Selective collision-induced desorption: Measurement of the -bonded C 2 H 4 binding energy on Ptˆ111 precovered with atomic oxygen

Transcription

1 Selective collision-induced desorption: Measurement of the -bonded C 2 H 4 binding energy on Ptˆ111 precovered with atomic oxygen D. Velic and Robert J. Levis Department of Chemistry, Wayne State University, Detroit, Michigan Received 20 November 1995; accepted 13 February 1996 Collision-induced desorption CID is used to selectively probe the well depth of one particular adsorbate surface potential energy surface in the presence of multiple adsorbates. Ethylene adsorbed at saturation coverage on Pt 111 precovered with atomic oxygen represents a system with three adsorbates: oxygen and two different adsorption forms of ethylene. Both the -bonded ethylene -C 2 H 4 and di- -bonded ethylene di- -C 2 H 4 species are formed at coverages of 0.25 ML preadsorbed atomic oxygen and 0.15 ML ethylene on Pt 111 at 100 K. Deconvolution of Al K x-ray photoelectron XP spectra suggests that C(1s) XP binding energy is ev for -C 2 H 4 and is ev for di- -C 2 H 4. The C(1s) XP spectra together with the CID result reveal that the coverage ratio of -C 2 H 4 and di- -C 2 H 4 at saturation at 100 K is 1:1. The selectivity of CID is demonstrated by desorbing only -C 2 H 4 using a neutral Xe atomic beam with translational energies ranging from 2.1 to 4.1 ev. In this translational energy range, di- -C 2 H 4 remains intact on the O/Pt 111 surface. The threshold energy for CID of -C 2 H 4 was determined by extrapolation to the Xe energy where the CID cross section is equal to zero. The threshold of ev can be related to the -C 2 H 4 Pt binding energy using classical collision mechanics. An upper limit for the -C 2 H 4 binding energy on O/Pt 111 is calculated to be ev from the CID threshold measurement, using a Pt effective mass of 1.5 Pt atom American Institute of Physics. S I. INTRODUCTION The collision-induced desorption CID experiment employs a hyperthermal, neutral, noble gas atom to eject an adsorbate from a surface into the gas phase. Applying gas phase molecular collision dynamics, 1 the CID experiment can be used to nonthermally probe the potential energy surface PES of adsorbate surface systems. Zeiri et al. 2 first suggested that analysis of the desorbate angular and kinetic energy distributions could be useful for determining an adsorbate surface binding energy. Beckerle et al. 3 then measured the cross section for CID of physisorbed methane on Ni 111 as a function of both the kinetic energy and angle of incidence of the hyperthermal Ar atom. Szulczewski and Levis 4 proposed that the threshold energy for CID is directly proportional to the adsorbate surface binding energy and the methane Ni 111 binding energy was calculated using a hard sphere/cube model. In this calculation, the mechanism of CID was approximated by three steps. The first step is a binary collision between the incident hyperthermal atom and the adsorbate. The second step is a subsequent impulsive interaction between the adsorbate and the surface. The third step assumes desorption of the adsorbate from the surface if the momentum component of the adsorbate translational energy directed away from the surface exceeds the binding energy. We have subsequently investigated the CID of the chemisorbed systems: ammonia on Pt 111 Ref. 5 and di- -bonded ethylene on Pt 111 Ref. 6 which represent systems with one chemical bond and two chemical bonds per adsorbate with the surface, respectively. In this paper the use of CID is extended to investigate a multiadsorbate system. Ethylene adsorbed on Pt 111 precovered with atomic oxygen C 2 H 4 /O/Pt 111 represents a system with three adsorbates as shown in scheme 1. These adsorbates are atomic oxygen and two different forms of ethylene, -C 2 H 4 and di- -C 2 H 4. The -C 2 H 4 adsorbate is, presumably, more weakly bound on O/Pt 111 than the di- -C 2 H 4 adsorbate, because the temperature programmed desorption TPD peak temperature is 185 K for -C 2 H 4 as compared with 265 K for di- -C 2 H Thus, the C 2 H 4 /O/Pt 111 system, having two C 2 H 4 adsorbates with clearly different binding energies, was chosen to investigate the selectivity of the CID technique. The -C 2 H 4 species is formed on Pt 111 only under specific conditions. In the present study, -C 2 H 4 is formed in the presence of preadsorbed atomic oxygen at 100 K. The coexistence of -C 2 H 4 with di- -C 2 H 4 has been previously demonstrated using high resolution electron energy loss spectroscopy HREELS at the C 2 H 4 saturation coverage when 0.23 monolayer ML of J. Chem. Phys. 104 (23), 15 June /96/104(23)/9629/11/$ American Institute of Physics 9629

