858
The Journal of Physical Chemistry, Vol. 82, No. 8, 1978
Y. Marcus and J. Naveh
Swelling of Anion Exchangers in Mixed Solvents. 4. Swelling and the Invasion of Lithium Chloride Y. Marcus" Department of inorganic and Analytical Chemistry, The Hebrew University, Jerusalem, Israel
and J. Maveh Israel Atomic Energy Commission, Nuclear Research Center, Negev, Israel (Received October 28, 1977)
The invasion of lithium chloride into the anion exchanger Dowex-lX8-Cl from, and the coincident swelling of the resin in, aqueous methanol, ethanol, l-propanol, and acetone has been measured. The data are compared with the invasion of hydrogen chloride in similar systems, and with the swelling in the absence of electrolyte, in order to elucidate the effect of the solvent composition on the invasion, and of the invasion on the total and selective swelling. Some data on the swelling in aqueous 2-propanol and allyl alcohol are also shown. It is concluded that the extent of ion pairing of the invading electrolyte, and of the fixed ion of the exchange site with the counterion, is the main factor determining the invasion and the swelling in the presence of the electrolyte.
Introduction Several years ago we published studies on the total and selective swelling1 of anion exchangers in aqueous organic solvent mixtures, on the densities2 of these swollen resins, and on the heat of ~welling.~ In order to complement the picture of the interactions occurring among the components of the systems that evolved in these studies, new results on the swelling of the anion exchanger in aqueous ethanol and propanol as a function of the temperature and in some further solvents (2-propanol and allyl alcohol) are reported here. More importantly, the invasion of the anion exchanger with an electrolyte from aqueous organic solvent mixtures, and the effect of the electrolyte on the total swelling and the solvent selectivity in swelling, have been determined. This information provides a means to better understand the interactions in the anion exchange resin-water-organic solvent-electrolyte systems. In so far as information on the behavior of the electrolyte in the aqueous solvent is known independently, its interactions inside the resin phase can be elucidated from the invasion and swelling data. In a recent study, Kim4 has investigated the invasion of hydrogen chloride into an anion exchanger from aqueous methanol, ethanol, and 1-propanol, and its effect on the swelling of the exchanger. Except for ethanol, the data were obtained at one electrolyte concentration only. In the present study, lithium chloride is used rather than hydrogen chloride, in order to avoid possible complications from hydrogen bonding of the hydrogen ions to chloride ions or hydrogen chloride molecule^.^ Three concentrations of electrolyte are investigated, and data for aqueous acetone solutions are added, which show rather unique features. Lithium chloride is selected mainly on the basis of its solubility in the solvent mixtures, and since na ion exchange, but only electrolyte invasion is to be studied, this dictates the use of the exchanger in its chloride form. The same batches of resin employed in our previous studies'-3 are used here, namely, an 8% cross-linked polystyrene-methylene-trimethylammonium type resin, Dowex-1, for the bulk of this study, and 4% cross-linked resin for some of the experiments. Experimental Section and Calculations Analytical grade Dowex-1x8, 20-50 mesh size, was employed, subjected to the following treatment before use: 0022-3654/78/2082-0858$01 .OO/O
it was washed in a column with a large excess of 1 M sodium hydroxide, water, 1 M hydrochloric acid, water, ethanol, 1M hydrochloric acid, and water, and then it was dried in a vacuum desiccator over phosphoric acid anhydride for several days. The capacity of the resin was determined by treating weighed portions with an excess of 1 M sodium perchlorate, and tit_rating the exchanged chloride ions. It was found to be C = 3.63 f 0.03 equiv (kg dry resin)-'. For a few experiments (swelling in aqueous 2-propanol) a similar resin of 4% cross-linking, having a capacity of 3.90 f 0.03 equiv (kg dry resin)-', and subjected to the same treatment, was employed. The