Hydrogen Bonding in Hydrogen Peroxide and. Water. A Raman Study of the Liquid State. Paul A. Giguere* and Hung Chen. DCpartement de Chimie, Universite ...
Hydrogen Bonding in Hydrogen Peroxide and Water. A Raman Study of the Liquid State Paul A. Giguere* and Hung Chen DCpartement de Chimie, Universite Laval, Qutbec, Canada G1K 7P4
The Raman spectra of pure, anhydrous H,02 and its deuteriated analogues, D 2 0 z and HDO,, were examined in the liquid state up to 40 "C and supercooled down to -75 'C. The 0-H (0-D) stretching bands show a broad, single-peaked contour, and a strongly temperature-dependent asymmetry. Decomposition by analog computer reveals two overlapping bands, about 100 cm-l apart, with different temperature behavior. In contrast with water, there is no indication in the hydrogen peroxide spectra of 'free,' i.e. non-hydrogen-bonded, OH groups, in keeping with the flexible geometry of the molecule and its electronic configuration. Based on these spectra, and by analogy with water, a two-state model is proposed featuring a continuum of hydrogen-bonded H 2 0 2molecules, with close-packed OH groups in two kinds of environments. The hydrogen bonds are slightly weaker (12-15"/0) than those in water. Hydrogen peroxide is the best analog for studying the structure of water.
INTRODUCTION ~~~~
~~~~
~~
~~
~
~
feasible to work at higher temperatures, almost up to the boiling point (150 "C).Still, it was possible to cover a temperature range of over 100°C thanks to the remarkable tendency of the concentrated solutions to supercool. Indeed, the extent of supercooling of the pure peroxides happens to be much larger than that of water, which seems limited to about -40 0C.9
The problem of the molecular constitution of hydrogen peroxide in the liquid state is of considerable interest by virtue of its close relationship to that, still unsolved, of ordinary water. However, very little work has been done on it, unlike the free molecule and the crystal, both of which have been studied extensively.' From an early x-ray diffraction study' a model was proposed, DISCUSSION consisting of close-packed OH groups, probably of the face-centered cubic type, each molecule being in contact with six or more neighbors. For obvious reasons, the present spectra are best interSpectroscopic investigations are hampered by the preted in comparison with those of water. From the chemical instability of the compound. For instance, in structural viewpoint, a major difference between them the infrared region, where information is still s k e t ~ h y , ~ - ~ derives from the much greater flexibility of the hydrogen the strong absorption of the liquid requires very thin peroxide molecule.lo Indeed, the internal rotation about films, which favors catalytic decomposition. As for the 0-0 bond is only weakly hindered, the trans barRaman spectroscopy, all previous work was done with rier being less than 5 kJ mol-'. As a result, the molecule mercury arc excitation and the photographic can yield to some extent in order to fit in the environThe long exposures promoted photochemical decompoment, giving a more compact packing than the water sition and degassing of the liquid, which flawed the molecule with its stiff HOH angle. Another major differspectra. With the advent of the laser and the photomultience in molecular interactions follows from the elecplier, these difficulties were largely overcome. The tronic configuration of the H 2 0 2molecule. Because of research reported here was started 10 years ago, but the approximately sp3 hybridization of the oxygen atom unfortunately had to be terminated before completion. orbitals, each OH group has two lone pairs available for However, incomplete as they are, the results provide hydrogen bonding instead of only one as in water. The sufficient information to warrant publication. It is hoped combination of these two factors effectively rules out that they may provide an incentive for further work. any 'free,' i.e. non-hydrogen-bonded, OH in liquid H202 at ordinary temperatures. For the spectroscopist, despite the greater complexity of the H 2 0 2molecule, two important simplifications are EXPERIMENTAL welcome in analysing its Raman spectra in comparison with those of water. First, the coupling between the two The preparation and handling of the liquid samples, the 0-H oscillators is very weak in the stretching modes, instrumentation and the technique for recording the as expected. For instance, the splitting between the spectra were the same as for the study of the small symmetric and antisymmetric vibrations is only 3crystals reported earlier.* The liquids, contained in an 4 cm-' in water.13 This permits a clear identification of 8 mm diameter flat-bottomed Pyrex tube, were heated intermolecular interactions, in contrast to water where only up to 40 "C, although with appropriate care it is they are blurred by the strong intramolecular coupling. Second, there is no possibility of Fermi resonance between the v1 stretching mode, and 2v2, the overtone * Author to whom correspondence should be addressed. CCC-0377-0486/84/0015-0199$03.00 @ Wiley Heyden Ltd, 1984
JOURNAL OF RAMAN SPECTROSCOPY,VOL. 15, NO. 3, 1984 1%
PAUL A. GIGUERE AND HUNG CHEN
Figurel. Raman spectra in the 0-H stretching region of liquid hydrogen peroxide (solid lines) and water (broken lines).
