JOURNAL OF CHEMICAL PHYSICS VOLUME 115, NUMBER 15 15 OCTOBER 2001 Transition state dynamics of the OH¿OH\O¿H2O reaction studied by dissociative photodetachment of H2O2À Hans-Jürgen Deyerl,a) Todd G. Clements, A. Khai Luong,b) and Robert E. Continettic) Department of Chemistry and Biochemistry, University of California, San Diego, La Jolla, California 92093-0340 ~Received 30 May 2001; accepted 27 July 2001! Photoelectron-photofragment coincidence ~PPC! spectroscopy has been used to study the 2 dissociative photodetachment of H2O2 2 and D2O2 . The observed partitioning of photoelectron and photofragment translational energies provides information on the dynamics in the transition state region of the reaction between two hydroxyl radicals: OH1OH→O( 3 P)1H2O. The data reveal vibrationally resolved product translational energy distributions for both the entrance channel OH1OH and the exit channel O( 3 P)1H2O upon photodetachment. The total translational energy distribution shows a convoluted vibrational progression consistent with antisymmetric stretch excitation of H2O in the exit channel and OH stretch in the entrance channel. The photoelectron spectra are compared to two-dimensional time-dependent wave packet dynamics simulations based on an anharmonic potential in the anion and a model collinear potential energy surface for the 2 neutral complex. The PPC spectra also yield the dissociation energies D 0 (H2O2 2 →H2O1O ) 2 2 51.1560.08 eV and D 0 (D2O2 →D2O1O )51.0560.08 eV. © 2001 American Institute of Physics. @DOI: 10.1063/1.1404148# I. INTRODUCTION The hydroxyl radical is the predominant oxidant in the troposphere, in combustion processes and in a wide range of other chemically active environments.3 The reaction between two hydroxyl radicals forming O( 3 P) and H2O is an exothermic reaction (DE r,0 K520.684 eV) 4,5 that is a minor source of H2O in hydrocarbon flames, while the reverse reaction is a chain branching process which results in flame acceleration at higher temperatures.6 The reverse reaction is also thought to be important in the mechanism of the thermal De–NOx method for removal of NOx . 7 In the context of atmospheric chemistry, the reverse reaction serves as a major source of tropospheric OH radicals.8,9 The hydroxyl radical is also known to play a vital role in stratospheric ozone chemistry through the HOx cycle.10 In addition, the reaction is of significant theoretical interest as it corresponds to the association of two open-shell radicals on a potential energy surface ~PES! governed by a complicated interplay of long range dipole–dipole and short range valence forces. Since these studies begin with the H2O2 2 anion, a brief review of the negative ion chemistry is necessary. Figure 1 shows a schematic energy diagram for the H2O2 2 /OH1OH 1e 2 system. The anionic reaction Photodetachment from hydrogen bonded anions of the type AHB2 provides a route to the exploration of the transition state region for the corresponding neutral bimolecular reactions A1HB→AH1B. Neumark and co-workers have made substantial progress in characterizing transition states by analysis of the photoelectron spectra of the bound anion precursor AHB2, where A and B have been both single atoms and molecular groups.1 Photoelectron-photofragment coincidence ~PPC! spectroscopy can extend these studies significantly by allowing characterization of the entire dissociative photodetachment ~DPD! event.2 DPD occurs when removal of an electron from a stable anion produces a neutral in a dissociative or metastable state that undergoes rapid dissociation. For stable negative ions in a nuclear configuration similar to that of the transition state of a bimolecular reaction, photodetachment can be used to access the transition state region. By combining the techniques of photofragment translational spectroscopy and photoelectron spectroscopy, the kinetic energies of the photoelectrons and photofragments can be measured in coincidence, providing a complete kinematic measure of the dynamics of DPD. In the experiments presented here, the dynamics of DPD 2 in H2O2 2 and D2O2 have been investigated, and yield insights into the dynamics of the reaction: OH1OH→O~ 3 P ! 1H2O. OH21OH→O21H2O has been the subject of a number of studies.11 The reaction is exothermic by 0.35 eV at 0 K, given the heats of formation DH f ,0 of OH2, OH, O2, and H2O. 4,12 Ion-molecule reaction studies by Lifshitz13 and Van Doren et al.14 using isotopically labeled compounds revealed very rapid isotope exchange near the statistical limit showing that at some point the oxygen atoms in the collision complex become equivalent. Lifshitz interpreted this in terms of an asymmetric double-well potential with two distinct isomers O2~H2O! ~1! a! Current address: Research Center COM, DTU, DK-2800 Kgs. Lyngby, Denmark. b! Current address: Sandia National Laboratories, P.O. Box 969, MS 9056, Livermore, CA 94551. c! Author to whom correspondence should be addressed. Electronic mail: rcontinetti@ucsd.edu 0021-9606/2001/115(15)/6931/10/$18.00 ~2! 6931 © 2001 American Institute of Physics Downloaded 29 Nov 2001 to 132.239.1.232. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp 6932 Deyerl et al. J. Chem. Phys., Vol. 115, No. 15, 15 October 2001 FIG. 1. Energetics diagram for the H2O2 2 /OH1OH system. The potential energy is shown as a function of a generic reaction coordinate. The values in italics are theoretically determined values from Ref. 31. and OH2~OH!, with O2~H2O! complex being the more The bond dissociation energy stable isomer.13 2 D 0 (H2O2 →H2O1O2) has been determined experimentally to lie within the range 1.08 eV,D 0 ,1.30 eV. 13 Johnson and co-workers studied reaction ~2! by photodissociation of O2 2 in the O2 2 ~H2O! cluster anion, observing nearly equal production of O2 and OH2. 15 There have only been a few theoretical calculations on this system,11,16 –19 which support the double-well potential energy surface with local minima best described as electrostatic ion/dipole complexes @i.e., O2~H2O! and OH2~OH!