Ion imaging studies of the Cl(2PJ) and Br(2PJ) atomic products resulting from BrCl photodissociation in the wavelength range nm

Transcription

1 Ion imaging studies of the Cl(2PJ) and Br(2PJ) atomic products resulting from BrCl photodissociation in the wavelength range nm Martin J. Cooper, Peter J. Jackson, Leon J. Rogers, Andrew J. Orr-Ewing, Michael N. R. Ashfold et al. Citation: J. Chem. Phys. 109, 4367 (1998); doi: / View online: View Table of Contents: Published by the AIP Publishing LLC. Additional information on J. Chem. Phys. Journal Homepage: Journal Information: Top downloads: Information for Authors:

2 JOURNAL OF CHEMICAL PHYSICS VOLUME 109, NUMBER SEPTEMBER 1998 Ion imaging studies of the Cl 2 P J and Br 2 P J atomic products resulting from BrCl photodissociation in the wavelength range nm Martin J. Cooper, Peter J. Jackson, Leon J. Rogers, Andrew J. Orr-Ewing, a) and Michael N. R. Ashfold School of Chemistry, University of Bristol, Cantock s Close, Bristol BS8 1TS, United Kingdom Benjamin J. Whitaker School of Chemistry, University of Leeds, Leeds LS9 2JT, United Kingdom Received 28 April 1998; accepted 9 June 1998 The near ultraviolet UV and visible photodissociation dynamics of BrCl have been explored using the technique of photofragment ion imaging at 26 wavelengths in the range 235 to 540 nm. Ion images of the Cl( 2 P 3/2 ), Cl( 2 P 1/2 ), Br( 2 P 3/2 ) and Br( 2 P 1/2 ) photofragments reveal both the angular distributions of photofragment velocities characterized by anisotropy parameters, and which of the four possible photofragment pathways are active at different wavelengths. The anisotropy parameters show extensive variation both with wavelength and for the different fragmentation channels, and these variations are interpreted largely in terms of excitations to the A 3 (1), B 3 (0 ), C 1 (1) and D(0 ) states as the wavelength is reduced. At wavelengths between 235 and 262 nm, the Br( 2 P 1/2 ) Cl( 2 P 3/2 ) channel is dominant and at 235 nm, characteristic of a parallel parent transition ( 0) and supporting previous assignments of the absorption in this wavelength range being due to the D(0 ) X 1 (0 ) transition. A minor channel forms Cl( 2 P 1/2 ) Br( 2 P 3/2 ) with an anisotropy indicative of the involvement of an underlying perpendicular absorption ( 1) to a state with 1. Br( 2 P 3/2 ) Cl( 2 P 3/2 ) and Br( 2 P 1/2 ) Cl( 2 P 1/2 ) fragmentation channels are not observed. Excitation in the wavelength range 320 nm to 410 nm results in Cl( 2 P 3/2 ) Br( 2 P 3/2 ) products with an anisotropy parameter of , consistent with assignment of the strong parent absorption to the C 1 (1) X 1 (0 ) transition. For photolysis wavelengths longer than 400 nm, the Cl( 2 P 1/2 )/Cl( 2 P 3/2 ) branching ratio increases with 1.0 to 1.4 for the Cl( 2 P 1/2 ), for Cl( 2 P 3/2 ) becomes less negative and for 450 nm, values lie in the range 0 to 0.2 and Br( 2 P 3/2 ) -parameters also increase. No formation of Br( 2 P 1/2 ) is observed. These observations are, in part, consistent with absorption via the B 3 (0 ) X 1 (0 ) transition, although the nonlimiting -parameter values imply a significant perpendicular contribution to the absorption spectrum. The measured anisotropy parameters for 410 nm are interpreted in terms of excitation both to an 0 state B 3 (0 ) and an 1 state A 3 (1) or C 1 (1), together with transfer of dissociating flux between states during the dissociation American Institute of Physics. S I. INTRODUCTION Understanding the photodissociation dynamics of halogen and interhalogen molecules following absorption of visible and near UV radiation is a considerable challenge both theoretically and experimentally because of the large number 23 of electronic states correlating with the four possible atomic asymptotes for ground state ( 2 P j ) atoms. 1 For BrCl, the atomic-fragment asymptotes, their thermodynamic threshold energies (E th ) relative to the zero-point level of the ground electronic state of BrCl, and the wavelengths corresponding to these thermodynamic thresholds are: 2 BrCl X 1 0 h Br Cl E th cm nm, 1 a Author to whom correspondence should be addressed. Tel: Fax: Electronic mail: a.orr-ewing@bristol.ac.uk Br Cl* E th cm nm, 2 Br* Cl E th cm nm, 3 Br* Cl* E th cm nm. 4 Halogen atoms labeled with and without an asterisk indicate formation of upper ( 2 P 1/2 ) and lower ( 2 P 3/2 ) spin-orbit components of the ground 2 P state, respectively. Detailed measurements of the perpendicular ( 1) or parallel ( 0) character of the above channels at different excitation wavelengths can provide much insight into the interplay of excitations to the various excited electronic states of BrCl in the Franck-Condon region, and of any subsequent nonadiabatic transitions between states as the atoms separate. As Fig. 1 shows, the ground state and no fewer than 22 electronically excited states of BrCl all correlate with Cl( 2 P) Br( 2 P) atoms, 1,2 yet its continuous absorption spectrum in the range nm has traditionally been /98/109(11)/4367/11/$ American Institute of Physics

3 4368 J. Chem. Phys., Vol. 109, No. 11, 15 September 1998 Cooper et al. FIG. 1. Diagram showing the low-lying electronic states of BrCl and their expected correlations to the four possible asymptotes corresponding to the different spin-orbit states of the Cl( 2 P) and Br( 2 P) atoms. The four-digit numbers on the left-hand side represent the electronic occupancies of the valence * * orbitals. interpreted in terms of excitations to just three 3 or four 4 electronically excited states. Such treatments, which assume that the total absorption can be deconvolved into the requisite number of overlapping band envelopes, each of which can be parameterized in terms of a simple analytic function, have led to the following suggested spectral assignments: D(0 ) X 1 (0 ) transition, with peak absorption at max cm 1 ( max 228 nm); C 1 (1) X 1 (0 ), with max cm 1 ( max 372 nm); B 3 (0 ) X 1 (0 ), with max cm 1 ( max 442 nm and, according to one treatment, 4 a weak contribution from the A 3 (1) X 1 (0 ) transition, with max cm 1 ( max 536 nm). The B X system has been studied by Clyne and coworkers in emission 5 and via laser induced fluorescence LIF. 6 9 These investigations confirmed the parallel i.e., 0 character of the B X transition and