3958
J . Phys. Chem. 1986, 90, 3958-3964
to the first overtone. For higher overtones, there is an approximate order of magnitude decrease for each quantum level of excitation. Line Widths. The simplest picture of overtone spectra would produce only one transition for each C-H upper state (ul = 0 v1 = 1, 2, 3, 4, 5, 6, ...). In reality one has additional absorption bands close to the main absorption. If there is a high density of vibrational and combination states and a substantial fraction of them interact resonantly, the vibrational spectrum will consist of broad overlapped bands in which the secondary transitions borrow some intensity from the main transition. The nature of the coupling between states will determine the overall appearance of the vibrational spectrum. For this particular molecule, CF3C= CH, the overtone absorptions involving the C-H stretch show only (at each side and separated from the main absorption) bands due to sum and difference transitions which originate from interactions with low-frequency modes, for example, the transitions found for (vu1 f vl0) and (vuI f v7). Also, the hot-band transition (u7 (v7, v v , ) ) which accompanies the main absorption (0 uul) increases its separation in energy from the main band as the quantum number of the upper level increases. The C-H vibrational overtone spectra obtained for CF3C=CH in the present investigation do not show resonant interactions in which the vibrational bands become broad due to superposition of many vibrational bands with slightly different band centers. An example of this would be saturated molecules like CF3H and (CF3),CH, where one lower molecular energy state is coupled with substantial line strength to several close-lying upper states. The bandwidths observed for C-H absorptions of C F 3 C ~ C H are relatively constant. Except for the full width a t half-maximum (fwhm) = 55 cm-’ observed for u1 = 5 , the fwhm for transitions to u1 = 2, 3, 4, and 6 are between 30 and 35 cm-I, which indicates there is no strong resonant couplings between these states and other vibrational fundamentals or overtones. Also, no apparent intensity enhancement of the combination bands due to resonance between the main absorption and the combination bands is observed. In general, the interactions that give rise to sum and difference bands and the interactions that produce hot bands (v7 (v7, u v I ) ) are very strong because these bands are observed for all levels studied. Although these couplings are strong, they are not resonant interactions in nature as is the case for saturated compounds. In compounds such as CF3H and (CF,)$H there is a Fermi resonance involving the C-H bending overtone and the C-H stretch. The main reason for the absence of this Fermi resonance in C F 3 G C H is the energy mismatch between the overtone of the bending mode and the C-H stretch.
-
--
-
Conclusions The spectrum of C-H and CEC overtones of 3,3,3-trifluoropropyne has been investigated with intracavity photoacoustic spectroscopy and standard infrared techniques. Three types of vibrational transitions are identified for C-H absorptions of this molecule: (a) the main overtone transition (0 uv,), (b) the hot band (v7 (u7 uv,)), and (c) combination bands of the type (vuI f vIo) and (uv, f u7). The hot-band absorptions v, (v7 uvI) observed for all the transitions from u = 1 to u = 6 show strong interaction between the C-H stretching mode ( u l ) and the C-H bending mode ( v . ~ ) . The same interaction between vibrational motions is also identified for other molecules with single acetylenic C-H stretching vibrations such as NECH and (CF3)3CC=CH. This strong coupling is nonresonant in nature. Combination bands such as vlvl f vl0 or ulvl f v7 are close in energy to the main absorption but they do not seem to be enhanced by resonance with it. The structure on top of the overtone absorptions u1 = 2 to v1 = 6 is probably due to a set of hot-band sequences that involve the main C-H transitions ( u l ) and hot bands of the CCEC bending transition (vl0). The bandwidths (fwhm) for all main overtone absorptions are approximately 30-35 cm-’, indicating that there are no new levels or combination states that exhibit a strong resonant interaction with this absorption at different levels of excitation. Fundamental bands due to C=C and C-H transitions have cross sections that are comparable. For overtones, there is a considerable reduction in the cross sections of CEC transitions compared with C-H cross sections. The band shape of the acetylenic stretching (vz) does not show a simple PQR structure. The shape of the band is mainly due to hot bands involving the C C S bending vibration (vl0) and the (v2) stretching vibration. The calculated anharmonicity constant (X2J is in agreement with an estimated value from the dissociation energy D and the harmonic frequency. 3,3,3-Trifluoropropyne represents an interesting example of isolated C-H absorptions where overtone transitions are not obscured by the presence of Fermi resonances between the main absorption and combination states.
