Research in Chemical Kinetics, Volume 1 R.G. Compton and G. Hancock (editors) © 1993 Elsevier Science Publishers B.V. All rights reserved.

1

Reaction dynamics of 0(3p), O(lD) and ΟΗ(2Π) with simple molecules

Piergiorgio Casavecchia, Nadia Balucani, and Gian Gualberto Volpi

Dipartimento di Chimica, Universita di Perugia, 06100 Perugia, Italy

Abstract The dynamics of elementary gas-phase bimolecular reactions can be probed in

detail using the crossed molecular beam scattering method with mass spectrometric detection. Exploiting the capability to generate continuous supersonic beams of very reactive species, the technique is applied to reactions of atomic oxygen, both in the ground and the first electronically excited state, and of hydroxyl radicals with simple molecules. A survey and discussion of recent experimental results from our laboratory are presented. Comparison with related experimental and theoretical work is also carried out.

1. INTRODUCTION

The dynamics of elementary gas-phase chemical reactions, intended as the microscopic study of individual collisional events, constitutes one of the major patterns of inquiry in modern chemical kinetics [1]. The field has witnessed tremendous progress during the last 30 years, made possible by the development of sophisticated microscopic experimental techniques and large-scale theoretical computations. The result is a significant enhancement of our understanding of the details of chemical reactions. In fact, molecular beams, lasers and modern spectroscopy allow us to investigate chemical reactions occurring at well defined collision energies, with reactants in selected quantum states. Recent advances in electronic structure calculations and theoretical methodology have provided potential energy surfaces to an ever higher degree of accuracy and, consequently, the results of the calculation of dynamical observables have become more reliable and are often guiding experimental efforts.

The techniques usually employed for experimental studies of the dynamics of elementary gas-phase chemical reactions fall essentially into two categories: crossed molecular beam (CMB) methods, and laser based spectroscopic methods.

2

In a typical CMB experiment, beams of atoms and molecules with narrow angular and velocity spread are crossed in a vacuum chamber and the angular and velocity distributions of the products are recorded after a single and well defined collisional event. Products are detected after they have travelled for some time from their formation zone (virtually, at time t->°°). The CMB method was introduced in the 1950s [2]. Widely applied in the 1960s [3, 4] to reactions of alkalies (the alkali era), it reached its maturity in the 1970s with the development of universal machines [5, 6], which, employing a rotatable mass spectrometer as detector, rendered the method of general applicability. In the 1980s, it reached a high level of sophistication [7-9], which was well expressed by a series of studies by Lee and co-workers and culminated in the Nobel Prize Award in 1986 to Y.T. Lee, D.R. Herschbach, and J.C. Polanyi [10]. Factors that have made the CMB method an extremely powerful tool for the investigation of reaction dynamics are the advancement in vacuum technology, the development of intense nearly monoenergetic atomic and molecular beam sources using supersonic expansion, as well as the pseudo-random chopping method for the velocity analysis of product molecules. The CMB method, in addition to the direct identification of product channels and of their relative importance, permits also the study of the effect of orientation of reactant molecules, the exploration of the lifetime of reaction complexes and of their subsequent decay dynamics, the derivation of the distribution of energy among various degrees of freedom and of the heights of both entrance and exit potential energy barriers, the identification of complex reaction mechanisms involving polyatomic radical products. While the measurement of the energy distribution of product molecules is not a unique capability of the CMB method (it can actually be better measured by laser spectroscopic techniques when applicable, as discussed below), information on the angular distribution of product molecules is usually only obtainable from CMB experiments.

As typical spectroscopic investigations of the dynamics of a chemical reaction are usually carried out in low pressure cells, nascent products are probed right after their formation (virtually, at time t->0), before they undergo collisional roto-vibrational relaxation [1, 11]. The collision free regime can still be attained even in a bulk environment using pump-probe laser methods, provided the pump-probe delay time lies well within the mean collision interval. Laser based spectroscopic methods for reaction dynamics studies were introduced in the 1970s by Zare and co-workers [12], and were literally widespread in the 1980s following the development and sophistication of laser sources and spectroscopic techniques. Studies using various linear or non-linear laser techniques, such as Laser Induced Fluorescence (LIF), [12,13], Resonant Enhanced Multiphoton Ionization (REMPI) [14, 15] and Coherent Antistokes Raman Spectroscopy (CARS) [16], provide detailed information on the quantum states of nascent reaction products. A similar type of information can often also be obtained from chemiluminescence [17] and, sometimes, from chemical laser

3

studies [18]. All these spectroscopic methods are often applicable to only a limited range of simple systems, since typically they rely on the availability of appropriate tunable lasers and on the knowledge of the spectroscopy of the product to be identified and probed. Progress is, however, continuously being made in the area of spectroscopic detection.

It has to be said that spectroscopic methods based on pump-probe laser techniques and the CMB method are, in general, complementary to each other. In the last few years, the combination of laser and beam methods has become more and more common, and has progressed in the direction of the use of lasers for the generation and/or state selection of reactants, for product detection, or a combination of both. The use of lasers for product detection in crossed beam experiments has been limited to LIF probing right at the collision center [12, 19]. Experiments in which a collision product is probed in its internal quantum states, as a function of scattering angle using a laser spectroscopic technique, have been already accomplished in a few cases in inelastic scattering studies [20]. They also appear to be feasible in reactive scattering studies, at least for simple selected systems. However, with the exception of the H + D2 —> HD + D reaction - for which reactive differential cross sections resolved in both rotational and vibrational states of the product were measured by Welge and co-workers [21] using the novel technique of Η-atom Rydberg time-of-flight spectroscopy - the ideal experiment, that is, the measurement of the reactive differential cross section for a single quantum state of the product starting from single quantum states of the reagents, has not been carried out yet. Examples of state-selected differential cross sections by LIF detection of products at the collision center and exploitation of the Doppler profile are also available [22, 23]. The novel product ion-imaging technique by REMPI, in principle capable of providing reactive [24] (and non [25]) differential cross sections for single quantum state and presently in its infancy, appears to be very promising.

A powerful complement to dynamic studies comes also from a new experimental strategy, in which molecular beam dynamics is explored in a cell. Polarized photodissociation can be used to generate superthermal "beams" of velocity aligned atomic photofragments. Doppler resolution of the polarized LIF spectra of nascent products can allow the determination of their scalar quantum state distributions and vector correlations. Critical, here, is the sub-Doppler resolution to unravel the dynamical information contained within Doppler profiles. The capability of the technique is well illustrated by recent work carried out in the laboratories of J. Simons [26] and G. Hancock [27]. However, to fully exploit the capability of the polarized sub-Doppler detection scheme, one should resort to rotationally cooled reagents, as can be obtained in molecular beams, since the work in bulk implies the limitation that molecular reactants can hardly be state selected.

The reaction dynamics of ground state atomic reagents has been by far the most extensively investigated, both by CMB and laser spectroscopic methods. During the last 10 years, that of electronically excited atoms has also been extensively

4

examined using LIF [28-34] and infrared chemiluminescence [35-37] in low pressure cells, LIF in crossed beams [34], LIF in laser-initiated half reaction within a pulsed beam [39], and also by infrared absorption [40]. Very little work has been performed under crossed beam conditions, except for a series of elegant experiments carried out in Y.T.Lee's laboratory, in which the reactivity and dynamics of electronically excited alkali and alkaline earth atoms have been investigated in detail exploiting crossed beam and laser techniques [9, 41]. With the development of continuous supersonic beams of 0(lD), it has also become possible to carry out dynamical studies of this important species in crossed beams [42]. The feasibility of these experiments was demonstrated about 10 years ago with the study of the 0(lD)+H2 and 0(lD)+CH4 reactions [43], but no further studies have been pursued since then. Studies of free radical reactions in crossed beams are limited to the pioneering work of Ross and co-workers [44] and Grice and co-workers [45] with effusive beams of CH3 and OH radicals, and, more recently, to a study in Grace's group [46] and a study in Lee's group [47] using a supersonic CH3 beam. Recently, nascent product internal state distributions of several free radical reactions have been determined in crossed pulsed beam experiments [19].

Up to now, in the field of reaction dynamics, the reactions of A + BC type have played a central role in the connection between experiment and theory. In particular, benchmark systems have been the reactions F + H2 and H + H2 and their isotopic variants. Following the considerable progress in theoretical methodology and computational capabilities, accurate comparisons between detailed experimental observables [8, 10a, 15,48] and exact quantum mechanical scattering calculations [49] have recently been carried out. The logical extension to more complex systems is that of the four-atom (diatom + diatom and atom + triatom) problem. Recently, the quantum scattering theory has also been extended to four-atom reactions [50-52] of the form AB + CD and calculations, although with still some degree of approximation, of state-selected integral and differential cross sections have become feasible. These theoretical developments have come at a propitious time because of the new experimental studies [53-69] on the reaction dynamics of four-atom systems.

The present review provides an account of some recent progress in the study of the reaction dynamics of electronically excited *D oxygen atoms and of hydroxyl radicals, which are two of the most important reactive transient species in the atmosphere and in combustion systems. The selected experiments discussed here are not a survey of the field; they have actually been carried out in the author's laboratory using the CMB scattering method with mass spectrometry detection. They are chosen as examples to illustrate the capability of the CMB method to obtain dynamical information not readily obtainable by other approaches, and to emphasize the complementarity of the CMB method with spectroscopic methods, showing how a more complete view and understanding of the mechanism and

5

dynamics of a chemical reaction can be reached when the information derived from the two approaches are made to coalesce. Moreover, the reactive scattering of four-atom systems is tackled with the investigation of those reactions for which theory has recently provided both ab initio potential energy surfaces [70 - 72] and classical [73-80] and quantum [50-52, 81-84] dynamical calculations, namely OH + H2 and OH + CO. The review is structured as follows. In Section 2 we introduce the experimental technique and discuss the generation of continuous supersonic beams of 0(3p, ID) and OH(2n) radicals. In Section 3 we outline the results of the reactions of O(^D) with hydrogen halides, while in Section 4 we explore the effect of electronic excitation on the reaction dynamics of atomic oxygen using the 0(3p, ID) + H2S system as an example. The reaction dynamics of OH + D2 and OH + CO are then examined in Section 5. A point to future developments and opportunities will be finally made in Section 6.

2. EXPERIMENTAL ARRANGEMENT

Figure 1 shows a schematic diagram (a top view) of our apparatus. Two supersonic beams of reagents are crossed in a region with background pressure in the 10'7 mbar range. A mass spectrometer, contained in an ultra-high-vacuum chamber, serves as detector of the reaction products. The mass spectrometer can be rotated in the collision plane around an axis passing through the collision center and the velocities of the particles can be derived from time-of-flight (TOF) measurements using a pseudo-random chopping disk. In the present experiments a selection of the velocities, as well as internal quantum states of the reactants by supersonic expansion, is achieved. The selection of the translational energy of the products also gives (by energy conservation) their internal (rotational + vibrational + electronic) energy. For the physical interpretation of the scattering process it is necessary to transform the angular and velocity distributions measured in the laboratory coordinate system to the center-of-mass (cm.) coordinate system [85]. This transformation is fairly straightforward:

Ilab(©>v)= I c.m.(^u)v2/U2

(where Θ and ν are lab angle and velocity, respectively, and û and u are the corresponding cm. quantities), i.e., the scattering intensity observed in the laboratory is distorted by the transformation Jacobian v^/u^ from that in the cm. system. Since an electron impact ionization mass spectrometric detector measures the number density of products, Ν(Θ), and not their flux, the actual relation between the lab density and the cm. flux is given by:

Nlab(©,v) = I c.m.(^u)v/u2.

6

Because of the finite resolution of experimental conditions, i.e. angular and velocity spread of the reactant beams and angular resolution of the detector, analysis of the laboratory data is usually carried out by forward convolution procedures over the experimental conditions of trial cm. distributions. The final outcome of a reactive scattering experiment is the generation of velocity flux contour maps of the reaction products, i.e., the plot of the intensity as a function of angle and velocity in the cm. system, Ic.m.C$> u) ( s ee Sections 3 and 4). This can be regarded as an image of the reaction.

ZJ L_ Ο σ CO ZD Ο CD Ο

Β

Figure 1. Top view (schematic) of the crossed molecular beam apparatus: (A) main scattering chamber; (B) beam source chambers; (C) beam differential pumping chambers; (D) triply differentially pumped rotatable electron-impact ionization quadrupole mass spectrometer detector; (E) pseudorandom time-of-flight disk.

7

The apparatus, which has been optimized for reactive scattering experiments, is the same previously used for measuring elastic and total (elastic + inelastic) differential cross sections for atom-atom and atom-molecule collisions to derive intermolecular potentials [86]. The basic design follows that of Lee et al [5]. The main scattering chamber is a welded aluminum alloy box with inside dimensions 110x110x57 cm, with a rotatable lid, supported by a 76.2 cm i.d. ball bearing, which carries the detector unit and its pumping system. The main chamber is pumped by a large two-stages (77 and 14 K) cryopump and two baffled diffusion pumps for a total pumping speed of about 11000 1/s. This set-up allows to keep a clean vacuum in the main chamber down to 3x10"7 mbar in operating conditions. The detector unit is made of 304L stainless steel and consists of three different pumping stages in three nested ultra-high-vacuum (UHV) chambers. The first two regions are pumped by a 220 1/s ion pump each, while the innermost region is pumped by a 80 1/s ion pump, a 170 1/s turbo-pump and a 400 1/s two-stages (77 and 13 K) cryopump. In addition, both the second and third regions are liquid nitrogen cooled. The detection system consists of an electron bombardment ionizer, a quadrupole mass filter and an off-axis secondary electron multiplier. The ionizer is located in the innermost region of the detector maintained in the 10"^ mbar pressure range. The three detector regions are all equipped with interchangeable slits, which allows us to vary easily the angular resolution of the detector. The detector unit can be closed off from the main chamber by means of a slide valve, sealed with a Viton O-ring, operated from the outside. The beam sources are placed in separately pumped side chambers, made of stainless-steel, which insert into ports in the main chamber. Usually each source is differentially pumped; however in some cases, as for the OH experiments, the secondary molecular beam source was brought close to the collision region, without differential pumping, to gain intensity. The source chambers and all the beam defining elements are readily interchangeable. The beam sources are pumped by unbaffled 6000 1/s diffusion pumps backed by 500 m^/h roots pumps, while the differential regions are equipped with 2400 1/s diffusion pumps.

The supersonic molecular beams of HC1, HBr, and H2S were generated in the present experiments by expanding pure gases at a stagnation pressure of 0.5-i-1.0 bar through a 100 μιη stainless-steel heatable nozzle, kept at about 250°C to prevent significant cluster formation. The HI beam was obtained by expanding pure HI at 0.3 bar through the same nozzle, which was kept at room temperature. The beams are skimmed by a 0.65 mm diameter stainless-steel skimmer and, after differential pumping, are further defined by a collimating slit. The angular divergence was 2.9°. The nozzle to collision center distance was 57 mm. Typical velocity spreads are about 20%. The hydrides are expected to be in the ground electronic and vibrational states, and, because of the rotational cooling during the supersonic expansion, they are in the lowest few rotational levels.

The supersonic molecular beams of D2 and CO were instead generated by expanding pure gases at a stagnation pressure of 4.5 bar through a 75 μπι diameter

8

stainless-steel nozzle kept at about 350°C. The beams were skimmed by a 1.0 diameter electroformed Nickel skimmer to an angular divergence of about 5°. Peak velocities were 2640 m/s for D2 and 1048 m/s for CO with FWHM spreads of 15%. The nozzle, mounted on an X-Y-Z translator, was moved closer to the collision center (ca. 40 mm) with respect to the previous arrangement, to gain intensity, reducing one stage of differential pumping. The D2 and CO molecules in the beams are estimated to be in the lowest few rotational (j=0,l,2) levels of the vibrational and electronic ground state.

While the generation of continuous supersonic beams of stable molecular species does not usually present difficulties, the production of radical beams of sufficient intensity remains a great challenge. In an apparatus equipped with a cw detector, such as a mass spectrometer, the optimum source should also be continuous. This condition rules out photolytic methods which are limited by the very poor duty cycle of the intense UV pulsed lasers which are required for photodissociating a suitable precursor contained in the beam. Instead, pulsed photolysis has been shown to work very nicely with pulsed beam sources, coupled with pulsed methods of detection such as LIF or REMPI [19, 24]. This consideration leaves essentially two ways of generating a cw beam of radicals: pyrolysis and electrical discharge. The thermal method is also not very suitable for generating intense cw beams of oxygen atoms or hydroxyl radicals, although an example of thermal generation of effusive beams of the two above mentioned species exist [87]. Of the two possible electrodeless discharges, the radio-frequency (RF) was found to be superior with respect to micro-wave (MW) discharges. Sibener et al [42] have shown that a RF discharge can operate at higher pressures and plasma temperatures than a MW discharge [88], so producing supersonic beams with lower velocity spread while giving a much higher degree of molecular dissociation, and hence permits achieving higher translational energies. Moreover, it was demonstrated that from a high-pressure RF discharge beam source also 0(lD) atoms can be produced when dilute mixtures of O2 in He are used [42], and this has opened the way to the investigation of 0(1 D) reactions in crossed beams [43]. For these reasons we have chosen the RF discharge method for our experiments.