solvents used were of analytical grade. The water contents of the nominally neat solvents were obtained by Karl-Fischer titration and taken into account in the preparation of aqueous solvent mixtures by accurate weighing. However, no effort was made to obtain 100% dry solvents, (thus, the x, = 1point was approached but not attained, x, being the mole fraction of the solvent in the external solution). A concentrated stock solution of lithium chloride in water was made by weighing the dried salt and water, and calibrated by titration. Proper aliquots were taken and mixed by weight with the aqueous solvents, so that the initial mole fraction of the solvent xi, varied in steps of 0.100 (accurate to f0.001) from zero to unity, and the nominal molality of the lithium chloride was miLicl = 0.010, 0.10, and 1.0 m. For the swelling experiments without lithium chloride, a known weight of dry resin wi7 was equilibrated with ca. 50 times its weight, wit of the aqueous solvent mixture. For invasion and swelling experiments, only a threefold excess by weight of the aqueous solvent solution of lithium chloride was used. The mixtures were shaken at room temperature, 22 f 1 " C , for 24 h in a glass vial. Some experiments were carried out in a thermostated bath a t different temperatures. The resin was then rapidly and quantitatively transferred to a glass tube with a sintered frit bottom, which was stoppered and centrifuged for 15 min at 5700 rpm. In those cases where the equilibrium temperature differed from room temperature (Table 11), the transfer and separation of the resin from the excess solvent were so rapid that the displacement of equilibrium was presumably rather small. In all cases, the swollen resin was weighed (wfr),and the density of the centrifugate ( p f ) was measured in a picnometer to better than 0.01 %. The 0 1978 American Chemical Society
The Journal of Physical Chemistry, Voi, 82, No. 8, 1978 859
Swelling of Anion Exchangers in Mixed Solvents 12
I
I
1
l
1
12 -
0.8-
I
0.6-
OD
0.2
0.4
0.8 x, 1.0
0.6
Figure 4. The swelling of Dowex-1x841 in aqueous 1-propanol containing LiCI. Curves as in Figure 2, (a) and (b) as In Figure 1. Flgure 1. The swelling of Dowex-1x8-CI in aqueous 1-propanol ----, 2-propanol (X4 rather than X8) -, allyl alcohol ----, and acetone (a) Absolute swelling, mol of H,O/equiv of resin = ii,, mol of solvent/equiv of resin = ii, against x,. (b) Selective swelling, R, vs. x,.
._..-..-...
OlOm LICL
O O l O m LiCl
10m LiCl
12
i1
Figure 2. The swelling of Dowex-1x841in aqueous methanol containing no LCI -, 0.010 m LiCl - - -, 1.O m LlCl -. (a) and (b) as In Figure 1.
--
12
I
1
I
I
00
05
~
1 0 00
X,
0 5 Xs
00
10
05
0
Xr 10
Figure 5. The invasion of lithium chloride into Dowex-1x8-CI from aqueous acetone -, 1-propanol ---, ethanol -----, and methanol ---. The function log F = log Y+*/?+~ = log (@,fi,,,/mL,,,2) against x,. The LiCl concentrations in the three sections are the nominal external concentrations.
a
-
0
08-
0.6-
0.0
02
0.L
0.6
O.SX,l.O
00
0.2
04
0.6
O.BX,l.O
Figure 3. The swelling of Dowex-1x8-CI in aqueous ethanol containing no LlCl -, 0.010 rnLiCl ---, 0.10 mLiCl -----, 1.0 mLiCl ---. (a) and (b) as In Figure 1.
density of the initial lithium chloride solutions pi (x8, miLicl) was also similarly measured. The lithium concentration cfLiain the equilibrium external solution was determined by atomic absorption spectrophotometry to better than 0.5%. From these data, all the necessary information is obtained. The density of the initial solution is fitted to the following empirical expression: -
Pi(Xs, miLiCd =
(1 + AmiLicl)(pw + BxiS+ CdS2+ DxiS3)
I
(1)
where pw is the known density of water, 0.9977 g ~ m - A~ , = 0.023 kg mol-' is almost solvent independent, and produces anyway only a small correction, and B, C, and
0.0 0.0
I
I
I
02
0.4
0.6
1
o.sx,io
on 00
I
02
I
OL
06
o.ax,~.o
Flgure 6. The invasion of lithium chloride into Dowex-1x8-CI from aqueous I-propanol -, ethanol - - -, and methanol (invasion of HCI from aqueous ethanol4 Shown is the interior molality of Invading lithium Ions, mu (a), and the molar ratio of invading llthlum Ions and resin exchange sffes, ii, (b), against x,, for a nominal external concentration of 1.O m LiCI.
- --
---sa-).