#. The most detailed study by Hrušák et al.,11 however, suggests a potential energy surface ~PES! for reaction ~2! in which the OH2~OH! complex is a very shallow minimum or rather a shoulder on the PES. Restricted one-dimensional calculations ~variation of the O–H–O distance for the transferred hydrogen atom! on the collinear anion surface were carried out by Neumark and co-workers20 and found to be consistent with the work of Hrušák et al.11 The neutral reaction ~1! has been the subject of numerous experiments. Measurements of the reaction rate over a wide range of temperatures show that the disproportionation of the hydroxyl radical exhibits non-Arhennius behavior,9,21 consistent with previous measurements for the reverse reaction.22 The reason for this non-Arhennius behavior has been debated with respect to the existence of a potential barrier along the reaction path. Wagner and Zellner suggested a barrierless reaction between two OH radicals in which the long-range attractive forces affected the temperature dependence of the reaction.9 More recent work on the kinetics of this reaction has been carried out over a wide range of temperatures and pressures by Troe and co-workers23 and by Bedjanian et al.24 The dynamics of the neutral reactions at high levels of vibrational excitation were recently reported by Pfeiffer et al.25 Using overtone excitation, they determined the vibrational and rotational state distribution of OH and OD for the reaction O( 3 P)1HOD(4 n OH). In addition, there have been a number of studies, both gas phase and in clusters, of the excited-state reaction O( 1 D)1H2O. 26 For the neutral reaction ~1!, several ab initio calculations on the potential energy surface have been reported due to its importance in the hydrogen oxygen combustion mechanism.27–29 In a multireference study of the long-range OH( 2 P)1OH( 2 P) potential, Harding showed that the four singlet surfaces correlating with ground-state OH radicals interact very strongly.30 In a recent high level ab initio analysis of the transition states on the lowest triplet H2O2 potential surface, Karkach et al.31 showed that the barrier for reaction ~1! is very small. They determined the electronic energy of the transition state for reaction ~1! to be 0.070 eV above the electronic energy of two separated hydroxyl radicals and the van der Waals minimum in the entrance channel, corresponding to a weakly bound hydroxyl radical dimer, to be 0.162 eV below it. The experiments reported here study the dynamics of reaction ~1! using PPC spectroscopy to prepare the system in the transition-state region by photodetachment of the O2~H2O! anion and subsequently determining the asymptotic partitioning of energy among the products. These experiments build on the previous transition-state spectroscopy photodetachment experiments of Arnold et al.20 In their experiment, the photoelectron spectra recorded for O2~H2O! and its deuterated analog were interpreted in terms of photodetachment of a single isomer of the anion. The photoelectron spectra were observed to change considerably upon deuteration, suggesting that motion of a hydrogen atom orthogonal to the reaction coordinate was responsible for the observed structure, confirmed by one-dimensional simulation of the photoelectron spectra based on ab initio potentials.20 The present experiments reveal vibrationally resolved total translational energy distributions in the dissociative photode2 tachment of H2O2 2 and D2O2 and the dissociation energy D 0 2 2 @ O ~H2O!→O 1H2O# . Ab initio and density functional calculations on both the anionic and neutral surfaces and twodimensional wave packet dynamics simulations of the observed photoelectron spectra on model potential energy surfaces are also presented. II. EXPERIMENT In these experiments, the DPD of H2O2 2 was studied by a coincidence measurement of the three photoproducts (electron1two neutral fragments!. The experimental method has been previously described in detail32,33 and will be only briefly reviewed here. A fast beam of mass-selected negative ions (H2O2 2 ) was intersected with a pulsed laser beam and the kinematic properties of the photoelectron and neutral fragments were measured in coincidence. The apparatus consists of a mass-selected anion source, a photoelectron spectrometer, and a photofragment translational spectrometer. High detection efficiencies and use of time and positionsensitive particle detectors for both the photoelectron and molecular fragments allow the correlated measurement of the kinetic energy of all the products. 2 H2O2 2 /D2O2 was generated by crossing a pulsed supersonic expansion of H2O/D2O vapor seeded in N2O with a 1 keV electron beam. Anions are formed in this source through secondary electron attachment and three-body stabilization,34 and cooled in the expansion to a vibrational temperature of '450 K.35 The anions passed through a skimmer into a dif- Downloaded 29 Nov 2001 to 132.239.1.232. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp J. Chem. Phys., Vol. 115, No. 15, 15 October 2001 ferentially pumped chamber where they were accelerated to beam energies of 2.5 or 4 keV. Two beam energies were used to rule out formation of any long-lived ~ms! neutral complexes. After acceleration, the ion beam was referenced to ground potential using a pulsed high voltage switch and the anions were separated according to mass by time of flight ~TOF!. Fast neutral particles were removed by guiding the ion beam over a beam block situated on the beam axis. 2 Anions at m/e534 (H2O2 2 ) or m/e536 (D2O2 ) were intersected at a right angle by the third harmonic ~258 nm! of a pulsed Ti:Sapphire fundamental ~1.2 ps FWHM! generated by a CPA2000 regenerative amplifier ~Clark-MXR, Inc.!. The laser was focused to a 0.5 mm diameter spot at the interaction region to give a fluence of '5–10 mJ/cm2 per pulse. The E vector of the linearly polarized output was parallel to the direction of the ion beam in all experiments. Photodetached electrons were detected by one of two timeand position-sensitive wedge-and-strip-anode electron detectors placed opposite each other and perpendicular to the ion beam. Measurement of the electron recoil angle in this experiment is essential to allow correction for both the Doppler broadening due to the fast ion beam and the actual flight path in the large-solid-angle detector. With these corrections, the electron kinetic energy in the center-of-mass ~CM! frame ~eKE! was determined. The photoelectron spectrometer has been shown to have an effective angular acceptance of ;20% of 4 p sr, with a resolution in eKE of ;5% DE/E at 1.3 eV.33 Residual ions remaining in the beam after the interaction region were electrostatically deflected out of the beam path to an ion detector, providing the TOF mass spectrum of the negative ions and a means of monitoring the ion beam intensity. Neutral photofragments recoiled out of the beam over a 104 cm flight path and impinged on a time- and positionsensitive multi-particle detector.33 Given the parent mass and beam velocity, this information yielded the photofragment mass ratio, CM translational energy (E T ) and recoil angles. The photofragment mass resolution is not sufficient, however, to resolve O1D2O from OD1OD products. An E T resolution of ;10% DE/E at 0.7 eV has been demonstrated in studies of the photodissociation of O2 2. In these experiments, the correlated kinetic energy and angular distribution of the products are recorded directly. All events that produce one electron and two neutral photofragments were analyzed as correlated information for DPD, yielding the N(E T ,eKE) correlation spectrum. Statistics based on spectrometer efficiency and count rate of the experiment ensure that the photoelectron and photofragments from each event are correlated. Under the conditions of this experiment, a typical event rate of 0.13 per laser shot resulted in false coincidences of '2%.36 III. RESULTS A. Photoelectron-photofragment kinetic energy correlation: H2O2À and D2O2À The N(E T ,eKE) correlation spectrum at 258 nm ~4.80 2 eV! for H2O2 2 and D2O2 are shown in Figs. 2~a! and 2~b!, respectively, as a two-dimensional histogram of the correla- Transition state dynamics of OH1OH 6933 FIG. 2. ~Color! Photoelectron-photofragment energy correlation spectrum 2 @ N(E T ,eKE) # for H2O2 2 ~a! D2O2 ~b! at 258 nm. The N(E T ,eKE) spectra are represented as two-dimensional false color histograms, with increasing intensity from blue to red. The diagonal lines marked KE MAX represent the maximum translational energy in the O1H2O1e 2 product channel, while the diagonal line marked II represents the maximum translational energy for OH1OH1e 2 . Diagonal combs showing energetically allowed H2O n 3 and OH vibrations are also shown. tion between E T along the x-axis and eKE along the y-axis. The one-dimensional N(E T ) and N~eKE! spectra shown are obtained by integrating the correlation spectra over the complementary variable, and represent the distributions measurable in a conventional noncoincidence experiment. The N~eKE! spectrum along the y-axis shows broad irregularly spaced features, which reproduce at lower resolution the published photoelectron spectra by Arnold et al. at 266 nm.20 As noted by Arnold et al., the peak positions and intensities in the photoelectron spectrum change upon deuteration. This indicates that the observed peaks are related to hydrogen motion of the neutral complex formed by photodetachment. Downloaded 29 Nov 2001 to 132.239.1.232. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp 6934 Deyerl et al. J. Chem. Phys., Vol. 115, No. 15, 15 October 2001 FIG. 3. Vibrationally resolved product translational energy @ N(E TOT) # spec2 tra for the DPD of H2O2 2 and D2O2 at 258 nm as labeled. The full lines represent the dissociation asymptotes into O1H2O ~I! and O1D2O ~I! marked as KE MAX in Figs. 2~a! and 2~b!, respectively. The dashed lines represent the dissociation asymptotes into the entrance channel OH1OH ~II! and OD1OD ~II!, respectively. Combs corresponding to the vibrational states of the products are shown as in Fig. 2. The N(E T ) spectra, which provide a measure of the photofragment repulsion, exhibit broad peaks centered near E T '0.10 eV for the H2O2 system and shift to higher kinetic energy E T '0.14 eV upon deuteration. Correlation of the broad peaks in the N~eKE! and N(E T ) spectra reveals a series of faint diagonal bands. All events that lie within a single diagonal band are characterized by a well-defined total kinetic energy (E TOT5E T 1eKE) because of energy conservation. Within a band, however, there is a range of kinetic energy partitioning between the three products ~electron and two photofragments!. The maximum translational energy, corresponding to the O( 3 P)1H2O product channel, is defined by: KE MAX5h n 2D 0 ~ H2O2 2 ! 2EA ~ O ! , ~3! where h n 54.80 eV is the laser energy, D 0 is the bond dis2 2 2 sociation energy H2O2 2 →O 1H2O or D2O2 →O 1D2O, and EA~O!51.462 eV is the electron affinity of oxygen.37 This limit represents the dissociation asymptote in the exit channel ~I! and is marked as KE MAX in Figs. 2 and 3. KE MAX is drawn at the same contour level ~15%! as for the 2 previously studied H3O2 2 and D3O2 , where the bond dissociation energy has been accurately measured,2 yielding 2 KE MAX (H2O2 2 )52.1960.08 eV and KE MAX (D2O2 ) 52.2960.08 eV at 258 nm. From Eq. ~3! we derive the 2 bond dissociation energies D 0 (H2O2 2 →O 1H2O)51.15 2 2 60.08 eV and D 0 (D2O2 →O 1D2O)51.0560.08 eV, assuming that the KE MAX limits correspond to ground-state neutral products produced from ground-state anions. The lower bond dissociation energy in the deuterated case indicates that the change in the zero point energy (DDZ PE50.10 eV) in the bound anion complex is smaller than DDZ PE in the separated O2 and H2O fragments upon 2 deuteration. This behavior is similar to the H3O2 2 /D3O2 sys2 tem where we found a DDZ PE of 0.15 eV. The dissociation asymptotes ~II! in the entrance channel leading to 2OH or 2OD are drawn 0.68 and 0.71 eV below KE MAX for H2O2 2 and D2O2 2 , respectively. This value corresponds to the exothermicity of reaction ~1!.4,5 The dominant features observed in the N(E T ,eKE) spectra are a series of four faint diagonal bands for H2O2 2 and six faint diagonal bands for D2O2 2 . To provide a point of reference, diagonal combs showing the asymmetric stretch ( n 3 ) states of the water and the vibrational states of the hydroxyl radical are shown on both figures. The simplest explanation of these features is that they correspond to dissociative photodetachment ~DPD! onto vibrationally adiabatic 2 curves as found in the H3O2 2 