The design (see Fig. 2) is similar to that described by Sibener et ah [42], and has been optimized for 0(*D) generation. This was achieved by using large nozzle diameters, high pressure and high RF power, and careful tuning of the RF electronic circuit. A parallel LC circuit made to resonate around 14 MHz is used to feed high levels of RF power into a plasma contained in a quartz nozzle, cooled with low electrical conductivity water. The tank coil (L) is differentially wound around the front of the quartz tube forming the nozzle, while a variable air capacitor (C) is mounted outside the vacuum. The RF power is supplied by a solid state home built radio-transmitter and amplified by a linear amplifier capable of providing RF output levels above 500 W. A variable ground tap (G) on the tank coil is used to produce a large stepdown of the plasma impedance to match that of the electronics (50 Ω), this

9

Figure 2. Sectioned side view of the radio-frequency discharge beam source mounted in the differentially pumped chamber of the crossed molecular beam apparatus (see text for details).

arrangement acting as an RF autotransformer. The location of the ground tap is most critical. Standing-wave-ratios (SWR) of less than 1.05:1 are routinely achieved, and this means that power coupling to the plasma is extremely efficient. Some of the ions emanated from the nozzle pass through the skimmer (made of boron nitride) and they are deflected by a suitable electrical field (D) placed in front of the collimating slit in the differential pumping region. A very important characteristic of this source is the plasma localization directly behind the orifice of the nozzle, which permits achieving very high degrees of dissociation (up to 95% with 5-20% 02/Ne gas mixtures, and up to 83% with 5% 02/He mixtures). This is obtained by operating the discharge in the capacitive mode by placing around the nozzle tip an electrically grounded block of aluminum (A). This feature and the nozzle cooling are the factors which limit atomic recombination before the expansion and reduce the quenching of electronically excited oxygen atoms. The presence in the atomic oxygen beam of significant amounts of 0(1D) was soon realized by observing [89] the production of OH from the reaction 0(lD) + H2 at a collision energy of about 3 kcal/mol, which is well below the barrier (about 8 kcal/mol) [90] for the ground state reaction 0(3p) + H2. The concentration of O(^D) in the beam is estimated to be in a few percent range with respect to that of 0(3p). The ratio of

10

0(1D) to 0(3p) atom concentrations is not found to vary significantly passing from 5% 02/He to 5% 02/Ne gas mixtures, when operating at 350 mbar and 260 W and using a 250 μπι diameter nozzle orifice. Thus, simply changing the gas mixture, supersonic beams of 0(3p, ID) atoms with translational energy varying in a wide range (0.18-̂ -0.8 eV) become available. The factors which determine the high degree of dissociation (plasma localization at the orifice, nozzle cooling, high power), associated to the small rate constants for quenching in He or Ne [91], are also the ones which permit the survival of 0(lD) in the beam. This may be a complication in the study of Ο(^Ρ) reactions, but it offers the possibility to study those of O ( I D ) .

When the corresponding 0(3p) reaction is endoergic or has a much smaller cross section (which is usually typical), it is possible to study only the 0(1 D) reaction: this is the case of the hydrogen displacement reactions of O ( I D ) with hydrogen halides discussed in the following Section. In other cases, where both triplet and singlet reaction channels are exoergic, the simultaneous presence of OpP) and 0(1D) in the beam can permit a direct comparison of the dynamics of the reactions of the two species in the same experimental conditions: this is the case illustrated in Section 4 with regard to the 0(3p, I D ) + H2S reactions.

The capability of localizing the discharge core in the region immediately behind the orifice is the reason why it has also been possible to generate an intense supersonic beam of OH radicals starting from dilute (2.5%) mixtures of water in He or He/Ne gas [57], The OH radicals are probably generated in the discharge by a variety of kinetic processes, which produce also other species. We have identified by mass spectral analysis, in addition to OH and undissociated water, also the presence of 0(3p), 0(1 D), H, O2, and H2. However, these other species do not interfere in the experiments to be discussed below, since the products of the reactions with D2 or CO may only come from single collisions with OH.

Typical operating conditions of the OH beam source are as follows. Pure helium is bubbled in water kept at 4°C to maintain a total pressure of 250-300 mbar. Peak velocities above 3000 m/s and velocity spreads of about 20% are attained. The OH radicals in the beam are expected to be in the lowest electronic state, since possible electronically excited OH will radiatively cascade to the ground state in much less than 1 μβ [92]. The OH beam is also expected to be rotationally cold, since rotational relaxation of OH by H2O is known to be almost gas kinetic (k=2.2x 10-10 cc molec-1 s"l) [93]. Vibrational relaxation of OH by H2O is known to be also rather fast (k=1.3^-2.6xlO""H cc molec-1 s_l) [94] and by Η and Ο atoms even faster (of the order of 10-10 cc molec"lsl) [95, 96]. Although we have some indication from our scattering results that OH in the beam is predominantly in the v=0 level, a detailed characterization of its internal quantum states is desirable and is presently pursued by spectroscopic means.

The beam velocities are measured by single-shot time-of-flight analysis to better than 1%, using a high-speed multichannel scaler and a computer-controlled CAMAC data acquisition system. Product velocity distributions are obtained at selected

11

laboratory angles using the cross-correlation TOF technique [97] with four 127-bit pseudorandom sequences. The flight length is 23.6 cm. Typical counting times vary from 10 to 100 minutes, depending on the signal level. High time resolution is achieved by spinning the TOF disk, located at the entrance of the detector, at 393.7 Hz corresponding to a dwell time of 5 per channel.

The angular dependence of the reactive differential cross section is measured by rotating the detector unit in the plane defined by the two beams. The laboratory angular range accessible to the measurements is about 145°. The laboratory angular distributions of products, Ν(Θ), were obtained by taking several scans of, typically, 30-200 s counts at each angle. The Ν(Θ) are time normalized, when warranted, by returning the detector to an arbitrary reference angle in order to account for possible long-term drifts in beam intensities and detector sensitivity. The secondary target beam (in the present experiments the beam of molecular species) is modulated at 160 Hz by a tuning fork chopper. The background and signal plus background counts are obtained from a pulse counting dual scaler, synchronously gated with the tuning fork.

3. R E A C T I O N D Y N A M I C S O F 0(1D) A T O M S

3.1. 0(1 D ) atoms: their origin, role, kinetics and dynamics The state of oxygen is metastable (τ > 100 s) since the magnetic dipole

allowed emission:

0 ( ^ 2 ) -» 0(3P2,l) + hv (λ = 630 nm, 636 nm)

has very low transition probability. Therefore the above emission is observed in the rarefied conditions such as those of the upper atmosphere (as part of the night airglow), where O(^D) is produced by photolysis of ozone in the intense Hartley continuum at wavelengths <310 nm through the spin-allowed process:

O3 + hv -> 0 ( ! D 2) + 0 2 ( 1Δ )

Because of the long radiative lifetime, collisional deactivation of 0(1 D ) through physical quenching to Ο(^Ρ) by N2 and O2 is dominant throughout the troposphere and stratosphere. In competition a small but significant fraction of 0(1D) reacts with trace-level species such as N2O, water, hydrocarbons, hydrogen halides, and pollutants such as chloro-fluoro-carbons (CFC), creating highly reactive radical species which are responsible for dramatics reduction of the Earth's ozone layer [98].

In addition to atmospheric chemistry, 0(1D) reactions are also relevant in high-temperature combustion processes [99], plasma etching processes [100], chemical laser developments [101], and may also influence the degradation of surface

12

materials of spacecrafts in low-earth orbit [102]. There is a need, therefore, for a detailed knowledge of the kinetics and dynamics of 0(1 D) interactions.

Although rate constants and their temperature dependence have been measured for a series of 0(lD) reactions [91, 103-105], less information is available on the dynamics of the microscopic processes. The rate constants of 0(*D) reactions are typically many orders of magnitude larger than for ground state 0(3P). This makes the role of 0(1 D) in many processes much more important than its abundance would suggest. It is interesting to note that the even more energy-rich 0(1 S) (4.2 eV against 1.97 eV of 0(lD)), is very often much less reactive than 0(1D) [91, 104]. The chemistry of 0(3p), 0(1D), and 0(1 S) is therefore markedly different and quite intriguing, and illustrates the unpredictable nature of excited state chemistry. Only recently studies by LIF [28-34] and Infrared Chemiluminescence using fast-time-resolved Fourier-transform-spectroscopy (FTR-FTS-IC) techniques [35-37] have begun to provide information at microscopic level for some 0(1 D) reactions. However, only the investigation of reaction channels leading to OH product has mainly been possible with these techniques, which exploit the ease of detection of OH species. In these experiments, 0(lD) is produced in a pulsed fashion, using available UV lasers. More recently, 0(lD) reactions with N2O and CH4 have been studied via the production of velocity-aligned 0(lD) atoms using sub-Doppler polarized LIF to probe the products and determine their scalar quantum state distributions and vector correlations [26]. Besides the pioneering work of Lee and co-workers [43] on the 0(lD) + H2 and CH4 reactions, no other CMB studies of 0(1D) reactions were reported. The first reactions of 0(1D) we have looked at are those with hydrogen halides.

3.2. The reactions O^D) + HX —» XO + H (X=CI, Br, I) The reactions of excited oxygen atoms with hydrogen halides can proceed

through two competing pathways:

O(lD) + HC1 -> CIO + H

OH + CI

ΔΗ£=-6.3 kcal/mol (la)

ΔΗ°=-44.4 kcal/mol (lb)

O(lD) + HBr r-> BrO + Η

-> OH + Br

ΔΗ£=-14.0 kcal/mol

ΔΗ°=-60.5 kcal/mol

(2a)

(2b)

O(lD) + HI ΙΟ + Η

OH + I

ΔΗ°=-25.0 kcal/mol

ΔΗ°=-76.2 kcal/mol

(3a)

(3b)

13

The exoergicities of both reaction channels increase along the series, reflecting the corresponding increase in the heat of formation of the hydrogen halide (AH£f (HX) = -22, -8.7, and 6.3 kcal/mol for X=C1, Br, and I , respectively [106]). Since 0( !D) lies 45.4 kcal/mol above the ground state, the Η-displacement channel is strongly endoergic for 0(3p) with all the hydrogen halides, while the OH forming channel is still exoergic for 0(3p) with HBr and HI, but slightly endoergic with HC1. The above reactions represent a significant mechanism by which chlorine and bromine in the relatively unreactive form, HC1 and HBr, are converted into chemically active radicals [98, 107]. Reaction (3) is of more interest in tropospheric iodide chemistry particularly in maritime regions [108]. In addition to being of practical relevance, reactions (1-3) are of interest to dynamicists since they form a class of three-atom reactions which are amenable to theoretical and experimental investigation.

Measurements of the absolute rate constants and product yields at 298 Κ have recently been reported for reaction (1) and (2) by Wine et al [109]. They found that channel (la) was about 35% of channel (lb); but, surprisingly, channel (2a) was found to be only a minor reaction pathway (k(2a)/k(2b) < 0.045). No kinetic data are available on reaction (3). Infrared chemiluminescence studies on the OH channel by FTR-FTS techniques [37] have been carried out on reactions (1-2), while the internal state distribution of the OH product has been investigated by LIF only relatively to reaction (1) [29, 33]. No spectroscopic studies are available on reaction (3). The results of these spectroscopic investigations have indicated very high product rotational and vibrational excitation with pronounced population inversion. No information was instead available on the reaction dynamics of the XO (X=C1, Br, and I) forming channel.

Ab initio calculations on the potential energy surface (PES) of reaction (1) are available. A classical trajectory study [110] of the reaction dynamics on a PES fitted to the ab initio points found a cross section for the CIO + H channel more than a factor ten smaller than for OH + CI at energies between 10 and 20 kcal/mol.

In our experiments we have been able to obtain direct information only on the XO (X=C1, Br, and I) channel. The signal at the kinematically unfavored OH mass is always < 0.01 of the signal at the XO mass for all three reactions and this has permitted us to estimate only a lower limit for the branching ratio of the cross sections (see below). Angular and velocity distribution measurements were carried out at one collision energy, Ec, for the CIO product [111], and at two different energies for BrO and IO products. In Fig. 3 we report the angular distribution, Ν(Θ), at the higher collision energy for the three systems. As can be seen, these distributions are peaked around the center-of-mass, as one would expect, but are fairly wide for such massive XO products left by a light Η-atom. This suggests that a large fraction of the available energy is released into translation. The overall shape of the distributions reflects the detailed dynamics of the reactive encounters. A significant variation is noted going from HC1 and HBr to HI. While the distributions

14

" τ 1 1 1 1 1 1 II r

0 ( 1 D) + HCI H> CIO + Η GO

' c Ec=12.2 kcal/mol

_ 0 ( 1 D) + HI -> ΙΟ + Η CO -»—»

' c Ec=13.6 kcal/mol

LAB SCATTERING ANGLE, ©

Figure 3. Laboratory angular distributions of the XO (X=C1, Br, and I) product from the 0(*D) + HX reactions at the indicated collision energies. Solid lines: calculation with best-fit cm. translational energy and angular distributions.

15

Figure 4. Time-of-flight spectra of the product XO (X=C1, Br, and I) at angles close to the cm. and for the collision energies of Figure 3. The time scale is absolute, the ion flight time and electronic offests having been taken into account. The flight length was 23.6 cm. Solid lines: calculation with best-fit cm. translational energy and angular distributions.

Flight time, t (microseconds)

16

of CIO and BrO exhibit a slightly more favored backward scattering (with respect to the 0(1D) beam), that of 10 is characterized by a more favored forward scattering in the lab frame. Product velocity distributions were measured at a number of selected lab angles (7 for HC1, 13 for HBr, and 11 for HI). Typical TOF spectra, taken at angles close to the center-of-mass for all the three systems at the highest E c, are reported in Fig. 4. The fast and slow peak structure that starts to appear in the BrO and 10 data is, despite their large mass with respect to the H counterpart, a clear indication that the translational energy distribution peaks away from zero and that the product has a high recoil energy.

Data analysis was carried out by fitting the product angular distributions and TOF spectra, using a forward convolution method, with the assumption of a separable form for the cm. frame product flux distribution I c # m. C f > , E')= T(0)xP(E'). The calculated lab distributions were averaged over beam velocities and collision angles, as well as the detector acceptance angle and the length of the ionizer. The P(Ef) and Τ(ϋ) functions which produce the best-fit (reported as solid lines in the Figs. 3 and 4) of both angular and velocity distributions, are shown as velocity flux contour maps superimposed on the Newton diagrams in Fig. 5, after a straightforward transformation to convert the flux distribution from an energy space to a velocity space. The results can be summarized as follows. The CIO + Η and BrO + Η channels are qualitatively similar and are characterized by a slightly anisotropic cm. angular distribution with more intensity in the backward direction. The angular distribution of the 10 channel features, instead, sideways scattering at low E c (see Fig. 6) and preferred forward scattering at high E c (see Fig. 5). The average fraction of total available energy disposed in translation is large in all cases: 43% for CIO, 49% for BrO (both at low and high Ec), 58% and 48% for 10 at low and high E c, respectively. While for CIO and BrO the P(E') peaks beyond the exothermicity of the reaction, for 10 the P(E') peaks at a value («18.5 kcal/mol) which is significantly lower (ΔΗ—25 kcal/mol). The decrease in the fraction of energy disposed in translation as the collision energy is raised in the 0(*D) + HI reaction indicates that the extra initial translational energy is more efficiently channeled into internal energy.

In Fig. 7 the energy level and correlation diagram which is relevant to the following discussion is shown for the 0(*D) + HC1 system. The energy levels of the excited states of the two isomeric forms, H0C1 and HC10, are taken from the SCF-CI calculations of Bruna ei al. [112]. Qualitatively similar diagrams are expected to hold for the 0(*D)+HBr and 0(*D)+HI systems too, for which only the energies of the asymptotes and of the ground state of the HOX isomer are known.

The large translational energy release in all three cases reflects the nature of the PES of bond breakage and it is an indication of the presence of a potential barrier in the exit channel [113]. Ab initio calculations do show, indeed, the presence of exit barriers for the H0C1/HC10 dissociation into C10+H (see Fig. 7).

The cm. angular distribution of CIO and BrO can be thought to arise from the

17

0(Ό)

Figure 5. Center-of-mass XO (X=C1, Br, and I) product velocity flux contour maps superimposed on the nominal Newton diagram for the indicated relative collision energies, from the best-fit analysis of the laboratory angular and TOF distributions. The velocity scale is the same for all three systems.

18

J # = 9 0 °

Figure 6. Center-of-mass 10 product velocity flux contour map superimposed on the nominal Newton diagram for the relative collision energy of 4.7 kcal/mol, from the best-fit analysis.

superposition of a backward-forward symmetric component and a backward scattering component. In the case of 0(lD)+HBr, it was possible (actually, necessary) to separate these two contributions which witness two different micro-mechanisms. In fact, at Ec=5.0 kcal/mol a satisfactory fit could not be reached using a single T(p) and P(E'), and only an isotropic symmetric Τ(ϋ), coupled to a P(E') with 48% of the energy in translation, and a strongly backward peaked Τ(ϋ), coupled to a P(E') with 77% of the available energy in translation, gave an accurate description of the experimental data [114]. The separation of the two contributions was made possible only by the high resolution of the present experiments. The backward-forward component can be related to the formation of a long-lived complex via an insertion mechanism forming the strongly bound ΗΟΧ(χΐΑ')

19

ο Ε \ σ ο

Ο or

L U

L U

LE.

