WATER POLLUTION CONTROL RESEARCH SERIES •
17010FSJ01/71
NITRATE REMOVAL
FROM
WASTEWATERS
BY
ION EXCHANGE
ENVIRONMENTAL PROTECTION AGENCY • WATER QUALITY OFFICE
-------
WATER POLLUTION CONTROL RESEARCH SERIES
The Water Pollution Control Research Series describes the
results and progress in the control and abatement of pollution
in our Nation's waters. They provide a central source of
information on the research, development, and demonstration
activities in the Water Quality Office, Environmental Protection
Agency, through inhouse research and grants and contracts
with Federal, State, and local agencies, research institutions,
and industrial organizations.
Inquiries pertaining to Water Pollution Control Research Reports
should be directed to the Head, Project Reports System, Office
of Research and Development, Water Quality Office, Environmental
Protection Agency, Room 1108, Washington, D. C. 20242.
-------
NITRATE REMOVAL FROM WASTEWATERS BY ION EXCHANGE
by
The Dow Chemical Company
Walnut Creek, California 94598
for the
WATER QUALITY OFFICE
ENVIRONMENTAL PROTECTION AGENCY
Project #17010 FSJ
Contract #14-12-808
January, 1971
Tor sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 2M02 - Price $1
-------
EPA Review Notice
This report has been reviewed by the Water
Quality Office, EPA, and approved for publication.
Approval does not signify that the contents
necessarily reflect the views and policies of
the Environmental Protection Agency, nor does
mention of trade names or commercial products
constitute endorsement or recommendation for
use.
11
-------
ABSTRACT
This report describes an exploratory experimental study of the use of
porous polymer beads containing a water-immiscible extractant system
for the removal of nitrate from waste waters.
NHR'
Alkylated amidines, R - C ^ ,, (R's are octyl through dodecyl), proved
^"* NR
to be a suitable class of compounds for the extractant system. The ami-
dines are relatively strong bases, and possess the advantage over the
simple aliphatic amines that they exist in the salt form in contact with
waste waters in the pH range of 7-8. They can, however, be readily re-
generated with alkalis, such as ammonia or sodium hydroxide.
The amidinium ion in the organic phase selectively extracts nitrate ion over
chloride ion by a factor of about 20 (i. e. , the nitrate/chloride ratio in the
organic phase is about 20 times the ratio in the equilibrium aqueous
phase), and nitrate over sulfate and bicarbonate by much higher
ratios. From typical municipal waste waters amidine systems will
therefore pick up mainly the nitrate ion.
Amidines dissolved in an aromatic hydrocarbon were absorbed in macro-
porous polystyrene beads and used to treat a synthetic municipal waste
water containing 62 ppm nitrate ion and 350 ppm chloride ion. Beds
of this material treated up to 70 bed volumes of water prior to
breakthrough of the nitrate in the effluent. The absorbed nitrate
ion was removed with either ammonia or sodium hydroxide. Soluble
losses of the extractant were in the range of 1-20 ppm, and could be
reduced to well below 1 ppm by proper choice of materials.
This report was submitted in fulfillment of Project Number 17010 FSJ,
Contract 14- 12-808, under the sponsorship of the Water Quality
Office, Environmental Protection Agency.
Key Words
Nitrates Anion Exchange
Denitrification Resins
Waste Water Treatment Tertiary Treatment
Municipal Wastes Solvent Extractions
Water Pollution Treatment
111
-------
CONTENTS
Page
ABSTRACT ill
CONCLUSIONS AND RECOMMENDATIONS 1
INTRODUCTION 3
Removal of Nitrates from Water 4
Basis of Extractant-in-Bead Approach 6
Solid Absorbents 7
Extractant Systems 8
SYNTHESIS OF AMIDINES AND GUANIDINES 13
Amidines 13
Starting Materials 15
Products f6
Guanidines 20
PHYSICAL CHEMISTRY OF THE AMIDINE EXTRACTION
SYSTEMS 23
Basis of the Experimental Methods 23
Apparent Basicity of the Amidines &3
Ion Sele ctivity 25
Soluble Loss of Extractant to Aqueous Phase. . . 2*5
Experimental Methods &6
Apparent pK of Extractants 2.6
Soluble Loss of Extractants 2',7.
Ion Selectivity of Amidines 2,7,
Results and Discussion 2,8
pH Dependence of the Extractant £•§
Ionic Selectivity of Extractants 3!9
Soluble Loss of Extractants 46
Chemical Stability 49
Soluble Loss of the Diluent 50
Water Extraction and Volume Change 51
ION EXCHANGE BEHAVIOR OF SOLID SUPPORTED
EXTRACTANTS 5>5.
Incorporation of Extractants into Macroporous
Copolymers 55
Solid Absorbent 55
Extractants 56
Preparation and Properties of Extractant
Loaded Resins 5,6
-------
CONTENTS
(Continued)
Extractant Absorption Capacity ............ 57
Column Studies ........................... 59
Experimental Procedure ................ 59
Analytical Methods ..................... 5'9
Results of Column Studies .................... 60
OUTLINE OF NITRATE REMOVAL PROCESS ......... 67
Process Description ....................... 67
Adaptation of Extractant-Loaded Beads to Column
Operation ............................. 67
Flow Rate .......................... 69
Nitrate Loading ....................... 69
Elution Performance ................... 7Q
Handling of Beads and Other Operational
Problems ......................... 71
Commercial Availability of Amidines ............ 73
Cost Projections .......................... 76
ACKNOWLEDGMENT ......................... 81
REFERENCES .............................. 83
APPENDIX I SYNTHESIS PROCEDURES ............ 87
APPENDIX II DETERMINATION OF ALKYLAMMONIUM
IONS IN AQUEOUS SOLUTION ......... 93
APPENDIX III pK OF AMIDINES ................. 95
cl
APPENDIX IV SOLUBLE LOSS OF AMIDINE KYDRO-
CHLORIDES .................... 99
v i
-------
FIGURES
Figure Page
1 Apparent pK of Amidines 30
ct
2 Apparent pK Values of Amidines 31
a
3 pK for Eh EhA in Chevron 3 34
a 2
4 Effect of pH on Extractant Form 35
5 Effect of pH on Extractant Form 38
6 Soluble Loss of Amidines 45
7 Soluble Loss of Eh EhA from Chevron 3 into
0. 010 M NaCl . 48-
8 Extraction of Water by Eh EhA in Toluene .... 53
£i
9 Nitrate Absorption by Extractant Loaded Macro-
porous Resins 61
10 Effect of Flow Rate and Column Packing on
Nitrate Absorption 62
11 Nitrate and Chloride Elution with 0. 5 NaOH ... 65
12 Nitrate and Chloride Elution with 0. 5 N NH OH. . 66
13 Schematic Diagram of Nitrate Removal Process. . 68
Vll
-------
TABLES
Table Page
I pK Values of Typical Organic Bases >
3.
1°
II Properties of Amides .................. 17
III Properties of Amidines ................. 19
IV Properties of Guanidines ................ 2-4
V Nitrate/ Chloride Selectivity of Amidines ...... 40
VI Nitrate/Sulfate Selectivity of Amidines ....... 41
VII Nitrate/Bicarbonate Selectivity of Amidines . . . 42.
VIII Expected Extractant Composition from Treatment
of Typical Waste Water ................ 44
IX Comparison of Ultraviolet and GLC analyses
for Chevron 3 in Aqueous Phases ......... 52'
X Extractant Absorption Capacities of the Macro -
porous Copolymer ................... 58
XI Nitrate Absorption Capacities of Macroporous
Resins ........................... 63
XII Nitrate Elution from MPA-2 with NaOH and
64
XIII Raw Materials Costs for Amidine Synthesis .... 75
XIV Basis for Cost Projection ................ 78
XV Cost Projection ...................... 79
Vlll
-------
CONCLUSIONS AND RECOMMENDATIONS
The major accomplishment of this project has been the elucidation
of the chemistry of high molecular weight amidine as liquid-liquid
extraction agents. The chemistry exhibited by these materials
identifies them as both unique and adequate materials for the removal
of nitrate from waste waters. They exhibit sufficient basicity to form
salts in contact with neutral aqueous solutions. They are, on the other
hand, sufficiently weak in base strength to be converted back to the
base form with either ammonia or alkali. They exhibit good selectivity
for nitrate ion over other common inorganic anions. Soluble losses
of the salt form to dilute aqueous solutions can be kept in the region
of one ppm or less. And finally, they are quite stable toward hydrolysis
by alkali, when used in hydrocarbon diluents. Soluble losses of the
amidines and the aromatic hydrocarbon diluent used in this study
were not as low as desirable for optimum economic performance.
However, on the basis of the studies, it is clear that minor modifications
would probably eliminate this factor as an economic problem.
Absorption of hydrocarbon solutions of amidines into porous polymer
beads provides a means of utilizing the chemistry of the amidines
to remove nitrate from water. Column experiments have demonstrated
that removal of nitrate can be effectively accomplished at low concen-
trations and in the presence of typical chloride concentrations. The
major drawback of the extractant-in-bead approach is the relatively low
(compared to conventional ion exchange) total exchange capacity of the
beads. This factor results in low feed water capacity per cycle, and
in low nitrate concentration in the eluate. The former leads to an in-
crease in resin bed requirement, whils the latter increases the volume
of eluate to be discarded or evaporated for discard.
Further improvement of the extractant-in-bead system can probably
be expected by further study of the processing variables. Use of two
beds in series, or the recycling of some of the lower concentration
portions of the eluate, or use of slower elution flow rates, may
all lead to higher eluate concentrations and lower evaporation costs.
_'„ some situations direct utilization of the eluate as a fertilizer
material might eliminate the evaporation requirement altogether.
A rough projection of processing costs, based on the data of this
study and an earlier cost estimate for ion exchange processing of waste
water for nitrate removal, indicate a cost in the range of 16<|: per
thousand gallons (16£/M gal) of feed. Of this, about 10 $ is required
for evaporation of the eluate to a 40% sodium nitrate-sodium chloride
solution. Further development may determine that this evaporation
requirement can be lowered, but it probably cannot be eliminated.
The ultimate processing cost can probably be expected, therefore,
to fall in the range of 5-10J:/M gal.
1
-------
Amidine compounds can probably be utilized in other ways than
absorption in polymer beads. The simplest competitive method would
be conventional liquid-liquid extraction.
On the basis of these conclusions, we recommend the following:
1. Development of this system should be continued to a limited
extent, in order to obtain a clearer picture of the nature
of the elution process. The work should be concentrated on
the effects of the operating variables such as flow rate,
bed depth, bead size and type, and eluant type and strength.
The importance of estimating the final eluate concentration
lies in the dominant position of the evaporation cost required
to reduce its volume for disposal or transport for use. A
secondary question, which should receive some effort, is
whether diluent-free beads (containing extractant only)
will be effective, since greater concentrations of extractant
may then be achievable.
2. Other potential means of utilizing the amidines for nitrate
removal should be considered. Conventional liquid-liquid
extraction is a significant possibility. This, and any other
promising ideas for utilizing the amidine chemistry in other
contacting systems should be explored in a preliminary
way and compared to the use of polymer beads prior to any
further development of the latter beyond that described in
recommendation 1.
-------
INTRODUCTION
Nitrogen Compounds in the Environment
Nitrogen compounds are fundamental components of all biological
systems. They are not only required for maintenance of these systems, but
are also found as waste products from the metabolism of biological
systems, and are released to the environment in the decomposition
of these systems after death. Because of this strong role played by
nitrogen, the functioning of biological systems and the maintenance of
delicately balanced relationships among them is often critically
dependent upon the nature and levels of nitrogen compounds existing
in the environment.
Until recent years man's major concern with nitrogen was with the
use of nitrogenous fertilizers to increase the yield of crops. The
quantity of nitrogen compounds in waste streams from his activities
has heretofore been sufficiently small that any effects on the environ-
ment have gone unnoticed. Within the last several years, however,
the increasing concentrations and quantities of both plant nutrients,
phosphorus and nitrogen, in water bodies have led to the recognition
of a number of environmental problems. Studies have identified the
major sources of these compounds, which, in the case of nitrogen, are
agricultural runoff and municipal and industrial waste water discharges(3).
Minor amounts are contributed byurba-n runoff and rainwater.
One major problem is the occurrence of nitrate compounds in ground
water. Nitrate ion can appear here either from use of nitrates as
fertilizers, or from the oxidation of other nitrogen compounds in
the soil by bacteria. Nitrates are quite soluble in water and are not
readily adsorbed on soil particles, and the percolation of rain or
irrigation water through the soil readily carries nitrates into the
ground water table . This situation has created a public health
problem in some areas, since nitrate is responsible for the condition
known as methemoglobinemia in very young infants (4). Nitrate is also
toxic to adults, though at considerably higher levels, and the U.S.
Public Health Service has set a limit of 45 ppm of nitrate for drinking
waters{5).
This introduction is intended to provide only a brief summary of
the nitrate problem. For a more detailed examination of the problem,
the reader should consult the proceedings of the Conference on
Nitrate and Water Supply, held in 1979 (1). A bibliography of
material on the sources and effects of nitrate is also available (2).
-------
Significant concentrations of nitrate in ground water have been
reported in various locations in California (4, 6} in several midwestern
and eastern states (.7, 8, 9) and abroad in Israel and Germany (10)
(11). Many of these contain nitrate at greater than 50 ppm and represent,
therefore, potential hazards not only to humans, but also to livestock
which may be required to use them.
A second major problem associated with the existance of nitrates in
water is the occurrence of algae blooms or explosive growths of
algae. The latter, being plants, require nitrogen and phosphorus
as nutrients, and utilize carbon dioxide in their metabolism. Under
certain conditions algae blooms occur in natural water systems,
after which the decomposition of the dead algae by bacteria consumes
oxygen and often results in extensive fish kills. Increasing frequency
of blooms, and their close association with sewage discharges has
implicated both nitrogen and phosphorus as contributors. High
levels of carbon dioxide have also been suggested as the crucial
factor (12). At this point the relative importance of these factors,
as well as others, has not been clearly established. It seems clear,
however, that increasing sewage loading of natural water bodies
by increasingly populous urban areas will probably require more
extensive treatment of sewage, probably including major reduction
of total nitrogen concentration.
Nitrogen occurs in sewage mainly as organic nitrogen or ammonia,
and passage through most secondary sewage plants, while decomposing
the organic material, still leaves the nitrogen mainly in the ammonia
form (13). Modifications in plant operation can, however, convert
most of the nitrogen to nitrate, if desired (14, 15, 16).
An additional source of nitrogen compounds in the environment is
agricultural drainage obtained from underground tile drain systems.
In California extensive drainage systems are used in the Central
Valley and volumes of up to 500, 000 acre feet of water containing
up to 100 ppm of nitrate are anticipated during the next 50 years (17).
A considerable controversy surrounds proposals to discharge this
waste stream into the Sacramento-San Joaquin river delta, and con-
siderable interest currently exists in possible means for eliminating
nitrate, from this water stream.
Removal of Nitrates from Water
Because of the emergence of the nitrogen problem considerable
research effort has been devoted to methods for eliminating ammonia
-------
and nitrates from water (1, 18). The major methods in existence or
under study include bacterial reduction of nitrate to nitrogen
(denitrification), utilization of nitrate by algae, air stripping of
ammonia, and selective ion exchange removal of ammonium ion.
These methods are summarized in several surveys of the
problem (1, 18, 19).
At this time the most promising method for nitrate removal appears
to be the anaerobic denitrification process. It has been known for
some time that under anaerobic conditions, in the presence of organic
matter, certain bacteria will reduce nitrate to nitrogen gas. The
process has been actively studied as a means of eliminating nitrates
from water (3, 18, 20). For removing nitrogen compounds from sewage,
an additional nitrification step beyond the normal secondary treat-
ment is required. In this step, a bacterial culture is maintained to
oxidize ammonium compounds to nitrates. The effluent from this
step is fortified with a suitable organic substrate, usually methanol,
and is conducted through an anaerobic pond or tank containing a culture
of denitrifying bacteria. For waste waters containing nitrate in the
range of 20 mg N/l a 90% removal is readily achieved with costs
in the range of 8 j*/M gal (1). The process has been operated on
scales as large as 1 mgd and a 300 mgd plant is being planned in
Washington, D. C. (18). The process has also been tested on subsurface
agricultural drainage at the Firebaugh test station in California,
apparently also with success (20).
Nitrate can also be removed from water with conventional anion
exchange resins (1, 21, 22, 23). Most commercial materials exhibit
a selectivity for nitrate over chloride of 2 or 3*, with the result that
a water containing nitrate to chloride ratio of, say, 0. 1, will result
in a loaded resin containing a ratio of 0. 2 or 0.3. The difficulty
arises in the fact that the absorbed nitrate is relatively difficult to
remove With the standard sodium chloride regenerant, and the problem
of disposing of the regenerant solution containing the nitrate still
remains. A cost estimate for such a process, including an additional
liquid-liquid extraction step to concentrate the nitrate into an ammonium
zircr'ate product, was prepared in this laboratory under a recent contract (24).
The total treatment cost, which applied to that treatment of agricultural
drainage water, was found to be about l6j£/M gal from which a credit
of about 3^ could be anticipated for the value of the ammonium nitrate.
* Selectivity is defined as the nitrate/chloride ratio in the resin
phase divided by the ratio in the solution phase. It can also be
defined as the ratio of distribution coefficients (resin/solution)
of nitrate over chloride.
-------
This estimate was based on a fairly large plant ( 100 mgd), but also
was based on a rather unfavorable water conposition (about 7.000 ppm
total dissolved solids). Application to a sewage stream, where less
dissolved solids are usually found, might be a more attractive pro-
position.
A specific task of that contract was also to estimate the effect of
using resins of higher selectivity for nitrate over chloride. Usi ng
a selectivity value of 20, the corresponding treatment cost for the
same waste stream was about6^/M gal, from which, again, a
credit of about 3£ could be anticipated for the ammonium nitrate
product.
With this information, the goal of the work reported here has been
to develop a material with a considerably higher nitrate/chloride
selectivity than is available in existing exchange resins. The approach
has been in the direction, not of a conventional exchange resin, but of
a porous resin bead containing a water-immiscible liquid ion-exchanger.
Basis of the Extractant-in-Bead Approach
The problem of development of more selective resin materials can
be broken into two parts: the search for a phenomenon which can
provide the desired selectivity under any conditions at all; and the
adaptation of this phenomenon to a resin system. Certain aliphatic
amine solvent extraction systems are known to exhibit fairly high
selectivity for nitrate over chloride and other common ions (25),
and selective extraction of nitrate from mixtures of anions based on
this phenomenon is easily conceivable.
