MOLECULAR PHYSICS, 1999, VOL. 96, NO. 11, 1659± 1665 The e ect of pressure on hydrogen bonding in water: IR study of m OD HDO at pressures of up to 1500 bar YU. E. GORBATY*, G. V. BONDARENKO, A. G. KALINICHEV and A. V. OKHULKOV Institute of Experimental Mineralogy, Russian Academy of Sciences, Chernogolovka Moscow Region, 142432 Russia (Received 5 August 1998; revised version accepted 16 November 1998) A seeming contradiction has been revealed between the behaviour of the shortest intermolecular separation rOO and that of the frequency of the OH vibration of H2 O under compression in liquid water. Decrease of rOO with increasing pressure up to 2 kbar seems to be evidence of strengthening of the hydrogen bonding, but the increase in the frequency of t OH H2 O seems contradictory. To clear up this question, measurements of the integrated intensity of IR absorption for t OD HDO have been made which de® nitely show that the energy of the hydrogen bonds decreases under compression at least up to 1.5 kbar. The decrease in the intensity is of about 20% , i.e., essentially more than the estimated accuracy of the measurements. Monte Carlo simulations of the compression of liquid water show that the behaviour of rOO, t OH, and the integrated intensity can be explained by the bending of hydrogen bonds. 1. Introduction Studies of the e ect of pressure on the hydrogen bonding in liquid water can give important information on the properties of this so far enigmatic liquid, and on the very nature of the phenomenon of hydrogen bonding. Since the pioneering experiments of the Neilson group [1, 2] and the outstanding work of Wu et al. [3] there have not been many experimental studies of the in¯ uence of pressure on hydrogen bonding in water [4± 9]. What is worse, often the data on the pressure e ect are contradictory: perhaps not surprising, taking into account the great di culty of conducting high pressure experiments and the small values of the observed changes. Thus far it is even hard to predict a general trend in the pressure e ect on hydrogen bonding. It has been shown, for example, that pressure a ects the pair correlation functions of water in the same direction as temperature, decreasing, in particular, the degree of tetrahedral ordering [4]. So, we could expect the weakening of hydrogen bonds. However, the shortest separation rOO between hydrogen bonded water molecules has been found to decrease with pressure [4, 5] up to about 2000 bar, and this fact seems to be evidence of strengthening of hydrogen bonds in view of the well known correlation between the shortest separation and the energy of hydrogen bond [10± 13]. Regret- * Author for correspondence. e-mail: [email protected] u tably, even the very fact of rOO decreasing with pressure remains disputable. In this paper we would like to draw attention to the ostensible contradiction between X-ray scattering data [4, 5] and results of high pressure Raman experiments [8, 9]. Using infrared absorption data and Monte Carlo simulations, we show that there is no contradiction and the observed phenomena can be explained in the framework of the earlier suggested model [5]. 2. Experimental The purpose of the experiment was to obtain the pressure dependence of the integrated intensity for the OD stretching band t OD of HDO. The spectra of t OD was obtained using the technique developed by us [14, 15]. The method is based on the use of high pressure high temperature cells with changeable path lengths. It allows us to get rid of the most common measuring errors arising, for example, from distortions of optimal geometry, light losses due to absorption and re¯ ection by sapphire windows, overlapping with the spectra of water vapour and CO2 from atmosphere, the `lens’ e ect, etc. A new custom made cell (to be described in detail elsewhere) has been used in the experiment. One of the sapphire windows can be moved with a driving mechanism attached to the cell. It consists of a lead screw and a system of levers. The driving system allows one to change the path length smoothly between 0 and 0.8 mm so that one full revolution of the screw Molecular Physics ISSN 0026± 8976 print/ISSN 1362± 3028 online Ñ 1999 Taylor & Francis Ltd http://www.tandf.co.uk/JNLS/mph.htm http://www.taylorandfrancis.com/JNLS/mph.htm 1660 Yu. E. Gorbaty et al. head corresponds approximately to a window shift of 40m m. To obtain