6 0

4 0

2 0 _

0 —

- 2 0

- 4 0

- 6 0

- 8 0

- 1 0 0

- 1 2 0

vHCIOÇA1)

pfa + HCI ^HQCl (k ' )

OfT)+HCI

//OH + CI

^0('D)+CIH

HCIO(XW)

Ηθα(ΧΆ)

Figure 7. The energy level and correlation diagram for the 0(lD)/HCl system. The energies of the H0C1 and HC10 states are the values for the equilibrium geometries of the ground electronic state, as calculated by Bruna et al. [112]. The energy level of the asymptote of the ground state reaction is also indicated.

molecule (see Fig. 7). If the HC10 isomer was formed, since only a small barrier at the most is calculated to exist for interconversion of HC10 into H0C1, it would convert to the much more stable H0C1 prior to dissociation into C10+H products. The backward scattering component originates from reactive encounters taking place by near-collinear collisions at small impact parameters. Since the calculations by Bruna et al. [112] indicate that the first excited ^A" state of HC10 appears to be inaccesible at the experimental energies, the direct abstraction of the halogen atom with rebound dynamics evidenziate the role played by H-X-0 configurations, with the oxygen atom attacking the halogen side of the hydrogen halide molecule and proceeding on the ground singlet surface of HC10. This backward contribution amounts to about 15% in the case of HBr (at both energies) and is presumably similar in the case of HC1, since the shape of the Ί(ϋ) is also similar in the two systems. No backward component is, however, observed in the reaction with HI, suggesting that the direct abstraction of the iodine atom plays a negligible role.

The qualitative shape of the cm. angular distributions can be understood in terms of the conservation of the total angular momentum J [4]. For a reaction which proceeds through the formation of long-lived complex, one will observe a symmetric cm. angular distribution sharply polarized at ϋ=0° and d=180° if the initial and final orbital angular momenta, L and Lf, are parallel or antiparallel. This situation occurs only when the product molecules are not rotationally excited or when M' (the

20

projection of the total angular momentum J along the final recoil velocity v') is null. In all other cases, i.e., when part of the total J is removed as j ' (rotational motion of the product molecule) and Mf is not null, L and L' are weakly correlated and the scattering angle ϋ emerges out of the collision plane and therefore the condition for sharp polarization at the poles is removed, while the backward-forward symmetry is maintained. In general, there will be a distribution of M1 and, therefore, dissociation of a complex leading to highly excited products will lead, at the most, to an isotropic Ί(ϋ). However, when geometric constraints are present in the collision complex, they will determine the more probable values of M'. If the collision complex dissociates with low Mf values, v' and ν are perpendicular to J and the angular distribution will be polarized along v. If, instead, the complex dissociates with high M1 values, then the final relative velocity v' is almost in the direction of J and hence perpendicular to the initial v: this gives rise to sideways scattering. The behavior exhibited by the 0(*D) + HI —> ΙΟ + Η reaction, in which sideways scattering is observed (see Fig. 6), appears to be described by the latter situation. The CIO product Ί(ϋ) shows, instead, only a weak polarization at û=0° and û=180°, while the Ί(ϋ) of BrO has an even weaker polarization at high E c and a null polarization (i.e., Τ(Φ) is isotropic) at low Ec. For the specific mass combination, angular momentum partitioning arguments predict high product rotational excitation for all three reactions of the series. In fact, as a consequence of the much smaller reduced mass of the products with respect to that of the reagents, L ' « L and the final rotational angular momentum j 1 is about equal to L, since the initial j is very small with respect to L in view of the fact that reagent molecules are formed in a supersonic expansion where considerable rotational cooling occurs. In the absence of a preferred dissociation geometry, high rotational excitation would, at most, lead to an isotropic scattering. The fact that symmetric sideways scattering is observed in the case of IO formation at low E c indicates that the reaction proceeds through the formation of a long lived complex and that M1 is very large. In other words, the direction of Η displacement is orthogonal to the plane containing the initial relative velocity vector (that is, v1 is parallel (or antiparallel) to L and thus perpendicular to v), resulting in very highly rotationally excited 10. This dynamical behavior, although it can be predicted, on the basis of total angular momentum conservation, has never been observed previously in A + BC reactions. The reaction O(^D) + HI is surely a favorable case, since the departing Η atom is very light and removes only a small fraction of J, so that the angular momentum related to the 0-1 rotational motion dominates the total angular momentum J. Herschbach and co-workers [4] have described a mechanism for sideways peaking in which an oblate complex dissociates along its symmetry axis. Situations occur in which the nature of the potential energy surface (i.e., the shape of the collision complex) and the conservation of angular momentum make the decomposition very similar to that of an oblate complex. It is interesting to note that the trend, along the series, in the polarization of the Ί(ϋ) function, which becomes progressively more sideways as

21

we move from HC1 through HBr to HI, reflects a trend in the geometry of the dissociating transition state, where geometric constraints appear to become more relevant going from HOC1 to HOI.

Since in the O^D) + HI reaction at Ec=4.7 kcal/mol about 58% of the available energy is released into product translation, and high rotational excitation also occurs, very little energy remains for electronic and vibrational excitation. The spin-orbit splitting of ΙΟ(2]Ί3/25ΐ/2) has recently been determined to be 6 kcal/mol and the vibrational spacing about 2 kcal/mol [116]. This would suggest that 10 is mainly formed in the lowest few vibrational levels of the ground ^113/2 electronic state, or in the ground vibrational level of both possible ^113/2 and ^U\/2 electronic states.

A lower limit for the ratio between the cross sections, σ, for formation of CIO and OH was calculated [111] to be a(C10)/a(0H) > 0.34 ± 0.10. A similar estimate [114] for reaction (2) gives σ(ΒιΟ)/σ(ΟΗ) > 0.16 ± 0.07. For reaction (3) no estimate was attempted, because the sensitivity to the OH channel decreases as the halogen atom mass increases due to the very unfavorable kinematics. However, on the basis of the relative signal intensities (IO « 2.5 BrO ~ 1 CIO), it is inferred that the BrO and IO channels should be as significant as the CIO channel, and, therefore, a large fraction of the corresponding overall reactions. This is in agreement with the results of recent bulk studies [109] for CIO, but not for BrO.

We may now compare information coming from different sources on the competing channels of the reactions 0(lD) + HC1 and HBr, and therefore attempt to draw a global picture. The dynamics of CIO formation appears to be very different from that observed for the OH channel in the same reaction. The observation of very hot rotational [29, 33] and vibrational distributions [33, 37a], with very pronounced inversion, and of a marked propensity for Π(Α')λ sublevel production in OH, led to the conclusion [33] that reaction (lb) must proceed with the insertion of 0(*D) into the H-Cl bond via the HOC1 (XlA%) surface to form highly excited HOC1, which rapidly dissociates into the product, with rupture of the O-Cl bond, before the available energy is randomly distributed among all degrees of freedom. The extremely high rotational and vibrational excitation of the products is thought to arise from initial excitation of the HOC1 bend during the insertion act, with the O-Cl bond-breaking occurring before this mode couples to the stretching modes of the complex. Statistical calculations [117], while simulating successfully the non-statistical population of OH rotations, could not explain the vibrational distributions. Trajectory calculations [110] reproduce very closely the experimental rotational distributions, but underestimate the vibrational excitation. The present results suggest that the formation of CIO takes place, in part, via the ground *A' surface correlating with HCIO, since the first excited state is calculated to lie significantly high in energy (see Fig. 7), and, in part, via the ground 1 A1 and/or first excited ^A" surface correlating with HOC1, forming an HCIO/HOCI intermediate, which dissociates into products, with rupture of the H-CIO/H-OCI bond. The

22

backward component can be associated to HCIO configurations, while the symmetric component to HOC1 configurations. The potential barrier that has to be surmounted in the exit channel would lead to the large translational excitation of the products. The experimental results support a similar interpretation also for the 0(1D) + HBr reaction. The different dynamics observed for the OH- and ClO-forming channels may therefore be ascribed to the different role played by the different singlet surfaces and the different configurations of the reaction intermediate. No clear indication about a possible role of singlet-triplet surface crossing has been obtained in studies of the OH channel. However, from a flow-reactor infrared-chemiluminescence investigation of the reverse reaction H + CIO, Wategaonkar and Setser [118] found that the OH + CI channel is favored over the HC1 + 0(3p) channel by a factor 4.5. From the vibrational energy disposal in OH and HC1, they inferred that the reaction proceeds via the HOC1 (X^A') ground state surface and that while OH is formed by adiabatic dissociation, HC1 is formed by a crossing from the X^A' to the 3A" surface late in the HCIO exit channel. However, they admitted that direct reaction on the 3Am HOC1 surface could also partly account for HC1 formation. Support to this interpretation comes from the fact that the energy calculated [112] for the equilibrium geometry of the excited 3 A" HCIO isomer is about 9 kcal/mol above the reactants.

All the experimental results available should facilitate attempts to model the dynamics of the O(^D) + HX reactions using realistic potential energy surfaces. Sophisticated ab initio calculations of the potential surfaces for the reaction 0 ( * D ) + HC1 have recently been carried out by Aquilanti et al. [119], and dynamical scattering calculations using quasiclassical and quantum methods on an empirical PES, based on a fit to the ab initio points, are currently under way [120].

A final comment regards the general mechanism of 0(1 D ) reactions. For a long time, they have been thought to proceed through insertion of the oxygen atom into the molecular bond to give a highly excited, possibly long-lived, intermediate, usually corresponding to a stable species. Whether the collision intermediate was sufficiently long-lived to allow complete energy randomization it has been a matter of some uncertainty. As matter of fact, the products of most of the reaction channels investigated to date by spectroscopic techniques are characterized by a non-statistical internal energy distribution. The Η-displacement channels examined here appear to be the first (but not the last, as it will be seen in the next Section) example of 0(1D) reactions observed to proceed through a formation of a collision complex, living several rotational periods. A more efficient trapping into the deep potential well of the singlet intermediate appears to occur in the H displacement processes, because of their lower exothermicity with respect to the H abstraction processes and of the presence of a significant barrier in the exit channel (see Fig. 7). A full dynamic description of O(^D) reactions, which have the characteristics of releasing a very large amount of energy into a strongly bound intermediate, needs to take into account the stability of the singlet intermediate and of its possible isomers, the

23

exoergicity of the process, the multiplicity of the electronic state of the product and the mass combination of the reagents.

4. EFFECT OF ELECTRONIC EXCITATION ON REACTION DYNAMICS: THE REACTIONS 0( 3P , iD) + H2S

Investigation of atom-molecule reaction dynamics has witnessed significant experimental and theoretical effort towards the understanding of the effect on reactivity of the internal (rotational and, especially, vibrational) energy of the molecule [1]. On the contrary, much less work has been done to explore and characterize the effect of the internal excitation of the atom. What characterizes the effect of electronic excitation is that the reaction is actually occurring on a different potential energy surface with respect to that of the ground state reagent. The ground and excited state surfaces will usually have different symmetry and may be topologically very different, and interaction between the surfaces may still occurr through non-adiabatic coupling.

Recently, by exploiting the capability of generating continuous supersonic beams containing both Ο(^Ρ) and 0(1 D), we undertook the study of the reactions of H2S with both atomic oxygen species in conditions where the contribution of the two electronic states were resolved in high-resolution angular and velocity distribution measurements. Here, we will give a summary of our results. Preliminary accounts have already appeared [121, 122] and a comprehensive publication is currently under preparation [123].

The reactions of atomic oxygen with hydrogen sulphide are the prototype for the atmospheric oxidation reactions of sulphur compounds, produced both biogenically and anthropogenically, and are also important in the various processes associated with the combustion of sulphur contaminated fossil fuels [124]. The reaction of 0(3p)and O^D) withH2S can occur via three competing pathways:

> HSO + Η Δ Η ^ -4 kcal/mol (4a)

0(3p) + H2S -> U OH + SH ΔΗ°=-11.8 kcal/mol (4b)

-> SO + H 2 ΔΗ°=-53.4 kcal/mol (4c)

O(lD) + H2S

j-> HSO + Η ΔΗ°= -49.4 kcal/mol (5a)

• OH + SH ΔΗ°=-57.2 kcal/mol (5b)

> SO + H 2 ΔΗ°=-98.8 kcal/mol (5c)

24

The energy level and correlation diagram for the above reactions are given in Fig. 8.

6 ( V E,

ο Ε \ σ ο

Ο

LU

40

20

0 —0(3P)+H2S l \ 0SH2(

5A

-20

- 4 0

-60

-80

—0('P) + HPS_

2—

HOSH(A)

Figure 8. The energy level and correlation diagram for the 0(3p, ID) + H2S system. E m and indicate the lowest and highest investigated collision energies.

Although the kinetics and mechanism of the reaction of 0(3p) with H2S have been investigated extensively [125, 126], very little information is available on the chemical dynamics. The recommended value for the room temperature rate constant is 1.8xl0~~14cc molec~l s~* and for the activation energy 4.3±0.4 kcal/mol [127]. Bulk studies [126] concluded that OH formation is the main channel (>50%), while the addition-displacement pathway leading to HSO was estimated to account for less than 20% of the overall reaction yield. Observation of H2 elimination, although from an energy point of view certainly possible, has never been reported. Grice and

25

co-workers [128] measured the HSO angular and velocity distributions in a low resolution CMB study at Ec=7.2 kcal/mol using a pure 0(3p) beam and observed a threshold energy of 3.4±0.5 kcal/mol. The HSO product was found to be nearly isotropically distributed, but slightly favoring the backward hemisphere. An extremely high product translational energy (about 86% of the total available energy) was derived. Agrawalla and Setser [129] measured the OH vibrational distribution in a flowing-afterglow reactor using IC and LIF techniques. The large fraction of energy released in vibration was found to resemble closely the energy disposal to HF and HC1 in F and CI atom reactions, suggesting a similar dynamics for H abstraction by 0(3P), F(2P), and C1(2P). This was rationalized on the basis of the similar kinematics associated with the H + L-H mass combination (H = heavy, L = light), overriding differences in the PES. The OH vibrational energy disposal in 0(3p) + H2S was found to be also similar to that observed in the corresponding 0(1D) reaction (see below) and it was speculated that the abstraction-like dynamics of 0(1D) was originating from a singlet-triplet surface crossing following the initial insertion of the Ο atom. Currently, Hancock and co-workers [27, 130] are looking at the OH channel via the production of velocity-aligned Ο(^Ρ) atoms and detecting products (OH and SH) by LIF.

The investigation of the kinetics and mechanism of the reaction of O(^D) with H2S has instead been much more limited. The reaction is supposed to proceed with a virtually zero activation energy, as many other O(^D) reactions, with the room temperature rate constant estimated [36, 37b] to be in the gas kinetics range («2.5x

10~10 cc molec~l s~l), that is about four orders of magnitude larger than for 0(3p). From a microscopic point of view, information only on channel (5b) leading to OH formation is available from LIF [31] and FTR-FTS-IC studies [36]. This channel is thought to account for about 50% of the overall reaction [31]. Nothing was known about the other reaction channels leading to HSO + Η and SO + H 2 products.

We have looked in detail at the Η displacement and at the H 2 elimination channels by measuring angular and velocity distributions at m/e=49 and m/e=48 for six different initial collision energies, in an energy range from 3.4 to 11.8 kcal/mol. Different energies were obtained by varying the velocity of the oxygen beam and, in some cases, also of the H2S beam.

Let us examine first the Η displacement pathway. At collision energies of 3.4 and 4.75 kcal/mol, that is lower than and comparable to the threshold of the 0(3p) reaction, the angular distributions at m/e=49 are found to have the same shape with a clear backward-forward structure (see Fig. 9a). From the widths of the lab angular distributions, which are consistent with the large exothermicity of the 0(*D) reaction, and from the fact that at room temperature the global cross section of the 0(3p) reaction is four orders of magnitude smaller than that of the 0(1 D) reaction, we conclude that the observed HSO (HOS) product comes essentially from the excited state reaction. The data in this low energy range can be fitted by using a single symmetric cm. angular distribution and a single translational energy

26

Τ 1 1 1 Γ Π Γ

C

1.0

® 0.5

0(1D) + H2S -» HSO + H

Ec=4.75 kcal/mol α

r

τ r (a)

= 0

m/e=49

"0(3Pf1D)

0.0

c ZJ

•e i.o CO

H,Sl

0(3P,1D) + H2S -» HSO + H (b)

Ec=8.15 kcal/mol a = Q.30

• m/e=49

© 0.5 h n i 3 D l 0(3P,1D)

0.0

H 2 ^

H h

Z3

. r i

0(3P,1D) + H2S -> HSO + H (0

Ec= 11.8 kcal/mol q = Q.70

1.0

0.5

m/e=49

~0(3Ρ,Ό)

0.0

HpSl

30 60

LAB SCATTERING ANGLE, ®

90

Figure 9. Laboratory angular distributions of the m/e=49 (HSO) product from the 0 (3p, ID) + H2S reactions at three different collision energies. Solid lines: calculated curves with best-fit cm. translational energy and angular distributions according to Eq. (6) with the indicated α value. The separate contributions from the 0(*D) and 0(3P) reactions are shown with dashed and dotted lines, respectively. At Ec=4.75 kcal/mol, only the 0(*D) contribution is present.

27

— ι 1 Γ

Ec=4.75 kca l /mo l

100 200 300 400

Flight time, t (microseconds)

Figure 10. Time-of-flight distributions of the m/e=49 (HSO) product at angles close to the cm. for the three collision energies of Figure 9. Solid, dashed and dotted lines have the same meaning as in Figure 9.