system. As discussed below, these adiabatic curves are located around the transition state for the neutral bimolecular reaction, and they correlate with the different vibrational states of the OH1OH products in the entrance or O1H2O products in the exit channel. Examination of the E T distribution for each correlation band shows some variation, and may represent an important observable for comparison with future dynamics calculations. This observation is consistent with the interpretation that these diagonal ridges correspond to different adiabatic curves in the transition state region correlated with the various product vibrational states, with some variation in slope relative to the asymptotic product channels. Thus in this half-collision experiment we directly probe the vibrational thresholds for the bimolecular reaction by measuring the photoelectrons and photofragments in coincidence. B. Total kinetic energy release spectra of H2O2À and D2O2À at 258 nm Another way to view the correlated N(E T ,eKE) data is by direct examination of the total translational energy spectra N(E TOT) generated by summing the photoelectron kinetic energy and the translational energy release for each event: 2 E TOT5E T 1eKE. The N(E TOT) spectra for H2O2 2 and D2O2 are shown in Fig. 3, and since evidence for diagonal structure was observed in the N(E T ,eKE) data, these spectra display more resolved structure than the N~eKE! photoelectron spectra alone. The dissociation asymptotes for the exit channels O( 3 P)1H2O or O( 3 P)1D2O ~I! which correspond to KE MAX and the entrance channels OH1OH or OD1OD ~II! are marked by the vertical lines. In this spectrum, the diagonal bands in the N(E T ,eKE) spectra appear as a resolved spectrum of the correlated product vibrational distribution. A series of combs corresponding to the vibrational states of the products, as in Fig. 2, show a good correlation with the observed features. Examination of the offset of the vibrational peaks from the assigned limits KE MAX52.19 eV for H2O2 2 and KE MAX52.29 eV for D2O2 2 in this spectra give a sum for the rotational energy and bending excitation in the lowest observed state of the H2O products of '160 meV and '200 meV in D2O. If the experiment had higher total translational energy resolution, detailed product state distributions could be extracted from the N(E TOT) spectra. In the D2O2 2 spectrum, Downloaded 29 Nov 2001 to 132.239.1.232. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp J. Chem. Phys., Vol. 115, No. 15, 15 October 2001 two vibrational features are visible in the energy range where only O1D2O products are accessible, which we assign to the asymmetric stretch of the D2O product based on the ab initio calculations discussed below. In the H2O2 2 spectrum, the first asymmetric stretch of H2O falls at nearly the same energy as the 2OH( v 50) threshold, leading to the large peak in the N(E TOT) spectrum at 1.46 eV. It is difficult to extract more detailed product state distributions with the limited resolution observed here. C. Electronic structure calculations To aid in the interpretation of these experiments, density functional theory ~DFT! and ab initio calculations were performed using the GAUSSIAN 98 program package38 for the anion and neutral complexes involved in these experiments. Complete-active-space self-consistent-field ~CASSCF! calculations39 for the anion complexes were also carried out to characterize the anion potential with the DALTON program package.40 1. Anion calculations For the anion reaction ~2!, two electronic potential energy surfaces ( 2 A 8 and 2 A 9 ! connect the OH( 2 P ı ) and OH2( 1 S 1 ) reactants to the O2( 2 P) and H2O( 1 A 1 ) products. The CASSCF calculations were performed using the augmented correlation-consistent polarized-valence-doublezeta ~aug-cc-pVDZ! basis set, with the lowest two doubly occupied molecular orbitals ~MOs! kept frozen and an active space of 12 MOs used to allocate 15 electrons. The imposed orbital occupation for the 2 A 9 state is ...(6a 8 ) 2 (1a 9 ) 2 (7a 8 ) 2 (8a 8 ) 2 (2a 9 ) 1 and for the 2 A 8 state is ...(6a 8 ) 2 (1a 9 ) 2 (7a 8 ) 2 (8a 8 ) 1 (2a 9 ) 2 . The CASSCF calculations on the O2~H2O! system considering both electronic states reveal four minima and two connecting transition states very close in energy as detailed in Table I. In agreement with the most detailed theoretical studies of this system11,20 the 2 A 9 O2~H2O! complex was found to be the most stable with a very flat potential energy surface in the proton transfer coordinate between O2 and OH2. From an energetic point of view, since dynamical correlation was not included, the calculated energy difference between these states is probably not meaningful, and it is likely that both electronic states are populated in the ion beam. DFT calculations by Hrušák et al. mentioned above, suggested a very anharmonic single well potential with a shoulder.11 DFT calculations with the standard Becke3LYP density functional41 were carried out in order to map the potential in the proton transfer coordinate used in the wave packet dynamics simulations discussed below. These results, for both the anion and the neutral, are given in Table II. Structures of the calculated stationary points for both the anion and the neutral are shown in Fig. 4. 2. Neutral calculations In reaction ~1! the interaction of two ground state OH( 2 P i ) radicals leads to four singlet and four triplet states.42 The hydroxyl radical disproportionation ~1! can occur adiabatically on three of the four triplet surfaces along a Transition state dynamics of OH1OH 6935 TABLE I. Optimized geometries ~bond lengths in Å, angles in degrees!, electronic energies ~Hartrees!, zero point energies ~Hartrees per particle!, and total energies (electronic energy1zero point energy) of the stationary points along the reaction path for the OH1OH2→O21H2O reaction in the anion @CASSCF~15/12!aug-ccpVDZ#. 