INDUSTRIAL AND ENGINEERING CHEMISTRY
1230
dicates that the light-scattering method can be useful in the investigation of stability additives in fuels. Tentative correlation of light-scattering results with other criteria of fuel stability indicates that the light-scattering method should be a useful tool for predicting in a short time the expected behavior of fuels in extended storage. ACKNOWLEDGMENT
The authors are deeply indebted to C. R. Singleterry of this laboratory for invaluable advice during the course of this study.
Vol. 47, No. 6
LITERATURE CITED
(1) Brice, B. A., Halwer, M., and Speiser, R., J . Opt. SOC. Amer., 40,
768 (1950). (2) Oster, G., Chem. Revs., 43, 319 (1948). (3) Smith, H. M., Bureau of Mines, Bartlesville, Okla., private communication, Dec. 31,1953. RECEIVED for review October 8, 1954. ACCEPTED December 22, 1954. The opinions and assertions contained in this article are the private ones of the authors and are not to be construed as reflecting the views of t h e Navy Department or the naval establishment a t large.
Anionic Softening of Water with StrongBase Anion Exchange Resins ROBERT KUNIN AND FRANK MCGARVEY Rohm & Haas Co., Philadelphia, Pa.
QUATERNARY ANION EXCHANGERS can
be
used for water soffening
. .. without acid resistant equipment . . .with a simple technique . , . atening costs comparable to other softmethods
A
LTHOUGH the cationic composition of water has been modified with cation exchangers for many years, similar processes for the exchange of anions have received little or no attention until recently. Fluorides, sulfides, carbonates, bicarbonates, and silica have caused difficulties in the use of many water supplies, These anions are not universally objectionable, but in certain applications their presence gives rise to difficulties. FLUORIDES
I n 1931 it was first reported by Smith, Lantz, and Smith (12) and Churchill (2) that the presence of fluoride in drinking water was related to certain bone diseases, the first symptom of which ~170sthe mottling of teeth. Since that time a number of workers have verified this finding, and have shown that in areas in which drinking water contains a fluoride concentration greater than 2 p.p.m. (as F), the inhabitants suffer from a mottling of the teeth, as well as lesions of the endocrine glands, the thyroid, liver, and other organs. A number of areas throughout the world must use for drinking purposes water with a high fluoride concentration. Several cases of mottling of the enamel of teeth have been reported in Algeria, Tunisia, Spain, Italy, Russia, South America, and the United States. A survey made in 1941 (5) reports the extent of fluorosis in the United States. Of the areas investigated, 360 had water supplies containing more than 2 p.p.m. of fluoride (as F), the Texas panhandle area being the worst found. Churchill’s investigation (2) in 1931 showed that some fluoride could be expected in most areas west of the Appalachians. The midwestern section is the
really bad area, and cases of fluorosis have been reported in the Dakotas, Kansas, Colorado, Arkansas, and Iowa as well as Texas. Although fluoride concentrations greater than 2 p.p.m. (as F ) are harmfuI, waters containing approximately 1 to 2 p.p.m. inhibit the formation of dental caries. Many communities have started programs of fluoride additions to their water supplies. I n some cases this addition has caused a hardship on the food and packing industry, since some processes-for example, jelly making-may involve a concentrating step which could increase the fluoride concentration in the final product to objectionable ranges of 10 to 20 p.p.m. ( 4 ) . Several,methods have been proposed for removing fluorides from water, the more important of which are: adsorption on basic calcium phosphates, precipitation with aluminum compounds and clay, precipitation with lime, and removal with aluminum-treated ion exchange resins. These processes have been reviewed critically by Maier (ZO), who has found them unsatisfactory for one reason or other. SULFIDES
Although there are few published data on sulfide contamination, it is recognized that this is an important and difficult problem. Sulfide concentrations as low as 0.5 pap.m. give an objectionable taste and odor to the water. Aeration will remove sulfide down to a concentration of 2 to 3 p.p.m. Further treatment must be used to produce a taste- and odor-free water. Chlorination and iron precipitation and coagulation are frequently used, but are not easily adapted to small plants and to individual domestic use. BICARBONATES AND CARBONATES
The presence of carbonates and bicarbonates in water supplies is objectionable, particularly where the water is to be used for preparation of carbonated beverages, or for boiler feed. The accepted practices for removing bicarbonate and carbonates are: lime precipitation, sulfuric acid neutralization, and neutralization with the hydrogen form of a cation exchange resin. These methods are objectionable in some cases because of the incompleteness of removal, difficultyin maintaining a constant pH, and necessity of acid-resistant equipment and of acid handling. SILICA
The presence of fractional parts per million of silica in water is objectionable when the water is to be used for high-pressure
INDUSTRIAL AND ENGINEERING CHEMISTRY
dicated that weakly basic resins might also accomplish this task, they cannot function in neutral media as the quaternary resins do and are thereby useless for water treatment, except for deionization applications. I n order to investigate this method, a series of quaternary anion exchange resins was studied using various synthetic and natural water supplies.