28

distribution which exhibits 40% of the total available energy in product recoil energy. This indicates that the H displacement reaction of 0(*-Ό) proceeds through the formation of a long-lived collision complex, presumably a thioperoxide (HOSH) intermediate, following the insertion act. Formation of a sulphoxide (OSH2) intermediate, following addition of Ο to the sulphur atom cannot be ruled out by our data: however, on the basis of theoretical calculations [131] on the relative stability of the two isomers, which find the sulphoxide form less stable than the thioperoxide form by about 20 kcal/mol (see Fig. 8), we find more plausible the formation of a long-lived complex with the more stable peroxide-like structure. The cm. angular distribution is perfectly symmetric, but shows only a mild forward-backward polarization. This can be appreciated in the product velocity flux contour map shown in Fig. 12. Even for this specific mass combination one expects a significant (L, j') correlation [4]. Indeed, mild peaking of the cm. angular distribution arising from the dissociation of long-lived complexes is known to be an indication of significant product rotational excitation. The departing Η atom can impart a significant torque either in a sulphoxide-like transition state geometry, where the Η would come off forming an angle of about 45° from the symmetry axis (see below), or in a thioperoxide transition-state geometry, where the Η is expected to be ejected close to the symmetry axis as well as from bent geometries. In contrast to previous, more approximate, calculations [132, 133], recent large-scale theoretical calculations have shown [134] that the HSO isomer is more stable than HOS by about 2 kcal/mol. While a sulphoxide transition state can only dissociate into HSO + H, a thioperoxide complex, besides giving HSO + Η following the breaking of the excited O-H bond, could also evolve towards the rupture of the weaker S-H bond producing HOS + H, if energy redistribution throughout the complex has time to occurr. This could in part occur in the present case. The fairly large fraction (40%) of energy deposited into product translation, more than expected on statistical ground, indicates the presence of a sizeable potential barrier on the exit channel and reflects a repulsive interaction between the separating product particles.

As the collision energy is raised above 6 kcal/mol, the experimental data show clearly that a more complex situation is occurring. This is not surprising considering that the cross section for O(^P) reaction becomes significant at E c well above the threshold [126-128], and that the concentration of Ο(^Ρ) is dominant in the beam. The angular distribution data at Ec=8.15 and Ec=l 1.8 kcal/mol are reported in Fig. 9(b) and 9(c), and TOF spectra close to the cm. angle are shown in Fig. 10 for the three energies considered in Fig. 9. For the purpose of illustration, we wish to use Fig. 11, which shows the m/e=49 lab angular distribution and velocity spectrum at 0=32° measured at Ec=l 1.8 kcal/mol, together with the most probable Newton diagram. The error bars are well within the dots. The angular distribution shows some structure which can be related to the important features of the velocity and angular distributions of the products from the Ο(^Ρ) and O(^D) reactions in the cm. frame of reference. The Newton circles in Fig. 11 delimit the maximum cm.

29

0( 3Ρ, Ό) + H2S

Ec= 11.8 kcal/mol

Figure 11. Laboratory angular distribution and velocity spectrum at 0=32° of m/e=49 (HSO) product at Ec=l 1.8 kcal/mol for the 0( 3P, ιΌ) + H2S reactions. The corresponding Newton diagram is shown, indicating the maximum speed of the products from the ground- and excited-state reaction when all the available energy is assumed to go into translation. Solid, dashed and dotted lines have the same meaning as in the Figures 9 and 10.

speed for HSO product formed from the 3 Ρ or oxygen reactions assuming that all available energy goes into translation, and hence define the lab range where the product from reaction of each of the two electronic states, is expected. The broad peak at 0=40° reflects a pronounced backward scattering (with respect to the Ο atom), but a significant shoulder is also present in the forward direction. For a detailed interpretation of the angular distribution, measurements of TOF spectra at high-resolution proved to be essential. TOF spectra were recorded every 2° from 0 =12° to 0=52°. Several TOF spectra close to the cm. angle show two distinct peaks,

30

and it was immediately clear from their positions and relative intensities that there was a process generating a slow, rather intense, back distributed product and, simultaneously, another process producing rather fast products in the cm. forward direction. The two processes may be attributed to 0(3p) and 0(*D). The angular and translational energy distributions in the cm. frame were determined for each reagent electronic state by forward convolution of trial distributions, according to the formula:

Icm.(fl> E') = [Ύ(ϋ) χ P(E')] + α χ [Τ(θ) χ Ρ(Ε')] (6) *D Ρ

The data quality, together with the wide range of energy investigated, permitted us to determine with considerable accuracy the cm. functions and the parameter a, which represents the weight of the 3p contribution. The value of α is 0.30±0.03 at Ec=8.15 kcal/mol, while it is 0.70±0.05 at Ec=11.8 kcal/mol. Values at other energies are in line with the expected trend. The dashed and dotted lines in the Figures 9-11 are the lab distributions generated by the best-fit cm. functions for reaction (4a) and (5a). While at Ec=3.4 and 4.75 kcal/mol the T(û) for O^D) reaction is backward-forward symmetric, at higher collision energy the distribution becomes more forward, reaching a ratio Τ(ϋ=180°)/Τ(ϋ=0°)=0.60 at Ec=l 1.8 kcal/mol. This finding is interpreted in terms of formation, at higher E c, of an osculating complex [135]. The estimated mean lifetime of the complex decreases from 4.7 rotational periods at Ec=6.69 kcal/mol to slightly less than one rotational period at Ec=11.8 kcal/mol [123]. The product flux contour map for the 0(*D) reaction at Ec=11.8 kcal/mol is shown in Fig. 12 (bottom left), and the more pronounced forward scattering with respect to Ec=4.75 kcal/mol (see Fig. 12 top) can be clearly seen.

Instead, the 0(3p) reaction is found to proceed through a direct mechanism with a pronounced rebound dynamics at all four energies investigated. At Ec=11.8 kcal/mol the cm. angular distribution is completely confined to the backward hemisphere and the P(E') indicates that a very large fraction («60%) of the available energy is released in translation (see Fig. 13). This suggests that the barrier to reaction is located in the exit channel. The results obtained for the 0(3p) reaction are in qualitative agreement with those of an earlier [128] low-resolution CMB study. The smaller uncertainties in our work [121-123], which are the consequence of a much higher resolution, have permitted us to determine the heat of formation of the HSO radical with an accuracy of better than one kcal/mol and to have a direct insight into the geometric orientations between the interacting particles during the reactive encounter, i.e., on the geometry of the transition state.

In general, when the average lifetime of the reaction intermediate is comparable to the collision time (direct reaction), the anisotropy of the product angular distribution provides direct information on the possible geometric constraints of the reaction intermediate. This is particularly true for a Η displacement reaction, in

31

Figure 12. Center-of-mass m/e=49 (HSO) product velocity flux contour maps superimposed on the nominal Newton diagrams for the 0(lD) + H2S reaction at Ec=4.75 kcal/mol and 11.8 kcal/mol (left) and for the 0( 3P) + H2S reaction at Ec=l 1.8 kcal/mol (right). The same velocity scale applies for all three contour maps.

which the light H leaves very quickly, virtually without perturbing the transition state geometry. Hence, if the experimental resolution is good enough to trace the departing angle and velocity of the H atom with respect to the center-of-mass and to the direction of the incoming atom, one really can probe the transition state geometry from the T(i!>). As only the heavier counterpart of the reaction products can

32

ΟΓΡ) + HPS - HSO + Η

1.0 h

0.5 h

0.0

Ε =11.8 kcal / m o l c

Ί Γ Ί Γ

J I I I ! L

0° 9 0 ° 180° C M . Scattering Angle , &

& = 180°

0 5 10 15

Translational Energy , E'T

Figure 13. Center-of-mass angular distribution (top) and translational energy distribution (bottom) of the HSO product from the 0(3P) + H2S reaction at Ec=l 1.8 kcal/mol. The geometry of the transition state is also shown schematically (right-hand-side); the departure directions of the H and HSO products with respect to the incoming Ο atom are indicated by arrows.

be conveniently detected in CMB experiments with mass spectrometry detection, high resolution is needed in angular and velocity distribution measurements. From the data shown in Fig. 13 we have concluded that reaction (4a) is direct and that the barrier is located in the exit channel. Once again, the initial L, which is essentially the total angular momentum of the system, is expected [4] to be carried away mainly as rotational motion of HSO. Thus J', the final total angular momentum, should be strongly correlated with L, and, on the other hand, L' will be determined mainly by

33

geometric relations between HSO and the departing H. If a large fraction of the exit potential energy barrier («4 kcal/mol for this reaction) is released as translational energy, as it is the case here, it is possible that L' largely comes from this repulsive energy release and could be quite substantial if the exit impact parameter is not small. Ab initio calculations [136] of the 0(3P) + H2S triplet PES and molecular-orbital considerations [128, 137] predict that dissociation of the transition state occurs through a planar configuration following the electrophilic out-of-plane attack of the oxygen atom on the lone pair of the sulphur atom. The H atom is expected to depart in a direction lying in the HSO plane to minimize the exchange repulsion and with a direction of about 45° with respect to the incoming oxygen, the HSH angle being about 90° (see right-hand-side of Fig. 13). As a consequence of linear momentum conservation, the detected heavier HSO product will recoil in the opposite direction, forming an angle of about 135° with respect to the incoming Ο direction. This is exactly what we observe experimentally (see Fig. 13). We think that this is a good example of how the shape of the product cm. angular distribution of a direct Η atom displacement reaction can mirror the geometry of the transition state. Another example of this type will be discussed in Section 5.1.

Determination of thermochemical quantities through the analysis of the high-energy fall-off in the product translational energy distribution is another outcome of CMB experiments. For reactions which are thermoneutral or weakly exoergic, the collision energy is a large fraction of the translational energy of the products, so that, with proper energy conservation considerations, one can trace back the minimum exothermicity of the reaction. The accuracy of the determination depends on the narrow velocity distributions of the reactants and on high-resolution of the TOF method for the measurements of velocities. We obtain for reaction (4a) an exothermicity of ΔΗ°=-4.0±0.7 kcal/mol, from which we deduce AH°f(HSO)=-0.9± 0.7 kcal/mol. Our estimate agrees with a previous similar determination [128], and reduces significantly the uncertainty. A recent theoretical estimate derived from ab initio calculations gives AHJf(HSO)= 0.3±3 kcal/mol [133].

In order to gain information on the SO + H2 reaction channel, angular and velocity distribution measurements were carried out at m/e=48. At the low E c these distributions are superimposable to those measured at m/e=49, and demonstrate that reaction (4c) and (5c) are not occurring, since clearly the detected SO + is a daughter ion coming from ionization of HSO formed in reaction (5a). At the highest E c, instead, the m/e=48 distributions were found to be different from those at m/e=49; specifically, the intensity of the peak attributed to the 0( 3P) contribution is reduced with respect to that arising from 0(1 D), but the overall width of the distributions is the same. This may be rationalized quantitatively assuming that the HSO products from the 0( 3P) and 0(*D) reactions fragment differently during ionization because of their very different internal energy content. In conclusion, SO formation (reactions (4c) and (5c)) does not seem to occurr appreciably for EC<11.8 kcal/mol.

34

This is not surprising since a three/four center elimination of molecular hydrogen is expected to occur through a very high energy barrier.

The results obtained from this study on 0( 3P, 1D) + H2S -» HSO + H demonstrate that the triplet reaction proceeds with a substantial energy barrier via a very short-lived, weakly bound, triplet intermediate and is characterized by a direct (rebound) mechanism, whereas the singlet reaction proceeds with virtually no barrier via a singlet surface that correlates with a strongly bound intermediate (presumably the stable HOSH molecule) and therefore is characterized by a mechanism forming a long-lived complex.

As already pointed out in Section 3.2, although 0(lD) reactions are thought to occur predominantly via insertion and to proceed via a deep singlet surface, formation of a long-lived complex has never been observed previously in dynamical studies of OH forming channels. A possible explanation has been put forth in the context of the 0(*D) + HX reactions (see previous Section). However, in contrast to the 0(*D) + HX systems, where the difference in exoergicity between the H displacement and the H abstraction channels is always very large (40-J-50 kcal/mol), in the 0(1 D) + H2S system, the formation of OH is only 8 kcal/mol more exoergic than that of HSO. If HSO formation goes through a long-lived complex via the ΓΑ' ground state surface (see Fig. 8), one would reasonably expect OH formation as well from the decomposition of a long-lived HOSH complex, if reaction proceeds via the *A' surface. Indeed, dynamical studies of OH energy partitioning from the reaction 0(1D) + H2S, have revealed that two different micro-mechanisms lead to two different OH internal state distributions, one statistical and the other highly non-statistical [31, 36]. The results indicate that the overall product vibrational energy distribution is bimodal, with a minimum at OH(v-l). As discussed by Sloan and co-workers [36], the two different micro-mechanisms giving rise to OH production would consist in a direct Η atom abstraction, which creates vibrationally excited OH, and in an insertion forming a long-lived HOSH (or H2SO) intermediate, which decomposes statistically after energy randomization, without producing much product vibrational excitation. The direct abstraction would predominantly populate OH(v'>l) and the insertion-elimination OH(v'<l). No proof, however, was given on the basis of the existing data. Clearly, dynamic calculations on realistic PES are desirable and would help to clarify the overall picture and elucidate the role played by the different potential surfaces.

5. REACTION DYNAMICS OF ΟΗ(2Π) RADICALS

Among free radical reactions, those of hydroxyl radicals are of great importance because of their primary role in atmospheric and combustion chemistry. The OH radicals dominate the chemistry of the troposphere in the same way that oxygen

35

atoms and ozone dominate the chemistry of the stratosphere [138]. Production of OH in the troposphere is driven by the UV photolysis of ozone

O3 + hv -> 0 ( ! D ) +O2

followed by the fast reaction

0 ( !D) + H 2 0 H> 2 OH

The radical OH is highly reactive and is involved in a great number of atmospheric chemical changes. In the unpolluted troposphere about 70% of the OH radicals react with carbon monoxide:

OH + CO -> CO2 + H (7)

and this represents one of the most important reactions in all tropospheric chemistry. In the troposphere OH is also involved in many free radical chain reactions which oxidize H2, CH4 and other hydrocarbons to H2O.

In combustion chemistry reaction (7) represents the second most important reaction, acting as the last step in the oxidation of hydrocarbons [139]. The oxidation of H2 by OH:

OH + H 2 -> H 2 0 + H (8)

is also extremely important, since it is the chain propagation step in hydrogen combustion [139].

In addition to their practical importance, reactions (7) and (8) are also of considerable theoretical interest. After the impressive progress, during the last few years, in both experiment [8, 10a, 15, 48] and theory [49] toward the solution of the atom + diatom reactive scattering problem, attention is now being focused on four-atom (diatom + diatom and atom + triatom) systems [50-69]. The solution of the quantum mechanical problem of reactive scattering involving four atoms can be considered the challenge of the 1990s. Obviously, the upmove in complexity from A + BC is essential to pave the way for the understanding of more complex chemical systems. In this context, the reaction OH + H2 can be considered as prototype of four-atom reactions, involving three hydrogen atoms and an atom of the first row of the periodic table, and therefore offers favorable possibilities for ab initio studies using quantum-mechanical methods. Also the reaction OH + CO has become a test case for the family of four-atom reactions.

High-quality ab initio calculations of the potential energy surface for reaction (7) and (8) have become available [71, 72]. Quasiclassical trajectory calculations of the dynamics of four-atom reactions have been implemented a number of years ago

36

[73-78]. Very recently, also 3-D quantum-mechanical dynamical calculations have been carried out on four-atom systems, with still some degree of approximation however; rate constants and state-specific integral cross sections were calculated [50-52, 81, 82]. In an effort to provide detailed data for a test of the ab initio potential energy surfaces, we have performed reactive differential cross section measurements for reactions (7) and (8). This has stimulated Clary and Schatz to carry out dynamical calculations of this observable, so that detailed comparison between experiment and theory has become feasible. Exact quantum dynamical calculations for OH + H 2 appear to be on the horizon, since they will shortly be within reach of modern computer capabilities: hence, the need for detailed experimental information on the dynamics of four-atom reactions. In this lies one of the major motivations of the experimental work on the OH + H 2 and OH + CO reactions very recently carried out in our laboratory and discussed below.

5.1. The reaction OH + H2 The reaction

OH + H 2 -> H 2 0 + Η ΔΗ°= -14.8 kcal/mol (8)

has been the object of extensive experimental and theoretical studies. Its energetics are outlined in Fig. 14, where the reaction correlation diagram is shown schematically. Kinetic bulk experiments have measured rate constants and isotope effects over an extended range of temperatures (250-1050 K) [140, 141]. Reviews of the kinetics of this reaction are available [142]. At room temperature the reaction is very slow (k29g= 6.1x10 ~15 cc molecule "1 s "1, corresponding to a cross section of 3.6x10~4 A2), but exhibits a non-linear Arrhenius behavior with a pronounced upward curvature at high temperature [140, 141]. The very large isotopic effect on the rate constants suggests a strong increase in the importance of tunneling contribution as temperature decreases. The experimental activation energy is 4.0 kcal/mol for OH + H 2 (5.3 kcal/mol for OH + D2) [141]. The effect of initial vibrational excitation of the OH and H 2 molecules on the reaction rate was explored in state-resolved kinetic studies [143-145].