-
+
-
-
+
Acknowledgment. We thank the National Science Foundation for support of this work under CHE82-06976 and CHE85-06957. We also thank Professor Mark Ratner for some useful discussions. Registry No. CF3C=CH, 661-54-1.
Spectroscopic Evidence for Spatial Correlations of Hydrogen Bonds in Liquid Water J. L. Green, A. R. Lacey, and M. G. Sceats* Department of Physical Chemistry, University of Sydney, N.S.W., 2006, Australia (Received: February 3, 1986)
The low-frequency shoulder of the OH stretching Raman spectrum is developed as a probe of in-phase collective motions in liquid water. Its relative intensity approaches that of ice I as the supercooled liquid temperature tends toward the conjectured thermodynamic singularity in the vicinity of -46 O C . The collective band appears despite the large disorder of the OH stretching frequencies in the liquid compared to the strength of the resonance coupling. The resonance condition required for collective OH motions leads us to conjecture that patches of water molecules with similar hydrogen bond energies, which are capable of sustaining the resonance, appear as water is supercooled toward T,. Introduction Coupled energy and density fluctuations appear as liquid water is supercooled.’*2The conjectured transportbs ( 1 ) Angell, C. A. In Wafer-A Comprehensive Treatise; Franks, F., Ed.; Plenum: New York, 1983; Vol. 7, Chapter 1. (2) Angell, C. A. Annu. Rev. Phys. Chem. 1983, 34, 593.
and X-ray scattering9 divergences have led to an explanation of many of the observed anomalous properties of water as well as R: .I.Angell, ; C. A. J. Chem. Phys. 1976, 65, 851. (4) Zhelenzny, B. V. R u s . J. Phys. Chem. (Engl. Transl.) 1969,43, 131 I; 1968, 42, 950. (5) Angell, C.A.; Sichina, W. J.; Oguni, M. J . Phys. Chem. 1982,86, 992. (3) Speedy,
0022-3654/86/2090-3958$01.50/0 0 1986 American Chemical Society
Spatial Correlations of H Bonds in Liquid Water its solvation properties.'.2.10.11 The structural origin of these fluctuations has been the source of considerable speculation but is far from resolved. Assuming that liquid water at any instant can be regarded as a single connected tetrahedral network of hydrogen bond^,^^-'^ it is reasonable to expect that the fluctuations will arise from the energetic and geometric properties of the network. If the network is modelled as being random in the sense that the energetic and geometric properties of neighboring bonds are uncorrelated, then there is no mechanism available for generation of the long-range correlations required to produce the conjectured thermodynamic singularity at T, = -46 O C 3 q 8 or below.2o The most straightforward mechanism for generating long-range correlations is to assert that there is an energetic correlation between hydrogen bonds, i.e., that a strong hydrogen bond induces, on average, stronger hydrogen bonds adjacent to it in the network. This is the basis of the "flickering cluster" model of Frank and Wen.21 The nonadditive induction terms in the water potential energy provides such a m e c h a n i ~ m . ~ ~The .