RCI t N o F # R F t N o C I REGENERPTION-2LE. DPOINT-I
1231
NoCl,FT'
BPM. F A S C o c o 3
EXPERIMENTAL STUDY
Resins Studied. Four quaternary anion exchange resins were selected for this study. Two of the resins, Amberlite IRA-400 and Amberlite IRA-410, are standard quaternary resins. The other two, Amberlite IRA-401 and Ambeilite IRA-411, are porous analogs of the same resins. A description of these resins is given in Table I.
Table I. Quaternary Anion Exchange Resins Investigated for Anion Exchange
Resin 20
40
60
COMPOSITION RATIO,
Figure 1.
80
too
[ T O T A L 2N,0NS] 100
Amberlite Amberlite Amberlite Amberlite
IRA-400 IRA-410 IRA-401 IRA-411
Apparent Density, Grams(Dry)/ M1. Description Very strong base 0.38 Strong base 0.35 Porous very strong base 0.28 Porous strong base 0.28
Total
Capacity Meq./MI: 1.20 1.25
0.95 0.95
Fluoride reduction with strongbase anion exchangers
boilers and turbines. Magnesia adsorption techniques or deionization by ion exchange using quaternary ion exchange resins are the accepted practices a t present. These methods are well described in the literature and are not applicable to the present study (IS). As a result of previous studies (8) on the equilibria of anion exchange in quaternary-type anion exchange resins, it became apparent that certain quaternary ion exchange resins might be capable of adsorbing fluoride, sulfide, and bicarbonate ions in exchange for chloride ions. Although other studies (9) have in-
Operational Procedure. The resins were studied in 1-inch columns containing 250 ml. of resin and the procedure employed was identical to that previously described (7). The exhaustion flow rate was limited to a value of 2 gallons per cubic foot per minute and the regeneration limited to 1 gallon per cubic foot per minute. Regenerations were achieved with 10% salt solutions, except where indicated. Two methods of column operation were investigated, concurrent and countercurrent. I n concurrent operation, exhaustion and regeneration were downflow. In countercurrent operation, exhaustion was upfiow and regeneration waa downflow. For the most part, the exhausting solutions were composed of fluoride, sulfide, and bicarbonates with sufficient chloride ion to produce the desired ratio. Fluoride concentrations were kept below 10 p.p.m. except in one case. End points were restricted to specific leakages except under conditions where leakage exceeded these arbitrary values. I n those cases, the end point was selected a t the point where a definite and rapid increase in anion was observed, Analytical Methods. Fluoride analyses were performed volumetrically according to the method of Rowley and Churchill (11)
- CONCURRENT
R E G EN E R A NT 10 % N 0 C I
0
3.5
0
=
m w
4 5 m-l
z .
3.0
COMPOSITION RATIO,
Figure 2.
[ TOTAL FAN,ON ]
Fluoride reduction with strong-base anion exchanger at low composition ratios
I
I
I
I
I
2
3
4
REGENERATION
Figure 3.
L E V E L , LE.
5
NoCIIFT.~
Fluoride elution
5
1232
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
Vol. 47, No. 6
0.02
.f
"r-.
Ib
0.18
0
2
0.16
li P
L
b- 0.12
t 0
8q
0.08
w
P 0 0
n
a A
Y
REGENERATION LEVEL,
LB
N a C l I FTa
Figure 4. Fluoride reduction capacity of Amberlite IRA-401 as a function of regeneration method Exhausting solution. 275 p.p.m. NaCl fluoride a8 CaCOt
EM
0.04
0.0
0. s
0
Figure 5.
CrtCOa, 19 P.P.m.
NaCIIFT.'
Fluoride removal capacity of Amberlite IRA401 as a function of regeneration level
Influent. 275 p.p.m. NaCl as CaCOa, 19 Regenerant 10% NaCl End-point dreakthrough
using thorium nitrate. An end point of 1.0 p.p.m. fluoride W&CI chosen for most of the runs. Sulfides were determined colorlmetrically with antimony potassium tartrate ( 1 ) and an end point of 0.5 p.p.m. was chosen arbitrarily for the sulfide breakthrough point, exce t under certain cases where higher leakages were observed. Eicarbonate concentrations were determined volumetrically by titrating to the methyl orange end point with 0.1N hydrochloric acid. An average leakage of 10% alkalinity W&E considered as the end point for the dealkalization studies.
e. 5
P .O
1.5
1.0
REQENERATION L E V E L , LE
prim.
fluoride as CaCOn
14
I2
h.