of the bending mode near 2800 cm-'. In the spectra of water, that question has been a subject of controversy in assigning the Raman bands near 3220 cm-'.
The 0-H stretching bands From mere visual comparison, the 0-H stretching bands of liquid hydrogen peroxide and water show some striking differences (Fig. 1). First, the former exhibit a onepeak intensity distribution instead of two for the water bands. Also, they are not so broad, their half-widths being barely 60% of those of water at the same temperature. Also, their frequency is more temperature dependent. Another conspicuous temperature effect is the change in shape of the peroxide bands. At 4 0 ° C and presumably even more so at higher temperatures, the bands are strongly asymmetric on the low-frequency side. This asymmetry decreases on cooling, so that at -55 "C it has almost disappeared. Then, at still lower temperatures, it crosses over to the high-frequency side of the maximum. Such behaviour can only be due to overlap of two or more subbands, with different temperature coefficients. To check this an analog computer decomposition was carried out using pure Cauchy or Gauss functions, or combinations thereof. An excellent fit was achieved with two bands of Gaussian contour, as shown in Fig. 2. Decomposition into three bands brought no significant improvement. From the temperature effect on the populations of the two states, A and B (Table 2), the corresponding intermolecular forces are in the approximate ratio of 5:3. Lastly, there is no sign in the spectrum of H202 of a weak shoulder corresponding to that near 3610 cm-' in the spectrum of water. Weak as it may be, this shoulder has played a key role in the controversy between the 'continuum' and the 'mixture' models for liquid Since its approximate frequency implies a red shift of about 50 cm-' from that of u1 = 3657 cm-' for the free water molecule, one would expect its equivalent in hydrogen peroxide to be around 3600-50= 3550cm-I. Its complete absence is an important clue 200 JOURNAL OF RAMAN SPECTROSCOPY, VOL. 15, NO. 3, 1984
cm-1 Figure 2. Decomposition by analog computer of the 0-H stretching bands of liquid hydrogen peroxide.
in devising a suitable model for liquid hydrogen peroxide, as discussed below. Polarization measurements showed in all cases a fairly constant ratio p=0.14 across the whole band, thereby confirming the weak contribution of the antisymmetric mode u5. The frequency of the latter is not known accurately from the meagre infrared data, but it should be close to that of v1 by analogy with the crystal' where the splitting is of the order of 35 cm-'.
The 0-D stretching bands The spectra of deuterium peroxide in the 0-D stretching region (Table 1)show the expected isotope effects, that is, a decrease in frequency, intensity and band width in the approximate ratio of 1.35. Therefore, they are not reproduced here. The asymmetry of the 0-D bands follows the same temperature pattern as the H202bands, likewise disappearing at about -50 "C, then reappearing at higher frequency below that temperature. Much more instructive are the vibrations of the hybrid molecule HD02. Figure 3 shows the 0-D stretching region of a 5% solution of D 2 0 2in H202,which was studied more closely than the 0-H bands in a similar solution in D202. As in the isotopic water molecule^'^, the frequencies of
HYDROGEN BONDING IN HYDROGEN PEROXIDE AND WATER
Table 1. Band parameters for the Raman spectra of liquid hydrogen and deuterium peroxides _ _ _ _ _ ~ ~ p - 1 )
HDO,
0202
"202
-
_
_
I
A":
T ("C)
v(-m-'l
100 130 150 160
250 260 265 270
40 -15 -55 -75
2525 2495 2480
.. ,180
15 -65
2070
~ I
%
T ("C)
100 115 120
180 205 220
40 -20 -50
v1
v1
3410 3375 3335 3310 2% 2800 2820 2% 1735
vw
1350
vO-H
-50
vw
3415 3340
-20 -20
-200
vO-D
2527 2515 2490
130 140 165
15 -1 5 -70
SOOH
-20
1405 1415
T ("C)
Av:
25
vvw
v2
p - ' l
-40
120 130
40 -55
-5
-80
-40
130
17
80
1022
-55
-30
1405 600D
-20
1033
v3
88 1 -480
25
vw
-120
-65
vvw
-25
-55
v4?
