Ion exchange processes in aqueous-organic media

zyxwvu zyxwvu zyxwvutsrq zyxwvutsrq Ion Exchange recesses in Aqueous- rganic Media J. L. Pauley, D. D. Vietti, C. C. Qu-Yang, D. A. Wood, and R e D. Sherrill Department of Chemistry, Kansas State College of Pittsburg, Pittsburg, Kans. Alkali metal ion exchange on Bio-Rad AG 50W X-1 resin in 0, 5, 10, 25, 50, 75, and 90% methanol-water, ethanolwater, and dioxane-water and alkali metal barium ion exchange in acetic acid-water systems were investigated at 25 OC. In part, observed changes in selectivity coefficient with solvent changes can be explained on the basis of dielectric constant effects on coulombic interactions, but the observed maxima or minima and changes in order of selectivity as the water content of the solvent is decreased can be accounted for by the electrostatic model only if it is assumed that ion solvation also changes in an appropriate manner. Some correlation between selectivity data and swelling data was noted. Radiotracer techniques were used to determine alkali metal exchange coefficients and gravimetric techniques for exchanges involving barium. From the solvent distribution data, there appeared to be no selective solvation of the resin phase. THEREhave been a number of studies of ion exchange in organic and mixed water-organic solvents, in recent years in an attempt to deduce principles determining ion selectivity (1-5). Mixed solvents have been preferred because, as a rule, equilibrium is reached very slowly, if at all, in strictly nonaqueous media. The solvent systems employed in the present investigation-methanol-water, acetic acid-water, ethanol-water, and dioxane-water-were chosen to provide a reasonable variation in solvent properties. An exchanger with a low degree of cross linkage was used to facilitate attainment of equilibrium in regions of low water content where resin swelling was limited. EXPERIMENTAL Materials. Solutions were prepared using reagent grade solvents and ion-free water. The dioxane was redistilled from Na2C03to remove any acid present. The potassium, sodium, lithium, barium, and cesium salts were all reagent grade and were dried under vacuum before use. The radioactive 22Naand 13*Cswere carrier free in the chloride form. The exchanger used was Bio-Rad AG 5OW-X1 (50-100 mesh) obtained in the hydrogen form. The resins were converted to the appropriate salt form with the hydroxide or carbonate of the desired salt as convenient. Resin capacities were determined by adding excess base and back-titrating the excess. Capacities were calculated as meq/dry g of the resin in the salt form. The Cs and Na form resins were spiked with carrier free 22Naor 131Cs (activity approximately lo6 cpm/g). The salt form resins were dried under vacuum at 120 "C before use. This drying did not lead to measurable changes in exchange capacity for the resins studied. Equilibrium Systems. Except for the barium exchanges, weighed samples of approximately 1 g of the dry spiked exchanger in the sodium or cesium form as appropriate were equilibrated at 25 "C with 30 ml of an approximately 0.1Nsolution of the chloride of the metal ion to be exchanged. When equilibrium had been reached, aliquots of the solution phase were taken to determine the radioactivity and thus the concentration of the sodium or cesium ions eluted from the resin phase. Attainment of equilibrium was assumed when further statistically significant change in the activity of the solvent phase could not be detected. An additional 24 hours beyond this time was allowed to assure that equilibrium was reached. The resin was then eluted with HC1 to determine the activity remaining on the resin to assure that an activity balance was maintained. For the barium exchanges reported, approximately 2 g of the resin in the barium form was equilibrated with 25 ml of approximately 0.1M solutions of the acetate of the metal ion to be exchanged. The barium ion eluted was determined gravimetrically as the sulfate. Differing batches of resin were used for the different solvent systems. Thus, only qualitative comparisons of a given exchange in going from one solvent system to another can be made. Comparison of changes in selectivity with change in water content for exchanges in a given solvent system and relative exchange coefficients of the ions for any given solvent system should, however, be