
ln * VS>
DEGREE OF CROSSLINKING
FOR
NON-POLYSTYRENE RESINS
93
-------
=lf,AITfc".k PLOT STRAT ,
N = 12 OUT OF 24 10.LOGeN' CL US. 14.XL1NKTNG
1.3S1C
1.4745
1 • 2606
1.1541
1.0473
H 1-
.50000
laooo
t. 70 oo
•80000
1.4000
i^LJ.NKINU
2.0000
FIGURE 11
SCATTER PLOT WITH LINEAR REGRESSION EQUATION
In rt^j vs. PELATIVE DEGREE OF CROSS LINKING
FOR
POLYSTYRENE RESINS
94
-------
(1) polystyrene resins in general
(2) tertiary > quaternary (for polystyrene resins only)
(3) macroporous > microporous (for non-polystyrene resins
only)
The nitrate/chloride preferences which these resins exhibit are
rather difficult to explain in terms of the expected changes in
free energies of ionic hydration and electrostatic interaction.
Three parameters indicative of the order of hydration energy:
(1) effective ionic radius in aqueous solution, (2) ionic
activity coefficient and, (3) limiting equivalent ionic con-
ductance are given in the tables below. For each parameter,
the ions are listed in the order of expected preference by the
resin phase in dilute solution (<0.05N).
TABLE 12 [34]. EFFECTIVE JONIC RADII IN
AQUEOUS SOLUTION; A, 25°C
(also referred to as
Hydrated Ionic Radii
and 0
Debye-Hiickel Ion-Size Parameter, a)
Chloride =3.0
Nitrate = 3.0
Sulfate = 4.0
Bicarbonate = 4.0
95
-------
TABLE 13 [34]. INDIVIDUAL IONIC ACTIVITY COEFFICIENTS (y . )
Of
Ions in Water at 25°C, .005N
Calculated from:
-logy. -
1 + Bai/T
where
: I = 0.5 T C.Z?
v 11
Bicarbonate = .927
Chloride .925
Nitrate .925
Sulfate .693
TABLE 14 [34]. LIMITING, EQUIVALENT, IONIC CONDUCTANCE
IN AQUEOUS SOLUTION AT 25°C
2
mho • cm /equivalent
1/2 sulfate = 80.00
chloride = 76.35
nitrate = 71.40
Bicarbonate = 44.50
Actually, ionic conductance is a measure of both the hydration
and electrostatic energy effects; the rate at which an ion moves
through water in an electric field is influenced among other
things by its hydration shell and the charge it carries.
Conductance does correctly predict that sulfate should be the
most preferred and bicarbonate the least preferred but incor-
rectly predicts as do the other two parameters that chloride
should be equally or more preferred than nitrate.
That polystyrene resins and relatively highly crosslinked
96
-------
resins exhibit the highest nitrate/chloride selectivity may have
to do indirectly with the water content of the resins - these
categories of resins being expected to contain relatively less
water due to their hydrophobia non-polar character.
Sulfate/Nitrate vs Nitrate/Chloride Selectivity
Considering all resins, low sulfate selectivity generally
corresponds with low chloride selectivity. This was originally
presumed to be a fortunate correlation? see Figure F9 where
r = -.53 for In a^ vs In acl. Among the weak base resins the
correlation is even higher with r,*= -.67 (Table F2). It is
presumed to be fortunate because minimizing the sulfate selec-
tivity also tended to minimize the chloride selectivity i.e.,
the effects of nitrogen position, functionality and porosity
on selectivity do not offset one another; however, as we shall
see later, moderate to high sulfate selectivity actually im-
proves the chemical efficiency of the nitrate removal processes.
Sampling Bias: "R" Group Size vs Nitrogen Position
The previous discussion has emphasized the importance of
functionality and nitrogen-in-or-out of the matrix in determin-
ing selectivity. Unfortunately, for this particular sampling
of resins, these two variables were highly correlated making it
more difficult to see intuitively which factor was most impor-
tant; see Figure F10 where r = -.82 for "R" Group Size vs
Nitrogen Position. There is some fundamental reason for this
correlation; all the quaternary amine resins have nitrogen out
of the matrix and this author is unaware of the existance of a
monofunctional quaternary amine resin where nitrogen is part of
the continuous structure .
Helfferich [60] mentions a polycondensation polymer of poly-
ethyleneamine and epichlorohydrin, but this resin has tertiary
and secondary amines mixed with the quaternary groups. Another
possible exception, Amberlite IRA 458, an acrylic gel Type 1
strong-base anion resin was not evaluated in this work.
97
-------
In spite of this fundamental problem, the correlation
could have been made less significant had there been more
tertiary amine resins with nitrogen in the matrix (like Amber-
lite IRA-68, acrylic-tertiary) and more polyamine resins with
nitrogen out of the matrix (like Amberlite IR-45, STY-DVB-
polyamine). The addition of such resins to the data set would
have improved the reliability (significance) of the correlations
but would not have changed their validity. This assumes/ of
course, that all resins having the same functionality and matrix
behave essentially the same with respect to selectivity - a
fact which has been observed and is amply demonstrated by the
composite isothersm (Figures 12-17).
Selection of Regression: The Final Statistical Result
Having established which independent variables were most
important in determining selectivity and having created the
necessary interval-scale dummy variables to represent the im-
portant categorical variables, the task remaining was to in-
corporate these into a simple, efficient predictive equation.
The selection of regression technique [42, 48] was used to
accomplish this objective. It is essentially the optimization
of a multiple regression analysis. The dependent variable to
be predicted is chosen along with the independent variables and
the desired levels of significance for inclusion and rejection
of the various independent variables. The program computes
the individual, simple, linear-regression coefficients (r's)
for each independent variable then chooses the highest one
which is significant at or below the level specified. The
partial correlation coefficients are then computed for the
remaining variables i.e., the ability of each of the remaining
variables to account for the remaining variance is determined.
Whichever one of these has the highest, partial correlation co-
efficient is then incorporated into the now multiple-regression
equation but only if it improves the multiple correlation co-
efficient (R) at or below the second level of significance
98
-------
0.20 O.UO 0.60 0.80
EQUIVflLENT FRflCTION SO, IN LIQUID PHflSE
EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
WEflK BflSE flNIQN RESINS 1, 5, 8 4 12
MflCROPOROUS RESINS, STTRENE-DVB MATRICES
TERTIflRY flMINE FUNCTIONflLITT
D= RESIN NO. 1. flMBERLITE IRfl 93
O= RESIN NO. 5. DOWEX MWfl-1
A= RESIN NO. 8, DUOLITE ES-368
-*- = RESJN NO. 12. lONflC flFP 329
FIGURE 12
COMPOSITE ISOTHERMS, 25° C, 0.005 N
1.00
99
-------
0.20 0.10 0.60 0.80
EQUJVRLENT FRflCTION S0» IN LIQUID PHfiSE
EQUIVRLENT FRflCTION CL IN LIQUID PHflSE
1.00
WEflK BflSE flNJON RESINS 6 4 9
GRflNULflR, MflCROPOROUS RESINS, PHENOLIC MflTRICES
POLYRMINE FUNCTIONflLITY
D= RESIN NO. 6. DUOLITE fl7
0= RESIN NO. 9, DUOLITE ES-561
FIGURE 13
COMPOSITE ISOTHERMS, 25° C, 0.005 N
100
-------
I 1 1 1
0.20 0.40 0.60 0.80
EQUIVflLENT FRRCTION SO^ IN LIQUID PHflSE
XCL, EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
WEflK flND INTERMEDIflTE BflSE flNION RESINS 4, 7
GEL RESINS, EPOXT-flMINE MflTRICES
POLTflMINE FUNCTIONflLITY
m = RESIN NO. 4. DOWEX WGR
© = RESIN NO. 7, DUOLITE fl-340
A = RESIN NO. 13. IQNflC fl-305
1.00
13
FIGURE 14
COMPOSITE ISOTHERMS, 25° C, 0.005 N
101
-------
0.20 O.UO 0.60 0.80
EQUIVflLENT FRflCTION S0« IN LIQUID PHflSE
EQUIVPLENT FflflCTION CL IN LIQUID PHflSE
1.00
STRONG BflSE flNION RESINS 15, 17, 21, 27 4 32
GEL flND MflCROPOROUS RESINS, STTRENE-DVB MflTRICES
TYPE I, QUflTERNflRY RHINE FUNCTIONflLITY
D= RESIN NO. 15, flMBERLITE IHFMiOO
O = RESIN NO. 17. flMBERLITE IRfl-900
A= RESIN NO. 21. DOWEX SBR
+ = RESIN NO. 27. IONRC RSB-1
X = RESIN NO. 32. IQNflC flFP-100
FIGURE 15
COMPOSITE ISOTHERMS, 25° C, 0.005 N
102
-------
0.20 0,10 0.60 0.80
EQUIVflLENT FRflCTION SO,, IN LIQUID PHflSE
EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
1.00
STRONG BRSE RNION RESINS 16, 19, 22, 24, 28 4 30
'IMPROVED POROSITY" GEL RESINS, STYRENE-DVB MRTRICES
TYPE I, QURTERNRRY RHINE FUNCTIONRLITY
Q = RESIN NO.
0= RESIN NO.
A= RESIN NO.
-I-« RESIN NO.
* = RESIN NO.
16, flMBERLITE IRfl-U02
19, OOWEX SBR-P (21-KJ
22, DOWEX 11
2U, DUOLITE fi-101-0
28, IQNflC R-6H1
RESIN NO. 30, IDNflC RSB-1P
FIGURE 16
COMPOSITE ISOTHERMS, 25° C, 0.005 N
103
-------
'"b.oo
0.20 O.UO 0.60 0.80
EQUIVRLENT FRflCTION 80,4 IN LIQUID PHflSE
EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
1.00
STRONG BflSE flNION RESINS 14, 18, 20, 23 £ 29
GEL flND MflCROPOROUS RESINS, STYRENE-DVB MflTRICES
TYPE II, QURTERNflRY-RMINE FUNCTIONflLITY
m= RESIN NO. 14. flMBEBLITE IRfl-910 (MR)
0 = RESIN NO. 18. flMBERLITE IRfl-410
A = RESIN NO. 20, DOWEX SflR
+ = RESIN NO. 23. DUOLITE fl-102-D
* = RESIN NO. 29, IDNflC flSB-2
FIGURE 17
COMPOSITE ISOTHERMS, 25° C, 0.005 N
104
-------
specified. The procedure is repeated until there are either no
significant partial correlations remaining or no significant
improvement in multiple R is possible. An example output for
the selection of regression of N2POSITN, RSIZE, CAPACTIY AND
o
XLINKING on Log oC for WBA resins is included in Appendix F for
review. A similar example for the selection of regression of
N
the same variables on Log acl is also included in Appendix F.
The resulting predictive equations are listed below with their
associated statistical parameters.
Sulfate/Nitrate Selectivity for all Resins;
In a^ = 2.30 3J2POSITN - 3.63 RSIZE + 9.44 (33)
R = 0.980
F = 317
SIGNIF = .0000
Sulfate/Nitrate Selectivity for WBA Resins
In a^ = 2.03 N2POSITN - 7.45 RSIZE + 17.44 (34)
R = .991
F = 261
SIGNIF = .0000
Nitrate/Chloride Selectivity for All Resins
In a^ = -0.371 N2POSITN + 0.206 XLINKING + 0.962 (35)
R = .772
F = 11.8
SIGNIF = .0007
105
-------
Nitrate/Chloride Selectivity for WBA Resins
In a^ = -0.413 N2POSITN + 0.381 XLINKING + 0.755 (36)
R = .919
F = 24.6
SIGNIF = .0002
The usefulness of the equations for "all resins" is that they
indicate the two most important variables out of the six
possible which influence the selectivities of anion resins in
general. For predictive purposes three additional equations
are offered. They are bas=d on the results of the analysis of
variance tests. See Table 8.
aN = 1.82 for all Type 1, Gel and MR SBA resins
aj? = 2.98 for all Type 1, ISO; and Type II Gel and MR SBA
resins
N
ou, = 3.14 for all SBA resins regardless of type or
porosity
Comparison of Predicted Selectivities to Measured Selectivities
Equations 39-43 were used to predict aN and cu-, for each
representative type of resin. These predictions are compared
to the average of the experimentally measured selectivities in
Tables 15 and 16 following.
PHASE I RESULTS SUMMARY: ANION RESIN SELECTIVITY STUDIES
Thirty-two anion resins from four U.S. manufacturers were
tested for nitrate,chloride, sulfate and bicarbonate selectiv-
ity; nitrate and chloride capacity, and organics bleed. Sul-
furic, nitric and hydrochloric acid titration curves were
constructed from equilibrium data for the weak base resins.
The resins comprised a variety of combinations of matrix,
106
-------
TABLE 15
PREDICTED AND MEASURED VALUES OF SULFATE/NITRATE SELECTIVITY: a
N
Resins
1,5,8,12
3
6,9
4,7.3
11
2
10
16,19,22
24,28,30
15,25,27
17,23
18, 2C,
23,29
14
Resin Description
STY-DVB, Tert. Amine, MR
STY-DVB, Polyamine, Gel
PEBNOL-HCHO-PA, Polyanune, MR
EPOXY-AMINE, Polyamine, GEL
ALIPHATIC-AMINE, POLYAMINE, GEL
ACRYLIC -AMINE, TERT .AMINE, GEL
ACRYLIC-AMINE, POLYAMINE, MR
STY-DVB, Quat. (I) Amine, ISO
STY-DVB, Quat. (I) Amine, GEL
STY-DVB, Quat. (I) Amine, MR
STY-DVB, Quat. (II) Amine, GEL
STY-DVB, Quat. (II) Amine, MP
Predicted
og *
3.08
12.7
97
97
97
23.4
97
2.98
1.82
1.82
2.98
2.98
Ave
Measured
-^
3.08
12.7
108
109
54
23.4
94
2.99
1.89
1.74
..94
3.26
%
Error
0.
0.
-10.
-11.
+ 80.
0.
+ 3.
0.
-•-4.
+5.
+ 1.
-9.
* Equations 39 and 41 ,42 were used to calculate the WBA and .-iBA resin
selectivities respectively.
With three equations, the sulfate/nitrate selectivities of 11 of the
relevant combinations of matrix, functionality, porosity and type
are predicted to within + 11% in the extreme range of s^lectivities
encountered: 1.8 to 1097 By extropoation, these equations should
correctly predict ( + 11%) the selectivities of 24 of the 26 possible
strong and weak-base resins. More importantly, only one equation
is required to correctly describe (± 11%) the sulfate/nitrate pre-
ferences of 6/7th of the weak-base resins: the primary subjects
of this study.
On an individual resin basis, the three equations "correctly" predict
a| for 28 of the 29 resins evaluated and whose selectivities (a^)
vary over the wide range of 1.71 to 137.
107
-------
TABLE 16
PREDICTED AND MEASURED VALUES OF NITRATE/CHLORIDE SELECTIVITY: a
N
'Cl
Ave
Predicted Measured
Resins
Resin Description
Error
1,5,8,12
3
6,9
4,7
11
2
10
16,28
21
17,32
29
14
STY-DVB, Tert. Amine, MR
STY-DVB, Polyamine, Gel
PHENOL-HCHO-PA, Polyamine, MR
EPOXY-AMINE, Polyamine, GEL
ALIPHATIC-AMINE , POLYAMINE, GEL
ACRYLIC-AMINE, TERT. AMINE, GEL
ACRYLIC-AMINE, POLYAMINE, MR
STY-DVB, Quat. (I) Amine, ISO
STY-DVB, Quat. (I) Amine, Gel
STY-DVB, Quat. (I) Amine, MR
STY-DVB, Quat. (II) Amine, GEL
STY-DVB, Quat (II) Amine, MR
4.56
3.11
3.02
2.06
2.06
2.06
3.02
3.14
3.14
3.14
3.14
3.14
UJL
4.33
3.89
3.00
1.85
2.25
1.89
3.85
3.22
2.90
3.19
3.64
2.85
+5.
-20.
+1.
+11.
-8.
+9.
-22.
-2.
+8.
-2.
-14.
+10.
* Equations 40 and 43 were used to calculate the WBA and SBA resin
selectivities respectively.
Here with two equations, the nitrate/chloride selectivities of
all 12 relevant combinations of matrix, functionality porosity
and type are predicted to within + 22% in the relatively narrow
range of selectivities encountered: 1.85 - 4.33. Again,
by extropolotion, the selectivities of all 26 possible weak and
strong base anion resins are expected to be correctly predicted
by these equations (+ 22%).
N
One equation is necessary to predict (+ 22%) the aC]_'s of all of
the weak-base anion resins.
The two equations predict (+ 22%) the ot^i values of all 19 of the
resins for which nitrate/chloride selectivity was evaluated.
108
-------
functionality, porosity, pKa, and capacity. Degree of cross
linking had not been directly specified or determined, so it
was related to porosity for data analysis. The following is a
list of the ranges of the independent variables:
Matrix: STY-DVB, Acrylic, Aliphatic, Epoxy, Phenolic
Functionality: quaternary, tertiary, polyamine
Porosity: microporous, macroporous, isoporous
Capacity: 0.98 — 2.54 meq/ml
pKa: 6.8 — 13
See Table 3 for complete details on resins tests.
Sulfate was always preferred over nitrate by all the strong
and weak-base resins tested which exhibited an extremely wide
c
range of selectivities: a^ = 1.71 to 137. Although these
separation factors can strictly be applied only at 0.005N, it
is expected that the selectivity trend will hold true for any
anion resin tested with groundwaters having total dissolved
solids up to at least 3000 ppm (0.06N as CaCO3). See Figures
A1-A32 (Appendix) and Variable Total Concentration Isotherms
(Figure 19).
Nitrate was always preferred over chloride by all the
anion resins tested although the range of preference was rela-
tively narrow: a!,, = 1.85 — 4.33, and, as exptected, was not
v*« J.
a function of total concentration. See the lowest isotherm
of Figure 19.
The average separation factor, ou, determined by the ratio
of areas technique proposed here provided an adequate descrip-
s
tion of the resin preference for sulfate at constant a at a
given total concentration (.005N) even though the least selec-
c
tive (aN = 1.7 to 3.7) resins yielded isotherms with inflection
points and would have required empirical cubic equations for a
good cirve fit. See Figures A33 and A34 and Reference 100,
Table 16-5. These more-or-less "S" shaped isotherms describe
109
-------
the sulfate/nitrate behavior in tertiary and quaternary STY-DVB
c
resins (Figures Al, A5/ A8, A12 and A14-A32) of modest (a =
2.0 — 4.0) sulfate preference as opposed to the apparently
inflectionless isotherms of resins with high sulfate selectiv-
ity (ajj = 13-137, Figures A2-A4, A7, A9, All and A13) . It is
proposed that the "S" shaped isotherms represent resins with
sites of unequal preference for sulfate as verified by the
generally higher preference (more convex curves) for sulfate
at low equivalent fraction of that ion; see Helfferich [60]
p. 183. The large tertiary and quaternary functional groups
pendant on the polystyrene matrix may less frequently be close
enough together to satisfy the divalent, sulfate ion than is the
case with polystyrene - polyamine resins and non-polystyrene
resins where the probablility of the two requisite functional
groups being close enough to satisfy divalency is expected
to be much greater.
A separate indication of unequal preferences between ions
of dissimilar valence is exemplified by the differences among
the titration curves for a given weak-base resin; see Figures
B1-B12. None of the polyamine resins have discernible inflec-
tion points for HC1 or HNCU whereas all but one (Figure B9)
have definite inflection points for H2SO4 indicating the pre-
sence of sites of nearly equal affinity for divalent anions
like sulfate but not for monovalent ions like chloride and
nitrate. As expected, all these polyamine resins showed great
affinity for sulfate over nitrate. The exceptional resin,
Duolite ES-561 (Figure B9) had been manufactured in such a way
so as to minimize the number of pendant amine groups (R.
Anderson, Diamond, Shamrock Chemical Co., Personal Communica-
tion) , a procedure which tended to produce fewer pairs of
sites preferring divalent anions.
I
Briefly, resins with relatively low sulfate selectivity
have modestly "S"-shaped isotherms explained by the -cendency
110
-------
of these resins to have sites with varying affinities for the
divalent sulfate ion. Resins of high sulfate selectivity have
smooth shaped isotherms explained by the expected preponderance
of pairs of sites available for divalent ion interactions.
The separation factor and, more specifically, the ratio-
of-areas technique provides an excellent description of a .
which is nearly constant, and independent of total concentration
and equivalent fraction as theoretically expected for monoval-
ent-monovalent, ion-exchange. See the chloride/nitrate iso-
therms in Figures 19, A1-A32 and A35-A36.
Ion-exchange hysteresis does not appear to have been very
significant either in nitrate-sulfate or nitrate-chloride
exchange. Essentially the same isotherm was arrived at regard-
less of the initial ionic form of the resin be it nitrate,
sulfate or chloride. See Figure 18.
Total concentration variations in the range of 0.002 to
——————————— g
0.008N (Figure 19) gave rise to separation factor (aN) varia-
tions as follows for a modestly sulfate selective, STY-DVB,
tertiary amine resin of the type one might choose for nitrate
removal service:
CT 4
0.002 N 5.2
0.005 N 2.8
0.008 N 1.8
Statistical Analysis of the Phase I experimental data
using analysis of variance, simple and multiple regression
analyses; scatter plotting and selection (optimization) of re-
gression yielded validated, predictive equations for sulfate/
nitrate and nitrate/chloride selectivities. This was accompli-
shed only after the relevant categorical variables, matrix and
111
-------
SULFflTE-NITRflTE
FROM NITRflTE FORM
FROM SULFflTE FORM
CHLORIDE-NITRflTE
FROM NITRflTE FORM
ORIDE-NITRflTE
FROM CHLORIDE FORM
0.20 O.UO 0.60 0.80
EQUIVALENT FRflCTION SO, IN LIQUID PHflSE
Xa. EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
WEflK BflSE PINION RESINS 8 4 12
DUOLITE ES-368, IDNflC flFP-329
MflCROPOROUS RESINS, STYRENE-DVB MflTRICES
TERTIflRY FIMINE FUNCTIONflLITY
D= RESIN INITIflLLY IN NITRflTE FORM
0= RESIN INITIflLLY IN SULFflTE FORM
+ - RESIN INITIflLLY IN CHLORIDE FORM
FIGURE 18
1.00
HYSTERESIS ISOTHERMS, 25° C, 0.005 N
112
-------
SULFRTE-NITRflTE
0.002 N
SULFflTE-NITRflTE
0.005 N
SULFflTE-NITRflTE
0.008 N
CHLORIDE-NITRflTE
^002. .005 «, .008 N
-+-
0.20 O.UO 0.60 0.80
EQUIVflLENT FRflCTION SO,, IN LIQUID PHflSE
EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
WEflK BflSE flNION RESIN NO. 12
IDNflC flFP-329
MflCROPOROUS RESIN, STYRENE-DVB MflTRIX
TERTIflRY flMINE FUNCTIONflLITY
03= 0.002 N TOTflL CONCENTRflTION
A= 0.005 N TOTflL CONCENTRflTION
+ = 0.008 N TOTflL CONCENTRflTION
FIGURE 19
VflRIflBLE TOTflL CONCENTRflTION ISOTHERMS, 25° C
1.00
113
-------
functionality, had been converted to interval scale, dummy
variables: nitrogen position (N2POSITN — related to coion
separation distance) and "R" group size (RSIZE), after much
preliminary analysis of the descriptive data. See Equations
37-43.
Matrix is the single most important factor in the determin-
ation of both a and a , and consequently of nitrate selectivity
in general. See "selectivity as influenced by matrix type" for
tertiary amines (Figure 20) and polyamines (Figure 21). If
the electrostatically active nitrogen atoms are in the contin-
uous polymer structure, as they are with all but the polystyrene
resins where the active nitrogen is pendant on the polymer
structure, then the resin is highly sulfate selective. This,
it is hypothesized, is due to the almost-guaranteed proximity
of two active nitrogen atoms which are expected to be separated
o o
by about 4.48 A in the polymer backbone. This distance, 4.48 A,
derives from the nitrogen separation distance of one ethylene
group in the amine monomers, diethylenetriame (DETA) and
triethylenetetraamine (TETA), commonly used to provide function-
ality and crosslinking in anion exchange resins:
o
4.48 A
o
4.48 A-
NH2— CH2 — CH2 — NH — CH2— CH2—
DETA
Fixed pairs of properly-spaced electrostatically-charged amines
will tned to prefer single, divalent anions for both entropic
and electrostatic reasons. With tertiary and quaternary amines
pendant on a polystyrene matrix, the natural electroselectivity
of multiple charged ions is reduced by the steric hinderance of
the large functional groups and the lesser probability of their
being properly spaced to interact with a divalent ion of fixed
size like sulfate. Summarily the nitrate/sulfate selectivities
114
-------
^).00
0.20 O.UO 0.60 0.80
EQUIVqLENT FRflCTION SO^ IN LIQUID PHflSE
EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
1.00
NO. 2j flCRYLIC RHINE. TERTIRRY-flMINE RESIN. HCL CflP. = I.H2 MEQ./ML.
NO. 12» STYRENE-DVB. TERTIflRY-flMINE RESIN, HCL CflP. =1.26 MEQ./ML.
CT = 0.005 N.
T = 25° C
FIGURE 20
SELECTIVITY flS INFLUENCED BY MflTRIX TYPE
(flLL TERTIflRY-flMINE FUNCTIONflLITY RESINS)
115
-------
0.20 O.HO 0»60 0.80
EQUIVRLENT FRflCTION SO, IN LIQUID PHflSE
EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
1.00
NO, 3i STYRENE-DVB, POLYflMINE RESIN, HCL CRP. "1.76 MEQ./ML.
NO. 4: EPOXY-flMINE, POLYRMINE RESIN, HCL CRP. » I.S3 MEQ./ML.
NO, 6j PHENOL-HCHO, POLYflMINE RESIN, Ha CRP. - K67 MEQ./ML.
NO. lOt HCRYLIC-flMINE, POLYflMINE RESIN, HCL CflP. =2.59 MEQ./ML.
Of = 0.005 N.
T = 25° C
FIGURE 21
SELECTIVITY flS INFLUENCED BY MflTRIX TYPE
(flLL POLYflMINE FUNCTIONflLITY RESINS)
116
-------
(a,, not ot ) can be ordered as:
Polystyrene > non-polystyrene resins
That nitrogen in the continuous structure should be associated
with low nitrate/chloride selectivity is not easily explained
nor is the observation that nitrate is always preferred over
chloride by all these anion resins. Their relative energies
of hydration as indicated by calculated activity coefficients,
and measured, limiting, ionic conductances, indicates that
chloride should be favored. Electrostaticlly their charges
are etqual but structurally they differ; nitrate is larger
[98, 67] and of greater polarizability than chloride [89],
and nitrate is polyatomic-planar while chloride is monatomic-
spherical [67]. The "water-structure-enforced, ion pairing"
described by Diamond [38] may well account for the unexpectedly
high resin preference for nitrate.
Coincidentally, the polymers with nitrogen in the matrix
are also more polar (and hydrophilic) than polystyrene resins;
the acrylic and aliphatic resins contain carbonyl groups while
the epoxies and phenolics contain hydroxyl groups. In Diamond's
view, large, poorly-hydrated, univalent anions tend to be
rejected from an aqueous phase and have higher activity coef-
ficients than predicted by the Debye-Huckel limiting law. His
specific reasoning is that
"...Such ions intrude into the water structure without
being able to orient the water molecules around themselves
into coordinate hydration shells; as a result the water
molecules nearest the ions are bound more tightly into the
water structure beyond them".