2 9630 D. Velic and R. J. Levis: Selective collision-induced desorption atomic oxygen was preadsorbed on Pt 111 at 92 K. In that study, a band measured at 720 cm 1 was assigned to the CD 2 wagging mode of -bonded ethylene by comparing the EELS spectra at 92 and 280 K. Ultraviolet photoelectron spectroscopy 8 UPS binding energies have been assigned for C 2 H 4 /O/Pt 111 as C C at 4.2 ev, CH2 at 6.6 ev, CH2 / C C at 8.9 ev and CH2 at 9.9 ev supporting the formation of -C 2 H 4. Near edge x-ray absorption fine structure NEXAFS features 10 in the spectrum of C 2 H 4 /O/Pt 111 were also assigned as * at 285 ev and CC * at a range from 294 to 301 ev using synchrotron radiation at an angle of 50 incidence. Examples of -C 2 H 4 bound to transition metals 11 can be described by the Dewar Chatt Ducanson model or the parameter 12 proposed by Stuve and Madix. For instance, the parameter has been proposed as a measure of the rehybridization of carbon from sp 2 to sp 3. The value of the parameter for C 2 H 4 in the gas phase was defined to be 0, representing sp 2 hybridization with an undistorted C C bond. Zeise s salt represents a Pt -C 2 H 4 bond with electron density sharing between the C C bond and the Pt atom and the parameter is For the C 2 H 4 /Pt 111 system at 100 K the parameter is 0.92, representing sp 3 hybridization with two strong C Pt bonds. 6 The -C 2 H 4 species was not formed on Pt 111 at 100 K, although this species has been formed and characterized by TPD and EELS on 1 1 Pt 110, Pt 110, 14 and on Pt This formation of -C 2 H 4 species for metallic coordination of 7 and below 16 was supported by extended-huckel calculations. The -C 2 H 4 species has been observed on Pt 111 only at temperatures below 52 K as characterized by TPD and UPS. 17,18 At temperatures above 52 K the -C 2 H 4 transformation into di- -C 2 H 4 occurred. Calculations 19 revealed no barrier between -C 2 H 4 and di- -C 2 H 4 on Pt 111. The -C 2 H 4 species has been observed at 100 K on O/Pt 111, 7 10 on Pt 111 precovered with cesium, 8,17 on Pt 111 precovered with potassium 8,10,20 22 and on Pt 111 precovered with potassium and oxygen. 23 All predeposited elements were in a submonolayer coverage and, presumably, their role was to perturb the Pt PES. This perturbation is typically described by electrostatic and donation/back donation models 22 and results in the formation of the -C 2 H 4 species. In this paper CID is investigated for C 2 H 4 on Pt 111 precovered with atomic oxygen. Oxygen has been observed to adsorb molecularly on Pt 111 as a peroxo species below 150 K Ref. 24 with a saturation coverage O ML. 25 Molecular oxygen was observed to partially desorb at 160 K and partially dissociate into atomic oxygen. The dissociation can take place even at lower temperatures 92 K when the molecular oxygen coverage is below the saturation coverage. 25 In the K temperature range 24 atomic oxygen is present with a saturation coverage O 0.25 ML and forms a 2 2 low energy electron diffraction LEED pattern. In the K temperature range 24 a subsurface oxide may form associated with surface impurities. 9 In the experiments reported in this paper, selective CID was used to collisionally desorb only the -C 2 H 4 adsorbate, leaving the di- -C 2 H 4 adsorbate intact on the O/Pt 111 surface. In addition to investigating the selectivity of the CID technique we also use CID to determine the -C 2 H 4 Pt 111 binding energy. The CID experiment is one of the few techniques including TPD, 26 isosteric measurements, 27 and microcalorimetry, 28 capable probing adsorbate-surface chemical binding energies. As a nonthermal tool this method can also shed some light on the desorption energy measurements 26,28 made using thermal techniques. The remainder of this paper is organized as follows. In Sec. II we briefly describe the experimental apparatus. In Sec. III the preparation of C 2 H 4 /O/Pt 111 system is described, as well as the characterization of the -C 2 H 4 and di- -C 2 H 4 species using x-ray photoelectron spectroscopy XPS. The difference in C(1s) XP binding energy between the -C 2 H 4 and di- -C 2 H 4 species is used to monitor the -C 2 H 4 surface coverage to demonstrate the selectivity of CID. In Sec. IV the binary collision energy transfer model is used to relate the threshold Xe energy for CID of -C 2 H 4 to the -C 2 H 4 Pt chemical binding energy. Sections V and VI provide a discussion and a summary of the findings, respectively. II. EXPERIMENT Experiments were performed in an ultrahigh vacuum UHV chamber, with a base pressure of torr, coupled to a hyperthermal neutral atomic beam source. The experimental apparatus has been described previously. 5 Briefly, the UHV system contains facilities for electronbeam heating, liquid nitrogen cooling, ion sputtering, quadrupole mass spectroscopy QMS, and XPS. All of the XP spectra reported here employed 230 W of Al K radiation at ev. The O(1s) XP spectra were used to verify a constant coverage of atomic oxygen. The C(1s) XP spectra were monitored to differentiate -C 2 H 4 from di- -C 2 H 4 and to quantify the -C 2 H 4 coverage before and after CID as described subsequently. The desorbed ethylene was monitored in the gas phase by comparing the QMS cracking pattern of backfilled C 2 H 4 with the QMS signal of collisionally desorbed C 2 H 4. The Pt 111 crystal was cleaned by cycles of argon sputtering 5 kv and annealing at 1200 K. Between individual experiments, residual surface carbon was removed by oxygen treatment 10 7 torr at 1000 K and afterwards the crystal was flashed to 1200 K. No impurities were detected at the XPS detection level 1%. Except where noted, all the experiments were performed with the sample maintained at a temperature of 100 K. The CID experiment is performed using a hyperthermal Xe beam produced via a supersonic free-jet expansion from a heated nozzle. 29 The nozzle system is constructed from a sealed molybdenum tube with a 100 m diameter conical orifice. A mixture of Xe seeded in either 1/2% or 1% H 2 was used to form the supersonic beam. Note that control experiments employing only the H 2 carrier gas revealed no change of the initial C(1s) and O(1s) XP spectra for the C 2 H 4 /O/Pt 111 system. The atomic beam nozzle could be heated up to 1200 K to vary the beam translational energy.