The importance and the prototypical nature of this reaction have also encouraged a variety of state-of-the-art dynamical experiments in recent years. All of them, however, have focused on the reverse endothermic reaction H + H20—> OH + H 2 (see Fig. 14). By exploiting the hot Η-atom technique, absolute reactive cross sections and OH internal state distributions were measured in the energy range from 33 to 58 kcal/mol by Wolfrum and co-workers [53a], by Honda et al. [53b] and by Kessler and Kleinermanns [53c]. Good agreement with theoretical results based on quasiclassical trajectory calculations [73, 74] on an ab initio potential surface was found in all these studies. While the spin-orbit components of OH(2]l) are populated statistically, a remarkable non-statistical partitioning over the A-doublet

37

5 0 p Kleinermanns a Wolfrum (1984)

40 k-

3 0 -„ 4 y0 H

- 2 0 L

Figure 14. Energy level and correlation diagram for the OH + D 2 reaction. Arrows from the right indicate the energies at which the reverse HOD + D (or H 2 0 + H) reaction has been investigated (see text). Main investigators are also specified. The arrow from the left indicates the collision energy of the CMB study of the direct reaction.

sublevels was observed. This has been intepreted in terms of a direct mechanism in which the reaction barrier is located in the exit channel. The preferred population of the A' state of OH (unpaired electron in the plane) indicates that the Η-atom attacks H 2 0 in the molecular plane and the resulting torque explains the high rotational excitation. Recently, a series of clever experiments involving selective initial excitation of the local OH stretching mode of H 2 0 (HOD) have also been carried out. Crim and co-workers [54] found a dramatic mode selectivity in thermal H-atom reactions with highly excited vibrational states of H 2 0 and HOD. Zare and co-

38

workers [55] found large differences in the product branching ratios for the HOD(OOl) and HOD(IOO) reactions with hot Η-atoms, depending on whether the OH or OD local mode was vibrationally excited. These effects have been examined theoretically by Kudla and Schatz, using quasiclassical trajectories [76], and by Clary [81] and by Bowman and Wang [82], using reduced dimensionality quantum mechanical treatments. Elegant as these dynamic experiments are, they do not provide a sensitive probe of the details of the potential energy surface along the minimum energy path. Very recently, inelastic ro-vibrational excitation of H 2 0 by 51 kcal/mol H atoms was investigated [56] and theoretically examined [76b].

Very limited experimental information is instead available on the dynamics of the direct process. The only information available comes from early flash photolysis studies [146] coupled to time resolved infrared detection, which did not observe any infrared emission from H 2 0 formed in reaction (8). From these studies only upper bounds to the fraction of available energy released into vibrational excitation could be inferred. In particular, Trainor and von Rosenberg [146] concluded that less than 11% of the reaction exothermicity was being partitioned into H 2 0 symmetric and asymmetric stretch excitation, and less than 18% into bend excitation. However, later studies of H 2 0 vibrational relaxation [147] suggested that the actual upper bounds are significantly higher (see Ref. 73). Inelastic collisions of OH by H 2 have been investigated in crossed beams [148] in connection with their importance in astrophysics (OH-maser).

High-quality, large-scale polarization configuration interaction calculations have been performed by Walch and Dunning [70] on the PES. The calculated barrier height is 6.2 kcal/mol (the vibrationally adiabatic threshold is 5.9 kcal/mol) and the saddle point is found to have a coplanar geometry and to be located in the entrance channel. Schatz and Elgersma [73] obtained an analytic PES, based on a fit of the ab initio points, and carried out quasiclassical trajectory calculations to determine product vibrational distributions. The same authors examined also the rotational, angular and projection distributions of the H 2 0 product [75]. Schatz also explored the effects of reagent vibrational excitation on reactivity [74]. Transition state theory has been applied to this reaction, using the same surface, by Isaacson and Truhlar to compute rate constants for reaction (8) and its isotopic analogue [149]. Some inadequacies (spurious wells) of the Schatz-Elgersma surface, in the asymptotic reactant region, were removed by Rashed and Brown [78a], who, using the modified surface, carried out quasiclassical trajectory calculations on the effect of the reagent energy distribution (translational temperature, and vibrational and rotational energies) on reactivity for reaction (8). The effect of reagent rotations was put under further scrutiny by Harrison and Mayne [78b], who used the Schatz-Elgersma and Rushed-Brown surfaces and also a modification of them, after eliminating some still remaining artifacts. The main conclusion of all these studies was that translation is the most efficient form of energy for promoting reactivity,

39

vibrational excitation of H 2 has a moderate effect, while vibrational excitation of OH and rotational excitation of either or both reagents has a very little effect on reactivity.

Recently, quantum mechanical scattering calculations of integral cross sections and rate constants were performed for the first time by Clary [51, 52] on both OH + H 2 and OH +D2, using the Schatz-Elgersma surface. It appears, from the generally good agreement with the experiment of the calculated rate constants for the direct reactions and of state-selected cross sections and product state distributions for the reverse reactions, that the Schatz-Elgersma potential surface is quite reliable, in spite of its above mentioned inadequacies (see Ref. 150).

We have undertaken a direct experimental investigation of the dynamics of reaction (8). For obvious reasons of simpler detection, we have looked at the isotopic variant:

OH + D 2 -» HOD + D.

In a first experiment [57] we have measured the angular and velocity distributions of the HOD product at a relative collision energy, Ec, of 6.3 kcal/mol. At this energy the integral reactive cross section is about 0.4 Â2 [52]. In order to improve the signal-to-background ratio, isotopically labeled water, H 2l^o, was used to generate l^OH. This allowed us to detect the HOD product at m/e=21, which has a very low inherent background in the detector.

The laboratory angular distribution for the HOD product is shown in Fig. 15, together with the most probable Newton diagram. The lab angle Θ is measured from the OH beam. In the center-of-mass (cm.) coordinate system, û=0° is the direction of the OH beam and represents the forward direction with respect to OH. The circle in the Newton diagram (see Fig. 15) represents the maximum cm. speed for HOD product assuming that the OH reactant is in v=0 and that all the available energy goes into product translational energy. The HOD angular distribution peaks sharply to the right of the cm. angle, indicating that the product is thoroughly back-scattered. Its space distribution is, at first sight, completely confined within the angular range predicted for OH in v=0, while OH(v=l) would originate a significantly larger Newton circle, leading to a lab angular range from ©=-15° until 0=36°.

Time-of-flight spectra were recorded at five different laboratory angles. The product angular distribution and TOF spectra were fit by the usual forward convolution procedure using a separable form for the cm. frame product flux distribution I c m( # , Ε') = Τ(ύ)χΡ(Ε'). The continuous line in Fig. 15 is the lab angular distributions calculated from the best-fit cm. functions. The best-fit cm. angular distribution is depicted as a solid line in Fig. 16. One can immediately see how the HOD product cm. angular distribution peaks in the backward direction (ύ =180°). This is the expected results for a direct chemical reaction dominated by

40

Figure 15. HOD product laboratory angular distribution from the OH + D 2 reaction at Ec=6.3 kcal/mol. The circle in the Newton diagram (drawn for OH(v=l)) delimits the maximum HOD speed when all the available energy is assumed to go in translation. The solid line represents the calculated angular distribution with best-fit cm. translational energy and angular distributions.

41

OH + D 2 • HOD + D

Ε = 6 . 3 k c a l / m o l c

C M . Scat ter ing A n g l e , #

Figure 16. Center-of-mass angular distribution (continuous line) of the HOD product from the OH + D 2 reaction at Ec=6.3 kcal/mol. The area within dashed lines represents the limits of cm. angular functions which still give acceptable fits to the data. Dots: theoretical results from quantum-mechanical scattering calculations by Clary [84].

collinear, or nearly collinear, geometries. The product cm. angular distribution observed bears strong similarities with that measured for the isoelectronic F + D 2

reaction [8] and for D + H 2 [48], and with that calculated for CI + H 2 [151], suggesting that the PES of these systems have a similar topology. All the above abstraction reactions appear to proceed dominantly through collinear, or nearly collinear, geometries, and the nearly collinear transition state gives rise to pronounced rebound dynamics, at least at low collision energies. The OH + H 2

42

reaction is therefore a simple hydrogen transfer reaction in which formation of the new bond occurs simultaneously with the breakage of the old bond.

For this reaction too it is possible to infer the geometry of the transition state directly from the marked anisotropy of the cm. angular distribution (see Sec. 4 and Ref. 122 and 137).Theoretical ab initio calculations [70] predict that the preferred geometry of the transition state is that shown in Fig. 17. If this is true, as a consequence of a direct collision the D atom is expected to depart in a direction lying in the HOD plane [56] and forming an angle of about 15° with respect to the incoming OH. For linear momentum conservation, the detected HOD will recoil in the opposite direction, forming an angle of about 165° with respect to the incoming OH (see Fig. 17). This is very much like what we observe experimentally, as can be seen in Fig. 16, where the experimental cm. angular distribution is not sharply peaked at θ=180°, but extends almost flat within 160° and 180°.

Figure 17. The geometry of the ab initio transition state is shown schematically. The incoming direction of OH, indicated by an arrow, defines û=0°. The velocity vectors of the departing D and HOD products are also represented, m and u stand for mass and cm. velocity, respectively.

About 32% of the total available energy is found to be channelled into translation. This suggests that a large fraction of the available energy is released into vibrational excitation, since large rotational excitation is not expected for nearly collinear dominated reactions. Similar effects were also observed in the related F + D 2 reaction [8].

It is interesting to compare the present results with those of dynamical calculations on the ab initio surface. Quasiclassical trajectory calculations [75] of the cm. angular distribution of the H 2 0 product from OH + H 2 at a translational energy of 4.6 and 6.9 kcal/mol, show good qualitative agreement with the HOD

43

distribution measured in the present experiment at 6.3 kcal/mol. The calculated distribution is also completely confined between 90° and 180° in the cm. frame. The calculated fraction of energy in translation varies from about 35% to about 50%, depending on the method of partitioning product vibrational states [73-75]. However, extension of these calculations to the OH + D 2 reaction at the energy of the present experiment are desiderable for direct quantitative comparison. Schatz and coworkers [80] are currently performing these calculations.

Very recently, using the latest version of his quantum method, Clary [83, 84] has extended his calculations to the differential cross section for the OH + D 2

reaction at several energies including the energy of our experiment. He finds a completely backward scattered HOD product at Ec=6.3 kcal/mol similar to that determined experimentally in the present work, falling within the experimental error of our determination (see Figure 16). Instead, the calculated fraction of energy released in translation is 0.70, which disagrees remarkably with respect to the experimental value of 0.32. This disagreement suggests that the potential energy surface has some shortcomings and may need some improving, or that some of the approximation in Clary's quantum method are not warranted, or both. Preliminary calculations by Clary [84] on a fitted surface in which the OH bond length is closer to the ab initio value, find much more product vibrational excitation than obtained on the Schatz-Elgersma surface. There is clearly a need for exact 3-D quantum scattering calculations on a more accurate potential energy surface. A new potential surface for modeling the dynamics of the OH + H 2 reaction has been recently proposed by Isaacson [150], This new potential form not only reproduces the ab initio information of Walch and Dunning [70], but also the new barrier shape computed by Dunning et al. [152]. It differs somewhat from the earlier form devised by Shatz and Elgersma and until now used in quasiclassical trajectory and also quantum dynamical calculations. Since it was found [150] that both barrier shape and the degree of reaction path curvature strongly influence the rate constants, it would be interesting to examine the effects on the dynamics as well. Very large scale, high-quality, ab initio calculations of a new potential energy surface for the two isoelectronic reactive systems OH + H 2 and F + H 2 are currently being performed by Werner [153]. Exact quantum dynamical calculations of the double differential cross sections for OH + H 2 may shortly be within reach of modern computer capabilities.

5.2. The reaction OH + CO The energetics of the reaction

OH + CO -> C 0 2 + Η ΔΗ° = -24.5 kcal/mol (7)

are outlined in Fig. 18, where the reaction coordinate diagram is shown schematically. Rate constants have been determined over a wide range of

44

temperatures and pressures [58a-b, 154]. The rate constant shows an interesting non-Arrhenius behavior with a very pronounced upward curvature above 500 K. This has been explained by the competition between forward reaction, redissociation and stabilization of an energized HOCO intermediate on the potential surface for the reaction [58, 154] . The first detection of the HOCO species in low temperature matrices dates back about 20 years [155], but it was not until very recently that spectroscopic observation of HOCO in gas-phase was made: roto-vibrational spectra were recorded and accurate structural information was provided [156]. Thermochemical information (i.e., the HO-CO bond energy) was also obtained recently from a photo-ionization mass spectrometry investigation [157]. Kinetic studies of reaction (7) at some state selected level have also been carried out [58a, 146, 158].

Recently, an explosion of experimental and theoretical interest at the dynamical level toward this reaction occurred, but most of the work was directed to the reverse endothermic reaction H + CO2 -> OH + CO (see Fig. 18). A review updated to 1989 can be found in Ref. 67. Using photolytically produced "hot" H atoms, absolute reactive cross sections and OH internal state distributions were measured in the energy range from 30 to 60 kcal/mol in gas-phase by Wolfrum and co-workers [60] and by Wittig and co-workers [62, 63] both in gas-phase and in van der Waals complexes formed in supersonic expansions, by using pump-probe laser techniques. In some cases the CO product was monitored [66]. Inelastic excitation of CO2 by H atoms was investigated [67] and, recently, state-resolved integral cross sections for the inelastic scattering of OH with CO were measured in a crossed beam study [159] Recently, Zewail and co-workers [68] in pioneering work proved that bimolecular reactions are accessible to time domain studies if one starts from van der Waals impacted reagents. The reaction H + CO2 was their first example. They looked in real time (using picosecond probing techniques) at the birth of OH (by LIF) following the breakup of the weakly bound CO2-HI complex by UV photolysis. More recently, Wittig and co-workers [64, 65] carried out similar experiments using femtosecond pulses. Lifetimes of the HOCO intermediate were determined as a function of photolysis wavelength, i.e., of the internal energy of the HOCO. Values ranging from 250 fs to about 4 ps were obtained [64, 65, 68] in the photolysis wavelength 235-263 nm. In these studies there exist some difficulties in the definition of the experimental conditions and also in the interpretation of the results.

Elegant as these dynamic experiments with hot Η-atoms are, they do not provide a sensitive probe of the details of the PES along the minimum energy path. Neverthless, all the above microscopic investigations of the H + CO2 reactive (and non) collisions have provided an astonishing amount of information that must all fit together to obtain the true picture of the dynamics of reaction (7).

As for the OH + H2 reaction, very limited experimental information is available on the dynamics of the direct process (reaction (7)). Early flash photolysis studies

45

4 0 r

30F

20

ο Ε

^ ιοί ο

our experiment

Quick a Tiee (1983)

Wolfrum ( I984 ) j

Rice a Baronavski (1991)

^ o l f r u m

(1985-89)

Zewail (1987-90)

Wittig (1986-93)

? 0 φ c

U J

OH+CO

•10

>0

-30 h -

- 4 0 1 trans-HOCO

C 0 2+ Η

c is -HOCO

Figure 18. Energy level and correlation diagram for the OH + CO reaction. Arrows from the right indicate the energies (or range of energies) at which the reverse reaction has been investigated (see text). Main investigators are also specified. The arrow from the left indicates the collision energy of the CMB study of the direct reaction.

[146, 158a], coupled to time-resolved infrared detection, did not observe any infrared emission from CO2 and concluded that CO2 is formed mainly in its ground vibrational state. In a very recent flash photolysis study Smith and coworkers [58 b,

46

c], using tunable diode laser absorption, found that only 6% of the total energy available to the H + CO2 product is channeled into vibrational energy at room temperature. This result was used to deduce the geometry of the transition state on the assumption that all the CO2 vibrational excitation starts as bending potential energy at the transition state; this leads to a more collinear HOCO than predicted theoretically. A partially state-selected integral cross section of 19 ±10 A 2 for reaction (7) was measured by Wolfrum [59] at a translational energy of 30.4 kcal/mol by detecting H atoms; but new measurements at the same energy give a value of about 2.4 A 2 [160].

Extensive ab initio calculations on the PES of the OH + CO reaction have been recently carried out [71, 72]. An empirical PES based on a fit to the ab initio points was used to carry out detailed quasiclassical trajectory studies [71b]. The theoretical surface exhibits no entrance channel barrier and an exit channel barrier of about 3 kcal/mol. Cross sections for HOCO complex formation, product energy partitioning, HOCO lifetime distributions, and thermal rate constants were calculated. The reaction appears to proceed through addition of OH to CO forming the trans HOCO isomer. Interconversion of trans HOCO to cis HOCO may occur readily, due to a low barrier, and product formation is predicted to occur only from the cis form. The saddle point for addition of OH to CO is early and broad, while the saddle point for elimination of H from HOCO is very narrow. The latter feature suggests the importance of tunnelling, which should play a significant role at low collision energies.

Despite the large body of experimental and theoretical data on the OH + CO and H + CO2 systems, the kinetics and dynamics of reaction (7) are still not well understood. Many questions are open, especially with regard to the microscopic dynamics of the direct process: (a) Is the reaction direct or is it going through a long-lived collision complex? (b) If it is going through a long-lived complex, what is its lifetime and what is the dynamics of its decomposition? (c) What is the energy partitioning? (d) What is the effect of the relative translational motion of the reagents on the reaction dynamics?

The direct experimental investigation of the dynamics of reaction (7) that we have recently undertaken, even if it is not completed yet, is providing usefiil results which address some of the above questions. We have measured angular and velocity distributions of the CO2 product. These results have stimulated quasiclassical and quantum-mechanical scattering calculations, so that direct detailed comparisons between experiment and theory have become feasible.

