6594
J . Phys. Chem. 1990, 94, 6594-6598
we deduce that their unusual sharpness and relative intensity are due to the quasi axiality of the tensor of the CH proton (see Table 11).
Correlation Times. The rotational correlation times of nitroxides in the fast motional regime are usually obtained from the spectra by the means of a simple expression for the line width as eq 6. This expression does not allow for the relaxation effects of the proton interaction anisotropies, and therefore it is interesting to compare these approximate values with those obtained by the means of the full theoretical expression eq 7. The more accurate value of T~ turns out to be smaller than the approximate one for Tempyo and larger for Tempol. This difference is due to the
different signs of the line-width coefficients in eq 7, and the sign of these coefficients is determined by the relative orientations of the A, and A, tensors. Therefore, the influence of the proton interaction anisotropy on the accurate determination of the average rotational correlation time T~ depends upon the geometrical structure of the radical. On the other hand, we found that the rotational anisotropy parameter N = R , , / R , is less sensitive to the accuracy of the simulation. In fact the same value was obtained by both analysis. Registry No. I , 3229-73-0; 11, 106367-37-7; 111. 2226-96-2; IV, 14691-88-4; V, 14691-89-5.
Raman Spectroscopic Study of MethanolCc Electrolyte Solutions in Liquid and Glassy States Shigeru Yamauchi and Hitoshi Kanno* Department of Chemistry, The National Defense Academy, Hashirimizu, Yokosuka, Kanagawa 239, Japan (Received: October 26, 1989; In Final Form: March 7 , 1990)
Raman OH stretching spectra were measured for glassy methanolic solutions of alkali-metal halides. The Raman spectra for the glassy solutions at liquid nitrogen temperature exhibited fine structures in contrast to the spectra of the solutions at room temperature, which showed only a single featureless envelope. Two relatively narrow Raman peaks (one at 3385 cm-' and the other at 3320-3340 cm-I) appeared and grew with an increase in halide concentration in glassy methanolic LiCl solutions. The higher frequency peak was assigned to the OH stretching Raman band arising from methanol molecules weakly hydrogen bonded to halide ions and the lower frequency one to the OH stretching Raman band due to methanol molecules not only hydrogen bonded to halide ions but also coordinated to cations by their oxygen lone pairs. It is shown that ionization of some electrolytes (NaBr, Lil, and NaI) is greatly suppressed in glassy methanolic solutions.
Introduction Raman spectroscopy is a useful technique for characterizing solvation structures of dissolved ions in aqueous and alcoholic solutions.'-7 Alcoholic solutions of electrolytes have been extensively studied by Raman4-7 and infrared8-" techniques. However, most of these studiese9 are confined to the investigations of alcohol solutions at ordinary temperatures. Strauss and SymonsloJ1 have made extensive infrared studies of alcoholic solutions of various electrolytes at low temperatures, which have revealed that fine structure appears in infrared O H stretching spectra for glassy alcoholic solutions. Kanno and H i r a i ~ h i 'have ~ , ~ ~reported that Raman spectra for glassy aqueous solutions usually give better resolved Raman bands than those for the corresponding solutions at room temperature. These observation^'^'^ lead us to expect that a Raman spectrum for a glassy alcoholic solution should have better resolved features, which may be more informative about solvation than that for the solution at ordinary temperatures. In fact, a well-resolved Raman O H stretching spectrum has been obtained for a glassy ethanolic solution of lithium ch10ride.I~ Thus, as a continuation of our studiesi2-I4of the Raman OH stretching spectra of glassy aqueous and alcoholic electrolyte solutions, we investigated the Raman spectra of glassy methanol solutions of alkali-metal halides as a funtion of salt concentration. As alcohols behave as a hydrophilic solvent and have solvation structures analogous to aqueous solutions, Raman results for glassy alcoholic solutions should give us deeper insight into solvation in both alcoholic and aqueous solutions. Experimental Section All alcoholic solutions were prepared by dissolving commercially available anhydrous salts in absolute methanol (>99.5%). Here *To whom correspondence should be addressed