Solvent extraction systems, however, while quite useful for processing
various industrial materials, are less attractive for processing
water. This is partly a matter of economics, since water, at, say,
50^/M gal. , is many times less valuable than most industrial process
solutions. However, in processing water; an additional considera-
tion arises, namely, that of providing an effluent containing a minimum
of organic matter which could contribute to a subsequent pollution
problem.
A major source of organic material from a liquid-liquid contact is
the physically entrained solvent system, left after coalescence
of the bulk of the extractant phase. One possible means of circum-
venting this problem is to use a solid absorbent in which is contained
the desired liquid extractant system, eliminating the need to disperse
the organic liquid to obtain liquid-liquid contact.
-------
Solid Absorbent Materials. The use of liquids adsorbed on porous
solid materials forms the basis of reversed phase partition chroma-
tography. The first published use of a gel type polymer bead seems
to have been that of Small (26, 27, 28), who used cross-linked poly-
styrene beads to absorb tributyl phosphate solutions in organic
diluents. These materials were found to be fairly effective for extract-
ing uranium from aqueous solutions, by analogy with the liquid-
liquid systems which have been used for the same purpose.
A patent has been granted for a similar process covering the use of
solid adsorbents containing aliphatic amines or esters of phosphoric
acid for removing metal ions from aqueous solution (29). A related
patent described the use of activated charcoal containing ethanolamine
as an absorbent for gaseous carbon dioxide (30).
Small used the gel type beads, which consist of a more or less uniform
polymeric matrix, with occasional cross-links to render the material
insol-nible. In principle, however, any hydrophobic material, including
activated carbon, could be used, which would absorb the water-
immiscible solvent system preferentially over water.
We considered the use of activated carbon, but an examination of the
major types currently available revealed that most materials possessed
much smaller pore sizes than we believed desirable. The use of the
gel type polystyrene beads appeared unattractive since the diluent
system must be one which swells the polystyrene matrix. We therefore
confined our attention to the so-called "macroporous" or "macro-
reticular" beads, which consist of a network of cross-linked polymer
structures, but contain also relatively large pores or voids, with
diameters up to several hundred Angstroms(31). The rationale
behind these materials was that such pores were small enough that
the organic phase would resist hydraulic or mechanical forces which
might tend to strip away the organic liquid. They would on the other
hand, be large enough to allow essentially bulk solution behavior;
i. e. , diffusion of dissolved materials would be about as rapid as in
a bulk liquid. This is of considerable practical importance, since
one of the characteristics of gel type ion exchange resins is that
dissolved species exhibit much lower diffusion coefficients in the
resin phase than in the solution outside.
-7'-
-------
Extractant Systems. While the polymer beads were available as
materials at the start of this study, none of the extractant systems
existed. Four major requirements must be satisfied by an extractant,
to be considered useful in this application. They are as follows:
1. It must be susceptible to being stripped or eluted of
anions with alkali;
2. It must exhibit suitable ion selectivity alone or in an
appropriate diluent;
3. It must be sufficiently basic to remain in the salt or
cationic form at the pH of natural waters (presumably in
the range 7-8);
4. It must be sufficiently hydrophobic to be strongly retained
by the bead, and exhibit only very small soluble losses
to the aqueous stream treated.
The first requirement is relatively easily satisfied by using a non-
quaternary amine (i. e. , tertiary, secondary, or primary) wherein
the presence of the proton in the substituted ammonium ion provides
pH sensitivity. Reaction of the amine salt with a base converts
the salt to a neutral molecule and results in removal of the inorganic
anion from the organic phase as follows:
R NH+ NO " + Na+ + OH~—>R N + HO + Na+ + NO ~ (1)
J J J £ J
(Barred species denote those in the non-aqueous phase. )
Adequate selectivity for nitrate also can be observed in the simple
aliphatic amine systems. We define the selectivity constant as the
equilibrium constant for the exchange reaction:
R+ X" < R+N0" + X" (2)
NO (NO ") (X~)
s 3 = - £
X (N03~) (X")
-8-
-------
This constant can also be thought of as the ratio of distribution
coefficient of nitrate to the coefficient of some other ion, X . It
is also equal to the nitrate/chloride ratio in the extractant divided
by the nitrate/chloride ratio in the aqueous solution.
Earlier work with amine extraction systems had indicated (32)
that primary amine systems were not particularly selective for
nitrate over chloride, but that secondary, tertiary, and quaternary
ammonium ion systems were selective, exhibiting constants of the
order of 20 in favor of nitrate. Studies of other ions were considerably
less detailed, but indicated that selectivity constants for nitrate
over sulfate and bicarbonate were in the range of 100-5 (;0 (33).
In the case Of sulfate, the constant S is the equilibrium constant for
the reaction:
2(BHN03) + SC>4
With respect to the basicity of the aliphatic amines, the situation
is less satisfactory. This subject has also been investigated rather
extensively by us (34), and a full discussion is beyond the scope of
this report. It will be useful, however, to outline the situation in
order to provide the basis of this application. Although we are con-
sidering a system in which the amine would be held in a bead, the
treatment is essentially identical to the case of the amine in a liquid
solvent system.
If we denote the free amine base by B, and the extracting anion by X ,
then in the equation
(BHX) B + H + X (5)
the dissociation constant is given by:
(6)
(BHX)
In order for the amine systems to function as liquid phase ion
exchangers, it is necessary to operate in a pH range where the
amine is primarily in the salt or ammonium form.
If we then wish to maintain 90% of the extractant in the salt form,
and if we assume treatment of a waste water containing 0. 01M
chloride (about 350 ppm) and a pH of 7, we find that
-9-
-------
10-7x0.01
1Q
-10
Thus, .potential extractants must exhibit proton dissociation constants
of 10 or less (pK of 10 or more) in order to be suitable for use
in this problem. Although the lower aliphatic amine salts do exhibit
pK values of this order of magnitude in water solution, the higher
molecular weight homologs behave somewhat differently, and no
pK values above about 8 or 9 have been reported (25, 34). Primary
amines function as the strongest bases in liquid-liquid systems, but,
as mentioned above, do not exhibit much selectivity.
At the time of writing of the original proposal on this subject, we
suggested that the requirement of stronger basicity should be met
by compounds of the amidine and guanidine types. The following
are typical examples.
R
NR'
NHR
R - N =
NHR'
NHR"
amidines
guanidines
Both types of compounds are known to be considerably stronger
bases than aliphatic amines. This is clearly shown by the aqueous
pK values (where K is the acid dissociation constant of the amine
salt) of some typical representatives of these three types of compounds
(35), which are given in Table I.
Table I
pK Values of Typical Organic Bases
— ^— **" * •™* ™ * "* "^^^^— i
Compound
-
Ethylamine
Diethylamine
Triethylamine
Acetamidine
Benzamidine
Guanidine
pK
r_a
10.63
10. 93
10.87
12.40
11.6
13. 6
-10-
-------
Although, as mentioned above, the pK values are not directly
usable in estimating behavior in liquids-liquid extraction systems,
it is clear that amidines are at least 10 times as strong bases as
the aliphatic amines, while guanidines are even stronger bases.
The synthesis portion of this study has accordingly been directed
at these two classes of compounds, with primary attention given
to the amidines.
-11 -
-------
SYNTHESIS OF AMIDINES AND GUANIDINES
Amidines
Amidines can be considered as the nitrogen analogs of carboxylic
acids, where the oxygen atoms of the latter have been replaced with
nitrogen atoms plus the required number of protons. Amidines
can be prepared in a number of ways, but all involve starting from some
derivative of the corresponding carboxylic acid. Some general
reviews of amidine chemistry, including their synthesis, have been
published (36, 37).
In approaching the problem of amidine synthesis for this study,
our na jor goal was to obtain sufficient material for small scale
studies by procedures which could be conveniently carried out in
the laboratory. However, we also wished to keep in mind the possibility
that, should the process under study prove to be a practical one, the
amidines should be capable of synthesis on a commercial scale by
methods which would be reasonably economical and flexible. We
selected, on this basis, the route involving halogenation of carboxamides
to imidoyl halides, followed by reaction of this intermediate with
an amine.
If an amide is allowed to react with any one of several inorganic
acid chlorides, the following reaction occurs:
Cl
-,+ r- I n+ -
RCONHR' + COC1 _ >[RC=0-COCl] Cl.^R -C = NHR J Cl +CC>2
NHR1
Cl
I
_jrL_> RC = NR' + HCI
imidoyl chloride (8)
The adduct, I, is metastable in most cases, and decomposes
readily in this case to carbon dioxide and the iminium chloride,
which in turn decomposes to HCI and the imidoyl chloride. Thionyl
chloride, phosphorus pentachloride, phosphorus oxychloride and
several other halides have been used to effect this reaction (38,
39,40).
-13-
-------
The imidoyl halide is a very reactive species, and reacts readily
with nucleophilic species, much as the analogous acyl halides do.
In the synthesis of amidines the reactant is an amine, giving the
reaction
R
Cl
I
- C = NR'
+ R"NH,
R C
;NHR'
•NHR"
Cl" (9)
Both reactions apparently proceed readily, and yields of over 90%
have been reported for both steps (40,41, 42). Yields actually reported
by various workers differ widely, and probably depend on the specific
compounds under study, but the general procedure seems capable
of producing relatively high yields of the desired products. We did
not attempt to optimize our procedures but nevertheless were able
to obtain yields in the 50-70% range.
The major source of difficulty appears to be a side reaction which
occurs at the imidoyl halide stage (38). When the alpha carbon of
the carboxyl group contains a proton, tautomerization of the imidoyl
halide can occur, the tautomer then reacting with more imidoyl
halide as fellow's:
RCH
Cl
I
- C =
II
NR1 <-
RCH =
Cl
I
C
III
NHR1
(10)
H + in
. RCH- - C
-> 2
NR1 C - CHR
I
Cl + HC1 (H)
NR1
IV
The compound IV is still an imidoyl chloride, and can react Avith the
amine in the final step of the reaction, thereby reducing the yield.
Many workers in this field have, therefore, used formic acid derivatives
(43), which have no alpha carbon, or aromatic acids (40) which have
14-
-------
no alpha hydrogen. There is some evidence that sterically hindering
the alpha carbon atom reduces the extent of this side reaction (38),
and most compounds synthesized in this study used 2-ethylhexoic acid
as the starting carboxylic acid, mainly for this reason.
The amides used as starting materials for the preparation of imidoyl
halides were prepared in the conventional manner, using the acyl
chloride and the appropriate amine in two fold excess. The reaction is
RCOC1 + 2 R'NH2 &ther v. RCONHR' + R'NH^l (12)
The amine hydrochloride was extracted from the ether with aqueous
hydrochloric acid, and the amide was obtained by evaporating the
ether.
Starting Materials
In most cases, starting materials were purchased from regular
supply houses. In a few cases, we used commercial materials
which consisted of mixtures, or were identified with trade names.
The neodecanamide derivative was prepared from neodecanoyl
chloride, purchased from both K & K labs, and from the White
Chemical Company. The prefix "neo" is used to denote the absence
of alpha hydrogens, but indicates nothing else about the structure
of the hydrocarbon chain. Proton magnetic resonance patterns of
the two samples indicated them to be quite similar, possibly identical,
and confirmed the absence of alpha protons.
2
A second material used in these preparations was Primene 81-R , a
primary aliphatic amine obtained from the Rohm & Haas Company.
This is a mixture of isomers obtained by placing a primary amino group
on a tertiary carbon of a branched chain hydrocarbon. This particular
material exhibits an equivalent weight corresponding to an average of
about 13 carbon atoms.
Mention of commercial products throughout this report does not
imply endorsement by the Federal Water Quality Administration.
-15-
-------
Products
The amines prepared in this program are listed in Table II.
Detailed synthesis procedures are listed in Appendix I.
The amides, which were colorless, slightly viscous liquids, were
characterized by elemental analysis and examination of their
infrared spectrum. The products were of fairly high quality and
were not further purified, but used directly in the next step. Some
of the amides listed in the table were prepared in anticipation of
their use in certain preparations which, because of lack of time,
were not pursued.
The imidoyl halides were prepared in solution, but were not isolated.
The amide and phosgene were mixed in toluene and allowed to stand
overnight to form the imidoyl halide, according to equation 8.
Excess phosgene was removed along with some diluent, under vacuum,
in order to eliminate reaction with the final amine. Finally, the
desired amine was added to the imidoyl halide solution and allowed
to react, according to equation 9, to form the amidine salt.
Excess amine was removed by contacting the toluene with hot aqueous
hydrochloric acid and then hot water. Ii the toluene were removed
from the mixture at this stage, the product usually had a very high
equivalent weight, 600 or more, instead of the desired value, which
was usually in the range of 400.
We found that the neutral impurity could be rejected rather well by
partitioning the crude product between hexane or pentane and an
ethanol-water medium. This system forms two phases as long as
the water content of the ethanol remains above about 20%, and the
ethanol remains almost entirely in the aqueous phase. In this system,
the amidine hydrochloride enters the aqueous ethanol phase, leaving
the neutral impurity in the hydrocarbon. Several washes of the
aqueous phase with hexane usually reduced the impurity to a level
which gave an equivalent weight within 10% of the theoretical. For
purposes of this study we arbitrarily established this as an acceptable
level. We found that it was also possible to convert the amidine to
the free base, in which form it preferred the pentane or hexane phase.
Successive washes of this phase with ethanol-water also removed the
undesirable impurity. The former procedure was preferable, since
the crude product is then obtained in the hydrochloride form, in
which form it is less susceptible to hydrolytic decomposition.
-16-
-------
Table II
Compound
N-(2-eth.ylhexyl)-2-ethylh.exanamide
2-ethylhexyl-p-t-butylbenzamide
n-octyl-n-octanamide
N-(2-ethylhexyl)-n-octanamide
N(2 -ethylhexyl) -neodecanamide
N-Dodecylneodecanamide
N-Octyl-2-ethylhexanamide
N-Dodecyl-2-ethylhexanamide
Properties of Amides
Elemental Analysis
Notebook Reference
GW-48-68: 107+140
GW-48-68:132
GW-48-68:133
GW-48 68:145
GW-1-70:53
GW-1-70:48
GW-19-70:68
GW-1-70:43
Yield C
Found 75.5
92-5% Calc 75.2
Found 78.5
Calc. 78. 9
72% Found 75.9
Calc. 75.2
-
95%
75%
89%
61%
H N
12.7 5.8
13.0 5.5
10.6 4.5
10.8 4.8
12.5 5.4
13.0 5.5
-------
One amidine was prepared by a different route, involving the addition
of a Grignard reagent to dicyclohexyl carbodiimide, as follows:
MgBr
i -^^^
(13)
MgBrOH + C8H17 (14)
The carbodiimide happened to be a convenient material at the time
it was done, but the carbodiimides are no longer commercially
available. For all other syntheses we have therefore used the imidoyl
chloride method.
The various amidine hydrochlorides prepared in this study are listed
in Table III. Detailed synthesis procedures for each of the methods
used are given in the Appendix.
Because of the awkwardness of using the full names of these compounds,
we have resorted to a shorthand description of them, as follows.
Aliphatic substituents are denoted by a two-letter abbreviation; e. g. ,
octyl = Oc, dodecyl = Do. For amidines the two substituents on
the nitrogen are given first, followed by the designation of the
carboxylic acid chain, and the capital letter A to denote the amidine
structure. For example, N-(2-ethylhexyl)-N' -(n-dodecyl)-Z-
ethylhexanamidine the structure of which is
C2H5 N - (2-ethylhexyl)
C H . CH - C
NH (n-dodecyl)
is designated as DoEhEhA, where Eh denotes 2-ethylhexyl, and Do
denotes n-dodecyl.
-18-
-------
Table III
Properties
of Amidines
Shorthand Approx.
Compound Designation Notebook Yield
N,N' Dicyclohexyl non- CyNoA GW 48-68:12
anamidine, hydrochloride
N,N' di(2-ethylhexyl)-2- Eh EhA GW 48-68:125
e th ylhexanamidine
, GW 1-70: 28
t—>
vO
N-(2-ethylhexyl)-N'- EhOcEhA GW48-68:142
(n-octyl)-2-ethylhexan-
amidine, hydrochloride
N, N'-di(2-ethylhexyl)n-oct- Eh OcA GW48-68.-146
anamidine, hydrochloride
N-di(n-Butyl)-N'-(2-ethyl- (Bu )EhOcA 1-70:13
hexyl) -2-2 -eth ylhex-
anamidine
N-Dodecyl-N'-(2-ethyl- DoEhEhA 1-70:21
hexyl)-2-ethylhexan-
amidine, hydrochloride
N-Dodecyl-N'-(2-ethyl- DoEhNdA 1-70:51
hexyl) -neodecanamidine
20%
22%
64%
20%
22%
29%
60%
35%
Found
Calc
Found
Calc
Found
Calc
Found
Calc
Found
Calc
Found
Calc
Elemental
C
70.
70.
79.
78.
71.
71.
72.
71.
73.
73.
80.
80.
8
7
4
7
9
5
0
5
5
2
4
2
H
11.
11.
13.
13.
12.
12.
13.
12.
12.
13.
13.
13.
6
6
4
7
8
8
0
8
7
0
7
7
Analysis Equivalent
N
7.
7.
7.
7.
7.
6.
6.
6.
6.
6.
6.
6.
Cl Weight
6
8
5
6
0
9
8
9
3
3
2
2
8.6 Found
9.9 Calc
Found
Calc
Found
Calc
8. 3 Found
8.8 Calc
8. 0 Found
8.8 Calc
Found
Calc
7.4 Found
7.8 Calc
Found
Calc
412
357
391
367
388
367
427
403
442
403
378
367
480
460
483
450
-------
Guanidines will be denoted in the same way, except that, since all
of the nitrogen atoms are equivalent, the order of stating the sub-
stituent groups is immaterial. The compound N, N1 -dioctyl-N"-
(2-ethylhexyl) guanidine is denoted Oc EhG.
While most yields shown in the table are relatively low, this is at
least partly the result of losses during the partitioning and workup
procedures. Actual yields of crude product would be significantly
higher, but we did not attempt to determine them.
One cause of low yield was presumably side reactions, of which one
is that shown in equations 10 and 11. We did not attempt to identify
these impurities, although we were able to identify considerable
amounts of the starting amide in some products. In later preparations,
the reaction time in both stages was increased, with a corresponding
increase in yield.
In addition to the elemental analysis and equivalent weight, the
infrared scans of the products were obtained, and in a few cases,
proton magnetic resonance scans were obtained, to confirm the
structures.
