Photochemistry of adsorbed molecules. Part 3.mdash;Localised atomic scattering in the photolysis of HI/LiF(001)

摘要

Faraday Discuss., 1993,96, 129-149 Photochemistry of Adsorbed Molecules Part 13.-Localised Atomic Scattering in the Photolysis of HI/LiF(OOl) V. J. Barclay, W-H. Hung, J. C. Polanyi,* G. Zhang and Y. Zeiri? Department of Chemistry, University of Toronto, Toronto M5S lA1, Canada We have measured the translational energy distribution, P(E;), for atomic H coming from 248 nm and 193 nm photolysis of HI adsorbed on LiF(001) at a coverage of 0.7 ML (monolayers). At both wavelengths P(Ek) showed evi- dence of three contributions as follows: (a) The most energetic H was desig- nated H(1); the energetics indicated that in this channel HI(ad) photodissociated to give ground-state I(2P3,2). (b) Fast H with approx- imately 1 eV lower peak energy was designated H(I*); in this case the energy corresponded to HI(ad), giving H + I*(2P1,2).(c) The third component was slow H observed down to ~0.5eV; it was interpreted as being inelastically scattered and was designated H(Ine1).For photolysis at 248 nm the highest energy component, H(I), had a peak translational energy (Ek), = 2.0 eV, and the second component H(I*) had (Ek), = 1.1 eV. For photolysis at 193 nm H(1) had (E;), = 3.4 eV and H(I*) had (Ek)p= 2.5 eV. These energies for the scattered H at each wavelength are the same as those reported for H recoil- ing from photolysed gaseous HI; it appears therefore that HI(ad) gives the contributions H(1) and H(I*) by elastic scattering. The yield ratio H(I)/H(I*) from HI/LiF(001) was comparable with that for the gas phase for 248 nm photolysis of HI/LiF(001), but was greatly reduced from its gas-phase value at 193 nm.Taken together with the enhanced H(Ine1) at 193 nm, this sug-gested markedly increased inelastic energy loss in collisions of the 3.4 eV H-atoms with the substrate and/or co-adsorbate. Theory, also reported here for the first time, predicted at 0.7 ML that HI(ad) would be tilted with the H-end down ca. 15" more steeply than for HBr(ad), but pointing at F-on LiF(001) as reported previously for HBr/LiF(001) [E. B. D. Bourdon et al., J. Chem. Phys., 1991, 95, 1361). This resulted in localised atomic scattering (LAS) offF-. Energy loss from the 3.4 eV H photorecoiling from HI(ad) and then colliding with F-in the substrate can be due (i) to the more complex trajectories that theory predicts for the case that HX is tilted downwards more steeply, (ii) to increased 'chattering' due to the high impact energy, and (iii) to inelasticity due to strong encounters between the photorecoiling H and adjacent HI(ad).1. Introduction The photochemistry of adsorbates is being pursued in a number of lab~ratories.'-~ A particular attraction of this line of research is the possibility that it offers for the study of inelastic and reactive encounters in an ordered layer. In an earlier experimental paper' we showed that in the case of HBr adsorbed on LiF(001) the H-atom points downward by 21 5" from the plane of the crystal in a hydrogen-bonded Br-H.--F- alignment, t Present address: Dept. of Physics, N.R.C.N., P.O.Box 9001, Beer Sheva, Israel 84190. 129 LAS in the Photolysis of HI/LiF(001) hence one can study the scattering of photorecoiling H off F-sites on the surface in what we have termed 'localised atomic scattering' (LAS).6-9 Theoretical values of 23" and 26" were calculated by alternative approaches for this Br-H downward The photochemistry of HBr/LiF(OOl) was previously studied at 193 nm.l0 The H-atom energy distribution was measured, as also was the anisotropic scattering, P(O'),of the fast (elastic) component of the H-atoms which was found to peak at 55" to the normal with respect to the LiF(001) substrate. This observed scattering angle of 55" is 14" less than that expected for specular scattering (69"); this was attributed to the effect of LAS off F-.In the present paper we give first results for the H-atom energy distribution from photolysis of HI/LiF(001) at 248 and 193 nm, and discuss our findings in the light of a simple classical trajectory calculation. In a subsequent paper we shall compare the experimental energy distributions reported here and also the angular distributions with a full dynamical study employing an optimised interaction potential. Localised scattering as an outcome of surface-aligned photochemistry has been posited independently in a theoretical study by McCarthy and Gerber, who considered the system IC1/MgO(OO1).l' The concept is also paralleled in a different context in studies of resonances and cage effects in the scattering of photorecoiling H from Ar within the gaseous complexes Ar-HX (X = Cl, Br).12'13 The Ar in the complex has a somewhat similar role to the F-surface atom encountered by the photorecoiling H from HX in the present work, i.e.it is struck in a restricted range of impact parameters due to prior alignment of HX. The relevant electronic states for the bound-to-free transitions in HI are given in Fig. 1. The lowest asymptote corresponds to the formation of H + I(2P3/2)(ground-state halogen atom); the higher asymptote to H + I*(2Pljz)(excited-state halogen atom). The difference in energy is 0.9 eV for E(I*)-E(I).I4 Dissociation to the more stable (lower) 4 3 2 h'2-0 -1 -2 -3 1.5 2.0 2.5 3.0 3.5 4.0 r/A Fig. 1 Potential-energy curves for HI (adapted from ref.15 and 17). The nature of the transition from ground to excited state is shown in parentheses: I= perpendicular; 11 = parallel. The atomic asymptotes for I and I* are indicated. V. J. Barclay et al. asymptote gives rise to H-atoms with higher translational energy, and dissociation to the upper asymptote gives a lower H-atom energy. In our earlier experimental study" of HBr/LiF(001) we found that the translational energy of the scattered H-atoms peaked at two energies corresponding to those from the gas-phase photodissociation, and con- cluded that these H-atoms had been elastically scattered from the surface. The yields of the high- and low-energy components in the H-atom translational- energy distribution from the HBr/LiF(OOl) at 193 nm corresponded closely to the 6/1 yield obtained for Br/Br* in gas-phase photolysis." Note that the energy of the recoiling H from HBr is 2.6 and 2.1 eV, a lower range than for H recoiling from HI at the same wavelength, due to the strong bond in HBr.Two elastically scattered peaks were once again obtained in the translational energy distribution of H from HI/LiF(001) at 3, = 248 nm and at 193 nm. At 193 nm the relative yields of the high- and low-energy H-atoms differed markedly from the gas phase. This finding is discussed here in terms of classical trajectory