3 D. Velic and R. J. Levis: Selective collision-induced desorption 9631 The temperature of the nozzle was measured by chromel alumel thermocouple attached to the outside of nozzle tube. The beam travels through a 1 mm diameter skimmer placed 2 cm away from the nozzle. The translational energy of the Xe beam was varied from 2.1 to 4.1 ev and was determined by a time-of-flight TOF technique using a quadrupole mass detector as described previously. 5,6 The QMS was used to determine the flux and translational energy distribution of the Xe beam prior to each experiment. For instance, a typical flux for the atomic beam at 3.0 ev of translational energy was Xe atoms/sec cm 2. Three angles of incidence, 0, 30 and 45, were used to investigate the angular dependence of CID for C 2 H 4 /O/Pt 111. III. RESULTS: XPS CHARACTERIZATION AND SELECTIVE CID The ratio of -C 2 H 4 and di- -C 2 H 4 is dependent on the atomic oxygen surface coverage, therefore the preparation of a constant O/Pt 111 system was essential. The atomic oxygen coverage was interrogated by XPS and was consistent with previously reported XPS characterizations. 24,30 The following procedure was used to prepare the atomic oxygen on the Pt 111 surface. The clean Pt 111 surface was saturated with 6 L 1 L 10 6 Torr sec of molecular oxygen at 100 K forming O ML. The sample was then annealed to 170 K to convert the molecular oxygen into atomic oxygen. The initial saturation coverage of oxygen atoms/cm 2 can be approximately doubled using extended temperature cycling 24 or external atomization. Therefore, the O 2 exposure and annealing process was performed only once to maintain the constant atomic oxygen coverage. Control experiments were performed to ensure that neither the ionization gauge nor the quadrupole mass spectrometer altered the atomic oxygen coverage. CID experiments were performedonlywhen O / O2 0.25/ %.In Fig.1, the O(1s) XP spectra for the a O2 and b O are shown. An atomic oxygen coverage of 0.25 ML, as shown in Fig. 1 b, was used in all CID experiments. After the atomic oxygen monolayer was prepared, the crystal was cooled to 100 K and C 2 H 4 was then adsorbed on the surface. To determine whether the Xe beam would desorb atomic oxygen from C 2 H 4 /O/Pt 111, the O(1s) XP spectra were monitored before and after the CID experiments. The typical O(1s) XP spectra for C 2 H 4 /O/Pt 111 system before and after CID are shown in Figs. 1 c and 1 d, respectively. No change in the O(1s) XP spectra was observed between Figs. 1 c and 1 d. This suggests that no CID of atomic oxygen occurred at the energies used in this study. As another control experiment we used a Xe atomic beam with translational energy of 4.1 ev to collisionally prepare a monolayer of atomic oxygen on Pt 111 from a saturation coverage of molecular oxygen as shown in Fig. 1 a, at 100 K. The spectrum is shown in Fig. 1 e and is almost identical with the thermally prepared atomic oxygen in Fig. 1 b normally used in all experiments. This also supports the observation that the atomic oxygen adsorbed on Pt 111 at 100 K is collisionally stable in the CID experiments. FIG. 1. O(1s) XP spectra for a the saturation coverage of molecular oxygen on Pt 111 at 100 K; b the saturation coverage of atomic oxygen prepared by annealing the molecular oxygen shown in a to 170 K, measured at 100 K; c the spectrum of atomic oxygen for C 2 H 4 /O/Pt 111 before CID; d the spectrum of atomic oxygen for C 2 H 4 /O/Pt 111 after a typical CID experiment; e the atomic oxygen coverage prepared using 4.1 ev Xe beam collisions from the saturation coverage of molecular oxygen on Pt 111. To calibrate the C 2 H 4 coverage on O/Pt 111, the C(1s) XP spectra were first recorded as a function of C 2 H 4 exposure as shown in Fig. 2. In each spectra the background due to the Pt 4d 5/2 feature has been subtracted. The integrated intensity of the C(1s) XP feature is plotted as a function of C 2 H 4 exposure in Fig. 3. Steininger et al. reported 7 that the saturation coverage of C 2 H 4 adsorbed at 92 K on Pt 111 precovered with 0.23 ML atomic oxygen was reduced to 60% of the saturation coverage on clean Pt 111. The saturation exposure for C 2 H 4 on O/Pt 111 was measured to be approximately 1.3 L as shown in Fig. 3. This value is consistent with 60% of the 2.2 L saturation exposure 31 reported for the C 2 H 4 adsorption on clean Pt 111. The C 2 H 4 saturation coverage on clean Pt 111 was reported 31 to be 0.25 ML. Assuming constant sticking coefficient and using our measured 60% coverage value, the C 2 H 4 saturation coverage on O/Pt 111 was estimated to be 0.15 ML. Steininger et al. 7 demonstrated the coexistence of -C 2 H 4 and di- -C 2 H 4 on 0.23 ML O/Pt 111 at 92 K at C 2 H 4 saturation coverage using EELS and TPD. At low C 2 H 4 coverages no desorption feature was observed for the more weakly bound -C 2 H 4 species and only at about 300 K did the di- -C 2 H 4 species desorb. The TPD data demonstrated that with increasing ethylene coverage, the di- -C 2 H 4 desorption feature saturated and became broader, eventually