Both chloride and equivalent weight determinations were made
titrimetrically in a water-acetone medium. Chloride was determined
by titration with silver nitrate in the standard manner. Equivalent
weights were dete rmined by first converting a sample of the compound,
dissolved in a diluent, to the free base by contact with dilute sodium
hydroxide. Titration with standard acid gave not only the equivalent
weight, but also an indication of the presence or absence of the excess
free aliphatic amine used in the final reaction step. Because of
its much weaker basicity presence of this material showed up as
a shoulder on the titration curve of the amidine.
The amidines prepared here were all pale yellow or colorless
liquids at ambient temperatures. The hydrochlorides, however, were
extremely viscous liquids, also pale yellow or colorless, with the
consistency of molasses. The one exception to this was the dicyclo-
hexyl analog, which was obtained as the bromide salt, and was a white
crystalline solid. During the course of this work, none of the liquid
hydrochlorides showed any tendency to crystallize.
Guanidines
The most convenient and versatile laboratory synthesis of substituted
guanidines involves conversion of a substituted thiourea to the S-
alkyl thiuronium salt, and subsequent conversion of this intermediate
-20-
-------
to the guanidine (44, 45). If the desired thioureas are not available,
they can be readily prepared from carbon disulfide and amines (46).
The reactions are:
H2°
2 RNH + CS + NaOH ^ (RNH) CS + NaHS + H_O
c, £. T f £ c,
(15)
(RNH) CS + C H Br
u L» D
EtOH
RNH.
RNH
C - S C,H,
> 25
Br" (16)
RNH.
RNH
,G - SC2H5
EtOH
Br + R'NH,
_>
r
RNH
C=NHR'
RNH'
(17)
C2H5 SH
Relatively good yields were obtained for all three steps. The final
product was subjected to the partitioning procedure described in
the amidine synthesis section, in order to improve the quality of
the product.
One additional synthesis involved the use of N, N1-bis-(cyclohexyl)
carbodiimide, which was available at that time. Formation of
a guanidine occurs rather simply by reaction with an amine, as
follows:
N = C =
+ RNH,
ONH-C=N
NHR
-21-
-------
The guanidines prepared oy tnese methods are listed in Table IV.
Detailed synthesis procedures are given in Appendix I. These
materials were usually obtained as the hydrobromides, but both
bromides and chlorides were waxy solids or extremely viscous
liquids. None were soluble in toluene to the extent of 0. 1 molar.
Since the amidines appeared to be suitable materials, the synthesis
of additional guanidines was discontinued.
-22-
-------
PHYSICAL CHEMISTRY OF THE AMIDINE EXTRACTION SYSTEMS
Prior to the start of this study we had planned to examine both amidines
and guanidines as potential extractants. As events have developed,
we have confined our work almost entirely to the amidines for two
major reasons. First, initial experiments with the amidines indicated
they fulfilled all of the requirements which we had set forth for such
materials. Second, the few guanidines which we prepared turned out
to be insoluble in toluene (in the hydrobromide form) at the level
of 0.1 M, which we considered to be a rough lower limit for practical
consideration. Solubility is observed at very high guanidine salt
concentration (above about 60-70%), but the solutions are quite
viscous, and of uncertain value in the systems of interest. We have
since found that these materials are soluble in toluene containing
various polar additives (e.g., alcohols). Conceivably, the nitrate/
chloride selectivity could be greater than for amidines, but we
have no data on this point.
Basis of the Experimental Methods
Apparent Basicity of the Amidines. One of the critical requirements
for the extractant in the systems under study is that it should be
a sufficiently strong base to exist as the salt form in contact with
neutral aqueous solutions. We have used the apparent acid dissociation
constant of the amine salt as a convenient measure of the relative
acid strength of the protonated amidine. This quantity is defined by
equation 6. The term "apparent" is applied here because the thermo-
dynamic activities of the free base and the amidine salt in the organic
phase are not known. We have, in fact, shown that in the analogous
aliphatic amine systems, the amine salt is extensively aggregated (34),
so that the simple buffer equation defining K is not obeyed; that is,
K is not a constant. Nevertheless, the apparent acid constant
is&a useful parameter for expressing the relative behavior of the
various compounds and their sensitivity to certain variables of interest.
The physical significance of K is most easily appreciated by
converting equation 6 to the logarithmic form
(BHC1) - + log (Cl") (15)
pK = pH + log
(B)
From this equation it can be seen that the pK is the pH of the aqueous
phase when the organic phase contains equal concentrations of the
salt and free base form, and when the aqueous phase is 1 molar
-23-
-------
Table IV
I
ts)
Properties of Guanidines
Notebook Approx.
Compound Reference Yield
N-(Primene 8 1R)-N' , N" di(o-tolyl) GW-48-68:41 25%
guanidine, hydrochloride
N, N1 dioctyl-N" butylguanidine, GW-48-68:90 -
hydrobromide
N, N',N" Trioctyl guanidine, GW-48-68:93 -
hydrobromide
N-(2-ethylhexyl) -N, N"-dioctyl GW-48-68:94 -
guanidine, hydrobromide
N,N',N"-tris(2-ethylhexyl) GW-48-68:105 75%
guanidine, hydrobromide
N(Phenethyl)-N',N"-di(2-ethyl- GW- 1-70:59 75%
hexyl)guanidine, hydrobromide
Halide,
Found
Calc
_
Found
Calc
Found
Calc
Found
Calc
—
C
74.
73.
_
63.
63.
63.
63.
63.
63.
_
9
4
7
0
9
0
3
0
1
1
1
1
1
1
H
9.
9.
_
1.
1.
1.
1.
1.
1.
_
1
7
8
2
5
2
3
2
N
8.
9.
_
8.
8.
8.
8.
9.
8.
—
1
2
8
9
8
9
0
9
X
6.7
7.7
_
16. 1
16.9
16.3
16.9
16. 1
16.9
„
Equivalent
Weight
Found
Calc
Found
Calc
Found
Calc
Found
Calc
Found
Calc
526
458
_
497
476
490
476
496
476
498
468
-------
(more accurately, unit activity) in the counter anion. Because our
interest lay in the potential application of these systems to waste
waters, we chose to make measurements in a standard system containing
0.010 M sodium chloride. This concentration, which is equivalent
to 585 ppm of sodium chloride, is more typical of the dissolved solids
concentration in municipal sewage. By making the appropriate
substitutions in equation 16 we can still determine the pK for the
amidine system under study.
Ion Selectivity. Removal of nitrate from waste waters by the
proposed process is based on exchange of nitrate in the water for
some other ion in the organic phase. Since several other ionic
species will be present in most waste water systems, and since
nitrate will be one of the minor species, it is important, obviously,
that the extractant system have a relatively high affinity for nitrate
ion as compared with other ionic species. Based on experience with
analogous aliphatic amine extraction systems (32, 33) we expected
the major interference to come from chloride ion, and our screening
experiments were based on determining the selectivity for nitrate
over chloride ion. The selectivity constant was defined in equation 3,
and was the parameter determined in screening for this property.
Other major ionic constituents of waste water include both bicarbonate
and sulfate ions, and selectivites involving these species were
determined in one case.
Soluble Loss of Extractant to Aqueous Phase. A major concern
in extraction systems is the loss of the extractant to the aqueous
phase. In the type of exchange systems of interest here we expect
the soluble loss to consist mainly of the salt form of the extractant,
due to its greater polarity. The distribution of extractant between
organic and aqueous phase will presumably be governed by the
equilibrium
(BHC1)
BH + Cl ^ BHC1 ; K = (BH+) (Cl~) (16)
- - -Mm S
In other words, the loss of the amidinium ion should be dependent
upon the concentrations of anions in the aqueous phase, higher con-
centrations tending to drive the organic cation back into the organic
phase. Data were obtained with the idea of determining the equilibrium
constant for equation 16 as a means of comparing different compounds
and of predicting losses under various conditions.
-25-
-------
Experimental Methods
In addition to the compounds synthesized in this project, two commercial
aliphatic amines were included in the physical chemistry study.
Primene JM-T is a highly branched primary amine, in which the
amino group is attached to a tertiary carbon atom. Its equivalent
weight indicates it to contain about 21 carbon atoms. Amberlite LA-2 is
a secondary amine obtained by attaching a dodecyl group to the nitrogen
atom of a lower molecular weight analog of JM-T. Its equivalent
weight indicates a total of about 26 carbon atoms.
Three hydrocarbon diluents were used in the studies. The toluene
was a standard material. Chevron 3 is a high boiling aromatic
hydrocarbon diluent obtained from the Standard Oil Company. The
boiling range midpoint is about 188°C and is stated to contain 100%
aromatic compounds. Cyclosol 73 is a product of Shell Oil Company
with a boiling range midpoint of 218°C, and is stated to contain 74%
aromatics, 10% naphthenes, and 16% aliphatics.
All experiments were carried out at ambient (about 22+^ 2 °C)
temperature. Aside from certain special procedures, described
below, analytical techniques involved mainly conventional procedures.
Water was determined by the Karl Fischer titration. Chloride was
determined by titration with silver nitrate for higher levels, and by
coulometric titration for lower ones. Titrations of organic phases
for either chloride or for equivalent weight were done potentiometrically
in an acetone/water medium.
Apparent pK of Extractants. A 0. 5 M solution of the amidine was
first titrated^in acetone-water solution to determine its exact con-
centration. Dilutions were made to four additional concentrations.
These solutions were shaken 30 minutes with twice their volumes of
0. 010 M aqueous NaCl solutions, each of which contained a known
amount of hydrochloric acid. In most experiments this was equal to
half of the total amidine, thus producing a system containing equal
concentrations of free base and salt form of the amidine. In some
experiments variable quantities of acid were added to achieve other
free base/salt ratios. After equilibration of the phases, pH was
read directly in the aqueous system, and the phases were centrifuged,
if necessary, and separated. Total aqueous chloride was determined
by coulometric titration, and the other quantities were calculated
by difference from initial values.
-26-
-------
Soluble Loss of Extractants. A 0. 1 M solution of the amidine hydro-
chloride was shaken 30 minutes with 0.010 M NaCl solution. The
aqueous phase was analyzed for amidine, using the picrate method.
This method is based on the extraction of the picrate salt of an alkyl-
ammonium cation into chloroform, followed by determining the absorb -
ance of the yellow picrate ion in the chloroform. The aqueous phase
is buffered at about pH 3 in order to ensure the existence of the amine
component in its cationic form. The procedure is described in
detail in AppendixII.
At least two successive contacts of the organic phase with 0. 010 M
sodium chloride were made to ensure that successive losses were
equal. This procedure is often necessary when dealing with impure
materials, since an impurity which is more soluble in water than the
principal compound would appear in initial contacts, and give a false
result. Successive contacts, however, can be used to wash such
impurities out. When successive contacts give the same or nearly the
same results, it can be assumed that the impurity has been eliminated,
and the observed values of loss can be identified with the principal
compound. Following this procedure, portions of the 0. 1 M amidine
salt solution were shaken with 0. 03, 0. 1, and 0. 3 M sodium chloride,
and the soluble losses were determined.
Loss of the hydrocarbon diluent was determined by shaking the straight
diluent with several successive portions of 0.010 M sodium chloride,
separating and either filtering or centrifuging the aqueous phase.
The soluble diluent was determined in one method by extracting the
aqueous phase with 1/25 of its volume of carbon disulfide and determining
the diluent content by gas-liquid chromatography. A second method in-
volved back extraction with cyclohexane instead of carbon disulfide,
followed by determination of the ultraviolet absorbance of the cyclohexane
solution.
Ion Selectivity of Amidines. A 0. 25 M solution of the free amidine base
was titrated to determine the exact concentration. Portions of the
solution were converted separately to the chloride and nitrate or other
oalt by shaking with the corresponding aqueous acid. These salt
solutions were mixed in varying proportions and shaken 30 minutes
with an equal volume of 0.010 M sodium chloride. After separation,
the organic phases were stripped with 1 M sodium hydroxide for
analysis. Both the equilibrium aqueous phase and the strip solution
were analyzed for nitrate by two methods. Chloride was determined
only in the strip solution, by titration with silver nitrate.
-27-
-------
Nitrate was determined by the Hach method and the modified brucine
method (47). The Hach method involved the use of a Hach nitrate
water testing kit, based on reduction of nitrate to nitrite, diazotization
and dye formation. Instead of the color comparator, we determined
absorbance with a standard spectrophotometer. Some difficulty
was occasionally encountered with the formation of cloudiness, which
of course tended to give slightly high results, and in some cases,
prevented use of this method.
The modified brucine method was also used, in order to have an
independent check on the nitrate values. We have in the past had
difficulty with this and similar methods using strong acid media,
which we attributed to reaction of traces of organic materials with
the nitrate in the strong acid medium. This phenomenon produces
low results. While the Hach and brucine methods generally agreed
fairly well, deviations were generally in the directions expected.
Determination of nitrate/bicarbonate selectivity required somewhat
more care. It was necessary to carry out the experiment at a pH of
8 to 8. 5, the approximate pH of a bicarbonate solution, in order to
ensure the absence of free carbon dioxide, which would interfere.
This pH restruction necessitated using an amidine solution which was
only partly in the salt form. Bicarbonate was determined in the
equilibrium aqueous phase simply by titration with acid. Determination
of bicarbonate in the organic phase involved stripping with sodium
hydroxide, and titration with acid in the usual manner.
Results and Discussion
pH Dependence of the Extractant. Values of the pK of the amidines
are shown in Figure 1 as a function of the concentration of the amidine
salt in the organic phase. The value of pK varies considerably over
the range studied, and equation 6, in which K is assumed to be
constant, is clearly not obeyed in this region. Tabular data from
which this curve and other pK curves were obtained is given in
Appendix III.
We have shown in earlier studies that in aliphatic amine systems the
same phenomenon is observed, and is due to aggregation of the salt
form of the extractant (34). It is possible, however, by plotting the
apparent pK against the concentration of the salt form, to obtain
information about the aggregation process. If one assumes the existence
of only one aggregated species, in addition to the monomeric form of
the salt, the following equation can be derived (34):
10g (BHC1) Total <17>
-28-
-------
where n refers to the number of monomeric units in the aggregate,
and (BHC1) denotes the total concentration of all forms of the
hydrochlonde; i. e. , the analytical concentration of the salt. Thus,
a plot of (BHG1) against pK gives a line of slope . Besides
the assumption of a single aggregated species, the equation also
assumes no interaction between salt and free base, and also assumes
that activity coefficients in the organic phase are constant throughout
the range of data.
The curves in Figure 1 all exhibit a rather pronounced positive slope,
indicative of considerable aggregation. Some exhibit a definite
curvature, with the slope at the upper end being decidedly greater
than 1. A slope of 1, it should be noted, indicates an infinite aggregate.
We believe that this is probably due to the fact that concentrations
are so high in this region that the various assumptions enumerated
in the preceding paragraph are simply not obeyed. Considering the
limited range of data, we believe that aggregation is clearly occurring,
but that quantitative conclusions about the aggregation process are
unwarranted.
Another indication of the lack of strict adherence to the assumptions of
equation 6 is given by the data of Figure 2. We have shown here pK
data for two systems, Eh EhA and DoEhEhA, each taken in two
different ways. One curve of each pair was obtained from systems
where the base/salt ratio in the organic phase was kept at 1. 0; the
second of each pair was taken from systems where the total amidine
concentration was constant, the ratio of base/salt being varied. Although
there is general agreement between the two curves of each pair, there
is significant divergence at the lower ends of the curves, indicating
that the variable free base concentration exerts some effect on the
equilibrium.
In spite of the difficulties in explaining precisely the origin and
magnitude of the deviations from equation 6 we believe the important
point is that the pK of a particular compound is a function of the
concentration of the salt form of the extractant, and that this concentra-
tion must be considered in working with the physical chemistry of
these systems. A corollary is that the ability to vary the concentra-
tion of the extractant in a system provides an additional degree of
freedom in achieving the desired basicity in an extractant system.
Of considerably greater interest, however, is the effect of molecular
structure on the apparent basicity of the compounds. The earlier
work with aliphatic amines led to the observation that increasing the
steric hindrance around the nitrogen atom decreased the apparent
base strength of the amine. An increase in the steric hindrance
-29-
-------
11
10
8
lope =
0.04
0.07 0.1
0.2
0.3
(BHCI).M
Figure 1. Apparent pK of Amidines
3,
Ratio (B)/(BHC1) = 1. 0;
Aqueous phase, O.OlOMNaNO .
Slope of dotted line = 1. 0.
-30-
Eh EhA in toluene
V Eh EhA in Chevron 3
• Eh^OcA in "
• EhOcEhA in "
+ DoEhEhA in "
Q DoEhNdA in "
D Cy NoA in toluene
-------
10
8
0.04
I I
0.07 0.1 0.2
(BHCI), M
0.3
0.5
Figure 2. Apparent pK values of Amidines
Si
Aqueous phase, 0.010 M NaCl; diluent,, Chevron 3.
• Eh EhA, (B) + (BHCI) = 0. 5 M
0Eh EhA, (B)/(BHC1) = 1.0
ADoEhEhA, (B") + (BHCI) = 0. 48 M
ADoEhEhA, (B~)/(BHC1) =1.0
-31-
-------
increases the difficulty of forming a. stable aggregate involving the
highly polar parts of the molecule, with the result that the salt form
is destabilized. This is equivalent to saying that the compound behaves
as a weaker base.
The same phenomenon appears to exist in the amidine series. For
example, the compound Eh EhA, in which all chains are branched,
is a weaker base (stronger acid, or lower pK ) than either EhOcEhA
or Eh OcA, both of which have one straight chain the molecule. The
compound DoEhNdA, which has a tertiary carbon alpha to the carbon
containing the nitrogen atoms, is a weaker base than any of the others,
some of which have secondary carbons, and one of which has a primary
carbon. Of some interest is the fact that the effect of complete
substitution on the alpha carbon atom, evident in the compound
DoEhNdA, is far more important than variations in the structure of
the substituents on the nitrogen atoms.
The effect of changing the diluent from toluene to Chevron 3 is shown
for the Eh EhA system, and is seen to be small. Thus, the use of
aromatic hydrocarbon diluents of various types should not affect
the chemistry greatly.
From . a practical standpoint, the important feature of this data is
that a considerable flexibility exists in selecting a system which
exhibits a desired pK . Variations in the structure of the amidines, as
we have shown, are not difficult to obtain, and provide a simple way
of producing the desired behavior.
Equation 6 shows the involvement of the counter anion in the extraction
reaction, and predicts that the value of pK should depend upon the
anion. In Figure 3 we have shown data comparing the chloride and
nitrate forms of the compound Eh EhA. The two curves for -0.919 M aqueous
salt concentrations show that the nitrate curve runs roughly parallel
to the chloride curve, but is displaced upward by about 1. 6 units.