a spectrum of a pure substance free from the aforementioned errors it is su cient to record two spectra at di erent path lengths in two successive runs. It is not necessary to know the values of path lengths. However, the di erence between them, that is, the shift of movable window ¢ , must be measured with the highest possible accuracy. For pure water at a density q kH2 O = M log [T 2 ( t ) / T 1 ( t q ¢1 )], ( 1) where kH2 O is the absorption coe cient, M is the molecular weight of H2 O, T 1 ( t ) is the transmission spectrum obtained with the larger path length, while T 2 ( t ) is obtained with a shorter path length. Deriving this simple equation, it is easy to see that corrections for all the errors mentioned above are cancelled out. To obtain the absorption coe cient kHDO ( t ) from the spectra of the HDO solution in H2 O we again have to measure two transmission spectra of the solution: T 3 ( t ) (with a longer path length) and T 4 ( t ) (with a shorter path length) with ¢2 the di erence in path lengths. Then log [T 4 ( t ) / T 3 ( t ¢2 )] = kH2O( t ) cH2O + kHDO( t ) cHDO, ( 2) where cH2 O and cHDO are the concentrations in mol cm of H2 O and HDO in the solution at density q . Solving equations (1) and (2) simultaneously, we have: 3 kHDO( t ) log [T 4 ( t ) / T 3 ( t = cHDO ¢2 )] McH2 O log [T 2 (t ) / T 1 ( t q cHDO ¢ 1 )]. ( 3) (In this case, the spectra T 1 ( t ) and T 2 ( t ) should be measured in the same spectral range as the spectra T 3 ( t ) and T 4 ( t )) . To measure ¢1 and ¢ 2 a measuring microscope with spiral scale was used that provided an accuracy of 0.5 m m. It is possible, however, to avoid measuring ¢ 2 using the condition M log [T 2 ( t ) / T 1 ( t q ¢1 )] = log [T 4 ( t ) / T 3 ( t cH2 O¢2 )] ( 4) that is valid for a spectral range over which no absorption from HDO occurs. Taking, x¢ 2 = ¢ 1 , we can calculate x as x= McH2 O log [T 2 ( t ) / T 1 ( t q log [T 4 ( t ) / T 3 ( t )] )]. ( 5) Because of the noise in the experimental spectra, it is better to calculate an average value of x using a su cient number of points i Figure 1. Illustration of the method for obtaining a spectrum of t OD HDO from four spectra: ( a) spectra of pure water; ( b) spectra of HDO in H2 O solution; and ( c) resulting spectrum of t OD HDO at a pressure of 1200 bar. x = ( i2 i1 ) 1 i2 i1 McH2 O log [T 2 ( t i ) / T 1 ( t i )] . q log [T 4 ( t i ) / T 3 ( t i )] ( 6) In the experiment described i1 corresponds to 1 1 1 2750 cm , i2 to 2850 cm , and i2 i1 = 100 cm . Then equation (3) may be rewritten as kHDO( t )= log [T 4 ( t ) / T 3 ( t cHDO x¢ 1 )] McH2 O log [T 2 ( t ) / T 1 ( t q cHDO ¢ 1 )]. ( 7) Figure 1 demonstrates an example of applying the algorithm. Note the increasing noise at low wavenumbers due to the very low transmittance of sapphire 1 in this spectral range. The same occurs near 2900 cm , where the absorption of H2 O increases strongly. Note also an overcompensation of the CO2 band near 2300 cm 1 . It is di cult to allow accurate compensation for this band since the content of CO2 in the atmosphere tends to ¯ uctuate. The spectra of t OD HDO were obtained in the spectral 1 1 range 1900± 2900 cm with a step of 1 cm using a Perkin-Elmer 983 ratio recording spectrophotometer. Regrettably, we can use only a cell designed to work at high temperatures (up to 500 ë C) in the pressure E ect of pressure on hydrogen bonding in HDO range restricted by 1500 bar. Therefore, the measurements at room temperature have been made in this pressure range with a step of 100 bar and with an accuracy of 5 bar. The integrated intensity of absorption A in the region of the OD vibration mode of HDO was calculated according to A= t 2 =2800 t 1 =2100 kHDO ( t ) dt . ( 8) The solution of 4 mol% HDO in H2 O was prepared using 99.8 wt% D2 O delivered by the `Isotope V/O’ company (Russia) . 