4 9632 D. Velic and R. J. Levis: Selective collision-induced desorption FIG. 4. Full width at half maximum of C(1s) XP spectra as a function of C 2 H 4 exposure on O/Pt 111 at 100 K. FIG. 2. C(1s) XP spectra for various C 2 H 4 exposures on O/Pt 111 at 100 K. a 0.3 L; b 0.5 L; c 0.7 L; d 1.0 L; e 1.3 L. The diamonds are measured data points and the thick solid line represents the best fit. The thin dashed line represents -C 2 H 4 and the solid thin line represents di- -C 2 H 4. The best fit of e represents the di- -C 2 H 4 and -C 2 H 4 deconvolution using Gaussians 1.9 ev FWHM at and ev, respectively. formed a doublet. At high C 2 H 4 coverages the desorption of -C 2 H 4 occurred. At the saturation coverage, -C 2 H 4 desorbed at 185 K and di- -C 2 H 4 desorbed at 265 K. This di- -C 2 H 4 desorption temperature was in good agreement with subsequent TPD measurements ranging from 235 K Ref. 9 to 275 K Ref. 8. Using EELS, Steininger et al. 7 FIG. 3. C(1s) XP integrated intensity as a function of C 2 H 4 exposure on O/Pt 111 at 100 K. The saturation is observed at 1.3 L, representing 0.15 ML of C 2 H 4. concluded that -C 2 H 4 was the initial form of adsorbed C 2 H 4 on O/Pt 111. The fact that no -C 2 H 4 desorbed at low C 2 H 4 coverages was attributed to a thermal conversion into di- -C 2 H 4 which consequently desorbed at higher temperatures. 7 Some insight into the surface reaction chemistry of C 2 H 4 on O/Pt 111 can be obtained by analyzing both the C(1s) XP binding energy shift and the change in the full width at half maximum FWHM of the C(1s) XP spectra as a function of C 2 H 4 exposure. The C(1s) XP binding energy slightly shifts from an initial value of to ev at C 2 H 4 saturation. The FWHM changes from 1.9 to 2.2 ev as the C 2 H 4 exposure is increased, as shown in Fig. 4. This is consistent with the presence of two C 2 H 4 species on O/Pt 111. Figure 2 displays the deconvolution of the spectra corresponding to 0.3, 0.5, 0.7, 1.0, and 1.3 L C 2 H 4 exposure. The low coverage spectrum shown in Fig. 2 a represents the initially adsorbed C 2 H 4 species and was fit with a 1.9 ev FWHM Gaussian. The spectrum shown in Fig. 2 b reveals an increase in the C 2 H 4 surface coverage with no change of FWHM indicating the presence of only the initially formed C 2 H 4 species. At 0.7 L exposure, as shown in Fig. 2 c, a measurable increase in the FWHM was observed. This increase is consistent with the presence of two C 2 H 4 species: -C 2 H 4 and di- -C 2 H 4. At 1.0 L exposure, shown in Fig. 2 d, the FWHM of C(1s) XP feature further increases due to the additional formation of the second C 2 H 4 species. Figure 2 e shows the C(1s) XP spectrum of the saturation C 2 H 4 coverage, deconvoluted into two peaks. Using Steininger s suggestion 7 of initial C 2 H 4 formation on O/Pt 111, we can assign the initially formed C 2 H 4 species as -C 2 H 4. The di- -C 2 H 4 formation contribution was then observed at 0.7 L exposure suggesting a stepwise formation: first -C 2 H 4 : second di- -C 2 H 4. Therefore the two features of C(1s) XP spectrum shown in Fig. 2 e are assigned for di- -C 2 H 4 at ev and for -C 2 H 4 at ev. This

5 D. Velic and R. J. Levis: Selective collision-induced desorption 9633 formation mechanism can be rationalized using donation/ back donation mechanism. 22 On clean Pt 111 at 100 K, C 2 H 4 exclusively adsorbs as di- -C 2 H 4 species 7 as soon as atomic oxygen is preadsorbed the Pt 111 PES is perturbed. This perturbation is described by the electronegative oxygen withdrawing of electrons from the Pt 111 surface. The adsorbing C 2 H 4 can either donate electrons into Pt d band and form the -C 2 H 4 species or get back donation from the Pt d band into the * orbital and form the di- -C 2 H 4 species. In a case of 0.25 ML O/Pt 111 the Pt surface electron deficiency is presumably high and forces C 2 H 4 to donate electrons resulting in the initial -C 2 H 4 formation. As soon as the lowest energy adsorption sites are titrated by -C 2 H 4, the formation of di- -C 2 H 4 is possible. Zhou et al. 21 investigated C 2 H 4 adsorption at 100 K on Pt 111 precovered with potassium. This study showed the increasing -C 2 H 4 /di- -C 2 H 4 ratio as a function of potassium coverage. At low less than 0.5 L C 2 H 4 exposures on 0.06 ML K/Pt 111 the -C 2 H 4 species thermally desorbed at 145 K. No di- -C 2 H 4 desorption was observed, which was attributed to the di- -C 2 H 4 thermal decomposition. At C 2 H 4 exposures higher than 0.5 L, the di- -C 2 H 4 desorption was observed at 285 K and the TPD features of both species increased steadily with C 2 H 4 exposure. This experiment together with the EELS investigation suggested an almost simultaneous formation of both ethylene species on 0.06 ML K/Pt 111. The perturbation of the Pt 111 electronic structure caused by electronegative atomic oxygen and electropositive potassium are clearly different. The fact that both C 2 H 4 /O/Pt 111 and C 2 H 4 /K/Pt 111 systems form -C 2 H 4 suggests that the adsorption mechanism is determined by more than one process. Since the -C 2 H 4 /di- -C 2 H 4 adsorption chemistry as a function of C 2 H 4 and preadsorbate surface coverages is complex, we focus on CID investigations at C 2 H 4 saturation coverage on O/Pt 111. At this coverage, both -C 2 H 4 and di- -C 2 H 4 coexist as clearly shown using EELS Ref. 7 and our XPS measurements. The CID scheme was used to further differentiate the two C 2 H 4 species and to selectively quantify the -C 2 H 4 surface coverage. The TPD measurements for the C 2 H 4 /O/Pt 111 system suggest that the -C 2 H 4 Pt binding energy is weaker than the di- -C 2 H 4 Pt binding energy. Therefore, we assume that a Xe atom having translational energy lower than the threshold for CID of di- -C 2 H 4 will selectively desorb only -C 2 H 4. Obviously, the Xe atom must also have translational energy above the