calculations and is attributed to increased inelasticity at the high translational energy (3.4 eV) of fast H in the system HI/LiF(001) at 193 nrn.2. Experimental The experiments were performed in a UHV chamber (base pressure 1 x lo-'' Torr) shown schematically in Fig. 2. The apparatus has been described in detail elsewhere." A doubly differentially pumped mass spectrometer rotates around the crystal, permitting measurement of time-of-flight (TOF) spectra at specified angles to the surface, and angular distributions at specified product translational energies. The TOF path length from crystal to mass-spectrometer ioniser was 175 mm. The LiF(001) surface was cleaved in air and then annealed in UHV at a temperature of ca. 700 K by radiation heating for at least 12 h prior to an experiment. The LiF crystal was mounted on a manipulator with xyz translation and polar rotation. The crystal could be cooled to 85 K by means of a closed-cycle helium refrigerator. The HI MASS EXCIMER LASER Fig.2 Schematic of the UHV chamber. The photolytic laser beam comes from below. The differ- entially pumped mass spectrometer was rotated around the cooled crystal giving time-of-flight spectra as a function of 0'. See ref. 10. LAS in the Photolysis of HI/LiF(001) (Matheson; 99%) was purified by freeze-pumpthaw cycles before being introduced into the chamber by background dosing onto the LiF(001) surface. An excimer laser (Lumonics TE860-4) was used to irradiate the adsorbate-covered surface at 5" glancing incidence (i.e. 85" from the surface normal). The laser output was 80 mJ per pulse at 248 nm and 45 mJ per pulse at 193 nm. The corresponding intensities at the crystal surface lay in the range 1-2 mJ cm-2 per pulse, due to losses at mirrors and due to the large spot-size at the glancing angle.In the TOF measurement the LiF crystal was held at a temperature of 85 K. Follow-ing ca. 600 laser shots, co-added to produce one TOF spectrum, the LiF crystal was heated to room temperature to remove residual adsorbate prior to re-dosing. This single dosing procedure contrasts with that used in our earlier photochemical study of HBr/LiF(001). The crystal samples purchased at that data showed efficient photodesorption of intact HBr, hence a constant coverage (roughly ascertainable) was obtained by balancing continuous dosing against photodesorption. The crystals used in the present work showed negligible photodesorption, perhaps due to lower impurity levels, hence this procecedure was no longer feasible.Instead we formed a stable adsorb- ate layer prior to photolysis, and following ca. 600 shots (this would correspond to ca. 10% dissociation for HBr but only ca. 1% for HI) the residue was removed and the crystal re-dosed. This modified method of dosing could give rise to a different adsorbate structure (since the adsorbate layer has time to equilibrate) and also to a different average degree of contamination of adsorbate with photoproduct (e.g. if the halogen was inefficiently photodesorbed it would accumulate following a period of continuous dosing of the cold crystal with concurrent photolysis). We therefore repeated our earlier translational energy and angular distribution measurements on HBr/LiF(OO 1) using this single dosing method; we could detect no sigr,ificant changes. Coverages were estimated as follows.Following exposure of the cold crystal to a known dose in Langmuir (L), the desorption yield was measured by integrating the 2.5 h2 2.0 3 4 1.5 m w I! 1.0 .-. a 0.5L F 0.0 0 0 1 23 4 56 exposure/L Fig. 3 Integrated yield from TPD at various coverages. (a) HBr; (b) HI. The slope changes at around 2 L, indicative of 1 ML coverage. V.J. Barclay et al. 133 temperature-programmed desorption (TPD). The TPD was obtained by ramping the sample temperature at a linear rate of ca. 1 K s-', with the mass-spectrometer axis along the normal to the surface.A TPD for HBr/LiF(001) can be found in our earlier paper." The monolayer and multilayer components were found to be less well separat- ed in the TPD for HI than that for HBr. Both HBr and HI showed a peak desorption yield at ca. 110 K corresponding to a heat of adsorption A,, H FZ 0.27 eV assuming a pre-exponential factor of 1OI2 (ref. 16). The integrated TPD yields are shown for HBr and HI in Fig. 3(a) and (b) as a function of the dose in L. It is evident that in both cases the yield function shows a change in slope at ca. 2 L; we ascribe this to a change from a layer in contact with the crystal (1 monolayer, ML) to an overlayer (1 ML). The doses of HBr (used in the check mentioned above) and of HI, both 1.5 L, are therefore considered to correspond to ca.0.7 ML. 3. Results 3.1 Energy Distributions Fig. 4(a) shows a typical TOF spectrum of H-atoms from HI adsorbed on LiF(001) at ca. 0.7 ML coverage, The photolysis wavelength was i= 248 nm, and the mass-spectrometer detector was set roughly at the peak of the H-atom angular distribution 45" to the normal. The corresponding translational-energy spectrum is shown in Fig. 4(b). A conversion factor of t2 (t = time; t-' corrects for the changing efficiency of the ioniser with velocity, and a further t3 converts the corrected TOF to an energy distribution) has the effect of magnifying errors at the low-energy end of the TOF spec- trum. The TOF shown in the figure represents the sum of three individual single-dose spectra, each taken with ca.600 laser shots directed at a freshly prepared surface. Fig. 4(b) shows two distinct peaks centred at 1.1 and 2.0 eV, corresponding, within experimental error, to the excess energies from the two HI photodissociation pathways in the gas phase at A =248 nm.15917 Since these same energies are present in the scat- tered H, we conclude that [as for HBr/LiF(001) at ?, = 193 nm in our earlier work''] there is efficient elastic scattering of H at the crystal. The ratio of the yield of scattered H at 1.1 eV to that at 2.0 eV is ca. 2/1. Photolysis to form I* gives the slower photorecoiling H, and photolysis to form I gives the faster H. The ratio of yields of I/I* in gas-phase photolysis at 248 nm is 54/46.'5*'7 This is within a factor of two of the observed ratio of 'slow' to 'fast' H leaving the surface at 45" [Fig.4(b)]. There is a small amount of inelastic scattering yielding very low energy H, indicated in Fig. 4(b) as H(Ine1). This is a broad distribution, extending from r-ughly 0 to 1 eV. The-inelastic events could involve collisions between photorecoiling H and the surface, or H in collision with co-adsorbed HI(ad). A similar low-energy tail on the two hot- atom translational energy peaks in the scattered H was observed for HBr/LiF(OOl), 13. = 193 nm, icour previous work." It was