Should the effect be considered applicable here it would help
the systematic, resin phase preference for nitrate over chloride
and the fact that nitrate is more preferred in more hydrophobia
(polystyrene) resins. See Diamond [38] p. 257, for supporting
arguments relating selectivity differences to the hydrophilic
117
-------
and hydrophobia character respectively of cation vs anion-ex-
change resins (polystyrene-sulfonic acid vs. polystyrene-
quaternary amine). Briefly, nitrate/chloride selectivities
are ordered as follows:
polystyrene > non-polystyrene resins
Functionality is nearly as important as nitrogen position
in determining sulfate selectivity but has no apparent effect
on the nitrate/chloride preferences of resins. See "Selectiv-
ity as Influenced by Functionality" for STY-DVB resins (Figure
22) and acrylic resins (Figure 23). Although the functionality
effect on sulfate selectivity has previously been attributed
to functional group basicity [13] no uniform correlation be-
c
tween basicity (pKa) and a^ was obtained here. Rather, the
size and steric hindrance produced by the functional groups
seem to be the determining factors; larger functional groups
tend to prevent the required proximity of a pair of nitrogen
atoms in addition to hindering the approach of the mobile coun-
terions to the positively charged nitrogen centers.
Briefly, &„, is independent of functionality but a^ is
TJ
greatly influenced by it. Nitrate/sulfate selectivities (a^
s
not aN) are ordered as follows:
Quaternary > Tertiary » Polyamine
That is well demonstrated by Figures 22 and 23
c
Capacity is not a significant variable for predicting GL*
N
or a ^ even though high capacity (equated with high internal
molality) should theoretically produce high sulfate selectivity.
c
Capacity was mildly correlated with
-------
1 1 1 1
0.20 O.UO 0.60 0.80
EQUIVRLENT FRflCTION SOJ, IN LIQUID PHflSE
EQUIVflLEMT FRflCTION CL IN LIQUID PHflSE
1.00
NO. 3, STYRENE-DVB. POLYflMINE RESIN, HCL CflP.
NO. 8| STYRENE-DVB, TERTIflRY-flMINE RESIN, HCL CflP.
NO. 21t STYRENE-OVB. QUflTERNRRY-flMINE RESIN. HCL CflP.
C, = 0,005 N.
T = 25° C
U76 HEQ./ML.
1.U3 HEQ./ML.
1.66 MEQ./ML.
FIGURE 22
SELECTIVITY RS INFLUENCED BY FUNCTIONflLITY
(flLL STYRENE-OVB MflTRIX RESINS)
119
-------
0.20 O.UO 0.60 0.80
Xj»t. EQUIVRLENT FRflCTION SO,, IN LIQUID PHflSE
X. EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
1.00
NO. 2; flCRYLIC-flMINE. TERTIRRY-flMINE RESIN. HCL CflP. = 1.12 MEQ./ML.
NO. 10; flCRYLIC-flMINE. POLYflMINE RESIN. HCL CflP. - 2.59 MEQ./ML.
Ct = 0.005 N.
T = 25° C
FIGURE 23
SELECTIVITY RS INFLUENCED BY FUNCTIONflLITY
(FILL nCRYLIC-flMINE MflTRIX RESINS)
120
-------
reliability once the major effects due to matrix and function-
/
ality have been accounted for. See Selection of Regression in
Appendix F.
Porosity is a major determinant of sulfate selectivity
among Type I strong-base anion resins where isoporous resins
with a relatively low degree of crosslinking are considerably
Q
more sulfate selective (aN = 2.98) than are the more-crosslinked
c
gel and macroporous resins (a = 1.82). Since hydration of the
sulfate ion is not very significant (c.f. limiting ionic
conductances Table 14) the screening by size due to higher de-
grees of crosslinking isn't expected to be significant.
Greater flexibility of the polymer matrix permitting freer
movement of the quaternary groups to pair-up with the divalent
ion is offered as a possible explanation although the size-
screening effect probably cannot be entirely ruled out. With
c
this single exception of porosity being related to
-------
degree of crosslinking and nitrate/chloride selectivity for
polystyrene resins. This is also as expected from the theory
that nitrate selectivity is influenced by hydrophobiaity which
probably doesn't change as a result of increased crosslinking
in STY-DVB resins. See Figures 20, 21, 10, and 11.
The basicity (pKa) of a resin doesn't appear to be a pri-
nary determinant of either sulfate/nitrate or nitrate/chloride
selectivity Although it is statistically and meaningful correla-
ted with sulfate selectivity when non-polystyrene weak-base
anion resins are considered. See Figure F6, Appendix. No new
information is gained regarding either selectivity of any
polystyrene resin except a Type II SBA resin from the knowledge
of basicity after the most important independent variables,
matrix and functionality, have been specified. Furthermore,
the pKa's of weak-base resins (polystyrene and non-polystyrene)
' O TVT
are unrelated to a or a_,. See Figures F4 and F5, Appendix.
Type II, strong base anion resins have higher sulfate
s s
selectivity («N = 2.99) than do the Type I resins (all = 1.82).
Since the major difference here is basicity, one would be
inclined to conclude that in this particular classification,
i.e., Gel and macroporous, SBA resins, that reducing the
basicity (pKa) increases the sulfate selectivity. However, for
SBA resins as a group, porosity is as important as Type in
determining selectivity; recall that the difference between
S S
isoporous (oC = 2.98) and MR or Gel (
-------
ants of sulfate/nitrate selectivity are matrix and function-
ality while matrix and relative degree of crosslinking are
the primary determinants of the magnitude of nitrate/chloride
selectivity. Nevertheless, within particular subclasses of
resins other factors such as type and basicity (pKa) do have
S N
significant further influences on (XN and acl- All of these
important relationships are summarized in the predictive
equations and summary Tables 17 and 18 below. Note carefully,
that in the tables, selectivities (separation factors) are
given in a^ (not a^) and acl since it is the purpose there to
summarize all of the influences on nitrate selectivity.
For Anion Resins in General; (Rough Estimate)
ajj = exp (2.3 N2POSITN - 3.63 RSIZE + 9.44) (37)
a^., = exp (-0.371 N2POSITN + 0.206 XLINKING + 0.062)
(38)
For Weak-Base Anion Resins; (ajj, +10% at .005 N)
(ou,, +20%, independent
of cone.)
S
N
aJJ, = exp (-0.413 N2POSITN + 0.381 XLINKING + 0.755) (40)
af7 = exp (2.03 N2POSITN - 7.45 RSIZE + 17.44) (39)
N
For Type I Gel and MR Strong-Base Anion Resins
a?T = 1.82 (+ 10% at .005 N) (41)
N —
For all Type I, Isoporous, and Type II, Gel or MR, SBA
Resins
a = 2.98 (±10% at .005 N) (42)
123
-------
For all strong base Anion Resins Regardless of Type or
Porosity
=3.14 (+20% independent of total cone.) (43)
Note: The N2POSITN dummy variable is indirectly related to
the fundamentally important variable—distance of
functional group separation. For polystyrene resins
with -pendant amine groups, containing one nitrogen
atom this distance is expected to be randomly
distributed, whereas it is controlled in polymers
with the functional groups incorporated into the
continuous structure by preselecting the amine
monomers used in polymerization. Furthermore, it is
conceivable to control the nitrogen separation dis-
tance and hence, the multivalent ion selectivity,
in any new polymer, polystyrene or non-polystyrene.
aS = Sulfate/nitrate separation factor
N
aN = Nitrate/chloride separation factor
\_oj_
** N2POSITN = 1.0 for resins with nitrogen in
the matrix (i.e., non-polysty-
rene resins)
**N2POSITN =0.0 for resins with pendant nitrogen
(i.e. polystyrene resins)
RSIZE =2.0 for polyamine resins
RSIZE = 2.19 for tertiary amine resins
RSIZE =2.36 for quaternary amine resins
XLINKING =0.5 for isoporous resins
XLINKING =1.0 for microporous resins
XLINKING =2.0 for macroporous resins
124
-------
TABLE 17
Variables Influencing Nitrate/Sulfate Selectivity a
N
N
++ = Greatly Increases ac
N
+ = Increases a_
N
0 = No Significant Effect on a_
XT "
= Decreases a_
N
= Greatly Decreases ag
N/A = Not applicable
Nitrogen in Polymer Backbone
Increasing "R" Group Size
Increasing Degree of Crosslinking
Macroporous as Opposed to Microporous
Isoporous as Opposed to Gel or MR
Increasing Capacity
Type I as Opposed to Type II
Increasing pKa
CO
c
•H rH
in to
0) H
cS
o c
•H
c c
(1)
--
++
0
0
N/A
—
N/A
+
H
< n
M vH
•P CO
(0 0)
^1 tf
•H
O
PH
(4)
N/A
++
0
0
N/A
N/A
N/A
0
8
2
^f
1 1
CO (U
rH ItJ
O«M
04 1 C
1 X-rH
c men
SIS
(5)
N/A
++
0
0
N/A
N/A
N/A
++
(U
(0
n)
CQ
H tn W
C C
rP d)
E-< W P!
(6)
N/A
N/A
+
0
__
N/A
N/A
N/A
Interpretation of Table:
Maximum Nitrate/Sulfate Selectivity is Associated with:
(1) Polystyrene, quaternary, low capacity, anion resins
(2) Non-isoporous, type I, strong-base anion resins
(3) Polystyrene, tertiary, low capacity weak base anion resins
125
-------
TABLE 18
N
Variables Influencing Nitrate/Chloride Selectivity a-,.
N
++ = Greatly increases a.,.
N
+ = Increases a ,
N
0 = No Significant Effect on ar,1
= Decreases a-,.
N
= Greatly decreases a_..
N/A = Not applicable
Nitrogen in Polymer Backbone
Increasing "R" Group Size
Increasing Degree of Crosslinking
Macroporous as Opposed to Microporous
Isoporous as Opposed to Gel or MR
Increasing Capacity
Type I as Opposed to Type II
Increasing pKa
m
c
•H iH
in a;
CO
<0
ffl to
1 C
D^-H
C M
O 0)
M tt
4J
Cfl
(2)
N/A
N/A
0
0
0
0
0
N/A
<0
to co
<0 C
ffl-H
1 CO
id K
(D
S
(3)
__
0
+
+
N/A
0
N/A
0
a)
c
0
M en
>i C
-P -H
CO CO
>i 0)
r-\ PS
0
ft
(4)
N/A
-
0
0
0
0
N/A
-
<1)
oj
S-l
>
4J
CO V
>i CO
rH rt
o mn
&4 1 C
1 AJH
C RttO
O (1)0)
!S&«
(5)
N/A
0
++
++
N/A
0
N/A
0
0)
en
(0
0
i
H cruo
c c
1-p (1)
EH Cfl K
(6)
N/A
N/A
0
0
0
0
N/A
N/A
Interpretation of Table:
Maximum Nitrate/Chloride Selectivity is Associated with:
(1) Polystyrene anion resins (Porosity and crosslinking aren't
included because although they are relevant to anion resins
in general, they are irrelavent to a^for polystyrene resins)
(3) Polystyrene Weak base resins
(5) Macroporous (highly crosslinked) weak base anion resins
126
-------
SECTION 6
PHASE II: MULTICOMPONENT CHROMATOGRAPHIC COLUMN STUDIES
OBJECTIVES
To determine if the separation factors (cu) developed from
binary equilibrium experiments can be used to predict the
chromatographic behavior of the nitrate, sulfate, chloride and
bicarbonate anions.
To determine whether or not differences exist in the
quality of the effluent waters from the two types of processes,
i.e., to compare single-bed, strong-base anions resin process
performance to two-bed strong acid, weak-base process perfor-
mance in chromatographic elution to nitrate breakthrough.
To determine the maximum possible chemical efficiencies
and actual overall chemical efficiences of various modifica-
tions of these two types of processes.
To determine if nitric acid and ammonium hydroxide can be
used for cation and anion bed regeneration so that the waste-
water, mostly ammonium nitrate, might be disposed of as a
fertilizer.
To establish the comparative seriousness of the iron
fouling problem in the single-bed and two-bed processes and to
determine how it is influenced by the type of regenerant used.
127
-------
To establish which of the thirty-two anion resins tested
are best for nitrate removal service by determining which resin
characteristics are most influential in maximizing the overall
process efficiency.
PROCEDURAL OUTLINE
C 1ST
(1) Select resins for column studies based on
-------
KJ
-CX-
Two
Plexiglas
Columns
2.54cmI.D.
1 .52m. (ong
Resin Depth
61 cm (Typ.)
i—XK
Feed water
Pump
0-450 ml/min
1001
Artificial
Ground -
water
c
A
T
I
0
N
•
C
o
L
U
M
N
To Waste
[Acid Pump
O-5Oml/min
A
N
I
0
N
*
C
0
L
U
M
N
Syphon
Break
Automatic Sampler
(24; 500 ml Bottles)
To Waste
NH4OH Pump
0-20ml/min
4%
NH4OH
1.14 N.
pH Meter
Strip Chart Recorder
tx= N.O. Valve
M= N.C. Valve
Figure 24 Experimental Column Set-up.
-------
(4) Choose range of operating conditions. The Dow [41],
Duolite [37], and Amberlite [104] Manuals were consulted as
were the manufacturers data sheets on the resins used. The
following representative conditions were chosen for the experi-
mental column runs:
Exhaustion Rate: 2.5 to 5.0 gal/min ft
3.0 to 1.5 min. superficial detention
time
103 ml/min in 1" dia x 24" deep column =
2.5 gal/min ft3
Backwash Rate: Sufficient to get 30-100% bed expansion
Backwash Time: Typically 5-10 minutes
Regenerant Rate: 0.25 to 0.50 gal/min ft3
10-20 ml/min, Downflow in 1" dia x 24"
deep column typical
100 ml/min, downflow in 2 1/2" dia x 30"
deep column typical
Minimum Regenerant Contact Time: 45 minutes
Regenerant Concentrations: 1.5 N HNO.,, 9.0% HNO,.
1.5 N HC1, 5.4% HC1
1.14 N NH4OH, 4.0% NH4OH
Regenerant Direction: Usually downflow, but upflow (HC1)
attempted for efficient cation
regeneration and Fe(OH)- removal
Displacement Rinse Rate: Same as regeneration flow rate
Displacement Rinse time: Sufficient to displace regener-
ant, typically equal to regener-
ant contact time.
Final Rinse Rate: Theoretically equal to service (exhaus-
tion) rate, typically used 1/5 service
rate overnight
Final Rinse Volume: Theoretically, 5-20 BV
Typically, 20-50 BV
(5) Establish compositions of artificial ground waters to
be used and procedures to make up 100 £ batches of these waters.
See tables below and discussion on "Test Water Composition" in
following section.
130
-------
TABLE 19. TEST WATER 1 FOR RUN 1
Na TEST WATER FOR PRELIMINARY ACID ELUTION RUN*
xi
1.0
.30
.20
.30
.20
Total Cations
Ion
Na+
so4=
N03
Cl
HC03~
& Anions
meq/1
5.0
1.5
1.0
1.5
1.0
5.0
ppm
115
72
62
53
61
363
C_ = 0.005 N = Total Concentration
NO
3-N = 14 ppm, XN = 0.20
Note: For actual chemical composition of test waters, see
Appendix Tables D2 and D3.
* Acid elution of the anion bed means that acids not neutral
salts were fed to the anion bed during the entire run, i.e.,
the H+ ion exchange capacity of the cation bed was greater
than the OH~ ion-exchange capacity of the anion bed. Neutral
elution means that neutral salts were fed to the anion bed
during single-bed operation or that the H+ ion-exchange
capacity was equal to or less than the OH ion-exchange capa-
city of the anion bed during two-bed operation.
131
-------
TABLE 20. TEST WATER 2 FOR RUNS 2 THROUGH 8
Na TEST WATER FOR TWO-BED ACID ELUTION RUNS AND
SINGLE-BED NEUTRAL ELUTION RUN
xi
1.0
.27
.27
.27
.18
Total Cations
Ion (i)
Na+
so4~
NO3~
Cl
HCO ~
& Anions
meq/1
5.5
1.5
1.5
1.5
1.0
5.5
ppm
126.5
72.0
93.0
53.2
61.0
406
CT = 0.0055 N = Total Concentration
NO
3-N =21 ppm; XN = 0.27
TABLE 21. TEST WATER 3 FOR RUNS 9 THROUGH 11
Ca-Mg-Fe TEST WATER FOR TWO-BED NEUTRAL ELUTION RUNS
X.
Cations <
.54
.27
.18
.Nil
J'27
Anions *\
.27
»
Total Cations
Ion
4.4.
Ca
4.4.
Mg
4.
Na+
Fe++
so4=
NO3
Cl
HCO3~
& Anions
meq/1
3.0
1.5
1.0
Nil
1.5
1.5
1.5
1.0
5.5
ppm
60
18
23
1
72
93
53.2
61
381
CT = 0.0055 N;
N = 21 ppm, X._ = 0.27
Hardness = 225 ppm as CaCO.
N
132
-------
Note: For actual chemical composition of test waters, see
Appendix Tables D2 and D3.
(6) Do preliminary test of system — Run 1. This first
run established that sharp effluent profiles could be produced
in this flow system at 2.5 gal/min ft (T = 3.0 min) exhaustion
rate with a bed depth of 25 inches (63.5 cm) and a total con-
centration of .005 N. Furthermore, a complete effluent profile
could be produced in 1000 bed volumes (50 hours) with the resin
of choice which at this point was Duolite ES-368 a MR, STY-DVB,
tertiary amine resin with a capacity of 1.4 meq/ml and a parti-
cle size distribution favoring the smaller particles (30-40
mesh). See Column Effluent Profile, Run 1, Figure Cl (Appendix)
(7) Using acid elution to eliminate possible low-capacity
effects, compare three, weak-base resins with very different
selectivities. A flow rate of 5 gal/min ft (T = 1.5 min) was
chosen to speed up the tests as the capacity of one resin
chosen was 3.0 meq/ml, nearly double that of Run 1. This also
offered the opportunity to determine the effect of exhaustion
rate on efficiency. The following resins with wide ranging
selectivities were chosen for these initial comparisons:
S N
Resin (X <*
Duolite ES-368 2.83 (low) 3.87 (high)
Duolite ES-374 94.0 (high) 3.85 (high)
Dowex WGR 137 (high) 1.99 (low)
(8) Again, using acid elution to eliminate hydroxide ion
interference, compare effluent profiles of resins with highly
different sulfate selectivities but similar chloride selectivi-
ties at a relatively low flow rate: 2.5 gal/min ft (T = 3.0
Resins chosen were
S N
Resin « a
Duolite ES-368 2.83 (low) 3.87 (high)
Duolite ES-374 94.0 (high) 3.85 (high)
133
-------
(9) Establish the column performance of the single-bed
strong-base anion system at the lower flow rate: 2.5 gal/min
ft3 (T = 3.0 min). A macroreticular SBA resin from the least
sulfate selective category was chosen -for this run:
„ . S N
Resin aN a_.
lonac AFP 100 1.76 2.97
(10) Attempt a neutral elution of the two-bed system by
balancing the cation and anion bed capacities by tailoring the
regenerant level of the cation bed. Amberlite IRC-120 cation
resin and Duolite ES-368 anion resin were utilized at the lower
flow rate, 2.5 gal/min ft . For these Runs (8, 9 and 10), the
true, simulated Ca-Mg-Fe groundwater was used to exhaust the
cation unit before it was regenerated at the level specified
prior to being used in the two-bed run. The following regen-
eration levels based on the anion bed capacity were achieved
for the cation bed which had a theoretically higher capacity
(40% higher TEC) than the anion bed.
Run Two-Bed Regeneration Level
8 600% of theory
9 120% of theory
10 . 24.0% of theory
See further discussion of "Regeneration Level" under Experiment-
al Methods.
(11) Establish column performance of Amberlite IR-45 weak-
base anion resin in two-bed neutral elution service at low flow
(2.5 gal/min ft ) with Ca-Mg-Fe simulated groundwater. The two-
g
bed performance of this moderately sulfate selective resin (c*N
= 12.7) in nitrate removal service has been reported in the
literature [47] with some unusual results and so it was included
for comparison purposes.
134
-------
(12) Define ion-exchange process performance criteria on
which to base judgements regarding the suitability or non-suit-
ability of a given process or set of process operating con-
ditions. The criteria selected were EM and EQ where:
E = Maximum possible chemical efficiency in nitrate
removal service.
E_ = Observed chemical efficiency in nitrate removal
service.
See further discussion of "Definitions of Process Efficiency"
under "Data Evaluation Methods".
(13) Establish acceptable breakthrough concentration of
nitrate. See "Effluent Nitrate Breakthrough Concentration"
following. The value chosen was 6.7 ppm (0.48 meq/1).
(14) Determine empirical efficiencies (EM and EQ) of
various runs by area measurement techniques on column effluent
profiles and by observations of regeneration efficiency.
(15) Graphically compare throughputs (T) of various runs as
measures of comparative process efficiencies.
(16) Establish comparative economics based on chemical
regenerant costs, disposal costs and efficiencies (EQ).
(17) Rank the various processes and the anion resins as to
their suitability for nitrate removal from ground waters having
concentrations of ions similar to the representative Ca-Mg-Fe
test water. Consider economics, organics bleed, possible iron
fouling and regenerant waste water disposal. •
(18) Calculate theoretical efficiencies (EM and EQ) using
the multicomponent chromatographic ion-exchange theory of
Helfferich and Klein [63, 64] for test waters of similar and
different compositions than those used in the column runs.
135
-------
EXPERIMENTAL METHODS: '(See also Appendix D)
The Criteria for Resin Selection
Of the weak base resins, the least sulfate selective were
also the least chloride selective; these were all the STY-DVB,
macroporous tertiary amine resins:
Resin o o
Amberlite IRA-93 3.75 4.86
Dowex MWA-1 2.67 4.43
Duolite ES-368 2.83 3.87
lonac AFP-329 3.07 4.14
The further choice among these for a representative of what was
expected would be the best column performance for a given class
of resins was based on relatively minor differences wrt particle
size distribution and organics bleed as measured by UV adsorp-
tion. Having considered all these variables, Duolite ES-368
was chosen primarily because of its finer, more consistently
sized particles. It did however have, as did all but lonac
AFP-329, significant organics bleed as indicated by UV adsorb-
ance. Furthermore about 10% of the beads had a tendency
to float during backwash, a possible problem which would have
to be considered seriously in any large-scale installation.
Effluent NO- Breakthrough Concentration
In a bypass blending system such as the one proposed (Fig.
1), the question arises as to what range of nitrate concentra-
tions is acceptable in the blended effluent and what factor
of safety is appropriate. The legal limit is 10 ppm NO-j-N and,
an appropriate blended NO^-N concentration might be one-half
that value. Assuming an influent concentration of 20 ppm NO3~N
to the ion-exchange process the flow split would be:
136
-------
Feed = Q
Flow to Columns = .75Q
Bypass Flow = .25Q
With such a split/ one would have to limit the N03~N break-
through concentration to 6.7 ppm (0.48 meq/1) so as not to ex-
ceed the 10 ppm allowable maximum in the blended water. The
appearance of that effluent concentration then determines the
end of a run. Clearly the economic efficiency of the process is
inversely proportional to the safety factor chosen. For
example, allowing 8 ppm N03~N in the blended water would permit
40%, as compared to 25% of the influent flow to be bypassed.
However this leads to a maximum allowable breakthrough concen-
tration of 3.3 ppm (0.24 meq/1) which would necessitate earlier
termination of the run. But, from the experimental data, the
loss in bed volumes treated to 3.3 ppm NO^-N breakthrough com-
pared to 6.7 ppm NO--N would be only about 10% which is more
than offset by the 60% gain 'in permissible, bypass flow (40% by-
pass compared to 25%). Nevertheless, operation to such a low
breakthrough concentration is not recommended since some pre-
liminary leakage of nitrate might be expected as was evidenced
in Run 10 (Fig. CIO) where a preliminary NO3-N plateau reached
a level of 0.24 meq/1 and would have necessitated premature
termination of the run.
For the reasons stated above, the evaluations of process
efficiency and economics in this work based on:
21 ppm (1.5 meq/1) NO3-N in Ground Water
25% bypass of Raw Water
5 ppm NCU-N in Blended Effluent
6.7 ppm N03-N breakthrough NO-j-N concentration
137
-------
Level of Regeneration in Two-Bed System
Weak-base resins are so selective for hydroxide ions that
they are nearly stoichiometrically regenerated (110% of theory)
even with weak bases like NH.OH. This advantage is lost how-
ever, when a weak-base resin follows a strong-acid cation bed as
it must in the treatment of nearly neutral to basic water
supplies. The reasoning has to do with the required neutraliza-
tion of regener"ant wastewaters: a requirement which dictates
that there be as much excess base as acid. Since regeneration
of the strong-acid cation bed is quite inefficient (300% of
theory) especially in divalent calcium and magnesium ion re-
moval service, the NH.OH saved during efficient/ anion-bed re-
generation must be expended to neutralize the inevitable excess
acid from strong-acid cation regeneration. This does not negate
the basic reason for choosing a weak-base anion resin to solve
the regenerant disposal problems, i.e., to allow a weak base
like NH~ to be the regenerant thereby eliminating the agricul-
turally undesirable cations, Na and K from the regenerant
wastewaters and providing instead the agriculturally desirable
NH. cation.
Because of this unavoidable disparity in cation and anion
bed regeneration efficiencies, a regenerant design procedure
based on the capacity of the anion unit was utilized. Once the
exchange capacity of the anion bed was specified, an amount of
cation resin was provided which had a total equivalent capacity
(TEC) at least 20% greater than that of the anion bed. The
level of regeneration specified was then expressed in terms of
per cent of theoretical anion bed capacity. That regeneration
level was applied to both beds assuming they were of equal
(anion bed) capacity. A very large excess of regenerant cannot
be applied indiscriminantly because once the capacity of the
anion bed is utilized completely by the strong acids: HC1,
HNO3 and H2SO4, its effluent will then be very acidic (pH = 2.4
138
-------
for these 0.005 N test waters) and that condition will always
occur before nitrate breakthrough as chloride is always the
first strong-acid anion to appear in the effluent. On the
other hand/ if insufficient acid regenerant is applied, the
cation bed will be exhausted before the strong acid capacity of
the anion bed is utilized. This results in a neutral influent
to the anion bed and a much reduced anion bed capacity in addi-
tion to a probable change in the separation factors among all
the ions of interest? both of these changes substantially in-
crease the difficulty of predicting the breakthrough profiles.
The difficulty of efficiently balancing the acid and base re-
generants is admittedly a disadvantage with any two bed ion-
exchange process, nevertheless it is not overly difficult and
once solved for a given installation should remain solved as
long as the composition of the ground water and the capacities
of the resins remain relatively constant.
Sample Collection During Column Runs
Automatic sampling from the one liter overflow vessel
(Fig. 24) was accomplished using a Manning Wastewater Sampler;
samples were usually taken once/hr except during Run 6 when
the rate was once/2hr. Typical length of a run was 48 hours
except for Run 6 which lasted 100 hours. Flow rates were
either 20 BV/hr (Runs 1 & 5-11) or 40 BV/hr (Runs 2-4) and of
such duration as to give 800-2000 BV of cumulative total flow.
Carbonic Acid & Bicarbonate Analyses During Runs
Because H2C03 and HCO~ weren't removed to any appreciable
extent by the anion resins/they appeared quite early in the
effluent profiles. For all the two-bed runs, H2CO3 was the
dominant species (effluent pH 2.3 — 4.6) and was easily lost
upon exposure to the atmosphere. To eliminate the loss due to
CO, volatilization, hourly, manual samples were taken directly
139
-------
from the flowing effluent stream by syringe and immediately
injected into the inorganic channel of the Beckman TOG analyzer
for CO- analysis. Once the H-CO., effluent profile had been
established, hourly samples were stopped and only an occasional
sample was taken. All H-CO., and HCO., analysis were accomplished
in this manner irrespective of the effluent pH.