threshold for CID of -C 2 H 4. Using these assumptions and knowing the translational energy threshold for CID of di- -C 2 H 4 /Pt 111 of 5.2 ev Ref. 6 we should be able to use a range of Xe translational energies less than 5.2 ev to effectively desorb the -C 2 H 4 species, presumably leaving the di- -C 2 H 4 species intact on the O/Pt 111 surface. Note that since we have not measured the di- -C 2 H 4 binding energy on O/Pt 111, we used the TPD temperature to provide some indication of the bond strength. The peak temperatures in the TPD spectra for di- -C 2 H 4 in C 2 H 4 /O/Pt 111 and C 2 H 4 /Pt 111 are 265 and 280 K, respectively. This suggests that the di- -C 2 H 4 binding in the C 2 H 4 /O/Pt 111 system is slightly weaker than FIG. 5. C(1s) XP spectra. a The initial coverage of C 2 H 4 on O/Pt 111 at 100 K used in all CID experiments ; b The final coverage of C 2 H 4 after a total of 600 s of CID 4.1 ev of Xe translational energy, 45 angle of incidence ; c the C 2 H 4 coverage after a total of 1800 s of CID 4.1 ev of Xe translational energy, 45 angle of incidence ; d the C 2 H 4 coverage after the initial coverage shown in a was annealed to 240 K. The two deconvolution features employed represent di- -C 2 H 4 thin solid line at ev and -C 2 H 4 thin dashed line at ev. in the C 2 H 4 /Pt 111 system. To be conservative, we used Xe translational energies of 4.1 ev and lower. As will be seen, these energies do not effect di- -C 2 H 4. Figure 5 a shows the saturation coverage of C 2 H 4 adsorbed at 100 K on O/Pt 111. This coverage was used as an initial coverage in all CID experiments. As described previously, background subtraction, peak integration and deconvolution were used to identify at least two C(1s) XP features as shown in Fig. 2 e. Figure 5 a shows the C(1s) XP spectrum of the initial state of C 2 H 4 /O/Pt 111 deconvoluted into the -C 2 H 4 and di- -C 2 H 4 features; this XP spectrum is identical with that shown in Fig. 2 e. Figure 5 b represents the C(1s) XP spectrum as a result of 600 s of collision-induced desorption of the C 2 H 4 /O/Pt 111 system shown in Fig. 5 a, the incident Xe atoms had 4.1 ev of translational energy. This spectrum shows a depletion of -C 2 H 4 feature, as assigned from Fig. 2 e, while the di- -C 2 H 4 feature remains unchanged. The virtually complete depletion of -C 2 H 4 feature is shown in Fig. 5 c. This spectrum represents a total of 1800 s of CID of the C 2 H 4 /O/Pt 111 system shown in Fig. 5 a using 4.1 ev of Xe translational energy. The fit of Fig. 5 c was done using a 1.9 ev FWHM Gaussian at ev and is consistent with the di- -C 2 H 4 feature in Fig. 5 a. Subsequent exposure of the system shown in Fig. 5 c to 4.1 ev Xe atoms results in no measurable change in the C(1s) XP spectrum. We also note that the result shown in Fig. 5 c is consistent with the hypothesis that Xe atoms of 4.1 ev translational energy have insufficient energy to collisionally desorb the di- -C 2 H 4 species from O/Pt 111 at 100 K. The assignment of C(1s) XP binding energies as ev for di- -C 2 H 4 and ev for -C 2 H 4 as shown in Fig. 2 e is consistent with the result of the selective CID experi-

6 9634 D. Velic and R. J. Levis: Selective collision-induced desorption ment. An additional thermal experiment described subsequently confirmed this observation. Note that a C(1s) XP binding energy of ev for di- -C 2 H 4 was reported 32 for the C 2 H 4 /Pt 111 system. Cassuto et al. 10 reported C(1s) XP binding energies for C 2 H 4 /K/Pt 111 at 95 K. In that experiment, high potassium coverages resulted in average C(1s) XP binding energies of ev for di- -C 2 H 4 and ev for -C 2 H 4. Those values are not in agreement with our C(1s) XP binding energy values, perhaps because of different initial and final state effects when C 2 H 4 is adsorbed on O/ and K/Pt 111. To further probe the identity of the C(1s) XP feature at ev, a saturated C 2 H 4 /O/Pt 111 system identical to that shown in Fig. 5 a was annealed to 240 K. The result of this experiment is shown in Fig. 5 d. This C(1s) XP spectrum has the same FWHM and XP binding energy as the di- -C 2 H 4 feature in Fig. 5 a strongly suggesting that the spectrum shown in Fig. 5 d represents the di- -C 2 H 4 species. This observation is also consistent with Steininger s 7 result that -C 2 H 4 thermally converts into di- -C 2 H 4. Note that the shift of C(1s) XP binding energy is expected after annealing to temperatures higher than 240 K due to formation of CO. The amount of thermally prepared di- -C 2 H 4 shown in Fig. 5 d represents about 2/3 of initial saturation C 2 H 4 coverage shown in Fig. 5 a. This suggests that approximately 1/3 of the initially adsorbed C 2 H 4 desorbed upon heating. This observation is in quantitative agreement with Steininger s TPD areas corresponding to -C 2 H 4 and di- -C 2 H 4 desorption. Note that at low C 2 H 4 coverages the -C 2 H 4 species does not desorb upon heating, but is converted into the di- -C 2 H 4 species. Recall that at the C 2 H 4 saturation coverage both thermal desorption and conversion into di- -C 2 H 4 is observed. This observation can be rationalized by assuming that, at C 2 H 4 saturation coverage, the sites for conversion of -C 2 H 4 into di- -C 2 H 4 are saturated, hence, the thermal desorption of -C 2 H 4 is observed. As soon as the -C 2 H 4 species desorb, the thermal conversion of -C 2 H 4 into di- -C 2 H 4 may occur. The difference between spectra of Figs. 5 c and 5 d essentially represents the