17
88 1
25
VT ,-
170
the two stretching modes of H D 0 2 lie higher than those of the parent molecules, a consequence of intermolecular coupling. As for their asymmetry, it is not only less marked than in the two parent molecules, but also less strongly affected by temperature. For instance, it is still visible on the low-frequency side, although appreciably reduced, at -70 "C (Fig. 3). Also noteworthy, the absence of a high-frequency shoulder in both the 0-H and 0-D stretching bands is still more significant than in the parent compounds. Indeed, in the spectra of water, the 0-H and 0 -D bands of HD O show very pronounced shoulders at 3610 and 2650 cm-', respectively. Compare Figs 3-6 in Ref. 15.
The bending bands Analysis of the OH bending bands of hydrogen peroxide is complicated by various overlapping. For instance, the marked asymmetry of the u2 band (Fig. 4) is due mostly, if not entirely, to the contribution of the antisymmetric vibration, v6, around 1350cm-I. That band is fairly strong, nearly one-fifth as much as the 0-H stretching bands, with a temperature coefficient of about -0.1 cm-' OC-'. This is in sharp contrast with the bending band of water, which is so weak in the Raman spectra that a recent extensive studyI6 could not detect any frequency or width change over a wide temperature range (from 4 to 300 "C). The overtone band, 2v2, well resolved in Fig. 3, is peculiar as it shows very little anharmonicity, and a large depolarization ratio, p = 0.65*0.1. The Raman spectra of the crystal exhibit similar anomalies.* In the spectra of D 2 0 2the shape of bending bands is affected by the strong 0-0 stretching
>-
t:
U) 2!
LI1
t
t
I I
I
I
I
2900
2700
2Mo
2300
cm-
1600
Figure 3. The 0-D stretching bands of HDOZin a 5 mol-Yo solution in H202.The 2u2 bands of the latter are shown around 2800 cm-'.
1400
cm-1
1:
Figure 4. The OH bending bands of liquid H20zat 40 "C.
JOURNAL OF RAMAN SPECTROSCOPY, VOL. 15, NO. 3, 1984 201
PAUL A. GIGUERE AND HUNG CHEN
I
I
I
lo00
1
8oo cm-1
Figure5. The OD bending bands and part of the 0-0 stretching band in deuterium peroxide at -55 "C.
Figure6. Raman scattering of liquid H202 in the low-frequency region.
band nearby (Fig. 5). Incidentally, the overtone of the latter was seen for the first time at 1735 cm-', thereby confirming its weak anharmonicity.
(Table 2) and that of the two main bands in the isotropic spectra of water (3420-3220 cmP1).l5 As for the hydrogen bonds themselves, there are various indications that, on average, they are slightly weaker than those in water. For instance, their length in the crystal," 2.8 A, compared with 2.76 A in ice,13 corresponds to a weakenin of 12-15% according to semi-empirical correlations." This may be rationalized in terms of the stronger polarity of the 0-H bonds in hydrogen peroxide. Luft21 once suggested much stronger hydrogen bonds for cyclic dimers ( H202)2in the liquid. Such dimers are believed to exist in an inert matrix.12 However, there can be no distinct dimers in a fully hydrogen-bonded continuum. In counterpart, one might imagine some six-membered rings, with strong coupling between two H202molecules, in equilibrium with looser, nine-membered rings involving three molecules. However, such a model cannot account for the observed asymmetry of the 0-H and 0-D stretching bands of H D 0 2 . A more appropriate model involves two kinds of environment for the OH groups, with slightly different hydrogen bonds, and intermolecular coupling. In the first, corresponding to subband A, the O-H...O are colinear, or nearly so, as in the crystal. Whether the donor hydrogen atom is directed towards one of the two available lone pairs, or half way between them, is a moot point at this stage. Statistically, the latter is more probable because all the lone pairs are engaged, giving a better distribution of the electrostatic charges, and parallel orientation of the 0-H dipoles. Incidentally, the frequency of subband A approaches that in 1:1liquid mixtures of hydrogen peroxide and p-dioxan.22 In the
The low-frequency bands The low-frequency region contains only two weak and diffuse bands. One is centered around 480 cm-' (Fig. 6), previously reported at 525 25 cm-' ( - 40 oC)6and at 550 cm-' (25 "C)' in H 2 0 2 . It must arise mainly from libration of the molecule about the 0-0 axis, with the hindered internal rotation mode, u4, contributing to the strong asymmetry towards higher frequencies, and by analogy with the infrared spectra of the crystal' and the vitreous s01id.l~Lastly, the very weak band extending from 150 to 190cm-' results from various hindered translations.
*
A MODEL FOR LIQUID H202 Because of the variable geometry of the hydrogen peroxide molecule, devising a model for the liquid becomes a problem in close packing of O H groups rather than of whole H,O, molecules. Also, because of the absence of 'free' OH, confirmed by the present spectra, any suitable model must involve a continuum of hydrogen bonds. Then, the temperature-dependent asymmetry of the 0-H bands, resolvable into two subbands, strongly suggests a two-state formalism, with slightly different hydrogen bonds and intermolecular coupling. It is possible to evaluate the degree of coupling in hydrogen peroxide by comparison with that in water, by means of the criterion of Snyder and Scherer," i.e. the frequency difference between the 0-H (0-D) stretching in the hybrid molecule H D 0 2 , and the mean of the in-phase-coupled v1 and v5 vibrations in the parent molecules. For the liquid, the data are not available, but for the crystal' this difference amounts to ca 65 cm-', viz. uO-H = 3236 cm-' in H D 0 2 , and &u,(A)= 3155 cm-' +v, ( E )= 3188 cm-'1 at 80 K, about half of that in ice. Incidentally, this is the same ratio as is observed between the splitting of the two subbands of the 0-H stretching in liquid hydrogen peroxide 202 JOURNAL OF RAMAN SPECTROSCOPY, VOL. 15, NO 3, 1984
Table 2. Band parameters for the computerized decomposition of the 0-H stretching bands of liquid hydrogen peroxide Temperature ("C)
Subband
40 40 -5 -5 -55 -55
A B A B A B
~(cm-l)
3352 3457 3320 3438 3295 3430
/
Area
105 65 120 70 150 75
270 100 310 110 375 115
HYDROGEN BONDING I N HYDROGEN PEROXIDE A N D WATER
other arrangement (subband B), the hydrogen atom is oriented towards the interstice between two neighboring H 2 0 2 molecules forming a bifurcated bond between three OH groups. The weaker binding expected for such a configuration could account for the large volume increase (16%) on melting the crystal.’ It must be remembered that these hypothetical structures are of the vibrationally averaged type, or V- structure^,'^ i.e. with very short-range order (a few molecular diameters) and extremely brief lifetimes (about lo-” s at ordinary temperatures). Reorientation of the OH groups through internal rotation about the 0-0 bond should be a cooperative phenomenon. The fast. proton transfer23 with H 5 0 2 +and H02- provides a mechanism for the reorientation. A direct test of this model might be achieved through neutron diffraction studies because of the different D...D and D...O distances in the two configurations.