2 A9 Oa – Ha Ha – Ob Ob – Hb /Oa Ha Ob /Ha Ob Hb electronic energy zero point energy total energy O2~H2O! @O–H–OH#2 –TS OH2~OH! 1.655 1.031 0.967 174.7 100.0 2151.011 202 0.022 469 2150.988 733 1.21 1.273 0.967 179.3 103.0 2151.006 791 0.016 955 2150.989 836 1.064 1.51 0.972 177.7 107.0 2151.007 802 0.019 614 2150.988 188 1.682 1.028 0.967 175.4 100.6 2151.009 879 0.022 594 2150.987 285 1.213 1.281 0.970 179.0 104.2 2151.004 717 0.016 940 2150.987 777 1.063 1.540 0.972 177.8 110.0 2151.006 025 0.019 721 2150.986 304 2 A8 Oa – Ha Ha – Ob Ob – Hb /Oa Ha Ob s Ha Ob Hb electronic energy zero point energy total energy planar C s reaction path. Based on the CASSCF calculations of the O2~H2O! anion, one-electron photodetachment from either of the low-lying 2 A 8 and 2 A 9 electronic states of the anion can form the 3 A 8 and 3 A 9 states separated by ;2.4 kcal/mol, with the 3 A 9 state lower in energy.42 Calculations using DFT were carried out but are not reliable due to problems with the energetics using the B3LYP functional and the 6-31111G(3d f ,2p) basis set. Instead, we used the energetics obtained from QCISD~T!/6-31111G(3d f ,2p) single point calculations43 for scaling the barrier height on the semi-empirical LEPS surface used in the time-dependent wave packet calculations discussed below. These results are collected in Table III, with the resulting calculated energetics presented in Table IV. The energetics in Table IV were computed using the QCISD electronic energies in conjunction with the zero-point energy corrections from the DFT calculations in Table II, and show generally good agreement with experimental values. More extensive quantum chemical calculations on this difficult system are beyond the scope of the present study. One important result of these and earlier calculations on this system is that the bond length between Ob and Hb ~the terminal hydrogen atom! is very close to that in free H2O, suggesting that excitation of the asymmetric stretch in the water product dominates. The transition state on the neutral surface is also found to be more bent (/Oa Ha Ob 5141° at the transition state as opposed to 178° in the O2~H2O! well at the DFT level of theory!, suggesting that bending excitation in the dynamics on the neutral surface may be important. D. Time-dependent wave packet simulations Arnold et al. analyzed the photoelectron spectra obtained for this system in terms of a one-dimensional Franck– Downloaded 29 Nov 2001 to 132.239.1.232. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp 6936 Deyerl et al. J. Chem. Phys., Vol. 115, No. 15, 15 October 2001 TABLE II. DFT/B3LYP optimized geometries ~units as in Table I!, vibrational frequencies ~cm21!, electronic energies ~Hartrees!, and zero point energies ~ZPE! ~Hartrees per particle! of the stationary points in the neutral and anion. Species ~state! OH( 2 P) OH2( 1 S) H2O( 1 A 1 ) O2~H2O)( 2 A 9 ) OH•~OH!( 3 A 9 ) van der Waals complex @ OHOH# ‡ ( 3 A 9 ) transition state Parameter O–H frequency ZPE electronic energy O–H frequency ZPE electronic energy O–H H–O–H frequencies ZPE electronic energy O a – Ha Ha – Ob Ob – Hb ,Oa Ha Ob ,Ha Ob Hb frequencies ZPE electronic energy O a – Ha Ha – Ob Ob – Hb /Oa Ha Ob /Ha Ob Hb frequencies ZPE electronic energy O a – Ha Ha – Ob Ob – Hb /Oa Ha Ob /Ha Ob Hb frequencies ZPE electronic energy Condon analysis and assigned the observed features to the O–H–O antisymmetric stretch motion of the transferred hydrogen in the dissociative @O–H–OH# complex.20 In the present study, we extend this treatment to a two-dimensional This work ~B3LYP/ 6-31111G(3d f ,2p)! 0.9739 3722 0.008 480 275.765 639 9 0.964 3766 0.008 580 275.830 312 3 0.961 105.2 1627, 3821, 3925 0.021 357 276.463 374 5 1.460 1.077 0.950 177.7 103.5 338, 560, 1164, 1550, 3864 0.021 316 2151.655 151 7 0.976 2.033 0.973 177.0 175.9 56, 138, 183, 423, 3692, 3734 0.018 739 2151.535 966 8 1.060 1.310 0.969 140.9 110.3 i795, 343, 432, 749, 2054, 3790 0.016 785 2151.537 101 5 Arnold et al. ~Ref. 20! 1.428 1.083 0.964 176.3 102.8 1.16 1.22 0.96 153.3 106.9 ¯ time-dependent wave packet simulation of the photoelectron spectra. Given an initial anion wave function and the potential energy surface for the neutral bimolecular reaction, one can simulate the photoelectron spectrum and compare the TABLE III. Electronic energies ~Hartrees! found in ~U!QCISD~T! calculations of the anion and neutral complexes. FIG. 4. Calculated @ B3LYP/6-31111G(d,p) # stationary points for the neutral OH1OH system and the CASSCF stationary points for H2O2 2. Structural parameters are given in Tables I and II. Structure UQCISD~T!/ 6-31111G(3d f ,2p) O( 3 P) O2( 2 P) OH( 2 P i ) OH2( 1 S g ) H2O( 1 A 1 ) OH21OH ( 2 A 9 ) supermolecule O21H2O ( 2 A 9 ) supermolecule O2~H2O! ( 2 A 9 ) complex OH1OH ( 3 A 9 ) supermolecule O1H2O ( 3 A 9 ) supermolecule OH•~OH! ( 3 A 9 ) vdW complex @ OHOH# ‡ ( 3 A 9 ) transition state 274.994 222 6 275.039 779 5 275.659 757 2 275.719 547 7 276.355 346 2 2151.379 198 5 2151.395 041 6 2151.435 635 9 2151.319 514 9 2151.349 568 8 2151.324 951 4 2151.315 270 5 Downloaded 29 Nov 2001 to 132.239.1.232. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp J. Chem. Phys., Vol. 115, No. 15, 15 October 2001 TABLE IV. Electron affinities ~EA!, reaction energies (DE r ), barrier height (E e ), association energy (E as), and the anion dissociation energy (D 0 ), all in eV. Calculated from the QCISD~T! electronic energies listed in Table III and the zero point corrections listed in Table II. EA~O2!~eV! EA~OH2!~eV! DE r ~eV! OH21OH→O21H2O D 0 ~eV! O2~H2O!→O21H2O DE r ~eV! OH1OH→O1H2O E as ~eV! OH1OH→OH•~OH! E e (eV) Calc. ~this work! Expt. Previous calc. ~Ref.! 1.24 1.62 20.32 1.11 20.70 0.15 0.12 1.462a 1.825b 20.353c 1.15d 20.684e — — 1.40 ~G2!f 1.87 ~G2!f 0.37 ~B3LYP!f 1.15 ~B3LYP!f — 0.162g 0.007g a Reference 37. Reference 12. c Reference 5. d This work. e Reference 4. f Reference 11. g Reference 31 ~electronic energy!. b results to experiment.44,45 This approach has been taken on a number of the other transition-state systems experimentally 46 studied by Neumark and co-workers, including FH2 2, 2 47 2 48 2 49 OHCl , OHF , and H3O , providing important insights into the properties of the potential energy surface in the vicinity of the transition state for the neutral bimolecular reaction. Simulation of the photoelectron spectra using the timedependent wave packet propagation technique on model potential energy surfaces can provide a qualitative understanding of the region of the potential energy surface probed by these experiments. These simulations use the dynamical wave packet formalism developed by Heller50 in concert with the wave packet propagation algorithm of Kosloff and Kosloff51 as implemented by Bradforth et al.44 Upon photodetachment, the anion wave function is projected onto the neutral potential energy surface, giving the starting point for the propagation. The initial wave packet uf~0!