identified the onset of predissociation in B state levels with v 7. From such experiments, Clyne and McDermid 6 estimated a dissociation energy for BrCl dissociating to ground state Br Cl atoms of D cm 1. The efficiency of this predissociation was observed to be rotational level dependent and to scale with J(J 1), implying that dissociation proceeds via Coriolis coupling to an 1 state. Two such nonradiative pathways can be envisaged coupling to the A 3 (1) state at the inner turning point of the B state potential energy curve or an outer limb coupling to the repulsive C 1 (1) potential both of which correlate to the ground state asymptote see Fig. 1. The relative importance of these two possible routes remains unclear, though the observed energetic onset of the predissociation does suggest that, for the latter mechanism to be operative, the bound-free crossing would have to occur at, or very close to, the energetic threshold for dissociation. From the foregoing it is clear that the diabatic B state potential must correlate to a higher dissociation limit, probably to Br Cl* channel 2. 5 Such is consistent with observations of diffuse vibronic structure associated with two higher vibrational levels v 8 and 9 of the B state in the wavelength dispersed E(0 ) B emission spectrum. 10 Tellinghuisen and co-workers used these data to produce the refined value for D 0 (Br Cl) cm 1 listed above, and to generate an effective adiabatic potential energy curve for the B 3 (0 ) state which is characterized by a barrier located at cm 1 above the ground state minimum i.e., 275 cm 1 above the ground state separated atoms asymptote. This barrier is attributable to the effects of an avoided crossing with a repulsive 0 state which correlates diabatically with the ground state atomic products. The parent absorption spectrum at wavelengths shorter than 545 nm, i.e., at energies above this exit channel barrier on the B state potential, shows no vibronic structure, suggesting rapid dissociation. The photodissociation of BrCl has received only sparse attention to date: Qian and co-workers 13 used REMPI of the Cl and Cl* atoms to deduce the Cl*/Cl product branching ratio following photolysis at six wavelengths between 389 and 500 nm. This branching ratio was observed to peak at phot 460 nm, where the ratio Cl*/ Cl Cl* 0.6, and to fall to 0.2 at 390 nm. This wavelength dependence was discussed in terms of possible nonadiabatic transitions among various of the 0 excited states. The nature of the low-lying electronic states of diatomic interhalogens (XY) was summarized by Child and Bernstein 14 and these authors suggested that states correlating to X Y*, with Y the heavier of the two atoms, will be hard to observe experimentally because their potential minima are displaced to longer internuclear separations than the ground state and the states are likely to be predissociated. The potentials for the different XY molecules typically show a B 3 (0 ) state that correlates diabatically with X* Y but which has a nominally avoided crossing with another 0 state giving an adiabatic correlation of the B-state to the X Y asymptote. For IBr, however, Child 15 argued that the crossing is best described in an intermediate coupling regime: Subsequent to excitation to energies above the I Br* asymptote, the dissociating molecules preferentially remaining on the diabatic curve leading to Br*. The theoretical treatment proved consistent with experimental measurements of the Br*:Br yield from IBr photolysis at wavelengths from 450 to 530 nm. 16,17 More recent photodissociation experiments on the interhalogens have probed the effects of shorter wavelength excitations, such as the photodissociation of ICl at wavelengths from 235 to 248 nm. 18,19 A detailed study of atomic orbital alignment and coherence effects for Cl atoms from 532 nm photolysis of ICl, however, is highlighting the complexity of the dissociation dynamics of the interhalogens in the vicinity of the B 3 (0 ) state. 20 In this study we make use of the technique of photofragment ion imaging 21,22 both to determine the symmetry of electronic absorptions leading to production of the various atomic fragments and to ascertain which of the above dissociation pathways are active. The ion image contains information on the angular distribution of state-selected photofrag-

4 J. Chem. Phys., Vol. 109, No. 11, 15 September 1998 Cooper et al ments with respect to the polarization vector of the photodissociating laser beam and hence on the value of in the absorption step and the radius of the image is determined by the photofragment recoil speed, from which the electronic state of the cofragment 2 P 3/2 or 2 P 1/2 can be deduced. We have recorded ion images at a total of 26 photolysis wavelengths in the range nm, detecting all four possible atomic fragments Cl, Cl*, Br and Br*. Parallel ion imaging studies of the wavelength dependence of the Br and Br* products arising in the photolysis of Br 2 an inevitable contaminant in most BrCl samples will be reported elsewhere. 23 The wavelength dependence of contributions from channels 1 4 and of the recoil anisotropies of the various atomic products provide insight into the evolution of the different excited states populated as the photodissociation wavelength is scanned, and provide further evidence for the role of nonadiabatic transitions during the fragmentation. II. EXPERIMENT The ion imaging apparatus is modeled on the design of Chandler and Houston, 21 but with a mutually orthogonal arrangement of the molecular beam, laser beam and time-offlight TOF axis. It consists of three differentially pumped chambers. A pulsed valve General Valve Series 9 and a skimmer located in the first chamber generate a supersonic, collimated molecular beam which is intersected in the second chamber by polarized laser pulses for BrCl photodissociation and for resonance enhanced multiphoton ionization REMPI of the atomic photofragments. The ions formed by the second laser pulse are extracted by focusing electric fields into a field free TOF tube, at the end of which they strike a position sensitive detector. BrCl was prepared by mixing Br 2 and Cl 2 gases in a 5l glass bulb and leaving the mixture for at least 12 hours to equilibrate. The resultant BrCl/Cl 2 /Br 2 mixtures