amount of -C 2 H 4 which does not thermally desorb but reacts to form di- -C 2 H 4. However, the deconvolution of C(1s) XP spectra shown in Figs. 5 a, 5 b, and 5 c suggests that the coverage of di- -C 2 H 4 remains constant during the CID experiments. Therefore we assume that the collision-induced cross section for -C 2 H 4 conversion into di- -C 2 H 4 is much lower than the collision-induced cross section for -C 2 H 4 desorption. Within the limit of our XPS measurements we observed no collision-induced conversion of -C 2 H 4 into di- -C 2 H 4. This observation is consistent with the assumption that the CID and TPD experiments access different reaction mechanisms, as will be discussed subsequently. The coverage study and the CID experiment not only differentiate the -C 2 H 4 species from the di- -C 2 H 4 species, but also quantify the surface coverage ratio of -C 2 H 4 and di- -C 2 H 4 /. Based on the adsorption study shown in Fig. 2, di- -C 2 H 4 forms at a C 2 H 4 exposure of 0.7 L which is at about 50% of the C 2 H 4 saturation exposure. This fact and our XPS deconvolution of Fig. 2 e suggest that / is approximately 1. Another estimate of / can be obtained from the results of selective CID experiment performed on the saturated C 2 H 4 /O/Pt 111 system. The estimate is obtained by comparing the C(1s) integrated area before CID as shown in Fig. 5 a and after CID as shown in Fig. 5 c. In essence, this experiment suggests the complete depletion of -C 2 H 4. The / ratio calculated from this CID experiment is 0.98 and is consistent with the adsorption study shown in Fig. 2. Using the estimated / 1, we propose that the -C 2 H 4 coverage is approximately ML at the C 2 H 4 saturation on O/Pt 111 at 100 K. IV. CID BINDING ENERGY DETERMINATION Our hypothesis is that the minimum Xe translational energy required to desorb -C 2 H 4 from the O/Pt 111 surface is proportional to the -C 2 H 4 Pt binding energy BE. This minimum energy is defined as the threshold energy E th.to determine the threshold for CID, the cross section for CID must be measured as a function of Xe translational energy. A plot of the CID cross section as a function of the incident Xe energy reveals E th. The determination of the cross section for CID begins with measurement of the integrated intensity of the C(1s) XP spectra before and after CID. Figures 5 a and 5 b display an example of the XPS analysis for a typical CID experiment. Figure 5 a represents the C(1s) XP spectrum for the initial C 2 H 4 coverage before CID and Fig. 5 b represents the C(1s) XP spectrum for the final C 2 H 4 coverage after 600 s of CID using a 4.1 ev Xe beam at 45 angle of incidence. Because a dissociation of C 2 H 4 was not detected using either XPS or QMS, we conclude that the desorption of -C 2 H 4 is the main product channel. To calculate the absolute cross section for CID of -C 2 H 4, we first subtract the di- -C 2 H 4 intensity from the C(1s) XP spectra shown in Figs. 5 a and 5 b. The subtraction is based on the fact that / 1 and allows us to obtain the absolute -C 2 H 4 coverage before i and after f CID. To determine the order of the CID process the logarithm of ( f / i ) was plotted as a function of the time allowed for CID Xe flux. We observe a linear dependence suggesting that the CID process follows first order kinetics. The absolute CID cross section CID is then calculated from CID ln f / i, 1 F Xe t CID cos i where i is the initial and f is the final surface coverage of -C 2 H 4, respectively, F Xe is the flux of incident Xe atoms, t CID is the time of CID, and i is the angle of incidence with respect to the surface normal. The factor of cos i multiplied by F Xe, measured perpendicular to the Xe beam axis, represents the number of Xe atoms that strike the Pt surface per unit surface area per unit time. The measured cross sections for CID are plotted as a function of Xe translational energies in Fig. 6. The angular dependence of collisional energy transfer into the C 2 H 4 /O/Pt 111 system was investigated using 0, 30, and 45 angles of incidence. The similarity in the energy depen-

7 D. Velic and R. J. Levis: Selective collision-induced desorption 9635 FIG. 6. The collision-induced desorption cross section as a function of Xe translational energy. The squares represent a data set for 30 incidence angle, the triangles represent a data set for 45 incidence angle, and the diamonds represent a data set for 0 incidence angle. The solid line is the empirical fit to the data using Eq. 2. The threshold energy is ev, with the parameters S and N dence of the cross sections for the three angles of incidence suggests that the CID of -C 2 H 4 proceeds via a total energy scaling mechanism. 6 To determine the threshold for CID from Fig. 6 the following empirical function 33 was used: CID S E i E th N, 2 E i where E i is the translational energy of the incident Xe atom, E th is the Xe threshold energy for CID, and S and N are fitting parameters. The general form of Eq. 2 was derived from statistical theories 1 for the simplest example of gas phase collision-induced dissociation: A BC A B C. This reaction represents Xe C 2 H 4 /O/Pt 111 Xe C 2 H 4 O/Pt 111 in this CID investigation. To obtain the threshold for CID of -C 2 H 4 the data shown in Fig. 6 were fit using Eq. 2 convoluted over the Xe translational energy distributions. 