valid. Selectivities. Selectivities calculated corresponded to the exchange A+ B*R = AR B*+ where A+ was sodium, potassium, lithium, or cesium and B* was tagged sodium or cesium. Selectivity coefficients were calculated according to (AIR, meq/dry g of salt form resin) (I%** cpmiml) KBA = (BR, cpm/dry g of salt form resin) (A+ meq/ml) + + The normalities of the salt solutions, A-, were calculated by difference from the original normalities and the concentration of radioactive ion following exchange. For the exchange in acetic acid-water solutions, selectivity coefficients were calculated in essentially the same way, except that the alkali metal ion concentrations appear squared because the barium ion is divalent. Also, because exchange of the €3' of acetic acid for the resin ions occurred to a limited extent (less than 2 % in all cases), the capacity of the resin was corrected for any hydrogen ion exchange. Solvent Distribution. Solvent distribution between the resin and solution phases was obtained by equilibrating the appropriate salt form of the resin with a slight excess of the particular solvent system above that needed to swell the resin and noting any changes in refractive index of the solvent. This was checked in some cases by equilibrating the resin with the solvent mixture, blotting the excess solvent, and distilling over the solvent retained in the resin phase. The composition was then determined from its refractive index. Swelling. A weighed amount of the dry resin in the appropriate salt form was equilibrated with the solvent mixture in a stoppered vessel. The swollen resin was then filtered, patted dry of excess solvent, and weighed. Swelling was calculated in terms of grams of solvent per milliequivalent of dry resin. This procedure does not lead to reproducible results in general in going from one resin batch to another, probably reflecting small differences in the degree of cross linking of the resins. Similarly, because a judgment is involved in the degree of dryness involved in removing excess solvent, the results obtained by different irrvestigators are not usually in complete agreement. Far the data presented here, for each solvent system the resins have the same batch number and were done by a single investigator. Thus, relative values for a given solvent system should be valid and permit valid comparisons of relative solvent swelling effects. zyxwvut (1) A. Ghodstinat, J. L. Pauley, Teh-Hsuen Chen, and M. Quirk, J. Phys. Chem., 70, 521 (1966). (2) R. G. Fessler and H. A. Strobel, ibid., 67, 2562 (1963). (3) R. Gable and H. Strobel, ibid.. 60, 513 (1956). (4) Yu I. Ignatov and N. A. Izamilov, Zhur. Fiz. Khim., 39, 2482 (1965). ( 5 ) A. Davydov and R. Skoblionok, ibid., 32, 1703 (1958). ANALYTICAL CHEMISTRY, VOL. 41, NO. 14, DECEMBER 1969 e 2047 I .2 A n * v r\ h U zyxwvutsrq zyxwvutsrqpon zyxwvut zyxwvu zyxwvuts zyxwvutsrq Q 8 0.4 0 0 Figure 1. Selectivity coefficient us. solvent composition for alkali metal exchange in dioxane-water mixtures at 25 "C KLicS = (D, KNacs = 0,KXcs = X, KNapqa = 0, xZipqa = A, K ~ N , = B9KCsva = V RESULTS AND DISCUSSION In general, the data show the same trends for each of the solvent systems in Figures 1-3. The self exchange coefficients are included to give an estimate of the reliability of the results. The data of Figure 4 are not compared as conveniently but show the same general characteristics. In all cases, the selectivity coefficients for exchanges involving lithium for another ion change in a nearly h e a r fashion with decreasing water content of the solvent. For the other exchanges, changes with solvent composition do not follow a simple pattern. In general, the net increase in selectivity coefficient with decreasing water content for the K-Na and Cs-Na exchanges and the decrease for the Li-Cs, Li-Na,'K-Ba, Na-Ba, and Li-Ba exchanges are as would be expected for the decrease in dielectric constant accompanying a decrease in water IO 2 0 30 40 5 0 6 0 7 0 Weight % Methanol 80 90 Figure 2. Selectivity coefficient GS. solvent composition for alkali metal exchange in methanol-water mixtures at 25 "C = 0, KLiN, = 8, KKN, = m, KCBN~= V content of the solvent (6-8). The change in the relative selectivity noted for the K-Na and Cs-Na exchanges in Figures 1 and 2, and the K-Cs exchange in Figures 1 and 3, and the maxima or minima observed for several of the