were diluted in argon typically 50% in 1 atmosphere prior to use. BrCl was photolyzed over the range of wavelengths from 235 to 540 nm, and two different laser systems were employed to span this range. The atomic fragments were probed by 2 1 REMPI using the following electronic transitions: 24,25 Cl 3p 5 ; 2 P 3/2 2h nm Cl 3p 4 4p 1 ; 2 D 3/2, 5 or Cl* 3p 5 ; 2 P 1/2 2h nm Cl 3p 4 4p 1 ; 2 P 1/2, each followed by Cl 3p 4 4p 1 h Cl e, and Br 4p 5 ; 2 P 3/2 2h nm or Br 4p 4 5p 1 ; 4 D 3/2, Br* 4p 5 ; 2 P 1/2 2h nm Br 4p 4 5p 1 ; 2 S 1/2, each followed by Br 4p 4 5p 1 h Br e, 10 The unavoidable mixture of BrCl, Br 2 and Cl 2 in the sample gas inevitably means that, in regions of strong Cl 2 or Br 2 absorption, images of Cl and Br atoms from BrCl photodissociation will be superimposed on images from, respectively, Cl 2 and Br 2 photolysis. The different bond strengths and cofragment masses of the homonuclear and heteronuclear halogen molecules ensure that at most wavelengths the velocities and hence the image diameters of the photofragments from different parent molecules can be separated, and indeed the photodissociations of Cl 2 and Br 2 provide calibrations of the proportionality between the radii of the ion images and photofragment velocities. To minimize contributions from Br 2 photolysis to the ion images obtained by detection of Br atoms, and from Cl 2 photolysis to images of Cl photofragments, gas samples rich in Cl 2 and Br 2, respectively, were prepared. UV/visible spectra of the BrCl/Br 2 /Cl 2 gas mixtures were subsequently recorded to quantify their composition. The other source of background signal, photolysis of BrCl by the probe laser, generally lies at much larger radius than the two-color photolysis and probe data of interest. Photolysis light in the range nm was provided by a reflection from a quartz flat of the fundamental signal beam of an optical parametric oscillator Spectra Physics MOPO The optical parametric oscillator was pumped by 450 mj of 355-nm radiation from a Nd:YAG laser Spectra Physics GCR-250. For wavelengths in the range nm, an excimer-pumped dye laser Lambda Physik LEXTRA and FL 2002 operating on DMQ, QUI and Exalite 428 dyes was used. Photolysis radiation of 355-nm was obtained from the third harmonic of a Nd:YAG laser. To generate wavelengths around 320 nm the signal beam of the optical parametric oscillator was frequency doubled in a KDP crystal. Photolysis wavelengths were calibrated using an optogalvanic spectrum of neon excited in a hollow cathode discharge. The probe laser wavelengths were generated by frequency doubling in a BBO crystal the fundamental of a dye laser either the excimer-pumped system or a Lumonics HD 500 pumped by a Nd:YAG Spectra Physics GCR-230 operating on LDS 489 Exciton dye. Photolysis wavelengths of 235 nm and at nm required the single excimerpumped dye laser both to photolyze BrCl and to probe one of the halogen atom fragments. The photolysis and probe lasers were focused into the chamber using lenses of focal length 30 cm and 5 cm, respectively. Typical photolysis laser energies varied from 1 mj per pulse at 235 nm to 5 mj per pulse in the visible. Ion images were recorded using a photolysis laser polarized vertically in the laboratory frame so that the electric vector of the radiation was parallel to the face of the detector. Additional images were taken with the photolysis laser polarized horizontally along the TOF axis ; the resulting annular images serve to confirm that the applied electric fields introduce no detectable image distortion. Checks were 9

5 4370 J. Chem. Phys., Vol. 109, No. 11, 15 September 1998 Cooper et al. FIG. 2. The absorption spectrum of BrCl between 200 and 600 nm solid line, adapted from Ref. 4. The dashed lines show proposed contributions also adapted from Ref. 4 to the spectrum from excitations from the ground X 1 (0 ) state to the A 3 (1), B 3 (0 ), C 1 (1) and D(0 ) states. made to test for image sensitivity to the probe laser polarization but no effects were discernible in the present study. The TOF mass spectrometer is a modification of our previous Wiley-McLaren configuration design. 19,26 The interaction region of the molecular beam and photolysis and probe lasers lies equidistant between two 125-mm diameter annular electrodes. These electrodes both have central 30-mm diameter apertures which, in the rear repeller electrode, is covered by 85% open area nickel mesh Buckbee Mears. These plates are separated by 40 mm and biased at, typically, 1000 V repeller plate and 1000 V, respectively. Four vertically mounted, equally spaced and suitably biased stainless steel shims positioned above and below the TOF axis help to ensure a uniform field gradient within this source region. The extracted ions are accelerated to 3900 V in the second stage of the spectrometer length 20 mm before entering the 560-mm long, field free TOF region all of which is held at this voltage. The annular extraction electrodes allow velocity mapping 27 of ions on the positionsensitive detector given such an appropriate choice of bias voltages, thereby enhancing the resolution of the resultant ion images. Ions exiting the TOF region strike a tandem, 40-mm active area microchannel plate MCP and UV enhanced fast phosphor P47 screen detector. The novel data acquisition procedure has only been summarized previously 19,26 and is described in more detail here; a rather similar real-time data collection and processing strategy has very recently been reported by Houston and co-workers. 28 The positions of ion impact on the twodimensional detector are recorded by a blue-enhanced S25 response CCD camera equipped with a gated image intensifier Photonic Science. The camera output is traditional EIA RS170 analogue interlaced video signal at 30 Hz. A home built 3 circuit reduces the video sync. output to a 10 Hz pulse train which is used to trigger with user selectable delays the pulsed valve, the photolysis and REMPI lasers, the gate of the CCD image intensifier and the digital signal processor DSP5000 card. Fragment ion selection is achieved by gating the CCD image intensifier pulse duration typically 250 ns at the time delay appropriate for the TOF of the ion m/z of interest. Data acquisition is via a suitably modified Tracker board DataCell Ltd.. On receiving a trigger signal, this system acquires the next frame and thence the x,y coordinates of all object pixels with intensity above a user set threshold value. A single ion impact illuminates several neighboring pixels. Thus the x, y coordinates are processed into groups defined as having some minimum separation between each group member and the centroid of each group is then calculated and stored in a memory buffer on the card. These data are then transferred to the memory of a 486 PC, on a shot-by-shot basis, where they are stored in a user defined experiment file and also accumulated into a virtual image buffer which forms a frequency-ofoccurrence image available for display at any time after the accumulation has started. The final image is constructed from an accumulation of these centroids while the probe laser frequency is repeatedly scanned across the full Doppler width of the REMPI transition. In contrast to the traditional ion imaging experiment based on direct accumulation of captured image data, real-time ion counting offers the advantages that, post thresholding and centroiding, each ion impact has been reduced to a single pixel address, and image distortion from such factors as nonuniform spatial response across the active area of the MCPs and/or the phosphor screen is much reduced. The three-dimensional distribution of photofragment velocities is reconstructed from the twodimensional image using a filtered back-projection method. 29 III. RESULTS A. Absorption spectra Two different samples of BrCl were prepared for the photodissociation studies: as described in the preceding section, samples rich in Cl 2 or rich in Br 2 were allowed to equilibrate. The resultant Cl 2 /Br 2 /BrCl mixtures were analyzed using a UV/visible spectrometer to determine the concentrations of the different species since analysis of some of the ion-imaging data presented in subsequent sections requires a knowledge of the mixture compositions. The UV/ visible absorption spectra were fitted to spectra of Br 2 and Cl 2 recorded using the same spectrometer, and to the known absorption spectrum of BrCl. 4 For the Cl 2 -rich sample, we established the relative concentrations for Cl 2 :Br 2 :BrCl to be 2.719:0.055:1, and for the Br 2 -rich bulb, this ratio is 0.125:1.1:1. From these ratios, a value K can be determined for the 298-K equilibrium constant for the reaction Br 2 Cl 2 2BrCl. 11 As a guide to the subsequent results and discussion, Fig. 2 shows the BrCl absorption spectrum in the wavelength range nm, together with the decomposition into excitations to the A 3 (1), B 3 (0 ), C 1 (1) and D(0 ) states proposed by Hubinger and Nee. 4 B. Short-wavelength photodissociation: nm One-color BrCl photodissociation and photofragment detection experiments were performed at nm with detection of Br* via 9, nm with detection of Br atoms via 8, nm with detection of Cl via 5, and

6 J. Chem. Phys., Vol. 109, No. 11, 15 September 1998 Cooper et al FIG. 4. Slices through the reconstructed, three-dimensional images of Cl and Cl* from BrCl photolysis at five photolysis wavelengths from 450 to 355 nm. Little or no Cl* was observed for wavelengths shorter than 400 nm. In all images, the linear polarization vector of the photolysis laser lies vertically. The double ring in the 355-nm image is a result of Cl formation from photolysis both of BrCl outer ring and Cl 2 inner ring. FIG. 3. Raw ion images and slices through the reconstructed 3-D images of Cl and Br* atoms from the photolysis of BrCl. Top: Cl at nm; Bottom: Br* at nm. In all images, the linear polarization vector of the photolysis laser is parallel to the vertical arrow. The right-hand scale shows how distance from the centres of the images relates to the photofragment speeds. The angular variation of the intensities of the reconstructed slices are used to obtain photofragment recoil anisotropy parameters, as described in the text nm with detection of Cl* via 6. Figure 3 shows raw, symmetrized 2-D ion images and reconstructed slices through the 3-D images, obtained from the 2-D images by filtered back projection, at wavelengths of nm Br* detection and nm Cl detection. Images of Cl* at nm were too large to fit completely on our detector and hence only partial images were recorded. The Br and Cl* signals were much weaker than the signal observed for Cl and Br* atoms under identical experimental conditions. Analysis of the radius of the single ring arising from detection of Cl from BrCl photolysis shows that the Cl has a mean speed of 2900 m/s with a distribution of speeds about this mean caused by the initial spread of parent internal energies and instrumental blurring. The single 2900 m/s Cl-atom speed is consistent with the Cl atoms being formed in conjunction exclusively with Br* atoms. The partial image of Cl* photofragments shows just a single photofragment speed and the larger radius than for Cl images demonstrates that the Cl* is formed with Br cofragments. The images for Br and Br* photofragments also show a single speed group, but the resolution of our instrument is insufficient to determine whether the Br/Br* is formed in coincidence with Cl or Cl* atoms, or both. This lack of resolution is a consequence of the smaller spin-orbit splitting of Cl( 2 P) (881 cm 1 ) compared to Br( 2 P) (3685 cm 1 ): at a photodissociation wavelength of 260 nm we calculate that the recoil speeds of Br atoms formed via channels 1 and 2 are, respectively, 1371 and 1341 m/s, and the resolution of our ion-imaging apparatus is estimated to be 100 m/s. The image obtained from detection of Cl at nm, demonstrates, however, that channel 1 forming Cl Br is not significant at this wavelength, and likewise Cl* detection at nm shows that channel 4 does not occur. Hence we propose that at the shortest wavelengths ( 235 nm) employed in the current study, photofragmentation of BrCl proceeds via channels 2 and 3 with little or no contribution from channels 1 and 4. The image intensities confirm that the dominant process is channel 3. The angular variations of the photofragment recoil velocities were derived from the reconstructed images in Fig. 3, and were fitted to the well-known velocity distribution function: 30 I v, 4 1 f v 1 P 2 cos, 12 where is the angle between the electric vector