4540
J. Phys. Chem. 1988,92, 4540-4542
diffusion of the N a ions is not so sensitive to the use of different long-range electrostatic forces, because both results of two runs with different long-range electrostatic forces are very similar as shown in Figures 10 and 11. 4. Conclusions
Although the framework is assumed to be rigid and only the N a ions are free to move, our molecular dynamics study on the Na-A zeolite system is successfully carried out. From this study, the following conclusions may be drawn: (1) The diffusion mechanism of N a ions in zeolite A is an actual bulk ionic In order to obtain the more realistic diffusion coefficient of Na ion, the larger model and the longer computing time may be
needed. (2) The NaII ion is more loosely bound to the A-type zeolite framework than NaI and NaIIIions. Hence, in the range of moderately low temperature, NaII ion can wander within a plane of the oxygen 8-ring instead of the fixed vibrations. (3) At higher temperatures, NaII and NaIIlions diffuse simultaneously (concerted transposition process). (4) The dynamic property of N a ions in the zeolite A framework is not sensitive to the long-range electrostatic forces. Acknowledgment. This work was supported in part by the Korea Science and Engineering Foundation and the Korea Research Center for Theoretical Physics and Chemistry. Registry No. Na, ion, 17341-25-2.
Hlgh-Temperature High-pressure Raman Spectra from Liquid Water G. E. Walrafen,* M. S. Hokmabadi, W.-H. Yang,+ Chemistry Department, Howard University, Washington, D.C. 20059
and G . J. Piermarini Ceramics Division, National Bureau of Standards, Washington, D.C. 20234 (Received: November 13, 1987)
Raman spectra were obtained from liquid water at 33 kbar and at 170 and 200 "C with a diamond anvil cell. Pressures were determined from the observed ice VII-water melting temperatures. Comparison of the present data, with data from Lindner and Franck, indicates that the OH-stretching peak position, A? in reciprocal cefltimeters, decreases with increasing pressure at constant temperature. The OH-stretching peak position, at 33 kbar and 170 OC (Hugoniot conditions), was plotted with Raman data obtained from shock-compressed water. This plot shows a rise in the OH-stretching peak position from about 3410 cm-' at 1 atm, to a maximum of about 3440 cm-I at about 90 kbar, followed by a decrease to about 3400 cm-I at 257 kbar. This maximum is thought to result from two competing effects: (1) hydrogen-bond rupture due to rising temperature and (2) decreasing nearest-neighbor 0-0distance due to compression, with a concomitant increase of the OH bond length and decrease in the OH-stretching force constant. Further analysis of the data suggests that the double-tesingle, hydrogen-bond, potential-well transformation, which is known to occur for linear hydrogen bonds, also occurs for severely bent, non-hydrogen-bonded, 0-H-0 configurations.
Introduction Raman spectra from shock-compressed water were reported recently.' Pressures, temperatures, and densities, respectively, ranged from 75 kbar, 367 OC, and 1.56 to 257 kbar, 1437 OC, and 2.00 g . ~ m - ~However, . pressures below 75 kbar were not involved,' Hence, the present Raman work was initiated to obtain data at lower pressures, e.g., 33 kbar. Previous high-pressure Raman work on liquid water, reported by Franck and Lindner,* was carried out a t temperatures to 400 OC and pressures to about 5 kbar. Other high-pressure Raman work on liquid water, reported by Walrafen,3 involved pressures to about 10 kbar, but at a temperature of only 28 OC. In contrast, the present Raman work involves both high temperatures and high pressures. Moreover, diamond-anvil techniques were employed instead of the more difficult shock-compression Raman methods, which are appropriate for very high pressures and very high temperatures.
'
Experimental Methods The diamond anvil cell was of the type described previously4 and used for Raman and X-ray work on ice VI1 to 360 kbar.5 This cell was heated electrically (6). Temperatures were measured by using a Pt-13% Rh thermocouple in intimate thermal contact with the diamonds.' The Raman instrumentation was described in ref 8 and 9. *Author to whom correspondence should be addressed. Permanent address: Institute of Physics, Chinese Academy of Science, Beijing, China.