~~ coupling of energetic and density fluctuations is, however, not easily visualized. Studies of random suggest that regions of weaker, bent hydrogen bonds allow topologically distant molecules in the network to penetrate close to a central molecule in the region, and the resulting nonbonded contact gives rise to an increase in coordination number, hence local density. An alternative structural mechanism is to consider the correlations among the ring structures" or polyhedra'0*24formed by closures of the three-dimensional network. In Stillinger's model,'0324the sharing of faces between adjacent polyhedra provides the mechanism for self-replication and, more recently, Speedy' has suggested that faces generated from stable pentagonal rings would lead to an appropriate coupling of energy and density fluctuations. Some evidence has been obtained from simulation^.^^ Both approaches have appealing features-in the former the energetic correlations between adjacent hydrogen bonds are readily visualized whereas in the latter the geometric, or density, correlations are well described, yet neither gives an adequate description of the important cross correlation of energy and density fluctuations. Speedy and Angel13 have argued that the well-known nonpairwise additive nature of the water molecule interaction,z2 which is a simple source of energetic bond-bond correlations, is not the driving force for the anomalies. The argument is that simulation@ using simple pairwise additive potentials which mimic
(6) Lang, E. W.; Llidermann, H.-D. Angew. Chem., Inr. Ed. Engl. 1982, 21, 315. (7) Cornish, B. D.; Speedy, R. J. J . Phys. Chem. 1984, 88, 1888. (8) Speedy, R. J. J . Phys. Chem. 1982, 86, 982, 3002. (9) Bosio, L.; Teixeira, J.; Stanley, H. E. Phys. Rev. Lett. 1981, 46, 597. (10) Stillinger, F. H. Science 1980, 209, 451. (11) Speedy, R. J. J . Phys. Chem. 1984,88, 3364. (12) Bernal, J. D.; Fowler, R. H. J . Chem. Phys. 1933, I , 515. Bernal, J. D. Proc. R. Soc. London, Ser. A 1964,280, 299. (13) Pople, J. A. Proc. R . Soc. London, Ser. A 1951, 205, 163. (14) Scats, M. G.; Rice, S. A. In Water-A Comprehensive Treatise; Franks, F., Ed.;Plenum: New York, 1983; Vol. 7, Chapter 2. (15) Rice, S. A.; Scats, M. G. J. Phys. Chem. 1981,85, 1108. (16) Gibbs, J. H.;Cohen, C.; Fleming 111, P. D.; Porosoff, H. J. Solution Chem. 1973, 2, 227. (17) Rahman, A.; Stillinger, F. H. J . Am. Chem. SOC.1973, 95, 7943. (18) Geiger, A.; Stillinger, F. H.; Rahman, A. J. Chem. Phys. 1979, 70, 4185. (19) Stanley, H. E.; Teixerira, J. J . Chem. Phys. 1980, 73, 3404. (20) Leyendekkers, J. V.; Hunter, R. J. J. Chem. Phys. 1985, 82, 1447. (21) Frank, H. S.; Wen, W. Y. Discuss. Faraday SOC.1957, 24, 133. (22) Rao, C. N. In W a f e r - A Comprehensive Treafise;Franks, F., Ed.; Plenum: New York, 1972; Vol. 1, Chapter 3. Eisenberg, D.; Kauzmann, W.
The Structure and Properties of Water;Oxford University Press: Oxford, U.K., 1969. (23) Berendsen, H. J. C.; Van der Velde, G . A. In 'CECAM Report of Workshop on Molecular Dynamics and Monte Carlo Calculations on Water", 1972, p 77. Del Bene, J. E.; Pople, J. A. J. Chem. Phys. 1970,52,4858. 1973, 58, 3605. (24) Stillinger, F. H. In Waters in Polymers; Rowland, S. P., Ed.;American Chemical Society: Washington, DC, 1981; ACS Symp. Ser. No. 127, pp 11-22. (25) Speedy, R. J.; Mezai, M. J . Phys. Chem. 1985, 89, 171.