The angular distribution of CO2 product at Ec=14.1 kcal/mol is shown in Fig. 19, together with the most probable Newton diagram. In order to improve the signal-to-background ratio, experiments were performed using H 2 ^ 0 . This allowed us to detect CO2 at m/e=46, which has about two order of magnitude lower inherent background in the detector than m/e=44. As can be seen, the angular distribution exhibits a backward-forward structure with more intensity in the forward (with

47

0° 10° 20° 30° 40° 50° 90°

LAB SCATTERING ANGLE, ©

Figure 19. CO2 product laboratory angular distribution from the OH + CO reaction at Ec=14.1 kcal/mol. The solid line represents the calculated angular distribution with the best-fit cm. angular and translational energy distributions shown in the Figures 20 and 21, respectively.

respect to the OH beam) direction, and it is quite broad for such an heavy product left by a light Η atom. This already suggests that the reaction is proceeding through the formation of an intermediate complex whose lifetime is of the order of its rotational period, and that a large fraction of the available energy is released into translation. Product velocity distributions were obtained at ten angles by the cross-correlation TOF technique. With the usual procedure of forward convolution seen in previous Sections, translational energy and angular distributions in the cm. coordinate system were derived. The continuous line in Fig. 19 is the lab angular distribution calculated with the cm. angular and translational energy distributions depicted in the Figures 20 and 21. From Fig. 20, one immediately sees that the cm. angular distribution, T('d), is not symmetric, which indicates that the collision complex lives a period comparable to its rotational period. The asymmetry in Ί(ϋ) allows us to make an estimate of the lifetime of the collision complex within the'Osculating model" for chemical reactions [135]. According to this model, the asymmetry in Ί(β) is related to the ratio of the mean complex lifetime, τ, to its

48

1.0

0.5 H

0.0

OH + CO—[HOCO]*—C0 2 + H

Ε = 14.1 kca l /mo l

Ε xper iment

QCT ca le .

Quantum cale.

/

/

t - H - I · + - V~ J

0 ° 9 0 ° 180°

C M . Sca t t e r i ng A n g l e . #

Figure 20. Center-of-mass angular distribution (continuous line) of the CO2 product from the OH + CO reaction at Ec=14.1 kcal/mol. Dotted line: quasiclassical trajectory calculations by Kudla and Schatz [161]. Dashed line: Quantum-mechanical (approximate) calculations by Clary [84]. The same ab initio potential energy surface was used.

rotational period, t r , which is used as a clock, through the relation Τ(Φ=180°)/Τ(ΐ> =0°)=exp[-(Tr/2x)]. From the experimental ratio Τ(ϋ=180°)/Τ(ύ=0°)=0.63, the model predicts a mean lifetime of 1.08 rotational periods. A reasonable estimate of the rotational period î r =2rcI /L m ax

c an be m ad e . The moment of inertia of the rotating complex, I, is calculated from the ab initio geometry [71b], and the maximum angular momentum L m ax may be obtained from the maximum impact parameter b m ax for complex formation, Ε Γ η α χ= μ ^ ι η£ Χ, being μ the reduced mass

49

and ν the relative velocity. Kudla and Schatz [161], using quasiclassical trajectories on the ab initio surface, found that the maximum impact parameter for the OH+CO reaction is about 1.6 Â at 14 kcal/mol translational energy, with the opacity function being approximately flat, and dying out to zero at 1.6 Â. Then a lower limit of 0.64 ps for the mean complex lifetime is obtained.

It is interesting to compare this value of the complex lifetime with other experimental and theoretical estimates. From their quasiclassical trajectory studies, Kudla et al. [71b] obtained an HOCO lifetime of 0.73 ps at a translational energy of 13.84 kcal/mol. Zewail and co-workers [68], in their experiments on the reverse reaction at an energy of 14 kcal/mol above the OH + CO asymptote, obtained a lifetime of 3.9 ps and, more recently, Wittig and co-workers [64, 65], in slightly

OH + CO • C 0 2 + H

Ec= 14.1 kca l /mol

E x p t .

Translational Energy , E | (kcal/mol)

Figure 21. Center-of-mass CO2 product translational energy distribution. The total energy available Εχ0χ is the sum of the reaction exoergicity (-ΔΗ) and the nominal collision energy Ec, plus the estimated internal energy of the reagents («0.4 kcal/mol). Dots: quasiclassical trajectory calculations by Kudla and Schatz [161] on the ab initio potential energy surface.

50

different experimental conditions and at a comparable energy, reported a value of 0.6 ps. The complex lifetime should be independent from the way the complex is formed, the only thing that matters is energy and angular momentum. In experiments involving initial photolysis of van der Waals complexes it is not possible to assign a well defined value of the HOCO energy because of the so called "squeezed-atom" effect [63-65].

The large fraction of available energy released into translation (about 64%, corresponding to 25 kcal/mol) indicates the existence of strong repulsive forces between H and CO2, after that the saddle point is passed (see Figure 18). The repulsion in the exit channel disposes essentially all the potential energy associated with the high exit barrier (see Fig. 18) into translational motion of the products and, therefore, only 36% of the available energy goes in internal degrees of freedom. For the specific mass combination of reaction (7), angular momentum partitioning arguments [4] predict L«j\ From trajectory calculations the maximum impact parameter for reaction at E c~14 kca/mol is 1.6 Â , which corresponds to a maximum initial orbital angular momentum Lm a x~94fi. Assuming j'=L=94 ft, one gets an average rotational energy <E'R>«9.8 kcal/mol, that is about 25% of the total available energy. This result leaves only 11% of the total available energy which can be carried away in vibrational degrees of freedom, with the conclusion that essentially only the CO2 bending (010) can be excited, with the symmetric stretch (100) and two quanta of bending (020) being barely accessible. Indeed, very little CO2 vibrational excitation (<6% of the available energy, corresponding to the first bending only) has been observed by Smith and co-workers [58b,c] in flash-photolysis studies. The very weak vibrational excitation suggested by these results and the flash-photolysis results appears to support a transition state HO-C-0 geometry more collinear than indicated by the ab initio calculations. The high rotational excitation determines also the mild polarization of the cm. angular distribution and is consistent with the bent H-OCO geometry of the ab initio transition state [71b], in which the H atom leaves at about 45°, with respect to the center of mass of the HOCO, thus exerting a strong torque to the CO2 moiety.

Very recently, Kudla and Schatz [161] have performed extensive quasiclassical trajectory calculations of the reactive differential cross section for OH + CO on the ab initio surface. They calculated the C 0 2 angular and translational energy distribution at the energy of the present experiment. Preliminary comparisons show quite a good agreement with the experimental results (see Figs. 20 and 21). The calculated average fraction (0.65) of energy in translation is practically equal to the experimental one (0.64). Clary [84] has recently carried out 3-D dynamical calculations within an approximate quantum mechanical scheme on the same potential surface, at several translational energies including that of our experiment. He finds that 65% of the available energy is released in translation, in agreement with the experiment. The calculated angular distribution, instead, is sharply polarized at ϋ=0° and 180° and appears to reproduce qualitatively only the

51

backward-forward structure (see Fig. 20). The extreme polarization of the calculated T(d) appears to be determined by the resonance-like structure which is exhibited by the quantum reaction probability function, which translates in a very small number of partial waves contributing to the reactive scattering. Larger scale quantum dynamical calculations, beyond the three degrees of freedom model presently employed and which include more internal quantum states of the reagents, are clearly needed.

6. FUTURE DEVELOPMENTS

The crossed molecular beams scattering method will undoubtedly continue to play a very important role in future investigations of reaction dynamics. There are obvious extensions of the work carried out so far, such as investigating a variety of other reactions of 0(lD), from the simple one with H2 in which the resolution of the OH product vibrational states can be attained, to those with chloro-bromo-fluoro-carbons for the channels leading to CIO and BrO formation. The effect of electronic excitation on atomic oxygen reaction dynamics can be explored in other systems too, such as 0(3p, I D ) + CI2, CH3SH, etc.. Complementary information is expected to come from laser spectroscopic investigation and/or infrared-chemiluminescence studies and, obviously, from theory.

The four-atom prototype reactions OH + H2 and OH + CO need further attention from both the theoretical and the experimental side. In order to gather a broader data base for comparison with theory, reactive differential cross section measurements will be carried out in our laboratory at more collision energies and on the isotopic variants. It would also be useful to measure the internal energy partitioning of the H2O product in the reaction OH + H2 for which the CMB results indicate a high degree of internal (essentially vibrational) excitation. This implies that the H2O product should chemiluminesce in the infrared. The reaction could then be studied, at least in principle, using the FTR-FTS-IC method. In order to detect the ground vibrational level population too, use of suitable spectroscopic, laser based, detection schemes will be required. This will most likely have to await proper laser sources and could be carried out in pump-probe type of experiments in cell or, albeit with more difficulty, in crossed beams. Similar things can be said in regard to the OH + CO reaction, for which possibilities of state-specific detection of CO2 by REMPI may become feasible in the near future following the exciting developments in ultra-high resolution VUV lasers [162]. We may expect to see dynamical investigations of the OH + H2 and OH + CO reactions using the new ion-imaging technique [24, 25], with H atom detection by REMPI and, when the required laser sources become available, state-specific detection of the molecular products will surely also be attempted. Further detailed information on the reverse of the OH + CO reaction is expected to come from dynamical studies in the time-domain. Of course, the goal of a detailed characterization of the PES for these

52

important reactions will also rely on theoretical effort, both at the level of electronic structure and dynamical calculations.

It should be possible to extend the investigation to the dynamics of other interesting reactions of OH, such as those with hydrocarbons (acetylene and methane) and with halogen-containing molecules.

There are also other atomic and radical species that one can reasonably think of generating using the radio-frequency discharge beam source. Clearly the discharge method is not a very selective way of making a beam of a specific radical, and one has to make sure that other constituents in the beam do not interfere in the reaction under study, but it has great advantages in the production of a continuous and intense supersonic beam.

Future developments in the area of radical beam generation are to be expected along the line of UV laser photodissociation of a suitable precursor in pulsed beams, which is capable of leading to a selective formation of many radicals. Recent work has the promises that using a high-repetition rate UV laser and the pulsed supersonic beam method, a pseudocontinuous beam of an internally cold radical with high purity, can be developed [163].

On the theoretical side, the formalism developed for handling the reactions of atoms and small radicals is being implemented on increasingly larger systems for comparison with experiment. The calculation of reliable PES by ab initio means remains a challenging endeavour, but continuous progress is to be expected. From the wealth of information that is being provided by chemical dynamics, kinetic studies and theory, we may expect an increasingly deeper understanding of atom and small radical reactions, which is essential to pave the way to understanding complex chemical systems.

ACKNOWLEDGMENTS The authors wish to thank all their colleagues of the Molecular Beam Group of

Perugia for the useful discussions, collaborative atmosphere and continuous encouragement. Our deep gratitude goes to all coworkers over the years whose names are listed with ours in the bibliography of this review. We thank G.C. Schatz and D.C. Clary for useful discussion and permission to report some of their results prior to publication. Financial support from the Italian "Consiglio Nazionale delle Ricerche - Progetto Finalizzato Chimica Fine" and "Ministero Universita e Ricerca Scientifica", is gratefully acknowledged. Grants from EEC, ENEA and NATO made possible several aspects of the research reported here.

REFERENCES 1 R.D. Levine and R.B. Bernstein, Molecular Reaction Dynamics and Chemical

Reactivity, Oxford University, New York, 1987. 2 T.H. Bull and P.B. Moon, Discussion Faraday Soc, 17 (1954) 54; E.H. Taylor

and S. Datz, J. Chem. Phys., 23 (1955) 1711; S. Datz and E.H. Taylor, in:

53

Recent Research in Molecular Beams, I. Estermann (ed.), Academic, New York, 1959, p. 157.

3 E.F. Greene, R.W. Roberts, and J. Ross, J. Chem. Phys., 32 (1960) 940; D.R. Herschbach, J. Chem. Phys., 33 (1960) 1870; D.R. Herschbach, Adv. Chem. Phys., 10 (1966) 319, and references therein; K.T. Gillen, C. Tiley, and R.B. Bernstein, J. Chem. Phys., 50 (1969) 4019.

4 W.B. Miller, S.A. Safron, and D.R. Herschbach, Discussion Faraday Soc, 44 (1967) 108.

5 Y.T. Lee, J.D. McDonald, P.R. Le Breton, and D.R. Herschbach, Rev. Sci. Instr., 40(1969) 1402.

6 T.P. Schafer, P.E. Siska, J.M. Parson, F.P. Tully, Y.C. Wong, and Y.T. Lee, J. Chem. Phys., 53 (1970) 3385; J.L. Kinsey, Int. Rev. Sci. Phys. Chem., 9 (1972)173; J.M. Parson, K. Shobatake, Y.T. Lee, and S.A. Rice, Faraday Discuss. Chem. Soc, 55 (1973) 344; D.R. Herschbach, Faraday Discuss. Chem.Soc, 55 (1973) 233; R: Grice, Adv. Chem. Phys., 30 (1975) 247; J.M. Farrar and Y.T. Lee, Ann. Rev. Phys. Chem., 25 (1974) 357, and references therein; D.St.A.G. Radlein, J.C. Whitehead, and R. Grice, Mol. Phys., 29 (1975)1813; P.A. Gorry, C.V. Novikov, and R. Grice, Chem. Phys. Lett., 49 (1977) 116; A.M. Rulis and R.B. Bernstein, J. Chem.Phys. 57 (1972) 5497; S. Stolte, A.E. Proctor, and R.B. Bernstein, J. Chem. Phys., 65 (1976) 4990; J.J. Valentini, Y.T. Lee, and DJ. Auerbach, J. Chem. Phys., 67 (1977) 4866; JJ. Valentini, M.J. Coggiola, and Y.T. Lee, Faraday, Discuss. Chem. Soc, 62 (1977) 232; R.B. Bernstein, Adv. At. Mol. Phys., 15 (1979) 167; M.R. Levy, Prog. React. Kinet. 10 (1979), and references therein.

7 Y.T. Lee and Y.R. Shen, Physics Today, (November 1980) 1; CH. Becker, P.Casavecchia, P.W. Tiedemann, J.J. Valentini, and Y.T. Lee, J. Chem. Phys., 73 (1980) 2833; S.J. Sibener, R.J. Buss, P.Casavecchia, T. Hirooka, and Y.T. Lee, J. Chem. Phys., 72 (1980) 4341; R.J. Buss, R.J. Baseman, G. He, and Y.T. Lee, J. Photochem., 17 (1981) 389; G.N. Robinson, R.E. Continetti, and Y.T. Lee, Faraday, Discuss. Chem. Soc, 84 (1987) 25; R. Grice, Acc. Chem. Res., 14 (1981) 37; P.E. Davidson, G.L. Duncan, and R. Grice, Mol. Phys., 44 (1981) 1119; S.M.A. Hoffman, D.J. Smith, and R. Grice, Mol. Phys., 49 (1983)621; N.C. Firth, D.J. Smith, and R. Grice, Mol. Phys., 61 (1987) 859; M.M. Oprysko, F.J. Aoiz, M.A. McMahan, and R.B. Bernstein, J. Chem. Phys.,78 (1983) 3816.

8 D.M. Neumark, A.M. Wodtke, G.N. Robinson, CC. Hayden, and Y.T. Lee, J. Chem. Phys., 82 (1985) 3045; D.M. Neumark, A.M. Wodtke, G.N. Robinson, C C Hayden, K. Shobatake, R.K. Sparks, T.P. Shafer, and Y.T. Lee, J. Chem. Phys., 82 (1985) 3067.

9 H. Schmidt, P.S. Weiss, J.M. Mestdagh, M.H. Covinsky, and Y.T. Lee, Chem. Phys. Lett., 118 (1985) 539; M.F. Vernon, H. Schmidt, P.S. Weiss, M.H. Covinsky, and Y.T. Lee, J. Chem. Phys., 82 (1986) 3045.

54

10 (a) Y.T. Lee, Chemica Scripta, 27 (1987) 215; Y.T. Lee, Science, 236 (1987) 793; (b) D.R. Herschbach, Chemica Scripta, 27 (1987) 327; (c) J.C. Polanyi, Chemica Scripta 27 (1987) 229; J.C. Polanyi, Science, 236 (1987) 680.

11 R.B. Bernstein, Chemical Dynamics via Molecular Beam and Laser Techniques, Oxford University, Oxford, 1982.

12 H.W. Cruse, P.J. Dagdijan, and R.N. Zare, Faraday Discuss. Chem. Soc. 55 (1973) 277; R.N. Zare and P.J. Dagdijan, Science, 185 (1974) 739.

13 J.L. Kinsey, Ann. Rev. Phys. Chem., 28 (1977) 349; R.N. Zare and R.B. Bernstein, Phys. Today, 33 (1980) 43; S.R. Leone, Ann. Rev. Phys. Chem., 28 (1984) 109; J. Wolfram, Laser Chem., 6 (1986) 125.

14 D.H. Parker, in: Ultrasensitive Laser Techniques, D.S. Klieger (ed.), Academic, New York, 1983.

15 D. Neuhauser, R.S. Judson, DJ. Kouri, D.E. Adelman, N.E. Schafer, D.A.V. Kliner, and R.N. Zare, Science, 257 (1992) 519, and refences therein.

16 D.P. Gerrity and J.J. Valentini, J. Chem. Phys., 82 (1985) 1323; JJ. Valentini, P.A. Aker, G.J. Germann, and Y.D. Huh, Faraday Discuss. Chem. Soc, 91 (1991) 173, and references therein.

17 J.C. Polanyi and D.C. Tardy, J. Chem. Phys., 51 (1969) 5717. 18 J.H. Parker and G.C. Pimentel, J. Chem. Phys., 51 (1969) 91. 19 (a) P.J. Dagdijan, in: Atomic and Molecular Beam Methods, G. Scoles (ed.),

Oxford University, New York, 1988, and references therein; (b) D.Patel-Misra, D.G. Sauder, and P.J. Dagdijan, J. Chem. Phys., 95 (1991) 955; D. Patel-Misra and P.J. Dagdijan, J. Chem. Phys., 96 (1992) 3232; (c) D.M. Sonnenfroh, R.G. Macdonald, and K. Liu, J. Chem. Phys., 93 (1990) 1478; R.G. Macdonald and K. Liu, J. Chem. Phys., 93 (1990) 2443.