0022-3654/88/2092-4540$01.50/0
Great care was taken with regard to water purity and reduction of fluorescence. The diamond-anvil cell was carefully loaded with dust-free distilled water. Also, both diamond anvils were set in platinum mounts, instead of the usual epoxy cement. A thin film of epoxy tends to move over the outer diamond faces at high temperatures, and this effect can give rise to fluorescence. The use of platinum mounts eliminated this problem. Also, the use of ruby was generally avoided in our work. Ruby was used initially to measure pressure, but it gives rise to strong fluorescence and should not be used in Raman work, if at all possible. Moreover, ruby is very slightly soluble in water at high temperatures, whereas diamond is totally insoluble. Because ruby and epoxy were both absent, it is evident that only the diamond could give rise to fluorescence in the present measurements. However, even the two (1) Holmes, N. C.; Nellis, W. J.; Graham, W. B.; Walrafen, G. E. Phys. Rev. Lett. 1985, 55, 2433. (2) Franck, E. U.; Lindner, H. Ph.D. Dissertation of the latter, University of Karlsruhe, 1970 (unpublished). (3) Walrafen, G.E. J. Solution. Chem. 1973, 2, 159. (4) Barnett, J. D.; Block, S.;Piermarini, G.J. Rev.Sci. Instrum. 1973, 44,
1.
(5) Walrafen, G.E.; Abebe, M.; Mauer, F. A.; Block, S.;Piermarini, G. J.; Munro, R. G. J. Chem. Phys. 1982, 77, 2166. (6) Piermarini, G. J.; Muriro, R. G.;Block, S.In High Pressure in Science and Technology; MRS Symposium Proceedings 22; Homan, C., MacCrone, R. K., Whalley, E., Eds.; North-Holland: Amsterdam, 1984; p 25. (7) The thermocouple was inserted into a small hole near the translating diamond mount plate. See Figure 1 of ref 6. (8) Walrafen, G. E.; Hokmabadi, M. S.;Yang, W.-H. J . Chem. Phys. 1986.85. 6964. (9) Walrafen, G.E.; Fisher, M. R.; Hokmabadi, M. S.; Yang, W.-H. J. Chem. Phys. 1986, 85, 6970.
0 1988 American Chemical Society
The Journal of Physical Chemistry, Vol. 92, No. 15. 1988 4541
Raman Spectra from Liquid Water
uo nm .o.o nm um Sam ym Anm-S
3.m
lam n m um Fm
Figure 1. Raman spectrum of liquid water at the Hugoniot condition of
I
um
um
Imo
aua
3-
am0
I
AY,U"-'
Figure 1. Raman spectra from liquid water at 33 kbar and 200 OC (upper) and from ice VI1 at 28 kbar and rcam temperature (lower).
Base-line estimates are shown below the speara: water (crosshatched)
33 kbar and 170 "C (upper) and the s p c t " transferred to a horizontal base line (below). The base line (dashed) was estimated, and then it was subtracted from the average signal. This yielded the lower spectrum. This spectrum was obtained by use of photon counting and computer plotting. Note that it is plotted in the opposite direction, compared to Figure 1. Excitation 488.0 nm at 600 mW.
and ice VI1 (solid line). Dc detection and a stripchart recorder were used for the upper (water) spectrum. Excitation 488.0 nm at 600 mW.
