A Combined Experimental and Theoretical Study of Resonance Ž

14 April 2000

Chemical Physics Letters 320Ž. 2000 499±506 www.elsevier.nlrlocatercplett

A combined experimental and theoretical study of resonance 1 emission spectra of SO22ž/ CÄ B Brad Parsons a, Laurie J. Butler a,1, Daiqian Xie b,2, Hua Guo b,) a The James Franck Institute and the Department of Chemistry, UniÕersity of Chicago, Chicago, IL 60637, USA b Department of Chemistry, UniÕersity of New Mexico, Albuquerque, NM 87131-1096, USA

Received 12 October 1999; in final form 24 February 2000

Abstract

1 Experimental and theoretical resonance emission spectra are obtained for a number of vibrational states of SO22Ž. CÄ B . The experimental emission spectra are dominated by Ž.n11n , n 22n , 0 bands, but some weak activity in the anti-symmetric stretch Ž.n 3 is observed. The calculated emission spectra based on an empirical near-equilibrium potential energy surface agree reasonably well with experiment for two lowest states investigated here, but fail to reproduce higher ones. Resonance Raman spectra are also calculated and agree well with an earlier experiment. q 2000 Elsevier Science B.V. All rights reserved.

1. Introduction nearly exponentially proportional to the excess energywx 5,6 , although some mode specificity involving 1 It is well-known that the CÄ B22 state of SO is the anti-symmetric stretch Ž.n 3 has been observed predissociative at wavelengths shorter than 219 nm, wx3,5,6,8,9 . Dynamically, the dissociation products, as manifested spectroscopically by an abrupt drop in SOŽ3 Sy.ŽqO3 P. , have been measured using vari- fluorescence yieldwx 1±6 . Interestingly, the predisso- ous experimental techniques between 218 and 193 ciation threshold coincides with the dissociation en- nmwx 7,10±15 . 1 ergy Ž.D01of the ground Ž XÄ A . electronic state, Although the existence of non-adiabatic interac- y1 1 which has been measured accuratelyŽ 45 725.3 cm tions between the CÄ B2 and nearby electronic states or 218 nm. wx 7 . The decay of fluorescence yield is is well-established, the detailed predissociation mechanism is not yet completely understood. Near 193 nm, experimentalwx 10±13 and theoretical wx 6,16 evidence supports the involvement of a repulsive ) Corresponding author. Fax: q1-505-2772609; e-mail: 1 state, which intersects with the bound CÄ B2 state at [email protected] large S±O distances. Kanomori et al.wx 12 identified 1 q Also corresponding author. Fax: 1-773-7025863; e-mail: it as a repulsive triplet state, but recent emission [email protected]. wx 2 Visiting Professor, Permanent address: Department of Chem- spectroscopy work 17 indicates the involvement of istry, Sichuan University, Chengdu, Sichuan, P.R. China. a repulsive singlet A1 state in the predissociation at

y1 Cs geometries near 50 000 cm . Ab initio calcula- predissociation pathways. This Letter is organized as wx 1 tions 6,16,18 confirmed the intersection of the CÄ B2 follows. The experimental setupŽ. Section 2 and state with both triplet and singlet states at respec- theoretical methodsŽ. Section 3 are first briefly out- tively 6700 and 9700 cmy1 above the dissociation lined. The measured and calculated spectra are then limitwx 6 . Near the predissociation threshold, how- compared and discussed in Section 4. The conclud- ever, the process seems to follow a different path- ing remarks are given in Section 5. way. Ivanco et al.wx 9 suggested that the dynamics is via internal conversion to the state X,Ä followed by prompt dissociation. This mechanism is supported by the observed exponential decrease of fluorescence 2. Experimental yield with the excess energywx 5,6 and statistical energy disposal in the productswx 7,15 , and consistent The instrumentation used to obtain the emission with the observation that the n 3 vibration is the spectrum of SO2 has been discussed previously promoting modewx 5,9 . The two predissociation path- wx17,24 . The 532 nm output of a Quantel YG580-C ways may coexist with each dominating in different series Nd:YAG laser operating at 20 Hz pumps a energy rangeswx 6 . Lambda Physik FL3002E dye laser with an intracav- Dispersed resonance emission and resonance Ra- ity etalon used to reduce the bandwidth of the dye 1 y1 man experiments via the CÄ B2 state have been laser to 0.05 cm . The dye used in these experi- reported in both gaswx 17,19±22 and condensed phases ments was a mix of DCM and LDS 698Ž. Exciton to wx 1 22 . Emission from the bottom of the CÄ B2 state generate light between 683 and 659 nm. Typically, potential energy surface provides information on the dye laser energies were 15 mJ per pulse. The 1 highly excited vibrational levels in the XÄ A1 state output of the dye laser was frequency doubled in a wx20,21 , which are difficult to access otherwise. Ray KDP crystalŽ. Lambda Physik FL 30 giving about 5 et al.wx 17 have recently reported a resonance emis- mJ per pulse. The 300 and 600 nm beams were 1 sion study of SO22Ž. CÄ B near 200 nm. They ob- frequency summed in a BBO crystalŽ. Inrad 561-044 . served strong transitions to states with odd quanta in A Pellin±Broca prism then separated the collinear n 3 the anti-symmetric stretching mode. Some activ- 226, 340 and 680 nm photons. n ity in odd 3 was noted in earlier emission studies The SO2 Ž. Matheson was introduced into the wx19,22 , but attributed to c-axis Coriolis coupling. vacuum chamber as a free jet expansion from an Since such transitions are symmetry forbidden if the IOTA ONE pulsed valveŽ. General Valve with an 1 excited electronic state is of pure B2 character, the open time of 200 ms. The incident laser pulse inter- appearance of strong emission bands into states with acted with the SO2 molecules about 1.0 cm from the odd quanta in was interpreted by Ray et al.wx 17 as an nozzle orifice. 1 indication that the CÄ B2 state is strongly mixed with Light scattered at 908 to the incident laser propa- 1 Y a A1 state at an avoided configuration crossing at gation direction was collected with an 4 focal length Y Cs geometries, thus allowing such emission. Theoret- lens and focused with an 10 focal length lens onto ically, Xie et al.wx 23 have recently developed a the entrance slit of a spectrometerŽ Acton SpectraPro near-equilibrium empirical potential energy surface 275. , which dispersed the light with a 2400 grrmm 1 for SO22Ž. CÄ B . The calculated resonance emission gratingŽ. Milton±Roy . The dispersed light was then spectra agree reasonably well with experimental data imaged onto an optical multichannel analyzer, OMA, for states in the red wing of the CÄÄ§X absorption Ž.EG&G PARC 1456B-990-HQ . The spectral reso- wx20,21 . lution is limited by the width of the entrance slit on In this Letter, we report a combined experimental the spectrometerŽ. 20 mm and the ruling of the and theoretical study of the resonance emission spec- grating. Spectral resolution is about 25 cmy1 FWHM 1 trum of SO22Ž. CÄ B just below the predissociation using this grating. threshold. The state-selected emission spectra pro- The varying linear dispersion across the spectrom- vide valuable insights into the geometric differences eter array was calibrated piecewise using the reso- of the two potential energy surfaces and possibly the nant laser line and the Hg lines at 253.6506 nm and B. Parsons et al.rChemical Physics Letters 320() 2000 499±506 501 at 237.832 nm. The calibration of the dye laser was 4. Results confirmed by the optogalvanic effect using a Mg and a Li hollow cathode lampwx 25 . 4.1. Experimental emission spectra

3. Theory Fig. 1 displays the dispersed resonance emission

spectra of SO2 at excitation wavelengths of 227.73, The resonance emission and Raman spectra re- 224.77, 222.92, 221.16, and 219.81 nm, with XÄ state ported here were obtained using the recently pro- vibrational assignments. The excitation wavelengths posed Chebyshev propagation methodwx 26 . The pro- were chosen to correspond to the CÄÄŽ.Ž. 1,1,0 , C 2,0,0 , cedure for calculating both the resonance Raman and CÄÄŽ.Ž. 2,1,0 , C 2,2,0 , and C Ä Ž. 3,0,0 vibrational states, emission spectra can be found elsewherewx 23,27 . In based on the experimental assignment of Ya- brief, correlation functions were first calculated in manouchi et al.wx 4 . The spectra are best referred to the Chebyshev order domain using the Chebyshev as resonance emission because of the relatively long recursion, and then Fourier transformed to yield lifetimes of these vibrational states. corresponding spectra. Following the recent work of A number of interesting observations is noted. Xie et al.wx 23 , the three-dimensional non-rotating First, the emission spectrum strongly depends on the 1 Hamiltonian was expressed in the Radau coordinates excitation wavelength, or more precisely on the CÄ B2 and the Condon approximation was invoked for the state vibrational quanta. This is expected since the transitions. The vibrational eigenstates on an empiri- Franck±Condon overlaps, which determine the in- cal ground electronic state potential energy surface tensity of the spectrum, are quite different for these were obtained in the previous work of Ma et al. vibrational states. Second, all the spectra are domi- wx 28,29 . nated by the Ž.Ž.n11n , n 22n , 0 or simply n 1, n 2,0 The near-equilibrium potential energy surface of progressions. The activity in the anti-symmetric 1 the CÄ B22 state of SO has recently been developed stretching mode is very weak, evidenced by only a by Xie et al.wx 23 by fitting experimental vibrational few emission bands to states with one quantum in wx frequencies 4 . The double-well potential is consis- n 3.Ž Note that all the CÄ state vibrational states from wx tent with previous model Hamiltonians 30,31 and which emission occurs have zero quanta in n 3.. The y1 gives a root-mean-squares error of 1.45 cm for 21 dominance of the Ž.n12, n , 0 progressions indicates low-lying vibrational frequencieswx 23 . Due to the that the excited state in this energy range is mainly 1 strong coupling between the stretching modes, the of B2 character. vibrational wavefunctions show some unusual fea- Specifically, the spectrum at 227.73 nmŽŽ 1,1,0 . tures, as discussed by several previous theoretical excitation. is very simple and dominated by the wx s studies 23,31±34 . The emission spectra from four Ž.Ž.n11,0,0 and n ,1,0 progressions; both peak at n 1 1 vibrational states near the bottom of the CÄ B2 sur- 3 and have similar intensities. Similar progressions face have been calculatedwx 23 . The reasonably good with higher bending quanta are also seen. At 224.77 agreement with the experiment of Yamanouchi et al. nmŽŽ 2,0,0 . excitation . , there are more such progres- wx s s 20,21 gave us some confidence in the empirical sions with n210±5 and peak positions at n 1or potential energy surface near its equilibrium position. 2. The Ž.n1,4,0 progression is the strongest and peaks s Since most of the vibrational levels used in the at n1 2. The spectra at 219.81 nmŽŽ 3,0,0 . excita- y1 fitting are below 2500 cm relative to the vibration.Ž. can also be assigned to n12,n ,0 progressions s s tional ground state, however, the accuracy of the with n210±6, with peaks at n 1 or 2. The potential energy surface is untested for moderately spectra at 221.16ŽŽ 2,2,0 . excitation . and 222.92 nm excited vibrational states. Furthermore, it has been ŽŽ2,1,0 . excitation . are, on the other hand, much shownwx 23 that the potential energy surface may more complex than the others, although most bands experience difficulties for states with excited anti- still belong to Ž.n12,n ,0 -type progressions. The com- ) symmetric vibration because few levels with n3 2 plexity is largely due to stronger emissions to states have been observed and included in the fitting. with large bending quanta Ž.n2 up to 11 . The pat- 502 B. Parsons et al.rChemical Physics Letters 320() 2000 499±506

1 Fig. 1. Experimental dispersed resonance emission spectra of SO22Ž. CÄ B at five excitation frequencies. B. Parsons et al.rChemical Physics Letters 320() 2000 499±506 503 terns observed here are similar to those found in an lated spectrum, but these emissions are absent in the earlier emission study by Yamanouchi et al.wx 20,21 . experimental spectrum. The discrepancies, also found The activity in the symmetric stretching and bending for the lower CÄŽ. 1,0,0 statewx 23 , are attributable to 1 1 modes can be readily understood since the CÄ B2 the inadequacy of the empirical CÄ B2 potential en- state has longer equilibrium bond lengths and a ergy surface in the anti-symmetric stretching coordi- smaller equilibrium bending angle than the XÄ state. nate. At higher energies, no meaningful comparison can be made between the calculated and measured spectra. We believe that this failure to reproduce the 4.2. Calculated emission spectra experimental spectra stems from the inaccuracy of 1 the empirical CÄ B2 potential energy surface. As Fig. 2 displays the emission spectra calculated for mentioned above, the experimental vibrational fre- 1 the five vibrational states of SO22Ž. CÄ B excited in quencies used in optimizing the potential energy the experiment. Since the energies of these states on surface are mostly below 2500 cmy1. The energies the empirical potential energy surface differ slightly of the CÄÄŽ.Ž. 2,1,0 , C 2,2,0 , and C Ä Ž. 3,0,0 states are from the experimental values, the excitation wave- 2285, 2644, and 2920 cmy1 above the ground vibra- lengths are not exactly the same as in Fig. 1. tional state. The spectra at the two lowest excitation energies reproduce the experimental data reasonably well. For the CÄŽ. 1,1,0 state, for example, the calculated spec- 4.3. Calculated resonance Raman spectra trum is dominated by the Ž.Ž.n11,0,0 and n ,1,0 pro- y1 1 gressions below 5000 cm , in excellent agreement If the lifetime of SO22Ž. CÄ B is short, the excita- with the experimental spectrum at 227.73 nm. These tionremission process becomes a coherent Raman two progressions are of similar intensities and peak processwx 35 . The resonance Raman spectrum via the s 1 at n1 3, again in agreement with the experimental CÄ B22 state of SO has been measured in a hexane spectrum. The calculated spectrum has significantly solutionwx 22 , where the solvent induced vibrational s higher intensities for the Ž.n12,n ,0 bands with n 2 relaxation is rapid. Interestingly, no transition to 2±6 than the experimental spectrum reported here. vibrational levels with odd n 3 quanta was observed Two peaks are present in the bending progression at in the experiment, even at a high energyŽ. 208.9 nm . s n2 0 and 5, reflecting the nodal structure of the In our calculation, the propagation was terminated CÄŽ. 1,1,0 wavefunction in the bending coordinate. before the excited state wavepacket recurs, which Similarly, there is an additional peak near 12 000 corresponds to a short stay on the excited state cmy1 Ž. not shown here owing to the nodal structure potential energy surface. The calculated spectra at of the wavefunction in the symmetric stretching co- three experimental frequenciesŽ 228.7, 223.1, and ordinate. This higher peak in the emission spectrum 208.9 nm. are given in Fig. 3. It can be readily seen has been observed by Yamanouchi et al.wx 21 , who that all the spectra are dominated by a progression in found that the peaks in n2 occur at 1±2 and 6±8, Ž.n1,0,0 , which peaks at the first overtone and de- consistent with our calculation. These authors did creases with n1. These characteristics are in excellent not, however, mention emission in the energy range agreement with experimental observation of Yang wx of this work. and Myers 22 . The dominance of the Ž.n1,0,0 series The agreement for the higher CÄŽ. 2,0,0 state is not indicates that the initial motion of the excited as satisfactory, although the major features of the wavepacket is primarily in the symmetric stretching experimental spectrum are reproduced. For example, coordinate. s the progressions Ž.n12,n ,0 for n 2 0±5 are present In contrast to the emission spectra, the Raman in the calculated spectrum, although their intensities intensities were found insensitive to the excitation do not quite agree with the observation. In addition, frequency. This is understandable based on the short significant intensities are found for emissions to the lifetime on the excited state. Under such circum- s 1 n3 2 anti-symmetric stretching state in the calcu- stances, the empirical CÄ B2 potential energy surface 504 B. Parsons et al.rChemical Physics Letters 320() 2000 499±506 B. Parsons et al.rChemical Physics Letters 320() 2000 499±506 505

emission bands to odd n3 levels that are observed at the near 50 000 cmy1 excitationwx 17 . For the low-lying vibrational states of the CÄ state excited 1 here one cannot exclude mixing of the CÄ B2 state with the ground electronic state. This would be consistent with the proposed internal conversion 1 wx pathway via the XÄ A1 state 9 , which is supported by a number of recent experimentswx 5,6 . Theoretically, reasonably good agreement is found for emission spectra from the CÄÄŽ. 1,1,0 and C Ž. 2,0,0 states. For states with higher energies, however, the agreement deteriorates. This inability to reproduce experimental spectra testifies to the inadequacy of the empirical potential energy surface at higher energies, which can be attributed to the lack of data on highly excited vibrational levels for the fitting process and to possible involvement of other electronic states. Both factors affect the potential energy sur- 1 ÄÄ§ Fig. 3. Calculated resonance Raman spectra of SO22Ž. CÄ B at face in configurations far away from the C X three excitation frequencies. Franck±Condon region. Indeed, the excellent agreement between the calculated and measured resonance Raman spectra and resonance emission spectra from is sufficient to accurately describe the resonance low-lying vibrational states indicate that the empiri- Raman spectrum. 1 cal CÄ B2 potential energy surface is reasonably reliable in the Franck±Condon region. This work clearly demonstrates the urgent need for high quality 5. Concluding remarks ab initio potential energy surfaces of this system.

In this work, we report experimental measure- Note added in proof: After this Letter was sent ments of resonance emission spectra of five vibra- to press, a relevant paper by Bludsky et al.wx 36 Ä 1 tional states of SO22Ž. C B that lie just below the appeared in the literature. These authors have de- predissociation threshold. These spectra are domi- 1 rived a three-dimensional potential of the CÄ B2 state nated by emissions to both the symmetric stretching from high-level ab initio calculations and determined and bending modes, although weak emissions to the corresponding low-lying vibrational energy lev- s n3 1 levels have also been found. The dominance els. This work strongly supports the suggestion made of the Ž.n12,n ,0 progressions indicates that the ex- in Ref.wx 17 that a repulsive singlet state is involved 1 cited electronic state is of predominantly B charac- 1 2 in the predissociation of the CÄ B22 state of SO at ter. However, the presence of emissions to levels high energies. with odd n3, which are not observed at lower excitation energies, may signal the onset of configuration interaction mixing with a nearby electronic state at Acknowledgements Cs geometries. The identity of this state is unclear at this moment, although mixing with the repulsive The work at University of Chicago was supported 1 3 A1 state is a likely explanation for the prominent by the Division of Chemical Sciences, Office of

1 Fig. 2. Calculated dispersed resonance emission spectra of SO22Ž. CÄ B for five vibrational states corresponding to the five excitation frequencies. 506 B. Parsons et al.rChemical Physics Letters 320() 2000 499±506

Basic Energy Sciences, Office of Energy Research, wx13 M. Kawasaki, H. Sato, Chem. Phys. Lett. 139Ž. 1987 585. wx US Department of Energy, under Grant No. DE- 14 P. Felder, C.S. Effenhauser, B.M. Haas, J.R. Huber, Chem. Ž. FG02-92ER14305. The UNM team was supported Phys. Lett. 148 1988 417. wx15 C. Braatz, E. Tiemann, Chem. Phys. 229Ž. 1998 93. by the National Science FoundationŽ. CHE-9713995 wx16 K. Kamiya, H. Matsui, Bull. Chem. Soc. Jpn 64Ž. 1991 2792. and in part by the National Natural Science Founda- wx17 P.C. Ray, M.F. Arendt, L.J. Butler, J. Chem. Phys. 109 tion of ChinaŽ. DX . We thank Prof. Yamanouchi and Ž.1998 5221. wx Dr. Katagiri for sending us the ab initio data of SO . 18 P. Nachtigall, J. Hrusak, O. Bludsky, S. Iwata, J. Chem. 2 Phys. 303Ž. 1999 441. wx19 J.C.D. Brand, D.R. Humphrey, A.E. Douglas, I. Zanon, Can. References J. Phys. 51Ž. 1973 530. wx20 K. Yamanouchi, H. Yamada, S. Tsuchiya, J. Chem. Phys. 88 wx1 H. Okabe, J. Am. Chem. Soc. 93Ž. 1971 7095. Ž.1988 4664. wx2 M.-H. Hui, S.A. Rice, Chem. Phys. Lett. 17Ž. 1972 474. wx21 K. Yamanouchi, S. Takeuchi, S. Tsuchiya, J. Chem. Phys. 92 wx3 T. Ebata, O. Nakazawa, M. Ito, Chem. Phys. Lett. 143 Ž.1990 4044. Ž.1988 31. wx22 T.-S. Yang, A.B. Myers, J. Chem. Phys. 95Ž. 1991 6207. wx4 K. Yamanouchi, M. Okunishi, Y. Endo, S. Tsuchiya, J. Mol. wx23 D. Xie, G. Ma, H. Guo, J. Chem. Phys. 111Ž. 1999 7782. Struct. 352r353Ž. 1995 541. wx24 M.F. Arendt, L.J. Butler, J. Chem. Phys. 109Ž. 1998 7835. wx5 A. Okazaki, T. Ebata, N. Mikami, J. Chem. Phys. 107Ž. 1997 wx25 D.S. King, P.K. Shenck, K.C. Smyth, J.C. Travis, Appl. Opt. 8752. 16Ž. 1997 2617. wx6 H. Katagiri, T. Sako, A. Hishikawa, T. Yazaki, K. Onda, K. wx26 R. Chen, H. Guo, Comput. Phys. Comm. 119Ž. 1999 19. Yamanouchi, K. Yoshino, J. Mol. Struc. 413r414Ž. 1997 wx27 H. Guo, Chem. Phys. Lett. 289Ž. 1998 396. 589. wx28 G. Ma, R. Chen, H. Guo, J. Chem. Phys. 110Ž. 1999 8408. wx7 S. Becker, C. Braatz, J. Lindner, E. Tiemann, Chem. Phys. wx29 G. Ma, H. Guo, J. Chem. Phys. 111Ž. 1999 4032. 196Ž. 1995 275. wx30 A.R. Hoy, J.C.D. Brand, Mol. Phys. 36Ž. 1978 1409. wx8 R. Vasudev, W.M. McClain, J. Mol. Spectrosc. 89Ž. 1981 wx31 T. Sako, A. Hishikawa, K. Yamanouchi, Chem. Phys. Lett. 125. 294Ž. 1998 571. wx9 M. Ivanco, J. Hager, W. Sharfin, S.C. Wallace, J. Chem. wx32 T. Sako, K. Yamanouchi, F. Iachello, Chem. Phys. Lett. 299 Phys. 78Ž. 1983 6531. Ž.1999 35. wx10 A. Freedman, S.-C. Yang, R. Bersohn, J. Chem. Phys. 70 wx33 S.C. Farantos, Laser Chem. 13Ž. 1993 87. Ž.1979 5313. wx34 R. Prosmiti, S.C. Farantos, H.S. Taylor, Mol. Phys. 82 wx11 M. Kawasaki, K. Kasatani, H. Sato, H. Shinohara, N. Nishi, Ž.1994 1213. Chem. Phys. 73Ž. 1982 377. wx35 E.J. Heller, Acc. Chem. Res. 14Ž. 1981 368. wx12 H. Kanamori, J.E. Butler, K. Kawaguchi, C. Yamada, E. wx36 O. Bludsky, P. Nachtigall, J. Hrusak, P. Jensen, Chem. Phys. Hirota, J. Chem. Phys. 83Ž. 1985 611. Lett. 318Ž. 2000 607.