& propagates as u f ~ t ! & 5exp~ 2iĤt/\ ! u f ~ 0 ! & , ~4! where exp(2iĤt/\) is the time evolution operator and Ĥ is the nuclear Hamiltonian for the neutral surface. The time autocorrelation function C(t) is defined by the overlap of u f (t) & and uf~0!&: C~ t !5 ^ f~ 0 !u f~ t !& 6937 Transition state dynamics of OH1OH ~5! TABLE V. LEPS parameters for the collinear neutral surface used in the time-dependent wave packet simulations. H2O OH O–OH De b re S 5.443 4.643 2.878 2.110 2.179 2.520 0.956 0.971 1.324 0.110 0.110 0.250 initio potential energy surface would be desired, this was beyond the scope of the present work, so the semi-empirical London, Eyring, Polanyi, Sato ~LEPS! surface was adopted.52 We consider only the lowest surface of 3 A 9 symmetry, but since the simulations are done in the collinear approximation, the electronic state is 3 P for the OH( 2 P) 1OH( 2 P)→O( 3 P)1H2O reaction. We constructed a collinear LEPS surface using the parameters listed in Table V, to account for DE e 520.81 eV for OH1OH→O1H2O and an activation barrier of E e 50.12 eV at an Oa – Ha – Ob distance of 2.37 Å and a Ha Ob distance of 1.31 Å consistent with the DFT and QCISD calculations on the neutral surface. The initial anion wave packet was generated using an anharmonic Morse function for the n 3 asymmetric stretch @ r e (Ha – Ob )51.08 Å, v e 51810 cm21, v e x e 5350 cm21# found by fitting the DFT/B3LYP energies generated in a calculation at a fixed equilibrium Oa – Ob distance of 2.54 Å. A harmonic function was used for the symmetric stretch n 1 , @ r e (Oa – Ob )52.54 Å, v e 5340 cm21#, also chosen to be consistent with the DFT calculations of the anion potential energy surface discussed above. Simulations were done for both the ground-state anion and also anions with one quantum in n 3 and n 1 , respectively, as illustrated in Fig. 5. The propagations were carried out for 300 fs and were checked for convergence by varying the time-step and grid size.53 As seen in the simulations for 2 H2O2 2 and D2O2 shown in Figs. 6~a! and 6~b!, the simulations from the ground state of the anion reasonably reproduce the number and spacing of the observed broad features in the photoelectron spectra in both the O1H2O and OH1OH channels. The theoretical simulations show little broadening except for the highest eKE features. In the figure, the simulations have been convoluted with the experimental resolution for the photoelectron spectrometer. These results are qualitatively consistent with the one-dimensional ( n 3 ) simulations carried out by Arnold et al. for both systems, and the assignment of the dominant features to specific n 3 levels and the Fourier transform of the autocorrelation function yields the photoelectron spectrum: s~ E !} E ` exp~ iEt/\ ! C ~ t ! dt. ~6! 2` In future efforts, it will be of interest to extend these calculations to include the product translational and internal state distributions that the coincidence experiments reported here provide, as previously illustrated for the OHF system.48 These simulations are restricted to two dimensions, the asymmetric and symmetric stretching motions of the O–H–O moiety, and assume that all nuclear motion occurs on a collinear potential energy surface. While a complete ab FIG. 5. Initial wave packets ~lowest contour contains 90% of the norm of the wave function! on the collinear LEPS potential energy surface ~contours at 0.1 eV intervals! for OH1OH 3A 9 for ~a! O–H–OH2 ~n 1 50, n 3 50!; ~b! O–H–OH2 ~n 1 50, n 3 51! and ~c! O–H–OH2 ~n 1 51, n 3 50!. The massweighted coordinates are x'3 @ r(Oa – Ob ) # and y5r(Ha – Ob ) in Å, following Bradforth et al. ~Ref. 44!. Downloaded 29 Nov 2001 to 132.239.1.232. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp 6938 Deyerl et al. J. Chem. Phys., Vol. 115, No. 15, 15 October 2001 vides a promising system for demonstrating vibrational state control of product branching ratios. IV. DISCUSSION 2 FIG. 6. Simulated photoelectron spectra for ~a! H2O2 2 and ~b! D2O2 . The 2 top simulation shown in ~a! is for H2O2 ~n 1 50, n 3 50!. The middle simulation shows the results for ~n 1 51, n 3 50!, illustrating the preference n 1 excitation in the anion exhibits for the O1H2O product channel. The bottom simulation shows the results for ~n 1 50, n 3 51!, illustrating the propensity for n 3 excitation to produce OH1OH reactants. The simulation in ~b! is for D2O2 2 ~n 1 50, n 3 50!. on the neutral surface is discussed in their work.20 Since these time-dependent simulations explicitly include the dissociative reaction coordinate they might be expected to reproduce the widths of the experimental features, however, they are found to be much narrower than observed experimentally. Simulations for the vibrationally excited anions are also shown for H2O2 2 in Fig. 6~a!, illustrating how excitation of one quantum in n 1 couples strongly with the O1H2O product channel, while one quantum in n 3 predominantly yields the OH1OH entrance channel. This control of product channel by parent vibrational excitation is similar to that previously predicted by Bradforth et al. for the Br1HI system,44 and shows that dissociative photodetachment of H2O2 2 pro- The data presented here provide detailed experimental insights into the half-collision dynamics in the dissociative 2 photodetachment ~DPD! of H2O2 2 and D2O2 . Figure 5 shows the mass-weighted two-dimensional potential energy surfaces used to model the neutral surface. The projection of the anion wave functions shown in this figure illustrate that these experiments directly probe the region of the PES near the transition state for the OH1OH→O1H2O reaction. The observation of diagonal structure in the N(E T ,eKE) spectra and the corresponding vibrationally resolved N(E TOT) spectra provide evidence for significant vibrational adiabaticity in this system, wherein excitation