0022-3654/90/2094-6594$02.50/0
the salt concentration was denoted by R (= moles of alcohol/moles of salt). The sample solution was placed in a 3-5 mm i.d. Raman cell with a flat bottom and then rapidly cooled by being immersed into liquid nitrogen. The overall cooling rate was approximately 4 X IO2 K/min. Glass formation was checked visually. Methanolic solutions are generally vitrifiable, even with a low salt concentration, despite the fact that pure methanol is almost impossible to vitrify with a cooling rate of 4 X IO2 K/min. Raman spectra were obtained with a JASCO NR-1100 spectrometer using -300 mW of the 514.5 nm line of a NEC argon ion laser as an exciton source. The glassy sample was kept at liquid nitrogen temperature during Raman measurements by using a (1) Lilley, T. H. In Water: A Comprehensive Treatise; Franks, F., Ed.; Plenum: New York, 1973; Vol. 3, Chapter 6. (2) Irish, D. E.; Brooker, M. H. In Advances in Infrared and Raman Spectroscopy; Clark, R. J . H., Hester, R. E., Eds.; Heydon: New York, 1976; VOI. 2, pp 212-311. (3) Brooker, M. H. In The Chemical Physics of Solvation; Ulstrup, J . , Dogonadze, R. R., Kalman, E., Karnyshev, A. A,, Eds.; Elsevier: The Netherlands, 1986; Chapter 4. (4) Kecki, 2.Spectrochim. Acta 1962, 18, 1155 and 1165. (5) Minc, S.; Kurowski, S. Spectrochim. Acta, 1963, 19, 330. (6) Hester, R. E.; Plane, R. A. Spectrochim. Acta 1967, 23A, 2289. (7) AI-Baldawi, S. A.; Brooker, M. H.; Cough, T. E.: Irish, D. E. Can. J. Chem. 1970, 48, 1202. (8) Adams, D. M.; Blandamer, M. J.; Symons, M. C. R.; Waddington, D. Trans. Faraday Soc. 1971, 67, 61 1. (9) (a) Symons, M. C. R.; Waddington, D. Chem. Phys. Lett. 1975, 32, 133. (b) Strauss, I. M.; Symons, M. C. R. Chem. Phys. Lett. 1976,39,471. (10) Strauss, 1. M.; Symons, M. C. R. Chem. Phys. Lett. 1976, 39, 471. (1 I ) Strauss, I. M.; Symons, M. C. R. J . Chem. Soc., Faraday Trans. I , 1971, 73, 1796; 1978, 74, 2146. (12) Kanno, H.; Hiraishi, J. Chem. Phys. Lett. 1979, 68, 46. ( I 3) Kanno, H.; Hiraishi, J. Chem. Phys. Lett. 1980, 72, 541. (14) Yamauchi, S.; Kanno, H. Chem. Phys. Lett. 1989, 154, 248.
0 I990 American Chemical Society
The Journal of Physical Chemistry, Vol. 94, No. 17, 1990 6595
Raman Study of Methanolic Electrolyte Solutions
Glassy s t a t e
Liquid state
> c .VI 0 C
-
1
pure EtOH
1
/
3400 3200 Raman shift I cm-l
3600
3000
Figure 1. Raman spectra in the OH stretching region for the H20. 40MeOH solution and pure ethanol. Note that the intensity scales of the two types (liquid and glassy states) of spectra are not the same. Thin curve, liquid state at room temperature; thick curve, glassy state at liquid
nitrogen temperature. specially designed Dewer vessel. Some ethanolic metal halide solutions were also prepared, and their Raman spectra were measured for comparison. Several attempts to make semiquantitative measurements of Raman intensities of methanolic LiX solutions were made by adding perchlorate ions to the solutions as an internal standard. However, the Y , band (at 932 cm-I) of perchlorate ions, which is usually used as an intensity standard, is far from an O H stretching Raman band and was found to be unsuitable for our purpose because reliable intensity measurements were difficult to make with our simple equipment for glassy samples. Furthermore, addition of perchlorate to a methanolic solution in the necessary quantity (molar ratio LiC104/LiCI -0.2) gives rise to significant changes in spectral contours, which seriously interfere with the quantitative intensity measurements. Therefore, Raman intensity measurements were abandoned in this study.