The amidine thus behaves as a stronger base toward nitric acid than
toward hydrochloric acid, or; to put it another way, the amidinium
ion in the organic phase has a greater affinity for the nitrate ion than
the chloride ion. The ion selectivity, S, can be derived from this
data by writing equation (6) in both the nitrate and chloride forms, and
then dividing one by the other. We get
- NO
(NO ") (Cl~) K 3
(NO -) (Cl )
-5 IS.
(18)
a
-32-
-------
which is simply equation 3 by which we defined the selectivity. From
this we can get
Cl
from which S is found to be about 40. Again, this equation makes
certain assumptions, mainly that there is no interaction between the
organic nitrate and chloride salts. Experimental determination of
the selectivity by direct exchange experiments gives somewhat lower
values, as we shall see.
The second pair of curves in Figure 3 was obtained with an aqueous
phase composed of ammonia and ammonium nitrate to simulate the
type of solution in a stripping system. The raw data are shown, but
because activity coefficients at 1. 0 M aqueous salt solutions are
significant, the data were also corrected by applying the activity
coefficients for ammonium nitrate to nitric acid in equation 6. This
is not a strictly proper use of activity coefficients, since the activity
coefficient of nitric acid in ammonium nitrate will be somewhat
different from the coefficient for ammonium nitrate itself. The mixed
activity coefficients were not available, and the resulting agreement
of the two nitrate system curves is to some extent coincidental.
The suitability of these systems to waste water processing can be
examined by first remembering that most natural waters, as well
as municipal sewage, will exhibit a pH in the range of 7 to 8, which
'is maintained by the action of the carbon dioxide-bicarbonate buffer
system. The pH of such systems can be raised somewhat by aeration
to remove carbon dioxide, which will result in a pH approaching
that of a bicarbonate solution, about 8. 0 to 8. 5. Similarly, the pH
can be readily lowered by carbonation. The behavior of two of the
extractant systems shown in the previous figures are replotted
'a Figure 4 in the form of titration curves; i. e. , pH is plotted as
a function of the ratio of salt and base forms of the extractant. We
have also shown similar curves obtained with Primene JM-T, a primary
amine, and Amberlite L/A-2, a secondary amine. As we indicated
earlier, these aliphatic amines behave as relatively w.eak bases in
liquid-liquid systems, and under the conditions shown are almost
entirely in the free base form at pH 7. In contrast, the two amidines,
-33-
-------
0
*
a
I I—i I I i I
0.01
0.05 0.1
(BHX), M
Figure 3. pK for Eh EhA in Chevron 3.
a Z
aqueous phase, 0. 010 M NaCl; (B~J7(BHC1) =1.0
aqueous phase, 0. 010 M NaNO ; (Bj/(BHCl) =1.0
aqueous phase, 1. 0 M ammonium (hydroxide + nitrate);
organic phase, 0. 5 M Eh EhA in Chevron 3.
same, but data corrected for aqueous activity effects.
-34-
-------
8
0 .2 .4 .6 .8
Fraction of Compound in Acid Form
Figure 4. Effect of pH on Extractant Form.
Aqueous phase, 0. 01 M NaCl; diluent, Chevron 3.
00. 86 M DoEhEhA
A 0.49 M DoEhEhA
• 0. 50 M EhEhA
1. 12 M Amberlite LA-2
A 1. 19 M Primene JM-T
Curved line, calculated for
0. 005 M NaHCC>
-35-
-------
Eh EhA and DoEhEhA, are mainly in the hydrochloride form at
pH i, with the latter compound being about 85% in this form at its
higher concentration. Since these data were obtained at O.OJOM
aqueous sodium chloride, this is approximately the behavior we
can expect in municipal sewage of this chloride content.
While this is probably borderline from the standpoint of satisfactory
operation we expect that under actual conditions the titration curves
would lie slightly higher in the plot, because of the greater preference
of the extractant for nitrate. This can be seen by rearranging
equation 6 to give
(B)
(BHXT + pKa + 10g (X) (20)
where X refers to the anion under consideration. If we examine the
position of any point on the titration curves of Figure 4, for example
where JB)~= (BHX), then
PHl/2= PKa + log (x ) (21)
and for any variations in pK or log (X~),
cl
A pH1/? ApKa + Alog(X") (22)
where pH , is the pH at the midpoint of the titration curve. Thus
the location of the curve depends not only on the pK of the extractant,
but also the aqueous concentration of the extracting anion. The data
of Figures 1 and 2 were taken at 0. 01 M chloride, and refer to the
condition of the extractant in the presence of chloride ion only. If
we consider a waste water with 0. 001 M nitrate ion (about 62 mg nitrate/1. ),
and note that Figure 3 indicates a higher pK for nitrate by about 1. 6
log units, we have
-36-
-------
Thus, we expect the nitrate titration curve to lie about 0. 6 pH unit
higher than the chloride curve. The value of 1. 6 for A pK implies,
as we discussed earlier, a selectivity for nitrate over chloride of
about 40. In actual exchange experiments values closer to 20 have
been observed, so that the A pH1 expected in an operating system may
be less than 0. 6, perhaps only half of this figure.
Furthermore, the variation available in the apparent pK of these
materials, illustrated by the data in Figure 1, makes it clear that
a compound with a higher pK could be easily prepared. We chose
to concentrate our studies on the two compounds Eh EhA and DoEhEhA
because a compromise may be necessary to allow stripping of the
system with ammonia, and because carbonation or a relatively
small amount of mineral acid could be used to reduce the pH to the
necessary point for extraction.
In the stripping step, we must determine the pH required to convert
the amidine into the free base form, in the presence of high nitrate
concentrations. The pK for nitrate systems is greater than for
chloride, and the titration curve for a nitrate system will lie higher
in the field than the chloride curves shown in Figure 4. In addition,
we wish to obtain a relatively high concentration of nitrate, and the
dependency dictated by equation 21 shifts the titration curve even
further upward. A single set of experiments was done with Eh EhA
and an aqueous phase of 1. 0 M ammonium hydroxide, and the data
are shown in Figure 5. Under these conditions, the titration curve
for the Eh EhA lies at about the same point as the curve for the
ammonium hydroxide/ammonium ion system. In other words,
the strip equilibrium
BHNO + NH4OH ^ NH4 + NO3 + B + H2O (24)
is a measurable one. This is also shown by examination of the
equilibrium constant,
(B) (NO.-) (NH ) K K
= K = „ — = 1.8xl(T K^ (25)
s
(BHNO ) (NH4OH) w
where K is the ionization constant of ammonium hydroxide, 1.8 x 10 ,
and K is the ion product of water, 1. 0 x 10 . From Figure 3
w
-37-
-------
11
10
8
0.2 0.4 0.6 0.8
Fraction of Compound in Acid Form
Figure 5. Effect of pH on Extractant Form.
Aqueous phase, 1. 0 M ammonium (hydroxide + nitrate);
Organic phase, 0. 5 M Eh EhA in Chevron 3.
| ammonium hydroxide/ammonium nitrate buffer system.
system
-38-
-------
we saw that pK for Eh EhA in the nitrate system varied from about
9 to 10. 5 in the region studied. K thus varies from about 1. 8 to
0. 05, depending upon the ratio of free base/salt in the organic phase.
From this equation, we can also readily predict that for stronger
ammonia solutions (which would produce stronger nitrate solutions
at equilibrium) the Eh EhA curve would lie still higher.
L*
Thus, in the presence of nitrate ion at concentrations of 1. 0 M or so,
it is not possible to completely strip the nitrate from the extractant
in a single contact. It is, of course , possible to do so by operating
the strip counter currently, wherein the entering ammonium hydroxide
solution, containing no nitrate ion, contacts the nearly completely
stripped extractant. In this way, a reaction with an equilibrium
constant of unity can readily by pushed to completion.
We have not made a more detailed analysis of the ammonia strip
because of the relatively small amount of data for this system.
Ammonia was used in one of the column runs, however, and was
found to be effective; the complete elution of the nitrate required
considerable ammonia solution. It can be predicted that the use
of a stronger alkali, such as sodium hydroxide or calcium hydroxide,
should allow complete stripping to be accomplished in the presence
of much higher nitrate concentrations. Time did not permit investiga-
tion of these systems, although these may be preferable to the use of
ammonia.
Ionic Selectivity of Extractants. The data for nitrate/chloride
selectivity are given in Table V for several of the amidines. Because
of the disagreement between the two nitrate analytical methods, the
precise values are subject to some uncertainty. The values generally
fall in the range of 20-30, with no clear dependence upon either the
ion composition of the organic phase, the diluent, or the particular
extractant under study. The selectivity constant was determined for
DoEhEhA at two extractant concentrations, 0. 25 M and 0. 89 M. Due to the
variation between the two methods there appears to be no significant
effect of the extractant concentration*
Selectivities for nitrate over sulfate, and for nitrate over bicarbonate,
were determined only for the compound Eh EhA. The data are given
in Tables VI and VII. The value of about 390 for nitrate/bicarbonate
clearly indicates a high preference of the extractant cation for the
nitrate. The value for sulfate of about 10 , of course, is also quite
high,but is not directly comparable, due to the more complex divalent/
monovalent exchange reaction.
In order to study carbon dioxide, a contact was made between 0. 25 M
Eh EhA (nitrate salt form) in Chevron 3 and an aqueous phase
containing 0.043 M NaHCO , 0.051 M NaNO , and 0.051 M CO .
-39-
-------
Table V
Nitrate /Chloride Selectivity of Amidines
organic phase: 0. 25 amidine in Chevron 3, except as noted
aqueous phase: 0. 01 M mixed sodium salts
Concentrations, M A_,i.,H,.ai
Compound
Eh EhA
2
DoEhEhA
DoEhEhA;
DoEhEhA;
Eh2OcA
EhOcEhA
(BHNOJ
0.224
0. 138
0. 058
0. 198
0. 124
0.050
(amidine) = 0.
0. 713
0.664
0.448
0.418
0. 179
0. 169
(Cyclosol 73)
0. 197
0. 124
0. 190
0. 119
0. 048
0. 198
0. 124
0. 050
(BHC1)
0.053
0. 130
0.206
0. 047
0. 121
0. 196
89 M
0. 175
0.433
0.694
0. 045
0. 118
0. 048
0. 117
0. 185
0. 049
0. 120
0. 193
(NO/)
0. 00165
0. 00047
0. 00013
0. 0025
0. 0023
0. 00072
0. 00058
0. 00023
0.00011
0. 0023
0. 0021
0. 00086
0. 00073
0. 00027
0. 00020
0. 0036
0.0031
0. 00099
0. 00078
0. 0017
0. 00175
0.00049
0. 00046
0. 00014
0. 0019
0. 00212
0. 00062
0. 00044
0.00018
0. 000087
(Cl )
0. 0083
0. 0089
0. 0095
0. 0127
0.0137
0. 0136
0.0116
0. 0146
0. 0153
0. 0151
0. 0170
0. 0090
0. 0104
0. 0117
0. 0108
0. 0134
0. 0136
SN°3
Cl
21. 3
22. 2
21.5
21.4
23. 3
19.5
24.2
15.6
31.5
20.5
21. 0
17.6
19.3
14.6
18. 6
18.4
21.3
18.0
22.9
20. 9
20.3
21.6
23.0
21. 7
27.6
22. 9
20.5
22.4
31.5
19.6
40.5
Method*
(B)
(B)
(B)
(H)
(B)
(H)
(B)
(H)
(B)
(H)
(B)
(H)
(B)
(H)
(B)
(H)
(B)
(H)
(B)
(H)
(B)
(H)
(B)
(H)
(B)
(H)
(B)
(H)
(B)
(H)
(B)
*H - Hach method; B = Brucine method
-40-
-------
Table VI
Nitrate/Sulfate Selectivity of Amidines
Organic Phase: 0. 25 M Eh EhA in Chevron 3;
L*
Aqueous Phase: 0. 020 N Mixed Sodium Salts
Concentrations, Normal
— NO
(BHNO ) (B H SO ) (NO -) (SO =) log S 3
32243 so4
0.256 0.00048 0.00092 0.020 6.51
0. 255
0.256
0. 0016
0. 0008
0. 0019
0. 0012
0. 058
0. 099
5.81
6.74
-41-
-------
Table VII
Nitrate/Bicarbonate Selectivity of Amidines
Initial Organic Phase: 0. 12 M Zh EhA-HNO , 0. 24 M Eh EhA
Ci J £t
in Chevron 3. Initial Aqueous Phase,
0. 10 M NaHCC>
Equilibrium Concentrations, M
NO
(BHN03) (BH2C03) (N03~) (HCO^) S
0.12 0.0033 0.0086 0.093 393
-42-
-------
Phases were analyzed as before, and slightly more total carbon
(CO + HCO ) was found in the organic phase than the aqueous.
Since bicarbonate extraction should be quite small, the amount can
be considered essentially all free carbon dioxide, and the distribution
coefficient of this species is then 1. 6. It is more realistic, we believe,
to express this as a simple distribution coefficient instead of an
exchange constant, since no ionic reaction is involved. The value
of 1. 6 is probably reasonable, since carbon dioxide is somewhat
more soluble in most organic solvents than in water.
We can use the data reported above to predict the behavior of amidine
systems in the treatment of a waste water. A typical composition
was discussed in the proposal for this study, and is reproduced in
Table VIII. From the mole ratios of ions in the water, and the
selectivity constants determined above, we have calculated the
expected composition of the organic phase after equilibration with
the water. These values are also given in the table. It should be
noted that the calculation for sulfate involves the organic nitrate
concentration, because of the character of the divalent/monovalent
exchange equilibrium. A value of 0. 5 M nitrate ion was assumed in
the organic phase.
The results clearly bear our the existence of chloride as the only
significant interference. For this waste water, we expect the organic
phase, and ultimately the nitrate waste, to consist of 88% nitrate ion
and 12% chloride. For waters of higher total dissolved solids, of
course, the chloride can also be expected to be higher, with a resulting
higher chloride content of the nitrate containing waste.
Other ions which maybe present in waste water include nitrite, phosphates,
silicate, and to a minor extent, borate. The ionization constant of
boric acid, 6. 4. x 10 , and that of silicic acid, about 2x10 ,
suggest that little extraction of either element will occur, since both
species will be present in most waste water as the undissociated acid.
The possibility exists that both phosphate and nitrite would be picked
up to some extent by the extractant, but this was not investigated.
One possible coextractant with nitrate ion is carbon dioxide. While
it is not an ionic salt, it is soluble in most organic solvents, and the
distribution coefficient of 1. 6 which we reported above for a typical
extractant system suggests that coextraction would certainly occur.
On the other hand, the amount would probably be small. The titration
curve for bicarbonate, shown in Figure 4, indicates that most dissolved
carbon would probably be in the form of bicarbonate, varying from
-43-
-------
Table VIII
Expected Extractant Composition from Treatment
of Typical Waste Water3-
Species Typical Waste Water
mg/1 mole ratio
X"/N03
Extractant Composition, for 0. 5 M
mole ratio
X-/NO,-
Eh0EhA
fraction of
total ions
Chloride 143
Nitrate 89
Bicarbonate 296
Sulfate 84
Silicate 43
Phosphate 25
Sodium 124
Potassium 12
Ammonium b
Calcium 66
Magnesium 19
b
2. 8
1. 0
2.4
0.6
0.4
0.2
0. 14
1. 0
0. 008
0. 0006°
0. 12
0. 88
0. 007
0.005
''Summary report on Advanced Waste Treatment, WP-20-A WTR-19.
Fed. Water Poll. Control Admin. , 1968, p. 46.
Distribution of nitrogen between ammonium and nitrate varies;
assumed to be all nitrate here.
Assumes 0. 5 M nitrate in organic phase.
-44-
-------
4)
C
E
<
0.01
0.03 0.1
(NaCD, M
Figure 6. Soluble Loss of Amidines
Organic phase, 0. 10 M hydrochloride salt of amidine.
^^ Cy NoA in toluene A EhOcEhA in Chevron 3
D Eh EhA in toluene f Eh OcA in Chevron 3
• Eh EhA in Chevron 3 ^ DoEhNdA in Chevron 3
-45-
-------
about 70% of the total at pH 7. 0 to 90% at pH 7. 7. With waste waters
averaging about 0. 005 M in bicarbonate/carbon dioxide, we would
not expect more than about 0. 003 M carbon dioxide in the organic phase
at equilibrium. For extractant solutions in the range of 0. 5 M nitrate
this represents a relatively insignificant coextractant. We anticipate
this should not complicate the extraction and stripping process, and
the absorbed carbon dioxide would be readily stripped out with the
alkaline stripping reagent.
Soluble Loss of Extractants. The dependence of soluble loss of amidines
on aqueous sodium chloride concentration is shown in Figure 6.
Data from which this curve was plotted are given in Appendix IV.
We have used a log-log plot, since equation 6 predicts that such a plot
should produce a straight line of unit negative slope. With some
conspicuous exceptions, this is the general shape of the curves, and
the results support the simple model of a dissociating ion pair,
which is the basis for equation 6. One additional point, not shown in
the figure, was obtained for the compound DoEhEhA at 0. 010 M
sodium chloride. The aqueous amidine concentration was found to be
about 8x10 M, which would fall just slightly below the lower
boundary of the figure.
We cannot explain the tendency of the curves to flatten out at the
higher salt concentrations. One possibility is the presence of some
impurity in the amidine, the distribution of which is not affected by
the sodium chloride, and which is masked by the higher amidine
losses at lower sodium chloride levels. Another possibility is that
at the higher sodium chloride levels the system simply does not obey
equation 6. Since the organic phase remains essentially constant in
composition during the series of experiments, any failures to obey the
equation are likely the result of some phenomenon occurring in the
aqueous phase. Either micelle formation or ion association effects
might be the source, but we have not attempted to pin this down any
further.
The important feature of the data in this figure is the fairly clear
separation of curves from amidines of different molecular weights.
The Cy NoA, with 21 carbon atoms, is clearly more soluble than
any of the 24 carbon compounds, by factors of about 10 to 30. The
cyclohexyl analog was actually run in toluene, which, judging by the
comparison between toluene and Chevron 3 solutions of Eh EhA, lowers
the distribution tendency toward water by a factor of about 2. Thus,
the difference between the 21 carbon compound and the 24 carbon
compounds is probably more nearly a factor of about 20 to 60.
-46-
-------
An exception to this correlation is the compound DoEhNdA, which has
30 carbon atoms, but which is only slightly less soluble than the
24 carbon compounds. While this result may possibly be due to some
experimental difficulty, we believe that it may very well be a real
effect. We encountered a similar problem with the compound
(Bu2) EhEhA, the hydrochloride of which was not soluble to the
extent of 0. 1 M in Chevron 3. It was soluble in 0. 010 M sodium
chloride to the extent of about 0. 04 M. We did not pur-sue this
observation, but we are led to speculate that the rather high steric
hindrance provided by the dibutylamine group may greatly reduce
the ability of the hydrochloride salt to aggregate in the organic phase.