The Journal of Physical Chemistry, Vol. 90, No. 17, 1986 3959 the tetrahedral character of the bonding give rise to anomalous thermodynamic behavior. Further studies should reveal whether this behavior survives when more realistic nonadditive potentials are used in the supercooled regime. Large divergences have not been seen with pairwise additive potential^,^^ but this could be associated with problems of configuration sampling.z In the structural models which invoke ring" or p ~ l y h e d r a ' ~ * ~ ~ structures, the mechanism whereby certain structures attain lower energies is not clear. There is a strong correlation of hydrogenbond energy with bond length and angle^,^*-^^ and it is not inconceivable that the consequence of formation of these structures is a spatial correlation of hydrogen-bond energies. The distinction between the two models would be reduced to a question of cause and effect-whether or not energetic hydrogen-bond correlation generates, or is the result of, low-density configurations of certain polyhedra or rings. From an experimental point of view, there are no probes which provide direct information on structural correlations. While small-angle X-ray scattering indicates the range of density fluctuations (-8 A at -20 OC9) it does not reveal the underlying structural entities. Relaxation and transport measurements can yield the time scales of the fluctuations and thermodynamics their magnitude, but neither reveals their origin.'*2*20Diffraction s t ~ d i e s which ~ ~ . ~measure ~ pair correlations are not sufficiently refined to pin down the three- or four-center correlations implicit in the bond-bond or ring (e.g., dihedral angle) correlations on which models are based. Indeed, adequate oxygen-oxygen pair distributions can be synthesized by assuming a uniform dihedral angle d i ~ t r i b u t i o n , ' ~ Jyet ~ - it~ ~is the dihedral angle which is the indicator of specific ring closures.25In this paper we conjecture that the low-frequency shoulder of the OH stretching Raman spectrum of water is an experimental probe of bond-bond correlations. Collective OH Stretching Motions in Liquid Water The Raman and infrared spectra of water in its condensed phases have been the focus of many s t ~ d i e s . ' ~ ~ 'In ~ -the ~ ~O H stretching region of the spectrum, the early of the spectrum were essentially in terms of normal modes of water molecules engaged in four or fewer hydrogen bonds, with complications such as Fermi resonance with the bending overtone also considered. That this approach is untenable was suggested by the ice I data of Haas and H ~ r n i gin~which ~ the force constant k3 for coupling of O H oscillators on adjacent molecules 0-H00-H was shown to be comparable to that for coupling the OH oscillators within a molecule -H-O-H.-, kz. This coupling produces profound effects observed in the Raman spectrum of ice I.14915*37-39 Theoretical s t ~ d i e s ' ~ *have ' ~ *confirmed ~~ that, at (26) Kataoka, Y.; Hamada, H.; Nod, S.; Yamamoto, Y. J . Chem. Phys. 1982, 77, 5699. (27) Jorgensen, W. L. J . Chem. Phys. 1982, 77,4156. (28) Falk, M.; Knopp, 0. In Water-A Comprehensive Treatise; Franks, F., Ed.;Plenum: New York, 1972; Vol. 2, Chapter 2. (29) Novak, A. Strucr. Bonding (Berlin) 1974, 18, 177. (30) Bosio, L.; Chen, S.-H.; Teixeira, J. Phys. Rev. A 1983, 27, 1468. (31) Englestaff, P. A.; Polo, J. M.; Root, J. H.; Hahn, L. J.; Chen, S.-H. Phys. Rev. Lett. 1981, 47, 1733. (32) Sceats, M. G.; Stavola, M.; Rice, S. A. J. Chem. Phys. 1979, 70, 3927. (33) Walrafen, G. E. In Water-A Comprehensive Treatise; Franks, F., Ed.;Plenum: New York, 1971; Vol. 1, Chapter 5. (34) Scherer, J. R. Advances in Infrared and Raman Spectroscopy; Heyden: London, 1978; Vol. 5, Chapter 3. (35) Scherer, J. R.; Go, M. K.; Kint, S. J. Phys. Chem. 1974, 78, 1304. (36) Haas, C.; Hornig, D. F. J. Chem. Phys. 1960, 32, 1763. (37) Wong, P. T. T.; Walley, E. J. Chem. Phys. 1975, 62, 2418. (38) Scherer, J. R.; Snyder, R. G . J. Chem. Phys. 1977,67,4794. (39) Sivakumar, T. C.; Rice, S. A.; Sceats, M. G. J. Chem. Phys. 1978, 39, 3468. (40) McGraw, R.; Madden, W. G.; Bergren, M. S.; Rice, S. A.; Sceats, M. G. J . Chem. Phys. 1978,69, 3483. (41) Madden, W. G.; Bergren, M. S.; McGraw, R.; Rice, S. A,; Scats, M. G. J . Chem. Phys. 1978, 69, 3497. (42) Bergren, M. S.; Schuh, D.; Sceats, M. G.; Rice, S. A. J . Chem. Phys. 1978,69, 3477.