Results and Discussion Methanol. Because pure methanol is very difficult to vitrify with the cooling rate (-4 X IO2 K/min) employed in this study, we could not obtain the glassy state of pure methanol. Therefore, the Raman spectra of an H20.40MeOH (MeOH = methanol) solution and pure ethanol were instead examined as alternatives for a Raman spectrum of glassy pure methanol. Figure 1 shows the Raman OH stretching spectra for the H2040MeOH solution and pure ethanol in both liquid and glassy states. Comparison of the spectra indicates that both the solutions give similar spectral changes from a liquid state to a glassy state. On vitrification the intensity of the low frequency region around 3 100-3300 cm-' (peak at 3 175 cm-l) increases in compensation for the decrease of the higher frequency region, implying the strengthening of hydrogen bonds among alcohol molecules from a liquid state at room temperature to a glassy state. This feature is essentially the same phenomenon as those of water and aqueous solutions on vitrification.Izl3 Addition of a small amount of water to alcohol induces only a small change in spectral contour. In fact, almost the same Raman spectra were obtained for both pure ethanol and an H20.40EtOH solution (EtOH = ethanol) although the Raman spectrum for the latter solution is not shown in this
Ramon shift I cm-'
Figure 2. Raman spectra in the OH stretching region for the LiCI. RMeOH ( R = 4-30) solutions in both liquid and glassy states.
report. Thus, it is concluded that the Raman OH stretching spectrum for glassy pure methanol must be very similar to that for the H20.40MeOH solution and that the Raman spectrum for the H20.40MeOH solution can be used as a starting point in discussing the effects of salt addition on the O H stretching spectrum. Methanolic Metal Chloride Solutions. Raman O H stretching spectra for the glassy LiCI-RMeOH solutions ( R = 4-30) are shown in Figure 2 together with those for the solutions in the liquid state. It is easily seen that Raman spectra for the solutions at room temperature give similar spectral contours while those for the solutions in the glassy state change significantly with increase in salt concentration. We are, of course, aware of the report by Abe and ItoIS that the O H stretching Raman spectrum of an alcoholic lithium halide solution increases in its intensity with a rise in halide concentration due to the pre-resonance Raman effect of solvated halide ions. From their Raman results, it is expected that Raman intensity will increase with an increase in halide concentration. In the case of aqueous solutions of LiX and CaX (X = C1, Br, and I), Kanno and HiraishiI6 observed a significant increase in the intensity of the whole water Raman spectra for the solutions in both liquid and glassy states. Thus, a Raman spectrum for a methanolic solution with a higher halide concentration should give a higher intensity than that for the one with a lower halide concentration. Therefore, it is evident that the measurement of Raman intensity is necessary for detailed discussion of the effects of salt addition on Raman spectral features. However, due to the experimental difficulty described in the Experimental Section, we have not made quantitative Raman intensity measurements but instead interpret the spectral changes associated with variation in salt concentration and vitrification. Although all the Raman spectra for the solutions at room temperature have a similar spectral shape, the peak maximum gradually shifts to higher frequencies with an increase in salt concentration, reflecting the weakening of hydrogen bonds among alcohol molecules and solvated ions (Table I). An important aspect in the results shown in Figure 2 is that a relatively large spectral change is observed on going from the liquid state at room temperature to the glassy state at liquid (15) Abe, N.; Ito, M . J. Raman Spectrosc. 1978, 7, 161. (16) Kanno, H.; Hiraishi, J. J . Phys. Chem. 1983, 87, 3664.
6596 The Journal of Physical Chemistry, Vol. 94, No. 17. I990 TABLE I: Peak Frequencies of the OH Stretching Raman Bands for the Alcoholic Electrolyte Solutions in Both Liquid and Glassy States glassy state liquid low-frequency high-frequency state band, cm-I band, cm-’ cm-I solvent solute Ra 3384 3336 MeOH LiCl 30 3385 3338 20 3384 3349 10 3324 337 I 3385 6 3326 3387 3391 4 3337 3410 LiBr 3360 20 3410 3373 10 3352 341 3 3385 6 3360 3400 Lil 20 3352 3419 IO 3378 3426 6 3387 341 I 3400 20 NaBr 341 8 Nal 20 3360 3435 10 3406 3386 3 300 3294 MgC12 20 ZnCI, 3385 3330 20 3372 3320 EtOH LiCl 30 3341 20 3371 3373 3360 10 3309 3360 LiBr 3390 20 3373 3392 10 3338
“ R is the molar ratio of solvent/solute.
nitrogen temperature. As is observed for pure alcohol and aqueous solution^,'^*^^ vitrification induces the shift of the whole O H stretching spectrum to lower frequencies. The shift is the reflection of the recovery of strong hydrogen bonds among alcohol molecules at low temperatures. This feature is a common property for hydrogen-bonded liquid systems.lJ7 A remarkable point in the Raman spectra for glassy solutions is that a sharp peak appears at 3385 cm-’ and grows rapidly with an increase in halide concentration. When the concentration of LiCI becomes about R = -20, another peak, which is relatively narrow but definitively broader than the peak at 3385 cm-I, appears at about 3324 cm-I and grows in compensation for the decrease of the low frequency Raman band at -3175 cm-I, which is ascribed to the OH stretching vibrations of methanol molecules having basic liquid methanol structure. In contrast with the sharp peak at 3385 cm-I, it gradually moves to higher frequencies, though only a little, with rise in salt concentration but shows a smaller growth rate than the peak at 3385 cm-I. These spectral changes with halide concentration are essentially the same as observed from glassy LiCI-EtOH solution^.'^ The frequency of the sharp peak is higher for a glassy methanolic solution than for a glassy ethanolic solution by 10-1 5 cm-’ (Table I). As the sharp peak is attributed to the O H stretching vibrations of alcohol molecules weakly hydrogen bonded to dissolved halide ions, it is considered that the hydrogen bonds between methanol molecules and halide ions are weaker than those between ethanol molecules and halide ions. Appearance of resolved OH stretching bands in methanolic bands in methanolic electrolyte solutions at low temperatures has been reported by Strauss and SymonsIoJI in their infrared studies. They observed several narrow resolved infrared bands of glassy methanolic solutions of electrolytes such as alkali-metal salts and tetraalkylammonium salts. They also suggested that those resolved infrared bands are obviously due to the OH stretching vibrations arising from solvated structures around anions and cations. As the selection rules for infrared-active vibrations are different from those for Raman-active vibrations,I8 direct inference from Strauss and Symons’ assignments of bands is not applicable to the Raman bands observed here. However, we agree with Symons’ interpretation” that narrow separated bands are assigned to the OH stretching modes of solvent molecules bonded to anions and/or (17) Kollman, P. A.; Allen, L. C. Chem. Reo. 1972, 72, 283.
(18) Nakamoto, K. In Infrared and Raman Spectra of Inorganic and Coordination Compounds. 3rd ed.; Wiley, New York, 1978; Chapter I ,
Yamauchi and Kanno Glassy state
MgCl2.20MeOH
I
3800
3330
3c 0
Raman shift I c m - ’ Figure 3. Raman spectra in the OH stretching region for the MgCI,. 20MeOH and the ZnC12-20MeOH solutions in the glassy state.
to both anions and cations. Hester and Plane6 also reported that all the electrolytes studied produced the shift of the OH stretching Raman bands to higher frequencies and that the rise of the O H frequency is in the order C104- > NO3- > > CI- > F. As reported in a previous paper,I4 the narrow peak at 3385 cm-l (3373 cm-l for an ethanolic LiCl solution) is assigned to the O H stretching vibrations arising from the O H groups directly and weakly hydrogen bonded to chloride ions. The assignment is supported by the fact that the peak appears at the same frequency in glassy M~C1~.20Cl~.20MeOH and ZnC12-20MeOHsolutions (Figure 3). A small cation dependence of the band frequency is the strongest experimental evidence that the band is due to the OH stretching vibrations of methanol molecules having no direct interaction with cations. The weakness of the band in the glassy MgCI2.20MeOH and ZnC12*20MeOHsolutions should be partly due to nonionization of dissolved MgCI2 or ZnC12 species and partly due to the formation of chloro complexes of metal ions. There is some experimental evidence that most chloride ions exist in either solvated ZnCI, species or complex ions (ZnC13- and ZnCI,*-) in the glassy ZnC12.20MeOH s o l ~ t i o n . ’ ~ It is to be noted that an OH stretching band due to “free O H groups” appears at about 3630 cm-I: Liddel and BeckerZ0observed a sharp peak at 3630 cm-’ in a methanol-CCI, system and subsequently Symons and Thomas2’ confirmed it in their infrared study. Therefore, the peak at 3385 cm-I may be called the O H stretching vibrations of “weakly hydrogen-bonded OH” or “anion solvated OH”. The sharpness of the peak indicates that the interaction of the methanol molecules weakly hydrogen bonded to halide ions with other adjacent methanol molecules and/or cations is very weak. It is a general observation that a strong hydrogen-bonded system gives a broad OH stretching Raman spectrum as seen for pure water and alcohols.22 Next we discuss the broader peak at about 3320 cm-I. In a previous paper,14 the peak was assigned to the OH stretching vibrations of alcohol molecules in second solvation spheres. The most important aspect of the results shown in Figure 2 and TabIe 1 is that there is a small positive frequency shift with increase in halide concentration. Furthermore, there is a clear cationic effect (19) Yamauchi, S.; Kanno, H. Unpublished data. (20) Liddel, U.; Becker, E. D. Spectrochim. Acta 1957, 10. 70. (21) Symons, M. C. R.; Thomas, V. K. J . Chem. Sor., Faraday Trans. I 1981, 77, 1883. (22) Pauling, L. In The Nature of the Chemical Bond; Cornell University, New York. 1960; Chapter 12
Raman Study of Methanolic Electrolyte Solutions
a
The Journal of Physical Chemistry, Vol. 94, No. 17, 1990 6591
I
Glassy s t a t e
d
Llquid sta'te
b
I-i R = 20
3800
3400
3000
3400
3000
Raman s h i f t I cm-'
Figure 5. Raman spectra in the OH stretching region for the LiBr. RMeOH ( R = 6-20) and the NaBr-2OMeOH solutions in both liquid and glassy states. Figure 4. Schematic solvation diagrams for the Raman bands at 3320 cm-' (a) and at 3385 cm-' (b) in the glassy LiCLRMeOH solutions.