DATA EVALUATION METHODS: DEFINITIONS AND MEASUREMENTS OF
PROCESS EFFICIENCIES
Maximum Possible Chemical Efficiency
Maximum possible chemical efficiency (E ) is defined simply
as yN the average equivalent fraction of nitrate on the resin
at the end of the run. Since yN varies with distance into the
bed, the weighted average value y,,, must be used to represent
the ratio of nitrate removed to all ions removed. In the
ideally efficient process this would of course approach 1.0
which would only be possible if nitrate were much preferred
over all other anions which it is not in these experiments.
meq NO- on resin at end of run
meq of ions on resin at e
meq NO~ in - meq NO-, out
Initial meq of all ions + meq of all ions In"
- meq of all ions"out
E = v = 1 (44)
M -*N Total meq of ions on resin at end of run
ci,ove -
yl = ?N -
Cidv
(46)
where subscripts: 1 = NOl, 2 = SO., 3 = Cl~, 4 = HCO~ and
C^ Q = influent (initial) phase cone, of component i and all
other symbols are as defined in the nomenclature. Run 7 is a
special case as regards determination of E,. since the resin was
JM
initially saturated with the exchanging counterion , Cl", which,
140
-------
upon exchange, was accounted for in the effluent. This is
contrasted to all the other runs where the resins were ini-
tially in the free base (FB) form and no measureable counter-
ions were released upon acid adsorption (or, alternatively,
upon ion-exchange where OH~" was considered the counterion which
was neutralized upon exchange). For Run 7 then:
ci,ove -
yi = *N = - 4
Qclv + I VeC,
,0
(47)
where QCIV = chloride capacity of bed = . #ml. Example
calculations of efficiencies (EM) for Runs 7 & 11 are given
M
in Appendix E.
Overall Chemical Efficiency
Overall chemical efficiency (Er.) is product of the
\->
maximum possible chemical efficiency (E..) and the observed
regeneration efficiency (ER) .
(48)
„ _ ^. „__. . mea total capacity of anion bed
ER = Regeneration Eff1C1ency = *^q anion regenerant applied
(49)
meq NOZ removed
EQ = Overall Chemical Efficiency = meq anion regenerant applied
(50)
Ordinarily E_. is near 0.9 because of the ease with which weak
R
base resins are regenerated but, as has been pointed out in
the "Level of Regeneration" discussion, the need to neutralize
the excess cation regenerant greatly reduces the overall regen-
eration efficiency.
Summarily, the lion's share of ion -exchange operating costs
141
-------
will be for regenerant chemicals and possibly for their dis-
posal. Each equivalent of nitrate removed from the water
supply will require I/EO equivalents of anion and, if appli-
cable, cation-regenerant chemicals. Three procedures were
studied here to improve the overall efficiencies of these
processes:
S N
(1) selection of resins with low aN and high a ,.
(2) chromatographic elution of the lesser preferred ions,
HCOl and Cl~, to increase yN, the average equivalent fraction
of nitrate on the resin at the end of a run and
(3) minimization of the excess regenerant utilized.
The maximum possible chemical efficiency, E , is very much
dependent on the ionic composition of the water to be treated
over which no control can be exerted in actual practice. With
Test Water 1, Table 19, for example, xg = 0.3, XN = 0.2 and EM
= 0.4 if no chloride or bicarbonate are removed while all
sulfate and nitrate are removed. Slightly better efficiency
is possible with Test Water 2 Table 20 where xg = .27, XN =
0.27 and E., = 0.5 if all the nitrate and sulfate and none of
M
the bicarbonate and chloride are removed. Even though these
efficiencies are moderately low, they would be much lower in
conventional deionization or ion-exchange service where the run
would terminate on conductivity breakthrough or when the the-
oretical capacity of the anion bed was exhausted. In these
instances chloride and possibly bicarbonate would still occupy
a significant portion of the exchange sites and E for Test
Water 1 could be as low as 0.2 or as low as 0.27 for Test Water
2. Generally the maximum possible chemical efficiency (E ) has
XN/ the liquid phase equivalent fraction of nitrate, as its
lower limit corresponding to the complete deionization case
without any chromatographic elution.
142
-------
VISUAL INTERPRETATION OF COLUMN EFFLUENT PROFILES
Plate aus and PLate au Con centrations
The effluent concentration vs bed volumes of effluent
curves for all eleven runs are plotted in Figures C1-C13 (appen-
dix C). Consider RUN 1 as typical of the general effluent
behavior of the four anions of interest and note that, as pre-
dicted, there are four plateaus each corresponding to one of
the anions, and that these plateaus are separated by rather
abrupt transition zones. The first component to appear is
always H2CO3 or HCO~ followed by Cl~, NO~ and finally SO~ the
most preferred species. Observe also that, as expected, all
species save for the most preferred SO^ appear at some time
in the effluent in concentrations from 20 to 300% higher than
in the feed water (CQ). Abrupt increases in concentration of
one component are always accompanies by a correspondingly
abrupt concentration decrease in a second component once the
H_CO3 has been eluted and true ion-exchange is maintaining the
total, liquid, effluent concentration at approximately that of
the influent concentration: 0.005 N for RUN 1 and .0055 N for
all others.
Nitrate Breakthrough Profiles
In all the low flow rate runs (2.5 gal/min-ft ), the
nitrate breakthrough curves are quite sharp but not vertical
meaning that the end of the run is rather abrupt but not with-
out warning - a desirable feature for a full-scale, nitrate
removal installation. It is also consistently observed that
some preliminary, though minor, breakthrough of nitrate occurred
with Duolite ES-368; see Runs 1, 5, 8, 9, and 10, all at 2.5
gal/min ft3. The problem is at its worst in Run 9 where the
influent became neutral prematurely due to incomplete regenera-
tion of the cation bed (120% of theory) and nonexistent under
143
-------
the high flow rate (5 gal/min ft ) acid elution conditions of
Run 3.
Recall that the nitrate breakthrough concentration has been
conservatively chosen as 0.48 meq/1 (6.7 ppm NO.,-N) and that
the effluent volume at that point is labelled Ve indicating the
end of the run for nitrate removal service.
DISCUSSION OF COLUMN RUN RESULTS
Phase II Data Summary; Column Performance Characteristics
The important results from the column runs are listed in
Table 22 below. Five different resins with sulfate/nitrate
selectivities varying in the range of 2.83 to 137 and having
nitrate/chloride selectivities in the range of 1.99 - 3.87
were evaluated at two different superficial detention times:
2.44 gal/min ft in 61 cm deep beds (T = 3.1 min) and 4.88
gal/min ft in 30.5 cm deep beds (T = 1.5 min). Runs 1-6
were acid elution experiments where the cation bed was much
larger than the anion bed to insure constant capacity and
provide ideal conditions for the prediction of efficiency and
S N
the determination of the effects due to varying aN and
Run 7 was a single strong-base anion column run for the purpose
of comparing the performance of this currently used NaCl-regen-
eration process to the proposed two-bed system. Runs 9-11 were
neutral elution runs with a Ca-Mg-Fe containing groundwater
under conditions which simulated as closely as possible those
expected in a full-scale, nitrate removal installation on
groundwater.
Factors Influencing E..; Maximum Possible Chemical Efficiency
Range of Variation of E (y )
There is a surprisingly narrow range of efficiencies
144
-------
TABLE 22: COLUMN PERFORMANCE CHARACTERISTICS
Run
No.
1
2
3
4
5
M
*> 6
ui
7
8
9
10
11
Flow
min.ft3
2.34
4.88
4.88
4.88
2.44
2.44
2.44
2.88
2.44
2.44
2.44
Minimum
PH
Final
Ph
2.5
2.5
2.5
2.5
2.4
2.4
2.5
2.5
2.4
2.4
2.3
2.5
6.1
7.4
2.8
5.8
4.5
6.7
4.6
O
4.7
5.5
Bed
Depth
cm
63.5
30.5
30.5
30.5
61.0
61.0
61.0
61.0
61.0
61.0
61.0
Resin Description
(Cation Regeneration Level)
Duolite ES-368
STY-DVB, Tert-Amine, MR
Duolite ES-374
Polyacrylic, Polyamine, MR
Duolite ES-368
STY-DVB, Tert. Amine, MR
Dowex WGR
Epoxy-Amine, Polyamine, Gel
Duolite ES-368
STY-DVB, Tert. Amine, MR
Duolite ES-374
Polyacrylic, Polyamine, MR
lonac AFP-100
STY-DVB, Quat.(I)Amine, MR
Duolite ES-368 (600%)
STY-DVB, Tert. Amine, MR
Duolite ES-368 (120%)
STY-DVB, Tert. Amine, MR
Duolite ES-368 (240%)
STY-DVB, Tert. Amine, MR
Amber! ite IR-45 (300%)
STY-DVB, Polyamine, Gel
S
"N
2.83
94.
2.83
137.
2.83
94.
1.76
2.83
2.83
2.83
12.7
N v v v
arn JCl S •'HCO-,
Cl 3
3.87
3.85
3.87
1.99
3.87
3.85
2.97
3.87
3.87
3.87
3.89
.13
.26
.20
.27
.16
.15
.14
.21
.31
.14
.08
.53
.36
.40
.37
.43
.44
.43
.40
.34
.44
.45
.00
.02
.01
.00
.00
.00
.01
.00
.02
.00
.03
**
Final
Column
yN Capacity
meq/ml
.34
.36
.39
.36
.41
.41
.42
.39
.33
.42
.44
1.65
2.93
1.36
1.62
1.48
3.12
1.03
1.39
0.84
1.15
1.61
t
Ve
BV
582
720
364
391
423
920
295
375
190
334
480
ir , __._. . _ *i v 7 /Lft — Qi mav»-F i r* -i a 1 Ha+an+ ir^n -Hmo i- m-* «i i 4-or
gal/min.ft
t Ve = Bed volumes of effluent to 0.5 meq/1 (CL-breakthrough (end of run)
** Final Column Capacity is greater than measured HC1 capacity because resin has higher capacity for sulfate which occupies
a significant fraction of the available sites at the end of the run.
Final pH refers to the pH of the system effluent at nitrate breakthrough.
Minimum pH was the minimum pH observed during the course of the run.
-------
(E = y ) among all the runs where valid comparisons might be
made. It is not intuitively obvious that such small variations
in the nitrate content of the spent resins should result from
such large differences in a.,. However, as it turns out/ a-,
is more important in determining yM than is a , and with a much
N
smaller range of values existing for ou, the narrow range is not
surprising after all.
Breakthrough Volume (Ve) and Bed Capacity (meq/ral)
Here is where the largest variations are found among the
resins. Simplified theoretical considerations dictate that
capacity shouldn't influence the chemical efficiency of these
processes because the important factor is E where:
„ _ meg nitrate on resin at end of run /..,
M Total meq of ions on resin at end of run
which is independent of capacity per se. However, the practical
considerations of bed size required and rinse volume required
definitely favor the high capacity resins which permit smaller
resin beds and less rinse volumes to be used. An exception to
this has been reported (D. Harrington, Dow Chemical Co., per-
sonal communication) for highly sulfate selective resins
(especially epoxy-amine resins) which require increasingly
larger rinse volumes with time, an effect reportedly occurring
only in waters where sulfate represents a large fraction of the
total anions present.
Nitrate Selectivity vs. Column Efficiency (EM or yM)
Q M
The effects of both aN and acl can be determined by com-
paring runs in which these were the only variables; c.f. Runs
2, 3, and 4; Runs 5 & 6; Runs 7, 10 & 11. Direct comparisons
among these runs are provided in the throughput graphs (Figs.
25 - 27) and by comparing efficiencies (E.. or yN) in Table 22
for these groups of resins. Runs 2, 3, and 4 yielded the most
non-ideal effluent profiles and this lowest efficiencies for
nitrate: 0.36, 0.39 and 0.36 respectively. The greatest
146
-------
portion of the differences in these efficiencies is attributed
to kinetic not thermodynamic considerations. The average
equivalent fractions of chloride on the resins were highest for
this flow condition (4.88 gal/min-ft ); the short columns (30.5
cm) and detention times (T = 1.53 min) apparently promoted
chloride removal beyond that expected due to equilibrium separa-
tion. Duolite ES-368 with low ajj and high acl did provide the
highest efficiency (Run 3, EM = .39) but the trend of the data
among the runs was not consistent. Dowex WGR was expected to
be the worst due to its high af, and low ar, but it performed
S N
equally as well as Duolite ES-374 with high aN and high acl.
See Figure 25. Again, these obviously non-ideal results with
Runs 2, 3, and 4 are attributed to the short columns and short
detention times.
The nitrate effluent breakthrough profiles of Runs 5 and 6
(Fig. 27) are nearly identical which was, at first, surprising
c
for resins with such different sulfate selectivities:
-------
8
m
8
: cJ
s
. 8
5 oi
cc
oc
l_
5 g
(_) .
z —
8
*» n
co O
8
B
El
fc
g
8
A =
h-
COLUMN RUN 2
COLUMN RUN 3
COLUMN RUN U
PI ^ f*^
J&
, a,=1.99. yN=0.36
-a^ = 2.83,a~=3.87, yN=0.39
^.00
0.50 1.00 1.50 2.00 2.50 3,00
T = THROUGHPUT * EQUIVflLENTS SOLUTION / EQUIVflLENTS EXCHflNGER
FIGURE 25
COLUMN EFFLUENT PROFILES (NITRflTE)
EFFECT OF SULFflTE flND CHLORIDE SELECTIVITIES ON COLUMN EFFICIENCY
3.50
-------
§
*
m
8
cJ
. §
5 oi
CE
OC
§L- +
fc
o^
0.00
COLUMN RUN 7
COLUMN RUN 10
COLUMN RUN 11
= 2.83, y = 0.42
0^ = 12.7,7^0.44
0.50 1,00 1.50 2.00 2.50 3,00
T = THROUGHPUT = EQUIVflLENTS SOLUTION / EQUIVflLENTS EXCHflNGER
FIGURE 26
COLUMN EFFLUENT PROFILES (NITRflTE)
EFFECT OF SULFflTE SELECTIVITY ON COLUMN EFFICIENCY
3.50
-------
G> = COLUMN RUN 5
COLUMN RUN 6
(4 = 2.83, yN = 0.41
0.50 1.00 1.50 2.00 2.50 3.00
T = THROUGHPUT = EQUIVHLENTS SOLUTION / EQUIVRLENTS EXCHflNGER
FIGURE 2?
COLUMN EFFLUENT PROFILES (NITRflTE)
EFFECT OF SULFflTE SELECTIVITY ON COLUMN EFFICIENCY
3.50
-------
terminated upon chromatographic elution of the ions less pre-
ferred than nitrate.
Exhaustion Rate, Bed Depth and Detention Time
Only one actual flow rate (103 ml/min) was employed in all
3
the runs. This corresponded to 20 BV/hr or 2.44 gal/min ft
in the deep, 24 in. (60.5 cm) beds and to 40 BV/hr or 4.88
gal/min ft in the shallow beds - 12 in. (30.5 cm). Very
significant differences resulted from varying the bed depth.
See especially Runs 2 and 4 (Figs. 25, C2 and C4) and note the
non-ideal effluent profiles. Both these Runs were made with
the kinetically slower microporous resins compared to Run 3
made at the same bed depth but with a macroporous resin. Com-
pare Runs 2 and 6 (Fig. 28) and note the drastic change in
sharpness and ideality of the effluent profiles which was
produced with this high capacity, microporous resin as a result
of increasing bed depth from 12 to 24 inches with a correspond-
ing increase in actual detention time from 0.75 to 1.5 minutes
(assuming a bed porosity of 0.5). These short fluid detention
times are misleading; the real effect is the doubling of the
length of the run from 50 to 100 hours thereby allowing much
more time for most of the bed to come to equilibrium.
For the development of reasonably ideal effluent profiles,
an exhaustion rate of 2.5 gal/min ft or less is recommended.
Rates higher than this reduce the chemical efficiency, yN/ in
nitrate removal service by allowing much more chloride to remain
in the bed; c.f. the y>T values between Runs 2 and 6 and between
N
Runs 3 and 5 below:
3 °m y
Run gal/min ft Bed Depth j_N
2 4.88 30.5 .36
6 2.44 61.0 .41
3 4.88 30.5 .39
5 2.44 61.0 .41
151
-------
to
-------
These changes due to exhaustion rate (bed depth) might not seem
large but they are larger than those produced by the selectivity
differences of the commercial anion resins used in these experi-
ments .
Regeneration Level vs. Efficiency (EM): Runs 8-11
A significant reduced column capacity and efficiency
(yN = 0.33) resulted from the low regeneration level (120%
of theory) of Run 9. Note also the unusual nitrate profile
with a significant premature nitrate peak which forced termina-
tion of the run at 190 BV. The final capacity was only 0.84
meq/ml compared to 1.48 meq/ml for the acid elution run (Run 5)
and 1.15 meq/ml for the successful neutral elution run (Run 10).
Run 8 represented an unsuccessful attempt at neutral elu-
tion: regeneration level = 600% of theory for Duolite ES-368.
The cation bed consequently had too much capacity and didn't
break through soon enough. Thus, the effluent pH from the
system dropped to 2.8, an unacceptable level in actual practice.
Run 10 was a successful neutral elution run where the
regeneration level was 240% of theory which resulted in an
efficiency (EM) of 0.42 and a final column capacity of 1.15
meq/ml for Duolite ES-368. The pH never dropped below 4.6,
the carbonic acid pH, and could have been raised to near neutral
by degasification to remove C02.
Run 11 was also a successful neutral elution run, this time
with Amberlite IR-45, a STY-DVB, polyamine resin with a much
higher sulfate selectivity (a? = 12.7) than the STY-DVB tertiary
S
amine resins represented by Duolite ES-368 (aN = 2.83). Further-
more, IR-45 is a gel resin with a higher, measured HC1 capacity
(1.76 meq/ml) than the macroporous ES-368 resin (1.43 meq/ml).
At a regeneration level of approximately 300% of theory, 480 BV
of Ca-Mg-Fe test water could be treated with an efficiency (yN)
153
-------
of 0.44 and a final column capacity of 1.61 meq/ml. This was
overall, the best performance of a weak base anion resin in
neutral or acid elution if only EM was considered as the per-
formance criterion.
All three weak-base resins with high nitrate/chloride
selectivity (otc, = 3.85) performed acceptably in nitrate removal
service irrespective of their sulfate/nitrate selectivities.
Although no direct comparisons are possible from the experimen-
tal data the trends are clear and the calculated performance
(91% of acid elution performance) of these resins in the pro-
posed, two-bed system is summarized in Table 23 below:
TABLE 23. CALCULATED COLUMN PERFORMANCE OF WBA RESINS IN
NITRATE REMOVAL SERVICE ON TEST WATER 3 ASSUMING 300%
REGENERATION LEVEL
Final
Resin
STY-DVB, Tert. Amine, MR
STY-DVB, Polyamine, Gel
Polyacrylic, Polyamine/ Gel
Column Capacity
meq/mA
1.35
1.61
2.84
BV
Treated
370
480
840
EM
..42
.44
.41
Clearly, there are minor differences among the efficiencies but
large differences in capacities and bed volumes treated. As
previously discussed, capacity per se doesn't influence the
operating cost of a system except through the volume of rinses
required per unit volume of water treated which should be high
for low capacity resins. Nevertheless, if the highly sulfate
selective resins do indeed require progressively longer rinses
with time in service, then the capacity advantage is lost. A
safe compromise might be the polystyrene polyamine resin with
j
moderately high sulfate selectivity and the highest, maximum
possible chemical efficiency (E = .44).
154
-------
Bicarbonate Removal vs. Efficiency—
There was never any significant bicarbonate removal by the
resins under any of the experimental conditions examined in
this study. This was expected from the knowledge of the weak
nature of carbonic acid (pKa = 6.3); the uncharged H2CC>3 species
dominated at pH's below 6.3. Even during Run 7, the single-bed
strong base anion run, there was only 1% HCO3 on the resin at
the end of the run. This indicates the low selectivity the
resin had for the HCOZ anion which was the dominant species
during that run in which the pH of the feedwater was 7.4.
The fact that bicarbonate was nearly completely eluted
before the nitrate breakthrough occurred enhanced the maximum
possible chemical efficiency, EM, for nitrate removal by re-
ducing the total number of species present on the resin at the
end of the run. Nevertheless, all resins removed H2CO3 com-
pletely from the first 40 to 200 bed volumes of effluent at
5 gal/min ft3 and some initial removal of HCO3 took place in the
single bed, chloride form run (Run 7). When H2C03 breaks .
through, the pH drops to near 4.5 which is unacceptably corro-
sive for a water supply. Usually, in a two-bed system, a
degasifier would be installed between the cation and anion
beds to remove CO0 under the very acidic conditions produced by
4*
the mineral acids present. This may not be good practice here.
A better location would probably be following rather than pre-
ceding the weak-base anion bed. Some beneficial kinetic effect
due to the presence of H2CO3 in column experiments has been
observed here and reported (I. Abrams, Diamond Shamrock Chemical
Co., Personal Communication). Apparently in column operation
H2CO3 is neutralized by the weak-base anions, whereupon the HCO~
anions are taken up thereby swelling the resin beads in the
lower reaches of the bed where they compete with no other anion;
finally, the swollen, bicarbonate-form resin takes up the next
most preferred species (chloride) by rapidly exchanging the HCO~
for it. Simply stated, the bicarbonate anion is a catalyst for
155
-------
the removal of the more preferred species in ion-exchange column
operation. Thus it is questionable whether removing CO2 from
waters before anion-exchange in nitrate removal is good or bad
design since the closer the approach to equilibrium the more
chemically efficient is the operation of this process.
Comparative Process Economics
For a continuous, single-bed strong-base anion exchanger
in nitrate removal service Holzmacher [66] estimated that the
NaCl regenerant cost, plus the cost of regenerant brine dis-
posal by trucking eight miles to a river before dumping, re-
presented 50% of the operating costs of the process. That
percentage can only increase when more expensive regenerants
are used as in the two-bed system or when truly legitimate
means are considered for sodium chloride brine disposal. With
the two-bed system the cost for chemical regenerants will be
higher but that increase will be offset by the nitrogen ferti-
lizer value of those regenerants. The following economic
analysis has been made with the conservative assumption that the
two-bed regenerants wouldn't be sold, rather they would simply
be given away to eliminate any disposal costs.
Table 24 compares the chemical costs of all the feasible
regenerants for use single-bed and two-bed nitrate removal
processes.
Assumptions Made in Regenerant Cost Calculations—
(1) Exhausted resin is 40% in the nitrate form at the end
of a run. E., = 0.4.
M
(2) Regenerants levels are 300% of theoretical: ER = .33.
In actual practice, sulfuric acid might have to be 400-500% of
theoretical in high calcium waters due to CaSO^ fouling and NaCl
might have to be greater than 400% of theoretical in high sul-
156
-------
fate waters.
(3) Raw water supply has approximate analysis of Test
Water 3, Table 21 where NCU-N is 21 ppm, C « .006 N, TDS - 400
ppm.
Actually EM and £„ won't change significantly at total
ru i\
concentrations up to about .06 N (TDS - 4000 ppm) if XN remains
constant, but the regenerant cost will increase or decrease
in direct proportion to the total concentration (C-) in meq/1.
(4) Chemical costs are calculated from published prices
on August 2, 1976 in the Chemical Marketing Reporter. Prices
are FOB production point, i.e., they don't include delivery
which can be significant in remote locations. However, deliver-
ed ammonia costs will not be significantly higher than indicated
due to the rather universal availability of anhydrous ammonia.
(5) Twenty-five, percent of the raw water is bypassed for
blending with the deionized water.
TABLE 24. CALCULATED CHEMICAL REGENERANT COSTS
(1 Ib-equivalent = 14 Ibs of nitrogen removed)
Regenerant
Chemical
H2S04
NaCl
NH3
HCl
NH4C1
$
Ib-equivalent
1.23
1.30
1.53
2.43
5.64
6.62
1000 gallon Treated
8.63
9.15
10.7
17.1
39.6
46.5 "
t
m Treated
2.28
2.42
2.78
4.52
10.5
12.3
Only two of the above chemicals, NaCl and NH4C1, can be used in
the single bed process while NH- and either HCl, HN03 or H2SO4
would have to be chosen for the two-bed process. Clearly, HNO-
157
-------
is very expensive compared to the other acids; furthermore, its
use makes possible disastrous nitrate pollution of water
supply should errors be made in valve switching during regenera-
tion. In addition, the experimental runs demonstrated that in-
ordinately long cation bed rinses would be required to bring the
NO->-N concentration to below 1 ppm. Thus, nitric acid is not
recommended even though it would much enhance the fertilizer
«
value of the regenerants. Sulfuric acid is more economical,
but if it must be used in 500% rather than 300% xs that advan-
tage is lost to the relatively more expensive but more efficient
HC1 in calcium ion elution.
The further comparisons between the two-bed and single-bed
processes in Table 25 are made with the following assumptions
in addition to those already listed:
1) HC1 and NH.OH regenerants for the strong acid-cation
and weak-base anion beds respectively.
2) NaCl regenerant for the strong-base anion column with
disposal cost equal to regenerant cost. See
Holzmacher [66, p. 212].
3) Regenerant volume for disposal comprises the actual
regenerant plus one bed volume of displacement rinse.
If all the rinses are collected for disposal and a
value of 50 gal/ft (6.7 BV) is assumed for the rinse
volume of each bed, the volume for disposal would
approximately double and the solids concentrations
would be correspondingly halved.
The regenerant plus disposal costs of the two-bed process
are about 50% higher than the single-bed process but it is un-
likely that the nitrate containing sodium salts from the single
bed process will be permitted to be dumped at such low cost
onto agricultural land or into any receiving waters except the
oceans. A further disadvantage is that iron fouling of anion
resins is Known to be a problem in the single-bed process [10]
158
-------
TABLE 25: ECONOMIC AND REGENERANT WASTEWATER COMPARISONS BETWEEN THE SINGLE-BED
AND TWO-BED PROCESSES
Single-BedTwo-Bed
Item Process Process
Regenerant Chemical Costs, <£/1000 gal H20 Supplied. ... 9.2 27.8
Regenerant Chemical Costs,
-------
where significant reduction in efficiency and capacity have
been observed. Some observations were made on that problem
during the column experiments.
Seriousness of the Iron Fouling Problem
It was rather surprising to observe the volume of Fe (OH) 3
produced from one ppm of ferrous iron upon oxidation by the air
during experimental runs 9-11 in which the Ca-Mg-Fe test water
was used. Admittedly the conditions were somewhat different
than those which would prevail in a full-scale ground-water ion-
exchange application as 02 from the atmosphere was readily
available whereas it would be less so in a closed, full-scale
system. Nevertheless, some 02 will be unavoidably introduced
into the ion-exchange beds during the regeneration and rinsing
steps; this oxygen will readily oxidize the ferrous iron and
precipitate Fe(OH), . In the two-bed system the iron hydroxide
fouling was limited to the cation bed where it was, visibly at
least, completely removed during regeneration with 1.5 N HC1 as
would be expected. This precipitated iron tended to cement
itself, and clogged the first 10% of the cation bed and visibly
penetrated 50% of that bed by the end of the run. The problem
would have been very serious in a single-bed system if this
amount of ferric iron had been involved. However, that was pot
experimentally substantiated here as there was no ferrous iron
in the test water for the single bed run (Run 7). Beulow 110]
emphasized the potential seriousness of this problem and reiter-
ated the solution prescribed by the resin manufacturers, i.e.,
remove the iron before neutral, ion exchange. That would signi-
ficantly increase the costs associated with the single-bed
process reducing its relative cost advantage over the two bed
process.
160
-------
Organic Extractables in the Resins
Because there currently appears to be a good deal of
justifiable concern over the presence of chlorinated hydro-
carbon carcinogens in water supplies at the ppb level, the
possible presence of organics leached from these synthetic
organic ion exchangers must not be overlooked. While examining
the UV spectra of acidic resin equilibrates during a search for
possible interferences to the nitrate-by-UV method, a number
of resins were observed to have produced what appeared to be
very significant amounts of UV absorbing organics in the aqueous
phase. To verify that these absorbance peaks in the 210 to
230 nm range were, in fact, due to organics, TOG analyses were
run on 100 ml acidic (pH = 3) resin equilibrates tumbled for
20 hours with 1.00 gm of the various air-dried resins. The
results of those analyses are listed in Table 26 below.