3. Results and discussion 3.1. Background Studying the e ect of pressure on the structure of liquid water with the energy-dispersive X-ray di raction technique (EDXD), Okhulkov et al. [5] have found that the average separation between nearest molecules rOO exhibits peculiar behaviour. It might be expected intuitively that rOO should decrease as a result of compression. Indeed, it decreases with the pressure rise up to 2 kbar, as ® gure 2 shows, but at higher pressures it begins to grow, virtually reaching the initial value at 4± 5 kbar. Some authors [6], however, reckoned this result as debatable. Indeed, in some previous experiments [1± 3] a very small contraction of rOO was found. Also computer simulation of the e ect of pressure did not predict an essential decrease in rOO with increasing pressure [16, 17]. We should note, however, that this rather subtle e ect would be di cult to reveal explicitly in neutron di raction or usual angular scanning X-ray di raction experiments. The energy-dispersive X-ray di raction technique used by Okhulkov et al. has certain advantages over the classical X-ray di raction method because of the high statistical accuracy achieved with shorter exposure times and due to the absence of any contribution from the container to the scattering intensity. This presents the possibility of using the long range part of the structure function si( s) (where s is the scattering parameter or wavefactor and i( s) is the structure factor) to obtain the value of rOO, using a large number of experimental points. The procedure is based on the fact that the remote part of the structure function for certain values of s is virtually the Fourier transform of the ® rst peak of the radial distribution function 4p rq 0 g( r) , where q 0 is the bulk density at given pressure and g( r) is the molecular pair correlation function. The procedure gives an additional advantage over the usual way to obtain rOO, from the position of the ® rst peak of the radial distribution function. The ® rst peak of g( r) is overlapped 1661 strongly with the more broad peak at 3.2± 3.3 A [4] and this results in a shift of the ® rst peak to larger r. However, if the structure factor i( s) or structure function si( s) is considered, these two distributions do not overlap in the remote part of inverse space for s higher 1 than 8± 9 A . It is well known that the position of the ® rst peak depends on the method of calculation of g( r) . In particular, modi® cation functions used frequently to force the structure functions to approach zero within the experimental range of s inevitably lead to decreasing of g( r) resolution and, consequently, to the additional shift of rOO. Therefore, there is not much sense in comparing absolute values of rOO obtained in a variety of experimental studies. Usually, they di er within 0. 05 A, i.e., larger than the e ect under discussion. More important is the relative change of rOO observed with variation of external conditions. Incidentally, Bellissent-Funel and Bosio [7] seem to con® rm qualitatively the behaviour of rOO observed by Okhulkov et al. [5]. We ought to mention that the values of r obtained in [5] are not exactly the separations between the nearest oxygen atoms. They correspond to the separations between the centres of electron distributions in the nearest molecules. However, as may be seen from the behaviour of the form factors derived by Narten and Levy [18], the electron distribution of the water molecule is nearly spherical. Okhulkov et al. have explained the speci® c behaviour of rOO in terms of `preferential’ structure ¯ uctuations taking the form of deformed local structures of the ice high-pressure polymorphs (similar ideas have been discussed by other authors [19, 20]). This suggestion originated from the fact of increasing rOO in the series ice I ! ice VIII [13, 21]. It was shown [5] with the analysis of the pair correlation functions that at low pressures the preferential structure ¯ uctuations of the type of ice I and ice III prevail while at higher pressure the short lived molecular con® gurations typical of distorted structures of higher pressure ice polymorphs dominate. This presents a reasonable explanation of the minimum in the pressure dependence of rOO presented in ® gure 2. The experimentally observed behaviour rOO leads one to assume that the strength of hydrogen bonds in compressed water increases ® rst and then after 2± 3 kbar decreases. Such a notion is based on the correlation between the energy of hydrogen bonding and the OÐ O separation [11± 13]. On the other hand, this correlation is not well de® ned. If a large number of hydrogen bonded systems is considered, the scatter of points in the correlation is very large, so that the correlation between rOO and the energy of bonding can only be guessed rather than seen clearly. This fact witnesses that besides rOO a large number of factors in¯ uence the energy of 1662 Yu. E. Gorbaty