A V = 3410.60 + 0.671026 P -0.004876
diamond anvils used were selected for low fluorescence from a total of twelve l/3-carat diamonds. This selection was accomplished by determining that the fluorescence intensity was less than the twephonon Raman intensity from diamond at about 25M) cm-I. The diamond-anvil cell was filled with water and pressurized at r m m temperature until ice VI1 was observed visually. Raman spectra were then obtained, and the ice VI1 pressure was determined from the position of the OH-stretching peak according to the relation reported in ref 5. The temperature of the cell was next slowly increased over a period of several hours until melting occurred. The onset of melting was detected immediately by viewing the back-reflected laser light on the wall about 3 m from the cell. (The noise level of the Raman spectra also increased sharply upon melting.) When melting occurred, motion of the back-reflected patterns from the diamond f a m started abruptly. The pressure of the liquid water was then determined from the melting temperature," that is, from the temperature a t which the very first motion of the back-reflected laser light pattern was observed. The pressure observed after melting was always somewhat higher than the initial pressure of the ice VI1 a t rmm temperature. In a typical example, the pressure of the ice VI1 was 28 kbar, but the melting temperature was 170 OC, corresponding to 33 kbar.1°
Experimental Results Raman spectra from liquid water at 33 kbar and 200 OC,'O and from ice VI1 at 28 kbar and r m m temperature,' are shown in Figure 1. The peak of the OH-stretching contour from liquid water occurs near 3460 f IO cm-l for theX(Z,X+Z)Ygeometry employed?> whereas the peak from ice VI1 occurs near 3284 cm-I. The Raman spectrum from liquid water shows considerable intensity above 3425 C I X - ' ? ~ compared to ice VII. This is expected from a liquid containing ruptured (bent and/or stretched) 0-H-O units?' compared to fully hydrogen-bonded ice Ia, whose I3H-O units are linear.s When the liquid OH-stretching peak position of Figure 1 is compared to the value of about 3505 cm-l observed by Lindner and Franck* a t 2.8 kbar and 200 "C, it is evident that the additional 30 khar of pressure, applied here, lowers the peak position by roughly 45 cm-I. Another Raman spectrum from liquid water a t 33 khar and 170 ' C is shown in Figure 2. The melting temperature of 170 OC was employed because it corresponds closely to the Hugoniot temperature a t 33 kbar." The upper spectrum of Figure 2 was averaged and transferred to a horizontal base line as shown below. (10) Bridgman, P.W. I.Chem. Phys. 1937.5.964. Pistorius. C . W. F. T: Rapoport. E.;Clark, I. B. J. Chem. Phys. 1968. 48, 5509. ( I 1) N. C. Holmes has estimated the Hugoniot temperature mrresprmding to 33 kbar to be about I50 & 20 "C, private communication. This agrees reasonably well with the Hugoniot temperature of 1170 'C obtained from interpolation of the data of ref 1.
50
100
150
P' + 0.000008 Pr
200
250
kbar
FTyre 3. Raman OH-stretching peak pasitions from liquid water to 257 kbar under Hugoniot conditions, obtained from Holmes, N. C.; Nellis, W. J.; Graham, W. G., see ref. I, plus the 33-kbar datum from Figure 2. It is emphasized that the cubic equation (dashed line) was dsolely to approximate the maximum pasition, which is -3440 cmP at -90 kbar. No further significance is to be inferred from the cubic equation. The temperature increases rapidly, and by about 14M) "C, in going from the left to the right in this figure. Temperaturesin 'C corresponding to the points shown, from left to right, are respectively 25, 170,367,567. 1187, 1207,
and 1437.