Green et al.
3960 The Journal of Physical Chemistry, Vol. 90, No. 17, 1986
t
AMORPHOUS SOLID H2O DEPOSITION AT
IO K
t / 3100
\
I
I
3300
3500
3700
J
FREOUENCY (CM') Figure 2. Comparison of the I,, (-) and I , (- - -) OH stretching Raman spectra of H 2 0 and HDO in D 2 0 at 9 0 "C reconstructed from the data of ref 35.
strong dependence of the OH stretching frequency on the hydrogen-bond p r o p e r t i e ~and ~ ~the ~ ~decrease ~ in hydrogen-bond DEPOSITION AT 150 K disorder as the liquid is c o 0 1 e d . ' ~ In ~ ~ice ~ I and HzO(as) the frequency spread due to bond disorder is small (as deduced from the width of the OH stretching spectrum of HDO in D2039947) relative to the intermolecular coupling via k3,so that the resonance &I30 K between adjacent OH oscillators occurs giving rise to the collective band for the in-phase motions through its large polarizability derivative. In hot liquid water at, say, 90 OC the spread of frequencies due to disorder is larger than the coupling to such an extent that the collective character is suppressed. Most modes 3600 3400 3200 3000 are then localized to one OH oscillator because the likelihood of resonance is small, so that no mode can induce a relatively large V" (cm-') polarization of the network. Hence the Raman spectrum will not Figure 1. Comparison of the OH stretching Raman spectrum of liquid exhibit an enhanced low-frequency shoulder. Scherer's data35in water, H20(as),and polycrystallineice at various temperatures. From Figure 1 clearly indicate that the low-frequency shoulder in Ill ref 15. disappears as the temperature is raised and at high temperatures the low-frequency edge of the spectral density of states for OH the spectrum is not too different from the I , spectrum. The motion, the band of considerable intensity which appears in the hypothesis developed above can be tested because it follows that Il Raman l spectrum arises from a collective in-phase stretching at high temperatures the Raman spectrum of HzO should tend motion of the water molecules.44 The band appears as a result toward that of the completely localized OH oscillators of HDO of the large net polarizability change of this vibrational mode. in DzO. In Figure 2, the I,, and I, Raman spectra of H 2 0 and The tetrahedral nature of the bonding is such that the appearance HDO in D 2 0 at 90 'C from ref 35 are displayed. Their similarity of this band is most strongly influenced by the intermolecular is clearly illustrated although the intensity in the I,lspectrum at coupling of OH oscillators, through k3, rather than the intralow frequencies suggests that some collective character persists. molecular coupling k2. A second test of the hypothesis can be carried out in liquid water The significance of the in-phase collective band in the IRaman ll at lower temperatures where the in-phase collective band is already spectrum of ice I is highlighted when that spectrum is compared well established. Introduction of OD oscillators into the O H with the spectrum of the density of states. The density of states network will disrupt the spatial extent of the resonance coupling is relatively f e a t ~ r e l e s s , ~as' is the observed infrared and will reduce the intensity of the in-phase collective band. The The infrared spectrum can be modelled without taking account data of Kint and Scherer4*and Wiafe-Akenten and B a n ~ iinl~~ of collective but this is unlikely to be the case for the dicate that this occurs. The introduction of such defects will be Raman spectrum. analyzed in a forthcoming paper by using the concepts introduced In Figure 1 we compare the Raman spectra of ice I and in this work. Analysis of the collective band can, in principle, give amorphous solid water H20(as)14*15*39 with Scherer's liquid water insight into the spatial correlations of hydrogen bonds in liquid data35from -10 to 90 OC. The dominant feature, as liquid water water. As liquid water is cooled, the spread of hydrogen-bond is cooled, is that the band on the low-frequency side of the Ill OH stretching frequencies should decrease as the hydrogenspectrum rapidly increases in intensity. We suggest that this band bond-energy distribution sharpens. The collective band intensity has the same character as the collective band in ice I at 3100 crn-'. should increase as the probability of resonances between adjacent Further, we assert that the band increases in intensity because O H oscillators increases. However, that increase will only be the resonance condition required for coupling of the OH stretching modest if the OH stretching frequencies on neighboring oscillators motions is obtained a t low temperatures. This results from the are uncorrelated. On the other hand, the appearance of patches of oscillators with correlated OH stretching frequencies would (43) Bergren, M. S.; Rice, S . A. J . Chem. Phys. 1982,77,5 8 3 . (47) Bertie, J. E.; Whalley, E. J . Chem. Phys. 1978,69,3497. (44)Walley, E. Can. J. Chem. 1977, 55, 3429. POLYCRYSTALLINE H 2 0
i-70
K
I
(45) Ikawa, S.-I.;Maeda, S . Spectrochim. Acta, Part A 1968,24A,655. (46)Reimers, J. R.; Watts, R. 0. Chem. Phys. L e f f .1983,94,222.
(48) Kint, S.; Scherer, J. R. J . Chem. Phys. 1978,69,1429. (49) Wiafe-Akenten, J.; Bansil, R. J . Chem. Phys. 1983,78, 7132.
Spatial Correlations of H Bonds in Liquid Water
The Journal of Physical Chemistry, Vol. 90, No. 17, 1986 3961
allow a stronger growth of the collective band intensity as the temperature is lowered. The goal of this paper is to determine this temperature dependence. In this section we have established the basis of our conjecture on a qualitative level. The assignment of the low-frequency band as a collective mode is by no means definitive and relies strongly on the similarity of the spectra of liquid water at low temperatures with that of ice I (particularly at higher temperatures than that displayed in Figure 1, vide infra). The theoretical analysis of the ice I Raman spectra clearly identifies the collective nature of this mode. In the following section we will quantify the collective mode contributions of liquid water relative to ice I. Determination of the Collective Mode Contribution The problem which arises in quantifying the collective band contribution to the IRaman ll spectrum is that it is not well resolved in the liquid water spectrum as shown in Figure 1, whereas it is well resolved in ice I at low temperatures. The complexity of the remainder of the spectrum in ice I suggests that fitting of the broad liquid water spectrum by a set of partly overlapped Gaussian, or other, curves could lead to an unwanted bias. Rather, we have analyzed the spectra by a spectral stripping procedure. We note that in ice I the collective band is strongly polarized whereas the depolarization ratio p = I,(w)/lll(o) of the remainder of the spectrum38 is relatively constant and approximately equal to that of the OH stretching spectrum of an isolated OH oscillator of H D O in D20, defined as pOH. This suggests that, to a first approximation, the strongly polarized collective band Ic(w) can be extracted by the difference spectrum
U w ) = Ill(@) - aI,(w)
YAVENUWBER
w,?0,w0.50.10.-10,-2~~c
-26,-10,10.30.60.70.90%
(1)
with a = [pOH]-l. On the low-frequency side of Ill(u),where I,(w) is small, the resulting spectrum is not sensitive to the exact choice of a, whereas the high-frequency contribution is very sensitive. We are interested in the relative strength of the in-phase collective spectrum defined by
It was noticed that incomplete cancellation at high frequencies using a constant value of a could give rise to unacceptable variation of C. As a refinement procedure, we assumed that &(a)was approximately symmetrical and simply reflected the well-defined low-frequency estimate of Ic(w) about the maximum of Ic(w) defined by (1) to produce the final estimate of the spectrum of the collective band. For ice I a t low temperature the collective band is sufficiently well resolved39that the assumption of a symmetrical band can be tested, by visual observation, to be reasonable (but not exact). The error in Cic,arising from this assumption could be as large as lo%, but it is our expectation that this will be a systematic error in water as well as ice. The value of C for ice I was estimated from the polarized single-crystal spectra of Scherer and Snyder,3s from the low-temperature polycrystalline unpolarized (i.e., Ill I , ) spectra of Sivakumar et al.39and Wong and W h a l l e ~ , ~and ' from the high-temperature polycrystalline polarized ice I spectra of this work (vide infra). The value of Ci, was found to be 0.54 f 0.02. An important point is that Ci, evaluated from this procedure is not temperature-dependent, indicating that the character of the collective mode is not destroyed by thermal disorder. The strong temperature dependence of the spectral width of Ic(w) for ice I has been analyzed to show that the collective mode is dephased by scattering of phonon modes having a mean frequency of 200 cm-1.39 The value of C(as) for amorphous ice H20(as) was determined from the unpolarized spectra of Sivakumar et al.39to be 0.53. This indicates that the character of the collective mode is neither destroyed by the static hydrogen-bond disorder of H20(as), nor by the difference in ring statistics expected for the random network in H20(as). The broadening of the band is accounted for by the static hydrogen-bond disorder.42 At high temperatures the Iand llI , spectra of liquid H 2 0 and H D O are similar, as shown in Figure 2 at 90 OC. The depo-
+
A
3800
3600
3400
3200
3000
2800
WAVENUMBER
Figure 3. Il (A) l and I , (B) OH stretching Raman spectra of liquid H20 at temperatures ranging from 90 to -25 "C.The spectra are normalized to the area of Zll(w).
larization ratio of HDO is nearly constant across the entire band except at high frequency, indicating that pOH is not strongly dependent on hydrogen bonding except for very weak hydrogen bonds. Thus the stripping procedure developed above for a frequency-independent value of pOH (as in ice I) is also reasonable in the liquid phase. The primary objective of this paper is to evaluate Ic(w) and C as a function of temperature. Scherer et a1.j5 have reported the OH stretching spectrum of H 2 0 from -10 to 90 O C , D'Arrigo et aLS0from -24 to 95 OC, and Yeh et aLsl from -21 to 2 OC. From those sets of data, from which IC(@)can be determined,j4q5' there are too few spectra to allow confident extrapolation toward the temperature of the thermodynamic divergence. Therefore we have measured the I,, and I , spectra of H 2 0 in the range from -25 to 90 OC at 1 atm. The spectra were taken from a Pyrex capillary having an 0.d. of 250 pm and i.d. of 100 pm which was sealed a t one end and filled with water.52 The 200-mW output at 514.5 nm of an argon ion laser was focused onto the capillary, which was thermostated to an accuracy of f0.5 O C . The glass fluorescence was trapped by total internal reflection and could be spatially filtered by using a 90-deg collection geometry. In view of the broad spectral widths, a spectral band-pass of 12 cm-' was shown to cause no spectral distortion, and was used to optimize spectral accumulation. The data are displayed in Figure 3A,B at several temperatures. The results are similar to those obtained by 0thers.35950J1The spectrum of ice I was obtained by cooling below -28 OC at which point the (50) D'Arrigo, G.; Maisano, G.; Mallamace, F.; Migliardo, P.;Wanderlingh, F. J . Chem. Phys. 1981, 75, 4264. (51) Yeh, Y.; Bilgram, J. H.; Kiinzig, W. J. Chem. Phys. 1982, 77, 2317. (52) Lacey, A. R.;McPhail, A. K.; Trafalski, Z. J. J. Phys. E: Sci. Instrum. 1985, 18, 532.
3962
The Journal of Physical Chemistry, Vol. 90, No. 17, 1986
Green et al.
Y
3100
z
,-
woo)