[the glassy LiCl solution (R = IO), 3324; the glassy MgC12solution (R = 20), 3294 cm-I], though in some solutions the band becomes obscured due the overlap with other Raman bands. This implies that there are some interactions between cations and alcohol molecules responsible for the band. The fact that the band is highest (3337 cm-I) in the R = 4 solution indicates that the alcohol molecules responsible for the band should exist mainly in the first solvation spheres because there should not be an appreciable amount of alcohol molecules in the ordinary second solvation spheres at such a high halide concentration. Summing up these considerations, it is concluded that the peak can be ascribed mainly to the OH stretching vibrations of methanol molecules interacting with both anions and cations. Since cations are expected to solvate by forming weak bonds with oxygen lone pairs of methanol molecules, the hydrogen bonds between halide ions and the methanol molecules coordinated to cations should be polarized and stronger than those between halide ions and the methanol molecules without interaction with cations. Radnai et al.23made an X-ray diffraction study of a methanolic MgClz solution and revealed that the first solvation shell around a CI- ion is composed of six methanol molecules. A recent molecular dynamic study of a methanolic MgCI2 solution by Tamura et al.24indicates that the first solvation shell of a CI- is much more pronounced in methanol than in water and that linear hydrogen bonds are formed between a CI- ion and its coordinated methanol molecules. The schematic solvation diagram for the interaction giving rise to the Raman band at -3324 cm-' is depicted in Figure 4 together with the one for the Raman band at 3385 cm-I. As already stated in a previous paper,I4 methanol molecules in second solvation spheres must also contribute to some extent to the Raman band at -3324 cm-' because their Raman bands should appear between the peak at 3385 cm-l and the one at 3175 cm-I. As seen in Figures 1 and 2, the intrinsic liquid methanol structure is gradually disrupted with an increase in LiCl concentration. In other words, the intensity of the low frequency region (3100-3300 cm-l) diminishes as salt concentration rises. This trend is the same as that observed in aqueous electrolyte solutions.I6 In vitreous ice, a sharp peak is observed at 3 110 cm-I and is due to intrinsic tetrahedral structure with strong hydrogen bonds.25 As frequently
reported,2J6an increase of salt concentration in an aqueous solution gives rise to an increase in the fraction of disrupted water structure and to the enhancement of the high wavenumber region in an O H stretching Raman spectrum. These parallel observations tell us an important feature of the spectral changes associated with vitrification. The peak frequency of the LiCb4MeOH solution is almost invariant from a liquid state at room temperature to a glassy state at liquid nitrogen temperature though there are large differences in spectral features. The highest peak in the LiCI. 4MeOH solution at ordinary temperatures is mainly ascribed to the methanol molecules directly hydrogen bonded to chloride ions so that the local liquid structure around halide ions should not change with vitrification. The most plausible cause for this contrast is the rapid exchange of methanol molecules directly coordinated to chloride ions with those outside of the first solvation spheres at room temperature. Although there seems to be no direct measurement of the exchange rate of methanol molecules between the first solvation sphere and the second one of a chloride ion, the smearing out of the sharp peak at 3385 cm-' in a Raman spectrum for a methanolic solution at room temperature indicates that the exchange rate of methanol molecules in the first coordination sphere with those in the outside sphere should be very rapid to make the methanol-halide complex too short-lived to vibrate as a distinct species. On the other hand, translational movements of methanol molecules (and chloride ions) are almost completely suppressed in a glassy LiX solution and the lifetime of the methanol->(- complex should be very long. In other words, the relaxation time for the exchange of a methanol molecule between an inner-sphere and an outside sphere of a chloride ion is much shorter in a liquid solution at room temperature than in a glassy solution, resulting in a broadening of the Raman peak at 3385 cm-' in the liquid solution. This explanation should be tested by Raman measurements with changing sample temperature up to room temperature. It is expected that the sharp peak at 3385 cm-I will gradually broaden with a rise in temperature to reach a broad Raman envelope at room temperature. Methanolic Metal Bromide and Iodide Solutions. Raman spectra of glassy methanolic LiBr (R = 20, 10, 6) and NaBr (R = 20) solutions are shown in Figure 5 together with those at room temperature. As seen in the bromide solutions, two resolved peaks are observed at 3350-3360 cm-l and at -3410 cm-I, respectively.