of the linearly polarized photolysis laser and the photofragment recoil direction, f (v) is the speed distribution of the detected photofragment, is the anisotropy parameter, and P 2 (x) (3x 2 1)/2 is a second Legendre polynomial. For photodissociation of a single BrCl quantum state, f (v) is, in principle, a delta-function for each photofragment channel. For a prompt dissociation, takes the limiting values of 2 following a parallel ( 0), and 1 following a perpendicular ( 1) photoexcitation. Note that in the analysis of our photofragment angular distributions we are neglecting possible effects of orbital alignment 31,32 of the atomic products, but we anticipate such effects to be small because of rapid hyperfine depolarization. Fits to the angular-distribution data for BrCl photolysis at these short wavelengths yield values of Cl at 235 nm and Br * at 262 nm. Here, we use subscript labels to distinguish anisotropy parameters for the different detected photofragments. Inspection of images recorded by detection of Br and Cl* reveals that these photofragments arise subsequent to a predominantly perpendicular transition, but the weakness of the signal and the large size of the Cl* image preclude a more quantitative analysis. These observations are consistent with the formation of Br* exclusively via channel 3 at 262 nm, as well as at 235 nm, and of Br formation via channel 2 at both 235 and 260 nm. C. Longer wavelength photolysis: 320 nm Numerous images for the different atomic fragments from BrCl photolysis were recorded at wavelengths between 320 nm and 540 nm. Figure 4 shows illustrative examples for

7 4372 J. Chem. Phys., Vol. 109, No. 11, 15 September 1998 Cooper et al. FIG. 5. Anisotropy parameters,, determined for the Cl, Cl* and Br fragments from BrCl photodissociation at photolysis wavelengths from 320 to 540 nm employed in this study. Error bars are estimated uncertainties in based on the reproducibility of the images and exceed the standard deviations obtained from fits to the individual photofragment angular distributions. For clarity, anisotropy parameters for photolysis wavelengths shorter than 320 nm are not shown in the figure: for 235 nm, Cl and Cl* 0; for 260 nm, Br 0, for 262 nm, Br* Cl and Cl* detection at 5 representative wavelengths. The images clearly evolve both in radius and in anisotropy as the photolysis wavelength is changed. Images were taken of the Cl, Cl* and Br photofragments to obtain as complete as possible a picture of the photodissociation processes, and were analyzed both to deduce the cofragment and hence the dissociation channel and the anisotropy of the recoiling photofragments via fits to Eq. 12. Attempts were made to image Br* atoms from photodissociation in this wavelength range but no signal was detected. The observed changes in the radii of the images are a result of energy conservation as the photolysis wavelength changes. Figure 5 summarizes the values of parameters obtained for the various photofragments. At 410 nm, the Cl products show a limiting anisotropy of Cl The limiting anisotropy parameters are indicative of Cl formation via prompt dissociation following excitation on a perpendicular ( 1) transition. From 415 to 450 nm the perpendicular anisotropy evolves gradually to a near-isotropic angular distribution ( Cl 0) and for wavelengths longer than 450 nm, Cl remains close to zero in the range 0 to 0.2 with no apparent wavelength dependence. In marked contrast, the Cl*-forming channel shows a positive, but nonlimiting anisotropy parameter Cl * 1 to 1.5 for wavelengths longer than 400 nm, with, perhaps, a gentle decline in Cl * values as the wavelength increases. At wavelengths shorter than 400 nm we detected little or no Cl*, consistent with the measurements of Cao et al. 13 of the branching between Cl and Cl* products from BrCl photodissociation. The angular anisotropy of the Br photofragments changes from a distribution characteristic of a perpendicular excitation ( Br 1) at wavelengths shorter than 400 nm to a preferentially parallel FIG. 6. Schematic potential energy curves for the low lying electronic states of BrCl pertinent to this work. The figure shows an avoided crossing between the B 3 (0 ) state and a higher lying 0 state, giving respective adiabatic correlations to Br Cl and Br Cl*. but, again, nonlimiting distribution as the photodissociation wavelength increases. The Br values change rapidly with wavelength from 400 to 480 nm, but there is an inevitable contribution to the Br images at the longer wavelengths from photolysis of the Br 2 present in the gas sample. The extent to which Br 2 photolysis affects the measured Br values is discussed in Sec. IV D. An interpretation of the strong variations of parameters both with probed photofragment and with photolysis wavelength is given in Sec. IV. IV. DISCUSSION As a guide to the analysis of our experimental data, Fig. 6 shows schematic potential energy curves for the electronic states of BrCl that have, to date, been implicated in its UV and visible photochemistry. Consistent with the correlation diagram in Fig. 1, we show the B 3 (0 ) state correlating diabatically with the Cl* Br asymptote rather than the Cl Br* asymptote tentatively suggested by Brown et al., 10 but with an avoided crossing with another 0 state leading to an adiabatic correlation to Cl Br. Figure 6 is also consistent with interpretations of the potentials of other interhalogens such as IBr and ICl, for which the B 3 (0 ) states are thought to correlate diabatically with the lower in energy of the two possible 2 P 3/2 2 P 1/2 asymptotes. As mentioned previously, and illustrated in Fig. 2, fits of the BrCl absorption spectrum permitted assignments 3,4 of portions of the continuous UV/visible band to excitations from the ground X 1 (0 ) state to the A 3 (1), B 3 (0 ), C 1 (1) and D(0 ) states at wavelengths from 540 to 230 nm. The discussion of our experimental results is divided into wavelength regions corresponding to likely excitation to these various states, starting with the short wavelength data. The discussion concentrates first on the measurements of Cl and Cl* anisotropies and fragmentation channels; at the end of this section we address the information contained within the Br data concerning branching ratios.