The models for liquid water The above model invites comparison with the numerous models proposed for water (see Ref. 13 for a review.) First, the continuum of hydrogen-bonded H202 molecules is different from the continuum model for water in that the latter assumes a single-state, uniform distribution of hydrogen bond strengths and geometry. To account for the two peaks in the Raman spectra of water at ordinary temperature^'^ the low-frequency one (3220 cm-I) was then ascribed to Fermi resonance between w1 and 2v2, the overtone of the OH bending mode. However, this is disproved by recent spectroscopic measurements under extreme conditions. For instance, in supercooled water at -24 “C the 3220 cm-’ band is by far the major whereas in superheated water (up to 300”C)16 it disappears gradually at the foot of the other band. However, over that wide range, the frequency and intensity of the w2 fundamental remain unchanged. Likewise, in ice, where the situation is also favorable for resonance, analysis of the spectra confirms that its contribution to intensity distribution is only a second-order effect compared with intermolecular coupling between the 0-H oscillator^.^^'^^ Another marked difference with the continuum model for water relates to the weak shoulder at the highfrequency limit (3610 cm-’) of the 0-H bands, generally assigned to ‘free’ O H groups. An alternative interpretation, consistent with the continuum models, calls for a higher density of states at the frequency limit.27 However, this is negated by the above spectra, which show no trace of a corresponding shoulder, particularly on the stretching bands of H D 0 2 , as explained above. Assignment of that shoulder to ‘free’ OH groups is now well established by the following observations: (a) addition of large anions, such as C104-, which disrupt the water structure because of their size, causes a dramatic
increase of the high-frequency shoulder;28 (b) its frequency is about the same as that of w1 of monomeric water molecules in organic solvents such as CCl, (3615 ~ m - ’ ) ;(c) ~ the OH- anion, a non-proton donor because of its negative charge,29 shows a strong and sharp band at that same frequency in the Raman spectra of aqueous solutions.30 That band, barely visible in infrared, has the same narrow shape (Av; = 60 cm-I) as those in (a). An estimate of the fraction of ‘free’ O H groups (not ‘free’ H 2 0 molecules!) at ordinary temperatures may be gathered from the band intensities of the ‘odd’ OH (OD) groups in the Raman spectra of HDO (compare Figs 3-6 in Ref. 15). They indicate values of about one third (at room temperature) to half (at 90 “C) of the 5% of ‘odd’ HDO molecules with one ‘free’ OH group (for ‘free’ OD groups the fraction is lower, in agreement with the observation that HDO molecules bond preferentially through OD3’).From the above, it seems safe to conclude that only a few percent of the water molecules have momentarily one ‘free’ OH group at ordinary temperatures. That fraction increases gradually with temperature, so that in superheated water, e.g. at 300 OCI6, the violent thermal agitation overcomes the directional (covalent) intermolecular forces in favor of isotropic Van der Waals interaction, as shown by the shift of the 0-H band towards 3600 cm-’. According to thermodynamics, liquid hydrogen peroxide and water are equally strongly associated. l o Therefore, the present spectra rule out a whole class of mixture models for water, namely those involving large fractions of the so-called ‘broken hydrogen bonds’ at ordinary temperature (cf. Falk and Ford32 for a partial list). That nebulous concept has led to some descriptions of water as a mixture of up to five distinct species, with different numbers of hydrogen bonds, from none to four (see Ref. 33 for a review). In contrast with the situation in hydrogen peroxide, a suitable model for liquid water should be of the two-state type (except for the few percent of molecules with one ‘free’ O H at room temperature). The first state has strong hydrogen bonds and intermolecular coupling of 0-H oscillators, as in ice, and the other has a different orientation of the 0-H groups as in the bifurcated hydrogen bonds proposed above for state B in hydrogen peroxide. Finally, as a corollary, the present data support the assumption of symmetrical bands for the 0 - H ~ tr e tc h in g .Their ’ ~ symmetry, like their great width, is a consequence of the direct coupling between the 0 - H and the O-..Ostretching vibrations. Acknowledgements Most of the experimental work was carried out by H.C., at the time a postdoctoral fellow of the National Research Council of Canada. The spectra of HDOz were recorded later by J. L. Amau.
