J. CHEM. SOC. FARADAY TRANS., 1994, 90(6), 817-823 Photolysis of HOBr and DOBr at 266 nm: OH and OD Product-state Distributions Nebil Shaw, Andrew J. Bell, Michael J. Crawford and Jeremy G. Frey* Department of Chemistry, University of Southampton, Southampton, UK SO95NH ~ The OH and OD rotational and vibrational product-state distributions have been recorded following the 266 nm photolysis of HOBr and DOBr, respectively, in a molecular beam. The photodissociation dynamics are similar to those observed for HOCI. The OH and OD distributions are rotationally cold and, for almost all of the states invesfigated, Gaussian in shape. Preferential population of the OH (or OD) 2113,2spin-orbit state was observed, together with a strong preference for the II(A') A-doublet levels.The alignment, 8:(02), tends towards the limit- ing value of -0.5 at the highest N levels that could be observed. In contrast to the photolysis of HOCI, about 10% of the OH fragment was observed in the first excited vibra- tional state, with a similar rotational distribution to the OH(v = 0) products. The OD product distributions from DOBr were found to be sensitive to the molecular-beam expansion conditions with a bimodal distribution obtained for the colder expansions. The photodissociation is consistent with excitation of a non-bonding electron to a t~ antibonding orbital which promotes rapid and direct bond fission via an upper state of A' symmetry. Relatively few results are available for the dissociation dynamics of the triatomic molecules HOX (X = F, C1, Br and I).However, an example of what is possible from studying the photodissociation of a triatomic molecule is provided by the impressive and extensive series of experiments and theoretical calculations that have been brought to bear on the direct and rapid photodissociation of H20 following excitation to its first electronically excited state.' The difficulty of performing accurate ab initio calculations on the HOX series makes comparison with experiments much more difficult than for H20. In the case of HOCl, on the theoretical side, only the most recent large-scale ab initio calculations have brought reasonable agreement between the observed and calculated UV absorption spectr~m.~,~ On the experimental side, the difficulties in producing pure HOCl led to considerable doubt in the experimentally observed absorp- tion spectrum at wavelengths >300 nm1.4 There is an even greater dearth of information on HOBr for which the recom- mended gas-phase absorption cross-sections are based on those for HOCl and a comparison with the solution spectra.Some of the complexities of the investigation of the HOX series compared to H20 are mitigated by the very fact that they are a series of similar compounds to observe and compare. For example, the photolysis of the HOX com-pounds is complicated by the spin-orbit states of the halogen atom; the dissociation is expected to populate both the X(2P3/2) and X(2P1,2)states, in contrast to the production of only H(2S) from water.Work on HOCl suggests that the pro- duction of a particular C1 atom spin-orbit state is correlated with the OH spin-orbit state.' The larger spin-orbit splitting in Br atoms may help to highlight these correlations. As well as being of fundamental interest in the photo- dissociation dynamics of small molecules, the study of the hypohalous acids is important in assessing the impact these species have in the environment. HOCl and HOBr are pro- posed as important species in stratospheric ozone depletion especially in the Antarctic ozone h~le.~.~ In principle, both species can act as a temporary reservoir for the reactive OH, C1 and Br radicals, but the importance of the reservoir will depend on the photochemical stability of HOX.The results presented in this article are part of our ongoing study of the H(D)OX series of Experimental The photolysis of low-pressure free-jet expansions of HOBr and DOBr was investigated by passing the vapour above the aqueous solutions through a 500 pm glass nozzle. The HOBr(aq) was formed by the reaction of liquid bromine with water in the presence of a suspension of red mercury(I1) oxide in an analogous manner to the preparation of HOC1;" DOBr was prepared in the same manner using D20. Experi-ments on DOBr were conducted using 100 Torr of both