induced by photodetachment is carried over into the final product states. The total kinetic energy spectra show more structure than the photoelectron spectra in both cases. The photoelectron spectrum @ N(eKE) # is determined by the Franck– Condon overlap between the bound anion wave function and the scattering wave function on the neutral surface. A steeply sloped repulsive surface will yield broad features in the photoelectron spectrum due to lifetime broadening. The N(E T ) spectrum is similarly governed by the region of Franck– Condon overlap with the neutral surface, but also by any subsequent transfer of energy from internal to external degrees of freedom in the dissociation process. The total kinetic energy release E TOT5E T 1eKE for any given event, however, is determined solely by energy conservation. Thus for a given DPD event the partitioning of the total available energy provides a measure of the resulting product internal state distribution after the DPD process. In the case where the dissociation dynamics of the neutral H2O2 system are vibrationally adiabatic, as revealed by the diagonal structure observed in the PPC spectrum shown in Fig. 2, each peak in the photoelectron spectrum correlates experimentally to a product vibrational level in the exit or entrance channel. The degree of vibrational adiabaticity observed in this hydrogen exchange reaction can be understood in terms of the potential energy surface in mass-weighted coordinates, as discussed previously by Neumark and co-workers for a number of systems.1,44 As discussed above, the ab initio calculations for the anion potential show that the O–H–O atoms are almost linearly arranged, with a spectator H atom attached to one of the O atoms. To a first approximation the outer OH group in the O–H–OH system can be seen as a spectator. For triatomic linear molecules, the skew angle on the massweighted potential energy surface is given by tan21@(mb(ma 1mb1mc))/(mamc)#1/2. Using m b 51 ~H! and m a 516 ~O! and m c 517 ~OH! the resulting skew angle is 19.5° while for O–D–~OD! it is found to be 26.6° in a linear configuration of the oxygen atoms and the transferred hydrogen atom. Thus the asymmetric stretch coordinate is nearly orthogonal to motion along the reaction coordinate and is additionally stabilized by an inward force from the potential barrier between the reactant and product valleys. These factors result in a very weak coupling between the asymmetric stretch and Downloaded 29 Nov 2001 to 132.239.1.232. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp J. Chem. Phys., Vol. 115, No. 15, 15 October 2001 the dissociation coordinate giving rise to the structure dominated by motion of the transferred H-atom, and a strong correlation between the asymmetric stretch states and the reactant and product vibrational states. The two-dimensional time-dependent simulations of the photoelectron spectra presented here are chiefly of value in a qualitative sense. They are consistent with the previous onedimensional simulations of Arnold et al.,20 confirming the important role played by the asymmetric stretch states of the neutral complex. The difficulties associated with calculating potential energy surfaces for both the anion and the neutral in this system precluded comparison with more realistic potentials. In addition, the restriction to two degrees of freedom, the assumption of a collinear reactive surface and the concomitant neglect of bending and rotational corrections are all potential reasons for the prediction of narrow features in the photoelectron spectrum. Bending corrections in the timedependent simulations when using a LEPS surface44 were evaluated but made little difference in the simulations, so are not presented. Finally, as the simulations show, vibrational excitation in the anion can lead to significant changes in the photoelectron spectra and the product branching ratio. In these experiments anions are produced and cooled in a supersonic expansion, but high frequency modes like the asymmetric stretch of the anion are often characterized by temperatures '450 K, so a few percent of the anions can be vibrationally excited as well. Thus a number of inhomogeneous effects may be dominating the lifetime of the neutral states accessed by photodetachment. It would be of great interest to carry out these experiments at higher energy resolution to observe directly the homogeneous broadening of these states. The width of the N(E T ) distributions and the observation of the diagonal bands in the N(E T ,eKE) correlation spectra suggest that the lifetimes may in fact be much shorter than predicted by the time-dependent simulations carried out here. V. CONCLUSION This paper reports a PPC study of the transition-state dynamics of the OH1OH→O( 3 P)1H2O reaction using the 2 dissociative photodetachment of H2O2 2 and D2O2 . Although the product channels cannot be cleanly resolved by photofragment mass in either system in these experiments, a bond dissociation energy for the anion is determined, defining the energetics and showing that when the OH1OH1e 2 DPD channel is accessible a considerable enhancement in signal intensity is observed. Ab initio and DFT calculations were used to characterize the anion surface near the equilibrium geometry and to provide parameters for scaling a collinear O–H–OH LEPS surface for the reaction. Two-dimensional wave packet simulations give reasonable agreement with the positions and intensities of the observed photoelectron spectra and also demonstrate the strong sensitivity of the product branching ratio on vibrational excitation of the parent anion. Experimentally, higherresolution coincidence experiments making use of tunablelaser photodetachment and threshold photoelectron detection will be of interest to examine the lifetimes of the dissociating Transition state dynamics of OH1OH 6939 states. Due to the multiple interacting potential energy surfaces in both the anion and the neutral, the H2O2 2 /H2O2 system represents a significant challenge to theory, but it is hoped that the results presented here will encourage further studies of this prototypical radical–radical four atom reaction. ACKNOWLEDGMENTS This work was supported by the Department of Energy ~DOE! under Grant No. DE-FG03-98ER14879. H.J.D. gratefully acknowledges partial support from a Forschungsstipendium sponsored by the Deutsche Forschungsgemeinschaft ~DFG! and the DOE. T.G.C. and A.K.L. were supported under AFOSR Grants No. AASERT F49620-97-1-0387 and F49620-96-1-0220. R.E.C. is a Camille Dreyfus TeacherScholar, an Alfred P. Sloan Research Fellow and a Packard Fellow in Science and Engineering. We also acknowledge Professor S. E. Bradforth for providing the time-dependent wave packet code and helpful discussions. We thank the National Energy Research Scientific Computing Center ~NERSC! for providing computation time. 1 ~a! D. M. Neumark, Acc. Chem. Res. 26, 33 ~1998!; ~b! R. B. Metz, S. E. Bradforth, and D. M. Neumark, Adv. Chem. Phys. 81, 1 ~1992!. 