REFERENCES 1. P. A. Giguere, Peroxyde d‘Hydrogene et folyoxydes d’Hydrogene, Vol. 4 of Complements au Nouveau Traite de Chimie Minerale de f . Pascal. Masson, Paris (1975).
2. J. T. Randall, froc. R. Soc. London, Ser. A 159, 83 (1937). 3. R. C. Taylor, J. Chem. fhys. 18, 898 (1950). 4. P. A. Giguere, J. Chem. fhys. 18, 88 (1950).
JOURNAL OF RAMAN SPECTROSCOPY, VOL. 15, NO. 3, 1984 203
PAUL A. GIGUERE AND HUNG CHEN 5. 0. Bain and P. A. Giguere, Can. J. Chem. 33, 527 (1955). 6. R. C. Taylor and P. C. Cross, J. Cbem. Phys. 24, 41 (1956). 7. S. A. Ukholin and M. Z. Pronina, Akad. Nauk. SSSR f i z . Inst. 244 (1959); Cbem. Abstr. 55, 24229c (1961). 8. J. L. Arnau, P.A. Giguere, M. Abe and R. C. Taylor, Spectrocbim. Acta, Part A 30. 777 (1974). 9. C. A. Angell, in Water: A Comprehensive Treatise, edited by F. Franks, Vol. 7, Chapt. 1. Plenum Press, New York (1983). 10. P. A. Giguere, J. Chem. Educ. 60,399 (1983). 11. P. A. Giguere and T. K. K. Srinivasan, J. Raman Spectrosc. 2, 125 (1974). 12. P. A. Giguere and T. K. K. Srinivasan, Chem. Pbys. Lett. 33, 479 (1975). 13. D. Eisenberg and W. Kauzmann, The Structure and Properties of Water. Oxford University Press, London (1969). 14. G. E. Walrafen, in Water: A Comprehensive Treatise, edited by F. Franks, Vol. 1, Chapt. 5. Plenum Press, New York, (1972). 15. J. R. Scherer, M. K. Goand S. Kint.J. Pbys. Cbem.78,1304(1974), 16. C. I. Ratcliffe and D. E. Irish, J. Phys. Cbem. 86. 4897 (1982). 17. P. A. Giguere and K. B. Harvey, Mol. Spectrosc. 3, 36 (1959). 18. J. R. Scherer and R. G. Snyder, J. Cbem. Phys. 67,4802 (1977). 19. W. R. Busing and H. A. Levy, J. Cbem. Pbys. 42, 3054 (1965). 20. G. C. Pimentel and A. L. McClenan, The Hydrogen Bond. Freeman, San Francisco (1960).
204 JOURNAL OF RAMAN SPECTROSCOPY, VOL. 15, NO. 3, 1984
21. N. W. Luft. Discuss. Faraday SOC.12, 266 (1952). 22. H. Chen and P. A. Giguere, Spectrocbim. Acta, Part A 29, 1611 (1973). 23. M. Anbar, A. Loewenstein and S. Meiboom, J. Am. Cbem. SOC. 80. 2630 (1958). 24. G. D'Arrigo, G. Maisano, F. Mallamace, P. Migliardo and F. Wanderlingh, J. Cbem. Phys. 75, 4264 (1981). 25. E. Whalley. Dev. Appl. Spectrosc. 6, 277 (1966). 26 M. S. Bergren and S.A. Rice, J. Cbem. Phys. 77, 583 (1982). 27. W. R. Wyss and M. Falk, Can. J. Cbem. 48,607 (1970). 28. G. E. Walrafen, J. Chem. Phys. 55, 768 (1971). 29. P. A. Giguere, Rev. Cbim. Min. 20, 588 (1983). 30. W. R. Busing and D. F. Hornig, J. Phys. Chem. 65, 284 (1961). 31. G. P. Ayers and A. D. E. Pullin. Spectrochim Acta, PartA 32.1641 (1976). 32. M. Falk and T. A. Ford, Can. J. Cbem. 44, 1699 (1966). 33. C. M. Davis and J. Jarzynski, in Water and Aqueous Solutions, edited by R. A. Horne, Chapt. 10. Wiley-lnterscience, New York (1972).
Received 9 November 1983