The only obvious trend in the data is that the STY- DVB,
quaternary amine resins (Nos. 14-32) produced much less TOG
than did the weak-base resins (Nos. 1-13). This may have been
due to the chemical forms of the previously air-dried resins
which were stored in tightly-capped, polyethylene bottles
prior to the extraction experiment. The weak-base resins were
stored in the free base form while the strong-base resins were
stored in the more stable chloride form.
By far the worst resin as measured by UV contamination,
visible contamination and TOG was the aliphatic polyamine resin
lonac A-260 (No. 11). Although not indicated in the table, two
other anion resins produced visibly, yellow-colored waters at
various times during the resin conditioning and batch equili-
brium studies, these were the phenol formaldehyde polyamine
resins: Duolites ES-561 and A-7. Also, the cation resin
Amberlite IR-120 yielded an orange-colored supernatant water
when stored in the hydrogen form. The point here is not so
161
-------
much to cite particular resins as being unacceptable, but
rather to point out that, visibly or invisibly, the organic
contamination does exist with all resins to some degree at
least when they are relatively new.
TABLE 26. ORGAN-IC LEACHED FROM "CONDITIONED" ANION RESINS
U-M
Resin No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
16
17
21
32
Description
STY-DVB, Tertiary Amine, MR
Acrylic-Amine, Tertiary Amine, GEL
STY-DVB, Polyamine, GEL
Epoxy-Amine , Polyamine , GEL
STY-DVB, Tertiary Amine, MR
Phenol-HCHO, Polyamine, MR
Epoxy-Amine Polyamine , GEL
STY-DVB, Tertiary Amine, MR
Phenol-HCHO Polyamine, MR
Acrylic Amine, Polyamine/ MR
Aliphatic Amine, Polyamine/ GEL
STY-DVB, Tertiary Amine/ MR
Epoxy-Amine/ Polyamine, GEL
STY-DVB, Quat. (II) Amine/ MR
STY-DVB, Quat. (I) Amine/ ISO
STY-DVB, Quat. (I) Amine/ MR
STY-DVB, Quat. (I) Amine/ GEL
STY-DVB/ Quat. (I) Amine, MR
ppm
TOC
14
5
35
13
16
26
25
46
30
20
90
33
19
8
6
4
3
4
RESIN CONCENTRATION * 0.9%
EXPOSURE TIME: 16 HOURS IN A 13 RPM TUMBLER
TEMPERATURE: 25°C
pH 5 2.5 (HC1)
INSTRUMENT: MODEL 915 BECKMAN TOC ANALYZER
162
-------
Many of the TOC values were alarmingly high in spite of the
fact that exposure time was 16 hours compared to the short 1 to
3 minutes typical detention time in an ion-exchange column.
Also, there was the possibility of organic particulates from
resin attrition due to tumbling — a non-representative condi-
tion with respect to column behavior. Nevertheless, one must
be concerned with these values as they are thousands of times
higher than the desirable levels even after the resins had been
"conditioned" by extensive backwashing and two service cycles
with 1.0 N NaOH and 1.5 N HCl including the appropriate, in-
termediate and final rinses. Resin manufacturers are aware of
this problem; Rohm and Haas [106] draws attention to it and
recommends a solution to be used in treating resins for use in
food and drug processing:
"Furthermore, Amberlite IR-45 contains trace quantities
of low molecular weight aromatic hydrocarbons which are
leached slowly from resin during service unless properly
pretreated...A most effective way in which the residual
aromatic material can be removed is to place the resin
in a column and pass steam at atmospheric pressure
down through the column allowing the condensate to
drain freely at the bottom. Ordinarily a matter of several
hours of such treatment after the entire bed has
reached steam temperature is sufficient to remove
virtually all the aromatics as well as any residual
traces of free amines and low molecular weight amino
compounds which may be left in the resin at the conclu-
sion of the manufacturing process."
Such a procedure would seem to be highly recommended for
resins prior to usage in water supply. Even so, the existence
of such a recommended solution doesn't eliminate what appears
to be a real need for research on the identification and quan-
tification of the organics leached from don-exchange resins in
water supply applications. One final note: the TOC results
163
-------
were highly variable and tended to change upon standing so care
must be exercised in the design of truly quantitative experiments.
The loss of volatile organics must be avoided as that seemed
to be one cause of the variability.
PHASE II RESULTS SUMMARY: MULTICOMPONENT CHROMATOGRAPHIC COLUMN
STUDIES
Nitrate/Chloride selectivity (a^) is the most important
selectivity in determining the relative amount of nitrate on
the resin at nitrate breakthrough, i.e., in determining the
maximum possible chemical efficiency (EM or y,T). This is both
M IN
good and bad: good because all the resins were nitrate selec-
tive with respect to chloride; bad because little variation
existed in the values of ou, among the thirty-two resins tested
N
(acl - 1.85 - 4.33) and no real significant effects on selectiv-
ity seem possible by further varying the important independent
variables—matrix and relative degree of crosslinking.
o
Sulfate/nitrate selectivity (a°) is nearly irrelevant in
determining the average equivalent fraction of nitrate on the
resin at the end of a run (yN). Surprisingly, slight increases
in yN are possible as a result of increasing rather than decreas-
ing the sulfate selectivity—ajj. The simple explanation offered
for this is that (1) ail the sulfate will be removed from the
feedwater regardless of its actual selectivity because it is the
most preferred species and (2) high sulfate selectivity promotes
a short sulfate-rich zone near the column entrance in which
almost no nitrate is removed thereby leaving essentially all of
that species to compete with the lesser preferred chloride in
the second equilibrium zone of the column which is where nearly
all of the nitrate is concentrated; see Fig. 5. Regardless of
the explanation, the effect of the selectivity of the most pre-
ferred species, sulfate, is predictably slight when the objec-
tive is to remove nitrate, invariably the lesser-preferred
164
-------
species,
E the maximum p_ossible chemical efficiency in nitrate re-
moval service has been defined as being equal to the average
equivalent fraction of nitrate on the exhausted resin (yN).
This y will be greater than x if the resin concentrates
nitrate by eluting the lesser preferred species (H9CO., and Cl")
^ J
in chromatographic fashion until nitrate breakthrough. As has
— S
just been discussed y-7 is not much influenced by OLT and only
N ,, N
moderately influenced by a , because of the narrow range of
values possible for a_,, among commercially available resins.
The most important influence on yM is, predictably, x...; when
it's low, process efficiency will be correspondingly low be-
cause the exhausted resin will comprise mostly sulfate and
chloride--species not intended to be removed; see Fig. 29.
In these studies the influence of x , at 2.5 gal/min ft , a , =
3.9 and x£ = 0.3 was as follows:
Liquid Phase Resin Phase
Equivalent Fraction Average Equivalent
of Nitrate Fraction of Nitrate Relative Efficiency
"V \T \7 /"V
XN yN YN/XN
.20 .32 1.70
.27 .40 1.48
Relative efficiency has been included to illustrate that yN is
not simply linearly related to XN- In addition to acl and XN/
the interrelated variables, exhaustion rate, bed depth and
superficial detention time (T), are quite significant. Short
detention times (T < 3.0 min), shallow beds (depth < 60 cm)
and high exhaustion rates (> 2.5 gal/min ft ) reduce yN by
causing relatively more chloride, apparently the kinetically
favored anion, to be in the resin at nitrate breakthrough.
That is summarized below for the condition where XN = .27, xg =
.27, xcl = .27 and a^ = 3.9:
165
-------
CD = COLUMN RUN 1
COLUMN RUN 5
= 0.20, 7 = 0.34
0.00
0.50 1.00 1.50 2.00 2.50 3.00
T « THROUGHPUT = EQUIVflLENTS SOLUTION / EQUIVflLENTS EXCHRNGER
FIGURE 29
COLUMN EFFLUENT PROFILES (NITRflTE)
EFFECT OF NITRflTE CONCENTRflTION ON COLUMN EFFICIENCY
3.50
-------
Resin
Exhaustion Rate T Depth
gal/min ft^ min cm ^N
2.44 3.1 61 .41
4.88 1.5 31 .39
Although xg was not a variable in the column experiments
it will greatly influence yN because all the sulfate fed to the
column will still be on it at nitrate breakthrough. When x_ is
high, the efficiency, yN, will be low.
Regeneration level influenced both the overall chemical
efficiency (EQ) and the maximum possible chemical efficiency
(EM). For the two-bed system, the regeneration level has been
defined based on the final anion column capacity. In practice,
the total equivalent capacity (TEC) of the cation bed must
equal or exceed the final anion bed capacity. It has been
determined here that a regeneration level of 300% of the theore-
tical HC1 required must be applied to the cation bed if calcium
and magnesium are the primary cations on the resin. Levels
much lower than that cause premature cation breakthrough, in-
creased pH and reduced anion bed capacity with smaller values
of yN at breakthrough. High regeneration levels on the other
hand maximize yN but cause unacceptably low effluent pH forcing
termination of the run. For each specific groundwater applica-
tion the sizes of the beds and the exact regeneration level
would have to be determined to insure maximum yN and a neutral
process effluent.
For the single-bed strong-base anion process regenerated
with NaCl it is expected that regeneration levels of 300% or
greater will be required for efficient regeneration. This is
based on published rather than experimentally determined infor-
mation.
167
-------
Net bicarbonate removal was zero, as expected for both the
two-bed and single-bed processes. Thus, high values of bicar-
bonate in the raw water don't measurably influence yV.. Bi-
carbonate and carbonic acid apprently have a catalytic effect
in columnar ion-exchange processes so it is tentatively recommen-
ded that the system degasifier be placed downstream from the
anion bed rather than preceding it.
The effluent concentration profiles at 2.5 gal/min ft3 were
very sharp but not quite vertical indicating that, at the end
of the run, nitrate breakthrough can be readily anticipated and
used to control the process. Unusually shaped profiles with
early and inefficient nitrate breakthrough resulted from pre-
mature pH increases in the weak-base anion column influent due
to insufficient cation bed capacity.
The final column capacitaejs and bed volumes of effluent to
nitrate breakthrough were of course, very much a function of the
advertised and measured capacities of the resins and the sulfate
concentrations of the feed waters. High column capacities can
improve the overall economic efficiency of an ion-exchange
process if they lead to lower rinse volume requirements but,
since high capacity resins also tend to be highly sulfate
selective and require progressively longer rinse volumes with
service time, that possible improvement in operating efficiency
is not guaranteed.
To compute the expected multicomponent column capacity of
a highly sulfate selective resin at nitrate breakthrough one
should assume that all the sulfate will be removed, and use the
advertised or measured H2S04 capacity (see Titration Curves,
Figs. B1-B13) for that fraction of the capactiy represented by
sulfate ions, and do likewise for HC1 and HNO., taken together
assuming in all cases that insignificant HpCCU will be on the
resin at the end of the run. The equation is as follows; It
168
-------
assumes that HC1 capacity = HNO_ capacity:
Expected Multicom-
ponent Column Capacity = (x') (H-SO, Cap.) + (1-x') (HC1 Cap.)
where x' = Equivalent fraction of SOT in raw water not including
HCO~. For Test Waters 2 and 3, x's = 0.33.
The overall chemical efficiency (E_) can be expected to be
about 13.3% for both the single-bed and two-bed processes. This
is based on the observed average equivalent fraction of nitrate
on the resin at the end of the runs (y"N) with a feedwater
containing the same equivalent concentration of nitrate/ chloride
and sulfate and an irrelevent amount of bicarbonate which under-
goes no net removal in either process. This overall chemical
efficiency has been defined as the equivalents of nitrate re-
moved per equivalent of regenerant supplied and is the product
of yN (or EM) and E , the regeneration efficiency, which has
been determined to be 0.33 based on a regeneration level of
300%.
A comparative process economic evaluation reveals that the
two-bed process with NH3 and HC1 as regenerants has chemical
plus disposal costs which are approximately 50% higher than the
single-bed process assuming an overall chemical efficiency of
13.3%, 25% bypass water, a feedwater with the composition of
Test Water 3 (Table 21) ', NaCl-NaNCU brine disposal by trucking
8 miles before discharging into a stream, and no disposal cost
for the high-nitrogen content wastewaters from the two-bed
process which are given away for their fertilizer value. See
Tables 24 and 25 for complete details of the comparative econo-
mic evaluation.
The advantages (+) and disadvantages (-) of the single-bed
and two-bed processes are as follows:
169
-------
Single-bed, strong-base anion with NaCl regeneration
(+) Simple, no balancing of beds and regenerants
{+) Low cost regeneration
(-) Very difficult and costly to dispose of regenerants in
non-coastal locations where natural evaporation is im-
possible
(-) Iron must be removed to prevent resin fouling
(-) Continuous nitrate analysis required for process
control
Two-bed/ strong-acidy weak-base NH^ S HC1 regenerants
(+) Partial softening in addition to nitrate removal
(+) No problem with iron fouling. Precipitated iron is
removed from the cation bed during each regeneration
{+) Regenerants wastewaters expected to be easy to dispose
of by land application as fertilizer
(-) Complex system: bed sizes and regenerants must be
balanced
(-) Degasifier for C02 removal required
(-) Continuous pH and nitrate analysis required for pro-
cess control
(-) High regenerant costs
Continuous ion-exchange processes of the pulsed-resin flow
type will be more difficult to control in chromatographic elution
to nitrate breakthrough because there will be a nitrate break-
through prior to every resin pulsing operation. That will
require a control decision based on nitrate analysis once every
few minutes compared to the once or twice-per-day decision for
a large fixed-bed operation. An efficient, continuous, two-bed
system of the type recommended here would seem to be unduly
complicated because of the requirements for balanced capacities
and chromatographic elution of both beds.
170
-------
Organic extractables present in the anion resins even after
"conditioning" gave rase to total organic carbon (TOG) concen-
trations in the 3-100 ppm range in acidic, aqueous solutions
containing about 0.9% resin agitated for 16-20 hours. It is
anticipated that the extractable organics in both cation and
anion resins represent a potential problem in water supply.
Research on the ppb level of organics associated with the ex-
isting and potential uses of ion exchangers in water supplies
definitely seems warranted in view of this TOC data and the
recent concern over organics in public water supplies.
Nitric acid is definitely not recommended as a regenerant
in the two-bed process even though it would greatly enhance the
fertilizer value of the regenerant wastewaters. It is too
costly, 46.5C/1000 gal treated water (12.3<:/m ), requires excess
cation bed rinsing to reduce nitrate and allows the possibility
of disastrous nitrate and acid pollution of the water supply in
the even of an operating error. Even though HC1 is more costly
than H-SO. it may be more economical where large excesses of
H2SO4 are required due to CaSO. fouling of the cation bed.
A ranking of anion resins for nitrate removal service is
given in Table 27, considering that high nitrate/chloride selec-
tivity high capacity and moderate sulfate/nitrate selectivity
are the desirable characteristics. Organic extractables as
evidenced by the TOC of resin equilibrates were not considered
in making the rankings because of the very preliminary nature
of those measurements. However, an asterisk (*) has been used
to indicate a resin producing markedly colored water in addition
to high TOC.
Although the resins are ranked in preference order, the
differences among the recommended resins are not large; they are
all expected to give nearly the same maximum possible chemical
efficiency E... Some overall process efficiency is gained by
171
-------
using high capacity resins while some might be lost with the
highly sulfate selective resins should they require rinse
volumes.
TABLE 27
RANKING OF RESINS FOR USE IN NITRATE REMOVAL SERVICES
Recommended
STY-DVB, Polyamine Resins
Amberlite IR-45
STY-DVB, Tertiary-amine, MR Resins
Amberlite IRA-93
Dowex MWA-1
lonac AFP-329
Duolite ES-368
STY-DVB, Quat. (I & II) Amines, Gel & MR Resins
lonac ASB-100, AFP-100, A-641, ASB-1P, ASB-2
Duolite, A-101-D, A-102-D
Dowex 11, SAR, SBR-P, SBR
Amberlite IRA-400, IRA-900, IRA-402, IRA-910, IRA-410
Acrylic-Amine, Polyamine, MR Resins
Duolite ES-374
Phenol-HCHO, Polyamine, MR Resins
Duolite A-7
Duolite ES-561
Not Recommended
Epoxy-amine, Polyamine, Gel Resins
Dowex WGR
Duolite A-340
lonac A-305
Acrylic-Amine, Tertiary Amine, Gel Resins
Amberlite IRA-68
Aliphatic-Amine Polyamine, Gel Resins
*Ionac A-260
172
-------
REFERENCES
1. Adams, G., P.M. Jones, and J.R. Millar, "Kinetics of Acid
Uptake by Weak-Base Anion Exchangers" J. The Chemical
Society, Inorg, Phys. Theor., p. 2543, (1969).
2. American Public Health Association, American Water Works
Pollution Control Federation, Standard Methods for the
Examination of Water and Wastewater, 13th Ed., American
Public Healt!T~Assoc., Washington, D.C. 20036, 1971.
3. American Water Works Association, Water Quality and Treat-
ment; - A Handbook of Public Water Supplies, 3rd Ed., McGraw-
Hill Book Co., Inc., New York, N.Y., 1971, Chapter 10.
4. Ames, L.L., "Some Zeolite Equilibria with Alkaline Earth
Metal Cations" American Mineralogist, V. 49, p. 1099, (1964).
5. Ames, L.L., "Zeolite Cations Selectivity," Canadian Minera-
logist, V. 8, p. 325, (1965).
6. Anderson, R.E. "On the Pickup of Hydrochloric and Sulfuric
Acids on a Polyamine Resin" in Polyelectrolytes and Their
Applications p. 263, D. Reidel Publishing Co., Dordrecht-
Holland, 1975.
7. Anderson, R.E., "A Countour Map of Anion Exchange Resin Pro-
perties" I&EC Product Res. & Dev., V. 3(2), p. 85, June 1964.
8. Aveni, A., G. Boari, L. Liberti, M. Santori and B. Monopoli,
"Sulphate Removal and Dealkalization on Weak Resins of the
Feed Water for Evaporation Desalting Plants." Desalination,
V. 16, p. 135 (1975).
9. Baumann, W.C., in Ion Exchange: Theory and Application, Edi-
tor F.C. Nachod, Academic Press Inc., New York, 1949.
W-l r
10. Beulow, R.W., K.L. Kropp, J. Withered, and J.M. Symons,
"Nitrate Removal by Anion Exchange Resins," Water Supply
Research Laboratory, National Environmental Research Center,
U.S. EPA, Cincinnati, Ohio, May, 1974.
11. Bingham, E.G., "Fertilizer Maker Stops Nitrogen," Water and
Wastes Engineering, P.E-4, November, 1972.
173
-------
12. Blalock, H.M., Social Statistics/ 2nd. Ed., McGraw-Hill,
New York, (1972) .
13. Boari, G., L. Liberti, C. Merli and R. Passino,
"Exchange Equilibria on Anion Resins," Desalination,
V. 15, p. 145-166, 1974.
14. Bolto, B.A. and R.E. Warner, "An Ion-Exchange Process
with Thermal Regeneration VII. The Rates of Neutraliza-
tion of Weak Electrolyte Resins." Desalination, V. 8,
p. 21-34, 1970.
15. Bolto, B.A./ "Sirotherm Desalination: Ion Exchange
with a Twist" Chemtech, p. 303, May 1975.
16. Bouwer, Herman; "Renovating Municipal Wastewater by
High-Rate Infiltration for Ground-Water Recharge"
Journal AWWA/ p. 159/ March/ 1974.
17. Bouwer/ Herman; R.C. Rice, and E.D. Escarcenga;
"Renovating Secondary Sewage by Ground Water Recharge
with Infiltration Basins", U.S. Environmental Protection
Agency, Water Pollution Control Res. Series, 16060 DRV,
U.S. GPO, 1972.
18. Boyd, G.E., J. Schubert, and A.W. Adamson, The Exchange
Adsorption of Ions From Aqueous Solutions by Organic
Zeolites. I. Ion-Exchange Equilibria." J. ACS/ V. 69,
p. 2818, Nov. 1947.
19. Boyd/ G.E., A.W. Adamson, and L.S. Myers, Jr., "Ion-
Charge and Exchange Rates with Ion-Exchange Polymers."
J. ACS, V. 72, Oct. 1950.
20. Boyd/ G.E. and B..A. Soldano, Self-Diffusion of Cations
in and Through Sulfonated Polystyrene Cation-Exchange
Polymers, J. ACS, V. 75, No. 24, p. 6091, Jan./ 1954.
21. Boyd, G.E., A.W. Adamson/ and L.S. Myers/ Jr., "The
Exchange Adsorption of Ions from Aqueous Solutions
by Organic Zeolites. II. Kinetics" J. ACS/ V. 69/
p. 2836, Nov. 1947.
22. Bregman, J.I., and J.M. Schackelford, "Ion-Exchange
is Feasible for Desalination'4 En vi ronmenta1 Sci. ji Tech.,
Vol. 3, No. 4, p. 336, 1969.
23. Case, A.A., "The Health Effects of Nitrates in Water"
Proc. 12th Sanitary Engineering Conf., Univ. of 111.,
Urbana, 111., February, 1970.
174
-------
24. Chemerys, J.C., "Effect of Urban Development on the
Quality of Ground Wate*r in Raleigh North Carolina,
USGS Professional Paper 575-B, pp. 212-216 (1967) .
25. Chemical Marketing Reporter, Schnell Publishing Co.,
August 8, 1976.
26. Chu, B., D.C. Whitney and R.M. Diamond, "On Ion-Exchange
Resin Selectivities," J. Inorg. Nucl. Chem., V. 24,
p. 1405, 1962.
27. Clazie, R., "Multicomponent Diffusion: Generalized
Theory, With Ion-Exchange Applications" Ph.D. Thesis,
Univ. of California, Berkeley, 1967.
28. Clifford, D.A., "Feasibility Study for Total Water
Management at Arcosanti Arizona" Report to the Cosanti
Foundation, University of Michigan, April 29, 1974.
29. Cole, J.A., Editor, Groundwater Pollution in Europe,
Proceedings of a conference organized by the Water
Research Association, in Reading, England, September 1972,
Water Information Center, Port Washington, N.Y., (1974).
30. Conway, D.E., J.H.S. Green and D. .Reichenberg, "The
Kinetics of Sodium-Hydrogen Exchange on a Monofunctional
Cation Exchange Resin Containing Carboxyl Groups",
Trans. Faraday Soc;., V. 50, p. 511, 1954.
31. Copeland, J.P., C.L. Henderson, J.M. Marchello,
"Influence of Resin Selectivity on Film Diffusion
Controlled Ion-Exchange" A.I.Ch.E.J., Vol. 13, No. 3,
May 1967, p. 449-52.
32. Copeland, J.P. and J.M. Marchello, "Film Diffusion
Contolled Ion-Exchange with a Selective Resin,"
Chem. Eng. Sci., Vol. 24, No. 9, Sept. 1969, p. 1471-4.
33. Crabtree, K.T., "Nitrate Variation in Ground Water,"
Univ. of Wisconsin (Madison), Water Resources Center,
Supplemental Report, 1970.
34. Dean, J.A. (Editor), Lange's Handbook-.of Chemistry,
Sections 5 & 6 - Analytical"Chemistry & Electrochemistry,
llth Ed., McGraw-Hill, New York, (1973).
35. deLaguna, W., Chemical Quality of Water - Brookhaven
National Laboratory and Vicinity, Suffolk County,
New York, USGS Bull., 1156-D (1964).
36. Diamond Shamrock Chemical Co., Duolite Ion-Exchange
Manual, Redwood City, CA, 1969.
175
-------
37. Diamond Shamrock Chemical Co., "Duolite Ion Exchange
Resins - List of Products Properties and Uses"
Redwood City, CA., 1970.
38. Diamond, R.M., "The Aqueous Solution Behavior of Large
Univalent Ions. A New Type of Ion-Pairing", J.
Phys. Chem., V. 67, p. 2513, Dec. 1963,
39. Diamond, R.M., and D.C. Whitney, "Resin Selectivity
in Dilute To Concentrated Aqueous Solutions" in Ion
Exchange; A Series o_f Advances, J.A. Marinsky Ed.,
Marcel Dekker Inc., New York, 1966.
40. Dorfner, K., Ion Exchangers: Properties and Applications,
3rd Ed., Ann Arbor Science, Ann Arbor, Mich., (1972) .
41. Dow Chemical Co., "Laboratory Manual on Ion-Exchange"
The Dow Chemical Co., Midland, Michigan, 1971.
42, Draper, N, and H. Smith, Applied Regression Analysis,
Wiley, New York, (1968) .
43. Eiseman, G., Cation Selective Glass Electrodes and
Their Mode of Operation", Biophysical J., V. 2, p. 259,
(1962) .
44. Eliassen, R., B.M. Wyckoff and C.D. Tonkin; "Ion-
Exchange for Reclamation of Reusable Supplies" J. AWWA,
pp. 1113-22, Sept., 1965.
45. Eliassen, R. and G.E, Bennett; "Anion Exchange and
Filtration Techniques for Wastewater Renovation",
J. WPCF, Vol. 39, No. 10, Part 2, pp. R 82-91, 1967.
46. Eliassen, R. and Tchobanogeous, "Chemical Processing
of Wastewater for .Nutrient Removal" J. WPCF, Vol. 40,
No. 5, part 2, p. R 172, 1968.
47. Evans, S., "Nitrate Removal by Ion-Exchange" J. WPCF,
V. 45, No. 4, pp, 632-36, April, 1973.
48. Fox, D.J. and K.E. Guire, "Documen ta t i o n for MIDAS:
Michigan Interactive Data. An^lysjLs System^ 2nd. Ed.,
Univ. of Michigan Statistical Research Laboratory,
Ann Arbor, Mich., (1973).
49. Gauntlett, R.B., "Nitrate Removal from Water by Ion-
Exchange" Water Treatment and Examination, V. 24,
p. 172, (1975).
I
50. George, W.O., and Hastings, W.W., "Nitrate in the
Ground Water of Texas", Trans. of Am. Geophys. Union,
V. 32, No. 3, pp. 450-456, (1951a).
176
-------
51. Gold, H. and A. Todisco, "The Avco Continuous
Moving Bed Ion-Exchange Process and Large Scale
Desalting", Desalination, 317 (1975), p. 97.
52. Graham, E.E. and Dranoff, J.S., "Kinetics of Anion
Exchange Accompanied by Fast Irreversible Reaction"
A.I,. Ch.E. Journal Vol. 18, No. 3, May (1972).
53. Gregg, J.C., "Nitrate Removed at Water Treatment
Plant", Civil Engineering - ASCE, p. 45, April (1973) .
54. Gregory, J. and R.V. Dhond, "Anion Exchange Equilibria
Involving Phosphate, Sulfate and Chloride", Water
Research, V. 6, 1972, pp. 695-702.
55. Grinstead, R.R. and K.C. Jones; "Nitrate Removal
from Waste Waters by Ion-Exchange"', Environmental
Protection Agency Water Pollution Control Research
Series No. 17010 FSJ, U.S. GPO, Washington, D.C.,
Jan., 1971.
56. Grinstead, R.R., R.M. Wheaton, T.E. Drisko and
K.C. Jones, "Feasibility of Removal of Nitrates from
San Luis Drain Waters by Ion Exchange" Report by the
Dow Chemical Co. to U.S. Dept. of Interior: Bureau
of Reclamation, STS No. PB180342, Oct. 9, 1968.
57. Guttman, I., S.S. Wilks, J.S. Hunter, Introductory
Engineering Statistics, 2nd Ed., John Wiley, New York,
(1971).
58. Hall, E.S., "Some Chemical Principles of Groundwater
Pollution" in Groundwater Pollution in Europe,
J.A. Cole, Editor, Water Information Center, Port
Washington, N.Y. (1974).
59. Harned, H.S. and B.B. Owen, The Physical Chemistry
of Electrolytic Solutions, ACS Monograph Series No. 137,
3rd Ed. Reinhold, New York (1958) .
60. Helfferich, Friedrich, Ion Exchange, McGraw-Hill Book
Co. Inc., New York, N.Y., 1962.
61. Helfferich, F., "Ion Exchange Kinetics," in Ion Exchange
£ Series of Advances, Vol. 1, J.A. Marinsky Ed., Marcel
Dekker In., New York, 1966.
62. Helfferich, F., "Ion-Exchange Kinetics, v. Ion Exchange
Accompanied by Reactions," J. Phys. Chem., V. 69,
63. Helfferich?6F.'G., "Multicomponent Ion Exchange in Fixed
Beds" I&EC Fund., V. 6, No. 3, p. 362, (1967).
177
-------
64. Helfferich/ F. and G. Klein, Multicomponent
Chromatography; Theory of Interference/ Marcel
Dekker, New York, 1970.
65. Bering, B., and B. Harding, "Diffusion in Ion-Exchange
Resins" A.I_.Ch.E. J., Vol. 9, No. 4, p. 495, 1963.
66. Holzmacher, R.G., "Nitrate Removal from a Ground Water
Supply" Water and Sewage Works, p. 210, July, 1971.
67. Huckel, W., Structural Chemistry o_f Inorganic Compounds,
Elsevier Publ. Co., New York, 1951.
68. International Atomic Energy Agency, "Operation and
Control of Ion-Exchange Processes for Treatment of
Radioactive Wastes", IAEA, Tech. Report Series No. 78,
Vienna, Austria, 1967.
69. Jensen, C.H., and R.M. Diamond, "Anion Exchange in
Concentrated Solutions", J. Phys. Chen., V. 69,
No. 10, p. 3440, 1965.
70. Klein, G., M. Villena-Blanco, T. Vermeulen, Ind. Eng.
Chem., Process Design and Development, V. 3, p. 280,
(1964) .
71. Klein, G., D. Tondeur, T. Vermeulen, "Multicomponent
Ion Exchange in Fixed Beds", ISEC Fund., Vol. 6, No. 3
p. 339 (1967) .
72. Korngold, E., "Removal of Nitrates from Potable Waters
by Ion-Exchange" Water, air, Soil Pollution, Vol.2,
No. 1, pp. 15-22, 1973, CA 79:45602.
73. Kreusch, E., and K. Schmidt, "Wastewater Demineraliza-
tion by Ion-Exchange", Environmental Protection Agency,
Water Pollution Control Research Series No. 17040 EEE,
U.S. GPO. Washington, D.C., Dec. 1971.
74. Kunin, R., and R.J. Myers, "Rates of Ion Exchange in
Ion Exchange Resins", J. Phys. and Colloid Chem.
Vol. 51, No. 5, pp. llll,~Tl947rr~
75. Kunin, R., and R.J. Myers, "The Anion Exchange Equilibria
in an Anion Exchange Resin", J. ACS, V. 69, p. 2874
(1947). ~~
76. Kunin, R and R.E. Barry, "Carboxylic, Weak-Acid Type
Cation Exchange Resin" Ind. Eng. Chem., V. 41, p. 1269
(1949).
i
77. Kunin/ Robert and R.J. Myers, Ion-Exchange Resins,
John Wiley & Sons, Inc., New York, pp. 57-58, (1950).
178
-------
78. Kunin, Robert; Ion Exchange Resins, 2nd Ed., John Wiley
& Sons, Inc., New York, 1958.
79. Lehmann, E.J., Ground Water Pollution, Part 2 Pollution
from Irrigation and Fertilization, NTIS No. PS-75/740,
U.S. Dept. of Commerce, Springfield, Va., Sept. (1975).
80. Liberti, L. and R. Passino, "Chloride-Sulphate Exchange
on Anion-Exchange Resins Kinetic Investigations. I."
J. Chromatography, V. 102, p. 155, 1975.
81. Lupa, Alan J., "The Kinetics of Ion-Exchange With
Simultaneous Film and Particle Diffusion," Ph.D.
Thesis, Chem. Eng., Northwestern Univ., 1967.
82. Marinsky, J.A., "Prediction of Ion-Exchange Selectivity"
J. Phys. Chem., Vol. 71, NO. 6, p. 1572 (1967).
83. Mathews, A.P., "Mathematical Modelling of Multi-
component Adsorption in Batch Reactors", Ph.D. Thesis,
Univ. of Michigan, Ann Arbor, 1975.
84. McKee, J.E. and H.W. Wolf; Water Quality Criteria,
2nd Ed., 1963, State Water Resources Control Board,
Sacramento, CA., Publication 3-A, April, 1971.
85. Meloan, C.E., "The Selective Removal of Nitrate and
Nitrite from Polluted Water," Prepared for Office
of Water Resources Research, NTIS No. PB-231471,
U.S. Dept. of Commerce, Springfield, Va., March 1974 .
86. Midkiff, W.S., "Reactions of a Strongly Basic Ion-
Exchange Resin in Dilute Aqueous Solution in a
Columnar System", Ph.D. Thesis, University of Mich.,
Ann Arbor, 1969.
87. Midkiff, W.S., and Weber, W.J., Jr., "Operating
Characteristics of Strong-Base Anion Exchange Reactors
Proceedings of the 25th Purdue Industrial Waste
Conference, May, 1970.
88. Moison, R.L., and H.A. O'Hern, Jr., "Ion-Exchange
Kinetics" Chem. Eng. Prog. Symposium Series, V. 55,
No. 24, p. 71, 19T9.
89. Moore, W.J., Physical Chemistry/ Third Ed., Prentice-
Hall, Inc., Englewood Cliffs, N.J., (1962).
90. Naber, J.E., and Wesseligh, J.A., "New Shell Process
Treats Claus Off-Gas", CEP, V. 69, No. 12, p. 29,
Dec., 1973.
179
-------
91. National Academy of Sciences Committee on Nitrate
Accumulation; Accumulation of Nitrate/ Washington,
D.C., 1972.
92. National Academy of Sciences and National Academy of
Engineering, "Water Quality Criteria 1972" Report of
Committee on Water Quality Criteria, Environmental
Studies Board NAS-NAE, EPA Ecological Research
Series No. EPA-R3-73-033, U.S. GPO, Washington, B.C.,
March, 1973.
93. Nightingale, H.I., "Statistical Evaluation of Salinity
and Nitrate Content and Trends Beneath Urban and Agri-
cultural Area—Fresno, Calif.," Ground Water, V. 8, No. 1,
pp. 22-28 (1970) .
94. O'Brien, W.J., "Chemical Removal of Nitrate from
Potable Water Supplies" Kansas Water Resources Research
Institute, Manhattan, Kansas, June (1968).
95. Osborn, G.H., Synthetic Ion Exchangers, Chapman &
Hall Ltd., London, p. 16 (1961).
96. Palmquist, W.N., and F.R. Hall, "Reconnassance of
Ground Water Resources in the Blue Grass Region,
Kentucky", USGS, WSP 1333, (1961).
97. Paterson, R., An Introduction to Ion Exchange, Heyden
and Son Ltd., London, 1970.
98. Pauling, L., The Nature of_ the Chemical Bond, Cornell
University Press, 1960.
99. Peroni, L.J., and J.S. Dranoff, "Ion-Exchange Equilibria
in a Ternary System" A.I^.CJi.E.J., Vol. 9, No. 1,
p. 42, 1963.
100. Perry, J.H., et. al., Chemical Engineers Handbook,
5th Ed., Chapter 16, "Absorption and Ion Exchange"
McGraw Hill, New York (1974).
101. Rao, M.G. and M.M. David, "Single-Particle Studies
of Ion-Exchange in Packed Beds: Cupric Ion-Sodium
System," A.I.Ch.E.J., V. 10, No. 2, p. 213, 1964.
102. Reddy, M.M. and J.A. Marinsky, "Ion-Exchange Selectivity
Coefficients in the Exchange of Calcium, Strontium
Cobalt, Nickel Zinc and Cadmium with Hydrogen Ion
in Variously Crosslinked Polystyrene Sulfonate Cation
Exchangers at 25° C", J. Macromol. Sci-Phys., V. B5(l),
p. 135, March 1975.
180
-------
103. Reichenberg, D., "Ion-Exchange Selectivity" in Ion
Exchange; A Series of_ Advances, J.A. Marinsky Ed.,
Marcel Dekker, Inc., New York, 1966.
104. Rohm and Haas Co. Inc., Engineering Manual for the
Amberlite Ion-Exchange Resins, Philadelphia, PA.r 1970.
105. Rohm & Haas, Technical Bulletin: Amberlite IR-120
Plus, The Rohm & Haas Co. Philadelphia, PA., Sept. 1972.
106. Rohm & Haas, Technical Bulletin: Amberlite IR-45
Technical Notes, The Rohm and Haas Co., Philadelphia,
PA., Feb. 1971. . ,,. „ i
107. Sabadell, J.E., "Nitrate and Nitrite Removal
from Raw and Ground Waters by Anion Exchange - A Review"
Presented at the 67th Annual AIChE Meeting in
Washington, D.C., August, 1974.
108. Sanks, R.L., and W.J. Kaufman, "Partial Demineralization
of Brackish Waters," National Symposium on Sanitary
Engineering Research, Design and Development,
Penn. State Univ., University Park, PA., 1965.
109. Schmidt, K.D., "Nitrates and Groundwater Management
in the Fresno Urban Area," J.AWWA, p. 146, March (1974).
110. Smith, G.E., "Nitrate Pollution of Water Supplies,"
Univ. of Missouri, Proceedings of 3rd. Annual Conf.on
Trace Substances in Environmental Health 1969, pp. 2/J-
287, (1970) .
111. Smith, T.G. and J.S. Dranoff, "Film Diffusion-Controlled
Kinetics in Binary Ion-Exchange", Ind. J En1._Chem.
Fundamentals, Vol. 3, No. 3., p. 1957 1964.
112. Snoeyink, V.L., and W.J. Weber, Jr., "Adsorption of
Strong Acids, Phenol and 4-Nitrophenol by Active Carbon
in Agitated Non-Flow Systems," Dept. of Civil Eng.
Report T-68-3, Univ. of Michigan, Ann Arbor, 1968.
113 Soldano a A and G.E. Boyd, "Self-Diffusion of
j-j-j. DOxaano, B.A. euiu va.^. *•"•"* • _«.^_0 » t afQ \7 7<^
Anions in Strong Base Anion Exchangers, J. ACS, .V. 75,
No. 24, p. 6099, Jan., 1954.
U4. st. Amant, P.P. and P.L. McCarty, "Treatment of High
Nitrate Waters," J. AWWA, Vol. 61, No. 12, p. 659,
"Nitrate Removal from
Water Quality Management
Va., Oct., 1970.
181
-------
116, Summers, W.K., and Z., Spiegel, Ground Water Pollution:
a Bibliography, Ann Arbor Science, Ann Arbor, Mich.,
1974.
117. Thompson, G.L., "Ground Water Resources of Nararro
County, Texas," Texas Water Development Board,
Kept, 160, 63 pp. (1972) .
118. Todd, O.K., The Water Encyclopedia, Water Information
Center, Port Washington, N.Y., 1970,
119. Tondeur D., and G. Klein, "Multicomponent Ion-
Exchange in Fixed Beds" I&EC Fund., V. 6, No. 3,
p. 351, August (1967).
120. Turner, J.C.R. and C.B. Snowdon, "Liquid-Side Mass-
Transfer Coefficients in Ion Exchange: an Examination
of the Nernst-Planck Model", Chem. Eng. Sci., V. 23,
p. 221, 1968.
121. Turner, J.C.R. and C.B. Snowdon, "Liquid-Side Mass
Transfer Coefficients in Ion-Exchange - the H+/Cu++-
Cl~ System" Chem. Eng. Sci., V. 23, p. 1099, 1968.
122. Univ. of California, Sanitary Engineering, Research
Laboratory Staff, "Optimization of Ammonia Removal
by Ion Exchange Using Clinoptilolite," Water Pollution
Control Research Series Report No. 17080 DAR 09/71/
U.S. GPO 1971.
123. University of Illinois, College of Engineering,
"Proceedings of the 12th Sanitary Engineering Con- ,
ference: Nitrate and Water Supply: Source and Control,
Dept. of Civil Eng., Univ. of 111., Urbana, 111..
Feb. (1970).
124. Univ. of Michigan Statistical Research Lab Staff, A
Manual of Elementary Statistics Using MIDAS, First~Ed.,
Statistical Research Laboratory, The University of
Michigan, Ann Arbor, 1975,
125. U.S. Dept. of Health, Education and Welfare, "Public
Health Service Drinking Water Standards," Public
Health Service Publication No. 956, U.S. GPO,
Washington, D.C., 1962.
126. U.S. Environmental Protection Agency, Manual on Methods
of Chemical Analysis of Water and Wastes, U.sTEPA
Office of Technology Transfer, Washington, D.C. 20460,
1974.
182
-------
127. U.S. EPA, "Interim Primary Drinking Water Standards,"
Federal Register, December 24, 1975.
128. Van Brocklin, L.P., "Ionic Migration Effects on
Liquid-Resistance - Controlled Ion Exchange" Ph.D.
Thesis, University of Washington, 1968.
129. Walitt, A.L. and H.L. Jones, "Basic Salinogen Ion-
Exchange Resins for Selective Nitrate Removal from
Potable and Effluent Waters," U.S. EPA, Cincinnati,
Advanced Waste Treatment Laboratory, 1970, U.S. GPO,
Washington, D.C.
130. Walker, E.H., "Ground Water Resources of the
Hopkinville Quandrangle, Kentucky," USGS WSP 1328,
(1956).
131. Ward, Paul C., "Existing Levels of Nitrates in Waters-
the California Situations," 12th Annual Sanitary Eng.
Conf., Univ. of 111., Urbana, 111., Feb., 1970.
132. Weber, O.W., I.F. Miller and H.P. Gregor, The
Adsorption of Carbon Dioxide by Weak Base Ion-Exchange
Resins," A.I^.Ch.E .J., V. 16, No. 4, p. 609, 1970.
133. Weber, W.J., Jr., Physiochemical Processes for Water-
Quality Control, Wiley Interscience, New York, 1972.
134. Weiss, D.E., et.al., "Thermally Regenerated Ion-
Exchange Process: an Aid to Water Management"
J. WPCF, V. 38, p. 1782, 1966.
135. Weiss, D.E., B.A. Bolto, R. McNeill, A.S. MacPherson,
R. Siudak, E.A. Swinton, and D. Willis, "An Ion-
Exchange Process With Thermal Regeneration II. Pro-
perties of Weakly Basic Ion-Exchange Resins."
Australian J. Chem., V. 19, p. 561, 1966.
136. Wilkander, L., Ann. Royal Agr^ Coll. Sweden, V. 14,
p. 1-171 (1956)1
183
-------
APPENDIX A
EQUILIBRIUM
ISOTHERMS
SULFflTE-NITRflTE
CHLORIDE-NITRflTE
0-20 0.40 0.60 0.80
EQUIVflLENT FRflCTION 50^ IN LIQUID PHflSE
EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
RESIN NUMBER 1
flMBERLITE IRfl 93, MflCROPOROUS RESIN
STYRENE-DVB MflTRIX
TERTIflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY=1.25 MEQ/ML )
1.00
FIGURE Al
25° C, BINRRY ION-EXCHRNGE ISOTHERM
184
-------
SULFflTE-NITRflTE
CHLORIDE-NITRflTE
^.00
V
•SSOU'
0.20 O.UO 0.60 0.80
EQUIVflLENT FRflCTION SO,, IN LIQUID PHflSE
EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
RESIN NUMBER 2
HMBERLITE IRfl 68, MICROPOROUS GEL
POLYflCRYLIC MflTRIX
TERTIflRY-flMlNE FUNCTIONflLITY
TOTflL CflPflCITY^1.6 MEQ/ML
1.00
FIGURE A2
25° C, BINflRT IDN-EXCHflNGE ISOTHERM
185
-------
e SULFRTE-NITRRTF
CHLORIDE-NITRRTE
,00
0.20 0.40 0.60 0.80 I. 00
S(M, EQUIVRLENT FRflCTION SO^ IN LIQUID PHR3E
\L, EQUIVRLENT FRflCTION CL IN LIQUID PHflSE
RESIN NUMBER 3
RMBERLITE IR U5, MICROPQROUS GEL
STYRENE-OVB MflTRIX
POLYRMINE FUNCTIONRLITY
TOTRL CflPflCITY=1.3 MEG/ML
FIGURE A3
25° C, BINflRY ION-EXCHRNGE ISOTHERM
186
-------
e SULFflTE-NITRflTE
CHLORIDE-NITRflTE
^.00 .10
x
0.20 O.UO 0.60 C.80
EQUIVfllENT FRflCTION SO^ IN LIQUID PHflSE
*a, EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
RESIN NUMBER U
DOWEX WGR, MICROPOROUS GEL
EPOXYcflMINE MflTRIX
POLYflMINE FUNCTIONflLITY
TOTflL CflPflCITY-l.O MEQ/ML
FIGURE A4-
25° C, BINflRT ION-EXCHRNGE"ISOTHERM
187
-------
SULFRTE-NITRftTE
CHLORIDE-NITRRTE
0.20 0.40 0.60 0.80
EQUIVflLENT FRflCTION SC^ IN LIQUID PHflSE
XCL, EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
RESIN NUMBER 5
DOWEX MWfl-1, MflCROPOROUS RESIN
STYRENE-DVB MflTRIX
TERTIflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY=l.l MEQ/ML
1.00
FIGURE A5
25° C, BINRRY ION-EXCHRNGE ISOTHERM
188
-------
e SULFflTE-NITRflTE
CHLORIDE-NITRflTE
0.20 O.UO 0.60 0.80
EQUIVflLENT FRflCTION SO^ IN LIQUID PHflSE
EQUIVRLENT FRflCTION CL IN LIQUID PHflSE
RESIN NUMBER 6
DUOLITE fl-7, MflCROPOROUS GRflNULflR RESIN
PHENOL-FORMflLDEHYDE MRTRIX
SECONDflRY-flMINE FUNCTIONflLITY
(POLYflMINE TITRflTION CURVE)
TOTflL CflPflCITY=2.U MEQ/ML
FIGURE AG
25° C, BINflRT ION-EXCHRNGE ISOTHERM
1.00
189
-------
SULPHflTE-NITRflTE
CHLORIDE-NITRflTE
0.20 O.UO 0.60 0.80 1.00
EQUIVflLENT FRflCTION SO^ IN LIQUID PHflSE
EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
RESIN NUMBER 7
DUOLITE fl 340, MICROPOROUS GEL
EPOXY-flMINE MflTRIX
POLYflMINE FUNCTIONflLITY
TOTflL CflPflCITY=2.6 MEQ/ML
FIGURE A7
25° C, BINRRT IQN-EXCHflNGE ISOTHERM
190
-------
SULFflTE-NITRflTE "
* CHLORIDE-NITRflTE ••
^).00
0.20 O.UO 0.60 0.80 1.00
EQUIVflLENT FRflCTION SO^ IN LIQUID PHflSE
EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
RESIN NUMBER 8
DUOLITE ES 368, MflCROPOROUS RESIN
STYRENE-DVB MflTRIX
TERTIflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY=1.3 MEQ/ML
FIGURE A8
25° C, BINRRY IQN-EXCHflNGE ISOTHERM
191
-------
UJ
uj
-------
« SULFflTE-NITRflTE
CHLORIDE-NITRflTE
0.20 0.10 0.60 0.80
EQUIVflLENT FRRCTION 30^ IN LIQUID PHflSE
XCL, EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
RESIN NUMBER 10
DUOLITE ES 374. MflCROPOROUS RESIN
POLYflCRYLIC MflTRIX
TERTIflRY-flMINE FUNCTIONflLITY
(POLYflMINE TITRflTION CURVE)
TOTflL CflPflCITY=3.0 MEQ/ML
FIGURE AIO
25° C, BINRRY ION-EXCHRNGE ISOTHERM
1.00
193
-------
o
o
UJ
tn
cr
z
Q_
Z
H-«
»••«
UJ
cn
cr
X
Q- o
00
"Z. • •
£0
-------
SULFflTE-NITRflTE
CHLORIDE-NITRflTE
0.20 O.UO 0.60 0.80 1.00
EQUIVflLENT FRflCTION 30^ IN LIQUID PHflSE
EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
RESIN NUMBER 12
IDNflC flFP 329. MflCROPOROUS RESIN
STYRENE-DVB MflTRIX
TERTIflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY«1.25 MEQ/ML
FIGURE A12
25° C, BINRRT ION-EXCHRNGE ISOTHERM
195
-------
SULFflTE-NITRflTE
0.20 0.40 0.60 0.80 1.00
EQUIVflLENT FRflCTION SO,, IN LIQUID PHflSE
RESIN NUMBER 13
IDNflC fl-305, MICROPOROUS GRflNULflR GEL
EPOXY-flMINE MflTRIX
POLYflMINE FUNCTIONflLITY (INCL. QUflT. flMINE)
TOTflL CflPflCITY=3.5 MEQ/ML
FIGURE A13
25° C, BINflRY lON-EXCHflNGE ISOTHERM
196
-------
SULFflTE-NITRRTE
CHLORIDE-NITRflTE
^.00 0.20 0.140 0.60 0.80
*sou, EQUIVflLENT FRflCTION 50^ IN LIQUID PHflSE
xa, EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
RESIN NUMBER 1U
flMBERLITE IRfl 910, MflCROPOROUS RESIN
STYRENE-DVB MflTRIX
TYPE II. QUflTERNflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY=1.0 MEQ/ML
FIGURE AM
25° C, BINflRY ION-EXCHRNGE ISOTHERM
197
-------
i 1 1 1 1 1 1
* SULFflTE-NITRflTE
0.20 0.40 0.60 0.80 1.00
Xsou, EQUIVflLENT FRACTION SO^ IN LIQUID PHflSE
RESIN NUMBER 15
flMBERLITE IRfl 400, MICROPOROUS GEL
STYRENE-DVB MflTRIX
TYPE I. QUflTERNflRT-RMINE FUNCTION0LITY
TOTflL CflPflCITY=1.4 MEQ/ML
FIGURE A15
25° C, BINRRT ION-EXLHRNGE ISOTHERM
198
-------
.00
SULFflTE-NITRflTE
CHLORIDE-NITRflTE -
0.20 O.UO 0.60 0.80
EQUIVflLENT FRflCTION 50^ IN LIQUID PHflSE
XCL, EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
RESIN NUMBER 16
RMBERLITE IRfl 1402. "IMPROVED" POROSITY RESIN
STYRENE-DVB MflTRIX
TYPE I. QUflTERNflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY=1.25 MEQ/ML
1.00
FIGURE A16
25° C. BINflRT ION-EXCHRNGE ISOTHERM
199
-------
* SULFRTE-NITRflTE
CHLORIDE-NITRRTE
0.20 0.10 0.60 0.80
, EQUIVflLENT FRflCTION 30^ IN LIQUID PHflSE
*CL, EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
RESIN NUMBER 17
flMBERLITE IRfl 900. MflCROPOROUS RESIN
STYRENE-DVB MflTRIX
TYPE I. QUflTERNflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY=1.0 MEQ/ML
1.00
y
FIGURE A17
25° C, BINRRT ION-EXCHRNGE ISOTHERM
200
-------
SULFflTE-NITRflTE
0.20 0.110 0.60 0.80 1.00
EQUIVflLENT FRRCTION SO,, IN LIQUID PHflSE
RESIN NUMBER 18
flMBERLITE IRfl 410, MICROPQROUS GEL
STYRENE-DVB MflTRIX
•TYPE II. QUflTERNflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY=1.35 MEQ/ML
FIGURE A18
25° C, BINflRT ION-EXCHRNGE ISOTHERM
201
-------
SULFflTE-NITRflTE
0-20 0.40 0.60 0.80 1.00
*sou, EQUIVflLENT FRflCTION 30^ IN LIQUID PHflSE
RESIN NUMBER 19
DOWEX SBR-P, "IMPROVED" POROSITY GEL
STYRENE-DVB MflTRIX
TYPE I. QUflTERNflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY=1.2 MEQ/ML
FIGURE A19
25° CP BINRRT ION-EXCHRNGE ISOTHERM
202
-------
I 1 ) 1
• SULFRTE-NITRflTE
.00
0.20 0.110 0.60 0.80 1.00
EQUIVRLENT FRACTION SOy IN LIQUID PHflSE
RESIN NUMBER 20
DOWEX SflR. MICROPOROUS GEL
STYRENE-OVB MflTRIX
TYPE II. QURTERNRRY-RMINE FUNCTIONRLITY
TOTflL CfiPflCITY=l.U MEQ/ML
FIGURE A20
25° C, BINRRY ION-EXCHRNGE ISOTHERM
203
-------
SULFRTE-NITRflTE
CHLORIDE-NITRflTE
.00
0.20 O.UO 0.60 0.80
EQUIVflLENT FRflCTION SO^ IN LIQUID PHflSE
EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
RESIN NUMBER 21
DOWEX SBR. MICROPOROUS GEL
STYRENE-DVB MflTRIX
TYPE I. QUflTERNflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY=l.i4 MEQ/ML
1.00
FIGURE A21
25° C, BINflRY ION-EXCHRNGE ISOTHERM
204
-------
SULFRTE-NITRflTE
I 1 1 1 1
0.20 O.UO 0.60 0.80 1.00
EQUIVflLENT FRflCTION SO IN LIQUID PHflSE
RESIN NUMBER 22
DOWEX 11. "IMPROVED" POROSITY GEL
STYRENE-DVB MflTRIX
TYPE I, QUflTERNflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY=1.2 MEQ/ML
FIGURE A22
25° C, BINflRY ION-EXCHRNGE ISOTHERM
205
-------
SULFflTE-NITRflTE
•+-
-4-
-+-
0.20 0.140 0.60 0.80 1.00
EQUIVflLENT FRflCTION SO IN LIQUID PHflSE
RESIN NUMBER 23
DUOLITE FM02-D. MICROPOROUS GEL
STYRENE-DVB MflTRIX
TYPE II. QUflTERNflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY^l.4 MEQ/ML
FIGURE A23
25° C, BINflRT lON-EXCHflNGE ISOTHERM
206
-------
t 1 1 1 1 1 J 1 1
e SULFflTE-NITRflTE
0.20 O.UO 0.60 0.80 1.00
EQUIVflLENT FRflCTION Sfy IN LIQUID PHflSE
RESIN NUMBER 2U
DUOLITE fl-101-D. "IMPROVED" POROSITY GEL
STYRENE-DVB MflTRIX
TYPE I. QUflTERNflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY=1.3 MEQ/ML
FIGURE
25° C, BINRRT ION-EXCHRNGE ISOTHERM
207
-------
SULFflTE-NITRRTE
0.20 O.UO 0.60 0.80 1.00
, EQUIVflLENT FRRCTION Sfy IN LIQUID PHflSE
RESIN NUMBER 25
DUOLITE fl-104. MICROPOROUS GEL
STYRENE-DVB MflTRIX
TYPES I 4 II QUflTERNflaY-flMINE FUNCTIONflLITY
TOTflL CflPfl/ITY=1.5 MEQ/ML
FIGURE A25
25° C, BINRRY ION-EXCHRNGE ISOTHERM
208
-------
SULFRTE-NITRflTE
0,20 O.UO 0.60 0.80 1.00
EQUIVflLENT FRflCTION SO^ IN LIQUID PHflSE
RESIN NUMBER 26
IONFIC fl-550, "IMPROVED" POROSITY GEL
POLYSTYRENE MflTRIX
TYPE II, QURTERNflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY=1.3 MEQ/ML
FIGURE A26
25° CP BINRRT IDN-EXCHflNGE ISOTHERM
209
-------
SULFflTE-NITRflTE
).00
0.20 0.140 0.60 0.80 1.00
EQUIVflLENT FRflCTION 5 IN LIQUID PHflSE
RESIN NUMBER 27
IDNflC flSB-1. MICROPOROUS GEL
STYRENE-DVB MflTRIX
TYPE I. QUflTERNflRY-flMINE FUNCTIONflLITY
TOTflL CflPaCITY=l.U MEQ/ML
FIGURE A27
25° Cp BINRRT ION-EXCHRNGE ISOTHERM
210
-------
SULFflTE-NITRflTE
* CHLORIDE-NITRflTE
°b.OO
0.20 O.HO 0.60 0.80
-------
SULFflTE-NITRRTE
CHLORIDE-NITRflTE
0.20 O.UO 0.60 0.80
EQUIVRLENT FRflCTION SO^ IN LIQUID PHflSE
EQUIVflLENT FRflCTION CL IN LIQUID PHflSE
RESIN NUMBER 29
IDNflC flSB-2. MICROPOROUS GEL
STYRENE-DVB MflTRIX
TYPE II. QUflTERNflRY-flMINE FUNCTIONflLITY
TOTflL CBPflCITY=1.52 MEQ/ML
1.00
FIGURE A29
25° C, BINRRT ION-EXCHRNGE ISOTHERM
212
-------
* SULFflTE-NITRflTE
"^.00
0.20 O.UO 0.60 0.80 1.00
EQUIVflLENT FRflCTION SO^ IN LIQUID PHflSE
RESIN NUMBER 30
IONRC flSB-lP. "IMPROVED" POROSITY GEL
STYRENE-DVB MflTRIX
TYPE I. QUflTERNflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY=1.35 MEQ/ML
FIGURE A30
25° C, BINRRT IQN-EXCHflNGE ISOTHERM
213
-------
I 1 1 1 1 1 ( )
e SULFflTE-NITRflTE
^.00
0.20 0.40 0.60 0.80 1.00
EQUIVflLENT FRflCTION SO IN LIQUID PHflSE
RESIN NUMBER 31
IQNflC fl-540. "IMPROVED" POROSITY GEL
POLYSTYRENE MflTRIX
TYPE I. QUflTERNfiRY-flMINE FUNCTIONflLITY
TOTflL CflPRD!TY=1.0 MEQ/ML
FIGUREA31
25° C, BINflRY ION-EXCHRNGE ISOTHERM
214
-------
» SULFflTE-NITRflTE
^.00
0,20 0.10 0.60 0.80 1.00
EQUIVflLENT FRflCTION SO^ IN LIQUID PHflSE
RESIN NUMBER 32
IQNflC flFP-100. MflCROPOROUS RESIN
STYRENE-DVB MflTRIX
TYPE I. QUflTERNflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY*1.2 MEQ/ML
FIGURE A32
25° C, BINflRY ION-EXCHRNGE ISOTHERM
215
-------
SULFRTE-NITRRTE
CHLORIDE-NITRRTE
.00 0.20 0.40 0.60 0.80 1,
XSQU. EQUIVflLENT FRflCTION 50^ IN LIQUID PHRSE
*CL. EQUIVflLENT FRRCTION CL IN LIQUID PHRSE
RESIN NUMBER 3
RMBERLITE IR 45. MICROPOROUS GEL
00
FIGURE NO. A33
COMPARISON OF CURVE FITTING TECHNIQUES
ASSUMING
CONSTANT SEPARATION FACTOR
216
-------
SULFflTE-NITRfl
CHLORIDE-NITRflTE
0.20 O.UO 0.60 0.80 1.00
EQUIVfiLENT FRflCTION S04 IN LIQUID PHflSE
EQUIVflLENT FRACTION CL IN LIQUID PHflSE
RESIN NUMBER 8
OUOLITE ES 368, MflCROPOROUS RESIN
FIGURE NO. A34
COMPARISON OF CURVE FITTING TECHNIQUES
ASSUMING
CONSTANT SEPARATION FACTOR
217
-------
§
•,
r*
H h
APPENDIX B
TITRATION
CURVES
*> UPPER CURVE
Q, MIDDLE CURVE
+, LOWER CURVE
HCL
0.50
1.00 1.50 2.00
RCID RDDED. MEQ./ML.
