J. CHEM. SOC. FARADAY TRANS., 1994, 90(6), 849-853 Study of the Structure-breaking Effect in Aqueous CsCl Solutions Based on H20/D,0 Isotope Effects on Transport Coefficients and Microdynamical Properties Antonio Sacco* and Hermann Weingartnert Dipartimento di Chimica, Universita degli Studi di Bari,4, Trav.200 Re David, 1-70126 Bari, Italy Bernd M. Braun and Manfred Holz lnstitut fur Physikalische Chemie und Elektrochemie der Universitat Karlsruhe, Kaiserstr. 12, 0-76128 Karisruhe, Germany We have studied the effect of H,O/D,O isotopic substitution upon various transport and microdynamical proper- ties in aqueous solutions of structure-breaking salts, using aqueous CsCl in the concentration range up to 6 mol kg-' as a representative example. We report on isotope effects upon the self-diffusion coefficients of water and of Cs+ and CI- ions, upon the reorientational correlation times of water deduced from 2H magnetic relaxation rates, and upon the magnetic relaxation rates of the quadrupolar relaxing ionic nuclei 133Cs+ and 35CI- in these solutions. Comparison is made with the isotope effectupon the viscosity, and some subtle differences are outlined.The most substantial one is related to the isotope effects upon 13'Cs+ and 35CI- relaxation which selectively probe motions in the first cosphere of the ions. These are substantially lower than those observed with the other dynamical quantities, showing that dynamical isotope effects in the first cosphere may differ substantially from those in the bulk. It is concluded that the structure-breaking effect extends to the first co-sphere of the ions.This paper is intended to contribute to our understanding of the 'structure-breaking ' effect' or 'negative hydration'2 of salts like KCl or CsCl in aqueous solutions. These terms are now widely used for describing the effect of large ions on the water structure, although we have little knowledge on the actual structural changes. In particular, there is a profound influence of structure-breaking ions on transport coefficients, evidenced by a decrease in viscosity, and described by nega- tive B-coefficients in the Jones-Dole expansion3 (tf/tfo) -1 = AC''2 + BC + (1) where tfo is the viscosity of pure water and C the salt concen- tration.By some convention on the splitting of B, the same notion is used for single-ion properties. Although there are cases with B < 0 in non-aqueous solvents: negative B-coeficients are above all characteristic for water. Transport theories which treat the solvent as a structure- less medium characterized by its macroscopic viscosity and relative permittivity' are presently incapable of predicting negative B-coefficients.6 Hence, any explanation has to account for the molecular nature of the solvent. In fact, it is possible to examine the molecular basis of the structure- breaking effect by studying quantities which specifically monitor translational or rotational processes in solution on the molecular level, or probe the local dynamics of solvent molecules near ions: 'H and 'H magnetic relaxation probe reorientational motions of water spin-echo or radiotracer self-diffusion experiments monitor the trans-lational motions of water molecules7 and of ions,* magnetic relaxation of quadrupolar ionic nuclei probes fluctuating electric field gradients at these nuclei, thereby providing information on the local dynamics of solvent molecules near ion^.^*'^ Here, we adopt a new approach for investigating the structure-breaking phenomenon by monitoring the effect of H20/D,0 substitution upon the above-mentioned dynami- cal quantities.Again, continuum models are doomed to fail On leave from the Institut fur Physikalische Chemie und Elektro- chemie der Universitat Karlsruhe, Germany.in the explanation of such isotope effects. One reason is that the relative permittivities of H20 and D20 are almost the same, so that in continuum theories both solvents give rise to the same long-range electrostatic effects on dynamcal proper- ties like the B-coefficient.6 Differences are then inadvertently related to the molecular nature of the solvent. To address the problem in more detail, we note that for pure water at 298 K the isotope effect upon the viscosity defined by Vr = tf(D,O)/tf(H20) (24 is q, = 1.228.'' Defining analogous quantities for the self- diffusion coefficient and reorientational correlation time of water by Dr, w = Dw(D2O)/Dw(H20) (2b) and 7Zr = ~2(D20)/72(H20) (24 we have DrIk m 72r E v,,'~-'~i.e. the factor 1.228 is the same for translational and rotational motions.Moreover, it exceeds by far what is expected from the so-called 'square- root-of-mass law','' which predicts Dr;t and tf, to be equal to [m(D,O)/m(H,O)] 'I2 = 1.05, where rn is the molecular mass. It is intriguing to explain this observation by a strong coupling of translational and rotational motions, which implies that D,,is closer to the ratio of the square roots of the moments of inertia, i.e. Dr;k x 1.38, rather than to m1/2.15*16We have recently confirmed this interpretation on a broader basis by considering transport processes in meth- anol,' 7*1* where an unambiguous correlation between the self-diffusion coefficients and moments of inertia of eight iso- topic species was found.The question is then how structure-breaking ions affect this coupling. Interestingly, for structure-breaking salts viscosity B-coefficients are more negative in D20 than in H,O, while for structure-formers they are almost In other words, the H20/D20 isotope effect decreases upon addition of structure-breakers, while for structure-formers it remains almost constant. However, in general, one could expect 850 dynamical isotope effects to differ both locally and with respect to the different modes of motion. This is the hypothe- sis to be tested in the present work. Some evidence for such a behaviour comes from the limiting ionic mobilities, which for structure-breakers exhibit a smaller isotope effect than pre- dicted from the bulk viscosity." Moreover, one of the present authorsi3 has reported magnetic relaxation rates of ionic nuclei, specifically of 87Rb+, 23Na+ and "Br- in water, finding a substantial decrease of isotope effects with increas- ing salt concentration.To investigate these questions in more detail, we have per- formed a systematic study of isotope effects upon trans-lational and rotational motions in aqueous CsCl solutions at salt concentrations from 0.1 to 6 mol kg-'. The properties of the '33Cs+ and 35Cl- nuclei make this salt particularly suited to magnetic resonance studies. We report in detail on the influence of CsCl on H20/D20 isotope effects upon the self-diffusion coefficient and reorientational correlation time of water.Moreover, we have monitored isotope effects on the self-diffusion of the Cs' and C1- ions. We define these isotope effects in analogy to eqn. (2a)-(2c) by Dr, cs = DCs(D2O)/Dcs(H20) (24 and D,c1 = DCl(D2O)/DC,(H2O) (24 where Dc, and Dc, denote the self-diffusion coefficients of the Cs' and C1- ions, respectively. Finally, we will report on H20/D20 isotope effects upon the magnetic relaxation rates l/Tl of the quadrupolar nuclei 133Cs+ and 35Cl-. These are defined by (1/T1)r, cs = (I/7'1)cs(D20)/( 1/T1 )Cs(H20) (2f1 and As will be seen below, the latter quantities reflect the local dynamics of solvent molecules near ions. Although intrinsic dificulties connected with some of these experiments do not allow measurements of isotope effects with the accuracy typical for some other quantities, say elec- trical conductances,' ' these experiments will provide impor- tant new information on the molecular basis of the structure-breaking phenomenon. Experimenta1 CsCl of 'suprapur' quality (Merck, Darmstadt) was dried at 200°C.Solutions in H20 (deionized and distilled) and D20 (Aldrich, D-content >99.8 at.%) were made up by weight. Samples used in relaxation time measurements were degassed by several freeze-pumpthaw cycles. The equipment and experimental procedures for obtaining highly accurate data are described in detail elsewhere : self-diffusion measurements by 'H,l7 and 2H and 133Cs 20,21 NMR spin-echo tech- niques; relaxation time measurements of 2H and of quadru-polar ionic nuclei;22 diaphragm cell experiments on C1-self-diffusion using the radiotracer 36Cl-.23 We draw particu- lar attention to the methods for obtaining highly accurate data, described in these papers. Results We first consider isotope effects upon the self-diffusion coeffi- cient of water in aqueous CsCl solutions.Two methods were