RESULTS
Fluoride Removal. The results obtained in the study of fluoride removal are summarized graphically in Figures 1 to 7. It Is apparent that porous exchangers Amberlite IRA-401 and IRA411 are superior to their less porous analogs. Figures 1 and 2 show the effect of influent composition on fluoride capacity. As the exchange reaction involves the relative affinity of the resln for chloride and fluoride, the study was confined to chloridefluoride solutions. Preliminary studies indicated that other anions have little effect on capacity for fluoride. The results shown in Figures 1 and 2 were obtained using concurrent regeneration techniques. Figure 3 shows that considerable tailing results during regeneration, which indicates that there would be some advantages to operation in a countercurrent direction. Figures 4 and 5 illustrate this point for a typical influent. Some typical breakthrough and leakage curves are shown in Figures 6 and 7. The data are striking, in that they indicate the ion exchange process to be effective and easily regenerated a t low level dosages of brine. The advantages of countercurrent operation over concurrent are low fluoride leakage and high capacities. Bicarbonate Removal. The data reported in Figures 8 to 11 summarize the information on the bicarbonate-chloride exchange. A comparison between the porous and nonporous quaternary ammonium anion exchangers is shown in Figure 8. Because the addition of small amounts of caustic to the salt regenerant has been found t o be effective (6, 6) for increasing the capacity, a series of runs was carried out to establish the utility of this method. Figure 9 shows typical exhaustion curves for the caustic-salt technique using IRA-410. The capacity values are summarized in Figures 10 and 11 for variable amounts of caustic in a salt-caustic regeneration. The total amount of caustic and salt used in these studies equaled 4 pounds per cublo foot. A water containing boy0 alkalinity, 5 p.p.m. free carbon
: a
10
Y 4 wI
w
5
Z
6
4
LB.
NeCIIFT3
LEAKAGE %
GRAINWF
e 0
I
12.0
c
2
5.0
15
10
6
VOLUME TREATED
20
-----. 262 25
30
P E R VOLUME OF RESIN B E D
Figure 6. Typical Iealiage curves for fluoride reduction with Amberlite IRA-411
dioxide, and a total concentration of 500 p.p.m. as calcium carbonate was used to evaluate the effluent pH as a function of pounds of sodium hydroxide added. These results are shown in Table 11.
Table 11. NaOH Lb./Cu. $oat 0 0.5
1.0
Influence of Caustic-Salt Regeneration on Effluent pH NaCl, Lb./Cu. Foot 4
3.5 3.0
Efiiuent pH IRA-411 IRA-410 6.6 6.5 10.8 8.5 11.2 10.8
INDUSTRIAL AND ENGINEERING CHEMISTRY
June 1955
1233
24
20
ce
.
W
*6
16
U
Y
a W 4
g
-
00 0 0
I2
In a
LL
0
a
4
k
0 Y
a
Y
i
t 0
4
ea 0
0
4 VOLUME
8
12
TREATED
16
20
24
I
PER V O L U M E OF R E S I N
Figure 7. Typical leakage curves for fluoride reduction by Amberlite IRA-410 Curve A B
C D
Reg. Level, Lb. NaCl/Cu.Ft. Sat. 1 2
Av. Leakage, %
4
4 18 9 6
>-
k L 2 a Y 2
a
Capacity Grains/Cu. 'Ft. 231
io4 217
Sulfide Removal. The data obtained on the use of strong-base anion exchangers for the removal of sulfides are represented graphically in Figures 12 to 14. The results are similar to those reported for fluorides and bicarbonates except for two points: the capacity is markedly dependent upon the influent pH, and the regeneration efficiency is improved greatly by the use of sodium bicarbonate. The capacity a t p H 10 is severalfold that a t pH 7.0. Here again, preliminary studies indicated that capacity was dependent on ion ratio and not ion type.
Figure 8.