(23) Radnai. T.; Kalman, E.; Pollmer, K. 2. Naturforsch. 1985,39A, 464. (24) Tamura, Y . ;Spohr. E.; Heinzinger, K. Preprint,
(25) (a) Li, P. C.; Devlin, J. P. J . Chem. Phys. 1973, 59, 547. (b) Rice, S. A . Top. C u r r . Chem. 1975, 60, 109.
6598
The Journal of Physical Chemistry, Vol. 94, No. 17, I990
J
I
3800
3400
3400
3000
Roman s h i f t
3000
Icm-’
Figure 6. Raman spectra in the OH stretching region for the L i l a RMeOH ( R = 6-20) and the NalaRMeOH ( R = 10. 20) solutions in both liquid and glassy states.
Their frequencies are higher by about 30 cm-’ than those of corresponding peaks for glassy methanolic LiCl solutions, indicating that the hydrogen bonds between Br- ions and their coordinated methanol molecules are weaker than those between CIand coordinated methanol molecules. As stated in the preceding discussion, cationic effects on a Raman O H stretching spectrum for alcoholic electrolyte solutions are small. This is reasonable because the interactions between a cation and its coordinated methanol molecules are affected indirectly via oxygen lone pairs of OH groups. Comparison of the Raman spectra for the glassy methanolic solutions of LiBr and NaBr indicates that there are a few distinctive differences in spectral features. One of the differences is that the peak at 3410 cm-’ is obviously less intense for the NaBr solution than for the LiBr solution at the same halide concentration. This is not a cationic effect but rather due to difference of ionization between LiBr and NaBr in methanol at low temperatures. It is evident that ionization of dissolved NaBr is less than that of dissolved LiBr in glassy methanol. The existence of nonionized MX species (M = metal ion, X = halide ion) is crucial in explaining some Raman spectra of methanolic solutions, in
Yamauchi and Kanno particular, at low temperatures. Raman spectra for the glassy methanolic LiI ( R = 20, 10, and 6) and N a l ( R = 20 and 10) solutions are shown in Figure 6 together with those at room temperature. In contrast with the glassy LiCl and LiBr solutions, no sharp peak is observed for the glassy methanolic Lil solutions. One possible cause is that the two sharp peaks observed in the LiCl and LiBr solutions merge into one broad envelope due to the large ionic radius of iodide ions. In aqueous solutions, however, large ions such as perchlorate and PF,- ions tend to give rise to a split sharp peak in the high frequency Raman OH stretching region.26 The presence of a sharp peak in the high frequency region is believed to arise from OH oscillators weakly bonded to large structure-breaking anions. Knowing that the structure-breaking effects of anions are in the order F- < C1- < Br- < CI0,- and that a glassy methanolic solution of LiCIO, ( R = IO) gives a sharp peak at 3472 cm-l [this frequency is much higher than that (at 3410 cm-I) for a glassy methanolic LiBr solution],*’the above-mentioned explanation must be rejected. Another possibility is that ionization of dissolved LiI is incomplete in methanol at low temperatures. There are two possibilities for the chemical form of nonionized lithium iodide in a methanolic solution at low temperatures: neutral molecular form and contact ion pair. If lithium iodide is dissolved as neutral Lil molecules in a glassy methanolic solution, an Li-I stretching Raman band is expected to be observed in a Raman spectrum in the low frequency region (