This pnxedure was uscd to obtain the most accurate value p i b l e of the peak position for use in a subsequent plot involving Hugoniot conditions. The OH-stretching peak position of the transferred spectrum is seen to occur near 3427 f 10 cm-I, which corresponds to a value about 60 cm-' lower than that obtained from our 170 "C interpolation of the Franck and Lindner data.2 The peak position from Figure 2 was combined with the OHstretching peak positions from shock-compressed water resulting from the work of ref 1. A plot of the OH-stretching peak position versus pressure was then constructed and is shown in Figure 3. The data of this figure, from 75 to 257 kbar, refer to densities and temperatures, respectively, of 1.56 p c m F and 367 "C, and and 1437 OC.' Note that all of the data of Figure of 2.00 3, which include the Figure 2 result, correspond to the Hugoniot conditions for liquid water. The noise levels of the high-pressure high-temperature Raman spectra, and the large spectral bandwidth, lead to considerable uncertainty in the OH-stretching peak position. The error bars of Figure 3 reflect this uncertainty, which is estimated to be a t least f10 cmP. A third-degree least-squares polynomial fit was applied to the data, as indicated by the dashed curve and the equation shown. Thisfit was used solely as an aid in estimating the maximum frequency of Figure 3, and we do not attribute mechanistic significance to this equation at the present time. The increase in the OH-stretching peak position from about 3410 cm-l a t I atm to about 3440 cm-l near 90 kbar occurs because temperature rise increases the peak positi0n.8~Here, the corresponding pressure increase is not so large as to negate the
4542
The Journal of Physical Chemistry, Vol. 92, No. 15, 1988
effect of temperature. However, the effect of pressure on the OH-stretching peak position dominates over the temperature effect between =90 and 257 kbar. The maximum of Figure 3 occurs as the result of this pressure dominance. Raising the pressure lowers the OH-stretching peak position. This occurs because an increase of pressure actually increases the 0-H distance. An increase in the 0 - H distance decreases the OH-stretching force ~ o n s t a n t . ~The 0-H distance increases because an increase of pressure increases the density, which decreases the 0-0 d i ~ t a n c e .The ~ 0-H distance change and the 0-0distance change are of opposite sign. As pressure rises, the proton moves outward in an attempt to form a completely symmetric, linear hydrogen bond, 0-H-O, whose 0-H distances are equal and half of the 0-0 d i ~ t a n c e . ~
Walrafen et al. and the contributions of strongly bent 0-H-0 configurations to dominate. The OH-stretching peak position occurs at 3400 cm-l at 257 kbar and 1437 OC. This low OH-stretching value indicates that a considerable increase in the OH bond distance has occurred. The increase in OH-bond distance is inferred from Badger's ruleIg and, more directly, from Raman and X-ray measurements of ice VIL5 From the density of 2.00 g.cme3 it is certain that some decrease in the nearest-neighbor 0-0 distance has occurred. Hence, it seems necessary to conclude that the double-well potential characteristic of an asymmetric, linear 0-He-0 unit is changing in the direction of a single-well, even when the 0-H-O angle is highly nonlinear. This means that the double-well to single-well transformation, which is well-established for linear hydrogen bonds when the nearest-neighbor 0-0 distance approaches 2.4 A, is thought to occur also, on the basis of the present data, for severely bent &H-0 units. The change in the shape of the potential may be independent, qualitatively, of the 0-H-O angle. Of course, it is probable that a pressure of 257 kbar (density of 2.00 g ~ c m - ~is) insufficient to produce completely symmetric hydrogen bonds, because a pressure approaching 500 kbar is required to produce symmetric ice at room t e m p e r a t ~ r e . ' ~ Nevertheless, it now seems certain that the effect of high pressure on all Raman components of the OH and OD stretching contours, both hydrogen bonded (HB) and non-hydrogen bonded (NHB), leads to a general lowering of f r e q u e n ~ y ~ *and ~ , ~thus + ~ ~to a general lengthening of both of the H B and NHB, OH and OD, distances.3*sMoreover, computer analysis of the high-temperature high-pressure Raman contours was accomplished previously.' This analysis indicates that the OH-stretching peak position at 257 kbar and 1437 OC arises almost exclusively from the NHB component,' and thus a downward N H B component shift of roughly 220 cm-' has occurred. This shift amounts to about 0.9 cm-'/kbar, which compares favorably with values reported in ref 3 and 5.