applied to obtain such data. First, we have supplemented conventional 'H NMR spin-echo measurements of H20 self- J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 diffusion by analogous experiments using the 2H resonance of D20. Secondly, in a new type of experiment, we have per- formed 'H spin-echo measurements in H20-D20 mixtures of different isotopic compositions, followed by linear e~trapolation~~to zero proton concentration, as shown for a typical example in Fig. 1.This procedure yields the tracer diffusion coefficient of HDO in D20, which in pure water differs by a factor 1.015 from the true self-diffusion coefficient of D20, owing to the different tracer mass.12 The same cor- rection is then also applied for aqueous solutions. Fig. 2(a) summarizes the resulting isotope effects upon the self-diffusion coefficients of water obtained by the two methods. Note that for a significant comparison of data in H20 and D20 one has to refer to equal 'aquamolalities', m*, defined by the number of mol of salt per 55.5 mol of solvent. Moreover, in order to enable a direct comparison with vis- cosity data, Dr;: rather than D,has been plotted. Every data point in Fig.2(a) is an average over frequent experiments with an estimated uncertainty of f0.015. For brevity, we do not quote the absolute values for the self-diffusion coefficients in H20 or D20, but refer to the data for CsCl in H20 obtained by Hertz and Mills25 which were reproduced by us to better than f1%. Values for D20 can then be recalculat- ed in conjunction with the isotope effects shown in Fig. 2(a). In order to obtain information on water reorientation in CsCl solutions we have monitored 2H relaxation. The spin- lattice relaxation rate, 1/Tl, is related to the reorientational correlation time z2 of the water molecules by l/Tl = 3/h~~(e~Qq/h)~z, (3) In pure D,O the nuclear quadrupole coupling constant (e2Qq/h) is near 250 kHz, so that z2 = 2.5 ps'4*26 [an addi- tional factor in eqn.(3) due to the asymmetry of the electric field gradient is negligible in water]. The isotope effect upon water reorientation is then obtained by comparing data for pure D20 with data for D20 (HDO) dissolved at trace con- centrations in H20. In practice, the latter figure is obtained by performing experiments at various mole fractions of D20 in H20-D20 mixtures, followed by extrapolation to zero deuteron concentration. Assuming that the quadrupole coup- ling constant does not depend on the isotopic composition, this allows the determination of isotope effects upon 72, regardless of an exact knowledge of the coupling constant. In fact, with pure water we have obtained a value of 1.223,suffi-2.o 0 0.2 0.4 0.6 0.0 t Xn Fig.1 Self-diffusion coefficient of protonated species in a 2 rnol (55.5 mol solvent)-' CsCl solution as a function of the isotopic com- position of water expressed by the mole fraction of protons. The lim- iting value at xH-+0 yields the tracer diffusion coefficient of HDO in D,O. After correction for the tracer mass (dashed line) this limiting value is equal to the self-diffusion coefficient of neat D,O in the CsCl solution. J. CHEM. SOC. FARADAY TRANS., 1994. VOL.. 90 85 1 I l.15t 1.10--2 4 6 6 h1.10 LJ 110 2 ll0lv 0 ' I 1.10 ' I 1.051 1.0511 !I 10L-' I ' ' 1 1.0-J m* m* Fig. 2 Dynamical H20/D20 isotope effects <, upon various tram- port coeficients and microdynamical quantities in aqueous CsCl solutions as a function of salt concentration. tr corresponds to (4 0,;;(6)~2r;(c)0,t.s; (d)DLh!; (l/T1)r+Cs; (f)(1/Tl)r,c1*The solid lines represent qr .All quantities are defined in eqn. (2a)-(2g).The salt concentration is in aquamolality units, i.e. rnol CsCl per 55.56 rnol of solvent. ciently close to the value 1.228 obtained for the bulk viscosity and inverse self-diffusion coefficient to justify this procedure. Results obtained in the same way for CsCl solutions are shown in Fig. 2(b).The estimated uncertainty is 0.01 5. Fig. 2(c) and (6) show isotope effects upon the self-diffusion coefficients of the Cs' and C1-ions. For absolute values obtained in H20 we refer to an earlier publication containing data for Cs' 2o and to the work of Hertz and Mills2' containing data for C1-.Again, for easy comparison with other quantities, inverse ratios 0,' have