If so, this would presumably increase the distribution tendency in the
direction of the aqueous phase. The significance of this observation
lies in the similarity of this compound to the compound DoEhNdA,
which is also rather hindered as a result of the tertiary alpha carbon
atom. This is clearly reflected in the relatively low apparent basicity,
shown in Figure 1. Thus the relatively high tendency to distribute
into the aqueous phase may also be a result of the high steric hindrance.
Some additional support for this hypothesis is to be found in the
comparison between the curves for Eh EhA and EhOcEhA, where the
latter compound is both less hindered and has a lower distribution
tendency toward water.
The effect of concentration of amidine salt in the diluent on the soluble
loss was determined for the compound Eh EhA, and the data are shown
in Figure 7. The data show that little increase in soluble loss is
observed with increasing extractant concentration above 0.1 M.
Equation 16 predicts, for constant aqueous chloride concentration,
a linear relationship between soluble loss and extractant concentration.
However, the increase in aggregation of the organic salt with increasing
concentration would be expected to stabilize the organic salt and reduce
the tendency to distribute into the aqueous phase. This is, qualitatively,
the observed effect.
The impact of soluble loss on the processing of a waste water occurs
in two ways: as an economic problem, and as a pollution problem.
The economic problem can be relatively easily estimated from the
following equation:
Economic Cost (//M gal) = 0. 83 x (soluble loss, mg/l)x (unit cost of
compound, $/lb) (26)
Using a value of about j>2 per pound of extractant (see next section)
and the value of 8 x 10 M as the soluble loss of DoEhEhA (about
4 mg/1), the equation gives a cost of about 7^/M gal due to soluble
-47-
-------
C B H C I >, M
Figure 7. Soluble Loss of Eh EhA from Chevron 3 into 0. 010 M NaCl
-48-
-------
loss of extractant. This is a fairly high cost, and would very likely
be unacceptable in most situations. However, this can be reduced
fairly easily by switching to a higher molecular weight extractant.
The data on soluble loss indicated that for each four carbons added
to the extractant, a soluble loss reduction of something like 30-fold
could be expected. Thus, the use of a second dodecyl group instead
of a 2-ethylhexyl in the compound EhDoEhA should reduce the soluble
loss to a level of about 0. 1 mg/ 1. The cost of the compound would
probably remain about the same, so that the economic loss represented
by this value would be well below 0. 2£/M gal.
We have made no attempt to resolve the question of what is an adequately
low concentration of an organic compound in a waste stream. One
landmark of some use in considering'the question is the fact that
most secondary sewage effluents contain something of the order of
10-20 ppm of organic materials. While this can be further reduced
with tertiary methods such as activated charcoal adsorption, there
is no agreement that this is necessary for most situations.
Of more concern in this respect is the matter of biological activity.
While preliminary biological activity tests indicated no highly
significant activity, we did not examine biodegradability or toxicity to
marine organisms. It is known that amidines hydrolyze to amides,
and thence to carboxylic acids, producing also at each step an amine.
Biological testing of all of these materials would probably be required
in future development work^
Chemical Stability. Although this aspect of extractant characteristics
was not defined as a separate property in the proposal, it is nevertheless
an important feature of the extractant system. It is generally known
that amidines are susceptible to hydrolysis to the corresponding amide,
and ftiis reaction represents at least a potential source of difficulty in
the use of these materials. However, the stability experiments
showed clearly that, in contact with either 0. 1 M ammonium hydroxide
or 0. 1 M sodium hydroxide no observable hydrolysis occurred in
eight months. The following calculation will show that hydrolysis
"hould be a negligible factor in the operation of a water treatment
process.
A reasonable flow rate of water through a resin bed is around 2 gpm/
cu. ft. , and if we assume the bed is on the treatment step of the
cycle for only 50% of the time, then in eight months (the duration of
the stability experiment) one cubic foot of resin will la ve treated
about 400, 000 gallons of water. If we assume 40% voids in the bed,
30-35 pore volume in the bead, and a 50% solution of extractant, a
-49-
-------
cubic foot of bed will contain about 6 Ib. of extractant. Assuming also
a degradation of about 1% (about the limit of detection) in the eight
month experiment, the bed is found to have treated about 6. 5 million
gallons of water per pound of material degraded. At a cost of $2 per
pound, the loss amounts of 0.03£ per M gal of water. This is a
negligible cost, but assumes the extractant can be used until complete
degradation occurs. In practice the formation of the amide and amine
might complicate the addition of new amidine to maintain the extractant
strength, and a more realistic assumption might be that the system
is discarded after about 10% degradation by washing it out of the beads
with fresh diluent. This is equivalent to saying that only one-tenth
as much water will be treated as in the above calculation, leading to
a cost of about 0. 3$/M gal. Even this would be a relatively minor
cost, and in terms of the above calculation, would require at least
10 x 8 months, or about seven years, to occur.
Soluble Loss of the Diluent. We have assumed that the extractant system
absorbed into the polymer bead would consist of a solution of the
extraction in a diluent. The main function of the diluent is to reduce
the viscosity of the extractants, which in the salt form are extremely
viscous liquids or waxy solids. It would be expected that diffusional
processes would be relatively slow in such media, and the resultant
column performance would be poor. Because of the desire to maintain
soluble losses in the part-per-million range, the only classes of
diluents which can be considered are the hydrocarbons, which are
both relatively inexpensive and low in water solubility. While the
aliphatic hydrocarbons are the least soluble, for a given molecular
weight, they are also poorer solvents, and our experience with highly
polar solutes like the amine salts, is that they are poorly soluble in
the aliphatics, and in addition lead to higher solution viscosities.
We have rather arbitrarily used a high-boiling aromatic hydrocarbon,
Chevron 3, in most of the studies, since it combines rather well
a high boiling point, high aromatic character, low cost, and relatively
low solubility.
Some initial data on diluent solubility were obtained early in the study
by contacting diluents with water, filtering the aqueous phase to
clarify it, back extracting the water phase with carbon disulfide, and
analyzing the latter solution with gas-liquid chromatography.
Subsequent work has indicated that diluent is apparently lost from
the aqueous phase during the filtration, either by absorption on the
paper, or by evaporation. A subsequent experiment was carried out
using only centrifugation as the method for clarifying the aqueous
phase. The experiment involved successive contacts of Chevron 3 with
0.010 M sodium chloride. The GLC analysis technique was used, and
was checked by another back extraction of the aqueous phase with
-50-
-------
cyclohexane, followed by determination of the ultraviolet absorbance
of the aromatic content. The data are shown in Table IX. The sixth
contact is considerably out of line, and we assume some contamination
was involved. Both methods indicate the 5th and 7th contact to be
in the same concentration range, although the UV figure is considerably
higher than the GLC figure. We believe the explanation lies in the
presence in the Chevron 3 of some nonhydrocarbon impurity. The
GLC pattern showed four major components of the Chevron 3 in the
standard solution. The back extracts, however, showed comparable
amounts of an additional component, which was sufficiently delayed
in the scan to indicate a different type of material. Such an observa-
tion could readily arise from slight oxidation of the hydrocarbon to
an alcohol, phenol, ketone, etc. This type of compound would give
a UV response similar to the hydrocarbon, but in this case was not
included in the analysis in the GLC method. Because of the much
greater water solubility of this type of an impurity, a significant
response could be observed in the aqueous phase, even though the
impurity is not detectable in the original hydrocarbon.
In the following section, data will be reported for resin column effluents,
which agree reasonably well with the above data. Because of the
ability of the GLC method to identify the hydrocarbon components,
we believe it is a more reliable method for diluent loss determination,
and the 21 ppm figure probably represents the diluent loss reasonably
well.
Some further confirmation of this value can be obtained from examina-
tion of the boiling range of Chevron 3, which is given by the Standard
Oil Co. as 182°C. to 205°C. This range contains the boiling points
of most alkyl benzenes containing four aliphatic carbons in the
substituent groups. Solubility data for some hydrocarbon material(48)
indicate that such compounds should exhibit a water solubility in the
range of 15-20 ppm, quite close to the observed GLC value. Since heavier
diluents are available, it would probably be desirable to switch to such
a material, possibly even to an aliphatic diluent, the soluble loss for
which would be much lower still.
Water Extraction and Volume Change. The water extraction experi-
ments were carried out mainly as an additional means of characterizing
the organic extractant system. The data for water extraction by Eh EhA
are shown in Figure 8, and show that there is a roughly linear correla-
tion between amidine hydrochloride concentration and water concentra-
tion. The curves would intercept the y-axis at a concentration of
about 0. 03 M, which is approximately the solubility of water in toluene
alone. Starting from this point, the slopes of the curves indicate that
-51-
-------
Table IX
Comparison of Ultraviolet and GL.C Analyses
for Chevron 3 in Aqueous Phases
ppm Chevron 3
Aqueous Contact Number UV GL.C
5 36 21
6 78 100
7 29 21
-52-
-------
0.20
0.16
0.12
»
CN
0.08
0.04
0.2
0.4
0.6
0.8
(BHCD.M
Figure 8. Extraction of Water by Eh EhA in Toluene.
LA
Aqueous phase, O.QlOMNaCl
(B) + (BHC1) = 0. 86 M
(B) + (BHC1) = 0.48 M
-53-
-------
about 0. 2 mole of water is taken up for every mole of amidine hydro-
chloride. This can be compared with earlier data on aliphatic amine
hydrochlorides, for which the water extraction varied widely, from 0. 05
to 4 moles per mole of hydrochloride (34).
In a separate experiment, corresponding roughly to the uppermost
point on the 0.86 M curve, we attempted to measure volume change
associated with conversion of the amidine from the free base form
to the hydrochloride form. The change, if any, was less than 2%
by volume.
-54-
-------
ION EXCHANGE BEHAVIOR OF SOLID SUPPORTED EXTRACTANTS
A second aspect of this project involved a study of the feasibility of
using amidine extractants absorbed in a macroporous copolymer to
selectively remove nitrate from aqueous solutions. The extractant -
in-bead approach was chosen because it minimizes two problems often
associated with conventional liquid-liquid extraction processes;
namely, the loss of extractant by physical entrainment and the forma-
tion of emulsions that coalesce slowly. The following discussion
describes the incorporation of extractants into macroporous copolymers
and their ion exchange behavior in packed columns.
Incorporation of Extractants into Macroporous Copolymers
Solid Absorbent. The solid absorbent used in this study as a support
for the liquid extractant was a macroporous styrene-divinylbenzene
copolymer resin that was obtained from the resin synthesis group of
The Dow Chemical Company. This material is produced as an inter-
mediate in the manufacture of macroporous ion exchange resins,
however, it does not contain any ion exchange functionality. Similar
macroporous copolymers are also available from the Rohm and
Haas Company.
Macroporous resins differ from the more conventional gel resins
in that they possess a true pore structure that is a significant fraction
of the total resin volume. Gel resins, on the other hand, are essentially
homogenous crosslinked copolymers that do not have any significant
pore volume. Macroporous resins have been produced with total
porosities as high as 50%, and with average pore diameters ranging
from 50 to 1500 A° , depending on the particular polymerization pro-
cedure. Because the macroporous resins are highly crosslinked, they
exhibit very little swelling in either aqueous or organic solvents.
Consequently, their capacity to absorb organic extractants from an
external solution is determined primarily by the pore volume that
is accessible to the extractant. In contrast, the absorption capacity
of a gel resin for an organic extractant depends almost entirely on
the amount of resin swelling that is produced by the extractant solu-
tion. A macroporous resin was chosen for this study because it was
found to have about a two-fold greater absorption capacity for the
extractants of interest than did the gel resins. It was also felt that
an extractant in a macroporous resin would exhibit better ion ex-
change kinetics than it would in a gel resin, although this point was
not investigated.
-55-
-------
The macroporous resin was supplied as spherical beads with a size
range of 18 to 80 mesh, which were further screened and the 30 to 40 mesh
fraction was used in all column experiments. The resin was washed
with acetone and water to remove any residual monomers and vacuum
dried at 60°C. The copolymer contained a nominal 20% divinylbenzene
crosslinking which gave a rigid polymer bead. Total porosity of the
resin was 26%, and 95% of the pores were less than 350A in diameter.
Extractants. At the beginning of this project, the high molecular
weight amidines were not available in large enough quantities to
permit their use in resin experiments. Consequently, we used two
other similar extractants that are commercially available to develop
the resin loading procedure and to investigate some of the column
operating parameters. The two substitute extractants were Amberlite
LA-2, a secondary amine with an equivalent weight of 387, and Aliquat
336S, a quaternary ammonium chloride with an equivalent weight
of 442. Neither of these two compounds meet all the criteria stated
on page 9 for a suitable nitrate extractant. For example, Amberlite
LA-2 is too weakly basic to be useful at the normal pH of natural
waters, and Aliquat 336S cannot be stripped with alkali. Nevertheless,
they are similar to the amidine extractants in many other respects
and they served as useful substitutes for our initial investigations.
When the amidine DoEhEhA became available in quantity later in the
project, we used it exclusively in the resin studies.
Preparation and Properties of Extractant Loaded Resins. A standard
procedure for incorporating liquid extractants into macroporous
resins was developed. The extractant was dissolved in a high boiling
aromatic hydrocarbon, Chevron 3, and two volumes of this solution
were mixed with one volume of 30-40 mesh resin for at least 24 hours.
The interstitial solution was removed by gentle suction filtration and
the resin was rinsed several times with an aqueous solution of
composition: 1 M NaCl, 0. 04 M NaOH and 0. 05% Triton X-100.
The non-ionic surfactant, Triton X-100, was added to the rinse solu-
tion to facilitate removal of the surface adhering organic layer from
the resin particles. The number of rinses was kept to a minimum,
consistent with obtaining a relatively free-flowing resin while at
the same time, maintaining maximum extractant absorption.
Only moderate success was obtained with this procedure in producing
a truly free-flowing loaded resin. Generally, the loaded resins tended
to form aggregates when placed in water because of the hydrophobic
nature of the surface adhering organic layer. We anticipated that
this behavior might impede optimum column performance due to the
-56-
-------
difficulty of obtaining close packing of the resin particles. It was
found that la. truly free-flowing loaded resin could be obtained
if most of the Chevron 3 diluent was removed by evaporation; however,
we were reluctant to employ this technique because we felt that the
presence of the diluent in the resin was necessary to maintain a
homogeneous solution of the extractant in the pores. Without a diluent,
the extractant salts are very viscous materials and if any ion exchange
could occur in the absence of a diluent, it would probably be extremely
slow. We intended to investigate the ion exchange behavior of a
diluent-free loaded resin, but time did not permit. Consequently,
in all work reported here, the loaded resins contained both a diluent
(Chevron 3) and an extractant.
Small (26), in his work on gel liquid extraction, reported that organic
swollen gel beads did not aggregate when placed in water if the beads
were previously surface sulfonated. The thin surface shell of sul-
fonated copolymer provided the bead with a hydrophilic surface that
was easily wet by an external aqueous solution. We attempted to
apply this same approach to eliminate the aggregation problems ob-
served with the loaded macroporous resin. Several batches of macro-
p orous copolymer were partially sulfonated with 96% H SO at 90-100°C.
Reaction times varied from 2-10 minutes, and the extent of sulfona-
tion achieved varied from 0. 5 to 7% of the maximum amount possible
(about 5 meq. /g. ). These materials definitely showed less tendency
to aggregate in water when loaded with an extractant, however, the
extractant absorption capacity of even the least sulfonated resin
was about 17% less than a comparable unsulfonated macroporous
resin. Because of this rather significant reduction in absorption
capacity, we waived further investigation of this approach in hopes
that the aggregation of the unsulfonated resin would not adversely
affect its column behavior.
Extractant Absorption Capacities. The amount of extractant absorbed
by the macroporous copolymer was determined for each of the three
extractants: Amberlite LA-2, Aliquat 336S, and DoEhEhA. Some
typical results are shown in Table X. Throughout this report, the
extractant loaded resins are designated with-the following code:
MP = macroporous, A = DoEhEhA, LA = Amberlite LA-2, Q = Aliquat 336S.
Resins that differed only in the amount of extractant absorbed are
distinguished by a number following the letter code.
-57-
-------
Table X
Extractant Absorption Capacities of the Macroporous Copolymer
Concentration of Absorption Capacity
Extractant Solution meq. /g. meq. /g.
Resin Extractant (meq/ml) Dry Resin Loaded Resin
DoEhEhA 0.971 0.887 0.424
DoEhEhA 0.560 0.354 0.151
Amberlite LA-2 1.100 0.852 0.440
Aliquat 336S 0.725 0.445 0.203
Dry resin refers to copolymer beads only.
The extractant absorption capacities were determined by washing
a weighed sample of loaded resin several times with acetone to remove
the extractant, followed by potentiometric titration of the combined
acetone washings. Amberlite LA-2 and DoEhEhA were determined with
standard HC1 and Aliquat 336S (chloride salt) with standard AgNO .
The acetone-washed resin was vacuum dried and weighed.
It is evident from Table X that the amount of extractant absorbed by
the macroporous copolymer increased as the concentration of the
external extractant solution increased. Because of the high viscosity
of concentrated extractant solutions, we arbitrarily chose 50%
(about 1 N) as the maximum extractant concentration. Although more
concentrated solutions would probably have given greater resin load-
ings, it was felt that the rate of ion exchange would be substantially
reduced due to the high viscosity of the extractant solution in the resin
pores. Since the aqueous phase does not permeate an organic loaded
resin bead to any significant extent, ion exchange must occur primarily
at the aqueous-organic interface on the bead surface. In the case of
the DoEhEhA resins, this requires that amidinium chloride must
diffuse from the interior of the bead to its surface and amidinium
nitrate must diffuse in the opposite direction. A highly concentrated,
and therefore highly viscous, DoEhEhA solution in the bead pores
would restrict the rate of diffusion within the bead, which would result in
a low rate of ion exchange. This argument would be true regardless of
the nature of the diffusing species.
-58-
-------
Column Studies
Experimental Procedure. Three of the four resins shown in Table X
were investigated in the column experiments: MPA-1, MPA-2, and
MPQ. Each of these materials was prepared using the 30-40 mesh
copolymer. The column consisted of a 1/2" diameter by 23" high
glass tube with adjustable bed s.upport plungers on each end . A
dual syringe metering pump was used to pump solutions through the
column at some predetermined flow rate from 2. 4 to 10. 4 ml/min.