8 J. Chem. Phys., Vol. 109, No. 11, 15 September 1998 Cooper et al A. Excitation wavelengths around 235 nm and 260 nm Based on the analysis of the BrCl absorption spectrum shown in Fig. 2, we might anticipate that photodissociation at wavelengths near 235 nm and 260 nm follows excitation to the D(0 ) state since the band in the absorption spectrum centered at cm nm has been assigned to a 0 X 1 (0 ) transition. 4 That being so, we would thus expect photofragments to show a limiting anisotropy characteristic of excitation on a parallel band i.e., 2. Such a value of is indeed observed for the production of Cl atoms which, as discussed in Sec. III B, are formed in coincidence with Br* but the open channel corresponding to formation of Cl* together with Br atoms shows an angular distribution characteristic of a perpendicular excitation ( Cl * 1). Very similar behavior was recently reported in a one-color ion-imaging study of the 235-nm photodissociation of ICl, for which only channels leading to I Cl* and I* Cl were observed, with these two pathways resulting from, respectively, perpendicular and parallel excitations. 19 From comparisons with calculated potential curves for the low-lying excited states of ICl, 18,33 the perpendicular excitation leading to I Cl* products was attributed to transition to a 1 (1) excited state which correlates adiabatically with the observed products. The upper state of the parallel excitation producing I* Cl products must have 0 symmetry, but could be ascribed to the 2 3 (0 ) state thought to make at most only a minor contribution or, more likely, the 1 3 (0 ) state. Our data for BrCl photodissociation show many parallels with the studies of ICl at similar wavelengths, and demonstrate that the BrCl absorption in the wavelength range nm cannot be solely due to absorption to a 0 state, but rather that it must also include a perpendicular component to a state with 1. Excitation to the C 1 (1) state can be discounted since it correlates with Cl Br products but a likely candidate is the 1 3 (1) state. The relative contributions of parallel and perpendicular excitations to the absorption spectrum cannot be quantified from our measurements since no effort was made to calibrate the magnitudes of the signals for Cl and Cl* formation against one another, but the relative weakness of the Cl* signal as compared with the Cl signal indicates that the perpendicular excitation is a minor contributor to the absorption at 235 nm. If, as is suggested in Sec. III B, Br* is formed together with Cl via channel 2 both at 235 nm and at nm, we might expect Br * 2.0 at the longer wavelength, consistent with Cl 2.0 at 235 nm. The cause of the nonlimiting, but positive value, Br * at nm, is most likely a consequence of the parallel excitation to the D(0 ) state being overlapped by a weaker perpendicular transition also resulting in Br* photofragments. What fractions of these Br* products are formed via channel 3 or 4 cannot be determined from our data. B. Excitation wavelengths from 320 to 400 nm The strong feature in the BrCl absorption spectrum peaking at 372 nm ( cm 1 ) and extending to 300 nm has been assigned to the C 1 (1) X 1 (0 ) transition. 4 The C 1 (1) state is repulsive in the Franck-Condon region and correlates with ground state Cl Br fragments channel 1. Cao et al. 13 reported that the formation of Cl* from BrCl photodissociation falls off sharply at wavelengths shorter than about 450 nm and the yield of Cl* is only 10% of the total Cl( 2 P) formed at 390 nm, consistent with excitation to the C 1 (1) state in the Franck-Condon region followed by the majority of the dissociating flux following this potential curve. We can readily understand our observation of Cl atoms formed with Cl , independent of wavelength over the range nm, and with Br cofragments only channel 1 if we attribute their formation to excitation on the perpendicular C 1 (1) X 1 (0 ) transition. The lack of observed Br* and Cl* products in this wavelength range strongly supports the notion that the fragmentation proceeds diabatically on the 1 (1) surface, with little or no transfer of flux onto potentials correlating with upper spin-orbit components of the two atoms. C. Excitation wavelengths from 400 to 540 nm To wavelengths longer than 400 nm, two features distinguish the BrCl photodissociation dynamics from those observed in the 320 to 400 nm range: i the yield of Cl* products increases but Cl* is formed only via channel 2 in conjunction with Br and ii Cl becomes less negative, reaching a value close to zero at 450 nm that is maintained for all longer wavelengths. No Cl* fragments were observed at wavelengths longer than 530 nm, consistent with the threshold for channel 2 being 529 nm. We interpret these observations as indicators of the declining contribution to BrCl absorption via the C 1 (1) X 1 (0 ) transition and the increasing contribution of the B 3 (0 ) X 1 (0 ) and A 3 (1) X 1 (0 ) absorptions as the photodissociation wavelength increases. We consider first the declining anisotropy of the Cl atoms. The BrCl absorption spectrum in the wavelength range 400 to 450 nm is traditionally interpreted in terms of contributions from both the C 1 (1) X 1 (0 ) and B 3 (0 ) X 1 (0 ) transitions with the former dominant at shorter wavelengths. 3,4 As the photolysis wavelength is tuned from 400 to 450 nm, the Cl atom velocities are thus anticipated to show a spatial anisotropy that evolves gradually from perpendicular 1, Cl 1 to zero where perpendicular and parallel components sum to give a net isotropic distribution of photofragments and then to parallel 0, Cl 2. The first phase of this evolution is consistent with the experimental results but the model of increasing absorption to the B 3 (0 ) state alone as the wavelength increases cannot account for the fact that Cl 0to 0.2 values are observed for all wavelengths longer than 450 nm. Three different mechanisms might explain the small magnitudes of Cl, and we consider each of these in turn. The first possibility for Cl 0 is that photodissociation near threshold leads to slow recoil of photofragments, with axial speeds of similar magnitude to the tangential speeds for molecular rotation. A classical expression for the reduction in as the axial recoil approximation breaks down was derived by Oldman et al. 34 This expression depends parametrically on the molecular beam temperature and we have calcu-

9 4374 J. Chem. Phys., Vol. 109, No. 11, 15 September 1998 Cooper et al. lated the factors by which parameters for all channels from BrCl photolysis will be reduced at a range of temperatures up to 100 K. Even at the warmest beam temperatures considered, the calculated reductions in photofragment anisotropies cannot account for the small values of Cl measured experimentally at wavelengths longer than 450 nm. Alternatively, anisotropy parameters substantially reduced from their limiting values can arise if the upper state in the transition has a lifetime that is long in comparison with the rotational period of the molecule, and indeed values of less than the limit expected for prompt dissociation can be used as a rotational clock to estimate predissociative lifetimes. 35 The threshold for channel 1 lies at cm 1 ( 555 nm) and excitation at 450 nm ( cm 1 ) gives an excess energy above this threshold of 4198 cm 1 which exceeds the threshold for channel 3 and is close to the threshold for channel 4. If we postulate that the low Clatom anisotropies for nm result from excitation to a bound state with lifetime longer than the molecular rotational period, then the hypothetical excited, bound state should correlate asymptotically with Br* Cl* products and must be predissociated by a state correlating to Br Cl products. Such a correlation would be consistent with the observation that the Cl values start to fall from near-zero toward 1 at wavelengths shorter than 450 nm, since the excitation energy then starts to exceed the threshold for channel 4. Some apparent contradictions, however, lead us to reject this mechanism. The first objection is that excitation to this postulated state above the channel 4 asymptote should form measurable quantities of Br*, yet Cl* and also Cl images at 450 nm show only Br cofragments. The second objection is that a bound state would be expected to give vibrational structure in the absorption spectrum but no such structure has been reported. It is conceivable, however, that the structure is washed out by a combination of dense rovibronic spectral features, line-broadening from rapid predissociation and isotope shifts for the four possible common isotopomers of BrCl. Finally, we note that predissociation of a long-lived state does not necessarily lead to near-zero anisotropy parameters: in the high-j limit, states accessed respectively by a perpendicular or parallel excitation should show anisotropy parameters 0.25 and At low J, these parameters can take more extreme values, but, for simultaneous excitations on P, Q, and R branches the parameters rapidly approach the high-j limit. 37 Thus we might expect that for predissociation of BrCl on a timescale longer than the rotational period, values of Cl 0 would not be observed. The full analysis is, however, complicated by the effects of hyperfine coupling caused by the nonzero nuclear spins of Cl and Br, which could be significant at low J in reducing Cl. The timescale for such hyperfine effects is likely to be hundreds of picoseconds or longer, and states with such long lifetimes should be evident using LIF or dispersed emission from higher-lying states, yet no such states have been reported. We thus propose that the mechanism for the origin of Cl values close to zero is a mixed perpendicular and parallel excitation to the A 3 (1) or C 1 (1) and B 3 (0 ) states, respectively. As is shown below, such a mechanism can simultaneously account for the Cl* branching fraction and the nonlimiting values of Cl * observed at these wavelengths. The rapid increase in the yield of Cl* that occurs as the photolysis wavelength is increased from 400 nm, and which, according to Cao et al., 13 peaks at about 60% of the total Cl( 2 P) yield at nm, was attributed by these authors to extensive nonadiabatic transitions following B 3 (0 ) state excitation. This picture was based on the speculation by Brown et al. 10 that the diabatic dissociation limit of the B 3 (0 ) state might be the Br* Cl asymptote rather than the Br Cl* asymptote. An avoided crossing with another, repulsive, 0 state, however, leads to a suggested adiabatic correlation of the B 3 (0 ) state with the Br Cl products of channel 1. Thus, in the picture of Cao et al., substantial flux must transfer from the adiabatic B 3 (0 ) state en route to Br Cl products to give the 60% Cl* yield. We invoke a slightly different picture in which the B 3 (0 ) state correlates diabatically to Br Cl*. This revised correlation does not substantially affect the arguments we present below to account for our data, but simplifies the interpretation. One way to account for the significant formation of Cl* Br products following B 3 (0 ) state excitation above the threshold for channel 2 is if the crossing of the B 3 (0 ) state and the 0 state correlating with Br Cl is only partially or weakly avoided cf. the mixed coupling invoked for IBr 15. The simplest route to Cl* formation is then via excitation to the B 3 (0 ) state and dissociation on the diabatic potential, in which case the Br Cl* products should exhibit a positive anisotropy parameter as we observe in our experiments. The anisotropy parameter does not, however, take the limiting value of 2 at any wavelength, so there must be a perpendicular channel also leading to Br Cl* products. The primary excitation for BrCl at wavelengths longer than about 440 nm is traditionally attributed to absorption on the B 3 (0 ) X 1 (0 ) transition. The most likely candidate for an overlapping perpendicular transition at wavelengths longer than 450 nm is excitation to the continuum of the A 3 (1) state, which correlates with Cl Br products. The 0 and 1 components of the 3 state will be mixed by molecular rotation through the operation of the S-uncoupling operator, 38 J S J S, and in principle for excitation of a near-equal mixture of the 0 and 1 components for a rotating molecule, the photofragments will show reduced anisotropy parameters very much as we report here for Cl and Cl*. For rotating molecules the correlations shown in Fig. 1 become inappropriate because of mixing of different components of the same multiplet electronic state. We thus use the term adiabatic to describe the state correlations of Fig. 1 in the absence of molecular rotation. The extent of the rotational mixing depends on the molecular beam temperature: for the halogen-rich mixtures used in the experiments reported here, with beam temperatures that we estimate could be as high as K, it might be significant, and will be greatest at short internuclear separations both because the magnitude of the coupling depends on the rotational constant, and because the potentials are neardegenerate. To illustrate how the excitation to 0 and 1 com-

10 J. Chem. Phys., Vol. 109, No. 11, 15 September 1998 Cooper et al ponents can give the observed anisotropy parameters, and to show how they are consistent with Cl* branching-fraction data, we approximate the molecular absorption at 450 nm as being to individual B 3 (0 ) and A 3 (1) states, and treat the mixing as flux transfer between the two states. In addition, we consider the role of higher-lying states that mix with these states and correlate to the Br Cl* asymptote. Two limiting-case models are formulated that illustrate the significance of nonadiabatic transitions during the dissociation. 1. Model for a weakly avoided crossing To illustrate how the reduced anisotropy of both Cl and Cl* recoil can arise from the schematic potentials of Fig. 6, we first consider a simple limiting case in which the crossing between diabatic 0 states is only weakly avoided. Most of the dissociating flux on the B 3 (0 ) state will then follow the diabatic pathway to Br Cl*. The dissociation dynamics can be treated as occurring via excitation to the A 3 (1) and B 3 (0 ) states, with transfer of flux between these two potentials at short internuclear separation, followed by evolution to atomic products corresponding to channels 1 and 2 on, respectively, 1 and 0 diabatic curves. We take the relative probabilities of absorption to the A 3 (1) and B 3 (0 ) states as P 1 with 1 1 and P 0 with 0 2, and allow for possible crossings of flux between the two curves at short range by defining P 01 as the probability that excitation to the 0 state results in formation of Cl Br products; P 10 is then the probability that excitation to the 1 state forms Cl* Br. If we neglect interference between wavepackets evolving on the two coupled electronic states, we can formulate expressions for Cl, Cl * and the % yield of Cl*, Cl * / Cl Cl *, determined by Cao et al. By searching the multidimensional space spanned by the various probabilities we can estimate the values for P 01 and P 10 necessary to account for our measurements. To simplify the formulation, we assume that values of P 01 and P 10 are equal, and find that anisotropy parameters Cl 0 to 0.2 and Cl * 1.2 to 1.4 are calculated if P , P and P 10 P The calculated % yield of Cl* is then 53%, close to the measurements of Cao et al. Thus, the nonlimiting experimental values of Cl and Cl * can be readily explained in the limit of weak interaction of the two 0 states, with a rotation-induced crossing between B 3 (0 ) and A 3 (1) state potentials with probability Model for a strongly avoided crossing The detailed analysis by Tellinghuisen and co-workers of emission on the BrCl E(0 ) B 3 (0 ) transition suggests that the crossing between the B 3 (0 ) state and the higher-lying 0 state is strongly avoided. The resultant adiabatic B 3 (0 ) state correlates to the Br Cl asymptote. This model is clearly in conflict with the above discussion that supposes a weakly avoided crossing of the B 3 (0 ) and 0 states. We can adapt the model from the preceding section by allowing for transfer of flux from the A 3 (1) and B 3 (0 ) state adiabats which correlate to Br Cl to potentials denoted as and that correlate with Br Cl*, with respective probabilities P 1 and P 0. The likely assignment of state, which couples to the B 3 (0 ) state, is the 0 state shown in Fig. 6, but coupling of the A 3 (1) state to this 0 state would require rotationinduced mixing at large internuclear separation. The A 3 (1) state more probably couples directly to an 1 state correlating with the Br Cl* asymptote one such state, denoted as 1 (1) is evident in Fig. 1 and sketched in Fig. 6, just as was proposed to account for the nonlimiting anisotropy parameters observed for Br Br* atoms formed by the photodissociation of Br The precise nature of the and states is not, however, critical to the current analysis. To keep the model simple and to isolate the effects of the adiabatic correlations and possible nonadiabatic transitions, the short-range rotational coupling of the A 3 (1) and B 3 (0 ) states is neglected. A similar search of the probability values to that described above yields a range of values that can satisfy the experimental measurements of Cl, Cl * and the % yield of Cl*. We find that P 0 ranges from 0.55 to 0.65 hence P 1 1 P 0 lies between 0.35 and 0.45, P to 0.8 and P to 0.4. The large values of P 0 show that this model requires a substantial transfer of flux from the adiabatic B 3 (0 ) state to the state i.e., probably the 0 state upon fragmentation. The above models must be regarded as limiting cases, with the likely origin of reduced magnitude anisotropy parameters being a consequence of rotational coupling at small internuclear separation, an adiabatic evolution of dissociating flux through the avoided crossing giving both parallel and perpendicular components to the Br Cl asymptote, and substantial flux transfer to one or more states correlating to Br Cl*. We conclude that the estimated excitation probabilities to the B 3 (0 ) and A 3 (1) states should be comparable over the wavelength range nm since both models require that P to Such a conclusion is consistent with the proposed strong mixing at short internuclear separation of the A 3 (1) and B 3 (0 ) states by molecular rotation. While at the longer wavelength end of this range, fits to the BrCl absorption spectrum see Fig. 2 suggest a roughly equal contribution from these two states, 4 at 450 nm the fits imply that excitation to the A 3 (1) state is negligible. For wavelengths from 410 to 450 nm, the nonlimiting Cl values are, as discussed previously, a result of mixed excitation to the B 3 (0 ) and C 1 (1) state. Other 1 states will lie higher in energy than the C 1 (1) state and cannot cross it, so cannot be responsible for the observed Cl values around 450 nm. The ion-imaging data thus cast doubt on the validity of the spectral fits that imply exclusive excitation to the B 3 (0 ) at wavelengths around 450 nm. We propose that the A 3 (1) state absorption extends to shorter wavelength than is suggested by Fig. 2, and this view is supported by the expected energy separation of the different components of the 3 state. The minima of the 3 (0 ) and 3 (1) states should be separated by an energy substantially less than the (881 cm 1 ) spin-orbit splitting of Cl atoms: for Hund s case a wave functions, the energy gap is estimated 38 to be 300 cm 1. Hence, the absorption maxima of the A 3 (1) and B 3 (0 ) states should lie