Ar and He driving gases to produce a colder supersonic expan- sion through a 200 pm glass nozzle. The apparatus used in these experiments is similar to that described previously for the 266 and 248 nm photolysis experiments on HOCl.7-9 Great care was taken to ensure that the OH fluorescence signals were linear with probe laser power.The probe laser energy was <1 pJ pulse-' at the molecular beam and verti- cally polarized. No fluorescence was observed with a pure water sample and the signal intensities recorded from the HOBr samples were determined to be linear with respect to the 266 nm laser power. The fluorescence signals were cor- rected for variations in the laser powers, the sample vapour pressure and the change in detection efficiency across the band due to the filters and the PMT response. Examples of the OH and OD laser excitation spectra obtained are shown in Fig. 1. At each laser wavelength the fluorescence signals from 20 laser shots were recorded, allowing an estimate of the error at each point to be made.This was carried forward to allow an estimate of the uncertainty in the peak areas and rotational level populations to be determined. Fragment Rotational Alignment The Ab2)(J)m~mentl~,~~can be estimated from the ratio of the intensities of the main transition and accompanying satel- lite line originating from the same lower level. However, the resolution of our dye laser (0.5 cm-'), in conjunction with the relatively large Doppler width (0.7 crn-l) arising from the high degree of translational excitation of the OH radical, means that it was very difficult to resolve satellite transitions for branches other than R,.The lowest ground-state rota- tional level for which these satellites can be resolved in our experiment is N" = 2, and by N" = 6 the intensities of the lines are too small to allow an estimate of 8:(02) to be made; the values obtained for N" = 2-6 are shown in Fig. 2. The value of Dg(02) tends towards ca. -0.5 as N" increases. In the subsequent analysis the N-dependent values of Dg(O2) were used, together with tabulated Einstein B coefficient^,'^ to extract populations from the line intensities, with the same values used for both spin-orbit components and A-doublet J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 t 306 307 308 309 310 31 1 312 wavelengthjnm Fig. 1 OH (a)and OD (b) laser excitation spectra recorded in the A-X band, 0-0 vibronic transitions.The OH and OD were generated by the 266 nm photolysis of HOBr and DOBr, respectively. Both these spectra were recorded using the vapour above an aqueous solution of hypo-bromous acid expanded through a 500 pm glass nozzle with no driving gas. The spectra cover the P, Q and R branches originating from both OH (OD) spin-orbit states. states. The OD fluorescence spectrum has many more blended lines and no value of /?;(02) could be reliably deter- mined. OH Vibrational and Rotational Populations The OH(u = 0) rotational-level populations from the pho- tolysis of HOBr are shown in Fig. 3. They can be seen to be well represented by Gaussian distributions, with the OH(211,/2) spin-orbit state more populated than the OH(211,,2) state.The n(A)state is increasingly favoured over the II(A‘) A-doublet state as N increases and the maximum of the Gaussian distribution occurs at larger N for the ll(A’) than the n(A’’) states. 1 2 3 4 5 6 7 N” Fig. 2 The OH alignment /3;(02) as a function of the rotational quantum number N in the OH(’H3,J spin-orbit state measured by comparison of the main R branch lines with their associated satel- lites. Bi(02) could not be determined for the other hdoublet com- ponent or spin-orbit state owing to the limited resolution of the laser and poor signal-to-noise ratios. Weak transitions were observed originating from OH(u = 1, N). To obtain significant intensities these lines were recorded using a much higher probe laser power (ca.50 p.I pulse-l), sufficient to saturate the transitions. The popu- lations for OH(u = 1, N”) are shown in Fig. 4. The fits in Table 1 show that for the H(A’) A-doublet levels the u = 0 and u = 1 rotational distributions are identical within the error bars. However, the II(A”) A-doublet component of the 2113/2 spin-orbit state has a slightly colder and broader dis- tribution than the equivalent u = 0 levels. A number of OH(u = 0) and OH(u = 1) line intensities were recorded under non-saturating conditions and this enabled us to esti- mate that CQ. 10% of the OH was produced in OH(v = 1). OD Rotational Populations The OD(u= 0) rotational population distributions from the photolysis of DOBr under three different expansion condi- tions (no driving gas, 100 Torr He and 100 Torr Ar) are shown in Fig.5 with parameters from the Gaussian fits listed in Table 2. The overlapped lines in the OD spectrum mean that it is not possible to derive populations from the Q2 lines. In all three cases the ’lI312 spin-orbit state is more populated than the and the II(A’) A-doublet states are favoured over the lI(A’’) as N increases. The population distributions obtained in the absence of driving gas in the expansion and those obtained with 100 Torr of He are very similar and are fitted quite well by a Gaussian distribution. The distribution in the ’llIl2 state ll(A‘) A-doublet component is an exception, having rather a flat top, suggesting possible bimodality. The results of the experiments using 100 Torr Ar are significantly different.The II(A”) Adoublet levels observed using the Ar expansion can be fitted by a Gaussian, though the peak is shifted to lower N compared with the He expansion. The II(A’) levels of the 2111,2spin-orbit state are clearly bimodal in the Ar expan- sion and a fit using two Gaussians is shown. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 819 78001 I a I T I I I i I N" N" 10 5 0 A I 11IV Fig. 3 Rotational population distributions and Gaussian fits for OH(v = 0, J) fragments from photolysis of HOBr: (a) 2113,2spin-orbit state, II(A') A-doublet component; (6) 2113/2spin-orbit state, ll(A") A-doublet component; (c) 2111/2spin-rbit state, ll(A') A-doublet component; (42n1/2 A-doublet corn- spin+-jrbit state, n(~") ponent N" N" .O 350 300 -- I ' I ' I ' I 1 ' 1 (c)-1 250 200 150 100 50 0 N'' Fig. 4 Rotational population distributions and Gaussian fits for OH(v = 1, J) fragments from photolysis of HOBr obtained under saturation conditions : (a) 2113/2spin-orbit state, ll(A') A-doublet component; (6) 2113/2 spin-rbit state, n(A") Adoublet component ; (c) spin-orbit state, n(A')A-doublet component. Populations of the ll(A") A-doublet component of the spin-rbit state could not be extracted owing to poor signal-to-noise.Spin-Orbit Populations The OH spin-orbit population ratio from the photolysis of HOBr shows some variation with N, Fig. 6. For the II(A') states the ratio rises to a plateau of ca.2.5 between N = 2 and 6, but the ll(A") manifold reaches a ratio of 4 in the same region. It is difficult to be sure that the differences between the two A-doublet components are significant as fewer pairs of lines could be resolved in the ll(A") states than for the ll(A') states. Considering the errors involved in these mea- surements it is better to consider the average spin-orbit population ratio over the N range where there is significant population of the rotational levels (N = 2-7) and to concen- trate on the n(A)states. The observed average over all rotational levels indicates a 2.5 :1 preference for the lower-energy OH('II,/,) states and is close to the statistical limit of 2 :1 if we assume that the pro- duction of Br('II,/,) is correlated with the production of OH(2113/2)and similarly for the 'II3/2 states.This is similar to the observations in HOCl. The spin-orbit splitting in Br is ca. 3677 cm-'. Even with a much higher resolution probe laser it is unlikely that this splitting could be resolved in the OH Doppler lineshape. The OD spin-orbit population ratios for the II(A') A-doublet State are shown as a function Of N in Fig. 7 for the three different beam conditions used in these experiments. Table 1 Parameters of Gaussian fits obtained from OH rotational distribution plots following the photodissociation of HOBr conditions state 2n3/2 n(A) 2n3/2 n(Ar‘) HOBr effusive beam, u = 0 2n1/2n(A) 2n1,2WA“) 2n3/2 n(Ar)HOBr effusive beam, u = 1 2n3/2 n(A“) 2n1/2n(A7 The Gaussians were calculated using the expression A exp[ -(x -The corresponding ratios for the II(A”) levels could not be evaluated as so few of the 2111,2(A”) level populations could be determined because of blended lines.The three plots are similar with an initial spin ratio of about 3, climbing to a maximum