2.50
RMBERLITE IRB 93, MflCROPOROUS RESIN
STYRENE-DVB MflTRIX
TERTIflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY =1.25 MEQ./ML.
FIGURE Bl
RESIN NO. 1 TITRflTION CURVES
218
-------
i 1 1 ,-H 1 1
UPPER CURVE
MIDDLE CURVE = H
+, LOWER CURVE = HCL
~b.oo
0.50 1,00 1.50 2.00
flCID flDOEO. MEQ./ML.
2.50
flMBERLITE IRfl 68, MICROPOROUS GEL
POLYflCRYLIC MflTRIX
TERTIflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY = 1.6 MEQ./ML.
FIGURE B2
RESIN NO. 2 TITRflTION CURVES
219
-------
o
o
0
-------
o
o
A UPPER CURVE =
Cl MIDDLE CURVE = HNV
+, LOWER CURVE = HCL
0.50
1.00 1.50
flCIO flDDED, MEQ./ML.
2.00
2.50
DOWEX WGR, MICROPOROUS GEL
EPOXT-flMINE MflTRIX
POLYflMINE FUNCTIONflLITY
TOTflL CflPflCITT =1.0 MEQ./ML.
FIGURE W
RESIN NO. 4 TITRflTION CURVES
221
-------
8
0.00
^ UPPER CURVE -
Q MIDDLE CURVE -
+. LOWER CURVE « HCL
0.50
t.OO 1.50 2.00
flCID flDOEO, M£Q./ML.
2.50
DOWEX MMR-1, MflCROPOROUS RESIN
STYRENE-DVB MflTRIX
TERTIflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY » 1.1 MEQ./ML.
FIGURE B5
RESIN NO. 5 TITRflTION CURVES
222
-------
UPPER CURVE =
MIDDLE CURVE = HNOa
+, LOWER CURVE - HCL
~0'.00 0.50 1.00 1.50 2.00
flCID flDDED. MEQ./HL.
2.50
OUOLITE fl-7, MflCROPOROUS, GRflNULflR RESIN
PHENOL-FORMflLDEHYDE MflTRIX
SECONDRRY-flMINE FUNCTIONflLITY
(POLYRMINE TITRflTION CURVE)
TOTflLCRPRCITYR||.UMEQ./ML.
RESIN NO. 6 TITRRTION CURVES
223
-------
UPPER CURVE
MIDDLE CURVE
+, LOWER CURVE = HCl
0.50
1.00 1.50 2.00
flCID flDDED, MEQ./ML.
2.50
DUOLITE R 340, MICROPOROUS GEL
EPOXY-flMINE MflTRIX
POLYflMINE FUNCTIONflLITY
TOTflL CflPflCITY =2.6 MEQ./ML.
FIGURE W
RESIN NO. 1 TITRflTION CURVES
224
-------
0.00
A UPPER CURVE =
Q MIDDLE CURVE =
+. LOWER CURVE = HCL
0.50
1.00 1.50 2.00
RCID flDDED, MEQ./ML.
2.50
DUOLITE ES 368, MflCROPOROUS RESIN
STYRENE-DVB MRTRIX
TERTIflRY-flMINE FUNCTIONflLITY
TOTflL CflPflCITY = 1.3 MEQ./ML. .
FIGURE B8
RESIN NO. 8 TITRflTION CURVES
225
-------
8
t.
r-
1 j.
4, UPPER CURVE -
Q MIDDLE CURVE
•f. LOWER CURVE - HCL
0.50
1.00 1.50
flCID BODED. HEQ./ML.
2.00
2.50
DUOLITE ES 561, MflCROPtiROUS GRflNULflR RESIN
PHENOL-FORMflLDEHYDE MflTRIX
POLYRMINE FUNCTIONflLITY
TOTflL CflPflCITY =2.0 MEQ./ML.
FIGURE B9
RESIN NO. 9 TITRflTION CURVES
226
-------
o
o
r^
o
o
CO
o
o
u5
E
o
o
$0
CO
o
o
t
CM
o
o
+, UPPER CURVE = H»SOu
-------
UPPER CURVE
MIDDLE CURVE » HNOg
+. LOWER CURVE * HCL
~o.oo
0.50
1.00 1.50 2.00
flCJD RDDED. MEQ./ML.
2.50
IONRC fl-260. MICROPOROUS GRflNULflR GEL
RLIPHflTIC-flMINE MflTRIX
POLYflMINE FUNCTIONflLITY
TOTRL CflPflCITY = 1.8 MEQ./ML.
FIGURE Bll
RESIN NO. 11 TITRflTION CURVES
228
-------
UPPER CURVE •
MIDDLE CURVE - HNO,
+, LOWER CURVE « HO/
"b.oo
0.50
1.00 1.50
flCID flDDEO, NEQ./HL.
2.00
2.50
IDNflC flFP 329, MflCROPOROUS RESIN
STYRENE-DVB MflTRIX
TERTIflRY-flMINE FUNCTIONRLJTT
TOTflL CflPflCITY =1.25 MEQ./ML.
FIGURE B12
RESIN NO. 12 TITRflTION CURVES
229
-------
8
in'
DUOLITE ES-368
STY-DVB, TERTIflRY-flMINE RESIN
TWO-BED SYSTEM, flCID ELUTION
COLUMN DIfl. = 1 INCH C2.5H CMJ
BED DEPTH = 25 INCHES (63.5 CMJ
FLOW RflTE = 2.3U GflL./MIN.FT.3 (3.2 MINJ
100.00
200.00 300.00
400.00 500.00 '600.00
BED VOLUMES OF EFFLUENT
700.00
800.00
900.00
1000.00
FIGURE Cl
RUN NO. 1, EFFLUENT CONCENTRflTION PROFILE
-------
s
• 4
in
8
DUOLITE ES-374
POLYflCRYLIC, POLYflMINE RESIN
TWO-BED SYSTEM. RCID ELUTION
COLUMN DIfl. = 1 INCH (2.5U CMJ
BED DEPTH = 12 INCHES (30.5 CM.)
FLOW RflTE * 1.88 GflL./MIN.FT.3 (1.53 MIN.)
flVE. SOu/NOg SEPflRflTION FflCTOR = 9H.O
flVE. NCfe/CL SEPflRflTION FRCTOR = 3.85
a
UJ
a?
HN
600.00 800.00 1000.00 1200.00
BED VOLUMES OF EFFLUENT
1400.00 1600.00 1800.00 2000.00
FIGURE C2
RUN NO. 2, EFFLUENT CONCENTRflTION PROFILE
-------
8
in
DUOLITE ES-368
STY-DVB, TERTIflRY-flMINE RESIN
TWO-BED SYSTEM, flCID ELUTION
COLUMN DIfl. - i INCH (2.54 CM.)
BED DEPTH « 12 INCHES (30.5 CM.)
f.3 Cl.
flVE. SOu/NOa SEPflRflTION FflCTOR = 2.83
flVE. NOa/CL SEPflRflTION FflCTOR
3.87
1.88 GRL./MIN.FT,
,53 HIM.)
100.00
200.00
300.00
100.00 500.00 600.00
BED VOLUMES OF EFFLUENT
700.00
800.00
900.00
1000.00
FIGURE C3
RUN NO. 3. EFFLUENT CONCENTRflTION PROFILE
-------
8
UJ
o
o
a
£88
UJ
8
DOWEX WGR
EPOXY RHINE POLTflMINE RESIN
TNO-BED SYSTEM, RCID ELUTION
COLUMN DIfl. = 1 INCH (2.5U CM.)
BED DEPTH = 12 INCHES (30.5 CM.)
FLOW RflTE = ii.88 GfiL./MIN.FT.3 (1.53 MIN.)
flVE.
flVE.
SEPflRflTION FflCTOR = 137.
/CL SEPflRflTION FflCTOR = 1.99
HNC
200.00 400.00 600.00
800.00
BED VOLUMES OF EFFLUENT
FIGURE C4
RUN NO. 14, EFFLUENT CONCENTRflTION PROFILE
1000.00 1200,00 UOO.OO 1600.00 1800.00 2000.00
-------
§
in
s
ss
MSO
u» Qo
** K-OI
DUOLITE ES-368
STY-DVB, TERTIflRY-flMINE RESIN
TWO-BED SYSTEM, flCID ELUTION
COLUMN OIR. = 1 INCH (2.5U CM.)
BED DEPTH « 24 INCHES (61.0 CM.)
FLOW RflTE » 2.W GflL./MIN.FT.3 (3,1 MIN.)
flVE. SOu/NQs SEPRRflTION FflCTOR = 2,83
flVE. NCL/CL SEPflRflTION FftCTOR = 3.87
HN
100.00
200.00
300.00
400.00 500.00 600.00
BED VOLUMES OF EFFLUENT
700.00
800.00
900.00
FIGURE C5
RUN NO. 5, EFFLUENT CONCENTRRTION PROFILE
1000.00
-------
8
•
in
3
•
§0
»-« •
t-m
|_ CM
o
o
o
o
DUOLJTE ES-374
POLYflCRYLIC, POLYflMINE RESIN
THO-BED SYSTEM, RCID ELUTION
. *
°0.00
COLUMN DIfl. = 1 INCH (2.511 CMJ
BED DEPTH = 2U INCHES (61.0 CMJ
FLOW FHTE = 2.m GflL./MIN.FT.3 (3.1 MINJ
flVE. SOu/NOg SEPHRflTION FflCTOR = 91.0
flVE. N%/CL SEPRRflTION FflCTOR = 3.85
I | I I I -I I I I I I ' I
200.00 400.00 600.00 800.00 1000,00 1200.00 1400.00 1600.00 1800.00 20022OKOO2400.00
BED VOLUMES OF EFFLUENT
FIGURE C6
RUN NO. 6, EFFLUENT- CONCENTRflTION PROFILE
-------
OJ
8
U)
a
UJ
flVE.
flVE.
/NQs SEPRRflTION FflCTOR
CL SEPflRflTION FflCTOR
©» HCOjf
+ * CL-
1.76
2.97
8
IDNflC flFP-100
STY-DVB, QUflT. (1) RESIN
SINGLE BED. NEUTRflL ELUTION
COLUMN DIfl. * 1 INCH (2.5U CM.)
BED DEPTH « 2t INCHES (61.0 CM.)
FLOW RflTE * 2.11 GflL./MIN.FT.8 (3.1 MIN.)
200
300 UOO
BED VOLUMES OF EFFLUENT
500
600
FIGURE C7
RUN NO, 7, EFFLUENT CONCENTRflTION PROFILE
-------
s
9
in
8
<*
1
58
•-•~J
,_CO
to
OJ
is
-CM
£
s
DUOLITE ES-368
STY-DVB, TERTIRRY-flMINE RESIN
TWO-BED SYSTEM
flTTEMPTED NEUTRflL ELUTIONr+
COLUMN DIfl. = 1 INCH (2.5y CM.)
BED DEPTH = 2U INCHES (61,0 CMJ
FLOW RflTE = 2,iW GflL./MIN.FT.8 (3.1 MIN.)
flVE. SOu/NO, SEPRRRTION FflCTOR « 2.83
flVE. NCU/CT SEPfiRflTION FflCTOR = 3,87
100.00
200.00
300.00
400.00 500.00 600.00
BED VOLUMES OF EFFLUENT
700.00
800.00
FIGURE C3
RUN NO. 8, EFFLUENT CQNCENTRflTION PROFILE
900.00
1000.00
-------
8
• •
in
8
DUOLITE ES-368
STY-DVB, TERTIflRY-flMINE RESIN
TWO-BED SYSTEM. NEUTRRL ELUTION
REGENERflTION^ 120X OF THEORY
COLUMN DIfl, = 1 INCH (2.5U CM.)
BED DEPTH = 2U INCHES (61.0 CM.)
FLOW RflTE = 2.«« GflU/MIN.FT.8 (3.1 MIN.)
flVE, SOu/NOj SEPflRflTION FflCTOR = 2.83
flVE. NOL/CL SEPflRflTION FflCTOR « 3,87
'
100.00
200.00
300.00
400.00 500.00
BED VOLUMES OF
600.00
700.00
800.00
900.00
FIGURE C9
RUN NO. 9. EFFLUENT CONCENTRflTION PROFILE
1000.00
-------
DUOLITE ES-368
STY-DVB, TERTIflRY-flMINE RESIN
TWO-BED SYSTEM. NEUTRflL ELUTION
REGENERATION = 2UOX OF THEORY
COLUMN DIfl. = 1 INCH (2.54 CM.)
BED DEPTH = 2U INCHES (61.0 CM.)
FLOW RflTE « 2.14U GflL./MIN.FT.8 (3.1 MIN.)
flVE. SOii/NO. SEPflRflTION FflCTOR = 2.
flVE. Ntt,/CL SEPflRflTION FflCTOR * 3.87
100.00 200.00
300.00
UOO.OO 500.00 800.00
BED VOLUMES OF EFFLUENT
700.00 800.00
900.00
1000.00
FIGURE CIO
RUN NO. 10, EFFLUENT CONCENTRflTION PROFILE
-------
§
iri
Max =5.13
8
flMBERLITE IR
STY-DVB, POLTflMINE RESftN
TWO-BED SYSTEM. NEUTRflL EILUTION
COLUMN DIfU « 1 INCH (2.54 CHI)
BED DEPTH * 24 INCHES (61.0 CM.)
FLOW RRTE - 2,VI GflL./MIN.FT.9 (3.I/MINJ
18
100.00
200.00
300.00
400.00 500.00 600.00
BED VOLUMES OF EFFLUENT
700.00
800.00
900.00
1000.00
FIGURE Gil
RUN NO. 11, EFFLUENT CONCENTRflTION PROFILE
-------
APPENDIX D
EXPERIMENTAL APPARATUS AND PROCEDURES
TABLE Dl.
U.S. ION EXCHANGE RESIN MANUFACTURERS
Manufacturer Trade Name
Dow Chemical Company Dowex
Functional Products and Systems Dept. Resins
Midland, Michigan 48640
Diamond Shamrock Chemical Co.
Noplo Chemical Division Duolite
1901 Spring St. Resins
Redwood City, CA 94063
lonac Division of Sybron Corp. lonac
Dirmingham, New Jersey 08011 Resins
Rhom and Haas Company
Fluid Process Chemicals Dept. Amberlite
Philadelphia/ PA 19105 Resins
NOTES: A complete list (1967) of World-Wide "Producers of
Ion-Exchange Materials" can be found on p. 85 of:
Operation and Control of Ion-Exchange Processes for
Treatment of Radioactive Wastes" Reference 68.
Other descriptions and sources of ion exchangers are
listed in the appendices of Dorfner's Ion-Exchangers
(1972), Reference 40, and Helfferich's Ion Exchange
241
-------
(1962) Reference 60.
TABLE D2.
CHEMICAL MAKE-UP OF Na TEST WATER
(See Also Table 20)
*
Chemi cal
NaHC03
NaN03
Na0SO,
2 4
NaCl
Formula
Weight
84.01
84.99
142.0
58.44
m moles
a
1.00
1.50
0.75
1.50
gms
100 £
8.401
12.75
10.65
8.77
TABLE D3.
CHEMICAL MAKE-UP OF Ca-Mg-Fe TEST WATER
(See Also Table 21)
*
Chemical
NaHC03
Ca(N03)2'4H20
MgS04
CaCl2'2H20
FeSO4*7H20
Formula
Weight
84.01
236.15
120.37
147.03
^278.03
m moles
SL
1.00
0.75
0.75
0.75
1 ppm Fe
gms
100 A
8.401
17.71
9.03
11.03
0.500
Order
of
Addition
1
2
3
4
5
Dissolve weighed salt in small quantity of distilled water
(approximately 1 &) before adding to 100 £ batch.
Dl: PROCEDURE FOR RESIN CONDITIONING
1) Place 500 mJl or 1000 mJl weak base resin sample in free
base form into 51 cm I.D. resin conditioning column (Figure Dl).
2) Backwash with tap water at 100% or greater bed expan-
sion for 5-30 minutes or until effluent appears clear and color-
less. ;
3) Exhaust resin sample downflow with 2-3 BV of 2.0 N HC1
during a 30-45 minute period; flow rate - 70 mJl/min for the 1 I
242
-------
resin samples.
4) Rinse downflow with 4-6 BV of distilled water for a
total rinse contact time of about 30 minutes; flow rate - 130
m&/min for 1 a resin samples.
5) Regenerate downflow with 2-3 BV of 1.5 N NaOH during
a 30-45 minute contact time; flow rate - 70 mA/min for 1 H of
resin.
6) Repeat step 4: distilled water rinse.
7) Repeat step 3: exhaustion with HC1.
8) Repeat step 4: distilled water rinse.
9) Repeat step 5: regeneration with NaOH.
10) Repeat step 4: distilled water rinse.
11) Drain column till water level is about 8 cm above resin
level.
12) Slurry the resin by rocking the stoppered column, then
transfer resin-water slurry to polyethylene bucket by multiple
rinsing of inverted column.
13) Decant supernatant distilled water and any floating
beads or debris then return wet resin slurry to original con-
tainer.
14) Place 100 m& of wet resin slurry into 350 mJl EC
fitted glass funnel on 1000 mil vacuum flask.
15) Rinse three times with about 60 m£ of distilled water
each time then wash continuously with a stream of about 50 m£
of distilled water from squirt bottle.
16) Draw air thru resin for about 2 minutes.
17) Transfer damp resin to polyethylene tray and air dry
for about 3-5 days at 25° C in walk-in incubator at about 50%
relative humidity.
NOTES: a) For strong-base resins in the chloride form and
strong-acid resins in the Hydrogen form, the sequence of
acid-base addition is reversed.
b) A total of six columns were used.
c) This procedure was carried out on 32 anion resin
samples and 4 cation resin samples. All resins used for
experiments were conditioned by this procedure.
D2: RESIN CONVERSION PROCEDURE
Objective: to convert the free base form of a weak base resin
to the nitrate form at .005 N equilibration concen-
tration .
1) Assemble resin conversion apparatus as in Figure D2 .
2) Place about 75 m£ of wet, free-base form, weak-base
anion resin into 21 mm. I.D. glass ion-exchange column.
3) Backwash resin with tap water and allow to settle.
Drain water to a level about 1 cm above resin bed.
4) Pass 400% stoichiometric excess (about 300 mJl) of 2 N.
HNO-, through the resin at about 1 gal/min ft3 (T = 7.5 min) for
a total contact time of about 45 minutes.
243
-------
5) Rinse with 10 BV (750 mA) of .00500 N. HNO3 at about
20 ml/min.
6) Allow resin to equilibrate with .00500 N. HN03 overnight
in the column.
7) Remove resin from column, filter on glass frit into
vacuum flask. Draw air through resin for about 1 min. after
washing resin with a stream of about 50 m£ of .00500 N. HNO3
from a wash bottle.
8) Rinse quickly with two separate, one BV quantities of
0.001 N. HN03 to remove adhering .00500 N. HNO-j.
9) Draw air through resin for about one minute.
10) Transfer dry, caked resin from filter to polyethylene
pan and air dry in walkin incubator for 2-4 days at ambient
humidity and 25° C.
11) Store air-dried, nitrate form resin in 4 02. wide-
mouth polyethylene bottles for use in isotherm experiments
and capacity determinations.
NOTES: a) This same procedure was used to prepare Chloride
and Sulfate form resins at .00200, .00500 and .00800 N.
with those acid and those concentrations being substitu-
ted where appropriate in the procedure above. In all
cases the initital conversion was done with 2N acid and
the final, fast rinse was done with 0.001 N. acid.
b) a total of six columns were constructed and typi-
cally 6 resins were converted simultaneously.
D3: PROCEDURE FOR EQUILIBRIUM ISOTHERM CONSTRUCTION
Objectives; To obtain experimental data at 25° C so that sul-
fate/nitrate and chloride/nitrate equilibrium iso-
therms might be developed for all the anion resins.
To obtain at least five equally distributed equili-
brium points for each isotherm at a total solution
concentration of 0.005 N.
Note: The example discussed here is for the construction of a
sulfate/nitrate isotherm starting with a resin in the
nitrate form (see Procedure D2) which is placed into
0.005 N H2S04 for equilibration.
1) Estimate the grains of resin in the nitrate form which
must be added to 100 ml aliquots of 0.005 N H2SO, to achieve
equilibrium sulfate concentrations (xs's) approximately equal to
0.1, 0.3, 0.5, 0.7, and 0.9. To accomplish this, the ion-ex-
change capacity and the separation factors for sulfate/nitrate
exchange must be known or estimated. In this work, the capa-
cities were known from measurements and published data, and the
separation factors were estimated (but the original estimates
were not very good in many cases). Equation E10 (from Appendix
244
-------
E, Example Calculation El) is then used to calculate the resin
weights required for each equilibrium point.
2) Add the calculated amounts of resin to the 100 ml
aliquots of H,,S04 in 125 ml French-square bottles and tumble at
13 rpm for 242hours at 25° C.
3) Analyze the equilibrated supernatants for sulfate and
nitrate using Procedures D7 and D8 of this Appendix.
4) Using the predetermined nitrate capacity of the resin,
calculate the meq of nitrate remaining on the resin at equili-
brium.
5) Calculate the meq of sulfate on the resin by measuring
the disappearance of sulfate from the liquid phase.
6) Knowing the meq of sulfate and nitrate in both phases,
calculate the equilibrium equivalent fractions of each of the
ions in each phase.
7) Construct the isotherms by plotting yg ys xg for each
equilibrium point for a given resin. See Appendix A Figures
Al— A32 .
Notes: Because there were so many data points, a HP-25 Program-
mable calculator was used to calculate the x, 's and y.'s
and separation factors from the experimental data.
Because the equilibrium data were to be used for several
data plots, they were stored in an MTS computer file for
use in generation of the isotherm plots: Text Figures
12 — 17 and Appendix Figures Al — A32. Alternatively,
resins in the sulfate form previously equilibrated with
0.005 N H2S04 may be used with 0.005 N HN03 solution to
construct the same isotherms. This procedure was
followed for construction of the lower hysteresis iso-
therm shown in Figure 18.
The procedures above are essentially the same for the
construction of the chloride/nitrate isotherms, but HC1
is substituted for H
D4: BICARBONATE SELECTIVITY DETERMINATION PROCEDURE
Objectives: To determine the carbonic acid/nitric acid selecti-
vities of a representative number of anion resins;
To determine if a significant amount of H2CO- is
taken up at low pH (2.5 - 3.0) by weak ana strong
base resins.
1) Weigh out 1.00 meq of resins in the Chloride form using
the experimentally determined capacities in meq/gm of air-dried
resin.
2) Transfer weighed resins to the mini columns (figure Dl)
filled with distilled water.
3) Assemble mini columns into the apparatus as shown in
Figure D4 . The cation column contained 120 ml of Duolite C-20
resin in the hydrogen form with an approximate capacity of 240
245
-------
meq. A 16 liter reservior of feed solution at a total concen-
tration of 5 meq/H (80 meq total), not shown in the photo, is
used to gravity feed the system. Five different feed solution
mixtures of sodium nitrate and sodium bicarbonate were used
having the following equivalent fractions of nitrate and bicar-
bonate :
Feed Solution XT,™
Label HCQ3
.1 .1
.3 .3
.5 .5
.7 .7
.9 .9
4) Pass approximately 1 £ of cation effluent through each
of the mini columns at a rate of about 5-6 mA/min. This pro-
vides a 400% stiochiometric excess and an exposure time of about
21/2 hours.
5) Drain the column down to the top of the resin.
6) Remove mini-columns from apparatus and regenerate by
adding 5 mJl of 2% NH4<3H (0.57 meq/ml) to each column using care
not to mix the resin and regenerant (Use 0.5 N. NaOH for strong-
base resins).
7) Drip this slowly through the column for 15-20 minutes,
i.e. about 1 nUl/4 min., into a 100 ml volumetric flask 1/2
filled with distilled water.
8) Repeat step 6.
9) Repeat step 7.
10) Rinse slowly with about 5 ml of distilled water over a
10 min. period.
11) Repeat step 10.
12) Rinse quickly, pinch clamp opened wide, with 5 m£ of
distilled water. .
13) Repeat step 12.
14) Repeat step 12.