Reduction of alkalinity by strong-base ion exchangers
DISCUSSION
The ability to conduct anion exchange reactions economically between anions of markedly different affinity is not too surprising. Of course, capacity values obtained are dependent upon the value of the equilibrium constant. Such constants have been reported ( 7 , 14) for quaternary anion exchange resins. Generally, the porous exchangers are less selective, so that reactions which have appreciable capacity on IRA-400 and IRA-410 would be expected to show a higher degree of efficiency per exchange site for the porous exchange. I n general, this effect has been observed. The linear dependence of capacity on composition ratio follows from the mass action relationship for expressing equilibrium. For the general reaction: 1
R Ar
+ Bs eR Br + As
The mass action equation may be written as follows:
I INFLUENT -
5 0 % NOHCO; 50 X NoCI TOTAL C0NC.- 5 0 0 P.FM. AS CocoI
Figure 9.
I
I
I
Typical leakage curves for deallralization with Amberlite IRA-410 Av.
Curve 1
2 3 4
Reg., Lb./ Cu. F t . Sat. 4 NaCl 3 . 5 NaCl 0.5 NaOH 3 NaCl 1 NaOH
Cr
Cs
=
=
Br As
where Bs is the concentration of anion B- in solution and Br is the concentration of B- on the resin a t equilibrium. Ar and As are concentration of A - anion on the resin and in solution under the same conditions. By rearrangement Equation 1 becomes
Av. Leakage, %
Capacity Kgr./Cu. f t .
8 10 6
4.1 3.5
Effluent pH 7.0 6.0
4
4.8
10.9
4.6
1
cso: 0 0
1
18
(3)
RAr
+
BB
8.5
Cr
Now if Ks
Phth. P.P.M.
Alkalinity,
++ Ar = total column capacity Bs = total electrolyte concentration
=
The expression for the ratio between total column capacity and capacity for a given ion can be derived by substitution of the following terms:
I
1
As
Bs
[As
+ K Bs
>>Bs, Equation 4 reduces to R -= BrCr
KBs
cs
INDUSTRIAL AND ENGINEERING CHEMISTRY
1234
COMPOSITION
RATIO
[ T O T AHL COAi,ON ] X 100
Vol. 41, No. 6
C O M P O S I T I O N R A T I O , [ T o T A ~ c ~ ~ I o100 N]
Figure 11. Alkalinity reduction with Amberlite IRA-411 using caustic-salt reduction
Figure 10. Dealkalization with Amberlite IRA-410 using caustic-salt regeneration
Actually Equations 1 and 5 are equivalent if Bs is small in comparison with As. Figures 1, 2, 8, 10, 11, 12, 13, and 14 show that an empirical relationship can be derived. The straight-line relationship may be used to derive this:
ments, a measurement of the slopes on all graphs will give an indication of relative affinity. For the most part the beds have been regenerated a t a saturation level, so that this condition has been met. The sulfide-chlorideexchange is not well defined, because monoor divalent sulfide ions may enter into the reaction, depending on pH. The following reactions are involved,
+ NaHS e RHS + NaCl (below pH 8.5) 2RC1 + NalS e (R)&3+ 2NaC1 (above pH 10) RC1
The superiority of the porous resins for those exchange reactions involving the monovalent ions, fluoride and bicarbonate, is in This shows that a measurement of the slopes of the capacity lines will give a value which can be related to the equilibrium constant. I n Table 111, the values for slope provided from the column capacity are given for the various exchangers for the different exchange reactions. This table is a convenient way to compare the relative affinities of the resins for various anions. As the correlation between empirical slope and equilibrium constant is valid only for reactions where K -P 1 or where As>> Bs,agreement between reaction content and equilibrium constant can be expected only for the fluoride chloride reaction where As>>Bs. I n no case would K be expected to approach unity. Although only a portion of the data will meet the above require-
Table 111.
RCI
.
n
::",>
+ NoHS+
INFLUENT
-
R H S +NaCI
TOTAL 200 FeM. PS caco,
e'
PH
LL
12
REGENERATION-
s LB.
- 10.0
NDCI~FT.'
Reaction Constants for Anion Interchange Reactions
Reaction Fluoride-ohloride
Bicarbonate-chloride
Sulfide-chloride
Resin Type IRA-400 IRA-401 IRA-410 IRA-411 IRA-400 IRA-401 IRA-410 IRA-411 IRA-400 IRA-401 IRA-410 IRA-411
W/Cr 0.085 0.133 0.123 0.183 0.195 0.310 0.201 0.387 0.382 0.440 0.290 0.250
K Relative 0.35 0.73 0.67 1.00 0.50 0.80 0.67 1.00 0.86 1.00 0.66
0.58
Equilibrium Constant (14)
0.05
o:i3
..
0.32
0:53
.. ..
.... ..
COMPOSITION
Figure 12.