(1/2 to 2 gptn/ft ). Effluent from the column was collected and analyzed,
and breakthrough curves were constructed from these analyses. The
composition of the feed solution was 0.001 M KNO (62 mg. NO /I), 0.01 M
NaCl(355 mg. Cl/ 1.) and pH 6. 5
Packing the extractant loaded macroporous resins into the column
presented some difficulties because of their tendency to aggregate
and because they were less dense than water. The procedure that
was adopted was to slurry a known weight of loaded resin with a IN
NaCl solution in the column. A long glass rod was worked up and down
through the resin slurry to remove air bubbles, and the salt solution
was allowed to slowly drain from the column . When the floating
resin column began to accumulate against the lower bed support, it
was lightly tamped in place with the glass rod, working from the
bottom to the top of the column. The salt solution was never allowed
to drain below the top of the resin bed. A final compression of the
bed was accomplished by pressing the upper bed support plunger
against the top of the resin bed. Bed heights ranged from 36 to 47 cm. ,
and column volumes ranged from 45 to 60 ml.
After loading the column with MPA-1 and MPA-2, several bed
volumes of 0. 1 N HC1 were pumped through the column to convert
all the free-base DoEhEhA to the hydrochloride salt. The interstitial
acid solution was then removed with several more bed volumes of
0. 01 M Nad until the effluent was neutral. Since the MPQ resin was
already in the chloride form, it was preconditioned with 0. 01 M
NaCl only.
Analytical Methods. Nitrate in the column effluent was determined
spectrophotometrically at 210 and 220 m \L (49) after extraction of
interfering ultraviolet absorbers with spectroquality cyclohexane.
A S ml. sample of effluent was shaken for several minutes with a
like amount of cyclohexane and centrifuged. The absorbance of the
aqueous phase was determined using 1.0 or 0. 1 cm. quartz cells
and a Gary 14 spectrophotometer. When necessary, the effluent
solutions were diluted to give absorbance readings less than 1.0.
Available from Chromatronix, Inc. , Berkeley, California
-59-
-------
The molar absorbtivity for nitrate at 210 mjji was 8200; and at 220 m p.,
3750. The absorbance ratio, A210-A22Q' was consistently 2- 18 ± °' °3'
Any values that fell outside this range indicated the presence of
ultraviolet absorbers other than nitrate, and required that the analysis
be repeated.
In general, the cyclohexane extraction removed all interferences and
the nitrate analyses were simple and rapid. One difficulty was
encountered early in the work due to leaching of the plasticizer from
the Tygon tubing used with the column. The plasticizer was not
extracted by cyclohexane and it interfered with the nitrate absorbance.
This problem was remedied by replacing the Tygon tubing with poly-
ethylene and Teflon tubing.
Chloride was determined by coulometric titration using a silver anode.
Chevron 3 was determined spectrophotometrically in the cyclohexane
extract that was obtained from the nitrate analysis. Chevron 3 in
cyclohexane showed a sharp absorption maximum at 222 m|j. with an
extinction coefficient of 0. 0885 1. mg. cm
DoEhEhA was determined by the picrate extraction procedure outlined
in Appendix II.
Results of Column Studies
Nitrate breakthrough curves are shown in Figure 9 for three extractant
loaded macroporous resins: MPQ, MPA-1 and MPA-2. The ratio of
nitrate concentration in the column effluent to that in the feedwater,
C /C , is plotted against the number of column volumes of feedwater
treated. Each of these curves was obtained at a flow rate of 5. 0 ml/
min. The initial nitrate breakthrough from the MPQ resin occurred
after only 30 column volumes of feedwater were treated, but nitrate
extraction continued for more than 200 column volumes indicating
a very slow rate of ion exchange for this resin.
The effect of flow rate on nitrate absorption was investigated using
the MPA-2 resin and the results are shown in Figure 10. The
breakthrough curves obtained at 2. 4 and 5. 0 ml/min. are nearly
identical, and as expected, nitrate breakthrough occurred slightly
later at 2. 4 ml/min. than it did at 5. 0 ml/min. Surprisingly, however,
at 10.4 ml/min. the initial nitrate breakthrough occurred later and the
amount of nitrate absorbed by the resin was larger than for either of
the two lower flow rate runs. As the data in Table XI shows, about 66%
of the extractant in MPA-2 was converted to the nitrate form at a
flow rate of 10.4 ml/min; whereas, only 54% and 52% conversion
-60
-------
0.2 -
50
100 150
Column Volumes
200
250
Figure 9.
Nitrate Absorption by Extractant Loaded Macroporous Resins,
Feed Composition: 10"3 M KNO , 10"2 M NaCl
Flow Rates: 5. 0 ml/min
Column Volume = 60 ml; Capacity = 0. 092 meq/ml
Column Volume = 45 ml; Capacity = 0. 139 meq/ml
Column Volume = 51 ml; Capacity = 0. 252 meq/ml
-------
40 60 80
Column Volumes
Figure 10. Effect of Flow Rate and Column Packing on Nitrate Absorption
Resin: 36. 5 g MPA-2
Feed Composition: 10"3 M KNO 10"2 M NaCl
| Flow rate = 2.4 ml/min; Bed Volume = 60 ml
Flow rate = 5.0 ml/min; Bed Volume = 60 ml
Flow rate = 10.4 ml/min; Bed Volume = 52 ml
-62-
-------
was obtained at 2. 4 and 5. 0 ml/min, respectively. The relatively
low nitrate absorption at the lower flow rates was apparently due
to severe channeling within the resin bed. AH of the flow rate
experiments with MPA-2 were done with the same sample of resin
(36. 5 g). Between each exhaustion cycle, the absorbed nitrate was
eluted and the resin was regenerated with 0. 1 N HC1. The amount of
nitrate absorbed by the resin was calculated from a knowledge of the
total nitrate applied to the column and the amount in the effluent after
completion of an exhaustion cycle. During the exhaustion cycles
carried out at 2. 4 and 5. 0 ml/min, the resin volume was 60 ml;
whereas during the run at 10. 4 ml, the resin volume was only 52 ml,
a difference of 13%. This contraction of the resin bed was caused
by the removal of air bubbles that had inadvertently accumulated
in the upper third of the bed just prior to the high flow rate run. The
air bubbles were removed by stirring the resin with a long glass rod,
and when the resin bed was recompressed, its volume was only
52 ml. Since no resin was lost during this operation, the resin
particles were more closely packed in the contracted bed than they
were in the initial 60 ml bed. As a result, channeling was less
severe during the high flow rate experiment and greater nitrate absorp-
tion was therefore obtained. These experiments indicate the importance
of using a well packed resin bed to obtain optimum column performance.
Table XI
Nitrate Absorption Capacities of Macroporous Resins
Resin
MPQ
MPA-1
MPA-2
MPA-2
MPA-2
Resin
Vol. (ml)
45
51
60
60
52
Flow
Rate
(ml/min)
5.0
5.0
2.4
5.0
10.4
Nitrate Absorbed
NO,
Extractant in
Column (meq. ) (meq. )
(meq/meq) S
Extractant) Cl
6.25
12.85
5.51
5.51
5.51
4. 14
6.96
2. 98
2. 86
3.62
0. 662
0.542
0.541
0.519
0.657
19.6
11.8
11. 8
10.8
19.2
Selectivity coefficients, S
NO
were calculated for all resins
investigated in the column studies and they are also given in Table XI.
These coefficients were calculated by dividing the nitrate/chloride
ratio on the resin by the nitrate/chloride ratio in the feed solution.
Chloride on the resin was taken as the difference between the amount
of extractant on the resin and the amount of nitrate absorbed. The
nitrate selectivity of MPQ was 19. 6, which agrees well with the values
reported by Grinstead and Davis (33) for Aliquat 336S in a liquid-
-63-
-------
liquid system. Likewise, the selectivity of 19.2 obtained for MPA-2
in the compacted bed is very near the average of the selectivities
shown in Table V for DoEhEhA; again in a liquid-liquid system.
Thus it was established that the nitrate selectivity of an extractant
in a macroporous resin should be nearly identical to that in a liquid-
liquid system. It is apparent from Table XI that the selectivities
calculated for MPA-1 and for MPA-2 in the 60 ml. bed are considerably
lower than would be expected for DoEhEhA in a liquid-liquid system.
These low selectivities are undoubtedly the result of inefficient
utilization of the resin due to poor column packing, and further
emphasizes the necessity of preventing channeling within the column.
The loss of DoEhEhA from the resin during an exhaustion cycle
ranged from 1. 2 ppm at the beginning to 0. 2 ppm at the end of the
cycle. Chevron 3 losses ranged from a high of 40 ppm to a low of
10 ppm with an average loss of about 20 ppm. At this point the loss
of diluent appears to present more of a problem than the loss of
DoEhEhA.
Nitrate and chloride were eluted from exhausted DoEhEhA columns
with aqueous NaOH and NH OH. Figure 11 shows the results of
eluting MPA-2 (60 ml bed) with 0. 5 N NaOH, and Figure 12 shows the
corresponding results for MPA-2 (52 ml bed) using 0. 5 N NH4OH.
In each case the alkali flow rate was 2. 5 ml/min, and essentially
all of the nitrate was removed with four column volumes of alkali.
The data in Table XII indicates that reasonably good agreement was
obtained between the amount of nitrate absorbed and the amount
subsequently eluted. The elution curves illustrate that both NaOH
and NH OH are equally effective reagents for nitrate elution from
an exhausted column, however, the volume of eluate required to
remove all the nitrate was larger than desired. The concentration
of nitrate in four column volumes of eluate was only about 12. 5 times
that in the initial feedwater. Additional studies should be conducted
to determine the conditions necessary to increase this concentration
factor.
Table XII
Nitrate Elution from MPA-2 with NaOH and NH4OH
Resin Nitrate Nitrate
Resin Volume (ml) Reagent Absorbed (meq) Eluted (meq)
MPA-2 60 0.5NNaOH 2.98 2.88
MPA-2 52 0. 5 N NHXOH 3.62 3.83
4
-64-
-------
Column Volumes
Figure 11. Nitrate and Chloride Elution with 0. 5 N NaOH
Resin: MPA-2 (60 ml bed volume)
Flow rate: 2. 5 ml/min
-65-
-------
0.08 -
Column Volumes
Figure 12. Nitrate and Chloride Elution with 0. 5 N NH OH
Resin: MPA-2 (52 ml bed volume)
Flow rate: 2. 5 ml/min
-66-
-------
OUTLINE OF NITRATE REMOVAL PROCESS
Process Description
The goal of this study has been to examine a specific concept for a
nitrate selective material which could ultimately be used in a practical
process for waste water treatment. It is desirable, therefore, to
attempt some evaluation of the potential of the concept at this stage.
The extractant-in-bead approach to nitrate removal has been based
on eventual adaptation to a conventional fixed bed ion exchange
process. This formed the basis of the earlier cost estimate for
selective ion exchange (24). In its simplest form such a process
would appear schematically as shown in Figure 13. A column of
resin beads would be used to treat waste water until breakthrough
of nitrate in the effluent occurred. This would be followed by a
backwash, and an elution with an alkaline regenerant, which might be
either aqueous ammonia or sodium hydroxide, to produce a nitrate
solution. The column would finally be rinsed to eliminate contamina-
tion of the treated effluent from the following cycle by nitrate. The
final eluate could be disposed of directly, if possible, or could be
concentrated for disposal or use elsewhere.
Adaptation of Extractant-Loaded Beads to Column Operation
No specific criteria were set forth in the proposal for evaluating the
success of the column operation of the extractant-in-bead system.
However, one measure of the achievement is the degree to which
column operations approached the conditions chosen for the preliminary
cost estimates in the earlier Bureau of Reclamation contract. In
that study the estimated cost of treating San Luis agricultural drainage
water was about 6$/M gal, without any credit for the proposed product,
ammonium nitrate. In view of projected costs for other nitrate
removal processes, this appears to be at least a competitive value.
The major assumptions of that estimate regarding process variables
--ere as follows:
1. feed water flow rate of 10 gpm/ft
2. nitrate loading of about 0. 5 equivalents/liter of
resin, obtained with a nitrate/chloride selectivity
of 20, and a total resin capacity of 1. 0 equiv. /liter.
3. regeneration with ammonium hydroxide to give a
15% ammonium nitrate eluate.
-67-
-------
Feed Water
Absorb
Back-
wash
Treated Water
NaOH
W
Elute
Evapo-
rator
T
Rinse
Product
40% NaNO3
Figure 13 Schematic Diagram of Nitrate Removal Process.
Diagram shows a single column in four steps of a cycle.
-68-
-------
Because of the novelty of the system of liquid extractants in porous
beads a number of problems had to be solved in order simply to
reach the point of carrying out resin column experiments. For this
reason, we have not been able to explore the desired range of
process variables under items 1-3 above. However, we believe the
column data do establish the technical feasibility of the system.
Certain questions have had to be left unanswered, and the optimum
form of the system, as well as the values of some of the parameters,
are unknown at this point.
Flow Rate. The best column performance was the 10. 4 ml/min. run
shown in Fig. 10, which exhibited a nitrate selectivity of 19. The
bed treated some 35 bed volumes of feed prior to detectable nitrate
breakthrough, at a flow rate equivalent to 2 gpm/ft , and was apparently
free from channeling problems. This resin consisted of batch MPA-2,
which was made from the lower concentration of amidine solution, and
the column contained only 5. 5 meq of exchange capacity. The column
prepared from a higher amidine concentration (MPA-1) actually
contained over twice the capacity (12.85 meq), but was not .operated
at the 10.4 ml/min. flow rate. In order to estimate the operating
capacity of a practical sized bed we shall first assume that the MPA-1 resin,
if operated at the higher flow rate, would have treated at least twice
the feed volume as did the MPA-2 resin, or about 70 bed volumes.
We shall also assume that a reasonable estimate of scaled -up
behavior can be obtained by comparing equal solution residence times.
Thus, the 2 gpm/ft. " run, which was done in a 16 inch deep bed, would
be equivalent to a 10 gpm/ft. run in an 80 inch deep bed. While
additional runs would be necessary to confirm this prediction, in
particular runs with deeper beds and faster flows, we believe it is
reasonable to expect that the 70 bed volume capacity could be achieved
with the design flow rate of 10 gpm/ft .
Nitrate Loading. The original design value of nitrate loading was
selected on the basis of capacities of conventional anion exchange resins,
which commonly run somewhat greater than 1. 0 milliequivalent/ml
of bed (meq/ml). We have achieved values approaching 0. 2 meq/ml,
and this may "well be about the upper limit in this type of system.
Trie limitation is imposed by two factors: the fact that the absorbed
extractant is mainly limited to the pore volume of the bead; and the
presumed need for a diluent in order to maintain good diffusion.
Pore volumes of macroporous beads currently available generally
fall in the range of 30-40% of the bead volume. The highest extractant
concentration in the absorbing solution has been about 1. 1 M (about
50% by weight). Taking 33% and 1 M as the figures, and allowing a
factor of 40% for interstitial volume in the bed (between beads),
we calculate 0. 2 meq/ml as a typical bed capacity.
-69-
-------
While increases in bead porosity may be obtainable, we are not aware
of any such materials. We have considered the possibility of eliminat-
ing the diluent, in order to increase the concentration of the extractant,
and had planned to prepare and test a bed of such a material. The
approach is an uncertain one, since this might immobilize the extractant,
which is normally a viscous liquid or wax when in the salt form. In
this case, probably only the extractant molecules adjacent to the
aqueous interface would be effective. Practical utilization of the
extractant molecules further inside of the bead would be prevented
by the slowness of diffusion ina highly viscous medium. Such a
limitation affects the column behavior in two ways. First, the lower
bed capacity reduces the number of bed volumes of water which can
be treated per cycle. Second, the concentration of nitrate in the
eluate, for a given eluate volume, will be lower.
The importance of cycle length is that during at least a small portion
of the cycle a resin bed must be undergoing regeneration and back-
washing, during which time it is, of course, not treating feed water.
The number of columns required in a particular installation is thus
an inverse function of the fraction of the cycle devoted to absorption.
The cost estimate assumed a treatment capability of 374 bed volumes
of water per cycle, with 90% of the time devoted to feed water. For
a treatment capability of 70 bed volumes, only 70/374, or 19% as
much time is devoted to feed water. If we assume the time devoted
to regeneration is a constant, then the fraction of the cycle devoted
0.19 x 90
to feed water is -———— , or about 63%. Thus, assuming other
conditions to be the same, the current system would require 0.90/0.63,
or 1. 4 times as much cross sectional area in resin beds as was
assumed in the estimate.
Elution Performance. The impact of the lower capacity on the eluate
concentration is difficult to judge. We anticipated that the equilibrium
between ammonia and the extractant system would be a measurable
one, leading to tailing of the elution wave. Because of its greater
base strength, no such effect is expected with sodium hydroxide.
The observation of significant tails on both ammonium hydroxide and
sodium hydroxide elution waves suggests therefore that elution is
therefore limited by a kinetic process, and suggests that lower flow
rates might reduce the tailing with either regenerant.
Other problems, such as mixing or channeling of flow in the bed,
can also cause tailing, however. In fact, this problem is one of the
major limitations on the ability to obtain high concentration eluates
from an ion exchange bed. In the case of a bed with a total capacity
of 0. 2 meq/ml, for example, complete elution with a base should
require just 0. 2 meq. of base per ml of resin. In principle, this can
-70-
-------
be done with as little as 0. 01 ml of 50% (f 20 M) sodium hydroxide per
ml or resin. However, this requires that
-------
Of course, if the frequency of this operation is low, its reflection in
the economics of the process may not be significant.
In the system studied here we encountered losses of the extractant of
the order of 1 ppm, and of the diluent of the order of 20 ppm. A
cubic foot of polymer beads, loaded with a 50% solution of extractant,
will contain about 6 Ib. each of extractant and diluent. If we make the
somewhat arbitrary assumption that a rejuvenation step would be
required after 10% of one of the components has been lost, this would
occur in this case after 0. 6 Ib. of diluent was lost, or after treatment
of only about 3600 gallons of water per cubic foot of bed. If we further
assume a five-foot deep bed treating 10 gpm/ft. , and the absorption
step occupying 50% of the cycle, the 10% diluent loss would occur after
only about 2 days of operation. As we pointed out earlier, switching
to an appropriate higher molecular weight diluent would greatly
reduce this loss. For a 1 ppm soluble loss of either component the
10% loss point would be reached in about 50 days. It is fairly certain,
particularly in the case of the extractant, that soluble losses can be
pushed down in the 0. 1 ppm range. It is probably possible also for
the diluent, but less data are available on this point.
The point which emerges from this discussion is that the major
impact of small soluble losses of the reagents may not be the economic
loss incurred, but rather the increase in complexity which results.
Several ways of managing the rejuvenation problem can be imagined.