20 U. Hefter and K. Bergmann, in: Atomic and Molecular Beam Methods, G. Scoles (ed.), Oxford University, New York, 1988; L.J. Rawluk, Y.B. Fan, Y. Apelblat, and M. Kiel, J. Chem. Phys., 94 (1992) 4205; S.D. Jons, J.E. Shirley, M.T. Vonk, CF. Giese, and W.R. Gentry, J. Chem. Phys., 97 (1992) 7831.

21 L. Schnieder, K. Seekamp-Rahn, F. Liedeker, H. Steuwe, and K.H. Welge, Faraday Discuss. Chem. Soc. 91 (1991) 259; L. Schnieder and K.H. Welge, Abstracts XlVth Int. Symposium on Molecular Beams, Asilomar, CA, June 7-12,1992, p. 27.

22 J.L. Kinsey, J. Chem. Phys., 66 (1977) 2560; E.J. Murphy, J.H. Brophy, G.S. Arnold, W.L. Dimpfl, and J.L. Kinsey, J. Chem. Phys., 70 (1979) 5910.

23 B. Girard, Ν. Billy, G. Gonedard, and J. Viguè, J. Chem. Soc. Faraday Trans. 2, 85 (1989) 1270; J. Chem. Phys., 95 (1991) 4056; J.M. L'Hermite, G. Ramat, and R. Vetter, J. Chem. Phys. 93 (1990) 434.

24 D.W. Chandler, Abstracts 12th Int. Symposium on Gas Kinetics, Reading, UK, 19-24 July 1992, p. Al.

25 A.G. Suits, L.S. Bentuyan, P.L. Houston, and B.J. Whitaker, J. Chem. Phys., 96(1992) 8618.

55

26 M.Brouard, S.P. Duxon, P.A. Enriquez, and J.P. Simons, J. Chem. Phys., 97 (1992) 7414; Abstracts NATO Advanced Workshop on Orientation and Polarization Effects in Chemical Reaction Dynamics, Assisi, Italy, 20-25 November, 1992, p. 021.

27 F. Green, G. Hancock, and A.J. Orr-Ewing, Faraday Discuss. Chem. Soc, 91 (1991) 79; G. Hancock and D. Summerfield, Abstracts NATO Advanced Workshop on Orientation and Polarization Effects in Chemical Reaction Dynamics, Assisi, Italy, 20-25 November, 1992, p. 013.

28 T.L. Burks and M.C. Lin, Chem. Phys., 33 (1978) 327; J.E. Butler, L.D. Talley,G.K. Smith, and M.C Lin, J. Chem. Phys., 74 (1981) 4501; A.C Luntz,R. Schinke, W.A. Lester, Jr., and H.H. Gunthard, J. Chem. Phys., 70 (1979) 5908; A.C. Luntz J. Chem. Phys., 73 (1980) 1143.

29 A.C. Luntz, J. Chem. Phys., 73 (1980) 5393. 30 K.-H. Gericke, F.J. Comes, and R.D. Levine, J. Chem. Phys., 74 (1981) 6106;

K.-H. Gericke and F.J. Comes, Chem. Phys. Lett., 81 (1981) 218; W.A. Guillory, K.-H. Gericke, and F.J. Comes, J. Chem. Phys. 78 (1983) 5993; K. Miknlecky and K.-H. Gericke, J. Chem. Phys., 96 (1992) 7490, and references therein.

31 S. Klee, K.-H. Gericke, and F.J. Comes, Chem. Phys. Lett., 118 (1985) 530. 32 J.R. Wiesenfeld, Acc. Chem. Res., 15 (1982) 110, and references therein; G.M.

Jursich and J.R. Wiesenfeld, J. Chem. Phys., 83 (1985) 910; E.M. Goldfield and J.R. Wiesenfeld, J. Chem. Phys., 93 (1990) 1030; CR. Park and J.R. Wiesenfeld, Chem. Phys. Lett., 186 (1991) 170; J. Chem. Phys., 95 (1991) 8166; C.B. Cleveland and J.R. Wiesenfeld, J. Chem. Phys., 96 (1992) 248.

33 CR. Park and J.R. Wiesenfeld, Chem. Phys. Lett., 163 (1989) 230. 34 D.G. Sander, J.C. Stephenson, D.S. King, and M.P. Casassa, J. Chem. Phys.,

97 (1992) 952; S. Satyapal, J. Park, R. Bersohn, and B. Katz, J. Chem. Phys., 91 (1989) 6873; S.G. Cheskis, A.A. Iogansen, P.V. Kulakov, I.Yu. Razuvaev, O.M. Sarkisov, and A.A. Titov, Chem. Phys. Lett., 155 (1989) 37; J.F. Cordova, CT. Rettner, and J.L. Kinsey, J. Chem. Phys., 75 (1981) 2742.

35 J.J. Sloan and P.M. Aker, in: Springer Proceedings in Physics, Vol. 4, Time Resolved Vibrational Spectroscopy, A. Labereau and M. Stockburger (eds.), Springer, Berlin, 1985; JJ. Sloan, P.M. Aker, and B.I. Niefer, SPIE-Laser Appl. Chem., 669(1986) 169; P.A. Aker and J.J. Sloan, J. Chem. Phys., 85 (1986) 1412; P.A. Aker, B.I. Niefer, J.J. Sloan, and H. Heydtmann, J. Chem. Phys., 87 (1987) 203; P.M. Aker, J.J.A. O'Brien, and J.J. Sloan, J.Chem. Phys.,84 (1986) 745; J.J. Sloan, J. Chem. Phys., 92 (1988) 18.

36 P.M. Aker, J.J.A. O'Brien, and J.J. Sloan, Chem. Phys., 104 (1986) 421; E.J. Kruus, B.I. Niefer, and J.J. Sloan, J. Photochem. Photobiol. A: Chem., 57 (1991)419.

37 (a) E.J. Kruus, B,I, Niefer, and J.J. Sloan, J. Chem. Phys., 88 (1988) 985; (b) J.J. Sloan, E.J. Kruus, and B.I. Niefer, in Selectivity in Chemical Reactions,

56

J.C. Whitehead (éd.), Kluwer Academic, Dordrecht, 1988, NATO ASI Ser. C, Vol. 245, p. 117.

38 A. Lebehot, S. Drawin, F. Aguillon, R. Campargue, and X. Chapuisat, J. Chem.Phys., 92 (1990) 7340; D.C. Scott, J. de Juan, D.C. Robie, D. Schwartz-Lavi, and H. Reissler, J. Phys. Chem., 96 (1992) 2509.

39 K. Honma, Y. Fujimura, and O. Kajimoto, J. Chem. Phys., 88 (1988) 4739. 40 T.J. Sears, G.E. Hall, and J.J.F. McAndrew, J. Chem. Phys., 91 (1989) 5201. 41 J.M. Mestdagh, B.A. Balko, M.H. Covinsky, P.S. Weiss, M.F. Vernon, H.

Schmidt, and Y.T. Lee, Faraday Discuss. Chem. Soc, 84 (1987) 145; P.S. Weiss, J.M. Mestdagh, M.H. Covinsky, B.A. Balko, and Y.T. Lee, Chem. Phys., 126 (1988) 93; A.G. Suits, H. Hou, and Y.T. Lee, J. Phys. Chem., 94 (1990) 5672; A.G. Suits, H. Hou, H.F. Davis, Y.T. Lee, and M.J. Mestdagh, J. Chem. Phys., 95 (1991) 8178; A.G. Suits, H. Hou, H.F. Davis, and Y.T. Lee, J. Chem. Phys., 96 (1992) 2777. See also: C.T. Rettner and R.N. Zare, J. Chem.Phys., 77 (1982) 2416; R. Duren, U. Lackschewitz, ans S. Milosevic, Chem. Phys., 126 (1988) 81; F.X. Gadea, J.M. L'Hermite, G. Rahmat, and R. Vetter, Chem. Phys. Lett., 151 (1988) 183; J. Cuvellier, P. de Pujo, J.M. Mestdagh, P. Meynadier, J.P. Visticot, J. Berlande, and A. Binet, J. Chem. Phys., 90 (1989) 7050.

42 S.J. Sibener, R.J. Buss, C.Y. Ng, and Y.T. Lee, Rev. Sci. Instrum., 51 (1980) 167.

43 P. Casavecchia, R.J. Buss, S.J. Sibener, and Y.Y. Lee, J. Chem. Phys., 73 (1980) 6351; R.J. Buss, P. Casavecchia, T. Hirooka, S.J. Sibener, and Y.T. Lee, Chem. Phys. Lett., 82 (1981) 386.

44 J.A. Logan, C.A. Mims, G.W. Stewart, and J. Ross, J. Chem. Phys., 64 (1976) 1804; D.L. McFadden, E.A. McCullogh, F. Kalos, and J. Ross, J. Chem. Phys., 59 (1973) 121; D.L. McFadden, E.A. McCullogh, F. Kalos, W.R. Gentry, and J. Ross, J. Chem. Phys., 57 (1972) 1351.

45 CF. Carter, M.R. Levy, and R. Grice, Faraday Discuss. Chem. Soc, 55 (1973) 357; L.C Brown, J.C Whitehead, and R. Grice, Mol. Phys., 31 (1976) 1069.

46 I. Veltman, A. Durkin, D.J. Smith, J.H. Williams, and R. Grice, Chem. Phys. Lett., 113 (1985) 425; S.M.A. Hoffmann, D.J. Smith, N. Bradshaw, and R. Grice, Mol. Phys., 57 (1986) 1219.

47 G.N. Robinson, G.M. Nathanson, R.E. Continetti, and Y.T. Lee, J. Chem. Phys., 89 (1988) 6744.

48 R.E. Continetti, B.A. Balko, and Y.T. Lee, J. Chem. Phys., 93 (1990) 5719 S.A. Buntin, CF. Giese, and W.R. Gentry, J. Chem. Phys., 87 (1987) 1443; Chem. Phys. Lett., 168 (1990) 513; H. Buchenau, J.P. Toennies, J. Arnold, and J. Wolfram, Ber. Bunsenges. Phys. Chem., 94 (1990) 1231; D.P. Gerrity and JJ. Valentini, J. Chem. Phys., 81 (1984) 1298; J.C. Nieh and J.J. Valentini, J. Chem. Phys., 92(1990) 1083.

49 W.H. Miller, Ann. Rev. Phys. Chem., 41 (1990) 245, and references therein; M. D'Mello, D.E. Manolopoulos, and R.E. Wyatt, J. Chem. Phys., 94 (1991)

57

5985; S.L. Mielke, R.S. Friedman, D.G. Truhlar, D.W. Schwenke, Chem. Phys. Lett., 188 (1992) 359; Y.-S.M. Wu and A. Kuppermann, Chem. Phys. Lett., 201 (1993) 178.

50 A.N. Brook and D.C. Clary, J. Chem. Phys., 92 (1990) 4178; Q. Sun and J.M. Bowman, J. Chem. Phys., 92 (1990) 5201; G.D. Billing, Chem. Phys., 146 (1990) 63.

51 D.C. Clary, J. Chem. Phys., 95 (1991) 7298. 52 D.C. Clary, J. Chem. Phys., 96 (1992) 3656. 53 (a) K. Kleinersmanns and J. Wolfram, Appl. Phys. B, 34 (1984) 5; A. Jacobs,

H.R. Volpp, and J. Wolfram, 24th Symposium (Int.) on Combustion, Sydney, Australia, 1992, in press; (b) K. Honda, M. Takayanagi, T. Nishiya, H. Ohoyama, and I. Hanazaki, Chem. Phys. Lett., 180 (1991) 321; (c) K. Kessler and K. Kleinermanns, Chem. Phys. Lett., 190 (1992) 145.

54 A. Sinha, M.C. Hsiao, and F.F. Crim, J. Chem. Phys., 92 (1990) 6333; J. Chem.Phys., 94 (1991) 4928; A. Shina, J. Phys. Chem., 94 (1990) 4391; F.F. Crim, M.C. Hsiao, J.L. Scott, A. Sinha, and R.L. Van der Wal, Philos. Trans. R. Soc. London A, 332 (1990) 259; M.C. Hsiao, A. Sinha, and F.F. Crim, J. Phys. Chem., 95 (1991) 8263.

55 M.J. Bronikowski, W.R. Simpson, B. Girard, and R.N. Zare, J. Chem. Phys., 95 (1991) 8647; D.E. Adelman, S.V. Filseth, and R.N. Zare, J. Chem. Phys., in press (1993).

56 CM. Lovejoy, L. Goldfarb, and S.R. Leone, J. Chem. Phys., 96 (1992) 7180. 57 M. Alagia, N. Balucani, P. Casavecchia, D. Stranges, and G.G. Volpi, J.

Chem.Phys., 98 (1993) 2459. 58 (a) J. Branning, D.W. Derbyshire, I.W.M. Smith and M.D. Williams, J. Chem.

Soc. Faraday Trans. 2, 84 (1988) 105; (b) M.J. Frost, P. Sharkey, and I.W.M. Smith, Faraday Discuss. Chem. Soc, 91 (1991) 305; (c) M.J Frost, J. Salh, and I.W.M. Smith, J. Chem. Soc Faraday Trans., 87 (1991) 1037.

59 J. Wolfram, Faraday Discuss. Chem. Soc, 84 (1987) 191; Spectrochimica Acta, 46A(1990) 567.

60 K. Kleinermanns and J. Wolfram, Chem. Phys. Lett., 104 (1984) 157; K. Kleinermanns, E. Linnebach, and J. Wolfram, J. Phys. Chem., 89 (1985) 2525; A. Jacobs, M. Wahl, R. Weller, and J. Wolfram, Chem. Phys. Lett., 158 (1989) 161.

61 CR. Quick, Jr. and J.J. Tiee, Chem. Phys. Lett., 100 (1983) 223. 62 G. Radhakrishnan, S. Buelow, and C. Wittig, J. Chem. Phys., 84 (1986) 727. 63 S.K. Shin, Y. Chen, S. Nickolaisen, S.W. Sharpe, R.A. Beaudet, and C. Wittig,

Advances in Photochemistry, Vol. 16, D. Volman, G. Hammond, and D. Neckers, (eds.), Wiley, 1991, pp. 249-363, and references therein; S. Buelow, G. Radhakrishnam, J. Catanzarite, and C. Wittig, J. Chem. Phys., 83 (1985) 444; C. Wittig, Y.M. Engel, and R.D. Levine, Chem. Phys. Lett., 153 (1988) 411; C. Wittig, S. Sharpe, and R.A. Beaudet, Acc. Chem. Res., 21 (1988) 341; Y. Chen, G. Hoffmann, D. Oh, and C. Wittig, Chem. Phys. Lett., 159 (1989)

58

426; S.K. Shin, Y. Chen, D. Oh, and C. Wittig, Philos. Trans. R. Soc. London A, 332 (1990) 361; S.K. Shin, C. Wittig, and W.A. Goddard III, J. Phys. Chem., 95 (1991) 8048.

64 S.I. Ionov, G.A. Brucker, C. Jaques, L. Valachovic, and C. Wittig, J. Chem. Phys., 97(1992) 9486.

65 C. Jaques, L. Valachovic, S. Ionov, E. Bôhmer, Y. Wen, J. Segall, and C. Wittig, J. Chem. Soc. Faraday Trans., in press (1993).

66 J.K. Rice and A.P. Baronavsky, J. Chem. Phys., 94 (1991) 1006; S.L. Nickolaisen, H.E. Cartland, and C. Wittig, J. Chem. Phys., 96 (1992) 4378.

67 F.A. Khan, T.G. Krentz, G.W. Flynn, and R.E. Weston, Jr., J. Chem. Phys., 92 (1990) 4876, and references therein; G.W. Flynn, Science, 246 (1989) 1009; G.W. Flynn and R.E. Weston, Jr., Ann. Rev. Phys. Chem., 37 (1986) 551.

68 N.F. Sherer, L.R. Khundkar, R.B. Bernstein, and A.H. Zewail, J. Chem. Phys., 87 (1987) 1541; N.F. Sherer, C. Sipes, R.B. Bernstein, and A.H. Zewail, J. Chem. Phys., 92 (1990) 5239.

69 M. Alagia, N. Balucani, P.Casavecchia, D. Stranges, and G.G. Volpi, J. Chem. Phys., submitted.

70 S.P. Walch and T.H. Dunning, Jr., J. Chem. Phys., 72 (1980) 1303. 71 (a) G.C. Schatz, M.S. Fitzcharles, and L.B. Harding, Faraday Discuss. Chem.

Soc, 84 (1987) 359; (b) K. Kudla, G.C. Schatz, and A.F. Wagner, J. Chem. Phys., 95 (1991) 1635; (c) K. Kudla, A.G. Koures, L.B. Harding, and G.C. Schatz, J. Chem. Phys., 96 (1992) 7465.

72 M. Aoyagi and S. Kato, J. Chem. Phys., 88 (1988) 6409. 73 G.C. Schatz and H. Elgersma, Chem. Phys. Lett., 73 (1980) 21. 74 G.C. Schatz and S.P. Walch, J. Chem. Phys., 72 (1980) 776; G.C. Schatz, J.

Chem. Phys., 74 (1981) 1133; G.C Schatz, M.C. Colton, and J.L. Grant, J. Phys. Chem., 88 (1984) 2971; H. Elgersma and G.C. Schatz, Int. J. Quantum Chem. Symp., 15(1981)611.

75 G.C. Schatz and H. Elgersma, in Potential Energy Surfaces and Dynamics Calculations, D.G. Truhlar (ed.), Plenum, New York, 1981, p. 311.