been plotted. The estimated accuracy is & 0.02 for Cs + and & 0.01 for CI -. respectively. The corresponding limiting values at infinite dilution of the ions can be calculated from the accuratelj known limiting ionic conductances ikl'means of the by Nernst-Einstein relation D, = F2/RTj., . Note that the resulting isotope effects = 1.199 and DrIhl= 1.215 at infinite dilution do not correspond to what is expected from the bulk viscosity.' ' Data for the magnetic spin-lattice relaxation rates of 133Cs+and 35Cl- dissolved in H,O have been measured in the context of a general research program of our joint labor- atories dealing with the fundamentals of the quadrupolar relaxation of ionic nuclei in aqueous electrolyte solutions.The results of these studies will be reported in detail else- here.^' We have supplemented these data by measurements of CsCl solutions in D,O. The resulting isotope effects are summarized in Fig. 2(e) and (f).They should possess an accuracy of & 0.02. While in all cases considered above, data for the limit of infinite dilution are available, this is not so for the ionic relax- 1.a 1 m u ", 1.10k. cv I I1.0 0.1 0.2 m* Fig. 3 Isotope effects ( l/Tl)r.cs upon 133Cs'relaxation rates at low salt concentrations. m* is measured in aquamolality units.ation rates. We have therefore attempted to extend the relax- ation time measurements to salt concentrations as low as possible to estimate relaxation rates at infinite dilution. In the course of these experiments sufficiently accurate results for isotope effects could be obtained for Cs' relaxation down to about 0.01 rnol (55.5 rnol solvent)-' in salt concentration, while the results for 35Cl -relaxation in this concentration range did not allow the extraction of isotope effects with the desired accuracy. Fig. 3 shows these low-concentration results for '33Cscrelaxation. The data should possess an accuracy of k0.03. Discussion It seems appropriate to consider first some aspects of dynamical isotope effects in pure water. The higher viscosity of D20 as compared with H20 is usually ascribed to a stronger 'structuredness' of D20due to stronger hydrogen bonding. In fact, 'HfH substitution changes the moments of inertia which affect the intermolecular vibrations.28 In partic- ular.the infrared librational band of pure H,O at 685 cm-' is shifted to 505 cm-' in D20,29which corresponds to the square root of the moments of inertia. This mode directly reflects the restraints produced by hydrogen bonds, which are, hence, stronger in D20 than in H20. Computer simula- tions show3' that such librational and also hindered trans- lational modes associated with the bending of bonds contribute significantly to the power spectrum of the velocity autocorrelation function which determines the self-diffusion coeficient, thus yielding a natural explanation for the translation-rotation coupling evidenced by the self-diffusion coefticien ts.As added salts change the hydrogen-bonded structure of water, one would expect the dynamical isotope effects to change also. This can indeed be extracted from viscosity data for alkali-metal chloride^.'^.^' These data show a substantial reduction of qr upon addition of structure-breakers. In the case of CsCl this amounts to a decrease from 1.228 in pure water to about 1.150 at 6 rnol (55.5 rnol solvent)-'. Fig. 2(a)-(f) show the resulting curve for qr for comparison with the data for the other dynamical quantities considered in this work. In contrast, the isotope effect decreases little upon the addition of structure-forming LiCI." No change or even a slight increase is observed with hydrophobic tetra-alkylammonium ions.6 In other words, the structure-breaking effect is larger in D,O than in H20, while structure formers show about the same effect in both solvents.The common explanation is that owing to the greater 'structuredness' of D20,the breaking effect is also more pronounced.6 How is this behaviour reflected by the microdynamical quantities considered in this work? To this end, we discuss first the isotope effects upon the self-diffusion coefficient and reorientational correlation time of water in Fig. 2(u) and (b). The major result is that the isotope effect upon water reorien- tation coincides with the behaviour of the