The simplest, operationally, would be that of using systems whose
losses could be kept in the 0. 1 ppm range or below, eliminating the
need for frequent attention. Another method is simply to add the
corresponding amount of material into the feed water, so that no
change in the bed occurs at all. Still another approach is to utilize
a bed of "empty", or extractant-free beads as a cleanup bed on the
downstream side of the processing beds. This bed would pick up the soluble
loss of extractant and diluent, and at some point, after absorption of
a substantial amount of the extractant system, could be switched to
the upstream side to act as one of the processing beds. At this stage
of development, we believe that the most practical means of handling
the problem lies in selecting compounds of extremely low soluble
loss. To this end the use of a less soluble diluent, possibly even
an aliphatic hydrocarbon, may be indicated. For the extractant, a
somewhat heavier material than the DoEhEhA is indicated. The com-
pound Do EhA (which we have not prepared) would probably be adequate.
Besides a greatly reduced soluble loss, the compound should exhibit
a slightly higher pK , which would also be desirable. Based on
SL
-72-
-------
these modifications in the extractant system, rejuvenation can probably
be handled adequately by a continuous addition to the feed water, as
described above. If not, periodic, but presumably infrequent, soaking
of the beads in a fresh extractant solution, either in place or in a
separate tank, would probably be required.
As a final observation, we believe it is important to point out that
the amidine extractants have a potential for application beyond their
use in polymer bead absorbent systems. We believe that, based on
the data obtained in this study, the amidines are suitable in every
major respect for application to the problem of nitrate removal.
The precise optimization of the structure has not been done, but we
believe that synthesis of one or two additional compounds would be
adequate, since predictions of the relevant properties of these materials
can be made.
The major problems have been encountered, not with the compounds
themselves, but with the means for employing them; that is, the
contacting operation. Further development of the extractant-in-bead
concept is obviously required, if this means of contacting is to be
successful. However, other means may be available. One obvious
type of operation is conventional liquid-liquid extraction. We expect
that the major difficulty with an extraction approach would be the
problem of reducing physically entrained extractant in the treated
water to the part-per-million level. Certain advantages would
become available in return, however, principal of which is the absence
of any limit on the final eluate concentration. Thus, relatively
concentrated nitrate eluates could be obtained with little difficulty.
Other more speculative methods for applying the selective amidine
systems to nitrate removal also exist, but are beyond the scope of
this discussion.
Commercial Availability of Amidines.
Except possibly for the lowest members of the amidine class, none
0.0 available as commercial materials. Any utilization of these
compounds in a process such as was studied in this project would
require development and synthesis of the desired compound for this
purpose alone. The cost of a material under these conditions would
depend not only on the raw material and processing complexity, but
also on the scale of manufacture. For these reasons, prediction of
the likely price of such a material is extremely difficult. Nevertheless,
it is possible to make some estimates of the order of magnitude of
cost of the final product. The following discussion is directed toward
the compound DoEhEhA, which was the most promising of the
-73-
-------
materials included in this study. However, since a variety of both
amines and carboxylic acids are available at comparable costs,
compounds of other structure and of other molecular weights could
readily be made at about the same price.
Carboxamides are available from a number of companies processing
fatty acids. They are usually the simple unsubstituted amides,
RCONH , and are prepared simply by heating the ammonium salt of
the acia to dehydrate it. While the acids themselves (e.g. stearic
acid, oleic acid) sell for 25-30jfc/lb. the amides are quoted in the range
of 50^/lb. Since the ammonia represents such a small portion of the
final product weight, and yields are probably close to theoretical,
the only significant raw material cost is that of the acid. The remaining
20-30^/lb. is attributable to the operating costs of the process. The
scale of production is not known.
A somewhat more useful situation is represented by the synthesis of
N, N-dibutyl-oleamide (DBO), which was studied on a pilot plant scale,
using a simple dehydration process (50). Yields of the crude material
ran about 95%. Using 53£ and 23d: per Ib. , respectively, as the costs
of the amine and acid, estimated costs for the product amide were
given as about 40c/lb. for production levels in the 2 to 10 million Ib.
per year range. Allowing about 35£ per Ib. for raw materials,
the processing cost turns out to be only something like 5j£ per Ib.
The difference between this figure and the 20ji estimated above for
fatty amides is very likely connected with a difference in production
scale.
The amide intermediate for the compound DoEhEhA would be prepared
in a similar manner from 2-ethylhexoic acid and 2-ethylhexylamine.
These two materials are quoted by Union Carbide Chemicals Corporation
at 32£ and 55^ per Ib. , respectively. Since the amide is composed
of about equal weights of acid and amine the raw materials cost of
the amide is roughly the average of the two costs, or about 45# per Ib.
Using the estimates of processing cost discussed above and allowing
for a 90% yield, the total cost of the amide intermediate, N-(2-ethyl-
hexyl)-2-ethylhexamide, would probably fall in the range of 50-70£/lb.
A model for the phosgene reaction step is more difficult to find.
Phosgene is used, however, on a relatively large scale. One major
use is in the production of diisocyanates, which are intermediates
in the manufacture of polyurethane plastics. Another use is in the
production of carbamate pesticides. Typical reactions are the follow-
ing:
-74-
-------
CH
NH + 2 COC1 -
Li t*
CH
NH-C=0 +2HC1
i
Cl
200°C
NCO + 2 HC1
methylene dianiline diisocyanate (MDI)
RNH + COC1
b Lt
NaOH
H2°
RNH - C = O + HC1
i
Cl
RNH - C = O
ONa
sodium N-alkyl carbamate
Economics on carbamate compounds are not plentiful. Information on
diisocyanates is somewhat more accessible; the MDI ii own above
sells for about $1 per Ib. , and is manufactured on a scale of about
5 million pounds per year. Using known costs for methylenedianiline and
phosgene of about 60j£/lb. and 15c//lb. , respectively, the raw materials
costs for MDI at 90% yield would run about 60^/lb. Processing and
other costs for this reaction, therefore, amount to something
like 40{:/Ib of product.
Synthesis of an alkylated amidine according to the methods utilized
in this study would be similar to the isocyanate synthesis. There is
in each case an initial reaction with phosgene, followed by a second
reaction involving an amine, in the amidine synthesis, or heat, in
the isocyanate synthesis. The raw materials costs would appear
as shown in Table XIII.
Table XIII
Raw Materials Costs for Amidine Synthesis
Theoretical
Unit Cost, Requirement,
Compound £/Lb. Lb. /Lb. Amidine
Amide 70 0. 60
Phosgene 15 0. 24
Dodecylamine 50 0.44
Total
Raw Material
Cost,
£/Lb. Amidine
42
4
22
68£
85£
(100% Yield)
( 80% Yield)
-75
-------
An 80% yield is probably not an unreasonable assumption. While we
obtained yields in the range of 60%, no attempt was made to optimize
the procedures, and in view of the literature reports, it seems reasonable
to expect that substantially higher yields could be obtained. Further
development of the synthesis procedures would be required.
Addition of a 40£ per pound processing cost would bring the final cost
of the amidine up to about $1. 25 per pound. This is probably a lower
limit, since consumption of the amidines would presumably be
relatively low. A 100 million gallon/day waste treatment plant, for
example, losing 0. 3 ppm of compound, would require only 75,000 Ib
per year as makeup. Should a number of plants require this compound,
or if some other use were to arise, the scale of manufacture could
result in a price in this range. The chances are that a smaller scale
synthesis would be required, and a reasonable estimate for a price
for this amidine is probably somewhere in the $2-5 per pound range.
Cost Projections
Preparation of a conventional cost estimate of the proposed process
at this relatively early stage of development would be premature.
We have achieved, however, some of the characteristics which were
used as assumptions in preparing the earlier cost estimate for the
Bureau of Reclamation contract (24). With suitable changes in that
estimate it is thus possible to develop an approximate projection of
processing costs for the extractant-in-bead process.
In Table XIV we give the basic assumptions which underlay the
earlier cost estimate, which pertained to a hypothetical resin with
a total capacity of 1.0 meq/ml, and a nitrate/chloride selectivity
of 20. The estimate was based on treating 100 million gallons/day
of agricultural drainage. We also show the corresponding assumptions
involved in treating the same volume of municipal sewage with the
extractant-in-bead system. We have made reasonable assumptions
for costs of materials and processing capabilities, based on the above
discussion. The major unproven assumptions involve a soluble loss
of extractant of 0. 3 ppm; a soluble loss of diluent of 1 ppm, and an
eluate concentration of about 1% sodium nitrate. While unproven, they
represent situations which can probably be realized with suitable and
straightforward changes in the process as it now stands.
Table XV contains the projection of costs. The lefthand column summar-
izes the costs developed in the Bureau of Reclamation estimate,
while the projections for the extractant-in-bead process are shown
on the right.
-76-
-------
Elimination of the ammonia and HC1 costs results from the elimination
of a somewhat involved purification process assumed in the earlier
estimate. No change was made in the power or labor costs. Steam
costs were increased by the same ratio as the increase in water
evaporation requirements.
Capital costs were increased by the factor 1.4, which represents
the? estimated increase in number of ion exchange columns due
to the lower fraction of cycle devoted to feed water. The resulting
total processing cost, 16£/M gal, represents an orde^r of magnitude
projection for this process, based on current capabilities.
From this projection it is possible to observe the character of
the major processing problems. The single most important one, of
course, is the high evaporation cost required by the relatively dilute
eluate anticipated. Attempts to reduce this cost can proceed in a
number of directions. One possibility is to increase the exchange
capacity of the beads. However, due to the nature of the beads and
the extractant system, it is unlikely this can be raised greatly;
complete elimination of the diluent, for example, would only raise
the capacity by a factor of 2. More promising approaches lie in the
manipulation of the operating variables. Further work may determine
that a smaller eluate volume will suffice. The 0. 5 bed volume assumed
here may be reduced by slower flow rates, or by recycling some of the
more dilute portions of the elution tail back through the feed water.
The use of a two column processing arrangement would probably
be desirable, both in increasing the volume of feed water per cycle,
and raising the ratio of nitrate to chloride in the loaded bed.
A secondary reduction in the projected cost can probably be anticipated
in the extractant item. While we assumed a loss of 0. 3 ppm, it is
probably reasonable to expect an even lower figure. Thus, with
proper choice of extractant, a significant portion of the 2£/M gal
might be eliminated.
-77
-------
Table XIV
Basis for Cost Projection
Plant Size, gal/day
2
Area Flow Rate, gpm/ft
Feed Water Composition, mg/1
NO-
Cl ~
% of Cycle Time on Absorption
Bed Vols. Feed Treatment per Cycle
Resin
Total Capacity, equiv/1 bed
Selectivity, NO~/C1~
Cost, $/cu. ft. (resin)
(contained extractant)
Resin Bead Life, Years
Extractant Loss, mg/1
" Cost, $/lb
Diluent Loss, mg/1
" Cost, $/lb
Elution
Agent
Eluate Concentration
Final Evaporated Product Cone.
Prelim. Cost
Estimate,
Bureau of
Reclam. Contract
100 x 10b
10
90
1000
90
374
1.0
20
50
Proposed for
Sewage
Treatment
100 x 106
10
62
355
63
70
20
25
12
10
0.3
2.00
1
0.03
25% NH (aqueous) ^-20% NaOH
15% NH.NO.
4 ;
1% NH.C1
4
54%
A Ql
1% NaNO,
40% NaNO.
Lb. HO Evap'd/MgalTr ated Water 4
100
-78-
-------
Table XV
Cost Projection
Estimate of
Bureau of Reclamation
Contract
Capital Cost
Direct Fixed
Resin
Total
Operating Costs
Raw Materials
NH 0.79
NaOH 1. 19
HC1 1.48
Preliminary- Cost
Projection for
Sewage
$4,230,000
2,826,000
$7, 056, 000
4. 51
Costs, £/Mga-1
1.5
3.8
Resin Replac. 1.05
Extract. "
Diluent "
Utilities
Power 0. 16
Steam 0.41
Operating Labor
Capital Costs*
Total
Credit - NH,NO, Sale
4 3
Net Cost
0.3
2. 0
0. 03
0.57 10.2
0. 16
10
0.50 0.50
0.85 1.2
6.4 15.7
3. 0
3.4 15.7
*Based on an amortization rate of 3. 1%/yr.
-79-
-------
ACKNOWLEDGMENT
This project was carried on at the Western Division Research
Laboratories of Dow Chemical U.S.A. , at Walnut Creek, California,
under the direction of Dr. Robert R. Grinstead. Dr. Kenneth C.
Jones was responsible for those portions of the work dealing with
the polymer bead systems.
The authors wish to acknowledge the contribution of Sigrid W. Snider,
who performed the organic syntheses,most of the physicochemical
work, and contributed Appendix I on synthesis procedures to the
final report.
We wish to also acknowledge the helpful discussions with the project
officer, R. A. Dobbs, of the Advanced Waste Treatment Research
Laboratory in Cincinnati, Ohio.
-81
-------
REFERENCES
1. Proceedings, Twelfth Sanitary Engineering Conference: "Nitrate and
Water Supply: Source and Control", Feb. 11-12, 1970. Univ. of Illinois.
2. F. C. Gunderloy, Jr., et al, "Dilute Solution Reactions of the Nitrate
Ion as Applied to Water Reclamation"; Report No. TWRC-1, 1968,
Federal Water Quality Administration.
3. "Sources of Nitrogen and Phosphorus in Water Supplies, "Task Group
Report, J. Am. Water Works Ass'n. 5^344 (1967).
4. A. F. Lenain, J. Am. Water Works Ass'n. 5J?, (8) 1049 (1967).
5. Public Health Service Drinking Water Standards, Revised 1962.
Public Health Service Publ. No. 956.
6, H. I. Nightingale, Groundwater 8_, (1), Jan-Feb (1970).
7. T. E. Larson, L. Henley, Final Report, Project 65-05G. U.S. Dept.
of Interior, Office of Water Resources Research, June 1966. See
also ref (1).
8. W. D. Keller, G. E. Smith, Geol. Soc. of America, Special Paper
90, (1967).
9. B. A. Stewart, et al, Env. Sci. & Tech. j_ (9), 736 (1967).
10. San Francisco Chronicle, Aug. 3, 1970.
11. W. Fresenius, F. J. Bibo, W. Schneider, Gas and Wasserfach 107,
(12), 306 (1966).
12. "The Interrelation of Carbon and Phosphorus in Regulating Hetero-
trophic and Autotrophic Populations in Aquatic Ecosystems"; P.C. Kerr,
Water Pollution Control Research Series Report No. 16050, Federal
Water Quality Administration, 1970.
13. E. F. Barth, et al, J. Water Poll. Control Fed. 38_ (7), 1208(1966).
14> E. F. Barth, et al, ibid. 4£(12), 2040(1968).
15< S. Balakrishnan, W. Eckenfelder, Water Research _3_ (1), 73(1969).
16> S. Balakrishnan, W. Eckenfelder, ibid_3 (2) 167 (1969).
}7_ "San Joaquin Master Drain", State of Calif. Dept. of Water Resources,
Bull. No. 127, Jan. 1965.
-83-
-------
18. "Nitrogen Removal from Waste-waters", progress report, Advanced
Waste Treatment Research Laboratory, Cincinnati, Ohio, May, 1970.
19. W. R. Samples, Chem. Eng. Prog. Symp. Series No. 78, 6_3 223 (1967).
20. P. P. St. Amant, P. D. McCarty, J. Am. Water Works Ass'n. 61,
659 (1969).
21. R. Eliassen, et al, ibid, _57, 1113(1965).
22. R. Eliassen, G. E. Bennet, J. Water Poll. Control Fed. 39, (10),
Part 2, R 82 (1967).
23. R. Eliassen, G. Tchobanoglous, ibid 40 (5) part 2, (1968).
24. R. R. Grinstead, et al "Feasibility of Removal of Nitrates from San
Luis Drain Waters by Ion Exchange", Final Report on Contract No.
14-06-D-6434, U.S. Bureau of Reclamation, 1968.
25. Y. A. Marcus. Chem. Revs. 63_, 139 (1963).
26. H. Small, J. Inorg. Nucl. Chem. 1£, 232 (1961).
27. H. Small, ibid 1_9_ 160 (1961).
28. H. Small, U.S. Patent 3, 146, 213 (1964).
29. M. B. Goren, U.S. Patent 3, 320, 033 (1967).
30. M. Manes, French Patent 1,548,786(1968).
31. K. A. Kun, R. Kunin, J. Polymer Sci. C No. 16, 1457(1967).
32. R. R. Grinstead, J. C. Davis, "Recovery of Salts from Saline Water",
U.S. Office of Saline Water Research and Development Report No.
406, 1969.
33. R. R. Grinstead, J. C. Davis, "Recovery of Salts from Saline Water",
U.S. Office of Saline Water Research and Development Report. No.
320, 1968.
34. J. C. Davis, R. R. Grinstead, J. Phys. Chem. 72^ 1630 (1968).
35. "lonization Constants of Acids and Bases", A. Albert, E. P. Serjeant,
Wiley & Co. , N. Y. 1962
36. R. L. Shriner, F. W. Neumann, Chem. Revs. 35_ 351 (1944).
-84-
-------
37. "Open Chain Nitrogen Compounds", P. A. S. Smith; W. A. Benjamin,
Inc., N. Y. , 1965, p. 177.
38, "The Chemistry of the Imidoyl Halides", H. Ulrich; Plenum Press,
N. Y., 1968
39. H. Eilingsfeld, et al, Angew. Chem. , Int'l Ed. (Sample Issue)
(May 1961), 45
40. M. Grdinic, V. Hahn, J. Org. Chem. 3_0 2381 (1965).
41, H. Eilingsfeld, et al, Chem. Ber. 96_2671 (1963).
42. C. Jutz, H. Amschler, Chern. Ber. %_, 2100 (1963).
43. H. Bredereck et al, Chem. Ber. 92, 837(1959).
44. "Organic Chemistry of Bivalent Sulfur", E. E. Reid, ed. Chem.
Publishing Co. , N. Y. , 1963 Vol. V.
45. E. Brand, F. C. Brand, Organic Syntheses III, 440
46. N. Bortnick et al, J. Am. Chem. Soc. 78_4358 (1956).
47. D. Jenkins, L. Medsker, Anal. Chem. 36 610 (1964).
48. C. McAuliffe, Nature, 200 1092 (1963).
49. R. Bastian, et al, Anal. Chem. 2% 1795 (1957).
50. L. W. Mazzeno, Jr., Ind. Eng Chem. Prod. Res. & Dev. 9^ No. 1,
42 (1970).
-85-
-------
APPENDIX I
SYNTHESIS PROCEDURES
N, N'-Dicyclohexyl octanamidine Carbodiimide Method.
The Grignard reagent was prepared in 300 ml. ether from 100 g.