exchanges as the water content of the solvent is decreased cannot be explained satisfactorily on the basis of dielectric constant alone. At very low water contents, failure to attain equilibrium may affect the results as suggested by the Na-22Na and Cs-la4Cs exchange data in Figures 1 and 3, but this cannot account for the observed effects where the water content is higher than about 20-25 %. Both selectivity coefficients and the logarithms of the selectivity coefficients have been plotted (6) J . L. Pauley, J. Amer. Clzem. SOC.,76, 1422 (1954). (7) G. Eisenman, Biophys. J., 2, 259 (1962). (8) G. Panchenkov, V. Govshkov, and M. Kulanova, Zhur. Fiz. Khjm., 32, 361 (1958). 70 t 50t \ . I 0 zyxwvutsr zyxwvut zyxwvut I I I I I I I I IO 20 30 40 50 60 70 80 Weight % Ethanol 90 Figure 3. Selectivity coefficients us. solvent composition for alkali metal exchange in ethanol-water mixtures at 25 "C K 2048 a, N = ~8, K~ : ~J =~ ~ K~ x, Figure 4. Selectivity coefficients us. solvent composition for alkali metal exchanges with barium ion in acetic acidwater mixtures at 25 "C o K = ~ ~ K C B=~ ~ ANALYTICAL CHEMISTRY, VOL. 41, NO. 14, DECEMBER 1969 K K ~ =, 0, KNag, = a, K L ' ~ =, B 0 zyxwvutsrqpo zyxwvutsr IO zyxwvuts zyxwvutsr 20 40 30 50 70 60 Weight % Methanol 80 90 Figure 5. Solvent uptake us. solvent composition for sodium form resin in 0.1M methanol-water solutions of alkali metal ions m, -s+ Lif = 0, Na+ = 8, K+ = A b I I I , I I 40 50 60 70 80 1 IO 20 30 i 90 Weight % Methdnol Figure 6. Solvent uptake us. solvent composition for alkali metal resinates in methanol-water mixtures zyxwvutsr zyxwvuts LiR against varying functions of concentration of solvent including mole per cent, volume per cent, and dielectric constant ( I , 5,8). None of these approaches appears to be significantly more useful than the plots presented here in suggesting any simple interpretation of the data. Several models have been proposed which will suitably explain the observed changes in selectivity coefficient with solvent if certain arbitrary assumptions are made (6-11). Unfortunately these assumptions cannot be readily verified by separate independent measurements. The electrostatic model proposed in earlier studies of water-DMF solutions (1) is not necessarily more suitable than others, but does have the advantage of simplicity. If this model is assumed, two factors are involved in the determination of selectivity coefficientsthe dielectric constant of the solvent and the solvated ionic radii of the competing ions. The decrease in dielectric constant as the water content of the solvent is decreased would be expected to result in a regular increase of selectivity for the preferred ion. But, if it is assumed that ionic solvation also changes appropriately with changes in solvent composition, the observed maxima or minima and changes in order of selectivity noted for K and Cs in Figures 1-3 would be accounted for by this model. = 0, hTaR= V, KR = El, CsR = A Swelling data (Figures 5-9) are not inconsistent with this model. In Figure 5, the exchange of the sodium ion of the resin by lithium in methanol-water mixtures results in an increase in solvent uptake as might be anticipated if the lithium ion is more highiy solvated. The changes in solvent uptake for other exchanges in which the sodium is replaced are also consistent with generally accepted views as to relative solvation of ions. The swelling data for the various ionic forms of the resin in the methanol-water, ethanol-water, and acetic acid-water systems in the absence of added salts (Figures 6, 7, 9) shows a reversal of swelling order for the cesium and potassium forms as the water content of the solvent is decreased, suggesting changes in the relative hydration of the two ionic forms of the resin. The inversion of swelling order for potassium and cesium resins in the ethanol-water and methanol-water systems parallels the inversion of selectivity for these ions below about 7040% water content of the solvent. For the dioxane-water system (Figure s), there is no reversal of order of solvent uptake for the potassium and cesium resins corresponding to the observed reversal of order of selectivity for these ions. However, differences in solvent uptake for the potassium and cesium forms is small over the total range of solvent compositions. For each of the systems studied, there was a