D are solvent-specific parameters. For a given aqueousorganic solvent with known parameters, the equation can be inverted, and xf, (mfLic, pf), the mole fraction of the solvent in the external solution a t equilibrium, can be
860
The Journal of Physical Chemistry, Vol. 82, No. 8, 1978
Y. Marcus and J. Naveh
TABLE I: Sample Calculations of Derived Quantities for the System Dowex-1x8-Water-Ethanol-0.1m Licl from Experimental Dataa XS
Pi. m'LiCl
wi W'
wt
0.000 0.9982 0.0785 12.012 4.007 7.007
1.0000
c Licl
0.0961
0.000 8.976 0.00 11.46
XfS
W'f ns
0.000
XS
z f ~ i c l 0.0965
nLi mc1
r Y* -
Yi:
0.0244 4.87 12.76 0.777 0.218
0.100 0.9628 0.1023 12.003 4.009 6.990 0.9636 0.1268
0.200 0.9322 0.1035 12.015 4.009 6.886 0.9294 0.1199
0.300 0.9044 0.1013 12.017 4.008 6.881 0.8964 0.1111
0.1028 12.007 4.009 6.740 0.8678 0.0994
0.100 8.986 0.099 8.86 0.101 0.1324 0.0112 4.89 3.13 0.707 0.400
0.208 9.101 1.52 7.10 0.176 0.1298 0.0200 5.08 5.98 0.654 0.268
0.330 9.107 1.76 6.46 0.214 0.1247 0.0268 5.09 8.78 0.596 0.201
0.459 9.239 1.82 5.78 0.240 0.1152 0.0607 5.39 24.7 0.531 0.108
0.400
0.8800
0.500 0.8586 0.1020 12.015 4.008 6.549 0.8462 0.0910
0.600 0.8432 0.1014 12.000 4.001 6.377 0.8298 0.0826
0.700 0.8262 0.1032 12.021 4.006 6.217 0.8160 0.0802
0.800 0.8116 0.1040 12.011 4.009 6.159 0.8050 0.0814
0.900 0.8000 0.1048 12.020 4.002 6.168 0.7964 0.0910
1.000 0.7874 0.1053 12.018 4.002 6.315 0.7882 0.0941
0.575 9.436 1.92 4.79 0.286 0.1079
0.677 9.587 2.11 3.71 0.363 0.1000 0.1071 6.22 66.9 0.439 0.054
0.775 9.771 2.22 2.77 0.445 0.0988 0.1227 6.70 84.0 0.392 0.043
0.861 9.822 2.49 1.85 0.574 0.1016 0.4150 6.88 76.5 0.347 0.040
0.935 9.815 2.88 0.92 0.768 0.1150 0.0584 6.77 29.7 0.300 0.055
0.998 9.666 3.46 0.02 0.994 0.1202 0.0428 6.21
0.0800 5.81 39.8 0.484 0.077
18.8 0.265 0.061
a The upper part contains the direct experimental data, the lower part the derived quantities. The parameters A = 0.0236, B = -0.394, C = 0.273, and D = -0.090 were obtained from eq 1.
log
0010 m LiCL 10
- ----_ _ _ _
0
-
05-
-10
-
,**
,' 10m L C l 01
-- - - _ _ _
-
00
02
04
06
08
x,
01
10
Figure 7. The activity coefficient y+ of lithium chloride invading Dowex-lX&CI from aqueous propanol -, ethanol ---, and methanol asainst x.. The LiCl concentrations on the curves are the nominal extern2 conceitrations. The circles are calculated for invading hydrogen chloride from aqueous ethanoL4
----
calculated iteratively, to an accuracy of f0.002. In Table I1 and Figures 1-8 this final equilibrium composition of the external solution is designated simply by x,. In the absence of lithium chloride the composition of the aqueous solvent in the resin phase f ais obtained as follows. The weight of organic solvent initially in the solution is wis = xi,wi/(xis+ (1 - xis)18.01/M3),where M s is the molecular weight of the solvent. The weight of the external solution at equilibrium is wf = wi + wir - wfr,and the weight of the solvent in it is wf, = xfswf/(xfs + (1 xfs)18.01/Ms). Hence the amount of organic solvent that swells the resin is wig - wfs, and the number-of its moles that swells 1 equiv of resin (there being wirC/103equivalents of resin in the mixture) is
n, = 103 (wiS- w f s ) / ~ , w i r C
-15-20
-02
(2) The corresponding quantities wiw and wfw for water are obtained by difference, leading to tiw and finally to Zs = ris/(ris + riw). For solutions containing lithium chloride, the initial amount of water and solvent w/ is obtained from w/ = w i / ( l + 0.0424rniLicl), while the equilibrium amount kg mor1 is obtained from w i = wf(l - 0.0424cfLic1pf),0.0424
-1;
-110
doiwa
015
O'S
1'0
Flgure 8. The function log B = log (mcly*) against log a = log (mHCl(ot L,CI)~I) for Dowex-1x8-CI for invasion of hydrogen chloride (heavy curves) and lithium chloride (thin curve) from water -, methanol - - - (curve arbitrarily normalized to log B = 0.30 at log a = o.oo), x , = o.55 aqueous ethanol and ethanol _ _ _ _ ,F~~ sources of data, see text,
-..-..-,
being the molar weight of lithium &loride. The equilibrium molality of the lithium chloride is mfLiCl= cfLicl/(pf- 0.0424cfLicl) (3) The invasion of the lithium chloride is obtained as follows. The molality of lithium ions in the resin is
-
mLi = (miLiclw[ - mfLiClwfr)(wi, - wfy) the molality of chloride ions is
(4)