(0. 52 moles) of 1-bromo-octane and 12. 6 g (0. 52 moles) of magnesium
in the usual manner. The solution was transferred to a dropping
funnel and added dropwise to a dry solution of 71 g. (0. 34 moles)
of dry dicyclohexylcarbodiimide in another 300 ml of ether. The
carbodiimide had been previously melted and filtered through sintered
glass to free it of a higher melting impurity which was found in
our samples.
After 24 hours water was added slowly with the flask packed in ice,
then aqueous HC1 was added until all the solid had dissolved. After
separation of the aqueous phase, the organic (2 phases) was evaporated
under reduced pressure. The residue had a very high KOH equivalent
weight, indicating a neutral impurity which was removed by shaking
with a large volume of pentane. A solid resulted, which was collected
by suction filtration. The solid was a mixed bromide-chloride salt,
so was shaken with pentane and 10% NaOH to convert to free base in
pentane solution (some toluene added to help the solubility). The
organic solution was filtered to remove entiained water, and anhydrous
HC1 bubbled through it. The hydrochloride split out as a liquid
phase, and solidified only after all solvent, including water, which
was probably responsible for its not crystallizing out of solution had
been removed under reduced pressure. Yield: 50%.
N-(2-Ethylhexyl)-2-ethylhexanamide
105 g (. 65 moles) of 2-ethylhexanoylchloride and 180 g. (1. 39 moles)
of 2-ethylhexylamine were each dissolved in 200 ml of ether and the
amine solution was added slowly to the acid chloride solution with
stirring. An ice bath helped control the copious fuming as well as
termperature. Stirring was continued without the ice bath for approx-
imately one hour after the addition was completed. The solution was
then shaken with excess HC1 . Water was added to effect a two-
phase system and the mixture shaken again. Another acidified water
wash was followed by two water washes, and the ether solution
filtered and evaporated under reduced pressure to a constant weight.
Yield 92-4%.
-87-
-------
Notes:
1. Needed only for the amount of starting amine greater than
twice the amount of starting acid chloride.
2. Aqueous washes were clear, and ether phase cloudy throughout.
N,N'-Di(2-ethylhexyl)-2-ethylhexanamidine. PCI Method.
133 g (0. 64 moles) of PCI was added to 163. 3 g (0. 64 moles) of
2-ethylhexyl-2-ethylexanamidine with mechanical stirring. The mixture
was heated to 120° externally (95° internally) for 3 hours. 51 g.
(0. 64 moles) of pyridine was added, then 83 g (0. 64 moles) of 2-
ethylhexylamine, slowly, with the temperature maintained at 160°
externally for 75 minutes longer. The heating was discontinued, and
when the temperature had dropped to below 100°, 42 g of water was
added slowly. Chloroform was added when the temperature had dropped
below its boiling point, the aqueous separated, and the organic washed
3 times with water at 45° -60°C . The organic was evaporated under
reduced pressure. The remaining oil was partitioned between 600 ml
of 75% aqueous ethanol and 300 ml of pentane, and the ethanol solution
washed 3 times more with 300 ml portions of pentane. On evaporation,
an equivalent weight determination showed the product contained 18%
neutral impurity. It was therefore converted to free base by shaking
with 300 ml pentante and 10% NaOH, and washing the pentane solution
3 times with 75% ethanol. Evaporation of the pentane solution, with
the temperature kept below 30° yielded the free base in 22% yield.
HC1 eq. wt. :391, calculated:367.
Prep, of N-Dodecyl-N'-(2-ethylhexyl)-2-ethylhexanamidine (DoEhEhA ) .
Phosgene Method.
Procedure: 46 ml (0. 65 moles) of phosgene was collected and dissolved
in 200 ml of chilled toluene, then transferred to a dropping funnel.
The solution was added to a solution of 78 g (. 31 moles) 2-ethylhexyl-2-
ethylhexanamide in 200 ml of toluene, in an ice bath. The rate of
addition was such that the temperature stayed below 30°. The system
was stirred overnight, after which excess phosgene was removed by
a partial evaporation in vacuo. 57 g (0. 31 moles) of dodecylamine
was dissolved in toluene and added dropwise to the reaction mixture,
keeping the temperature AX 30°. (A solid or 2nd phase split out).
The system was stirred over a weekend .
An excess of HC1 was added and the mixture transferred to a separate
funnel and washed 3 times with boiling water .
-88-
-------
The final organic solution - still very cloudy - was gravity filtered
through several paperJilters, and evaporated in vacuo to a constant
weight in a water bath at 80-90°C.
The material was dissolved in/V 4 ml/gm of 80% ethanol (by volume -
hereafter called ethanol solution) and shaken with ^half that volume of
hexane three times. The ethanol solution was then evaporated in
vacuo as the toluene solution was above.
Eq. wt (Cl"):480, theo: 459 Yield: 60% based on 96% purity
Footnotes to DoEhEhA Prep;
1. Progress of this reaction was followed by samples periodically
withdrawn and evaporated by blowing N over them to evaporate
the solvent and phosgene. An I.R. scan of the residue showed the
relative strengths of bands at 5. 9 (x (desired) and 6.1 (j, (starting
material) and incidentally at 5. 7 (undesired).
2. Time necessary undetermined - all runs went overnight at this
point, and the ones with best yields went 2 or 3 days due to week-
ends or holidays.
3. To convert excess amine to hydrochloride salt which is soluble in
warm water.
4. The high temp, aids this slow separation - it is advisable to keep
it hot (70°-80°) during the separation of phases.
5. This material can be tested for purity by converting a weighed
sample to the free base by shaking with dil. aq. NaOH and toluene,
then titrating (electrometrically in 90% acetone and water) to get
equivalent weight. Any remaining dodecylamine will show up
here as a second break (weaker base) on the titration curve.
6. Near the end the product was very thick and bubbled over into
the receiver frequently.
-89-
-------
N-(Primene 81R)-N', N"-di(o-tolyl)guanidine. Carbodiimide Method.
CH,
CH
CH.
N = C = N
20 g (0. 1 moles) of Primene SIR was added slowly to 22.4 g (0. 1 moles)
of di(o-tolyl)carbodiimide, swirled, (the temp, rose to 75°) and clamped
in the steam bath with a drying tube for 24 hours. The mixture was
cooled, taken up in 400 ml of toluene and stirred as anhydrous HC1
was bubbled through. The hydrochloride solution was washed 2 times
with 11. of water and evaporated under reduced pressure. Yield:45%.
1, 3-di(2-ethylhexyl)thiourea - 258 g (2 moles) of 2-ethylhexylamine
were added to 96 g (1. 26 moles) of carbon disulfide and 250 ml of
water with good stirring. The temp, went to 70°. After two hours,
90 g of 50% NaOH (1. 12 moles) was added, causing the thick mixture
to become even thicker. The mixture was heated to reflux 3 hrs.
It was thin again, and yellow by the time it reached reflux, and orange
30 min. later. After standing overnight, the mixture was acidified
with HC1 and turned yellow again and was extracted with chloroform.
The chloroform solution was washed 3 times with hot water, the 1 st
wash containing HC1. The chloroform solution was filtered and evaporated
in vacuo. Yield: 240 g = 80%.
S-Ethyl-N, N' -di(2-ethylhexyl)thiuronium bromide
80 g (0. 27 moles) of 1, 3-di(2-ethylhexyl)thiourea and 33 g (0. 30 moles
of ethyl bromide were mixed in 50 ml of ethanol and heated to 55-60°
for 1-1/2 hrs. The solvent and excess ethyl bromide were removed in
vacuo to a constant weight.
Yield 99 g = 91%. Analysis C = 57. 9 H = 10. 6, N = 7. 2, Br = 19. 1, S = 7. 7
Calculated C = 55. 8, H = 10. 1, N = 6. 8, Br = 19. 5, S = 7. 8.
N-Phenethyl-N', N"-di(2-ethylhexyl)guanidinium bromide
48 g (0. 12 moles) of S-ethyl-N, N1-di(2-ethylhexyl)thiuronium bromide
were dissolved in 200 ml of ethanol and 71 g (0. 59 moles) of phenethyl
amine added. The mixture was heated at 60° overnight. An excess
of aqueous HBr was added and the mixture boiled to 90° to remove
ethanol and ethyl mercaptan. Chloroform was added to the residue,
-90-
-------
and the solution washed 3 times with hot water after the original
aqueous phase was removed. The chloroform was evaporated under
reduced pressure, and the crude product partitioned between 625 ml
of 80% ethanol and 300 ml of hexane. The ethanol solution was
washed 2 times more with 300 ml of hexane and the ethanol solution
evaporated under reduced pressure to a constant weight. KOH equiv.
wt. ; 498 Calculated eq. wt. : 468.
-91-
-------
APPENDIX II
DETERMINATION OF ALKYLAMMONIUM IONS IN AQUEOUS SOLUTION
PICRATE METHOD
Reagents:
1. 2 x 10 M picric acid in a pH 3 citrate/phosphate buffer
solution*
2. Reagent grade chloroform
Procedure:
1.
2.
3.
4.
Pipet 5 ml. each of the picrate reagent and chloroform into
a 20 ml vial.
Pipet 5 ml. of aqueous sample (approximately neutral) into
the reagents. Also prepare a reagent blank using 5 ml. of
distilled water instead of the sample.
Shake 5 min. , centrifuge, and collect the organic phase.
Read absorbance at 410 mjj. in a 1 cm. cell. The amine
concentration is determined from a curve for the particular
amine. This is made by shaking chloroform solutions
in the range of 0-5 x 10 M amidine with the buffered
picrate solution and reading the absorbance of the organic
phase. Blanks should typically be less than 0.02.
* 458 mg picric acid, 16. 7 g. citric acid monohydrate, and 5. 84 g.
anhydrous disodium phosphate in 1 liter of water. Solution is
0. 079 M in citrate, and 0. 041 M in phosphate.
-93-
-------
APPENDIX III
pKa of Amidines
Table III-l Apparent pKa of Amidines in 0. 010 M NaCl
Concentrations, M
Compound (B)
in toluene
Cy2 NoA 0 051
0. 077
0. 102
0. 153
0.255
Eh2EhA 0.0503
0. 0754
0. 1005
0. 1508
0.2513
in Chevron 3
Eh2EhA 0.0514
0. 0776
0. 103
0. 153
0.256
0.446
0. 371
0.256
0. 121
0. 046
Eh20cA 0.051
0.076
0. 103
0. 154
0. 257
EhOcEhA Q Q45
0. 067
0. 090
0. 134
0.224
(BHX)
0.285
0. 055
0. 0785
0. 128
0.229
0.049
0. 074
0. 100
0. 150
0.250
0.049
0. 074
0. 100
0. 152
0.255
0. 050
0. 125
0.255
0. 375
0.450
0. 051
0. 766
0. 102
0. 153
0.255
0.051
0. 0766
0. 102
0. 153
0.255
PH
7.95
8.08
8.20
8.40
9.00
6.42
6. 70
6.76
7. 02
7. 13
6.42
6.57
6.71
6. 90
7. 10
7.60
7.35
7. 10
6.95
6.70
7.27
7. 30
7.47
7. 57
7.82
6. 94
6. 92
7. 14
7. 22
7.54
pKa
9.54
9.78
9.92
10. 14
10.55
8.40
8.68
8.75
9.01
9. 12
8.40
8.53
8.68
8.88
9.08
8.65
8.86
9.08
9.44
9.69
9.26
9.29
9.46
9.56
9.81
8.99
8.70
9. 18
9.27
9.59
-95-
-------
APPENDIX HI
Compound
DoEhEhA
DoEhNdA
Table
III-l (Cont'd)
Concentrations, M
(B)
0.098
0. 146
0. 244
0.436
0. 351
0.244
0. 118
0. 046
0. 780
0. 647
0.430
0. 212
0. 084
0.052
0.078
0. 104
0. 155
0. 260
(BHX)
0. 102
0. 153
0. 255
0. 048
0. 133
0. 255
0.366
0.438
0. 084
0. 212
0.434
0. 652
0. 780
0.052
0.078
0. 104
0. 156
0.259
pH
6. 81
09
40
68
49
40
00
44
90
85
70
7. 23
6. 82
7.
7.
7.
7.
7.
7.
6.
7.
7.
7-
5.45
5.80
5.75
5.50
6.25
pKa
8.83
9. 11
9.42
8.62
9.06
9.42
9.49
9.41
8.92
9. 37
9.69
9.70
8.77
7.45
7.80
7.75
7. 50
8.25
-96-
-------
TABLE III-2
pKa OF ALIPHATIC AMINES IN 0. 010 M NaCl
Compound
Primene JM-T
Concentrations, M
(B)
(BHC1)
pKa
1. 17
0.89
0.60
0.30
0. 12
0. 12
0. 30
0.60
0.89
1. 17
6.50
6.30
6.00
5.69
5. 00
7.51
7.83
8. 00
8. 16
7.99
Amberlite LA-2
0.91
0.70
0.49
0.28
0.06
0.21
0.42
0.63
0.84
1. 06
4.85
4.6
4.3
3.9
3.2
6. 21
6.4
6.4
6.4
6.4
-97-
-------
TABLE m-3
APPARENT pKa OF Eh EhA /CHEVRON 3 IN NITRATE SYSTEMS
Concentrations, M
Compound
(B)
(BHX)
pH
pKa (NH )
Aqueous: 0. 010 M Sodium Nitrate
0. 097
0. 145
0. 242
0. 101
0. 150
0.252
8.25
8.45
8.74
10.26
10.46
10.76
Aqueous: 1. 0 M (NH + NH NO ); Organic 0. 50 M Amidine
(N?3lL
0. 78
0.22
0.48
0.42
0.28
0. 017
9.42
9. 72
9.96
10.4
10.3
9. 1
.45
. 62
.75
.54
.36
.24
0.54
0.36
0.24
-98-
-------
APPENDIX IV
SOLUBLE LOSS OF AMIDINE HYDROCHLORIDES
Organic diluent, Chevron 3, except as noted; aqueous phase,
sodium chloride
Concentrations, M
Compound
CvNoA-HCl (toluene)
Z
Eh EhA • HC1 (toluene)
2
Eh0OcA ' HC1
2
EhOcEhA • HC1
Eh^EhA ' HC1
2
DoEhNdA- HC1
DoEhEhA • HC1
(BHC1)
0. 083
0. 087
0.091
0. 092
0. 1013
0. 1014
0. 1015
0. 1015
0. 100
0. 100
0. 100
0. 100
0.096
0.096
0.096
0.096
0. 100
0. 101
0. 101
0. 101
0.20
0.60
1.00
0. 104
0. 104
0. 104
0. 104
0. 098
0.098
0.098
(Cl~)
0. 015
0. 033
0. 101
0.300
0.0102
0. 0301
0. 10
0. 30
0. 0102
0.0301
0. 100
0.300
0. 0102
0.0301
0. 100
0. 300
0.0105
0. 0302
0. 101
0.300
0. 01
0. 01
0. 01
0.010
0.030
0. 10
0. 30
0.010
0. 010
0.010
(EH*)
_3
5.2 x 10 ,
_3
3.0 x 10
8. 0 x 10"4
2.8 x 10"
_4
1.8 x 10 _
— J
8.5 x 10
3. 1 x 10"
1.8 x 10"
_4
1.5 x 10 _
_ n
5.7 x 10
3.5 x 10"
3.5 x 10"
_4
2.2 x 10
1. 16 x 10"
5.3 x 10"
2. 1 x 10"
_4
4.9 x 10
1. 92 x 10
6.75 x 10"_
_ T
4.25 x 10
_4
5.3 x 10
6.6 x 10
8.8 x 10"
-5
2. 8 x 10
4.6 x 10"
1.3 x 10"
1.0 x 10"
-6
8 x 10 6
13 x 10~,
3x 10"b
K
s
1060
880
1130
1090
4
5.5 x 10
4.0 x 10
3. 3 x 10
1.7 x 10
4
6.5 x 10
5.3 x 10,
4
2.9 x 10*
9.5 x 10
4
4.3 x 10
2.75 x 10
1.8 x 10
1.5 x 10
4
1. 94 x 10
1.74 x 10
1.48 x 10
7.92 x 10
4
3.8 x 10
9. 1 x 104
11.4 x 10
5
3. 7 x 104
7.5 x 104
8.0 x 10
3.5 x 10
In6
1.2 x 10 ,
0.75 x 10
3.3 x 10
-99-
-------
1
Accession Number
w
5
Organization
« Subject Field & Group
5D
SELECTED WATER RESOURCES ABSTRACTS
INPUT TRANSACTION FORM
Dow Chemical U.S.A. , Western Division Research Laboratories,
2800 Mitchell Drive, Walnut Creek, California 94598
Title
Nitrate Removal from Waste Waters by Ion Exchange
i /"\ Authors)
Grinstead, Robert R.
Jones, Kenneth C.
16
21
Project Designation
Proiect #17010.
Note
Contract #14-12-808
22
Citation
23
Descriptors (Starred First)
#Nitrates, *Denitrification, *Anion Exchange, Waste Water Treatment,
Municipal Wastes, Water Pollution Treatment, Resins, Tertiary
Treatment, Solvent Extractions.
25
Identifiers (Starred First)
Amidines
27
Abstract -phis repOrt described an exploratory experimental study of the use of porous
polymer beads containing a water-immiscible extractant system for the removal of
nitrate from waste waters. Alkylated amidines proved to be a suitable class of compounds
for the extractant system. They are relatively strong bases, and exist in the salt form
in contact with waste waters in the pH range of 7-8. They can, however, be readily
regenerated with alkalis, such as ammonia or sodium hydroxide.
The amidinium ion in the organic phase selectively extracts nitrate ion over chloride
ion by a factor of about 20 (i. e. , the nitrate/chloride ratio in the organic phase is about
20 times the ratio in the equilibrium aqueous phase), and nitrate over sulfate and bi-
carbonate by much higher ratios. From typical municipal waste waters amidine systems
will therefore pick up mainly the nitrate ion.
Amidines dissolved in an aromatic hydrocarbon were absorbed in macroporous
polystyrene beads and used to treat a synthetic municipal waste water containing 62 ppm
nitrate ion and 350 ppm chloride ion. Beds of this material treated up to 70 bed volumes
of water prior to breakthrough of the nitrate in the effluent. The absorbed nitrate
ion was removed with either ammonia or sodium hydroxide.
Abstractor
Grinstead
Institution
Dow Chemical U.S.A.
WR:I02 (REV. JULY 1969)
WRSIC
SEND, WITH COPY OF DOCUMENT. TO: WATER RESOURCES SCIENTIFIC INFORMATION CENTER
U.S. DEPARTMENT OF THE INTERIOR
WASHINGTON. D. C. 20240
* GPO: 1 970-389-930
------- |