et al. Figure 2. Pressure dependence of the shortest intermolecular separation rOO between hydrogen bonded water molecules. The line is a guide to the eye. hydrogen bonds. In this sense there is no direct relationship between rOO and the energy of hydrogen bonds. Nevertheless, it seems intuitively that the average energy of bonding should increase as rOO decreases. The frequency of OH stretching vibrations is also a strong function of the hydrogen bond strength. The higher the energy of hydrogen bonding, the lower is the frequency t OH . So, one ought to expect the behaviour of t OH to be qualitatively the same as that of rOO. However, as seen in ® gure 3, the pressure trend of t OH is opposite to the expected one. Here the Raman data on the maxima of the band position [8] and centres of gravity [9] for the OH stretching band of water are shown. It should be noted, though, that because of the large scatter of experimental points, Walrafen and Abebe [8] approximated their data with a linear ® t. Still, a combination of the data from [8] and [9] presented in ® gure 3 plus the behaviour of the halfwidth of the OH band contour found in the work of Cavaille et al. [9] de® nitely indicate a singularity in the pressure range 2± 3 kbar. This circumstance seems to con® rm the singularity found for rOO. However, taking t OH and halfwidth as criteria for the strength of the hydrogen bond in compressed water, one can only state that up to 2± 3 kbar hydrogen bonds weaken while at higher pressures they strengthen. So, the question arises: do we have increasing or decreasing the strength of hydrogen bonding in water under compression? Figure 3. Pressure dependence of the frequency of H2 O stretching vibrations: circles, position of the band maximum [8]; and squares, centre of gravity of the band [9]. 3.2. Spectroscopic results To answer the above question, we have studied the behaviour of the integrated intensity of absorption by OD vibrations of HDO. This characteristic is very sensitive to the energy of hydrogen bonds. The integrated intensity of t OH decreases approximately by a factor of 25± 30 at the transition of water from the solid state to the gaseous [10], while the frequency shift is of about 10% only. Because of the aforementioned experimental limitations, we could check only the left descending branch of the pressure dependence of rOO, shown in ® gure 2. However, even the limited pressure range 1± 1500 bar is enough to reveal the main trend of the pressure e ect on hydrogen bonding at the beginning of compression. Figure 4 shows the pressure dependence of the integrated intensity A for t OD HDO. It demonstrates convincingly that A, and consequently the energy of hydrogen bonds, decrease with the pressure rise. The decrease in A over the pressure range explored is 20% . This value is much higher than the estimated error of the intensity measurements ( 5- 6% ) but it is, in fact, very small, bearing in mind the full range of the possible variation of the integrated intensity. The data for low pressures are in a good agreement with those reported previously [22± 25]. In general, measurement of intensity in any kind of spectroscopy is always a much more di cult task than analysis of band positions, halfwidths, or band shape. However, E ect of pressure on hydrogen bonding in HDO Figure 4. t 1663 Pressure dependence of the integrated intensity of HDO. The lines are guides to the eye. OD when studying hydrogen bonded systems, the integrated intensity provides the most explicit information about the degree or probability of hydrogen bonding in the system [10]. We were not able, for example, to use for this purpose the position of the maximum of the OD band because the scatter of the data was so great in so narrow a pressure range. Thus, we have arrived at the conclusion that at least up to 1.5± 2 kbar hydrogen bonding weakens with compression of liquid water and shortening rOO. It is for future experiments to reveal what is going on at pressures higher than 2 kbar. Relying upon the data shown in ® gure 3, we dare to predict that the integrated intensity and, consequently, the strength of hydrogen bonds should increase. 