5 The fit resulted in E th ev, where S and N The threshold Xe energy for CID of -C 2 H 4 can be related to the -C 2 H 4 Pt binding energy by accounting for energy transfer in two binary collision events as shown in Fig. 7. In our model the -C 2 H 4 Pt binding energy determination is based on energy conservation before and after the collision events. The initial and final energies of the system can be represented by the expression, E Xe,i E Xe, f E Pt,em E C2 H 4,f BE, 3 where E Xe,i is the initial energy of the incident Xe atom before collision, E Xe, f is the final energy of Xe after collision with the -C 2 H 4 adsorbate, E Pt,em is the energy transferred FIG. 7. The selective CID scheme. Panel a represents the C 2 H 4 /O/Pt 111 system with the Xe atom as shown in Scheme 1. The gray spheres represent the Pt 111 surface. The black sphere represents the preadsorbed atom of oxygen. The white sphere of C 2 H 4 structure bound to Pt 111 through a single bond represents the -C 2 H 4 adsorbate. The shadowed sphere of C 2 H 4 structure bound to Pt 111 through two bonds represents the di- -C 2 H 4 adsorbate. The unfilled sphere in the gas phase above the C 2 H 4 /O/Pt 111 system represents the Xe atom. Panel b displays the collision between the Xe atom and the -C 2 H 4 adsorbate. Panel c displays the collision between the -C 2 H 4 adsorbate and the Pt 111 surface. Panel d displays the ejection of -C 2 H 4 adsorbate from the O/Pt 111 surface into the gas phase. into the surface after collision with the -C 2 H 4 adsorbate, E C2 H 4,f is the final energy of the -C 2 H 4 adsorbate after ejection from the O/Pt 111 surface, and BE is the binding energy of the -C 2 H 4 adsorbate on O/Pt 111. Our model for CID of -C 2 H 4 from O/Pt 111 is shown in Fig. 7. Figure 7 a represents the initial state of the system and all of the energy is contained in the translational energy of the incident Xe atom as measured using the time-of-flight TOF technique. Figure 7 b represents the first binary collision event between the incident Xe atom and the -C 2 H 4 adsorbate. We account for this energy transfer using the hard sphere approximation with head on collision impact parameter 0. The energy of -C 2 H 4 immediately after collision, with the incident Xe atom, E C2 H 4,1, is calculated using 4m Xe m C2 H 4 E C2 H 4,1 E Xe,i m Xe m C2 H 4 2, 4 where m Xe is the mass of Xe atom and m C2 H 4 is the mass of the C 2 H 4 molecule. The Xe atom impulsively transfers a quantity of energy into -C 2 H 4 and Xe scatters away from the system with energy E Xe, f E Xe,i E C2 H 4,1. 5 We assume that essentially no energy is transferred into the internal modes of -C 2 H 4 at the threshold for CID. This assumption is based on the adiabaticity parameter, which is defined by the ratio of the collision time to the period of motion for a given internal mode. 1 The hyperthermal Xe atom travels with a velocity of 1900 m/s at the threshold for CID and the interaction distance in the collision can be estimated to be approximately 2 Å. Therefore the collision time

8 9636 D. Velic and R. J. Levis: Selective collision-induced desorption is approximately 100 fs. Comparison with the C H vibrational period of 10 fs suggests that the collision is within the adiabatic limit. 1 In this limit essentially no energy is transferred into the internal modes of -C 2 H 4. Further support for a hard sphere energy transfer can be found in simulations 34 of CID of N 2 from W 100 using an Ar beam. These calculations suggested that less than 4% of the incident translational energy channeled into internal energy at the threshold for CID. In addition we also note that, essentially no excitation of CH 4 at threshold was evident during CID from Ni In the second collision represented by Fig. 7 c, energy is transferred from the translationally energetic -C 2 H 4 adsorbate into the Pt surface. We treat the Pt lattice involved in the second collision as a particle of some effective mass, m eff see subsequent discussion. The energy transferred from the -C 2 H 4 adsorbate into the Pt lattice, E Pt,em, can be calculated from 4m C2 H 4 m eff E Pt,em E C2 H 4,1 m C2 H 4 m eff 2. 6 The energy retained in the scattered -C 2 H 4 molecule after collision with the surface, E C2 H 4,2, can be calculated from the difference of Eqs. 4 and 6 : 4m Xe m C2 H 4 4m E C2 H 4,2 E Xe,i m Xe m C2 H 4 C2 H 4 m eff m C2 H 4 m eff 7 If the energy retained in the -C 2 H 4 adsorbate is larger than the -C 2 H 4 Pt binding energy, the -C 2 H 4 adsorbate can desorb from the O/Pt 111 surface as shown in Fig. 7 d. Therefore the -C 2 H 4 final translational energy with which -C 2 H 4 ejects from the surface is given by E C2 H 4,f E C2 H 4,2 BE. 8 FIG. 8. The calculated -C 2 H 4 Pt binding energy as a function of the Pt surface effective mass. The line was generated using Eq. 7 and the measured threshold energy of 2.4 ev. At the threshold for collision-induced desorption E C2 H 4,f is exactly zero. If E C2 H 4,f is zero, Eq. 8 defines the binding energy of -C 2 H 4 as being equal to E C2 H 4,2. Thus at threshold, we can substitute the value for E C2 H 4,2 given by Eq. 7, into Eq. 8 to obtain the BE. Note that the value for E Xe,i in Eq. 7 is simply the threshold energy of ev experimentally determined from Fig. 6. We note that an identical solution can be obtained from Eq. 3 when E C2 H 4,f is set equal to zero and the values of E Xe,i, E Xe, f, E Pt,em are employed. Using 1.5 Pt atom as m eff of , m Xe of 131, and m C2 H 4 of 28 the -C 2 H 4 Pt binding energy is calculated to be ev. In the calculation of BE the effective mass is the only estimated