R A T I O , [TOTA:
~N1ON].lOO
Sulfide removal by strong-base anion exchangers
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
June 1955
1235
Table IV. Economics of Anionic Softening of Water with Anion Exchange Resins Water Composition, P.P.M. (as CaCos) HardTotal ness oonon. F HCO: 9 274 373 .. 183 .. 160 230 10 .. .. 150 380 30
Reg. Level Cost0 Lb. NaCl/ Cents/1000 cu. ft. gal. 4.0 10.7 1.0 5.7 4 . 5 NaCl 21.0 S 0 . 5 NaOH 0 Based upon cost of salt a t 1 cent per pound and NaOH at 3 cents per pound. These values are estimates from actual installations.
..
*
..
Anion Removed HCOI F
agreement with previous studies (8) on the effects of porosity and degree of cross linking upon the equilibrium constant and distribution coefficients. As the distribution coefficients of most monovalent exchange systems tend to approach unity a t high porosities or low degrees of cross linking, the adsorption of these ions that are usually adsorbed with difficulty a t low porosities is increased.
Capacity Gal./Cu.’ Ft. 380 175 284
Amberlite Resin IRA-410 IRA-411 IRA-410
14
I2 o mI-. \
00 10 V
In 8 L’
y” ).
E
0
6
2
3 w s (L
-I 3
-
2
0
0
20
40
COMPOSITION
20
0
40
60
BO
100
C O M P O S I T I O N RATIO [ T O T A ~ S A N , O N )
Figure 13.
100
CaCOl
for removing objectionable anions from water supplies is the low capacity, in terms of anions removed, that is realized under certain conditions. However, as anions such as fluorides and sulfides are present in only trace quantities, the capacity, in terms of gallonage treated, is high. The fact that the operation can be conducted in equipment that need not be acid-resistant and with a technique as simple as water softening makes the method extremely attractive. I n addition to the simplicity of the method, the economics (see Table IV) of the regeneration are favorable. ACKNOWLEDGMENT
The authors wish t o acknowledge the assistance of Mrs. Paul Klaas during much of the experimental work. NOMENCLATURE
As Ar Bs Br
100 *
100
Figure 14. Effect of regeneration with sodium bicarbonate on sulfide capacity of Amberlite IRA-400
Cs = total electrolyte concentration K = mass action equilibrium constant K1 = empirical reaction constant LITERATURE CITED
A disadvantage in the use of the quaternary anion exchangers
-
80
Effect of pH on sulfide capacity of Amberlite IRA-400
Influent. NaCl, NazS 4- HzS, 200 p.p.m. Regeneration. 5 lb. NaCl per cu. ft.
&
60
R A T I O [,o,A,”sA,,,oN]
= concentration of anion A in solution = concentration of anion A on resin
= concentration of anion B in solution = concentration of anion B on resin Cr = total column capacity
Pub. Health Assoc., New York, “Standard Methods for the Examination of Water and Sewage,” 8th ed., Section 24,
(1) Am.
p. 167,1936.
(2) Churchill, H. V., IND. ENQ.CHEM.,23, 996 (1931).
(3) Goodwin, R.C., and Litton, J. B., Ibid., 33, 1046 (1941). (4) Jacobs, M. B., “Chemistry and Technology of Food and Food Products,” Vol. 2, p. 408,Interscience, New York, 1944. (6) Rittredge, A.,Proc. Am. Power Conf., 15, 567 (1953). (6) Kittredge, A.,Trans. 13th Water Conf. (Pittsburgh), 1953. (7) Kunin, R.,and McGarvey, F., IND.ENG.CHEM.,41, 1265 (1949). (8) Kunin, R.,and Myers, R. J., Discussions Faraday Soe., 7, 114 (1949). (9) Kunin, R., and Myers, R. J., J . Am. Chem. SOC., 69,2874 (1947). (10) Maier, F. J., Am. J. Pub. Health, 37, 12 (1947). (11) Rowley, R.J., and Churchill, H. V., IND.ENQ.CHEM.,ANAL. ED., 9, 551 (1937). (12) Smith, M., Lants, E., and Smith, H, Arizona Agr. Expt. Station, Tech. Bull. 32, 354 (1931). (13) Thompson, J., and McGarvey, Fl, Power, 94, 100 (1950). (14) Wheaton, R., and Bauman, W., IND.ENG. CHEM.,43, 1088 (1961). R ~ C E I V Efor D review February 21, 1951. ACCEPTED December 30, 1964. Presented before the Division of Colloid Chemistry, Symposium on ApplicaCHEMIUAL tions of Ion Exchange, at the 118th Meeting of the AMERICAN BOOISTY, Chicago, Ill., 1950.