of 5 at N z 6 or 7 before falling again. The average spin ratio is thus slightly higher for OD than for OH. In order to facilitate the spin-orbit comparison the populations, summed over the A-doublet levels, were fitted to a Gaussian distribution (Table 3). A-Doublet Populations The degree of electronic alignment (DEA), see eqn. (l), is a measure of any preferential production of a A-doublet com- ponent : where P,,,,(X = A‘, A”) correspond to the populations of the two A doublet states with the same N.The DEA shows a trend towards a limiting value of -1.0 as N increases for the photolysis of both HOBr and DOBr and for all beam condi- tions (Fig. 8 and 9). The maximum alignment observed is similar for both molecules (-0.8). The OD data from both the undriven expansion and the He driven expansion give very similar DEA distributions. For the Ar-driven expansions fewer lines were recorded but it seems that the magnitude of Table 2 Parameters of Gaussian fits obtained from OD rotational distribution plots after the photolysis of DOBr conditions state DOBr 100 Torr He DOBr 100 Torr Ar Coefficients as in Table 1. Table 3 conditions HOBr sum A-doublet DOBr sum A-doublet ~ ~ ~~ Coefficients as in Table 1.J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 A m AN 1.OOO f0.039 3.88 f0.16 4.30 f0.29 0.775 f0.042 2.81 f0.20 3.37 f0.35 0.330 f0.015 4.25 f0.15 3.80 f0.26 0.222 f0.004 2.14 f0.08 3.66 f0.13 0.712 f0.039 4.09 f0.22 4.31 f0.52 1.OOO f0.054 0.00 f0.15 4.62 f0.29 0.263 f0.016 4.42 f1.00 3.80 f1.50 N)2/(AN)2]. the DEA is reduced, reaching only -0.5. This may indicate some relaxation of the nascent distribution. Discussion The heat of formation of HOBr, A H --80 kJ mol-’; the O-Br bond energy is 234 kJ mfol’?ihe O-Br bond in HOBr is slightly weaker than the corresponding 0-C1 bond in HOCl (249 kJ mol-’).15 The O-Br bond strength in HOBr is very similar to that in the OBr radical (Dg = 231.3 kJ mol-’) where the C10 bond strength (Dg = 265.4 kJ mol-‘)I6 is significantly larger than the corresponding bond in HOCl.The photodissociation of HOBr to produce ground-state OH and Br products is energetically feasible for wavelengths <511 nm. At 266 nm (NAhv= 449.7 kJ mol-’) the excess energy (ignoring the internal energy of the HOBr) is 170 kJ mol-’. The heat of formation of DOBr is not available, but the O-Br bond energy is not expected to be significantly different from that in HOBr. The majority of the OH and OD fragments from HOBr and DOBr, respectively, are pro- duced in low rotational levels of the ground vibrational state. This means that most of the available energy from the pho- tolysis is channelled into relative translational motion.The Doppler widths of the OH transitions are consistent with the high kinetic energy release. The basic photodissociation dynamics of HOBr parallel those of HOCl. Even in the impulsive limit the larger mass of the Br atom compared with C1 is not sufficient to explain the difference in A 15 AN 1.OOO f0.062 5.76 k0.20 3.90 f0.31 0.379 f0.020 2.55 f 0.77 6.90 f0.95 0.246 f0.013 6.88 f0.25 5.28 k0.50 1.OOO f0.250 6.63 f0.86 4.35 f0.71 0.274 f0.015 3.62 k0.80 6.73 f1.42 0.263 f0.022 7.52 f 0.68 6.76 f1.54 1.OOO f0.070 1.10 f2.21 9.57 f2.90 0.379 +_ 0.038 3.97 f3.17 5.69 f5.44 0.455 f0.023 0.00 f0.30 5.53 f0.48 0.302 f0.021 9.30 f0.24 2.31 f0.43 Parameters of Gaussian fits obtained from a sum over A-doublet states for OH/OD rotational population distributions state A N AN ~~ ~ 2n3/2 LOO0 f0.035 3.35 f0.16 4.04 f:0.25 2n1/2 0.308 f0.012 3.38 f0.14 3.98 f0.22 effusive 2n312 1.OOO f0.079 6.21 f0.35 4.82 & 0.61 Helium 2173,2 1.OOO f0.063 6.45 f0.43 5.50 f0.93 ~~ 40 v 0 N" N" 75 C .-0 c m 550 P d---..fopaa .z 25 c Q 9? -KZLJ0U N" N" 75 50 25 0 2 4 6 a 10 12 N" Fig. 5 Rotational population distributions and Gaussian fits for OD fragments from the photolysis of DOBr: (a) effusive beam,2113/z spin-orbit state, n(A)Adoublet component; (b) effusive beam,2n312spin-orbit state, n(A") Adoublet component; (c) effusive beam,2111/zspin-orbit state, n(A)Adoublet component; (4He backing gas, 'lljlz spin-orbit state, n(A)Adoublet component; (e) He spin-orbit state, n(A") Adoublet component; (f)He backing gas, 2111/2backing gas, *nJlz spin-orbit state, n(A)Adoublet component; (g) Ar backing gas, 'lI3 spin-orbit state, n(A)Adoublet component; (h)Ar backing gas, 'II3/2 spin-orbit state, ll(A") Adoublet component; (i) Ar backing gas, 2111/z spin-orbit state, n(A)Adoublet component fitted with two Gaussians.Populations could not be determined for the n(A") component of the 2111,2spin-orbit state owing to overlapping lines. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 I I I I I I I I 1 I 12345678910 N" Fig. 6 Spin ratios 2113/f: 2111/2over both A-doublet components for OH fragments following the photolysis of HOBr: (0)ll(A), (0)n(A") the fraction of OH(v = 1) population between HOCl and HOBr.The ground-state OH bond length and the bond angles in HOBr are similar to those in HOCl. This leads us to believe that the different product vibrational-state distribu- tions arises from differences in the excited-state geometries. The relationship between parent XO-H bond length and nascent OH vibrational excitation will be further explored in the study of HOI photolysis. The bimodal nature of the population distribution obtained from the experiments with 100 Torr of Ar driving the molecular beam expansion is certainly worthy of comment. Bimodal product rotational-state distributions P 949 N" N, r F c! t? PP N" N, r F N,t? 01'2'4'6'8'lo'1'21 : 2n1,2N" Fig. 7 Spin ratios 2113/2 for OD fragments following the photolysis of DOBr: (a) effusive beam, (b) He backing gas, (c) Ar backing gas 0.0 4 -0.4 n -0.8 N" 0 1 2 3 4 5 6 7 8 9 10 N" Fig.8 Plot of degree of electron alignment (DEA) for OH frag- ments following the photolysis of HOBr and evaluated as (Pn(A,,) -Pn(A*))/(Pn(A,*) where P,(,, refers to the population of the + Pn(A,)), A-doublet component for given N. This ratio tends to +l for %-Pn(At)and to -1 for Pn(A,,lPn(A,,) Q Pn(Af).The DEA for OH('lI,,,) is shown in (a)and for OH(211,,2) in (b).The solid line in (b)is a scaled prediction for the dependence of the DEA on N arising just from angular momentum considerations. have been calculated in the photodissociation of ICN, albeit for much higher rotational excitations.17 The higher N com-ponent of the bimodal OD distribution is similar to the dis- tribution obtained from the experiments with no driving gas or with He.It is possible that the new lower N component is due to some relaxation process which is converting some of the nascent distribution to produce population on low N levels. Certainly Ar has a higher collision cross-section for rotational relaxation than He, but at the local pressure in the beam is such that even the fast-moving OD radicals will have had little chance to collide in the 30 ns between pump and probe laser pulses. We also suspect that the major source of relaxation would be the water present in the expansion.The effect of the water, which will depend on the pressure in the expansion, should be the same for 100 Torr of He as for 100 Torr of Ar. However, the He expansion is similar to the undriven expansion, ruling out any relaxation effects. The more polarizable Ar will produce a significantly colder expansion than the He beam. The reduction in the parent HOBr temperature would be expected to produce a small shift in the product rotational distribution to lower N. This effect has been observed in HOCl, but this shift would not be expected to be large enough to account for the observed low N distribution and does not explain the high N component. Bimodal distributions have been recently observed in the NO product rotational-state populations following photo-dissociation of [CH,ONO], and [(CH,),ONO], clusters.'8 This has been interpreted as being due to the superposition of two NO fragment rotational distributions, one produced from the free monomer and the other from NO production from the clusters.It is possible to form clusters of HOBr -Ar or HOBr H,O in our Ar beam, though we estimate at the J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 0.0Lo -0.5 1 --_0.0 ---02 4.5 1 P ppP -1 .o N" 0 0B -0.5 / no -1 .o 0 2 4 6 8 10 12 N " Fig. 9 DEA plot for OD fragments following the photolysis of DOBr: (a) effusive beam, (b) He backing gas, (c) Ar backing gas. Values for the 2111,2spin-orbit state are unavailable owing to over- lapping lines in the n(A)manifold.pressures of argon that we are using that there would be <5% of the HOBr clustered. Some of the rotational excitation in OH and OD is likely to come from the small additional torque provided by the anisotropy of the excited-state potential." This additional torque will also produce a broadening of the Gaussian dis- tribution and may thus account for the larger ANoD/AN,, ratio commented upon previously. Classical trajectory studies, however, gave a distribution that is significantly colder than the observed distribution7 and a full quantum- mechanical study including the effects of the electronic angular momentum is needed. The widths of the Gaussian distributions are similar for the chlorine and bromine com- pounds, with the deuteriated species having the slightly wider distribution. The ratio of the Gaussian width for the OH and OD distributions is about 1.4, very similar to the ratio found for the photolysis of HOCl and DOCl." The preference for the A-doublet A' states seen in the pho- tolysis of both HOBr and DOBr is consistent with the initial excitation of an antibonding c orbital followed by a rapid in-plane fission of the OC1 bond.This results in the singly occupied OH (OD) p orbital being in the plane of the OH rotation. Angular momentum considerations show that for low N there will not be a significant preference for one A-doublet state over the other. For a dissociation which has a dynamical stereochemical control a preference will show up in the A-doublet populations for the higher rotational states.This general trend is observed for all the OH and OD popu-lations investigated. However, the degree of electronic align- ment does not tend to the limiting value of -1 and in most cases does not follow the angular momentum predictions. This means that the dynamical preference for the formation of the in-plane orbital is not complete. The observation of some of the vector correlations and their N dependence would be useful; unfortunately the limited probe laser resolution prevented their observation via the Doppler pro- files. Conclusions The results presented here show the similar nature of the photodissociation of (D)HOCl and (D)HOBr at 266 nm.The work indicates that photodissociation occurs following exci- tation to an electronic state of A' symmetry which ab initio calculations have shown to be strongly repulsive in the 0-X bond. However, significant differences are observed, most noticeably the observation of OH vibrational excitation in the nascent photofragment and bimodality in the Ar-cooled spectrum of OD from DOBr. N.S. and M.J.C. acknowledge the SERC for studentships and A.J.B. acknowledges the Royal Society for postdoctoral funding. References 1 V. Engel, V. Staemmler, R. L. Vander Wal, F. F. Crim, R. J. Sension, B. Hudson, P. Andresen, S. Hennig, K. Weide and R. Schinke, J. Phys. Chem., 1992,%, 3201, and references therein. 2 S. Nanbu, K.Nakata and S. Iwata, Chem. Phys., 1989,135,75. 3 S. Nanbu and S. Iwata, J. Phys. Chem., 1992,%, 2103. 4 J. B. Burkholder, J. Geophys. Res., 1993,98,2963. 5 A. J. Bell, S.Boggis, J. M.Dyke, J. G. Frey, R. Richter, N. Shaw and M. Tabrizchi, J. Chem. SOC., Faraday Trans., 1994,90,17. 6 J. P. D. Abbatt and M. J. Molina, Geophys. Res. Lett., 1992, 19, 461. 7 G. Poulet, M. Pirre, F. Maguin, R. Ramaroson and G. Le Bras, Geophys. Res. Lett., 1992, 19, 2305. 8 A. J. Bell, P. R. Pardon, C. G. Hickman and J. G. Frey, J. Chem. SOC., Faraday Trans., 1990,86,3831. 9 C. G. Hickman, A. Brickell and J. G. Frey, Chem. Phys. Lett., 1991,185,101. 10 C. G. Hickman, N. Shaw, M.J. Crawford, A. J. Bell and J. G. Frey, J. Chem. SOC., Faraday Trans., 1993,89,1623. 11 F. A. Cotton and G. Wilkinson, Advanced Inorganic Chemistry, Wiley-Interscience, New York, 4th edn., 1980. 12 C. H.Greene and R. N. Zare, J. Chem. Phys., 1983,78,6741. 13 R. N. Dixon, J. Chem. Phys., 1986,85, 1866. 14 W. L. Dimpfl and J. L. Kinsey, J. Quant. Spectrosc. Radiat. Transfer, 1979, 21,233. 15 JANAF Thermochemical Tables, NSRDS-NBS 37, National Bureau of Standards, Washington DC, 2nd edn., 197 1. 16 K. P. Huber and G. Herzberg, Molecular Spectra and Molecular Structure ZV. Constants of Diatomic Molecules, Van Nostrand Reinhold, New York, 1979. 17 J. Qian, C. J. Williams and D. J. Tannor, J. Chem. Phys., 1992, 97,6300. 18 E. Kades, M. Rosslein, U. Bruhlmann and J. R. Huber, J. Phys. Chem., 1993,97,989. 19 R. Schinke and V. Engel, J. Chem. Phys., 1985,83,5068. Paper 3/05599J; Received 16th September, 1993