76 K. Kudla and G.C. Schatz, Chem. Phys. Lett.; 193 (1992) 507; J. Chem. Phys., in press (1993).

77 G.C. Schatz and J. Dyck, Chem. Phys. Lett., 188 (1992) 11. 78 (a) O. Rashed and N.J. Brown, J. Chem. Phys., 82 (1985) 5506; (b) J.A.

Harrison and H.R. Mayne, J. Chem. Phys., 88 (1988) 7424. 79 G.C. Schatz, Latsis Symposium 1992 on Intramolecular Kinetics and Reaction

Dynamics, ΕΤΗ Zurich, 20-23 October 1992, p. 61. 80 G.C. Schatz, private communication. 81 D.C. Clary, Chem. Phys. Lett., 192 (1992) 34. 82 J.L. Bowman and D. Wang, J. Chem. Phys., 96 (1992) 7852; D. Wang and J.L.

Bowman, J. Chem. Phys., 98 (1993) 000. 83 D.C. Clary, Abstracts NATO Advanced Workshop on Orientation and

Polarization Effects in Chemical Reaction Dynamics, Assisi, Italy, 20-25

59

November, 1992, p. 036; I. Last, A. Baram, H. Shichman, and M. Baer, ibid., p. 034.

84 D.C. Clary, private communication. 85 Y.T. Lee, in Atomic and Molecular Beam Methods, G. Scoles (ed.), Oxford

University, New York, 1987, Vol. 1. 86 L. Beneventi, P. Casavecchia, and G.G. Volpi, J. Chem. Phys., 85 (1986)

701 l;in Electronic and Atomic Collisions (Invited Papers XVIIICPEAC, Brisbane, 1991), W.R. MacGillivray, I.E. McCarthy, and M.C. Standage (eds.),IOP, Bristol, 1992, p. 527, and references therein; L. Beneventi, P.Casavecchia, G.G. Volpi, CR. Bieler, and K.C Janda, J. Chem.Phys., 98 (1993) 178.

87 H.H. Lo and W.L. Fite, Chem. Phys. Lett., 29 (1974) 39; R.B. Cohen, P. Majeres, and J.K. Raloff, Chem. Phys. Lett., 31 (1975) 176.

88 P.A. Gorry and R. Grice, J. Phys. E: Sci. Intrum., 12 (1979) 857. 89 N. Balucani, L. Beneventi, P. Casavecchia, D. Stranges, and G.G. Volpi, in

Ecological Physical Chemistry, C Rossi and E. Tiezzi (eds.), Elsevier, Amsterdam, 1991, p. 539.

90 R.N. Dubinsky and D.J. McKenney, Can. J. Chem., 53 (1975) 3531. 91 K. Schofield, J. Photochem., 9 (1978) 55. 92 H. Okabe, Photochemistry of Small Molecules, Wiley, New York, 1978, p.

196. 93 K.H. Gericke and F.J. Comes, Chem. Phys., 65 (1982) 113. 94 J.E. Spencer and G.P. Glass, Int. J. Chem. Kin., 9 (1977) 97; I.W.M. Smith

and M.D. Williams, J. Chem. Soc. Faraday Trans. 2, 81 (1985) 1849; G.A. Raiche, J.B. Jeffries, KJ. Rensberger, and D.R. Crosley, J. Chem. Phys., 92 (1990) 7258.

95 J.E. Spencer, H. Endo, and G.P. Glass, 16th Symposium (Int.) on Combustion, The Combustion Institute, Pittsburgh, PA, 1976, p. 829.

96 R. Zellner and W. Steinert, Chem. Phys. Lett., 81 (1981) 568. 97 K. Skôld, Nucl. Instr. Methods, 63 (1968) 114. 98 R.J. Cvetanovic, Can. J. Chem., 52 (1974) 1452; F.S. Rowland and M.J.

Molina, Rev. Geophys. Space Phys., 13 (1975) 1; J.A. Davidson, H.I. Schiff, G.E. Streit, J.R. McAfee, A.L. Schmeltekopf, and C.J. Howard, J. Chem. Phys.,67 (1977) 5021; I.S. Fletcher and D. Husain, Can. J. Chem., 54 (1976) 1765; R.G. Prinn, F.N. Alyea, and D.M. Cunnold, Ann. Rev. Earth, Planet. Sci., 6 (1978) 43; J.R. Wiesenfeld, Acc. Chem. Res., 15 (1982) 110; RJ. Cicerone, Science, 237 (1987) 35.

99 S.P. Walch and C. McMichael Rohlfing, in Supercomputer Algorithms for Reactivity, Dynamics and Kinetics of Small Molecules, A. Laganà (éd.), Kluwer, Academic, Dordrecht, 1989, p. 73.

100 Materials Research Society Proceedings, Vol. 29. Plasma processing and synthesis of materials, J. Szekely and D. Apelian (eds.), North-Holland, Amsterdam, 1984.

60

101 G.D. Downey, D.W. Robinson, and J.H. Smith, J. Chem. Phys., 66 (1977) 1685; A.B. Callear and H.E. van den Bergh, Chem. Phys. Lett., 8 (1971) 17; M.C.Lin, J. Phys. Chem., 75 (1971) 3642.

102 Thirteenth Space Simulation Conference, NASA Conference Publication 2340, NASA Scientific and Technical Information Branch, Springfield,

Virginia, 1985. 103 K. Schofield, J. Phys. Chem. Ref. Data, 8 (1979) 25;R.E. Hine and J.T.

Herron, Progr. Reaction Kinetics, 8 (1985) 1;R.J. Cvetanovic and D.L. Singleton, Rev. Chem. Intermediates, 5 (1984) 183; M.C. Lin, in Reactions of Small Transient Species, M.A.A.Clyne and A.Fontijn (eds.),Academic,London, 1983.

104 F. Stuhl, J. Chem. Phys., 73 (1980) 3978. 105 W.B. De More, S.P. Sander, D.M. Golden, M.J. Molina, R.F. Hampson, M.J.

Kurylo, C.J. Howard, and A.R. Ravishankara, Chemical Kinetics and Photochemical Data for Use in Stratospheric Modeling, Evaluation No. 9, JPL Publication 90-1, NASA, 1990.

106 D.L. Baulch, R.A. Cox, P.J. Crutzen, R.F. Hampson, Jr., J.A. Kerr, J. Troe, and R.T. Watson, J. Phys. Chem. Ref. Data, 11 (1982) 493 , Supplement 1.

107 J.E. Thompson and A.R. Ravishankara, Int. J. Chem. Kin., (1993) in press. 108 W.L. Chameides and D.D. Davis, J. Geophys. Res., 85 (1980) 7383; M.E.

Jenkin and R.A. Cox, J. Phys. Chem., 89 (1985) 192; S.P. Sander, J. Phys. Chem., 90 (1986) 2194; R.E. Stickel, A.J. Hynes, J.D. Bradshaw, W.L. Chameides, and D.D. Davis, J. Phys. Chem., 92 (1988) 1862; H. MacLeod, C. Balestra, J.L. Jourdain, G. Laverdet, and G. Le Bras, Int. J. Chem. Kin., 22(1990) 1167.

109 P.H. Wine, J.R. Wells, and A.R. Ravishankara, J. Chem. Phys., 84 (1986) 1349.

110 R. Schinke, J. Chem. Phys., 80 (1984) 5510. 111 N. Balucani, L. Beneventi, P. Casavecchia, and G.G. Volpi, Chem. Phys.

Lett., 180 (1991) 34. 112 P.J. Bruna, G. Hirsch, S.D. Peyerimhoff, and R.J. Buenker, Can. J. Chem., 57

(1979) 1839. 113 J.C. Polanyi and J.L. Schreiber, in Physical Chemistry: An Advanced Treatise,

W. Jost (ed.), Academic, New York, 1974, Vol. VIA. 114 N. Balucani, L. Beneventi, P. Casavecchia, G.G. Volpi, E.J. Kruus, and J.J.

Sloan, Can. J. Chem., in preparation. 115 M. Alagia, N. Balucani, P. Casavecchia, and G.G. Volpi, to be published. 116 M.K. Gilles, M.L. Polak, and W.C. Lineberger, J. Chem. Phys., 95 (1991)

4723; J. Chem. Phys., 96 (1992) 8012; J.P. Bekooy, W. Leo Meerts, and A. Dymanus, J. Mol. Spectrosc, 102 (1983) 320.

117 K. Rynefors, P.A. Elofson, and L. Holmid, Chem. Phys., 100 (1985) 53. 118 S.J. Wategaonkar and D.W. Setser, J. Chem. Phys., 90 (1989) 6223. 119 V. Aquilanti, S. Cavalli, G. Grossi, V. Pellizzari, M. Rosi, A. Sgamellotti, and

61

F. Tarantelli, Abstracts XI Int. Symposium on Gas Kinetics, Assisi (Italy), 2-7 September 1990, p. 026; V. Pellizzari, Ph.D. Thesis, Universita di Perugia, 1992.

120 A. Laganà, private communication. 121 N. Balucani, L. Beneventi, P. Casavecchia, D. Stranges, and G.G. Volpi, J.

Chem. Phys., 94(1991) 8611. 122 P. Casavecchia, N. Balucani, L. Beneventi, D. Stranges, and G.G. Volpi,

Faraday Discuss. Chem. Soc, 91 (1991) 147; Faraday Discuss. Chem. Soc, 91 (1991) 383.

123 N. Balucani, L. Beneventi, P. Casavecchia, D. Stranges, and G.G. Volpi, J. Chem. Phys., in preparation.

124 G.S. Tyndall and A.R. Ravishankara, Int. J. Chem. Kin., 23 (1991) 483, and references therein.

125 G. Liuti, S. Dondes, and P. Harteck, J. Am. Chem. Soc, 88 (1966) 3212; D.A. Whytock, R.B. Timmons, J.H. Lee, J.V. Michael, W.A. Payne, and L.J. Stief, J. Chem. Phys., 65 (1976) 2052; I.R. Slagle, F. Baiocchi, and D: Gutman, J. Phys. Chem., 82 (1978) 1333; R.R. Smardzewsky and M.C. Lin, J. Chem. Phys., 66 (1977) 3197; R.R. Smardzewsky, J. Chem. Phys., 68 (1978) 2878.

126 D.L. Singleton, R.S. Irwin, W.S. Nip, and R.J. Cvetanovic, J. Phys. Chem., 83 (1979) 2195; D.L. Singleton, G. Praskevopoulos, and R.S. Irwin, J. Phys. Chem., 86 (1982) 2605.

127 R.F. Hampson, Report No. FAA-EE-80-71, U.S. Dept. of Transportation, 1980.

128 A.R. Clemo, F.E. Davidson, G.L. Duncan, and R. Grice, Chem. Phys. Lett., 84 (1981) 509; F.E. Davidson, A.R. Clemo, G.L. Duncan, R.J. Browett, J.H. Hobson, and R. Grice, Mol. Phys., 46 (1982) 33.

129 B.S. Agrawalla and D.W. Setser, J. Chem. Phys., 86 (1987) 5421. 130 G.Hancock, Faraday Discuss. Chem. Soc, 91 (1991) 151. 131 H. Wallmeier and W. Kutzelnigg, J. Am. Chem. Soc, 101 (1979) 2804. 132 P.L. Moore Plummer, J. Chem. Phys., 92 (1990) 6627; A.B. Sannigrahi, K.H.

Thunemann, S.D. Peyerimhoff, and R.J. Buenker, Chem. Phys., 20 (1977) 25; B. Sannigrahi, S.D. Peyerimhoff, and R.J. Buenker, Chem. Phys., 20 (1977) 81; R.J. Buenker, P.J. Bruna, and S.D. Peyerimhoff, Israel J. Chem., 19 (1980) 309.

133 B.T. Luke and A.D. McLean, J. Phys. Chem., 89 (1985) 4592. 134 S.S. Xantheas, private communication; J. Chem. Phys., to be submitted. 135 G.A. Fisk, J.D. McDonald, and D.R. Herschbach, Faraday Discuss. Chem.

Soc, 44 (1967) 228. 136 A.D. McLean and G.S. Chandler, J. Chem. Phys., 72 (1980) 5639; K. Krishan,

J.B. Binkley, R. Seeger, and J.A. Pople, J. Chem. Phys., 72 (1980) 650. 137 Z.Z. Zhu, J.J.W. McDouall, D.J. Smith and R. Grice, Chem. Phys. Lett., 188

(1992) 520. 138 M.J. McEwan and L.F. Phillips, Chemistry of the Atmosphere, Edward Arnold

62

Ltd., 1975; R.P. Wayne, Chemistry of Atmospheres, Clarendon, Oxford, 1985. 139 J. Warnatz, in Combustion Chemistry, W.C. Gardiner, Jr. (ed.), Springer, New

York, 1984, Chap. 5; J.A. Miller and G.A. Fisk, Chem. Eng. News, 65 (1987) 22; I. Glassmann, Combustion, Academic, New York, 1977.

140 I.W.M. Smith and R. Zellner, J. Chem. Soc. Faraday Trans. 2, 70 (1974) 1045.

141 A.R. Ravishankara, J.M. Nicovich, R,L, Thompson, and F.P. Tully, J. Phys. Chem., 85 (1981) 2498, and references therein.

142 J. Warnatz, Sandia National Laboratories Report No. SAND83-8606, Livermore, CA, 1983; D.L. Baulch, D.D. Drysdale, D.G. Home, and A.C.

Lloyd, Evaluated Kinetic Data for High Temperature Reactions, Butterworths, London, 1972, Vol. 2; G. Dixon Lewis and D.J. Williams, in Comprehensive Chemical Kinetics, CH. Bainford and C.F.H. Tipper (eds.), Elsevier, Amsterdam, 1977, Vol. 17, p. 1.

143 J.E. Spencer, H. Hendo, and G.P. Glass, 16th Symposium (Int.) on Combustion, The Combustion Institute, Pittsburgh, PA, 1976, p. 829.

144 G.C. Light and J.H. Matsumoto, Chem. Phys. Lett., 58 (1978) 578. 145 R. Zellner and W. Steinert, Chem. Phys. Lett., 81 (1981) 568. 146 D.W. Trainor and CW. von Rosenberg, 15th Symposium (Int.) on

Combustion, The Combustion Institute, Pittsburgh, PA, 1974, p. 755. 147 J. Finzi, F.E. Hovis, V.N. Panfilov, P. Hess, and C.B. Moore, J. Chem. Phys.,

67 (1977) 4053. 148 P. Andresen, N. Aristov, V. Beushausen, D. Hausler, and H.W. Liilf, J. Chem.

Phys., 95 (1991) 5763; J.J. Ter Meulen, Abstracts Ninth European Conference on Dynamics of Molecular Collisions (MOLEC IX), Prague, August 30-September4, 1992, p. 25.

149 A.D. Isaacson and D.G.Truhlar, J. Chem. Phys., 76(1982) 1380; 77(1982) 3516.

150 A.D. Isaacson, J. Phys. Chem., 96 (1992) 531. 151 J.M. Launay and S.B. Padkjaer, Chem. Phys. Lett., 181 (1991) 95. 152 T.H. Dunning, L.B. Hardin, and E. Kraka, in Supercomputer Algorithms

for Reactivity, Dynamics and Kinetics of Small Molecules, A. Laganà (éd.), Kluwer, Dordrecht, 1989, p. 57.

153 J. Werner, private communication. 154 F. Westley, J.J. Herron, R.J. Cvetanovic, R.F. Hampson, and W.G. Mallard,

NIST Gas Kinetics Database, Ver. 1.1 (1989), and references therein; I.W.M. Smith and R. Zellner, J. Chem. Soc. Faraday Trans. 2, 69 (1973) 1617; R. Foster, M.J. Frost, H. Hippler, A. Schlepegrell, and J. Troe, Faraday Discuss. Chem. Soc, 91 (1991) 406; W.B. De More, Int. J. Chem. Kin., 16 (1984) 1187; A.J. Hynes, P.H. Wine, and A.R. Ravishankara, J. Geophys. Res., 91 (1986) 11815, and references therein; CW. Larson, P.H. Stewart, and D.M. Golden, Int. J. Chem. Kin., 20 (1988) 27; M. Mozurkewich, JJ. Lamb, and S.W. Benson, J. Phys. Chem., 88 (1984) 6435.

63

155 D.E. Milligan and M.E. Jacox, J. Chem. Phys., 54 (1971) 927; M.E. Jacox, J. Chem. Phys., 88 (1988) 4598.

156 H.E. Radford, W. Wei, and T.J. Sears, J. Chem. Phys., 97 (1992) 3989; T.J. Sears, W.M. Fawzy, and P.M. Johnson, J. Chem. Phys., 97 (1992) 3996.

157 B. Ruscic, M. Schwarz, and J. Birkowitz, J. Chem. Phys., 91 (1989) 6780. 158 (a) D.W. Trainor and CW. Rosenberg, Jr., Chem. Phys. Lett., 29 (1974) 35;

(b) T. Drier and J. Wolfram, in 18th Symposium (Int.) on Combustion, The Combustion Institute, Pittsburg, PA, 1981, p. 801.

159 D.M. Sonnenfroh, R.G. MacDonald, and K. Liu, J. Chem. Phys., 94 (1991) 6508.

160 J. Wolfram, private communication. 161 K. Kudla and G.C. Schatz, private communication and to be published. 162 A.H. Kung and Y.T. Lee, in Vacuum Ultraviolet Photoionization and

Photodissociation of Molecules and Clusters, CY. Ng (ed.), World Scientific, Singapore, 1991, p. 487.

163 S. Nourbakhsh, K. Norwood, H.-M. Yin, C.-L. Liao, and CY. Ng, J. Chem. Phys., 95 (1991) 946; K. Norwood, S. Nourbakhsh, G.-Z. He, and CY. Ng, Chem. Phys. Lett., 184 (1991) 147.