viscosity, while at high salt concentrations the self-diffusion data lie systemati- cally above the viscosity curve.Based on the estimated uncer- tainties, the latter effect is near the limits of experimental error, but its systematic appearance for data obtained by two different methods seems to ensure that it is real. In other words, the isotope effect upon self-diffusion of water is less affected by addition of a structure-breaker than that upon the viscosity and water reorientation. In this sense, a presumed breaking of hydrogen bonds also decouples the rotational and translational motions. Let us next consider the isotope effects upon the self- diffusion coefficients of the Cs+ cations in Fig. 2(c) and (4, respectively. As noted, in these cases the infinite dilution values do not coincide with what is predicted from the bulk viscosity.As discussed in detail by Kay and Evans,'' this indicates that in the cosphere of ions the isotope effect is dif- ferent from that in the bulk. It is then interesting to see what happens at finite salt concentrations, and above all in the high-concentration regime, where all water molecules are in the cosphere of ions. Our results indicate that and Dr.&approach qr. This is particularly obvious from the data for Cs' diffusion, as in this case at low concentrations the difference is sufficiently large to show up in our measurements, and its disappearance with rising m* is clearly visible. This is more difficult to prove for CI-, where at zero salt concentration the difference is smaller, and possibly not beyond the limits of uncertainty of our experiments.In any case, within the limits of uncertainty of our data no difference between qr and Dr;kl is seen at high salt concentrations. +Finally, we consider 33Cs and "Cl- spin-lattice relax- ation. We note that apart from the application in the study of isotope effects, these data are important in their own right, as they give significant information on ion-solvent and ion-ion interactions.lo The latter aspects are treated el~ewhere.~' For quadrupolar nuclei in monoatomic species, the relaxation rate l/Tl may be written as" (4) where A is a constant for the given nucleus, (VL) is the mean square electric field gradient q at the nucleus and z, is the correlation time for fluctuations of q.Electrostatic theory assumes that the q arises from the electrostatic moments of solvent molecules and from the charges of counter ions. The detailed expressions are comple~,~*~~*'~ but we can use here the simple result that, at the level of accuracy needed here, (VL) is the same in H20 and D20? Hence, (l/Tl)r reflects isotope effects on the correlation time z,. In fact, zq is a particularly interesting probe for local dynamics, as theory predicts (V;) to exhibit a very short- range R-' distance dependence for the ion-solvent inter-action. Also, we know from computer simulations of ionic" and uncharged" monoatomic solutes, that zq reflects at least three modes of motion, namely short-time librational motions of water molecules, lateral motions which disturb the symmetry of the first cosphere and reorientational motions of water molecules in the first cosphere.Hence, zq probes pro- cesses which are quite different from those reflected by the dynamical quantities discussed above. As a complication, there appears, however, in eqn. (4) an additional term (l/7'l)remfor the background contribution due to the remaining relaxation mechanisms. Owing to its large quadrupole moment, 35Cl- relaxes exclusively by quad- rupolar interaction. For '33Cs the quadrupolar interaction + is weaker, but the common belief is that, nonetheless, '33Cs+ J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 relaxes only by the quadrupolar mechanism." This may be justified for most applications, but in the context of the very subtle effects considered here, we cannot disregard the possi- bility of a small 133C~-1H dipolar contribution in H20, say of the order of 3%.Owing to the lower magnetic moment of 2H, this dipolar contribution would be absent in D,O, so that the data could indicate an isotope effect up to 3% smaller than actually present for zq. The major result of the data in Fig. 2(e) and (f)is that in the entire concentration range the isotope effects