mcl = mLi + wi,C/(wi, - wfr) The function r is defined as
(5)
-
-
r = ELiZcl/mfLic12
(6) Both mLi ( x s , miLicI)and I? (x,, miLic1)are convenient measures of the invasion. Results An example of the calculation of the various relevant quantities from the experimental data xis,pi, wi, m'LiC1, wIr, wf,, pf, and cfLiclis shown in Table I for the case of aqueous ethanol and 0.10 m lithium chloride. The derived quantities are shown in Table I1 and in Figures 1-6 for all the systems studied.
The Journal of Physical Chemistty, Vol. 82, No. 8, 1978 861
Swelling of Anion Exchangers in Mixed Solvents
TABLE 11: The Temperature Dependence of the Interior Composition 2,of Aqueous Ethanol and 1-Propanol Swelling the Anion Exchanger Dowex.lX8 Ethanol 1-Propanol TtmP, x , = 0.15 X, = 0.05-0.06 x, = 0.10 X, = 0.12-0.14 C x, = 0.05 x, = 0.10 4 7 23 45
0.096
0.141
0.150
0.209
0.190 0.270
0.096 0.169 0.220 0.227
75
Table I1 shows the composition of the swelling solvent for a resin swollen a t various temperatures from dilute aqueous ethanol or 1-propanol. The dilute solutions were reinvestigated because this is the region where the water structure was found1 to play the major role. The absolute swelling by solvent A, (x,) and by water it, (x,), and the selective swelling R, (x,), were obtained for 2-propanol, for 2-propen-1-01 (allyl alcohol), and for acetone: and are shown in Figure 1. The general behavior is very much like that previously obtained' for 1-propanol, but there are some differences in the behavior of the various mixtures. The presence of lithium chloride has notable effects on the swelling, both the absolute and the selective, which naturally increases with the electrolyte concentration, as shown in Figures 2-4 for methanol, ethanol and 1propanol. (For acetone, the only concentration employed, miLicl= 0.01 m, was too small to give important effects.) Particularly noteworthy is the increase in swelling by water in aqueous propanol as lithium chloride invades the resin, while the swelling by the propanol itself is hardly affected (Figure 4a). The invasion of lithium chloride into the resin is depicted as log l? (x,,miLic1)in Figure 5, and noteworthy is the steep increase in this function as x, increases for propanol and acetone (it is less steep for ethanol), and the maximum exhibited in the ethanol and propanol systems (in methanol a very shallow maximum is seen). The molality of the invading chloride in the resin mLi and the number of moles of invaded lithium chloride per equivalent of resin, Ab, are shown in Figure 6 for an external molality mfLiClnear unity, for comparison with the results of Kim4 for hydrogen chloride. The actual equilibrium external molality varied from 1.03 to 1.12 for methanol, from 0.84 to 1.15 for ethanol, and from 0.71 to 1.12 for propanol, due to the removal of water, solvent, and salt into the resin. If the activity coefficient of the electrolyte in the resin yr is defined appropriately,6 the Donnan law yields
R,
mLimC1 Y '.i
0.261 0.245 0.246 0.203 0.205
0.185 0.195
78
- - -
0.091 0.115 0.198
= mfEC:
Y**
(7)
from which the expression of I' = yh2/yh2follows, or
Y* = y*r-l l2
(8)
Here the 7's are the mean stoichiometric ionic molal activity coefficients, disregarding any solvation or association (ion pairing). In eq 7, the resin-phase activity coefficient includes a contribution from the pressurevolume term exp(.lrbLic1/2RT),where a is the osmotic pressure, and bLic1 the molar volume of lithium chloride in the resin. This term is quite sma114v6and neglecting it does not affect appreciably the meaning of this activity coefficient. Activity coefficients of lithium chloride are available for solutions in the neat solvents methan01,~-~ ethanol,lOJ1 l-propanol,1° and acetone12 (and for mixtures of methanol
and acetone13 that permit extrapolation). However, no data are available for solutions in the aqueous-organic solvent mixtures, These can be estimated from analogous data for hydrogen chloride solutions, which are available for aqueous methanol,14J5ethanol,l"l7 and l - p r ~ p a n o l . ' ~ ~ ~ ~ These data can be expressed as y+(m~cl, x,) which are in all cases linear with x, to a sufficiently good approximation (f3%).20 A similar linearity is therefore assumed also for lithium chloride, which in water and the neat solvents behaves very similarly to hydrogen chloride. From such calculated values of y+ (x,, mLicl),the corresponding values for the activity coefficient of the electrolyte invading the resin phase, y+, were calculated, using the r values and eq 8, and are shown in Figure 7.
Discussion Swelling in the Absence of Invasion. In our previous publication' we pointed out that the preferential swelling of the anion exchanger by alcohols at low concentrations is due to water-structure effects. At very low concentrations, the water structure is even enhanced by the alcohols, and it persists up to ca. 0.25 mole fraction of alcohol. Enthalpy is lost and entropy is gained when the alcohol is transferred from the external, water-rich and hence structured, solution into the internal solution, where much less structure is permitted by the resin skeleton and functional groups. This transfer is facilitated, the larger the hydrophobic part of the alcohol, and the more tight the water structure is. The data in Table I1 show the coexistence of endothermic and exothermic transfer reactions, the former due to destruction of the enhanced water structure, when the structure promoting agent, the alcohol, leaves the structured external solution, and the latter due to specific interactions with the resin. Such exothermic specific interactions3are also exhibited by the solvents compared in Figure 1. The comparison between the three-carbon solvents 1-propanol, 2-propanol, allyl alcohol, and acetone is highly significant. The presence of the double bond in allyl alcohol causes a strongly enhanced selectivity for the solvent compared to the saturated isostructural 1-propanol (Figure lb). However, this is not due solely to a larger swelling by the organic solvent, but is also due to a much smaller uptake of water. A direct interaction between the double bond and the aromatic .Ir-electronsystems could be responsible for the former trend, and this, apparently, causes a collapse of the free space available in the resin (it acts as a further cross linking), and reduces the ability to take up free water, beyond the solvating requirements of the fixed and counterions. The total uptake of solvent and water for all the four solvents does not fall below 3 mol (equivalent of and this may represent the solvation of the completely ion-paired exchange site. Acetone and 2propanol, which have their polar group in the middle of the molecule and present to the resin skeleton two methyl
862
The Journal of Physical Chemistry, Vol. 82, No. 8, 1978
groups, are intermediate between 1-propanol and allyl alcohol. The possibility of hydrogen bonding of the hydroxyl group with the chloride counterion in 2-propanol, compared with its absence in acetone, may account for the better sorption of the former on the resin. The effect seen in Figure 1 is beyond that ascribable to the lower cross linking of the resin used for 2-propanol (4%), since similar data for 4 and 8% cross-linked resins for 1-propanol1and acetone3 show only a minor effect of the cross linking. Swelling in the Presence of Invasion. The presence in the resin of invading lithium chloride has a general depressant effect on the swelling (Figures 2a, 3a and 4a). The effect is naturally very small at 0.010 m nominal lithium chloride in the external solution, and increases with 0.10 m and even more with 1.0 m LiC1. However, the effect is almost wholly on the sorption of water, that of the solvent being hardly affected. The sorption of water is depressed at low solvent concentrations (x, < 0.3) and enhanced at high solvent concentrations (0.5 < x, < 0.9), the latter effect being most noticeable with 1-propanol. The selective swelling curves f , (x,), Figures 2b, 3b, and 4b, merely reflect these changes in the water sorption. The preferred swelling by the solvent at x, < 0.2 becomes more pronounced, and the flat portion of the curve (with 2, even decreasing somewhat on increasing x,) is lengthened, in the presence of electrolyte. These obsevations can be explained as follows. In the water-rich portion (x, < 0.3) of the diagrams, where the lithium chloride is preponderately dissociated to ions, its ions are preferentially hydrated, so that the water activity is relatively decreased, and water is less able to swell the resin. The decrease in rt, is reflected in increased 2, as already stated. In the solvent-richer portion of the diagram (0.5 < x, < 0.9) the lithium chloride is highly ion paired, and no Donnan potential prevents its massive invasion of the resin, see below. The ion pairs still prefer hydration over solvation by the solvent, as long as sufficient water is present ( x , < 0.9), and carry the water along with them into the resin. Only when water is very scarce does solvation with the alcohol overcome hydration,23and indeed beyond x, = 0.9 water is no longer Preferentially sorbed and it, in the presence of the invading lithium chloride is no longer larger than in its absence. Kim4 has studied the effect of invasion of hydrogen chloride, rather than of lithium chloride, on the selective swelling by water and the alcohols, using nominally 1 m hydrogen chloride in the external solution.24In agreement with our results, he also finds only a slight effect of the invading electrolyte on the sorption of the solvent (a generally depressing effect) and a greater effect on the sorption of water. However, while his selective swelling curves resemble ours in methanol and ethanol, this is not the case for 1-propanol, where he does not find enhanced water uptake in the presence of electrolyte in the solvent-rich region, 0.5 < x, < 0.9, There could very well be a real difference between the two electrolytes, which otherwise behave quite similarly, in that the ion-paired hydrogen chloride can be as effectively solvated by propanol as by water, through hydrogen bonding C3H70(H)-.H+Cl-, which is not available for lithium chloride, where only dipole interaction should be operative or solvation of the anionic end of the ion pair. Invasion from Mixed Solvents. Invasion of lithium chloride into the resin is facilitated by the lowering of the Donnan potential barrier as the fraction of the low-dielectric-constant solvent in the swelling mixtures increases. The molality of electrolyte in the resin increases on increasing x, up to ca. 0.7, but then decreases (Figure 6). No
Y. Marcus and J. Naveh
such decrease was noted by Kim4 in his experiments with hydrogen chloride invasion, but his data extend, actually, only up to x, = 0.7 or 0.8, and for the very alcohol-rich region, the extent of invasion was evaluted by extrapolation, so that a maximum could have been missed. Invasion by lithium chloride is generally lower than by hydrogen chloride, but is impressive enough in that the concentration of the invading electrolyte (at 0.3 < x, < 0.8) in the swollen resin in ethanol is somewhat, and in 1propanol is even more than two-fold higher than its external concentration. This effect is carried to extremes in the case of invasion from aqueous acetone, where the electrolyte is quite effectively removed by the resin from the solution (Figure 5 ) . This removal of electrolytes, and specifically of lithium chloride, has already been noted long ago, in the pioneering work of Katzin and Gebert,25*26 and of Kennedy and Davies.22 No salt absorption was noted by these authors in a completely dry system, which corresponds to our observation of decreasing fiLior r values for very organic solvent-rich mixtures (x, > 0.8). Although with acetone, at the 0.01 m electrolyte level, our data show that whereas x, approached 1.000 as closely as could be ascertained from the known density of pure acetone, the resin contained only 0.20 mol of acetone, but as much as 1.51 mol of water, per equivalent of resin (i.e., 2, = 0.12), and as much as 0.206 mol of lithium chloride (6000 times the external concentration). Thus the high log r noted in our experiments depends on the presence of appreciable amounts of water in the resin, as already The invasion cannot, however, be expressed simply as a dissolution of the electrolyte in the water component of the swelling mixture, as has been ~uggested.