15) Make up to 100 ml with distilled water.
16) Repeat steps 1-15 for each of the five feed solutions
listed in step 3.
17) Analyze for nitrate by the UV method which has been
demonstrated by experiment to be free from interferences from all
the anions and cations present in these regenerant solutions.
18) Analyze for bicarbonate using the inorganic carbon
channel of the Beckman TOC analyzer using 100 microliter samples.
NOTES: a) Weak base resins 1-6 & 8-10 and strong-base resins
16, 19 and 21 were analyzed in this fashion. Very little
H2C-3 was ta^en UP even at 0-9 equivalent fraction of
HCOl" in feed. The range of H2CO3 uptake was 2-5% of the
total resin capacity or this pH \2.4-3.0).
b) A definite kinetic effect was noted; H-CO- did pro-
mote the uptake of HN03. For example the resins went from typi-
246
-------
— -** J>xo ^-^-''j ^f **-t*-rh-^-; • **rf ^— ** •- j__y At. «— t&jx^i«9 cut ^-M ** •*• v cio-^ii i* ^ j. aw L>.
of HCOZ greater than 0.10 to facilitate the stoichiometric
uptake of HNO3 under these conditions.
D5: TITRATION CURVE DETERMINATION PROCEDURE
Objectives: To establish titration curves for HC1, HNO- and
H-SO. for each of the weak-base anion resins: these
curves to be used later in the analysis of selec-
tivity data and to help model resin behavior in
fixed bed processes.
TO determine the pKa's of each of these resins. To
compare the capacities of each for Chloride, Nitrate
and Sulfate as a function of pH.
1) Prepare the following standard acids and check by tit-
ration with 1.000 NaOH to pH 7.0 and 4.5. Standardize to 4
significant figures.
1.000 N. HC1
1.000 N. HNO-
1.000 N. H2SO,
2) Weigh out 500 mg (+2 mgf of air-dried free-base form
resin into 125 ml, square, glass, wide-mouth bottles: forty-
two separate weighings for each different resin.
3) Into each of the above bottles place one of the follow-
ing amounts of standardized HC1, HNO^, or H-SO..
Bottle No. Meg of Acid Ml of Dist. H0
1 0.0 100
2 0.30 100
3 0.70 99.0
4 1.10 99.0
5 1.40 98.5
6 1.60 98.5
7 1.80 98.0
8 2.00 98.0
9 2.20 98.0
10 2.40 97.5
11 2.60 97.5
12 2.90 97.0
13 3.30 97.0
14 3.70 96.0
4) Place bottles into tumbler (Figure D3 ) and rotate for
16 to 24 hours at 13 rpm.
5) Measure and record pH of equilibrated samples using
potentiometric, strip-chart recorder to determine when pH
reading has stabilized. Do not rinse electrode with water be-
tween samples; simply touch a Kimwipe to the bottom of the com-
247
-------
bination electrode to absorb adhering sample before immersion
into new sample.
6) Determine density of air-dried, free-base form resin by
weighing 3.500 gm into glass bottle, adding 100 ml distilled
water, tumbling overnight and measuring volume in 10 ml graduate
after light tamping and settling for 10 minutes.
7) Plot pH vs meq acid added and pH vs meq/ml of resin.
See Figures Bl thru B13. +
8) Determine capacities at any pH by assuming+H ion
activity = H concentration in 100 ml liquid; and H ion added,
but not in liquid, is in solid phase (resin). See Table Bl for
resin capacity comparisons.
NOTES: a) Duolite ES-368 floats and adheres to pH electrode.
b) Densities of Chloride forms of strong base resins
were determined as in step 6 except 3.00 gm resin used.
D6: HC1 CAPACITY DETERMINATION PROCEDURE
(Generally the same procedure is used
for HNO3 and H2SO4 Capacity)
1) Into Al weighing disnes weigh out duplicate 0.600 gm
(+1 mg) air-dried, Chloride form of resins previously equilibra-
ted with .00500 N HC1.
2) Quantitatively transfer the weighed resin samples into
mini-columns (Figure Dl) and cover with distilled water.
3) Carefully add 10 ml of 2 N. HNO3 to the column and drip
slowly through resin (total contact time of 15 to 30 minutes)
collecting the HNO- in a beaker.
4) Repeat step 3 collecting the regenerant HNO^ in the
same beaker.
5) Rinse quickly with 10 ml of .005 N.HNO- collecting rinse
in same beaker.
6) Repeat step 5 with another 10 ml of 0.005 N.HNO3.
7) Rinse quickly with 10 ml of distilled water into same
beaker.
8) Titrate the total regenerant volume including rinses
with standardized 0.100 N. AgNO- to + 290 mV end point with
Fisher Automatic Titrator using double junction calomel reference
electrode and Ag/AgS specific ion electrode. See Potentiometrie
Titration Method for Chloride for full details.
NOTES: a) Blank for step 8 = 20 ml of 2N. HNO- + 20 ml of .005
N. HNOq + 10 ml of distilled water.
b) Standard = 20 ml of 2 N. HNO, + 20 ml of .005 N. HNO3
+ 10 ml distilled water + 4.00 ml^of 1.000 N. NaCl.
c) To determine HN03 capacity, start with air-dried,
nitrate form and elute with HC1. Dilute and analyze for
nitrate by UV method.
d) To determine H2SO. capacity, start with air-dried,
sulfate form resins and elute with HC1. Dilute and ana-
lyze for sulfate by modified turbidimetric method.
248
-------
D7: MODIFIED TURBIDIMETRIC METHOD FOR SULFATE
Reference: Standard Methods, 13th Ed., p. 334
1) Place 100 ml sample in 300 ml Erlenmeyer flask.
2) Add 5 ml conditioning reagent.
3) Add one "scoop" (0.2 - 0.3 ml) of reagent grade barium
chloride.
4) Shake by hand swirling occasionally for one minute.
5) Allow 4 additional minutes for turbidity to develop with
no additional agitation.
6) Set 10 ppm sulfate to read "100" on 0-100 scale of Hach
turbidimeter (Model 2100A) using 25 ml sample and no spacer in
the reading chamber.
7) Read turbidity of all standards and samples after
exactly 4 minutes of turbidity development following initial
1 minute agitation period.
8) Plot NTU vs ppm sulfate and read off samples
Typically,ppm = 1.0 + 0.09 NTU
NOTES: a) Linear range is 2-10 ppm sulfate.
b) Standards typically 2, 4, 6, 8, and 10 ppm sulfate.
c) Method described in Std. Methods was quite time-
consuming, insensitive, and didn't give reproducible
results presumably because light absorption rather than
reflected light was being measured.
d) See Std. Methods for preparation of reagents and
standards.
D8: ANALYTICAL METHOD FOR NITRATE BY UV ABSORPTION
Reference: Standard Methods, 13th Ed., p. 237
1) To 50 ml sample add 1.00 ml of 1.0 N. HC1.
2) Using square, 1 cm, silica cuvets measure absorbance at
220 nm. Do this for all standards and samples before changing
wavelength setting on UV-VIS spectrophotometer.
3) Measure absorbance at 275 nm.
4) Calculate ppm nitrate concentration from calibration
curve plotted using corrected absorbance:
AbBcorr = Ab8220 " 2 (Ab275>
Typically, ppm =4.00 Abs
corr
NOTES: a) Linear range is 1-4 ppm nitrate.
b) Standards of 1 and 3 ppm usually run.
c) Set zero on spectrophotometer using distilled water
blank with acid added.
d) Use same cuvet for all measurements in single beam
spectrophotometer.
e) In every experiment the effects of possible interfer-
ences was checked. Where organic extractables from the
249
-------
resins were high,nitrate was also high and dilution elim-
inated serious interference.
D9: POTENTIOMETRIC TITRATION METHOD FOR CHLORIDE
Reference: None; method developed here and possible interfer-
ences checked out.
1) Make sample to be titrated up to approximately 50 ml
in a 150 ml beaker with teflon-coated magnetic stirring bar.
2) Titrate with 0.0141 N. AgN03 (.500 meq Cl/ml) using
Fisher Automatic Titrimeter to + 290 mV end point. This was
previously determined to coincide with the inflection point in
the_ ml titrant added vs mV plot. Potential due to increase in
Ag ion was measured using double junction (nitrate-external)
calomel, reference electrode (Orion 90-02-00) with Ag/AgS solid
state specific ion electrode (Orion 94-16A) .
NOTES: a) Sensitivity is 125 mV/ml titrant added at inflection
point for .0141 N. AgNO_.
b) AgNO- standardized against 1000 ppm NaCl.
c): For concentrated Chloride solutions use 0.100 N.
AgNO..
d) It is optional but not necessary to acidify with 5
ml of 1:1 HNO_ if pH is below 7.
e) This method was used for all capacity determinations
and isotherm determinations. Chloride in column effluents
was determined by combination chloride electrode (Orion
96-17) .
DlO : CARBONIC ACID AND BICARBONATE DETERMINATIONS
Reference: For Instrument operations: Beckman Model 915 Total
Organic Carbon Analyzer Instruction Manual
Objective: To determine H^CO^ (dissolved but volatile C02) in
"
acidic ion-excnange column effluents and HCO" or CO-
in neutral to basic effluents.
1) Withdraw 100 microliter sample of column effluent dir-
ectly from flowing effluent stream by submerging syringe tip 4
cm below surface in overflowing 100 ml beaker in which effluent
tube is submerged. Rinse syringe at least twice by discharging
contents anywhere except back into beaker. Draw sample up
slowly to avoid CO2 bubbles.
2) Inject into inorganic channel of TOC analyzer previously
standardized with Na^CO- or NaHCO- (freshly prepared and kept
stoppered) . ^ J J
3) Make at least duplicate injections or repeat until re-
producible peaks are obtained.
4) From peak height, determine mm/£ C02 evolved.
NOTES: a) Peak height vs concentration curve is approximately
250
-------
5.1cm
j—11mm I.D. Std Wall
Tl Pyrex Tubing
15cm
6cm
JL
Mini-Column for
Capacity Determinations
V
-V=14ml.
•Coarse Pyrex Frit
- 3/8" O.D. Tubing
-*—Pinch Clamp
-Eyedropper Tip
Std. Wai I
-Pyrex Glass Pipe
Solution Volume
2.1.0 min.
Resin Conditioning Column
Resin
11 max.
Figure PI
Gloss Ion-Exchange
Columns
/- Pyrex Wool
5cm
#11 Stopper Taped to Column
Pinch Clamp
^-5/l6"O.D. Tubing
Plexiglas Support
251
-------
Separatory Funnel
i
Figure °2
Resin Conversion
Apparatus
40
i
cm
8.5cm
6cm
I
[]
\J
L.— -i—
60
50
30
20
10
V
11
fl
-1 \J
S
—
^V
'
#3 Rubber Stopper
21 mm I. D., Standard Wai I
Pyrex Tubing
Ion-Exchange Column
V=140ml
Conversion Solution
75ml Resin (Typical)
Coarse Pyrex Glass Frit
10mm O.D.
7/16 "O.Dx 1/16" Wai I, Gum
Tubing
'Pinch Clamp
10 mm 0. D. Tip
252
-------
BICARBONATE SELECTIVITY
APPARTUS
ANION MINI-COLUMNS
Figure D3
•ISOTHERM TUMBLER
WITH
125 ML BOTTLES
CATION COLUMN
253
-------
to
U1
—
FIGURE D5
COLUMN RUN FLOW SYSTEM
-------
1/4-20x1
SS.Socket Hd Cap Screws
SS Washers
1/4"
Threaded
1/4-20
ev
H
vs-
•••
i
,*--
7
•
J
1
•^
I
...
^^^^»
<=>
r— i
--
0
-
!
• ^
J— ^ 1/
^U_ ..i>
-^
LJ
J-
r— —
^.
—!**•
*l
}
\t
\t
-4
.0
c
p
-1"
S
^ r-t
'4" IV
'2"x1
=
IPT
-3/4" Dia. Plexiglas Boss
—6 Screws, 30° B.C., 3" Dia.
— Buna N "0"Ring
1/8"Tk.x2"Dia.
S^~\
8" Dia Hole
Flow Distribution Ports
62" Dia.
low Distribution Assy.
Dia. 60x60 Mesh
S Wire Cloth
-Plexiglas Column
8-32x5/16 l"l.D.x1/4" Wai 1x5'Long
SS. Socket Hd.
Cap Screw
l"Dio. Plexiglas Ion-Exchange Column Details
FIGURE D6
1" Dia. Plexiglas Ion Exchange Column Details
3/32" Teflon Cord
1-1/2" Dia. Overlapping Circle
52x52 Mesh
Saran Retaining Screen
1-3/4" Dia.
(Typ)
1/4" Male Connector
PPYE
3 Connectors
255
-------
1/4-20x1"
SS. Socket Hd Cap Screws
SS. Washers-
1/4"
Threaded
1/4-20 —
7
7
I
Flow Distribution Assy.
-1/4" Tubing Tee, PPYE
:1/4"MPT
-1/2"x1-3/4"Dia Plexiglas Boss
-6 Screws, 30°B.C.,4"Dia.
8-32x5/1(3
SS. Socket Hd
Cap Screw
•Buna N"0"Ring
1/8" Tk.x 3-1/2 Dia.
1/8" Dia. Hole
-4 Flow Distribution Ports
.062" Dia.
2-1/2"Dia.60x60 Mesh
SS Wire Cloth
-«Plexiglas Column
2-1/2"l.D.x1/4"Wallx5'Long
2-l/2"Dia. Plexiqlas Ion-Exchange
Column Details
FIGURE D7
2 1/2" Dia. Plexiglas Ion-Exchange Dolumn Details
3/32"Teflon Cord
2-3/4" Dia. Overlapping
Circle
ig
\
5
— \~_
,
— ->. ^
-S
—
-------
linear in 0-2 mm/A range.
b) Reproducibility at 1 mm/£ was + 2%.
c) Typical standards were 0.5, 1.0, 1.5 and 2.0 mm/ 5,.
mm/1 - millimoles/1
APPENDIX E
CALCULATIONS AND DERIVATIONS
Find $ = gms Resin. Given XA/ ^ AQf gQ
El: EXAMPLE CALCULATION FOR BINARY ION EXCHANGE ISOTHERM
Objective: To determine the wt. of Resin ^W, gins) containing
only counterions "B" at cone. Bgrtie^/gm which should
be added to a given Volume (V, liters) of solution
containing only counterions "A" at a cone. A, meq/&
to produce the desired equilibrium equivalent frac-
tion of A, i.e., XA-
Also Solve for XA = f(ot, 3, AQ , BQ)
Given the following contraints on the Equilibration Process.
(a) Constant total concentration in the liquid phase; i.e.
AO = Total initial Cone, of A=CT total liquid phase
cone. — final and initial
AO+BQ = CT; AQ = CT when BQ = 0
A+B = CT = AQ
Dividing by: CT or AQ
(b) Constant total concentration in the solid phase; i.e.
BO = total initial cone, of B on resin = Q, the
constant capacity of the resin
A+B = Q
AO+BO - Q; BO = Q when AQ = 0
Dividing by: Q or BQ
(c) Constant separation factor: aB/ an approximation
257
-------
(d) Mass balance on A, i.e. liquid phase loss of A = Solid
phase gain of A.
V (AQ -A) = WA (E4)
where: A, B = Cone, in liquid phase (meq/&)
A, B = Cone, in solid phase (meq/gm)
AO,BO/AQ,BO = Initial concentrations (meq/£; meq/gm)
CmfC- = Total concentrations (meq/£, meq/gm)
XA,XB = Equivalent Fractions, Liquid Phase
y^'Y-Q = Equivalent Fractions, solid phase
V = Volume of liquid phase (&)
W = Weight of Resin (gms)
3 = W/V (gms of resin/£)
Q = Resin capacity (meq/gm)
a = GL = Separation Factor (Dimensionless)
o
Solution:
from (2) and (4), where CT = AQ, CT = BQ
and • (An-A) = Wy A
A0B0 U A0B0
V _ W
"S = ^ YA
(1-*A) (E5)
from (2) XB = 1-x,
from (2) and (5)
a
A
YB = 1 - -IT- (1-x,) (E6)
3BQ A
from (3) , (5) , (2) and (6)
- A0
A (1 - XA>
A0
x ri -
XA 11
258
-------
" XA>
(E7)
X
In std. Quadratic form:
*0
A0B0
0
x. - ax,. _ _
A A BQA0
(1 -
XA)
= 1 - 2X
(E8)
- axA + ax - 1 + 2xA - x = 0
0
(a-l)x2 + (a3B0 - a + 2)x. -1=0
A T "
(E9)
Solving [8] for 3, the Desired ratio of gms. resin/£ solution
2 2
- 2X
- (1-cQx: + (a-2)x? + 1
x
A
(E10)
Example problem: Find the amounts of resin in NO- form to add
to a solution of 0.005 N. H2SO. that an equi-
valent isotherm might be developed with 5
Data points.
(a) Given: Resin originally in NO., Form
AQ = CT = 0.005 N. = 5.00 meq/A
B Q = Q = CT = 3.39 meq/gm HN03
A S04
= 2*5° — ESTIMATE1
Solution: Generate a table of 3 values xa values using equation
(10)
259
-------
XA~XSO.
0.
0,
0,
0,
0,
1
3
5
7
9
Calculated 3
6.1062
1.9961
1.0324
0.5183
0.1540
gms of resin in the N03 form (pre-
viously equilibrated with .005 N.
HNO.) to be added to 1 £ of 0.005
N. H2SO. to achieve equilibrium
at approximately the XA (i.e.
x,,_ ) values shown.
J4
(b) Given: Resin originally in the SO. form
AQ = CT = 0.005 N. HN03 = 5 meq/£
B~0 = CT = Q = 3.39 meq H2S04/gm
= 0.400 SAME ESTIMATE
Solution: Generate table as in part (a) using eq.
V = V
A NO.
0.1
0.3
0.5
0.7
0.9
£
AS IN PART (a)
(10)
Calculated 3
31.1947
7.0551
2.58
0.9166
0.1885
gms of resin in the SO^ form (pre-
viously equilibrated with 0.005
N. H2SO.) to be added to 1 A of
0.005 N? HNO- to achieve equili-
brium at approximately the x
(i.e. XNQ ) values shown.
E3: DERIVATION AND JUSTIFICATION OF a.
Ratio of Isotherm Areas Related to Separation Factor
R = f (a)
Objectives:
To mathematically relate the ratio (R ) of the
area below the isotherm to that aboveathe isotherm
to the separation factor (a). That relationship
will be used to determine the best fit, averaged
a given the measured areas below and above the
experimentally determined isotherm. To justify the
use of the separation factor a as opposed to the
selectivity coefficient as a measure of the resin
phase preference of one ion over another.
260
-------
1.0
AREA I
(above isotherm),
i-y±
dx
i AREA II
(below isotherm)
FIGURE El
Example Isotherm
Definition of separation factor: a^"
3 xi xj (1 " *v
-------
Relating the ratio of areas to the separation factor
where:
Area II
Area I
r
Jn
3x
dx
r1 (3 +
"J0 ~
1 - Z
3x
dx
r1
=1 TT^-
Dx
dx
Integrating the above expression for Z:
z = 3 + 1
3
3x)
Z = (3 + 1) y - —
Or, in terms of the separation factor, a:
Z =
a
a
(a - 1)
(a - 1)
In a
Then, solving for R in terms of a
cL
Ra =
- Z
(or - a - alna)
(a - I)2
1 - (a2 - a - alna)
(a - I)2
The above expression can't be solved explicitly for a, so, after
R is experimentally determined, the best fit a is obtained by
trial and error or from a plot of R vs a.
Justification for Use of a and Not
t\ —
The separation factor ag indicates directly the preference
of a given phase, in this case the resin, for the superscript
ion in question. It is the ratio gf the distribution of ion A
to that of ion B.
262
-------
A _ ratio of fractions of ion A between solid and liquid
B ~ ratio of fractions of ion B between solid and liquid
Although the experimentally determined separation factors
for divalent/monovalent (SO./NO-) exchange were not constant,
the ratio of areas technique has provided a means by which a
best fit factor can be determined. This then represents the
preference the resin has for one ion over another over the entire
range of equivalent fractions at some constant total concentra-
tion.
The selectivity coefficient KB at constant total concentra-
tion GO is the ratio of the squarea distribution of the monova-
lent species to the distribution of the divalent species; as
such it is influenced by the units of Q, resin capacity, and CQ,
the total liquid phase concentration. Consider the following
example of univalent-divalent exchange:
2B
2RB + A"
B
B
B
'B
x
B
Assuming the resin has no real preference for either ion, then:
yB = XB
rA
,B
Typically Q = 1 eq/l resin and CQ = 0.005 eq/£ solution. Then:
KB = 0.005
If the units of CQ are given as meq/Jl then:
B
Either of the above choices of units for Cn yields a selec-
tivity coefficient which infers a large preference by the resin
phase, first for ion A then for B neither inference being
correct, as the resin has equal affinity for each ion. The
separation factor being independent of CQ and Q correctly infers
no preference with
-------
E4: COLUMN EFFICIENCY
EXAMPLE CALCULATION
7,
N
Cl
= 3.54
Run No:
Resin: lonac AFP-1QQ, STY-DVB (I) MR
Resin Volume: .310 1/BV; Flow Rate: 2Q BV/hr.
Titration Capacity: meq/ml @ pH 2.3: HC1 =1.07
BV @ 0.48 meq/1
NO
-j-Breakthrough = 295
150Q mec[*BV x .31ITT.
~
15.51
HCO.
HCO
HCO
Cl
Cl
Cl
NO.
N0
SO
, Influent Cone,
> Influent Area;
j Effluent Area;
Influent Cone:
Influent Area:
Effluent Area:
Influent Cone:
Influent Area:
Effluent Area:
Influent Cone:
Influent Area:
BV A 29.98in
1.0 meq/1
2 2
5.906 in * 15.51 meq/in = 91.594 meq
5.74 in2 * 15.51 meq/in2 = 89.027 meq
1.5 meq/1
H-CO-, on Resin
QC1V
= 2.57 meq
= +331.70
2 ^J" 2
8.858 in * 15.51 meq/in = 137.39 meq
27.34 in2
1.5 meq/1
8.858 J"2
* 15 -51 meq/in = 424 .04 meq
HC1 on Resin = 45.05 meq
2
* 15 »51 meq/in = 137.39 meq
* 15.51 meq/in2 = 3.26 meq
HNO3 on Resin = 134.13 meq
1.5 meq/1
8.858 in2 * 15.51 meq/in2 = 137.39 meq
on Resin = 137.39 meq
.21 in
i
1
2
3
4
An ion
HC03
Cl
NO"
S0~
Total
meq
on
Resin
2.57
45,05
134.13
137.39
319.14
Eff.
yi
.01
.14
.42
.43
1.0
meq/1
in
Solution
1.0
1.5
1.5
1.5
5.5
xi
.181
.273
.273
.273
1.0
Relative Eff.
yi/xi
.055
.513
1.54
1.58
264
-------
Run No:
Resin:
E5; COLUMN EFFICIENCY
EXAMPLE CALCULATION
11, o£ + dj^ = 3.97
Amberlite IR-45, STY-DVB, Polyamine, Microporous Resin
Resin Volume: .310 1/BV; Flow Rate:
Titration Capacity: meq/1 @ pH 2.3:
BV @ 0.48 meq/1 NO3-Breakthrough
1500
20_ BV/hr.
HC1 - 1.70, HNO.
1.70,
= 2.13
480
= 15.619
HCO
Influent Cone.
HC03 Influent Area:
HCO"
1.0 meq/1
9.543 in2
Effluent Area: 8.69 in
Cl
Cl
Influent Cone:
Influent Area:
Effluent Area:
NO3 Influent Cone:
NO~ Influent Area:
NO~ Effluent Area:
SO5? Influent Cone:
4
SO~ Influent Area:
1 .5 meq/1
14.315 in2
11.62 in2
14.315 in
* 15*619 meq/in^ =
* 15.619 meq/in2 =
on Resin =
* 15.619 meq/in =
* 15.619 meq/in2 =
HC1 on Resin =
* 15.619 meq/in =
* 15.619 meq/in2 =
HN03 on Resin =
* 15-619 meq/in'
on Resin
149.057 meq
135.729 meq
13.33 meq
223.585 meq
181.493 meq
42.09 meq
223.585 meq
3.124 meq
220.46 meq
223.585 meq
223.59 meq
i
1
2
3
4
An ion
HCO3
Cl
N0~
SOT
— , — 4
Total
meq
on
Resin
13.33
42.09
220.46
223.59
499.47
yi
.03
.08
.44
.45
1.0
meq/1
in
Solution
1.0
1.5
1.5
l.S"
5.5
xi
.181
.273
.273
.273
1.0
Relative Eff.
yi/xi
.166
.293
1.612
1.648
265
-------
E6; EXAMPLE CALCULATION OF PREDICTED y
_ Meg NO3
Objective: To predict YN = meq total on resin at nitrate
breakthrough
Assumptions:
(a)
(b)
(c)
= 1.0, I y± = 1.0
i
Constant separation factors, a.'s.
Three plateau zones
(d) Two abrupt transition zones
(e) Instantaneous equilibrium
(f)
Q » C : Q= 1.0 meq/mlf CQ = .005 meq/ml
N
(g) Bicarbonate separation factor « 1.0: a^ « 1.0
SO
ZONE 1
(Sulfate)
y NO,
i *
yCM c.
yN2 No3
ZONE 2
(Nitrate)
yci,2 Cl
ZONE 3
(Chloride)
U.u ••* mi or lAcsin - ~
ZONE 4
(HC03)
.0
-------
Relevant Equations;
(a) c = X*y1
=
I a x.
S,n N,n Cl,n
= 5 meq for zone 1
(d) S + S + S,,, = meq on resin in zone n
S/n N,n ex / n
Initial Conditions:
XN,! = *25 These conditions are equivalent to those
XC1,1 - .375 existing during Run 1 with the
x = .375 assumption that HCO3 is a non-com-
S / 1 "^
N _ ponent, an assumption justified on
the basis that at equilibrium there
Cl _ _
aN = .26 is ^insignificant HCO,~ (or H0CO.,)
w JL J
S 00-5 ^-n any zone of interest.
°LJ = 2.83
Resin capacity = 1 meq/ml.
Calculations:
(a) Find the equivalent fractions of NO.,, SO. and Cl on
the resin in zone 1.
XVT
yN,l
v
xs + < "N + ° cl
267
-------
.25
(2.83) (.375) + (1) (.25) + (.26)(.375)
Given y , = .18 and assuming constant separation
factors, find Y , and Y_,, ,.
o f -L L. J. / -L
Yy V
SO ^^VT G C
.. _ O JN .1 ,T — „." °
AN
XS
(2.83) (.375) (.18)
1/1 (.25)
, , = .75
Cl YC1 ^ thus y^_ Cl XC1 YN
08
C1
(.26) (.375) (.18)
(.25)
YC1,1 = .07
(b) Given 5 meg total in liquid entering zone 1 and
•»
equivalent fractions in liquid in zone 1, find
meq of each component in the liquid.
LC1 1 = XC1 1 (5mec^ = 1-875 meq
1 = ^ 1 tSmeq) = 1.25 meq
Lo i = xo T (5meq) = 1.875 meq
la f JL o r J-
268
-------
(c) Assuming all SO4 removed in zone 1, find ml resin in
zone 1.
(L_ ,) =- (resin capacity) = ml resin zone 1
s.i ys,i
11 o-7c or, \ 1.0 meq total 1 ml resin = 2.5 ml resin
(1.875 meq SO4) ,64 me<* SQ4 1 meq total
The meq of N03 and Cl removed in zone 1 can be
found:
^-, , = ¥„.,, (2.5 meq) = (.07) (2.5 meq) =.175 meq Cl
Cl i J- <--!• / -L
N 1 = YN 1 (2'5 meq) = (*18) (2'5 meq) = *45 meq N°3
(d) From above, find the meq of NO- and Cl remaining in
the liquid and hence entering zone 2.
