E. M. ARNETT,H. C. KO, AND R. J. MINASZ
2474 the above ratio then becomes 1.04, a value which is not significantly smaller than the average value for such solvents as n-CBH14, CCl,, and C~HB.Therefore, we suspect that the Dllovalue reported by Stokes, et al.,17 may be slightly smaller for some reason. It seems that this kind of comparison may not be too sensitive to detect possible solute-solvent interactions. I n P-11, we toluene solutions as a mixed solhave used ethanol vent. Since Dl10 for toluene is slightly larger than we assumed, we have added a new datum in this solvent system. This datum, together with new Dl?value for pure toluene, has established more firmly the Dll0rza/T vs. x2 relation, which is now very similar t o that for ethanol benzene solution. However, the general conclusion given in P-I1 needs no alteration. There exist
+
+
Solvent Effects in Organic Chemistry.
Salts in Glycerol Acetate Binaries by
still larger discrepancies between our values and the available data (e.g., the Dllo values in n-C6H14 and mxylene). The validity of our data will be checked further in our future studies on diffusion both in pure and mixed solvents. I n conclusion, the present study has convinced us to conclude that a relative decrease in D1l0V23/T values will be observed whenever the interactions of dilute solute with one solvent differ nonnegligibly from those with another solvent. Acknowledgments. The authors are grateful to Mrs. T. Ozasa-Nakajima for her cooperation in some of the experiments (CC1, C6Ha solution) and t o Professor N. Watanabe and Mr. H. Touhara for their interest and discussion.
+
XIV.
23Na
Solvation of Sodium
Nuclear Magnetic
Resonance Spectroscopy and Thermodynamics1 by Edward M. Arnett,* H. C. KO,and R. J. Minasz Department of Chemistry, University of Pittsburgh, and Mellon Institute, Pittsburgh, Pennsylvania 16818 (Received February 16, 1978) Publication costs assisted by the Ofice of Saline Water, Department of the Interior
Heats of solution are reported for sodium chloride, sodium tetraphenylboron, and tetraphenylarsonium chloride in a series of water-diacetin (glycerol diacetate) mixtures. The results for NaCl are combined with literature values of free energy of transfer to give entropies of transfer for this salt from water to each mixture. Kmr studies of 23Na+ion from sodium tetraphenylboron are also reported in water-diacetin binaries, and the results are contrasted with those estimated for the single ion behavior of Naf using the (invalid) assumption that (C6H5)4Asf is equivalent t o B(C6H&-. The enthalpy results for the water-diacetin system are compared to those for heats of transfer of NaB (CeH6)4in acetone-diacetin, ethyl acetate-diacetin, and acetone-triacetin (glycerol triacetate) binaries. Exothermic interactions are found for NaB(CaH& and Mg(ClO& with cellulose acetate and glucose pentaacetate in acetone solution.
Introduction Cellulose acetate membranes are unusually effective for desalinating sea water by reverse osmosis.2va Kraus and his colleagues attempted to elucidate the reasons why cellulose acetate should reject salts under conditions of high water flux by studying the solubility of a number of common electrolytes in binary solvent mixtures composed of water and the mono-, di-, and triacetates of glycerol.* In the interest of gaining a further insight into this system we have measured heats of solution for sodium The Journal of Physical Chemistry, Vol. 76, N o . 17, 1978
chloride (the principal salt in sea water) in a number of water-diacetin mixtures. When our measurements (!) Supported by grants from the Office of Saline Water and National Science Foundation (GP-6550-X) . The 23Na work described here was made possible through NIH Grant RR-00292 supporting the Mellon Institute nmr laboratory. (2) U. Merten, Ed., “Desalination by Reverse Osmosis,” The MIT Press, Cambridge, Mass., 1966. (3) S.Sourirajan, “Reverse Osmosis,” Academic Press, New York, N . Y . , 1970. (4) K . A . Kraus, R . J. Raridon, and W. H. Baldwin, J . Amer. Chem. SOC.,86, 2571 (1964).