upon the quadrupolar relaxation rates are substantially smaller than that upon the viscosity. Moreover, the curve for the 13'Cs+ cation lies below that for 35Cl-. Qualitatively similar obser- vations have been made in our previous work on 2'Na+, 87Rb+ and 'lBr- relaxation in aqueous alkali bromide and iodide sol~tions,~~*~~ but as CsCl is a stronger structure- breaker, all effects are more pronounced, and from an experi- mental point of view, these larger effects enable a more detailed investigation.Let us consider first the isotope effects at high dilution, where ion-solvent interactions will prevail. While it is impos- sible to perform measurements at concentrations low enough to extract accurate infinite dilution values, the data in Fig. 2(e) and (f)and the low-concentration data in Fig. 3 show that neither the viscosity ratio of 1.228 nor the somewhat lower ratios of the limiting mobilities, 1.199 and 1.215 for Cs' and C1-, are approached.This is also true, if a small dipolar contribution to "'Cs relaxation is taken into account. Hence, in the direct neighbourhood of an ion, the isotope effect upon quadrupolar relaxation rates is quite dif- ferent from that in the bulk. This is to a minor degree already evidenced by the limiting ionic mobilities," but is obviously reflected in a more pronounced manner by quadrupolar relaxation, where due to the R-' dependence of the mean- square field gradient the nearest neighbourhood of the ion is probed more selectively. In view of these results, it is hard to escape the conclusion that the structure-breaking effect extends to the first cosphere of the ions. The question whether the structure-breaking effect has its origin in the first or second cosphere has been debated for some time,2*35-39 starting with the well-known model of Frank and Wen" which assumes the existence of a first hydration sphere with rather immobilized water mol- ecules, surrounded by a second region, where structure-breaking takes place.While it appears to be impossible to test this model by direct experiments, molecular dynamics simulations of the translational and reorientational dynamics of water mol-ecules near iodide ions" have indicated that molecular motions are also accelerated in the first hydration shell. This corresponds to an earlier interpretation of magnetic relax- ation data by Engel and Hertz2 and of diffusion data by Sam- 0il0v.~~The picture is" that, for large ions, the ordering influence exerted on water molecules in the first hydration shell by ions becomes comparable to the influence of the sur- rounding water molecules.In this case the energy barriers will be flattened out, and the thermal motions will break up the local structure. With respect to cationic hydration, this is also confirmed by recent neutron scattering data for the water structure around K ions.40 These have yielded a com- + paratively broad and featureless first hydration shell around K+ ions, at contrast with well resolved and ordered shells for the smaller Li+ and Na+ ions. Finally, let us consider the behaviour at high salt concen- trations. From Fig. 2(e) and (f) we find little concentration dependence.As at the same time qr decreases, the isotope effects on the ion relaxation approach those on the other dynamical quantities. Whether eventually they will become J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 853 the same depends largely on whether 133Cs+relaxation is 4 K. Crickard and J. F. Skinner, J. Phys. Chem., 1969,73,2060. purely quadrupolar. Unfortunately, the latter question is difficult to answer. Theoretical estimates of the '33C~-1H dipolar interaction give some support to such a contribution, but owing to the RP6dependence of the dipolar interaction much rests on the Cs'-0 distance used in these calculations. It appears also to 6 7 8 K. Ibuki and M. Nakahara, J. Chem. Phys., 1985,85,7312. K. Ibuki and M. Nakahara, 2. Naturforsch., Teil A, 1991, 46, 127.L. Endom, H. G. Hertz, B. Thull and M. D. Zeidler, Ber. Bun- senges. Phys. Chem., 1967, 71, 1008. H. G. Hertz, M. Holz and R. Mills, J. Chim. Phys., 1974, 71, 1355. be impossible to prove its existence experimentally by NOE measurements. However, there is an interesting indirect argu- ment in favour of a small dipolar contribution: One could expect that when hydration spheres overlap, isotope effects upon cationic and anionic relaxation become the same.l3 This has indeed been found in earlier work on s7Rb+ and 9 11 12 13 14 H. G. Hertz, Ber. Bunsenges. Phys. Chem., 1973, 77,477. M. Holz, Progr. N.M.R. Spectrosc., 1986, 18, 327. R. L. Kay and D. F. Evans, J. Phys. Chem., 1965,69,4216. R. Mills, J. Phys. Chem., 1973, 77, 685, M. Holz, J.Chem. SOC., Faraday Trans. 