marked decrease in solvent uptake for all forms as the water concentration was decreased. The swelling data for the acetic acid-water system show no simple relationship between swelling and selectivity coefficients (Figure 9). No reversal of selectivities was noted corresponding to zyxwvutsrqp zyxwvutsrq (9) S.Linderman, J. Phys. Chern., 70,814 (1966). (10) J. A. Marinsky, ibid.,71, 1572 (1967). (11) H. S. Frank and Wen-Yang Wen, Disc. Faraday Soc., 1957, 133. I .o 08 g Solv 06 per me4 Resin 04 02 O L 2b IO zyxwvutsr I 30 40 510 60 Weight % Ethdnol 80 i 0 910 Figure 7. Solvent uptake us. solvent composition for alkali metal resinates in ethanol-water mixtures LiR = 0, NaR = 8, KR = a, CsR = A 0 . I I I I I I I IO 20 30 40 50 60 70 Weight % Dioxane 90 80 Figure 8. Solvent uptake us. solvent composition for alkali metal resinates in dioxane-water mixtures LiR = 0 , NaR = V, KR = El, CsR = h ANALYTICAL CHEMISTRY, VOL. 41, NO. 14, DECEMBER 1969 0 2049 zyxwvutsrqp zyxwvutsrq zyxwvut zyxwvut zyxwvuts c zyxwvutsrqpon the reversal of swelling orders at about 3 5 x acetic acid for potassium and cesium resins, and 6 0 x acetic acid for the potassium and sodium resins. It was noted, however (Figure 4), that differences in relative selectivities of the resin for each of the ions, compared to barium, becomes smaller as the water content is decreased and the greatest decrease in potassium selectivity, compared to lithium and sodium, occurs near the region where reversal of swelling orders is first noted (12). Again, the observed data can be explained on the basis of the electrostatic model if appropriate changes in relative hydrated radii with changes in solvent composition are assumed. In all the systems studied, the various salt form resins showed little or no selective solvent uptake. This was somewhat surprising although in agreement with the observations of Reichenberg and Wall for resins with a low degree of crosslinking (13). The method chosen to determine differences in solvent composition between the resin and solution phases should have been adequate to detect composition changes of the order of 2-3 %. Thus it would appear that any significant changes in solvent composition due to selective solvent uptake should have been noted. CONCLUSIONS From this and other investigations of ion exchange in mixed solvent systems, it is obvious that changes in the solvent system have a marked effect on relative ion selectivities. This should have significant consequences for the use of ion exchangers for analytical separations. It is also apparent that quantitative prediction of these solvent effects is quite complex. As might be anticipated, dielectric constant effects appear to be significant, but any comprehensive explanation of effects must include a consideration of ionic solvation, solvent-solvent interactions, solvent-resin interactions, sol- 0.40.75 0 0.60 60 0 g Solv 045 per meq Resin 0 30 I 0 10 I I I I 20 30 40 50 , 60 , 70 I 80 I 90 100 Weigh+ % Acetic A c i d Figure 9. Solvent uptake us. solvent composition for alkali metal resinates in acetic acid-wafer mixtures LiR = 0 , NaR = V, KR = II, BaR = h vent structure changes and, at least for more highly crosslinked resins, the degree of cross-linking. Correlations between selectivities and resin swelling appear attractive but, again, there appears to be no simple correlation. Several of the proposed models for prediction of selectivity are capable of offering qualitative explanations of observed effects, but it would appear that no completely satisfactory model for quantitative predictions is yet available. RECEIVED for review July 22, 1969. Accepted September 24, 1969. The partial support of the Kansas Academy of Science for this research is gratefully acknowledged. Taken in part from dissertations presented to the graduate school of Kansas State College of Pittsburg in partial fulfillment of requirements for the Master of Science Degree. Presented in part to the Division of Physical Chemistry of the ACS Southwest Regional Meeting, Little Rock, Ark., December 1967. R.D. Sherrill is a National Science Foundation Undergraduate Research Participant. zyxwvut zyxwvuts zyxwvut (12) Zh. L. Vert and G. P. Grinberg, Zhur. Obshei Khim., 24, 953 (1954). (13) D. Reichenberg and W. Wall, J. Chem. Soc., London, 1956, 3364. 2050 e ANALYTICAL CHEMISTRY, VOL. 41, NO. 14, DECEMBER 1969