3.3. Monte Carlo simulations To explain the seeming contradiction between the behaviour of rOO and the strength of hydrogen bonds with compression of liquid water, isothermal± isobaric Monte Carlo simulations have been performed, using the TIP4P e ective site± site potential. There have been, of course, other attempts to study the e ect of pressure, using MD and MC calculations [16, 17, 27± 30], but the present study was aimed at a comparison of the simulated and experimental pair correlation functions [5] during the step by step compression of water. The simulations were performed in the pressure range 0.001± 10 kbar for the NPT ensemble of 216 molecules in a cubic cell with periodic boundary conditions. A detailed discussion of the results has been given elsewhere [31]. Here we con® ne ourselves to a few Figure 5. Results of Monte Carlo simulations: ( a) rOO as a function of pressure; ( b) energy of hydrogen bond versus pressure; and ( c) the pressure dependence of the average angle of hydrogen bonds. particular results most relevant to the problem. We could not, of course, anticipate good qualitative agreement of the simulated data with experiment. The TIP4P potential is too sti at short molecular distances and gives essentially even shorter values of rOO at ambient pressure. This is a well known drawback of many intermolecular water± water potentials used in computer simulations, and that may be a reason for the low rate of contraction usually found in computer experiments. It should be also noted that the phase diagram of TIP4P water does not correspond quantitatively to the real diagram of water. Furthermore, it can hardly be expected that computer simulations are able to reproduce so a complicated a phenomenon as the preferential variations of the nearest environment (structure ¯ uctuations [5]) involving a large number of water molecules. Still, e ectively computer simulations can help in understanding and interpreting experimental data. Figure 5( a) shows the pressure dependence of the shortest separation between the oxygen atoms of the neighbouring hydrogen bonded water molecules. For the sake of simplicity, only geometrical criteria of hydrogen bonding [32] were used. Despite the experimentally observed minimum, rOO is not reproduced, and one can see two distinct pressure ranges with the 1664 Yu. E. Gorbaty et al. di erent contraction rates. This may be considered as a weak re¯ ection of the peculiarity observed in ® gure 2. However, an even more important ® nding is that the average potential energy of hydrogen bonding UHB decreases with the pressure rise, as may be seen in ® gure 5( b). Undoubtedly, the reason for this e ect is the deviation of hydrogen bonds from their favourable near-linear arrangement at compression of liquid water. Figure 5( c) shows a noticeable decrease in the OHO angle H with increasing pressure. Thus, we may consider the bending of hydrogen bonds as the same reason for both the shortening of rOO and weakening of the hydrogen bonds. This means that there is no contradiction between the behaviour of rOO and the value of the red shift of the OH(OD) stretching vibrations in liquid water. It is also clear that the correlation between rOO and the energy of the hydrogen bonding is not universal and must be used cautiously. 4. Conclusion In conclusion, we would like to summarize the main inferences drawn from the present study. A seeming contradiction has been revealed between the behaviour of the shortest intermolecular separation rOO and that of the frequency of OH vibration of H2 O with compression of liquid water. Decrease of rOO with the pressure rise up to 2 kbar may be considered as evidence of strengthening of hydrogen bonding, but the increase in the frequency of t OH H2 O seems contradictory. To clarify this, measurements of the integrated intensity of IR absorption for t OD HDO have been made which de® nitely show that the energy of hydrogen bonds decreases with compression at least up to 1.5 kbar. The decrease in the intensity is about 20% , i.e., essentially more than the estimated accuracy of measurements. On the other hand, the e ect, though quite explicit, is rather small compared with the full range of possible change in the intensity of absorption. Therefore, it is not surprising, that the variations of rOO and t OH are also small, just slightly exceeding the limits of experimental error. Monte Carlo simulations of the compression of liquid water show that the behaviour of rOO, t OH , and the integrated intensity can be explained by the bending of hydrogen bonds taking place during the initial