1, 1978,74,644. D. Lankhorst, J. Schriever and J. C. Leyte, Ber. Bunsenges. Phys. Chem., 1982,96,215. 'Br -relaxation in aqueous RbBr.' The present results for CsCl can only be reconciled with this expectation by assuming a 133C~-1Hdipolar contribution of about 3% over the entire concentration range. In any case, there remains the fact that even at 6 mol (55.5 mol solvent)-', where each water molecule is in the solvation 16 17 18 H. J. Tyrrell and K. R. Harris, Difusion in Liquids, Butterworths, London, 1984. H. L. Friedman, in Molecular Motions in Liquids, ed. J. Lascombe, Reidel, Dordrecht, 1974, p. 87. H. Weingartner, M. Holz, A. Sacco and M. Trotta, J. Chem. Phys., 1989,91, 2568. M. Holz, H. Weingartner and A. Sacco, Ber.Bunsenges. Phys. sphere of an ion, and where hydration spheres begin to overlap, the isotope effects are still substantially higher than one would predict from the square-root-of-mass law. In other words, the translation-rotation coupling evidenced by such high values still persists in concentrated solutions, although to a lower extent than observed in the pure solvent. How 19 21 22 Chem., 1990,94,332. A. G. Ostroff, B. S. Snowdon Jr. and D. E. Woessner, J. Phys. Chem., 1969,73,2784. B. M. Braun and H. Weingartner, J. Phys. Chem., 1988,92, 1342. M. Holz and H. Weingartner, J. Magn. Reson., 1991,92, 115. M. Holz, A. Sacco and M. Trotta, J. Solution Chem., 1990, 19, 193. these effects are quantitatively related to the different hydro- gen bonding in H,O and D20 remains to be investigated by theory or simulations.We note, however, that in contrast to the common belief, large H/D isotope effects upon transport coefficients are not limited to water or to hydrogen-bonded protons in non-aqueous solvents. Isotope effects which 23 24 26 27 28 H. Weingartner, B. M. Braun and J. M. Schmoll, J. Phys. Chem., 1987,91, 979. H. Weingartner, Ber. Bunsenges. Phys. Chem., 1984,88,47. H. G. Hertz and R. Mills, J. Chim. Phys., 1976,73,499. B. C. Gordalla and M. D. Zeidler, Mol. Phys., 1986, 59, 817. M. Holz, X. Mao and A. Sacco, to be published. G. E. Walrafen in Water. A Comprehensive Treatise, ed. F. exceed by far what is predicted from the rn''2-law have also been found for other molecules with strongly directional interactions, we quote dimethyl sulf~xide/[~H,]dimethyl sulf- oxide with 0,' = 1.12 as opposed to rn'" = 1.038 as a typical example.' 29 31 Franks, Plenum, New York, 1972, vol.1, ch. 5. D. A. Draegert, N. W. B. Stone, B. Curnutte and D. S. Williams, J. Opt. SOC. Am., 1966, 56, 64. G. Szasz and K. Heinzinger, J. Chem. Phys., 1983,79, 3467. A. Selecki, B. Tyminski and A. G. Chmielewski, J. Chem. Eng. Data, 1970, 15, 127. 32 H. G. Hertz, M. Holz and A. Sacco, Chem. Scr., 1989,29,291. The authors are grateful to the Ministry of University and of Scientific and Technological Research (MURST), Italy, for financial support. H.W. thanks the Consiglio Nazionale delle Ricerche (CNR), Italy, for a visiting fellowship at the Uni- versity of Bari. 33 34 J. Schnitker and A. Geiger, 2. Phys. Chem. (Munich), 1987, 155, 29. M. Holz and C. K. Rau, J. Chem. SOC., Faraday Trans. 1, 1982, 78, 1899. H. S. Frank and W-Y. Wen, Discuss. Faraday SOC., 1957, 24, 133. 36 H. L. Friedman, in Water. A Comprehensive Treatise, ed. F. References 37 Franks, Plenum, New York, 1973, vol. 3, ch. 1. Faraday Discuss. Chem. SOC., 1977,64;general discussions. 1 H. S. Frank and M. W. Evans, J. Chem. Phys., 1945,13, 507. 2 G. Engel and H. G. Hertz, Ber. Bunsenges. Phys. Chem., 1968.72. 808. 38 39 A. Geiger, Ber. Bunsenges. Phys. Chem., 1981,85, 52. 0.Ya Samoilov, Discuss. Faraday SOC.,1957,24, 141. G. W. Neilson, Z. Naturforsch., Teil A, 1991,46, 100. 3 R. A. Robinson and R. A. Stokes, Electrolyte Solutions, Butter-worths, London, 1969. Paper 3/07041G; Received 29th November, 1993