LC1 2 = LC1,1 ~ SC1/1 = 1>875 ~ *175 =1-7 meq Cl
LN,2 = L * S = 1'25 -
The equivalent fractions of Cl and NO3 in the liquid
entering zone 2 are:
XC1,2
1.7
XC1,2 =
.8 + 1.7
XC1,2
269
-------
LN,2 + LC1,2
N,2 .8+1.7
XN,2 = -32
(e) Using constant separation factors and the equivalent
fractions in the liquid entering zone 2, find the
equivalent fractions on the resin in zone 2.
Cl x_,
< cl
YN,2 (1) (.32) + (.26) (.68)
YN,2
From constant separation factor y_, 0 is calculated,
L*J_ / £,
Cl . YC1 XN thus ^Cl_ Cl *C1 YN
"
N xci
= (.26) (.68) (.64)
C1,2 (.32)
(f) Assuming all remaining NO- is removed in zone 2,
find ml resin in zone 2.
270
-------
(L -) rr (resin capacity) = ml resin zone 2
N'J *N,2
, o m^ Mn \ 1 meg total 1 ml resin = 1.25 ml resin
(.8meqN03) .64 meq NO3 1 meq total
1.25 ml resin is equivalent to the removal of 1.25
meq total.
The meq of Cl removed in zone 2 can now be found.
SC1 2 = YC1 2 (1'25 mec3) = <-36) d.25) = .45
(g) The liquid entering zone 3 contains the following
meq of Cl.
LC1,3 = LC1,2 ' SC1,2 = 1'7 ~ '45 ' 1'25 me£3
Since all the chloride is removed in zone 3 and the
resin capacity is 1 meq/ml resin, zone 3 must
contain 1.25 ml of resin.
(h) Find y~N which represents the average of the equiv-
alent fractions of NO, on the resin in zones 1 and
2. This is found by weighting the value of yN with
respect to the amount of resin in the zone in which
it occurs.
_ _ (yN,l) (ml resin zone 1) + (yN,2)-(ml resin zone 2)
y ~ ml resin zone 1 + ml resin zone 2
V - M8) (2.5) + (.64) (1.25)
YN 2.5+1.25
yN = .333 Predicted
yN = .340 Experimentally Observed (Run 1)
271
-------
to
'J
to
TABLE Fl
DATASET FOR STATISTICAL ANALYSIS
2 05
M W
co g
H S
p* E3
2
1
2
3
4
5
6
7
8
9
to
11
12
13
J4
15
•| 7
CQ 2
a
(VI)
3 . /?••
23.40
12.70
137.00
2,67
108,00
82,90
2,83
109.00
94.00
54,00
3,07
108.00
3.26
1.89
3. 09
1 - ~VJ
18 2.40
19
20
21
2?
23
24
25
26
27
2P
• ••>
1.53
1.15
1.67
2.54
1.43
1.22
2.59
1.81
1 . .M
1.51
1.31
1 .5.1
.1,16
1,10
1 . ~V5
1 . •',.'
A. 50
1 .66
i . 1 "*
t . 48
t >31-
-0.
0,
1 , 39
1 , J 1
i . ^3
i -'I 1
O
t 07
id
W
cu
CV4)
7 . 7-')
11.10
7.90
7.90
7.60
7.70
8.70
7.80
6.80
9.90
10.60
8.50
9,00
13.00
1.3-00
1 3 , 00
13.00
13- ' 0
13. ^«(.
13.00
1 3 . 00
(3 00
I 'i . 00
1 '" 00
13.00
1 3 . 00
j. 3 • 00
1 i "r
1 '• •><"'
i 3 )<•>
1 3 . 0',
13-00
EH
H
§
H
£
CV5)
3 00
Jf.OO
'.i.OO
2 . 00
3.00
2.00
2.00
3.00
2 . 00
2.00
2,00
3.00
2.00
4.00
4.00
4 00
4 - ,»••
<4 .)O
4 •"* t '
4.00
4.. r*(>
4 . '•HI
-1 o .
4 0 ( '
4 • 00
.^ ,.,.
» ,•10
T v"
il ' *
n , •>•
1 . C:-r
4 Of'
X
H
^~1
. §
eve)
1 .00
2.00
1,00
4,00
1. .00
3,00
4,00
1.00
3.00
2.00
5.00
1,00
4.00
1.00
J. .00
! .00
1 00
EH
M
CO
o
Xg
MH
&
fV75
2.00
1.00
1.00
1,00
:> . oo
2.00
1.00
2 . 00
2.00
2.00
1.00
2.00
1.00
2.00
1 .00
3 . 00
J , ',.H;
1 00 1 00
J. .<•••
1 ^I'l
I ,00
I . -JO
i -00
1 oO
; oo
:| <>••
1 ^00
1 »<>0
i /;<:
1 . M.I
I t < j i
t ..>f>
i . s.'0
1.00
1 00
< . 00
1 .00
3 u'..
-o.
•0
i 00
3 , 00
• .00
*,00
-0,
2-00
EH
w o
0
§K
0
H £3
CVBJ
2.00
1.00
2,00
1,00
2,00
1.00
1 .00
2,00
1,00
1,00
1 .00
2.00
1.00
2 . 00
2.00
<> . 00
2 . 00
2.00
'*' . 00
2.00
2,00
2.00
7,00
2 , 0''1
-0.
-0
7.00
2 00
2. 00
"> , 00
•0 <
2.00
8
c
iH
(V9)
1.32
3.15
;1.54
4.92
.98
4.68
4.42
1,04
4.69
4.54
3.99
1.12
4.68
1.18
.64
1.13
.54
,88
.1.09
1.11
,64
1.21
1 . 1 B
,95
-0.
-0.
.63
! .20
1,11
.95
-O.
.5?
i-H
2 U
C
1-1
fyiQ7
1.58
.64
1.36
.6"
1.49
1.21
.53
1 .35
.97
1.35
.81
1,42
-0.
1.05
-0.
1.13
1.23
-0.
-0-
-0.
1 . 06
0.
--0,
-0.
•-0 •
0.
-n
1 . 20
\ . 29
-0.
0.
1 , 0°
PU
D
an H
M
= CO
«
(yin
2.19
2.19
2.00
2.00
2.19
2.00
2.00
2.19
?,00
2.00
2.00
2.19
2.00
2.36
2.36
2.36
2.36
1.36
2.36
2.36
2.36
2.36
2.36
2.36
2.36
2,36
2.36
2.36
2. >>6
2 . 36
2 36
2 - 56
22
H O
CJ H
EH CO
M O
2 Pi
CV121
0.
1.00
0,
1 - 00
0.
1.00
1 . 00
0,
1 . 00
1.00
1.00
0.
1.00
o.
0.
0.
0.
0.
o.
0.
0.
0.
0.
o.
-0.
-0-
0 .
0.
0.
o.
-0 .
0
z
HI WH
EH taw
W W t-3
2 QX
CV141
2.00
1.00
1.00
1.00
2 . 00
2.00
1.00
2.00
2.00
2.00
1.00
2.00
1,00
2.00
1.00
.50
2.00
1.00
.50
1.00
1.00
.50
1.00
.50
-0.
0
1 .00
.50
1.00
-50
- o ,.
? , 00
w
t-3
Jj
M >
cn f-d
HHH
ra
n &
> O
IT1 M
K>
X
/d
H ^
HH
cn
-------
N-- 12
JO
TABLE F2
CORRELATION MATRIX FOR WEAK-BASE RESINS
icTF.'Jis CASES-CASE*:i 13
Rl? .0100-' .7079
3.CAF-AI 1TY 1 .
-------
TABLE F4
>J
42k
VARIABLE
1.S04/NQ3
2.N03/CL
3,CAPACITY
4.PKA
9,LOGeS/N
10.LOGN/CL
S f
ll.RSIZE
12.N2PGSITN
14.XLINKING
CORRELATION MATRIX FOR
FICIENTS
@ .0500=
1*0000
-,4652
,5311
-,5094
,9413
-.4813
-.7901
.8824
.0369
1.
S04/N03
.4555 RB
1.0000
-.3303
-.2051
-.5002
.9884
.1952
-.6202
.4576
2.
N03/CL
.0100= .5751
1,0000
-.1940 1.0000
,633? -.5298
-.3596 -.1231
-.6090 ,7973
.5964 .3876
-.1602 -.3663
3. 4,
CAPACITY PKA
ALL ANION RESINS
1,0000
-.5319 1.0000
-.8825 ,2540
.9485 -.6522
-.0689 .4520
9. 10,
LOG S/N LQQ.ty
1.0000
-.7683
-.0973
11,
N/CL RSIZE
1 ,0000 .
-.0601 1.0000
12. 14,
N2POSITN XLINKING
-------
TABLE F5
CORRELATION MATRIX FOR POLYSTYRENE RESINS
;:-orw;:i A- ION coFR-icrENT;:, STRAT "NITROGEN:;?
H- tZ df:= 10 R@ .0500-- .'-;
.0100= .7079
3.
4,
9.
1. •>
I 1
14-
14.
VAP [ABU-
CAP ACT! V
PKf,
i, L)li. S/N
i utfc N/(.:L
-;:i7E
vUPOSUN
XL INKING
1,0000
-.10158 1.0000
,4927 -.4850
- . 1937 - -864?
-.3812 8991
-o, -o.
-,,3611 -.41?/
3. 4.
CAPACITY PKA
1,0000
.3587 1.0000
-,7583 -.7054 1.0000
-0. -0. -0, 0.
-.3008 .3338 -.1915 -0 , I. 0000
<> 10. 11. 12. 1.4.
i. n&ftS/N LOG^N/CI.. RSI7F NPPOSIIN XL.INKING
TABLE F6
N- 7
CORRELATION MATRIX FOR NON-POLYSTYRENE RESINS
,,Ot:FriC:EENTS STRATA-NITROGEN :i
-- S RG» . OSOO-. .V545 R0 .0100- .8741:1
VARIABI. E"
3.rAPAC.; , Y
4.PKA
*. L.OCL. S/N
IN
1.4, XL INK ING
I. . 0000
.2798 1.0000
.1345 .8190 1.0000
-lf58A -.1757 3 "86 1,0000
,3324
.0017
.5861
.4780
.1. 4,
CAPACITY PKA
0/4° -.3595 1,0000
0,
.4628
9.
. 8 V 4 3
10.
..X5"J6 0,
11.
( CIOeS/N LOGeN/CI. RSI7R.
12.
1 .0000
J.4,
275
-------
SELECTION OF REGRESSION CASES = CASE*: ;L»13 )
LOG OF SULFATE/NITRATE SELECTIVITY RELATED TO ALL THE INDEPENDENT
VARIABLES OF INTEREST FOR WEAK-BASE ANION RESINS — OPTIMISATION
PlNfiT-TSIS AT STEP 1 FOR 9. LOG S N N- 13 OUT OF 13
SOURCE DF SUM OF SQRS MEAN SQUARE- F-STAT S1GNIF
REGRESSION
ERROR
TOTAL
MULTIPLE R- ,93486
VARIABLE
CONSTANT
12.N2POSITN
REMAINING
3. CAPACITY
ll.RSIZE
14.V14
ANALYSIS AT STEP 2
SOURCE
REGRESSION
ERROR-
TOTAL
1 27*385 27.385 76.282
11 3,9489 .35899
12 31.334
R-SQR= ,87397 SET= .59916
0000
PARTIAL COEFFICIENT STD ERROR T--STAT SIGN1F
1.4015 26795 542303
.93486 2.9833 .34157 8.7340
PARTIAL. SIGNIF
,25354 ,4265
-.92235 ,0000
-.11621 .7191
FOR 9. LOG S N N= 13 OUT OF 13
0003
oooo
DF SUM OF SQRS MEAN SQUARE F-SfAf SIGNIF
2 30.744 15.372 260,80
10 .58943 .58943 -1
12 31,334
0000
MULTIPLE R= ,99055 R-SQR= .98119 SE= .24278
VARIABLE
CONSTANT
ll.RSIZE
12.N2POSITN
PARTIAL
-.92235
,95976
COEFFICIENT
17.442
-7.4538
2.0274
STD ERROR
2.1275
-98731
,18759
T-STAT
8.1985
• 7.5496
10.807
SIGNIF
,0000
.0000
.0000
REMAINING
PARTIAL
SIGNIF
3. CAPACITY
14.V14
REGRESSION OF
STEP R-SQR
1 ,87397
2 .98119
-.39349
.18226
9. LOG S N USING
.2312
,591?
FORWARD
STD ERROR * VAR
.59916
.24278
1 ,12
2 11
SELECTION
VARIABLE.
.N2PQ8ITN
.RSIZE
PARTIAL
IN ,93406
IN - ,92?35
SIGNIF.
.0000
.0000
276
-------
SELECTION OF RFGRFSSTON CASES--CA_SE.:fj_l-
13 1
LOG OF NITRATE/CHLORIDE SELECTIVITY RELATED TO ALL THE INDEPENDENT
VARIABLES OF INTER
ANALYSES AT STEP F
SOURCE
REGRESSION
ERROR
TOTAL
MULTIPLE R= .77540
VARIABLE;
CONSTANT
12.N2F'OSrrN
REMAINING
3. CAPACITY
ll.RSIZE
14.V14
ANALYSIS AT STEP 2
SOURCE
REGRESSION
ERROR
TOTAL
EST FOR WEAK-BASE ANION RESINS—OPTIMIZATION
F OR 10. LOG N CL N=
DF SUM OF SQRS
1 .89825
10 ,59573
11 1.4940
R-SQR= .60124 SE=
PARTIAL COEFFICIENT
1,4404
-.77540 -.55495
PARTIAL SIGNIF
,05000 ,883V
-,16796 .6216
,78221 ,0044
FOR 10, LOG N CL N=
DF SUM OF SQRS
2 1.2627
9 ,23123
11 1.4940
12 OUT OF .1.3
MEAN SQUARE F-STAT B.TGNIF^
.89825 15,07(3 ,0030
,59573 -1
,24408
STD ERROR T-STAT SIGNIF
,10915 13,196 ,0000
,14292 -3.8830 .0030
12 OUT Of- 13
MEAN SQUARE F-STAT SIGNIF
,63137 24,574 .0002
.25692 -1
MULTIPLE R= ,91936 R-SQR- .8452*2 SE- .16029
VARIABLE
CONSTANT
12.N2POSITN
14.V14
PARTIAL
-.80640
.78221
COEFFICIENT
,75501
-.41353
.38073
STD ERROR
,19557
,10109
,10109
T-STAT
3.8606
-4.0908
3.7666
SIGNIF
,0038
.0027
.0044
REMAINING
PARTIAL SIGNIF
3, CAPACITY
U.RSIZE
REGRESSION OF
STEP R-SQR,
1 ,60124
2 .84522
.32791
-.50651
,3550
,1352
10, LOG N CL USING FORWARD SELECTION
STD ERROR *
,24408
,16029
VAR VARIABLE
1 12.N2POSITN IN
2 14.V14 IN
PARTIAL SIGNIF
-.77540 .0030
.•W21 .0044
277
-------
N= ,_>9 OUT Oh ,32. 9., OGeS/N VS. .3 . CAP AC f TY
1,4132
,53649
,98000
1 •
1.6240 2.2680 CftPAC.1 FY
i .9460 2.S900
FIGURE Fl
SCATTER PLOT WITH LINEAP REGRESSION EQUATION
In o(^ vs MEASURED HC.l CAPACITY
FOP
WEAK AND STRONG BASE RESINS
278
-------
SCATTER PLOT CASE ~,= CASE* t 1-1?
N== 1? f)U T OF i 3 9. i OGC S/rv V'j
L.OG,»S/N
4.9200
4.1324
3.3448
1.7697
,98208
l .A/40
i.2680
-f-
CftPACITT
2.5900
FIGURE F2
SCATTER PLOT WITH LINEAR REGRESSION IQUATION
lr. o(g vs. MEASURED HCI CAPACTT\
FOP
WEAK 3ASf, P£SINS ONIY
279
-------
SCATTER PLOT CASES=CASE»: 1.4-32
N- 16 OUT OF 19 9.LOGeS/N US- 3,CAPACITY
L.QG«>S/N
1 . 2t49 *
* * *
,94354
.80786
1.0200 i.2760
1.1480
1.5320
t.4040
FTGURF F3
SCATTER PLOT WITH LINEAR REGRESSION EQUATION
In o(S vs. MEASLRED HC1 CAPACITY
N
FOR
STRONG BASE RESINS ONLY
280
-------
N- 29 OUT OF r? 9.LQG S/ VS. 4. PIS A
1.2899
I.4132
.53649
6.8000
8.0400
9.2SOO
11.760
10.520
KK.A
1* 000
FIOURF F4
SCATTER PLOT WITH LINEAR REGRESSION EQUATION
In «jjj vs. PKa
FOR
WEAK AND STRONG BASE RESINS
281
-------
•>CATTER PLOT ( ASES=CASE# : 1 - 13
N L3 OUT OF 13 9.LOG.b/N MS. 4.PKA
LOG, 3/N e '
4.920C
4. 1324
J.344S
2,5572
1-7697
*
*
* *
-• 1 1 1 1 1 1
1 I 1 1
6.8000 8.5200 10,240 PKfl
7.6600 V.3800 1:1.100
FIGURF F5
SCATTER PLOT WITH LINEAR REGRESSION EQUATION
In e(jj VS . pKa
FOR
WEAK BASE RESINS ONLY
282
-------
PLO'f bTRAr=NITROi3rN; j
N- P OUT OF 8 9 . LC)iJ6
9 200
^.5665
4.2131
< O "=-. v x.
^j * ' vJ ~ O
3 .11 " '
'c *•>
V *
V~» N'^
*> ff\
-------
SCATTER PLOT STRAT=MATf.'TX :
°F 24
1.2608
I.1541
! .0473
VS. 4.PNA
>.60"0
8.6800
9,7600
10.840
13.000
FIGURE F7
SCATTER PLOT WITH LINEAR REGRESSION EQUATION
In &--, vs . pKa
FOR
POLYSTYRENE RESINS
284
-------
SCATTER PLOT
l.Oa.N/Ci
1 .5810
N^ 19 OUT OF 32 lO.LOGeN/£l OS. ll-KSIZI:
olystyrene.
! .3710
1.1609
.95070
.74071
•*
<-
2.0000
_i c
2. I 440
-5.0720
2.216<>
2.2880 RS1ZE.
Z. 3600
FTGURF F8
SCATTER PLOT WITH LINEAR REGRESSION EQUATION
In etc! vs. FUNCTIONAL GROUP SIZE
FOR
WEAK AND STRONG BASE RESINS
285
-------
SCATTER F'LOl
N= 19 D'Jl OF 32 9
-------
SCATTER PLOT
N= 29 OUT OF 32 ll.RSIZE VS. 12,N2FOSITN
RSIZE
2.3600
2.2880
2.21AO
2.1440
2.0720
2.0000
16
Nitrogen position is a, dummy variable
presumably related to the distance of
functional group separation
+
0.
.40000
.80000
.20000 .60000
1.00 = Nitrogen in Polymer Backbone
0.00 = Nitrogen Pendant
N2PQSITN
1.0000
FIGURE F10
SCATTER PLOT WITH LINEAR REGRESSION EQUATION
FUNCTIONAL GROUP SIZE vs. NITROGEN POSITION
FOR
WEAK AND STRONG BASE RESINS
287
-------
GLOSSARY
"as CaCQ3": Normality (N) can be converted to calcium car-
bonate equivalents. There are 50 mg of CaC03 per milli-
equivalent. Any 0.005 N solution contains 5 milliequiva-
lents/1 or the equivalent of 250 mg/1 of CaCO.-.
bed: The ion-exchange resin contained in a column. Water to
be treated by ion-exchange is passed downward through the
column.
breakthrough: The appearance of a sharp increase in the con-
centration of an ion in the effluent from the bed.
capacity: The total number of ion-exchange sites available per
unit volume of resin measured in equivalents/1 or milli-
equivalents/ml. Resins were equilibrated with 0.005 N
acids (HC1, H2SO4^or HNO3) for the experimental capacity
determination^. 'Tlhis was done to simulate the expected
capacities in typical groundwater applications.
chromatographic elution: Continued application of the feed
water to an exhausted ion-exchange bed so as to "elute" or
sequentially drive off those less-preferred feed water
anions previously removed during the exhaustion cycle. In
this operation, the ions being driven off the resin are
separated into zones in which the aqueous concentration of
the primary ion in a given zone exceeds the concentration
of that ion in the feed water.
downflow regeneration: Cocurrent regeneration, i.e., the
regenerant solution is passed down through the bed in the
same direction as the feed water was passed through the
bed.
effluent profile: A plot of the effluent concentration of an
ion or ions vs. the volume of effluent water from the bed.
elution: The displacement of non-preferred ions previously
removed from the feed water by continued application of
the feed water or an "eluting solution" containing an ion
or ions more preferred by the ion exchanger.
288
-------
23
equivalent: One gm equivalent (6.023 x 10 ) of ionic charges
in the aqueous phase or that number of fixed charges in
the resin phase.
equivalent fraction: That fraction of the total negative or
positive charges present which is due to a given ion. If
xs = 0.27, then 27% of the negative ionic charges in a
given volume of water are due to sulfate ions.
exhaustion: The step in an ion-exchange cycle in which the
undesirable ions are removed from the water being treated.
The resin bed is said to be "exhausted" when the ions
originally on the resin have been essentially completely
exchanged for feed water ions.
functionality: A description of the nature of the amine groups
attached to the resin matrix which give an anion resin its
ion exchange properties, e.g., quaternary amine
functionality.
ion-exchange: A physicochemical process in which ions in the
water being treated replace and are exchanged for ions in
a solid phase (the resin). In the single-bed process,
nitrate, the pollutant ion, is placed on the resin phase
in exchange for an innocuous ion such as chloride.
isoporous resins: Resins having slightly greater uniform
porosity than typical microporous resins.
isotherm: A constant temperature plot of resin phase concen-
tration of an ion vs. the water phase concentration of
that ion. In a binary isotherm, e.g., sulfate/nitrate,
the resin phase exchange sites not occupied by sulfate are
occupied by nitrate. Similarly, the significant anions in
the water which are not sulfate are nitrate.
macroporous resins (also referred to as macroreticular resins):
Very porous resins whose beads comprise aggregates of gel
resins with large internal v
-------
milliequivalent: (Abbreviated meq.) 1/1000 of an equivalent.
An 0.005 N solution contains 0.005 equivalents/1 or 5
meq/1.
porosity: A measure of the degree of openness of the polymer
matrix which is related to the nature and degree of
crosslinking.
regeneration: The displacement from the exhausted ion-exchange
resin of the undesirable ions removed from the water
during the exhaustion cycle. Performed by passing through
the bed, a relatively concentrated (1 N) solution of the
ion desired on the resin.
regeneration level: A measure of the inefficiency of regen-
eration expressed here in %. The level indicates the
amount of regenerant which must actually be applied com-
pared to the amount theoretically required. For downflow
regeneration a level of 300% is typically required; that
means a 200% excess of regenerant must be applied.
selectivity: A measure of the relative affinity for one ion
over another exhibited by the resin. In this report
selectivity (relative affinity) is measured by the sepa-
ration factor, a. This a should not be confused with the
selectivity coefficient, K.
selectivity sequence: A listing of ions as preferred by the
ion exchanger ordered from most preferred to least
preferred.
separation factor (binary): The ratio of the distribution of
ions between the water phase and the resin phase. ag N is
the ratio of the distribution of sulfate ions between
phases to the distribution of nitrate ions between phases.
If ac M > I, the resin prefers sulfate over nitrate.
o / IM —
service flow rate: The' rate of application of feedwater to the
resin bed. Because the exchanger capacity is related to
the volume3 of resin, the rate is usually specified as
gal/min ft or volume of feed water per volume of resin
per unit time. With proper units this is reciprocal
superficial detentioji time. Recommended exhaustion rates
are 1-5 gal/ min ft corresponding to detention times of
from 7.48 to 1.50 minutes.
softening: In ion exchange, a process by which polyvalent
cations, e.g., calcium, magnesium, and iron are exchanged
for a monovalent cation such as hydrogen or sodium.
290
-------
spent regenerant: A wastewater containing the excess regen-
erant ions and the undesirable ions removed from the
exhausted resin. Its volume will be determined by the
volume of rinses included as "spent regenerants."
strong-base resin: An anion exchange resin containing fixed
positively charged quaternary amine functional groups
which prefer all common anions over hydroxide ions.
Simply, a resin which tends to readily give up hydroxide
ions in exchange for nearly any other anion. The capacity
of strong-base resins to exchange ions does not depend on
the presence of excess hydrogen ions (acidity) to form the
positively charged exchange sites as is the case with
weak-base resins. Thus, they may be used as ion
exchangers in acid, neutral, and basic solutions.
superficial detention time (t): The time a particle of feed
water spends in the empty resin bed assuming plug flow.
It is calculated as the empty bed volume divided by the
feed flow rate.
upflow regeneration: Countercurrent regeneration, i.e., the
regenerant solution is passed up through the bed in a
direction opposite to that taken by the feedwater.
Countercurrent regeneratipn is reportedly more efficient
than cocurrent regeneration because the most preferred
ions are not driven through the entire bed.
weak-base resin: An ion-exchange resin comprising primary,
secondary, or tertiary amine functional groups or a mix-
ture of those groups which acquire positive charges when
excess hydrogen ions (acidity) are present. These charged
sites can exchange anions if the feed solution remains
acidic. Thus, these resins are said to "adsorb" acids.
In neutral to basic solutions they have no charged sites
and consequently no significant anion exchange capacity.
They are readily regenerated with weak bases or even
neutral water solutions.
291
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-78-052
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
NITRATE REMOVAL FROM WATER SUPPLIES BY ION EXCHANGE
5. REPORT DATE
June 1978 (Issuing Date)
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Dennis A. Clifford*
Walter J. Weber
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
The University of Michigan
Ann Arbor, Michigan 48109
10. PROGRAM ELEMENT NO.
1CC614
11. CONTRACT/GRANT NO.
Grant No. R-803898
12. SPONSORING AGENCY NAME AND ADDRESS
Municipal Environmental Research Laboratory—Cin.,OH
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
13. TYPE OF REPORT AND PERIOD COVERED
Final 8/75 - 12/76
14. SPONSORING AGENCY CODE
EPA/600/14
15. SUPPLEMENTARY NOTES
Project Officer: Thomas J. Sorg 513/684-7228
*Presently with University of Houston, Houston, Texas
See also Executive Summary. EPA-600/8-77-015
77004
6. ABSTRACT
Anion exchange using synthetic organic resins is a proven and practical
technology for the removal of nitrate from water supplies. However, disposal of the
spent regenerant brine solution containing nitrate is a potential problem. Two
processes were examined in detail in this report—single-bed strong-base anion
exchange with NaCl regeneration and two-bed strong-acid, weak-base ion exchange with
HCl and NH.OH regeneration. Both systems must be operated to nitrate breakthrough to
minimize regeneration costs. The two-bed process is one and one-half to two times
as expensive to build and operate as is the single-bed process, but produces
softened low-TDS, low-nitrate water, and has a readily disposable, spent regenerant
with fertilizer value. Important design considerations were found to include the
nitrate and sulfate concentrations in the raw water, the service flow rate, the resin
bed depth, and the nitrate/chloride selectivity of the resin. The sulfate, nitrate,
chloride, and bicarbonate selectivities and multicomponent column behavior of the
anion resins available from U.S. manufacturers were examined and are reported in
detail. An important peripheral finding was that significant quantities of non-
volatile organics were leached from "clean" resins into the treated water.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
COS AT I Field/Group
Water treatment—ion exchanging, Water
supply, Ion exchanging, Ion exchange
resins, Demineralizing, Nitrate deposits-
inorganic nitrates, Sulfates, Chlorides,
Cost estimates, Experimental data
Nitrate removal, Ion
exchange—two-bed
process
13B
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
308
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
292
•&• U. ^ GOVERNMENT PRINTING OFFICE: 1978 — 757-140/1334
------- |