SOLVATION OF SODIUM SALTSIN GLYCEROL ACETATEBINARIES are combined with those of Kraus, a complete thermodynamic analysis for the transfer of this salt from pure water to 50 mol diacetin is achieved. T o compare the gross thermodynamic changes with the varying solvation of the sodium ion itself, we have also obtained 23Namagnetic resonance data over the same range and have attempted to estimate single ion heats of transfer for sodium ion for comparison, We hope that by the combined use of thermodynamic and spectral techniques it will be possible to broaden our understanding of the interaction of alkali and alkaline earth metals and various types of membrane materials. Such knowledge might serve as a predictive basis for the development of new membranes with considerably enhanced effectiveness.
Experimental Section Apparatus. The calorimetric equipment and procedures used for this study were essentially the same as described previously5 except that two 250-ml calorimeters were used, and they were immersed in a water bath controlled at 25 f 0 . 2 O . 6 The two thermistors, one in each calorimeter, were connected to the Wheatstone bridge, and the difference in temperatures between the calorimeters was registered on the recorder. Thus, runs could be made alternately on both calorimeters. All experiments were made a t 25 f 0.5". Accuracy of the calorimeters was checked against accepted values for the heat of solution of potassium chloride' or ethanol* in water once every 2 months. Materials. Acetone was of reagent grade, stored over molecular sieve Type 4A and used without further purification. Ethyl acetate, diacetin, and triacetin (Fisher Scientific Co.) were distilled at appropriate pressures and stored over molecular sieves. Refractive indices agreed well with literature values. Diacetin was apparently a mixture of 1,2- and 1,3-glycerol diacetates. Different batches of distilled diacetin all showed boiling points (132-133" (6 mm)) and refractive indices which were not only consistent with each other, but also with those reported by Kraus, et aL4 Calorimetric measurements using different batches of diacetin gave consistent results. Sodium tetraphenylboron (Fisher Scientific Co. reagent grade, 99.5%) was vacuum dried at 80" for 24 hr. Tetraphenylarsonium chloride was obtained by heating the dihydrate (Aldrich Chemical Co.) to consistent weight a t 150" under vacuum and then analyzed for chloride content. Both gravimetric and volumetric methods agreed well (8.41 and 8.48 wt % C1, respectively) within experimental error of the theoretical value, 8.47%. Sodium chloride was dried at 150". Glucose pentaacetate (Eastman Organic Chemicals) and cellulose acetate (Eastman Organic Chemicals) were used without further purification after drying in a vacuum oven. Reagent grade magnesium perchlo-
2475
rate (Fisher Scientific Co.) was analyzed for Mg content by the EDTA titration method and found to be 99.25% pure. 23NaMagnetic Resonance Measurements. The 23Na nmr spectra were obtained with a modified Varian DP-60 spectrometer operating at a frequency of 15.871 MHz. The magnetic field was held constant by an external lock on the proton resonance of water at 60.000000 MHz. The spectra were taken in the absorption mode under frequency sweep conditions. The radiofrequency was generated by a General Radio Co. frequency synthesizer Model GR-1164A. The signal was phase-detected using a Princeton Applied Research lock-in amplifier with a l-kHz modulation for base line stabilization. The locking and observing frequencies were phase-locked to prevent any drift between them. The frequency of the signal was read directly from an Anadex Model CF-2OOR electronic counter. The sample tube had an outside diameter of 15 mm. Corrections due to differences in bulk susceptibilities between samples were expected to be small and thus not applied to the chemical shifts. A sweep rate of about 1 Hz/sec was used, and each sample was swept at least twice in both increasing and decreasing field directions. A shift to lower frequency corresponds to an upfield chemical shift and represents increased screening of the 23Na nucleus. Spinning the sample tube caused extreme sharpening of the 23Naresonance signals.
Results Heats of solution of NaC1 in the HpO-diacetin binary mixtures are listed in Table I together with the derived thermodynamic values of transfer from H 2 0 t o each HzO-diacetin mixture. Tables I1 and I11 contain the heats of solution of NaB(CsH& and (C6H5)4AsCl in the HzO-diacetin binary mixtures and the derived heats of transfer of each salt from HzO to each binary. Owing to low solubilities of NaCI, NaB(C&)4, and (CeH5)4AsCl in pure diacetin, the values for the heats of solution of each of these salts in that solvent were extrapolated. Assuming that B(CeH&- and (CSH5)4As+ have equal heats of transfer (see Discussion), it is possible to calculate single ion heats of transfer for Na+, C1-, B(CeH5)4-, and (CeH6)4As+ from experimentally determined (heats of solution) values. The values of single ion heats of transfer for these ions and also heats of transfer for these three salts are plotted against mole fraction of diacetin and are presented in Figure 1. (5) E.M.Arnett, W. G. Bentrude, J. J. Burke, and P. McC. Duggleby, J. Amer. Chem. SOC.,87, 1541 (1965). (6) E.M.Arnett and J. J. Campion, ibid., 92, 7097 (1970). (7) (a) G. Somson, J. Coops, and M. W. Tolk, Recl. Trav. Chim. Pays-Bas, 82,231 (1963); (b) R.J. Irving and I. Wadso, Acta Chem. Scand., 18, 195 (1964). (8) F. Franks and B. Watson, J. Sei. Instrum., 1, 940 (1968). The Journal of Physical Chemistry, Vol. 76, No. 17, 1072
E. M. ARNETT,H. C. KO, AND R. J. MINASZ
247 6 +4,
I
I
I
I
I
I
I
I
1
I
Table I11 : Heats of Solution of (CeHs)4AsC1 in HzO-Diacetin Mixtures Wt % diacetin
diacetin
AHsoln
0.0 9.9 20.6 30.4 40.0 50.2 60.1 70.2 80.3 89.2 100.0
0 * 000 0.011 0.026 0.043 0.064 0.093 0.133 0.194 0.294 0.458 1.000
-1.53 i 0.12 1.36 i:0.13 1.83 f 0.41 1.81 4 0.47 1.40 f 0.06 1.00 f 0.03 0.57 4 0.04 0.04 4 0.41 - 0 . 5 7 i 0.13 -1.1340.10 -1.67a
-4
-5
I
0.1
0.2
0.3 0.4 0.5 0.6 0.7 Mole Fraction, Diocetin
0.9
0.8
1.0
Figure 1. Observed and estimated heats of solution in aqueous solutions of diacetin.
Table I : Heats of Solution and Thermodynamics of Transfer of NaCl in HzO-Diacetin Mixtures Mole Wt % fraction, diacetin X , diacetin
0.0 9.9 20.2 29.6 44.6 56.6 79.9 85.6 91.0 100.0
0.000 0.011 0.025 0.041 0.076 0.117 0.290 0.377 0.508 1.000
AHeolno
6AGoa'b
8AHoa
S(TA8)""
0.99 f 0.03 0.91 f 0.04 0.81 i:0.02 0.72 f 0.03 0.49 f 0.01 0.34 f 0.02 - 0 . 9 O f 0.09 -1.O5i:OO.O5 -1.24iO.10 - 1.40"
0 0.02 0.05 0.07 0.12 0.19 0.49 0.67 1.01
0 -0.08 -0.18 -0.27 -0.50 -0.65 -1.89 -2.04 -2.23 -2.39
0 -0.10 -0.23 -0.34 -0.62 -0.84 -2.38 -2.71 -3.24
a
5
wt % diacetin
X, diaoetin
AHaoln
0.0 9.9 20.1 29.9 40.0 46.2 55.2 60.1 70.2 79.8 89.2 100.0
0.000 0.011 0.025 0.042 0.064 0.080 0.112 0.133 0.194 0.288 0.458 1.000
-4.71f0.14 -2.50f0.15 -2.02 f 0.07 -2.46f0.06 -4.28i: 0.10 -5.62i0.21 -6.62f0.34 -7.54 f 0.26 -8.24f0.26 -8.83 i 0 . 2 1 -9.19f0.16 -9.50"
a
SAH
0 2.21 2.69 2.25 0.43 -0.91 -1.91 -2.83 -3.53 -4.12 -4.48 -4.79
The Journal of Physical Chemistry, Vol. 78, No. 17, 197.9
0 2.89 3.36 3.34 2.93 2.53 2.10 1.57 0.96 0.40 -0.14
Extrapolated.
Wt % diacetin
diaoetin
0 12.9 25.3 35.0 45.1 54.5 65.1 74.8 83.7 87.6 92.5 100.0
0 0.046 0.101 0.151 0.213 0.283 0.380 0.495 0.629 0.699 0.791 1.000
X,
AHBOIL,
-13.9140.44 -15.12i0.19 -14.84f0.02 -16.17 4 0.69 -15.44f0.12 -14.65&0.24 -14.50i0.03 -14.10i:0.12 -13.9740.25 -13.72f0.16 -13.19f0.36 - 12.85"
Extrapolated.
Table V: Heats of Solution of NaB(C6H6)4in Ethyl Acetate-Triacetin Mixtures
Extrapolated. Probably in error, see text.
For nonaqueous binaries we have studied heats of solution of NaB(CeH5)4 in acetone-diacetin, ethyl acetate-diacetin, and acetone-triacetin mixtures. Results are presented in Tables IV, v, and VI. Values for AHsolnof the salt in pure diacetin and triacetin, respectively, were extrapolated from these plots.
8AH
Table IV: Heats of Solution of NaB(C&s)4 in Acetone-Diacetin Mixtures
All thermodynamic values in this and subsequent tables are in kcal/mol. b Calculated from activity coefficient data of Kraus, et uL4 Extrapolated.
Table 11: Heats of Solution of NaB(CeHs)c in H20-Diacetin Mixtures
X,
a
wt % dia0 etin
diacetin
0 10.3 20.0 34.9 46.2 54.8 64.9 75.0 86.1 91.0 100.0
0 0.054 0.111 0.212 0.300 0 378 0.481 0.600 0.756 0.835 1.000
X,
I
AHaoln
-14.57 f 0.17 -15.43 i 0.47 - 16.30 f 0.15 -17.39i0.10 -15.22f0.19 -14.7040.14 - 1 4 . 2 0 i 0.08 -13.46dz0.37 -12.81f0.28 -12.654~0~32 -12.25a
Extrapolated.
Heats of solution of NaB(CeH5)4 and Mg(C10& in acetone-glucose pentaacetate and acetone-cellulose acetate mixtures were also measured. Results are
SOLVATION OF SODIUM SALTSIN GLYCEROL ACETATEBINARIES Table IX : 23Na Resonance of 0.05 iM NaB(CgH5)4 in H20-Diacetin
Table VI: Heats of Solution of NaB(CgH6)a in Acetone-Triacetin Mixtures
wt
%
triacetin
X, triacetin
AHsolo
0 12.1 23.9 33.0 44.9 55.2 64.8 75.0 84.8 91.2 100.0
0 0.035 0.077 0.116 0.178 0.247 0.329 0.444 0.597 0.734 1.000
-13.91f0.44 -14.38f0.04 -14.29f0.11 -14.61 f 0.26 -14.19f0.59 -14.35f0.50 -14.85 f 0.67 -15.23f0.23 - 1 5 . 0 2 f 0.31 -15.80f 0.26 -16.18“
wt
%
diacetin
a
Extrapolated.
listed in Tables VI1 and VIII. Table IX gives chemical shifts for 2sNa+in the H20-diacetin system.
Table VII: Heats of Solution of NaB(CgH5)d and Mg(C104)~ in Acetone-Glucose Pentaacetate (GPA) Mixtures
0 5.3 10.0
16.9 22.9 28.2 33.2 33.8
0 0.008 0.016 0.029 0.042 0.055 0.069 0.070
-13.91=tOO.44 -14.24f0.82 -14.33 f 0.06 -14.63rt 0.26 -14.58 f 0.35 -14.36 f 0.56
2477
-39.36f0.50 -39.08zt0.70 -40.74 f 0.72 -42.48f0.99 -40.64 f 1.01 -42.50 i: 0.76 - 4 1 . 8 6 f 1.01
-14.35 f 0.30
Table VIII: Heats of Solution of NaB(CeH& and Mg(ClO& in Acetone-Cellulose Acetate (CA) Mixtures Wt % CA
AHsoln,
AHsoln,
NaB(CsHs)r
MdClO4)e
0 3.1 6.0
-13.91f0.44 -14.9Ort0.53 -15.21f0.03
-39.36rt0.50 -45.79 f 1.25 -45.82f0.60
Discussion The Water-Diacetin System. Presented in Table I are the three thermodynamic properties of transfer, 6AGo, 6 A H o , and 8(TASo) for sodium chloride from pure water to the aqueous diacetin mixtures shown. Gibbs free energies are derived directly from the ac, ~ the exprestivity coefficients of Kraus, et ~ l . through sion AGO = -2.303RT log y, and entropy data are obtained by the combination of A G O and our enthalpy values. Kraus, et al., included in their convention for the activity coefficient of the solute the fraction of water in the mixture. This would have the effect of
0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 50.0 60.0 70.0
X, diacetin
0 0.005 0.011
0.018 0.025 0.033 0.042 0.052 0.064 0.093 0.133 0.193
Frequency of WJa’, 8AH(Na+)Q
HAC
0 -0.10 -0.20 -0.36 -0.54 -0.72 -0.94 -1.18 -1.44 -2.05 -2.67 -3.23
869.8 & 0 . 2 869.7 3I 0 . 2 868.8 I 0 . 2 867.9 i0 . 2 866.7 & 0.3 866.1 & 0.3 865.1 I 0 . 3 864.2 i 0 . 4 863.6 I 0.4 861.2 I 0.5 856.4 & 0 . 6 851.6% 1.0
Frequency a Assuming that ~ A H ( ( c ~ ~ = ~ 6AH~B(c,~,,4-). )~A~+) synthesizer set a t 15.871 x 106 Hz, c Chemical shift to lower frequency represents increased shielding of the Z3Na nucleus (see text).
making A G O indeterminant at zero water content but will have relatively little effect on 8AG” or 8(TASo) across the range which we have reported. It is clear from the data in Table I that the gradually decreasing solubility of sodium chloride as the water content is decreased is due to an entropy effect rather than to an unfavorable change in enthalpy. Indeed sodium chloride is solvated increasingly exothermally as the diacetin concentration is increased. We attribute this primarily to the endothermic effect of structure breaking by this salt in pure water which is also manifested by the greater gain in entropy due to structure breaking in pure water than in the less aqueous mixtures. Such effects are found in many other aqueous binary systems .9-l1 I n the hope of analyzing the enthalpy changes further, we have measured the heats of transfer for sodium tetraphenylboron and tetraphenylarsonium chloride in these same mixtures (Tables I1 and 111). The peculiar extremum behavior seen for sodium tetraphenylboron and tetraphenylarsonium chloride in Figure 1 is typical of nonelectrolytes and hydrophobic ions of large diameter in highly aqueous organic binaries.6t1l It is related to an enhanced degree of structure in the medium a t the composition where the inflection point occurs.11-18 However, the detailed interpretation of (9) H . S. Frank and M. W. Evans, J. Chem. Phys., 13, 507 (1945). (10) E. M. Arnett and D. R. McKelvey in “Solute-Solvent Interactions,” J. F. Coetaee and C. D. Ritchie, Ed., Marcel Dekker, New York, N. Y., 1969. (11) E. M. Arnett in “Physico-Chemical Processes in Mixed Aqueous Solvents,” F. Franks, Ed., Heinemann Educational Books, Ltd., London, 1967. (12) F. Franks and D. J. G . Ives, Quart. Rev., Chem. SOC.,20, 1 (1966). (13) E. M. Arnett and D. R. McKelvey, J . Amer. Chem. Soc., 8 8 , 2598 (1966).
The Journal of Physical Chemistry, Vol. 76, No. 17, I972
2478 such curves is beyond the scope of present solution theory. If the assumption is made that the heats of transfer for the tetraphenylboride and tetraphenylarsonium ions are exactly equal from pure water to any of the mixtures we have studied, one may calculate “single ion enthalpies of transfer.”l3 These are also presented in Figure 1. The strength of this assumption has been examined by several workers, 14,16 and it is unreliable in low dielectric solvents. Recent studies (which were unpublished when the present experiments were performed) also render it dubious for ~ a t e r . ‘ ~ J ’Unfortunately, the tetraalkylboride ions which are recommended14’17as superior for such a comparison are hydrolyzed in aqueous media. Comparison of Nmr Data with Thermodynamics. Unlike the chemical shift of the proton, that of the 23Na nucleus is dominated by a paramagnetic term.’8 This usually leads to an increasingly large downfield shift in response to greater interaction with an electron donor in contrast to the corresponding upfieId shift for greater screening of the proton. The 23Xanucleus also has a quadrupole moment. If the nucleus is in an asymmetric environment, this may produce considerable line broadening which is not subject to ready-quantitative interpretation. Bloor and Kiddl8and P O ~ O V have found direct linear correlations between the basicities of various solvents (against aqueous acid or antimony pentachloride) and the W a chemical shift produced by those solvents. We have also observed such correlations in nonaqueous media.21 In general the strongest bases produce the largest downfield shifts. However, some polydentate ligands cause upfield shifts presumably because of an increased diamagnetic contribution. It is, therefore, interesting to see that although the addition of diacetin to water produces an exothermic trend in the heat of transfer, which seems to indicate increasing solvation of the sodium ion, this is accompanied by an upfield shift which is cleanly proportional to the mole fraction of diacetin (Table IX). We have clear independent evidence from other measurements21 that water, when present at high dilution in an inert solvent, is strongly coordinated t o SOdium ion and interacts much more exothermally with it than does diacetin. Thus, in terms of what the sodium ion “sees” in its immediate solvation layer the large downfield W a shift in water is consistent with all available data in suggesting that water should be a stronger complexing agent than any of the water-diacetin mixtures. Why then does the curve for the single ion enthalpy of Na+ (Figure 1) show an exothermic trend for the solvation of sodium ion as the water content is decreased? The simplest of many possible explanations for this apparent discrepancy is that it is a manifestation of the failure of the single-ion assumption (B(CSH&-) = (As(CsH&+) through which the AHtransvalues for Na+ were derived. The Journal of Physical Chemistry, Vol. 76, No. 17,1972
E. M. ARNETT,H. C. KO, AND R. J. MINASZ An excellent linear correlation was obtained between the free energy of transfer for sodium chloride and the corresponding W a shift of sodium tetraphenylboron. Owing to the complicated contributions of entropy and enthalpy to the free energy change, this correlation may be fortuitous. The ability to explore these correlations further is limited by the poor solubility of the sodium salts in diacetin-rich solutions and by the very broad peaks obtained by 23Naresonance in such media. Nonaqueous Binaries. The results presented in Tables IV through VI11 for heats of solution of NaB(C6H6)4 in various nonaqueous binaries were developed to explore a number of points. Sodium tetraphenylboron is not soluble enough in diacetin to allow a direct heat of solution measurement with our apparatus. I n view of the special solvating ability which small amounts of water might exert on sodium ions in diacetin, we wished to have independent estimates for the extrapolated value (-9.50 kcal/mol) for NaB(CsH6)d in pure diacetin presented in Table I1 for the water-diacetin system. Accordingly the acetone-diacetin and ethyl acetate-diacetin systems were explored, and extrapolations to -12.85 and -12.28 kcal/mol were obtained, respectively. These are in ’fairly ~ ~ good ~ ~ agreement with each other but are grossly at odds with the much more endothermic value obtained from the aqueous binary system. We believe that the true value lies between -12 and --13 kcal/ mol and that the reason for the Iarge error in an endothermic direction for the value extrapolated from the aqueous binary lies in the fact that 0.458 mole fraction diacetin (Table 11) is still a highly structured solvent compared to organic binaries. The endothermic error is the result of the large structure-breaking energy required for dissolving salts in highly aqueous binaries. This has its peak at about 20 wt % (0.025 mole fraction) diacetin but is still effective in the 0.458 mole fraction solution. Tables IV and V show behavior similar to that which we have reported before for the solution of XaB(CeH5)r in binaries of dimethyl sulfoxide with acetone and with dioxane.ll I n all four cases this salt interacts more exothermally with a mixture of intermediate composition than it does with either pure component. This contrasts sharply with the endothermic maxima which (14) R . Fuchs, J. L. Bear, and R . F. Rodewald, J . Amer. Chem. Soc., 91, 5797 (1969).
(15) R.Alexander and A. J. Parker, ibid., 89, 5549 (1967). (16) C. Jolicoeur and H. L. Friedman, J . Phys. Chem., 75, 165 (1971). (17) J. F.Coetaee and W. R . Sharpe, ibid., 75, 3141 (1971). (18) E. G.Bloor and R . G. Kidd, Can. J . Chem., 46, 3425 (1968). (19) R . H.Erlich, E. Roach, and A. I. Popov, J . Amer. Chem. SOC., 92,4989 (1970). (20) R . H.Erlich and A. I. Popov, ibid., 93, 5620 (1971). (21) E . M.Arnett, H. C. KO,and R . J. Minasz, unpublished results.
ESRSPECTRA OF ZWITTERIONRADICALS
2479
are typical of aqueous binaries.6 It also differs from the acetone-triacetin binary (Table VI) in which a steady exothermic trend is found towards triacetin, the superior solvating medium. This superiority of 2.27 kcal/mol (16.18 - 13.91) is worth comparing with that of ethyl acetate (14.57 - 13.91 = 0.66 kcal/mol) over acetone. If triacetin were behaving as three ethyl acetate residues bound together, we might expect it to be superior to acetone by about 2.0 kcal/mol. The fact that triacetin exceeds this by about 0.3 kcal/mol might be due to the fact that the acetate residues are vicinal to each other on the three-carbon backbone of the molecule, thus permitting a chelation effect such as is found in the binding of alkali metal ions by other polybasic ligands.22 The results in Table VI1 show that on a mole fraction basis glucose pentaacetate (a possible model for the fundamental molecular unit of completely acetylated cellulose) is comparable to triacetin in its ability to solvate sodium ions. Table VI11 shows that in dilute acetone solution cellulose acetate is superior to diacetin, triacetin, and glucose pentaacetate in its interaction with sodium tetraphenylboron. Since cellulose acetate
is partially hydroxylated and also carries ether linkages between its glucose residues, these groups may be contributing exothermally to the solvation of sodium ions. Magnesium perchlorate is used as an additive in casting cellulose acetate reverse osmosis membranes. K e ~ t i n has g ~ ~proposed that cations of high charge such as Mg2+form complexes with the hydroxyl and acetate groups of cellulose acetate and contribute to the swelling of the membrane by subsequent hydration after immersion in water. The results in Tables VI1 and VI11 show clearly that Mg(C10& interacts strongly with cellulose acetate in acetone and that cellulose acetate is far superior to glucose pentaacetate on this score. It remains to be seen how much this difference is due to conformational factors and how much is due to the specific superiority of hydroxyl and ether functions in solvating cations. In any event, Kesting's postulation of complexing between this salt and cellulose acetate is clearly supported. (22) J. J. Christensen, J. 0. Hill, and R.M. Izatt, Science, 174, 459 (1971). (23) R.E. Kesting, J . A p p l . Polym. Sci., 9, 663 (1965).
Electron Spin Resonance Spectra of Zwitterion Radicals and Isoelectronic Anion Radicals1 by Ronald P. Mason" Department of Chemistry, Cornell University, Ithaca, Neur York 14860
and John E. Harriman Department of Chemistry, University of Wisconsin, Madison, Wisconsin 65706 (Received January 12, 1972) Publication costs borne completely b y The Journal of Physical Chemistry
Electron spin resonance spectra have been obtained for the p-nitrobenzyltrimethylammonium zwitterion radical and the 2,2-dimethyl-l-p-nitrophenylpropaneanion radical in acetonitrile at -40" and for the p nitrophenyltrimethylammonium switterion radical and the p-nitro-tert-butylbenzene anion radical in dimethyl sulfoxide at room .temperature. All radicals were generated electrolytically. Hyperfine coupling constants are reported and discussed in terms of probable radical structures. I n a recent note, Knolle and Bolton2 have discussed differences in the fluorine hyperfine splitting in two isoelectronic and essentially isostructural radicals, one a neutral nitroxide and the other a ketyl. We report here an investigation of the Droton hvDerfine CouDIing constants of two pairs of isoelectronic para-substituted nitrobenzene radicals. One member of each pair is an
-
" I
I
V
anion radical with an alkyl substituent while the other is a zwitterion radical containing a trimethylammonium substituent. The radical pairs are -0&CsH&H&(1) This research was supported by the National Science Foundation under Grants No. GP-7512 and GP-9539. (2) W. R. Knolle and J. R. Bolton, J. Amer. Chem. Soc., 91, 5411 (1969).
The Journal of Physical Chemistry, Vol. 76, No. 17, 1972