ascribed in the earlier experimental study to the effect of H +HBr(ad) collisions, since the magnitude of the 'tail' was found to increase with coverage.The ratio of yields for the two peaks in the scattered H energy distribution that we ascribe to elastic scattering at 248 nm, H(1) at 2.0 eV, and H(I*) at 1.1 eV, varied with the angle 0' at which the scattered H was measured. The ratio H(I)/H(I*) was ca. 2 from 0 to 50°, decreasing to unity in the region 50-90". This appeared to be due to a some- what broader angular distribution for H(I*) than for H(1). Fig. 5(a) and (b)show the TOF and the corresponding translational energy distribu- tion for H-atoms scattered from HI/LiF(001) at a markedly shorter photolytic wave- length, il = 193 nm. The detection angle was 57.5", and the coverage was 0.7 ML. The LAS in the Photolysis of HI/LiF(001) 0 5 10 15 20 25 time of flight/ps 00 05 10 15 20 25 30 translational energy/eV Fig. 4 (a)Experimental TOF spectrum of H-atoms detected at 8' = 45" after 248 nm photolysis of HI/LiF(001).(b)Translational energy distribution derived from (a),showing peaks due to product formation of I and I*. The gas-phase ratios, I/I* (ref. 15, 17) are shown as vertical lines, normalised to the higher peak. spectrum shown is (as before) the sum of three spectra, from three separate dosings of the crystal. The qualitative appearance of this energy distribution was insensitive to the angle of observation. We expected in this case to observe an elastic peak, H(I), at 3.4 eV, and a second peak, H(I*), at 2.5 eV since these are the measured energies of H-atoms recoiling from HI(g) at A = 193 nm. These two elastic peaks are indeed evident in Fig.5(b). However, the ratio of the yields of these peaks (in contrast to the situation at R = 248 nm) differs markedly from the ratio of I/I* in the photolysis of HI(g) at the same wavelength, 193 nm. The H(1) peak in the gas pha~e'~?'~ is nine times the H(I*) peak, whereas the H(I) peak for H scattered from LiF(001) is 0.25 times the H(I*) peak, a discrepancy of 36 times. We discuss the origin of this discrepancy in Section 4.3 below. There is an additional change in the qualitative nature of the translational energy distribution of H from HI/LiF(001) at 0.7 ML as between Fig. 4(b)(A = 248 nm) and 5(b)(A = 193 nm). Inelastic scattering has altered from being a relatively minor pathway at V. J.Barclay et al. h2 15 Y I I I I I 0 5 10 15 20 25 30 time of flight/ps n I m v 06-Q w.-v) 04-+-0) .-C 0 2-0 o--00 05 10 15 20 25 30 35 40 45 50 translational energy/eV Fig. 5 (a) Experimental TOF spectrum of H-atoms detected at 0' = 57.5" after 193 nm photolysis of HI/LiF(001). (b) Translational energy distribution derived from (a), showing peaks due to product formation of I and I*. Gas-phase ratios, I/I* (ref. 15, 17), are shown as vertical lines, normalised to the higher peak. 248 nm to being the major pathway at 193 nm [compare H(Ine1) in each figure with the elastically scattered H at the right in that figure]. These two new features at 2= 193 nm, loss of the rapidly moving H(1) and gain in +H(Inel), are likely to be related.For the calculated angle at which the photorecoiling H from HI(ad) impacts on the surface F-the energy loss may be greater at 3.4 eV collision energy than at 2.5eV, for reasons noted in Section 4.3. Inelastic collisions with co-adsorbed HI(ad) will also result in greater energy loss in the case that the colliding species is moving faster, espezially if the enhanced translation allows reaction or abortive reaction to occur in H + HI(ad) encounters as is the case here. Before presenting some model calculations illustrative of inelastic processes that could be responsible for energy loss from the 3.4 eV H, we consider other possible explanations for the observed marked diminution in H scattered from the surface at 3.4 eV.136 LAS in the PhotoZysis of HI/LiF(001) The first alternative explanation is a possible experimental artefact. Following electron-impact ionisation H+ is accelerated into a 115" deflector prior to entering the quadrupole field of the mass spectrometer. We have considered the possibility that low- energy ions may be deflected whereas high-energy ions are lost. The relative transmis- sion efficiency of the deflector has been measured (ref. 18, Fig. 2.5).If the ion acceleration voltage is set to 2 eV below the optimal value, the transmission is found to be virtually flat for incoming atoms of 0-4 eV. As a further check we have recorded an approximate TOF on a machine that has the mass spectrometer in line with the crystal (no deflector); we obtained a spectrum in qualitative accord with Fig.5(b). Another explanation for the anomalous ratio of the elastic-peak heights in Fig. 5(b) could, in principle, be a dramatic change in the relative yield of I/I* in the photolysis of HI(ad) as compared with HI(g). This could arise from a change in the nature of the crossings and avoided crossings between electronically excited states, so that disso- ciation takes place to a different asymptote. Since no marked change in I/I* was observed at A = 248 nm, nor in Br/Br* at A = 193 nm in our previous work, it appears improbable that the surface induces a change in branching ratio, due to the above effects, of well over an order-of-magnitude in I/I* at A = 193 nm. A third possible explanation for the dramatic change in branching ratio at 193 nm concerns the effect of the surface in polarising the incident photolysis radiation (initially unpolarised).Owing to surface reflectance, the p-polarisation of the laser field is enhanced by 37%, in comparison with s-polarisation. A 'horizontal' molecule, i.e. one lying with its principal axis parallel to the plane of the surface, would thus have its perpendicular transitions enhanced. Conversely, a 'vertical ' molecule, sitting perpen- dicular to the surface, would have its parallel transitions enhanced. In the most conser- vative estimate, if all adsorbate were parallel to the surface, the branching ratio for I/I* for HI(ad) photolysed at 193 nm would be 9 x 1.37/1 i.e. 12/1; whereas if all adsorbate molecules were vertical, the branching ratio would be 9/1.37, i.e. 6/1.A median tilt angle of 0 = 130"(FWHM = 30") is predicted for HI/LiF(001) as a result of theoretical calcu- lations reported in the next section. Since half of the adsorbate molecules would have 0 somewhat less than 130" and the other half would be tilted at 0 somewhat greater than the median, the small effect of UV polarisation should be undetectable on the average, whereas the modification in the yields from the two pathways (fast/slow H) is very large. We have, in addition, empirical evidence that laser polarisation is not the cause of the large anomaly in the fast/slow H-atom yield ratio at 193 nm, since we have obtained H-atom TOF at angles to the normal ranging from 10 to 80" (with fixed laser angle-of- incidence).Within a factor of ca. 2 the yield ratio is as depicted in Fig. 5(b). It follows that reflection enhancement of the p-polarisation of the incident light can be ruled out as the dominant source of the anomaly in the ratio of H(I)/H(I*) at 3.4 eV. 3.2 Angular Distributions We shall discuss the angular distributions for the high-energy elastically scattered H-atoms in detail in a subsequent paper (see Section 3.1, above). We summarise their major features here. For 248 nm irradiation of HI/LiF(001) the fastest H (2.0 eV) gave an angular dis- tribution that peaked at 8= 45 k 5". (The yield of H was integrated over k0.5 eV centred on 2.0 eV.) The form of P(0') was represented by a slow rise from 0 to 45" and a steep decline (to 50% by ca.55") in the range 45-90". The integrated yield at 45" was roughly double that at 245". For 193 nm irradiation the fastest H (3.4 eV) gave a markedly broader angular distribution. (The yield of H was, once again, integrated over f0.5 eV, but centred this time on 3.4 eV.) The signal-to-noise was poor due to the feeble signal at 3.4 eV, ascribed in the foregoing text to inelastic scattering. The P(0') exhibited no clearly defined peak; V. J. Barclay et al. 137 the maximum yield lay in the range 20-50". The breadth of the angular distribution was indicative of a significantly modified interaction with the LiF surface at 3.4 eV as com- pared with 2.0 eV. In the following sect@ we link the modified dynamics at 3.4 eV to more complex encounters, H + Surf and H + HI(ad), resulting in energy loss that dimin- ishes the yield of 3.4 eV H-atoms.A minority of such more complex encounters (H colliding with the heavy atoms, F or I) can deflect the H trajectory without the loss of 0.5 eV. In this case the scattered H will be included in the P(0') for 3.4 eV, but one would expect the angular distribution to be broadened, as indeed observed. 4. Theoretical In order to provide an acceptably realistic, yet simple, model of the high-energy hydro- gen scattering from the LiF(001) surface and from adsorbed molecules, we have per- formed a stochastic classical trajectory (SCT) calculation, using the generalised Langevin equation, which includes 'ghost particles' in the substrate, as described in ref.19. These ghost particles allowed the surface to exhibit motion and thereby to exert random damping forces on adsorbates, as in ref. 8 and 20, and in contrast to the 'frozen' surface used in ref. 9 (quantum scattering calculation). The equilibrium geometry for HBr/ LiF(001) has been studied extensively in ref. 7. Forms for the potential were recommend- ed, which, with some modifications, have been used in the present study. An earlier study* of HBr/LiF(001) using this general form of the potential gave a dynamical simulation of LAS, i.e. of the photolysis of a single HBr adsorbate molecule with the H-atom directed at a preferred location on the corrugated surface. This study predicted the form of non-reactive H-atom scattering patterns near the 'zero coverage' limit, in which the predominant scattering is from the crystal surface.The mass ratio was such that very little energy transfer between the hot H-atom and the surface occurred so that collisions were nearly elastic. In the present case the photon energy was 193 nm, i.e. 148 kcal mol-l. About 72 kcal mol-' of this energy was consumed in breaking an HI bond, leaving an excess energy Ex, = 76 kcal mol-l, as determined from Ex, = E(hv) -Do, where Do is the bond-dissociation energy. In photolysis, most of the excess energy becomes translational energy of the H-atom, E,, which is modified to Ek following collision with the surface and/or adsorbate. A fraction of the excess energy was used to overcome the energy of physisorption, AQad, w5kcal mol-'.In the calculation, further excess energy, ca. 10 kcal mol-', arose from the repulsion between atomic H in electronically excited HX and the surface. This repul- sion, was related to the fact that at the instant of photolysis, the H was changed from being molecular H within HX to atomic H. The energy available for translational excitation of the departing atomic H was therefore' In the present work we consider a coverage of one molecule and also 0.7 ML. The 0.7 ML coverage consisted of 14 molecules, with 1'-(6-represents a negative charge of less than unity) located over Li'. The 14 molecules were distributed on adjacent sites within an 18-site area of LiF consisting of a 7 ion x 7 ion slab with periodic boundary condi- tions.8 At 0.7 ML the hot H-atom will in many cases collide with HX(ad) before leaving the surface; provision is made for this in the calculation by including both inelastic and reactive encounters with H with adsorbate, as described below.4.1 Potential The interaction potential is described in three parts; the physisorption potential (adsorbate-surface and adsorbate-adsorbate), the reactive potential, and the switching LAS in the Photolysis of HI/LiF(001) function that switches between atomic and molecular environments as the photofrag- ment interacts with co-adsorbate. The physisorption potential-energy function used7-9g20 for HBr/LiF(OO 1) was applied once more to HI/LiF(001).It can be summarised as : The interaction of the adsorbate molecule with the LiF(001) surface (Vas) and with other adsorbates (Vaa) was described by potentials which consisted of electrostatic (T/el) and non-electrostatic (Vnel)contributions whose pairwise functional forms were described in ref. 7. The parameters used in that work for the potential functions were intended to describe HBr/LiF(001) with HBr in its ground electronic state. We have, in the present work, replaced these parameters by values appropriate to HI(ad); see Table 1. The fol- lowing is a brief description of the terms in the physisorption potential. The electrostatic interaction between the adsorbate and surface, V::, was the prin- cipal contributor to the overall stability.As estimated from Monte Carlo calculations of one monolayer coverage, the electrostatic adsorbate-surface interaction contributed ca. 75-85% of the total energy of HI/LiF(001) physisorption, as compared with ca. 80-95% of the total energy of HBr/LiF(001) physis~rption.~~~' It is thus important to model this term accurately. In the two point-dipole model chosen to model the adsorbate molecule the charge cloud on the molecule was approximated as two point-dipoles, pH and px, situated at the nuclei of an adsorbate molecule. The point-dipole was taken as pi = piu, where u is the unit vector pointing along the bond axis of the molecule to which the nucleus belongs. The adsorbate-surface electrostatic potential was described by the interaction of these point-dipoles fixed on the H and X sites with the permanent electric field at the (001) face of the LiF ionic crystal.Lennard-Jones and Dent have derived the form of this electrostatic potential.22 The values used for the HBr point-dipoles are as in ref. 8. The values used for the HI point-dipoles were apportioned to give the experimen- tally known molecular dipole23 and calculated quadr~pole;~~ they were pl= 1.3672 Dt at the hydrogen site and p2 = -0.9195 D at the iodine site. With no repulsive potential to counteract the electrostatic term, the adsorbate dipoles would fall into the surface ions. The force to balance the strong ion-dipole forces was provided by the non-electrostatic potential. In constructing the non-electrostatic adsorbate-surface potential, Vzl, we used the damped Tang-Toennies potential2 which was summed as a 2D Fourier series26 over (001) layer of the solid.7 The c6 and Table 1 Calculated Tang-Toennies gas atom plus surface-ion potential parameters [Lennard-Jones (6-12) parameters for the adsorbate HI] Tang-Toennies A/eV /J/%.-' c,/eV A6 c,/eV A' H(atom)-Li+ H(at om)-F -H(mo1)-Li + H(mo1)-F -I-Li + 174.6 174.5 189.0 177.1 1680.0 3.776 2.877 4.50 3.28 4.156 0.125 2.585 0.125 2.585 2.435 0.196 3.404 0.196 3.404 3.914 I-F - 630.0 3.092 57.90 122.0 Lennard-Jones (6- 12) &/kcal mol - 4 H-H 0.0266 2.735 1-1 0.561 3.89 H-I 0.122 3.310 t 1 D x 3.335 64 x C m. V.J. Barclay et al. c8, and Born-Mayer parameters needed for the H-Li+, H-F-, Br-Li+ and Br-F- interactions are as given in ref.8 (Table I). Values for the H-surface ion interaction parameters are given for both atomic and molecular environments. This is because the charge distribution around an atom depends on whether it is a free atom or is bound to another atom. Thus, a distinction should be made between atomic and molecular states of the H and X atoms when assigning parameters for their interactions with the LiF surface. In practice, however, the types of environment (and the switching between them; vide infra) were found to be significant only for the H-atom. The non-electrostatic parameters for the I-atom were obtained from the polarisabil- ities and the HI-HI dispersion coefficients.The derivation followed that for the Br-atom parameters7.* and is discussed in more detail elsewhere.21 The values used are given in Table 1. Where more than one adsorbate molecule was present, inclusion of the adsorbate- adsorbate interactions (electrostatic, nonelectrostatic, and reactive) was necessary. In previous work the Vgf term was modelled by three fractional point charges along the axis of HBr.7 The point charges were chosen so as to give the correct dipole and higher multipoles of HBr. One of these point charges, q3 (Fig. 1, ref. 7), was located 1.29 8, beyond the centre of the Br atom, away from the H. This was unphysical. In the present dynamical study we replaced the three point-charges by two point-dipoles located on the I-atom and H-atom, respectively, This was thought to be advantageous in order that the charges q3 on adjacent adsorbate molecules did not repel and obstruct reorganisation of the adsorbate during equilibration.In the previous work using Monte Carlo techniques27 this problem was less severe since molecules were free to flip through 90 or 180”during equilibration. The energy of interaction between two adsorbates modelled by two point-dipoles, described above, is based on the standard dipole-dipole interaction term, as in ref. 28 : The adsorbate-adsorbate point-dipole interactions were calculated between each nucleus of each pair of molecules a distance rij apart, with four interaction terms per adsorbate pair. Finally, the last term of the physisorption potential in eqn.(2), the non-electrostatic adsorbate-adsorbate term, V/antl, was modelled using a Lennard-Jones (6-12) function; the parameters are given in Table 1. We turn next to the reactive potential, H + H’X. Such a collision can have three outcomes: H + H’X +HX + H’ (exchange), +HH’ + X (abstraction), or -+H + H’X(v’, J’) (inelastic scattering). Although the branching ratios are not the subject of this paper, abortive exchange or abstraction, or full exchange reaction, was a significant source of H(Ine1). The interaction of the atomic photofragment H with its nearest molecular neighbour was modelled using a London-Eyring-Polanyi-Sat0 (LEPS) potential,29 as in ref. 30. The Morse parameters D,,p and re ,and Sat0 parameters S, for H + HI were for H-I D,= 73.78 kcal mol-’, p = 1.750 k’,re = 1.604 8,, and S(H1) = 0.0915, as given in Table I of ref.31. For H-H the corresponding parameters were 109.458 kcal mol-’, 1.9413 A-’, 0.742 8, and 0.353, as given in Table I of ref. 30. The closest HX molecule to a photorecoiling H situated in the forward hemisphere (i.e. along the direction of motion) is designated the ‘target molecule’. The interaction potential with this target molecule was changed from the adsorbate-adsorbate physi-sorption potential to the reactive potential following collision with the surface when a target molecule was identified. The need for a switching function is apparent from the following considerations. The interaction of H and H’ in an H + H’X(ad) reactive or inelastic encounter as already 140 LAS in the Photolysis of HI/LiF(001) noted is described by a LEPS function.The incoming H must gradually be reduced in size from its atomic to its molecular radius, and must at the same time have its point- dipole increased from zero at infinite separation to the value appropriate to HX at equilibrium re. This affects the interaction of H with surrounding atoms. Concurrently the departing H’ goes through reciprocal changes. In order to represent these changes in a smooth fashion we have scaled the atomic radius and the point-charge to the bond order (the ‘fraction’ of a bond, first introduced by Pauling3,). The gradual shift from ‘atomic’ to ‘molecular’ H interaction with the neighbouring environment was accom- plished by means of a ‘switching function’ based on that employed by Shustor~vich.~~ Details of the switching function used are to be found in another paper from this labor- atory.,’ The slow-moving photorecoiling X is assumed to be non-reactive.Once reaction occurred, for computational simplicity, the product H-atoms were not allowed to react with another adsorbate. 4.2 Trajectories The calculations reported here were of two kinds. In the first idealised case, a single adsorbate molecule was artificially fixed at a certain geometry above an Li’ ion site, and not allowed to thermalise. In the second more realistic case, several molecules rep- resenting ca. 0.7 ML coverage were distributed over a section of the crystal and allowed to thermalise.Thermalising the ensemble consisted of integrating the initial configu- rations of the adsorbate particles for tens of thosands of time-steps of 10 au (1 time au z 2.4 x s) until thermal equilibration to within 10 K of the surface tem- perature of 100 K was achieved. Ghost particles (see Section 4.1) permitted the exchange of energy with the substrate during equilibration.’ For both idealised and realistic cases, photolysis was initiated as described in ref. 8, and the trajectory was integrated with a time-step of 1 au for up to 20000 steps or until the product atom and/or molecule were desorbed. Both idealised and realistic trajec- tories included ghost particles, in order to allow exchange of energy between the adsorb- ate and the surface.For realistic calculations, the identity of the target molecule was checked every 100 time au along a trajectory, and reaction was considered to have occurred once the photofragment had been closest to a certain target atom for three vibrational periods of the newly formed molecule. Both HX and H, were observed as reaction products; a detailed description of the reactions occurring in sub-monolayer surface-aligned photochemistry forms the basis of a separate report from this group.20 Desorption for either reactive or non-reactive cases was considered to have occurred after the molecular product was 4.5 8, from the surface, or the H-atom was 10 8, from the surface. Over 1600 trajectories with 0.7 ML coverage were sampled. For this substantial coverage the pattern of adsorption is significant.Previous theoretical studies of HBr/ LiF(001) supported the hypothesis of a ‘checkerboard’ pattern (Fig. 6), where the ratio of adsorbate molecules per unit cell is 1 : 2, with alternate cells having no adsorbate molecule. The unit-cell size of LiF(001) is 4.03 A whereas the lattice constant of solid HBr (Phase I1 in this temperature range) is34 3.96 8,. We modelled the HI as being in a checkerboard adsorption pattern for which only alternate sites are occupied. This mini- mised the difference in AadH between HBr and HI (ca. 10% difference). Experimentally, it was found that AadH (HBr) z AadH (HI) to within 5-10% assuming the same pre- exponential factor. Although the lattice constant for solid HI at 80 K, 4.27-4.32 is larger than that of the LiF(001) unit cell, since only alternate sites are adsorbed there is sufficient room on the crystal surface to accommodate a checkerboard HI/LiF(001), albeit with a larger z-height variation than for HBr.Islanding was promoted in the initial distribution of the adsorbate by choosing adjoining adsorption sites. During V.J. Barclay et al. Fig. 6 Calculated equilibrium geometries of HX adsorbed on LiF(001) at 100 K. Top-down view. The lattice ion positions and the van der Waals radii of H and X are drawn to scale. (a) HBr/ LiF(001); (b)HI/LiF(001). equilibration the I moves very little; only the H moves substantially. For the given coverage and various energies, the following approximate numbers of trajectories were run: 800 at 3.4 eV, and 400 at each of 2.5,2.0 and 1.1 eV.4.3 Theoretical Results Fig. 6 shows, for comparison, top-down views of typical adsorption patterns for HBr and HI on LiF(OO1). The surface ions are represented only by ' + ' and ' -'. The aster- isks indicate the atomic centres. The van der Waals radius36 of each adsorbate atom has 142 LAS in the Photolysis of HI/LiF(001) been drawn in. Atoms which were obscured by another atom, when seen from this perspective, are drawn with a broken line. The geometries shown in Fig. 6 were calculated by the stochastic classical trajectory (SCT) approach described in the previous section. Comparison of the tilt angles with those obtained by the Monte Carlo method used in earlier work7 shows that the X-H bond was tilted down slightly more steeply in the SCT approximation (2-3").We attrib- ute this to the small extension that occurs in the non-rigid X-H bond, in the SCT calculation; the stretched bond hydrogen-bonds a little more strongly to the underlying F-. Although the adsorption patterns for HBr [Fig. 6(a)] and HI [Fig. 6(b)] bear great similarity, two differences are apparent. The HBr adsorbate molecules are in better regis- try with the underlying lattice (the Br-atoms sit more squarely atop the Lif ions) than their HI counterparts. Secondly, the tilt angle of HI adsorbed on LiF(001) is greater than that of HBr. This is evident from the fact that the H-atoms are more completely obscured in Fig.6(b).This difference in tilt angle, 8, is more clearly evident in the dis- tributions, P(8),shown in Fig. 7. The tilt angle is almost coverage-independent in the case of HBr but changes signifi- cantly for HI. Energy minimisation for the case of an isolated 0 K molecule showed tilt angles of 116" for HBr8 and 118" for HI. The lattice constant for HI (in contrast to HBr) is so close to the unit-cell size of LiF(OOl), that the adsorbate-adsorbate interaction is sensitive to coverage. The isolated HI adsorbate sits with its I-end ca. 2.65 A above the surface. As more HI molecules are added in a checkerboard pattern to the LiF(001) surface to increase the coverage to 0.7 ML, the I-end rises 0.35 A, whereas the H-end rises by only 0.2 A from the surface.Fig. 7 shows this leads to an increased HI tilt angle of 130" (12" steeper tilt) at 0.7 ML. By contrast, with increasing coverage to 1.0 ML the peak in the HBr tilt angle is constant to within 4". 4.3.1 Inelastic Collisions fi +Surf The significance of tilt angle to the scattering dynamics is demonstrated in Fig. 8, which shows two trajectories with Ex, = 3.4 eV and the I-atom set at a single selected position 15 1 70 90 110 130 150 Oldegrees I i; Fig. 7 Calculated polar angle 8 of thermalised ensembles of adsorbate molecules on LiF(001) at T = 100 K for HBr (-) and HI (----) V. J. Barclay et al. 0 2 4 6 8 x = y cut/A x = y cut/A Fig. 8 Calculated trajectory of an H-atom after photolysis of HI(ad) to show the effect of increas- ing tilt angle. The cut along the LiF surface is in the (111) plane.The I-atom is fixed to be at (1.282, 0.997, 3.440) with the coordinate system defined as in Fig. 6. The H-I bond length is fixed as its gas-phase value23 of 1.609 A. Dashed lines indicate van der Waals atomic36 and ionic3* radii, and the solid lines show the non-electrostatic corrugation contours for the surface at 20 kcal mol-intervals. Excess energy of photolysis = 3.4 eV. Numbers along the trajectories indicate 5 fs intervals. (a) Initial tilt angle 8 = 130" leads to a direct scattering event; (b) initial tilt angle 8 = 145" leads to an indirect scattering event. slightly displaced from a Li' site (see caption). These are representative calculations both in respect of the position of the I-atom and the I-H ti the actual value of the tilt angle depends sensitively on the (x, y) position of the I-atom over the unit cell.In Fig. 8(a) the H-atom is tilted down from the normal to the surface plane by 8 = 130". The subsequent H-atom trajectory corresponds to 'direct' reflection at the surface (DIR in LAS in the PhotoEysis of HI/LiF(001) the figure). In Fig. 8(b),where the tilt angle has been made greater, 0 = 145",the H-atom trajectory reflects back almost normal to the surface and hence is 'indirect' (IND), colliding with the surface and the parent I-atom. A systematic variation of the tilt angle of the idealised LAS event with I directly over Li+ and the I of HI 3.4 A above the centre of Li', is shown in Fig.9, where the scattering angle 8' is plotted as a function of the tilt angle 8 for a 20" range in 8. Four distinct regions are observed, I-IV, in this small range of angles; they differ in the sign of the slope, dsl/dO. The open circles on Fig. 9 denote eight tilt angles for which the corresponding trajec- tories are shown, according to region, in Fig. 10. Examination of Fig. 10 shows that each successive region of the scattering curve in Fig. 9 corresponds to one additional turning point in the trajectory of the scattering H. The turning points are due to collisions with the surface, 's', or with another adsorbate molecule, 'a'. Region I has s only; I1 has s followed by a; I11 s, a, s and IV s, a, s, a. Fig. ll(a) gives the energy of the scattered H-atom as a function of initial tilt angle.The scattering angle (taken from Fig. 9) is shown as a broken line for comparison. In region I the H-atom suffers a glancing collision with the surface with result that rela- tively little energy is transferred. As the tilt angle increases the energy transfer becomes markedly more efficient, to the point where an H-atom initially at 6 eV can lose up to half its energy. (This, of course, takes no account of restrictions on energy transfer that may arise from phonon quantisation.) Fig. ll(b) shows the amount of translational energy gained by the I-atom and by the surface. Most of the energy lost from H goes to the surface. Since the H and I masses are very different, kinetic energy transfer to the I is not great.Through conservation of momentum, the kinetic energy transferred to the I-atom in a single encounter is expected to be only ca. 0.05 eV. The substantially larger energies being transferred at large 0 are due to multiple encounters. The amount of energy lost can be seen in Fig. 12 to depend not only on the initial tilt angle 0, but also on Ex,, which increases in the figure from 1.1 to 6.0 eV. The degree to which the H-atom can penetrate the potential-energy contours depends on its initial energy, Exs.The inner-energy contours of the surface corrugation have a smaller radius 40 70 135 140 145 150 %/degrees Fig. 9 Dependence of scattering angle, 8', on the initial tilt angle, 8,for a selected: zI = 3.43 8, and Ex,= 6.0 eV.The Roman numerals, I-IV, indicate four different scattering regions corresponding to I-IV in Fig. 10. The numbers 1-8 indicate the corresponding trajectories in Fig. 10. V.J. Barclay et al. 2 4 6 8 113 2 4 6 8 10 r/A rlA 6 6 4 4 2 2 0 0 2 4 6 8 10 2 4 6 8 10 r/A r/A Fig. 10 Trajectories of 6 eV H-atoms from scattering events with I-H tilt angles given in Fig. 9 designated by number (1-8) and region (I-IV). The open circle denotes the initial H-atom position (the I-atom is not shown). Dashed lines indicate the ionic38 radii. Collision with the surface is denoted 's'; collision with the parent I-atom is denoted 'a'. (I) Direct scattering, with only one (glancing) collision with the surface; (11) steeper initial tilt leads to collision with the surface, followed by collision with the parent I-atom; (111)still steeper tilt angle leads to repeated collision with the surface; (IV) chattering (two surface collisions, s) is occurring.The effect of a small change (1") in the initial tilt angle is shown in each panel (the steeper tilt angle leads to the trajectory denoted by the lighter line). of curvature than the outer-energy contours. This affects the scattering angle at the first encounter with the surface. As a consequence, at high energy the H-atom is reflected more nearly normal to the surface and collides with the parent I-atom, leading to greater energy loss (see below). Thus, the shape of the Ek(8)curve in Fig. 12 depends on Ex,.The high-collision-energy case (6.0eV) exhibits the earliest onset of increased energy loss, i.e.indirect scattering begins at the smallest 8. Fig. 13 illustrates the effect of increasing Ex, on the scattering dynamics. With the same tilt angle, chosen as 8 = 154", for Ex,= 3.4 eV the reflected H has a single glancing collision with the parent I [Fig. 13(a)], whereas for Ex, =6.0eV [Fig. 13(b)] the H is scattered directly back toward I and chatters several times between I and the surface. The resultant final translational energy was E!,. =2.2 eV (energy loss 1.2 eV) in the former case and Ek =2.8 eV (energy loss 3.2 eV) in the latter. 4.3.2 Inelastic Collisions fi +HI(ad) The solid line in Fig. 14 shows preliminary calculated results for scattered H-atom energy distributions simulating 193 nm photolysis at 0.7 ML coverage.We have LAS in the Photolysis of HI/LiF(001) 6 20 . (a) I I 30 5- 40 v) 50 2 4 -.:.. 0 60 9 ?E 70 3- .. *..; E'(H) - 80 90 135 140 145 150 1 5 Oldegrees Fig. 11 (a)Energy of the scattered H-atom E& as a function of initial tilt angle (solid line; left-hand ordinate axis). The scattering angle is shown as a broken line (right-hand ordinate axis). (b) Energy of the scattered I-atom and the energy absorbed by the surface, as a function of initial tilt angle. The panels designated I-IV indicdte the same regions as in Fig. 9 and 10. l'00120 130 140 O 150 1 tl/deg reesk L4777-Fig. 12 Energy of the scattered H-atom, Ek, as a function of initial tilt angle, 8,for four different excess photolysis energies, Ex,, as noted V.J. Barclay et al. L 4 0 \ /\/+\ -/ \""!"~'!"''!"'~'"'0 included two channels for HI photolysis, combined according to the gas-phase branch- ing ratio15*17 of 9/1. The values of Ex, governing the energy of the collision with the surface were 3.4 eV for H(1) and 2.5 eV for H(I*). The simulation gave a ratio of fast scattered H-atoms [the H(1) channel] to slower H-atoms [the H(I*) channel] that was about one order of magnitude less than the gas-phase ratio. This high degree of inelas- ticity corresponds qualitatively to the observed loss of the high-energy peak in the scat- tered H energy distribution; see Fig.5(b). According to these model c_alculations the major pathway for inelastic energy loss from scattered H stems from H +HI(ad) collisions. We have divided the trajectories LAS in the Photolysis of HI/LiF(001) 1 2 3 4 E;/eV Fig. 14 Heavy solid line: calculated energy distributions for H-atoms after photolysis of HI/ LiF(001) at 0.7 ML with A = 193 nm, showing peaks due to the I, I* and inelastically scattered channels. For reference, the gas-phase ratios (ref. 15, 17) are_ indicated as vertical lines normalised to H(I*). The broken line indicates the contributicn from H + Surf collisions, and the light solid line gives the contribution from H + HI(ad) collisions (see text). that go to make up the computed P(ET)curve in Fig.14 into a component aFribable to H + Surf collisions (broken line) and a second (larger) component due to H + HI(ad) (thin solid line). The criterion for separation was the magnitude of the impact parameter with respect to a 'target' HI(ad). If, after collision with the surface a trajectory had an impact parameter in excess of 2.5 8, we considered that it would transfer a minor amount of energy to the target molecule. This impact parameter was selected as being slightly in excess of the maximum impact parameter that led to reaction. It has the effect of separating the encounters into two categories with markedly different attributes. For b 2.5 (dashed curve, Fig. 14) a large fraction of the trajectories result in an energy loss of only ca.0.2 eV and a small fraction have an energy loss of 1-2 eV. By contrast for trajectories with b 2.5 8, the most probable outcome is an energy loss in the region of 1-2 eV. It appears, therefore, that the major contribution to the energy loss from the 3.4 eV H-atoms at 0.7 ML, according to this model, comes from + HI(ad) inelastic encounters. This type of inelastic encounter is explored in a forthcoming paper.20 At lower coverages, particularly in the absence of islanding, the major energy loss can be expected to arise from collisions with the surface. In studying H-atom angular distributions by recording P(W) for only the highest-energy H-atoms, even at ca. 1 ML, one is selecting those H that were reflected from the surface but did not encounter a co-adsorbate molecule (see Fig.14). We have made use of this in our previous work" on HBr/LiF(001) in order to probe the potential in the neighbourhood of F-at the LiF surface. In a subsequent paper on H + Surf in HI/LiF(001) we shall discuss the form of P(@)for the fast elastically scattered component of H, as obtained from both experiment and theory.37 We thank Darrick Heyd and Erik Jensen for kindly verifying some TOF measurements on a differently configured apparatus. We thank David Jack for his assistance with the HI parameters and Colin Stanners for helpful discussions. We are also indebted to the Natural Sciences and Engineering Research Council (NSERC) of Canada, the Ontario V. J. Barclay et al. 149 Laser and Lightwave Research Centre, and the National Networks of Centres of Excel-lence Program for their support of this work.References 1 W. Ho, in Desorption Induced by Electronic Transitions, DIET IV, ed. G. Betz and P. Varga, Springer- Verlag, Berlin, 1990, p. 48. 2 X-L. Zhou, X-Y. Zhu and J. M. White, Surf. Sci. Rep., 1991,13,73. 3 J. C. Polanyi and H. Rieley, in Dynamics of Gas-Surface Interactions, ed. C. T. Rettner and M. N. R. Ashfold, Royal Society of Chemistry, London, 1991, p. 329. 4 J. C. Polanyi and Y. Zeiri, in Laser Spectroscopy and Photochemistry on Metal Surfaces, ed. H. L. Dai and W. Ho, in the press. 5 P. M. Blass, R. C. Jackson, J. C. Polanyi and H. Weiss, J. Chem. Phys., 1991,94,7003. 6 J. C. Polanyi, Faraday Discuss. Chem. SOC., 1991,91,451.7 J. C. Polanyi, R. J. Williams and S. F. OShea, J. Chem. Phys., 1991,94,978. 8 V. J. Barclay, D. B. Jack, J. C. Polanyi and Y. Zeiri, J. Chem. Phys., 1992,97,9458. 9 V. J. Barclay, J. C. Polanyi, Y. Zeiri and R. Kosloff, J. Chem. Phys., 1993,98,9185. 10 E. B. D. Bourdon, C. C. Cho, P. Das, J. C. Polanyi, C. D. Stanners and G. Q. Xu, J. Chem. Phys., 1991, 95, 1361. 11 M. I. McCarthy and R. B. Gerber, J. Chem. Phys., 1990,93,887. 12 R. Alimi and R. B. Gerber, Phys. Rev. Lett., 1990,64, 1453. 13 J. Segall, Y. Wen, R. Singer, C. Wittig, A. Garcia-Vela and R. B. Gerber, Chem. Phys. Lett., 1993, 207, 504. 14 C. E. Moore, Atomic Energy Levels as Derived from the Analysis of Optical Spectra, Vol. I, National Bureau of Standards, 1949. 15 Z.Xu, B. Koplitz and C. Wittig, J. Phys. Chem., 1988,92, 5518. 16 D. A. Redhead, Vacuum, 1962,12,203. 17 G. N. A. Van Veen, K. A. Mohamed, T. Baller and A. E. de Vries, Chem. Phys., 1983,80, 113. 18 C. D. Stanners, Ph.D. Dissertation, University of Toronto, 1990. 19 J. C. Tully, G. H. Gilmer and M. Shugard, J. Chem. Phys., 1979,71, 1630. 20 V. J. Barclay, D. B. Jack, J. C. Polanyi and Y. Zeiri, J. Phys. Chem., in the press. 21 V. J. Barclay, D. B. Jack, J. C. Polanyi, Y. Zeiri and R. R. Lucchese, unpublished. 22 J. E. Lennard-Jones and B. M. Dent, Trans. Faraday SOC., 1928,24,92. 23 K. P. Huber and G. Herzberg, Molecular Spectra and Molecular Structure Vol. ZV. Constants of Diatomic Molecules, Van Nostrand, Princeton, 1979. 24 A. D. Buckingham and P.W. Fowler, J. Mol. Struct., 1988,189,203. 25 K. T. Tang and J. P. Toennies, J. Chem. Phys., 1984,80,3726. 26 W. A. Steele, The Znteraction of Gases with Solid Surfaces, Pergamon, Oxford, 1974. 27 N. Metropolis, A. W. Rosenbluth, M. N. Rosenbluth, A. H. Teller and E. Teller, J. Chem. Phys., 1953, 21, 1087. 28 J. 0. Hirschfelder, C. F. Curtiss and R. B. Bird, Molecular Theory of Gases and Liquids, John Wiley, New York, 1954. 29 H. Eyring and M. Polanyi, Z. Phys. Chem. (Leipzig) B, 1931, 12, 279; S. Sato, J. Chem. Phys., 1955, 23, 592; 1955,23,2465; Bull. Chem. Soc. Jpn., 1955,28,450. 30 J. C. Polanyi and R. J. Williams, J. Chem. Phys., 1988,88,3363. 31 P. M. Aker and J. J. Valentini, Isr. J. Chem., 1990, 30, 157. 32 H. S. Johnson, Gas Phase Reaction Rate Theory, Ronald Press, New York, 1966. 33 E. Shustorovich, Acc. Chem. Res, 1988,21, 183. 34 Gmelin Handbook of Inorganic and Organometallic Chemistry, Br Supplement Vol. B 1, Springer, Berlin, 8th edn., 1990, p. 144. 35 F. A. Mauer, C. J. Keffer, R. B. Reeves and D. W. Robinson, Chem. Phys., 1965,42, 1465. 36 A. Bondi, J. Phys. Chem., 1964,68,441. 37 V. J. Barclay, W-H. Hung, J. C. Polanyi, Y. Zeiri and G. Zhang, unpublished. 38 R. D. Shannon, Acta Crystallogr., Sect. A, 1976,32, 751. Paper 3/04280D; Received 19th July, 1993