2 H.-J. Deyerl, A. K. Luong, T. G. Clements, and R. E. Continetti, Faraday Discuss. 115, 147 ~2000!. 3 R. Atkinson, Chem. Rev. 17, 227 ~1998!. 4 B. Ruscic, D. Feller, D. A. Dixon, K. A. Peterson, L. B. Harding, R. L. Asher, and A. F. Wagner, J. Phys. Chem. A 105, 1 ~2001!. 5 M. W. Chase, C. A. Davies, J. R. Downey, D. J. Frurip, R. A. McDonald, and A. N. Syverud, J. Phys. Chem. Ref. Data Suppl. 14, 1 ~1985! ~JANAF Tables!. 6 J. Warnatz, in Combustion Chemistry, edited by W. C. Gardiner, Jr. ~Springer, New York, 1984!. 7 J. A. Miller and C. T. Bowman, Prog. Energy Combust. Sci. 15, 287 ~1989!. 8 K.-H. Gericke, F. J. Comes, and R. D. Levine, J. Chem. Phys. 74, 6106 ~1981!. 9 G. Wagner and R. Zellner, Ber. Bunsenges. Phys. Chem. 85, 1122 ~1981!. 10 World Meteorological Organization, Global Ozone Research and Monitoring Project, Report No. 16 ~1985!. 11 J. Hrušák, H. Friedrichs, H. Schwarz, H. Razafinjanahary, and H. Chermette, J. Phys. Chem. 100, 100 ~1996!. 12 P. A. Schulz, R. D. Mead, P. L. Jones, and W. C. Lineberger, J. Chem. Phys. 77, 1153 ~1982!. 13 C. Lifshitz, J. Phys. Chem. 86, 3634 ~1982!. 14 J. M. Van Doren, S. E. Barlow, C. H. DePuy, and V. M. Bierbaum, J. Am. Chem. Soc. 109, 4412 ~1987!. 15 M. A. Buntine, D. J. Lavrich, C. E. Dessent, M. G. Scarton, and M. A. Johnson, Chem. Phys. Lett. 216, 471 ~1993!. 16 C. M. Roehl, J. T. Snodgrass, CA. A. Deakyne, and M. T. Bowers, J. Chem. Phys. 94, 6546 ~1991!. 17 S. Humble, I. Demachy, and P. C. Hiberty, Chem. Phys. Lett. 247, 126 ~1995!. 18 S. J. K. Jensen and U. K. Kläning, Chem. Phys. Lett. 241, 404 ~1995!. 19 R. Benassi and F. Taddei, Chem. Phys. Lett. 204, 595 ~1993!. 20 D. W. Arnold, C. Xu, and D. M. Neumark, J. Chem. Phys. 102, 6088 ~1995!. 21 R. Zellner, J. Phys. Chem. 83, 18 ~1979!. 22 E. A. Albers, K. Hoyermann, H. G. Wagner, and J. Wolfrum, 13th Symp. ~Int.! on Combustion ~The Combustion Institute, 1971!, p. 81. 23 D. Fulle, H. F. Hamann, H. Hippler, and J. Troe, J. Chem. Phys. 105, 1001 ~1996!. 24 Y. Bedjanian, G. Le Bras, and G. Poulet, J. Phys. Chem. A 103, 7017 ~1999!. 25 J. M. Pfeiffer, E. Woods III, R. B. Metz, and F. F. Crim, J. Chem. Phys. 113, 7982 ~2000!. 26 N. Tanaka, U. Nagashima, M. Takayanagi, H. L. Kim, and I. Hanazaki, J. Downloaded 29 Nov 2001 to 132.239.1.232. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp 6940 Deyerl et al. J. Chem. Phys., Vol. 115, No. 15, 15 October 2001 Phys. Chem. A 101, 507 ~1997!; H. Tsurumaki, Y. Fujimura, and O. Kajimoto, J. Chem. Phys. 110, 7707 ~1999!; K. Imura, M. Veneziani, T. Kasai, and R. Naaman, ibid. 111, 4025 ~1999!; R. Sayos, C. Oliva, and M. Gonzalez, ibid. 113, 6736 ~2000!. 27 E. P. Dougherty and H. Rabitz, J. Chem. Phys. 72, 6571 ~1980!. 28 U. Maas and J. Warnatz, Combust. Flame 74, 53 ~1988!; U. Maas and S. B. Pope, ibid. 88, 239 ~1992!. 29 L. B. Harding and A. F. Wagner, 22nd Symp. ~Int.! on Combustion ~The Combustion Institute, 1988!, p. 983. 30 L. B. Harding, J. Phys. Chem. 95, 8653 ~1991!. 31 S. P. Karkach and V. I. Osherov, J. Chem. Phys. 110, 11918 ~1999!. 32 K. A. Hanold, C. R. Sherwood, M. C. Garner, and R. E. Continetti, Rev. Sci. Instrum. 66, 5507 ~1995!. 33 K. A. Hanold, A. K. Luong, T. G. Clements, and R. E. Continetti, Rev. Sci. Instrum. 70, 2268 ~1999!. 34 M. A. Johnson and W. C. Lineberger, in Techniques in Chemistry, edited by J. M. Farrar and W. H. Saunders ~Wiley, New York, 1988!, Vol. 20, p. 591. 35 M. C. Garner, K. A. Hanold, M. Sowa Resat, and R. E. Continetti, J. Phys. Chem. A 101, 6577 ~1997!. 36 R. E. Continetti, in Photoionization and Photodetachment, Adv. Ser. Phys. Chem. Vol. 10B, edited by C. Y. Ng ~World Scientific, Singapore, 2000!, pp. 748 – 808. 37 H. Hotop and W. C. Lineberger, J. Phys. Chem. Ref. Data 14, 731 ~1985!. 38 M. J. Frisch, G. W. Trucks, H. B. Schlegel et al., GAUSSIAN 98, Revision A.9, Gaussian, Inc., Pittsburgh, PA, 1998. 39 F. Bernardi, A. Bottini, J. J. W. McDougall, M. A. Robb, and H. B. Schlegel, Faraday Symp. Chem. Soc. 19, 137 ~1984!. 40 T. Helgaker, H. J. Aa. Jensen, P. Joergensen et al., ‘‘Dalton release 1.0 ~1997!, an electronic structure program.’’ 41 A. D. Becke, J. Chem. Phys. 98, 1372 ~1993!; ibid. 98, 5648 ~1993!; C. Lee, W. Yang, and R. G. Parr, Phys. Rev. B 37, 785 ~1988!. 42 L. Harding, J. Phys. Chem. 95, 8653 ~1991!. 43 J. A. Pople, M. Head-Gordon, and K. Raghavachari, J. Chem. Phys. 87, 5968 ~1987!. 44 S. E. Bradforth, A. Weaver, D. W. Arnold, R. B. Metz, and D. M. Neumark, J. Chem. Phys. 92, 7205 ~1990!. 45 R. B. Metz and D. M. Neumark, J. Chem. Phys. 97, 962 ~1992!. 46 D. E. Manolopoulos, K. Stark, H.-J. Werner, D. W. Arnold, S. E. Bradforth, and D. M. Neumark, Science 262, 1852 ~1993!. 47 M. J. Davis, H. Koizumi, G. C. Schatz, S. E. Bradforth, and D. M. Neumark, J. Chem. Phys. 101, 4708 ~1994!. 48 R. N. Dixon and H. Tachikawa, Mol. Phys. 97, 195 ~1999!. 49 W. H. Thompson and W. H. Miller, J. Chem. Phys. 101, 8620 ~1994!; D. Wang, J. Z. H. Zhang, and C. Yu, Chem. Phys. Lett. 273, 171 ~1997!; D. C. Clary, J. K. Gregory, M. J. T. Jordan, and E. Kauppi, J. Chem. Soc., Faraday Trans. 93, 747 ~1997!; M. Igarashi and H. Tachikawa, Int. J. Mass. Spectrom. 197, 243 ~2000!. 50 E. J. Heller, Acc. Chem. Res. 14, 368 ~1981!. 51 D. Kosloff and R. Kosloff, J. Comput. Chem. 52, 35 ~1983!. 52 J. Espinosa-Garcia, J. Phys. Chem. A 105, 134 ~2001!. 53 S. E. Bradforth, Ph.D. Thesis, University of California, Berkeley ~1992!. Downloaded 29 Nov 2001 to 132.239.1.232. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp