Collective vibrational modes in strong electrolyte solutions at high concentrations

Collective vibrational modes in strong electrolyte solutions at high concentrations

Solid State Communications, Vol. 23, pp. 489—492, 1977. Pergamon Press. Printed in Great Britain COLLECTIVE VIBRATIONAL MODES IN STRONG ELECTROLYTE...

414KB Sizes 0 Downloads 16 Views

Solid State Communications, Vol. 23, pp. 489—492, 1977.

Pergamon Press.

Printed in Great Britain

COLLECTIVE VIBRATIONAL MODES IN STRONG ELECTROLYTE SOLUTIONS AT HIGH CONCENTRATIONS -

M.P. Fontana Istituto di Fisica and GNSM, Universita’ di Parma, Italy and Istituto de Fisica, Universita’ di Messina, Italy and G. Maisano, P. Migliardo and F. Wanderlingh Istituto di Fisica and GNSM, Universita’ di Messina, Italy (Recieved 21 March 1977 by R. Fieschi) We report evidence for the existence of solute related collective vibrational modes in- strong Il—I electrolyte solutions at high concentrations both in H20 and D2O. The evidence was obtained by studying the polarized Raman spectra of solutions of CdC12 and NiC12 Checks were also performed by taking the Raman spectra of the relative hydrated salts. Our results are explained in terms of vibrational densities of states corresponding to collective vibrational modes in a solute connected middle range lattice. IN THIS LETTER we present evidence for solute-connected collective vibrational excitations in aqueous solutions of strong Il—Iidentification electrolytes. as The evidence denmainly consists of the vibrational sity of states of diffuse spectral features in the low frequency (~ 500 cm’) Raman scattering from the solutions: This work constitutes a confirmation and a generalization of a preliminary report on NiC1 2 aqueous solutions [1]. We have taken polarized Raman spectra of solutions of CdC12, NiC12, CuC12, SrC12, ZnC12 and Cu(NO3)2, and of the respective hydrated salts. The spectra were taken as a function of concentration in the range from ca. 0.1 M to saturation. For brevity we shall report here only some of the results obtained for CdCl2 and NiCl2 the reasons for choosing these two materials will appear clear in the context of the discussion. A more complete report is in preparation. Aqueous solutions of NiCl2 and CdCl2 were prepared and characterized by standard techniques. For NiC12 solute concentration was also controlled spectrophotometrically, usingofthe known This absorption the blue-green region thewell spectrum. opticalin absorption also made necessary the correction of the Raman spectral intensities for absorption both of excitation and scattered light. Raman spectra were taken with a standard system. Excitation was usually with 5145 A or 4880 A light; power at sample site was always inferior to 100mW. Furthermore the beam was slightly 489

defocused in order to minimize noise and instabilities due to laser-induced turbulence. Spectral resolution 1: scattering was usually set90at degrees a bandpass 3 cm light propagating geometry was withofincident vertically and parallel to the entrance slits of the double monochromator. The solution was kept in an optical cell kinematically fastened to the sample holder so that it could be repositioned without altering alignment The alignement, effects of eventual fluctuations of incident power and the polarization response of the apparatus were checked periodically using the 453 cm~ line of CC14, for which the degree of depolarization was found to be p = 0.01. For each set of runs the spectrum of pure thrice distilled water was taken, for reference purposes, at the beginning and at the end of each set. In Fig. 1 we show the low frequency spectra of saturated solutions of CdC12, NiC12, CrC!2 and H2O as recorded. Note the presence of a strong broad peak in CdC12 and NiCl2 and its absence in SrC12. In the same figure we also report the spectrum obtained for a single crystal of CdC12 .2.5H2O grown from1.solution; Note This peak has again reported the prominent at 230incm been as the peak only peak the Raman spectrum of anhydrous CdC1 2 [2]. We have also taken Raman spectra of the solutions in D2O, in order to detect possible isotopic shifts of any of the spectral features. We have detected shifts only for the low frequency (w ~ 150 cm’) peaks in the CdC12.2.5D2O single crystal, also grown from solution. The other features,

490

VIBRATIONAL MODES IN STRONG ELECTROLYTE SOLUTIONS



-

Cd CL2 soLutions

-

a=Cd Cl2 2.5H20 crystaL b=Cd CL2 saturated c=NiCL2

-.

•‘

d=SrCL2 e=H20

a = 5.08 M b~= = 3.05 2.50

-

d = 2.03 e— 1

..

f

-

~:

‘~‘~‘ \ ‘~-,.

I

~

Vol. 23, No.7

zero

a

g

-

=

1.02

= ~

50

~-.‘•.— ‘.

‘~

b ~

~

r

-

e

-

50

150 -

250 350 1) RAMAN SHIFT(cm

450

Fig. I Experimental Raman spectra for the saturated solutions of(b) CdC12, (c) CiC12, (d) SrCI2 and for(e) pure water. Excitation power. 100mW, countmg time/ channel 0.6sec; scan speed 100cm /sec. (The top spectrum is from a single crystal of CdCI2, 0.2 5H2O (a). and particularly the peak at 230 cm’ for the hydrated CdC12 crystal and for the CdCl2 and NiCl2 solutions, and the 380 cm_i peak for the NiCl2 solution, were completely unaffected by the isotropic substitution. All the solution spectra have a diffuse character, with strong low frequency contribution. In such a situation it is important to correct the measured intensity for the factor [n(w) + 1]/w where n(w) is the Bose—Einstein population factor [3]. Furthermore, since we are mainly interested in the solute-related contributions to the scatteringintensity, some form of subtraction of the low frequency contributions due to water must be performed. The problem of accounting for the waterrelated contribution to the low frequency spectrum is not simple, since to date there is no clear experimental evidence of the effect of the solute, especially at high concentrations, on such spectrum. For lack of better information, we have performed the subtraction assuming that the spectrum of water in the solution is the same (savethe for suitably intensity) to that ofreference pure water. Thus we have used normalized water spectrum. In Fig. 2 we show then the difference spectra for CdC1 2 solutions, corrected for the [n(w) + 1] 1w factor, as a function of solute concentration. As a check of our

50 150 RAMAN 250 SHIFT(cm1) 350 450 Fig. 2. Corrected difference spectra for CdCl 2 solutions as a function of molar solute concentration. The integrated (50—SOOcm’) and peak (230cm~)intensities are linear in concentration for c> I M. normalization and difference taking procedures, we have also applied the [n(w) + 11/w correction to the raw experimental spectrum; the resulting spectral density did not differ from those reported in Fig. 2 to any relevant extent. Similar results have been obtained for NiCl2 solutions, although the 230 cm’ peak is not as well pronounced as for CdCl2. In NiC12 furthermore the low frequency contributions to the spectrum tend to increase relatively to the 230 cmi peak as the concentration is lowered, as already reported in reference [1]. The 230 cm_i peak is found to be strongly polarized (p = 0.11 for CdCl2, p = 0.2 for NiC12). The spectrum of the degree of depolarization is reported in Fig. 3 vs CdCl2 concentration. We note the continuous variation ofp across the spectrum, and the essential independence on concentration of p (230) down to about 1 M (see inset in Fig. 3). Below 1 M p tends rapidly to the value 3/4 (total depolarization when the incident light is totally polarized), also shown by the low frequency scattering of pure water. For NiC12, the behaviour of p is qualitatively similar, although the presence the 1 of distorts totally polarized additional peak at 380 cm the spectral dependence of p at the higher frequencies. The main point we shall try to prove in discussing the preceding data is that the relative spectral features represent a vibrational density of states, similar to that

Vol. 23, No.7

VIBRATIONAL MODES IN STRONG ELECTROLYTE SOLUTIONS .8

491

Cd CL2 soLutions

a=5~Mb~i~

2MOLAI~TY

050 e=H20 150

250 350 I 0 450 1 RAMAN SHIFT (cm~1)

~

550

Fig. 3. Concentration dependence of the depolarization spectrum for CdCl 2 solutions. The inseet shows the concentration dependence of p(230). Note the sharpness within experimental error of the upturn of p(230) for c ~ 1 M. found in the Raman spectra from amorphous materials [2]. Such density of states would stem from the existence of collective vibrational excitations in a medium range lattice of Cl—Cd-—Cl ions. The existence of such a medium range ordering in NiC12 saturated solutions has already been conjectured on the basis of neutron diffraction [41. Let us begin by briefly summarizing the ionic cornplex concentrations. structure that is As likely exist in the cations, solutionsthere at high for to most divalent is a strong tendency to hydration, with the hydration number varying from 2 to 10 and over; in most cases the number is close to 6 with an octahedral coordination (5). The anionic structure is more diverse and complicated. For the case of halide anions, and divalent cations such as Zn, Cd, Ni, there is a tendency at high concentrations to form complexes of the form Me (halide)~2, with a tetrahedral structure. Thus, for concentrations above, say, 0.1 M the solutions we have studied are very likely composed of Me (H 2 and Me(Q)~2,where Me = Ni, Cd and n = 6. On20) the basis of the spectra of the anhydrous CdCl 2 crystal, the absence of shift in the D20 solution, and the lowindegree of depolarization, 1peak the solutions to a totally we assign the 230 cm~ symmetric mode corresponding originally to stretching of the CdCl~2complex. Next, we consider the large width of the peak and its independence of concentration for c ~ 1 M. An isolated ionic complex especially if it does not involve directly water molecules would yield a reasonably narrow peak (say a 10—30 cm_i). Furthermore, upon increase of c towards saturation, the width should —



increase (and the spectral shape eventually change) as the ionic complexes begin to interact. This is well known from the literature for ions such as NO~,see for instance reference [5] Considering that the number of free water mo!ecules for concentrations near saturation, is not much greater than that of the ionic complexes, there will certainly be strong interaction thecm~1 ions.peak Thusand the its 1)oamong f the 230 large width (ca.on100 cm~ independence concentration in the range from 1 M to saturation indicates that it cannot be due to isolated ionic complexes. Rather, it must reflect the dispersion of the original vibrational frequency of the Cd—Cl bond, due to coupling of many of the complexes. It is thus not a peak in the ordinary sense of the word, but rather a convolution of peaks, i.e. a vibrational density of states modulated by some coupling function, such as is observed in the Raman spectra of amorphous materials. These assertions are confirmed by the behaviour of the depolarization ratio. First, we note that the low value of p(230), besides indicating that we are dealing with a -

totally symmetric vibration of a highly symmetric complex, implies also that such a vibration “internal”such to a the complex; in the language of crystal isvibrations, vibration would correspond to the optical phonon. In the low frequency region, we expect to fmd contributions from “acoustic” vibrations, i.e. vibrations external to the complexes; such vibrations, as the real acoustic vibrations in crystals, are much more sensitive to intercomplex interaction, i.e. they show more dispersion; their corresponding density of states would not be expected to show peaks. Furthermore, their intrinsic

492

VIBRATIONAL MODES IN STRONG ELECTROLYTE SOLUTIONS

Vol. 23, No.7

higher anisotropy would lead to a higher depolarization, as is actually found. Finally, the fact that the degree of depolarization varies slowly and smoothly in passing from the “acoustic” to the “optical” region is further indication that the observed spectrum is indeed connected to a vibrational density of states,

structure factor [8]. If this is done, one fmds a systematic difference between structure factors of divalent transition metal ions and, for instance, those of the alkaline earths (such as Sr). Such a difference, which is attributed to an additional interaction energy, can be calculated and yields the value for the case of NiCl2, of

In conclusion, our data confirm and extend previous static [4] and dynamic [1] evidence for middle range correlations and ordering in NiCl2 solutions near saturation. The observed effects (strong 230 cm_i peak for CdC12, down to no peak for SrC12) are in scale with a classification of Il—I electrolytes that can be obtained from the relative intensity of a solute-related pre-peak observed in X-ray diffraction from the solutions [6]. The conclusions reached here are also in agreement with recent viscosity and partial molar volume determinations [71- The existence of collective vibrational modes at high concentrations must also be reflected in the thermodynamics of the solutions. For instance, it is possible to analyze data concerning the chemical potential of the solutions in terms of the long wavelength limit of the

0.7 Kcal/mole; this value turns out to be in very good agreement with the frequency of the dominant feature in the CdCl2 and NiCl2 Raman spectra, i.e. the 230 cm~ peak. Thus the existence of collective excitations also finds a thermodynamicjustification. The existence of quasi-lattices in highly concentrated solutions has obvious and vast implications in most of their properties and particularly on theur transport properties, the interpretation of which may have to be revised. Furthermore, the fundamental question remains open as to what causes some divalent ions such as Cd, to show such pronounced structuring effects, whereas others, such as Sr, do not seem to have any noticeable effect in this respect.

REFERENCES 1.

FONTANA M2., Solid State Commun. 18, 765 (1976).

2.

LOCKWOOD D.J., Light Scattering Spectra ofSolids (Edited by WRIGHT G.B.), p. 75. Springer-Verlag, New York (1969).

3.

See for instance, SHUKER R. & GAMMON R.W.,Phys. Rev. Lert. 25,222 (1970).

4. 5.

ENDERBY J.E., HOWELLS W.S. & HOWE R.A., Chem. Phys. Lett. 21, 109 (1974). Most of the information on Il—I solutions we have used can be found in the monograph by GUGGENHEIM E.A. & STOKES R.H., Equilibrium Properties ofAqueous Solutions ofSingle Strong Electrolytes. Pergamon Press, New York (1969). Recent measurements of compressibility by ultrasonic method (see reference [7]) show that n may be a decreasing function of concentration in the range 0.1 M up to saturation.

6. 7,

DOROSH A.K. & SKRYSHEVSKH B.F.,Z. Strukt. Khimii 8,348 (1967). FONTANA M.P., MAISANO G., MIGLIARDO P. & WANDERLINGH F. (submitted for publication inPhys. Chem. Liq. CUBIOTFI G., MAISANO G., MIGLIARDO P. & WANDERLINGH F. (submitted for publication in J. Phys. C). The technique of deriving the structure factor in the long wavelength limit from thermodynamic parameters has been previously discussed in the context of binary alloys [see, e.g. BHATIA A.B. & THORNTON I.E., Phys. Rev. B2, 3004 (1970)].

8.