~'The ratio of electrolyte to water varies moderately up to x, = 0.6 for methanol and ethanol, or 0.8 for 1-propanol, but increases rapidly above that, so that another explanation is necessary. If Figures 5 and 6 are compared with Figure 7 , where Tiis shown, it is seen that the pronounced maxima in the former have disappeared in the latter, and very shallow minima replace them (probably due to errors in the estimation of yi and are insignificant). That is, the activity coefficient of the electrolyte in the resin approaches asymptotically some low value as x, increases. Thus, the extremum in r, fiLicl,or rtLi depends on a sluggish decrease of y+ for the external solution, as x, increases, whereas very low values of 7, are reached in the resin much faster. This decrease can be wholly ascribed to ion pairing, as pointed out for the homogeneous alcoholic solutions of lithium by othersSmIn view of the hydrocarbon skeleton and the tight space inside the anion exchange resins, the dipoles of the solvent cannot align properly and the dielectric constant is much lowered. Thus ion pairing should occur for much lower x, values than in the outside solution. This is true for both fixed-ion-counterion pairing at the exchange site, and to lithium and chloride pairing in the imbibed solvent. This extensive ion pairing even at low x, can also explain the notable initial decrease in water sorption at low x , (Figures 2-4), but since these two phenomena are interconnected, it is impossible to say which is the cause and which the effect. A final word should be said about the dependence of the invasion of the electrolyte on its concentration. Figure 5 leaves the impression that the effect is largest at the lowest concentration, and decreases as mLiClincreases. This is, however, illusory, since as Figure 7 shows the decrease in y+ is steepest and the ion pairing is the most pronounced at the highest concentration, as expected. However, it is noteworthy that on the whole the invasion of hydrogen
Swelling of Anion Exchangers in Mixed Solvents
chloride or of lithium chloride from their (nearly) anhydrous solutions in methanol or ethanol proceeds similarly to that from purely aqueous solutions. The function log d (log a), where d = fic17+ and a = mHCl(?, Lic1) y+ is shown in Figure 8 for hydrogen chloride in water,29 rnethan01,~Ox , = 0.55 aqueous ethan01,~and ethanoP1 and for lithium chloride in water32and in ethanol.31 The curves are similar, slightly inflected, hence third degree polynomials, but over a limited range of concentrations corresponding to 0.05-1.0 m (displaced along the log a axis because of different y+) simpler dependencies have been found. (First degree log d = 0.23 + 0.62 log a for the x , = 0.55 aqueous ethanolic hydrogen chloride data: second degree for methanolic and ethanolic hydrogen chloride.30) Thus, whether ion paired or dissociated, the electrolyte invading the resin provides an effective activity that varies over a wide range, and this affects the distribution of trace metal complexes in a manner already discussed in other publication^.^^^^ A general overview of the results reported in this paper and in our previous publi~ationsl-~ leads to the following conclusions, concerning the interactions responsible for the swelling of a dry anion exchange resin in mixed solvents. Depending on the conditions (counterion and cross linking of resin, nature and composition of solvent, presence of electrolyte),most, or all, of the following interactions occur. (a) That solvent which causes preferential solvation (usually water) is the one that solvates the ions (counterions and fixed ions). (b) Ions which become solvated change from the ion-paired state in which they occur in the dried resin to the partially or totally dissociated state. ( c ) Between dissociated ions of the same sign, Coulombic repulsion operates, which causes the opening up of the coiled chains of the resin skeleton. (d) Water and the organic solvent enter the spaces created between the chains, and constitute the “free solvent”. (e) Solvents with an organic “tail” which has an affinity to the resin skeleton (including aromatic groups) are sorbed on the skeleton, decreasing the space available for water. (f) Solvents which enhance water structure in dilute solution are pushed into the structureless resin phase. (g) Electrolytes distribute themselves so that they prefer the phase of lower dielectric constant but with polar solvating solvents, so that their enthalpy is lowered and entropy is increased most by ion pairing. References and Notes (1) Y. Marcus and J. Naveh, J . Phys. Chem., 73,591 (1969). (2) Y. Marcus, J. Naveh, and M. Nissim, J. phys. Chem., 73,4415(1969).
The Journal of Physical Chemistty, Vol. 82,No,
8, 1978 863
(3) Y. Marcus and J. Naveh, IsraelJ. Chem., 10,899 (1972). (4) J. I. Kim, J. Inorg. Nucl. Chem., 37, 239 (1975). (5) L. I. Katzin and E. Gebert, J . Am. Chem. SOC.,75,801 (1953);J. L. Ryan, Inorg. Chem., 2,348 (1963). (6) Y. Marcus and A. S. Kertes, “Ion Exchange and Solvent Extraction of Metal Complexes”, Wiley-Interscience, London, 1969,pp 267-274. (7) P. A. Skabichevskii, Zh. Fiz. Khim., 43,2556 (1969). (8) A. M. Shkodin and L. Ya. Shapovaiova, Izv. Vyssh. Uchab. Zaved., 9,563 (1966). (9) S.Minc and J. Jastrzebska, Rocz. Chem., 42,719 (1968). (10) A. M. Shkodin and L. Ya. Shapovalova, Ukrain. Khirn. Zh., 36,449 (1970). (11) S.Minc, J. Jastrzebska, and J. SieminskaJurek, Rocz. Chem., 43, 1093 (1969). (12) A. M. Shkodin and L. Ya. Shapovalova, Izv. Vyssh. Uchab. Zaved., 12, 134 (1969). (13) A. M. Shkodin and L. Ya. Shapovaiova, Ukrain. Khim. Zh., 35,254 (1969). (14) R. Parsons, “Handbook of Electrochemical Data”, Butterworths, London, 1959. (15) I. T. Oiwa, Sci. Rep. Tohoku Univ., Ser. 7 , 41,47 (1957);J . Phys. Chem., 60,756 (1956). (16) H. S. Harned and B. B. Owen, “Physical Chemistry of Electrolyte Solutions”, 3rd ed, Reinhold, New York, N.Y., 1958. (17) K. H. Pool and R. 0. Bates, J . Chem. Thermodyn., 1, 21 (1969). (18) R. N. Roy, W. Vernon, and A. L. M. Bothwell, Electrochim. Acta, 17,5 (1972). (19) R. N. Roy, W. Vernon, J. J. Gibbons, and A. L. M. Bothwell, J. Chem. Thermodyn., 3,883 (1971). (20) In fact, the data Y + ~(x,) , for 1 mHCI presented by Kim in his Table I‘ are linear with x , to better than &1.3%. (21) In acetone at x , > 0.9 the total swelling seems to fall below 3 mol equiv”, but swelling and other sorption reactionszzare very slow, and possibly equilibrium was not reached. (22) J. Kennedy and R. V. Davies, J. Inorg. Nucl. Chem., 12,193 (1959). (23) Y. Marcus, J. Chem. Eng. Data, 20, 141 (1975);J . Phys. Chem., 80,2451 (1978). (24) The actual equilibrium concentrations were given for the case of aqueous ethanol (Table I of ref 4),varying from 1.02to 1.31 mas x , varied from 0 to 1. (25) L. I. Katzin and E. Gebert, J. Am. Chem. SOC., 75,801 (1953). (26) Y. Marcus in “Ion Exchange and Solvent Extraction”, J. A. Marinsky and Y. Marcus, Ed., Vol. 4,Marcel Dekker, New York, N.Y., 1973, Chapter 1, pp 9-45. (27) C. W. Davies and B. D. R. Owen, J . Chem. SOC., 1676 (1956);R. G. Greene and J. S. Fritz, US. Atomic Energy Commission Report no. IS-1153 (1965). (28) A. M. Shkodin, L. P. Sadovnichaya, and V. A. Podolyanko, Nektrokhmh, 4,718 (1968);Ukr. Khim. Zh., 36,449 (1970);Y. Marcus, N. Ben-Zwi, and I. Shiloh, J . Solutlon Chem., 5, 87 (1976). (29) K. A. Kraus and G. E. Moore, J. Am. Chem. SOC.,75,1457 (1953). (30) Y. Marcus and E. Eyal, J. Inorg. Nucl. Chem., 34,1667 (1972). Only an indirect value of log B (log a ) is given, based on the distribution of tracer perrhenate anions between the resin and the methanolic hydrogen chloride solutions. An arbitrary normalization term log B (log a = 0.00) = 0.30is here employed, in order to fit the methanol data into the family of curves. (31)J. Penciner, I. Eliezer, and Y. Marcus, J. Phys. Chem., 69,2955 (1965). (32) F. Nelson and K. A. Kraus, J. Am. Chem. SOC.,80,4154 (1958); D. Maydan, Ph.D. Thesis, Hebrew University, Jerusalem, 1962;Y. Marcus and D. Maydan, J . Phys. Chem., 67,979 (1963).