stage of compression. At the same time, MC simulations do not reproduce the experimentally observed behaviour of the shortest intermolecular separation above 3± 4 kbar. A possible reason for this fact is the excessive rigidity of TIP4P intermolecular potential. The degree of hydrogen bond bending de® nes the structures of high pressure solid polymorphs of water and the P- T regions of their stability [33]. In terms of preferential structure ¯ uctuations, we may say that in the range of pressures studied the structure ¯ uctuations of ice I type dominate. The distortion of the geometry of hydrogen bonds with further increase in pressure should lead to the appearance of the short-lived structure variations typical of the distorted structures of `higher’ types of ice with hydrogen bonds bent even more. It would be very interesting to obtain the pressure dependence of the integrated intensity of HDO or H2 O (D2 O) stretching bands at pressures higher than those achieved in this experiment. It may be predicted that there should be a change in the behaviour of the integrated intensity above 2± 3 kbar. The authors are grateful to Professor G. E. Walrafen for kind permission to cite Raman data. This material is based upon work supported by the US Civilian Research and Development Foundation under Award No. RC1-170. Grants from INTAS 96-1989 and from RFBR (97-05-65956 and 97-03-32587) are greatly appreciated. References [1] Neilson, G. W., Page, D. I., and Howells, W. S., 1979, J. Phys. D, 12, 901. [2] Gaballa , G. A., and Neilson, G. W., 1983, Molec. Phys., 50, 97. [3] Wu , A. Y., Whalley, E., and Dolling , G., 1982, Molec. Phys., 47, 603. [4] Gorbaty, Y u . E., and Demianets, Y u . N., 1985, Molec. Phys., 55, 571. [5] Okhulkov , A. V., Demianets, Yu . N., and Gorbaty, Y u . E., 1994, J. chem. Phys., 100, 1578. [6] Ohtaki, H., R adnai, T., and Y amaguchi, T., 1997, Chemical Society Reviews, 41. [7] Bellissent-Funel , M. C., and Bosio, L., 1995, J. chem. Phys., 102, 3727. [8] Walrafen, G. E., and A bebe, M., 1978, J. phys. Chem., 68, 4894. [9] Cavaille, D., Combes, D., and Zwick , A., 1996, J. Raman Spectrosc., 27, 853. [10] Gorbaty, Y u . E., and Kalinichev, A. G., 1995, J. phys. Chem., 99, 5336. [11] Pimentel, G. C., and Mc Clellan, A. L., 1960, The Hydrogen Bond (San Francisco: Freeman). [12] Sokolov , N. D., 1972, Zhurnal Vsesoyuznogo Khimicheskoko Obschestva, 17, 299 (in Russian) . [13] Eisenberg , D., and Kautzman, W., 1969, The Structure and Properties of Water (Oxford University Press). [14] Babashov, I. V., Bondarenko, G. V., Gorbaty, Y u . E, and Epelbaum, M. B., 1970, Pribory I tekhnika eksperimenta, No. 2, 215 (in Russian). [15] Gorbaty, Y u . E., and Bondarenko, G. V., 1993, Rev. sci. Instrum., 64, 2346. [16] Stillinger, F. H., and R ahman, A., 1974, J. chem. Phys., 61, 4973. E ect of pressure on hydrogen bonding in HDO [17] Impey, R. W., Klein, M. I., and Mc Donald , I. R., 1981, J. chem. Phys., 74, 647. [18] Narten, A. H., and Levy, H. A., 1971, J. chem. Phys., 55, 2263. [19] Vedamuthu, M., Singh, S., and R obinson, G. W., 1995, J. phys. Chem., 99, 9263. [20] R obinson, G. V., and Zhu , S.-B., 1994, Reaction Dynamics in Clusters and Condensed Phases, edited by J. Jortner et al. (Dordrecht: Kluwer) p. 423. [21] Minceva -Sukarova , B., Sherman, W. F., and Wilkinson, G. R., 1984, J. Phys. C, 17, 5833. [22] Swenson, C. A., 1965, Spectrochim. Acta, 21, 986. [23] Wyss, H. R., and Falk , M., 1970, Can. J. Chem., 48, 607. [24] Frank , E. U., and R oth, K., 1967, Discuss. Faraday Soc., 43, 108. [25] Iogansen, A. V., and R oz enberg, M. Sh., 1978, Opt. Spectrosc., 44, 49 (in Russian). 1665 [26] Janscoí , G., Bopp, P., and Heiz inger , K., 1984, Chem. Phys., 85, 377. [27] Paí linkaí s, G., Bopp, P., Janscoí , G., and Heinzinger, K., 1984, Z. Naturforsch., 39a, 179. [28] R eddy , m. R., and Berkowitz , M., 1987, J. chem. Phys., 87, 6682. [29] Madura , J. D., Pettitt, B. M., and Calef, D. F., 1988, Molec. Phys., 64, 325. [30] Bagchi, K., Balasubramanian, S., and Klein, M. L., 1997, J. chem. Phys., 107, 8561. [31] Kalinichev, A. G., Gorbaty, Y u . E., and Okhulkov , A. V., 1999, J. molec. L iquids, in press. [32] Kalinichev, A. G., and Bass, J. D., 1997, J. phys. Chem A, 101, 9720. [33] Kamb, B., 1968, Structural Chemistry and Molecular Biology, edited by A. Rich and N. Davidson (San Francisco: Freeman) p. 507.
© Copyright 2025 Paperzz