Discussion It is now generally accepted that the 0-H distance of an ordinary, asymmetric, linear hydrogen bond, 0-H-0, increases as the 00distance of that bond decreases.12J3 Moreover, the 0-H distance continues to increase until a symmetric (linear) hydrogen bond, 0-H-0, forms at about 440 kbar and room temperature.14 The 0-0 distance of such a symmetric hydrogen bond is about 2.4 A.13 This means that the 0-H bond distance has increased from roughly 1 to 1.2 A, an increase of roughly 20%. Clear evidence has been presented recently which indicates that the linear, partially covalent hydrogen bonds of liquid water are nearly all ruptured above a pressure of 257 kbar and a temperature ~ those of 1437 OC.' However, the density is 2.00 g ~ c m -under conditions, and thus other strong, cohesive interactions must be present between the H 2 0 molecules, e.g., electrostatic (multipole), induction, and dispersion forces. The hydrogen bond has long been thought to involve partial covalency.Is An alternative description of the situation is that a small amount of electronic charge is transferred between 0(2) and the proton of the 0(1)-H--0(2)linkage.I6 Intermolecular Raman scattering from the 0-0stretching motion of the hydrogen Summary bond would not be very likely to occur with sizeable intensity, if partial covalency or charge transfer were totally a b ~ e n t . ~ ? ~ Raman data obtained from liquid water at 33 kbar and 170 Collision-induced Raman scattering is possible, but this form of OC were added to previously unpublished shcck-compression data scattering yields featureless spectra;" Le., it does not produce to obtain a plot of the OH-stretching peak position versus pressure, intermolecular peaks such as those observed near 170 ~ m - ' . ~ , ~ all under Hugoniot conditions. This plot indicates a maximum Quantum mechanical calculations indicate that the amount of value of about 3440 cm-' at about 90 kbar and a final value charge transfer in the 0-H-0 linkage decreases as the 0-H-O (observed at 257 kbar, 1437 OC, and a density of 2.00 g ~ c m - ~ ) angle decreases below 18Oo.l8 Thus one would expect the phoof 3400 cm-I. The 0-H-0 angle is thought to be severely bent nonlike peaks in the intermolecular Raman spectrum, e.g., the ~ , a decrease of the nearest-neighbor at a density of 2.00 g ~ m - and 170 cm-' peak due to 0-0 stretching of linear, and nearly linear, 0-0distance is also expected at this density. An increased OH partially covalent O-H-.O units, to decline in intensity with distance is associated with the 3400-cm-' OH-stretching peak increasing temperature. This intensity decline is observed, and value. Hence, an increased OH distance is linked to the decreased it measures the rupture of hydrogen bonds.*q9 A corollary of this 0-0distance of this bent 0-H-0 configuration. The situation observation is that the contribution of strongly bent 0-H-0 is entirely analogous to that found for linear hydrogen bonds, configurations must rise as the temperature rises. Further, under except that the partial covalency is diminished, and other forms the extreme conditions of Figures 2 and 3, one would expect the of interaction become dominant. contributions from linear 0-H-0 configurations to be very small Acknowledgment. This work was supported by contracts from the Office of Naval Research. N. C. Holmes, W. J. Nellis, and (12) Speakman, J. C. MTP International Review of Science, Physical W. B. Graham are thanked for providing the data for Figure 3, Chemistry Series One;Roberston, J. M., Ed.;Butterworths: University Park, and in particular the help provided by N. C. Holmes is very greatly Baltimore, 1972; Vol. 11, pp 1-31. appreciated. (13) Stillinger, F. H.; Schweizer, K. S. J . Phys. Chem. 1983, 87, 4281. Schweizer, K.S.; Stillinger, F. H. J . Chem. Phys. 1984, 80, 1230. Registry No. H,O,7732-18-5. (14) Polian, A.; Grimsditch, M. Phys. Rev. Lett. 1984, 52, 1312. (15) Frank, H. S. Proc. R . SOC.London, A 1958, 247, 481. (16) Whalley, E.; Klug, D. D.J . Chem. Phys. 1986, 84, 78. (17) Walrafen, G. E.; Hokmadbadi, M. S.;Yang, W.-H., manuscript in preparation. (18) Rao, C. N. R. In Water, A Comprehensive Treatise; Franks, F., Ed.; Plenum: New York, 1972; Vol. 1 , Chapter 3.
(19) Badger, R. M. J . Chem. Phys. 1934, 2, 128; J . Chem. Phys. 1935, 3. 710
(20) Walrafen, G. E. J. Chem. Phys. 1971,55, 5137. (21) Walrafen, G. E.; Abebe, M. J . Chem. Phys. 1978, 68, 4694. (22) Abebe, M.; Walrafen, G. E. J . Chem. Phys. 1979, 71, 4167.