parameter and represents a certain number of Pt atoms involved in the collision between -C 2 H 4 and Pt surface. The effective mass of the Pt lattice during the scattering process is a complex function of incident particle energy, angle, particle-pt PES and Pt Pt PES. To model 35 the effective masses as function of velocity and particle mass we use well-defined Ar Pt 111 and Xe Pt 111 scattering experiments. A simplification of the physics involved in determining a given effective mass was made by proposing that the velocity of the incident particle in comparison with the speed of sound in the Pt lattice governs the effective mass. At the CID threshold energy of 2.4 ev, the -C 2 H 4 molecule obtains a velocity of 3100 m/s after the collision with the incident Xe atom. Since C 2 H 4 and Ar have a comparable mass in comparison with Xe and C 2 H 4, at least we use the Ar scattering experiments to estimate the effective mass of the Pt lattice during the C 2 H 4 Pt collision. At 3100 m/s the Ar scattering studies suggest that the Pt effective masses are calculated to be 1.5 Pt atom using simulations 36 and 2.0 Pt atoms using scattering experimental results. 37 We conclude that at the -C 2 H 4 velocity of 3100 m/s the Pt effective mass is approximately 1.5 Pt atom. The -C 2 H 4 Pt binding energy is plotted in Fig. 8 as a function of Pt effective mass using Eq. 7. This functional dependence of BE and effective mass provides an estimate of the sensitivity of calculated BE to effective mass. The calculated -C 2 H 4 Pt binding energy varies on average by 10% with a change in the estimated effective mass 0.5 Pt atom. However, a reliable model for calculating the effective mass can increase the accuracy of CID adsorbate-surface binding energy determination. 35 V. DISCUSSION The CID experiment can be used to determine the adsorbate-surface binding energy. 4 6 This CID binding energy measurement is one of few techniques, including TPD, 26 isosteric, 27 measurements, and microcalorimetry 28 capable of probing a chemisorption adsorbate-surface PES. We propose that TPD and CID/microcalorimetry can probe complementary portions of the -C 2 H 4 Pt PES and we will

9 D. Velic and R. J. Levis: Selective collision-induced desorption 9637 show that the CID and microcalorimetry experiments reveal comparable binding energies for -C 2 H 4 on Pt surfaces. Stuck et al. 28 recently reported the first microcalorimetric measurement of a heat of reaction for an adsorbate bound to a single crystal surface. In that study the C 2 H 4 /Pt system was investigated at 300 K. The formation of -C 2 H 4 was observed to occur reversibly at above 0.6 ML coverage. The heats of reaction for -C 2 H 4 and di- -C 2 H 4 were directly measured to be 120 kj/mol 1.24 ev and 136 kj/mol 1.41 ev, respectively. The -C 2 H 4 heat of reaction represents the amount of energy released due to -C 2 H 4 adsorption. Thus the heat of adsorption is analogous to the minimum translational energy required to desorb -C 2 H 4 in the CID experiment. Collision-induced desorption of -C 2 H 4 is similar to the microcalorimetry experiment, at the C 2 H 4 coverage of 0.6 ML and above, in that the ethylene molecule is ejected intact into the gas phase without accessing alternative intermediate adsorbate states. In a thermally activated experiment such as TPD, alternative intermediate states are often accessed and one can not unambiguously determine the intermediates accessed prior to desorption. Hence, the determination of the activation energy for TPD is always model dependent. 38 For instance, the activation energy for thermal desorption is proportional to the TPD peak temperature for first order process. As an example of irreversible reactivity, we note that the TPD study of C 2 H 4 /Pt Ref. 14 revealed three desorption temperatures for C 2 H 4 of 220, 280, and 415 K. The desorption at 220 K was assigned to the -C 2 H 4 species. In the C 2 H 4 /Pt system, there is no doubt that multiple reaction pathways are accessed during the thermal desorption process, so a determination of the absolute binding energy for such a system would be difficult if not impossible. One question that must be addressed involves the difference in the activation energy measured by TPD and the binding energy measured by CID. The TPD experiment provides an activation energy for desorption of an adsorbate. If there are no precursor states accessed during thermal desorption of the adsorbate-surface system, the measured activation energy should be equal to the binding energy. If there are precursor states, implying local minima in the PES of adsorbatesurface system, it is likely that the activation energy corresponds to the PES barrier for the deepest local minima for a nonequilibrium process. The binding energy or the bond dissociation energy is proposed to be proportional to the threshold energy measured by the CID experiment. Unlike TPD, excitation to a local minima has no contribution to the collision-induced desorption threshold energy because the frequency of excitation roughly 1 Hz is far slower than the rate of deexcitation from the local minimum. It is probable, then, that the TPD and CID experiments probe different portions of the PES. Differences in the CID and TPD measurements have been observed previously for the di- -C 2 H 4 /Pt 111 system. Using TPD, Salmeron and Somorjai 26 measured a desorption feature at 280 K which was attributed to the di- -C 2 H 4 species. The activation energy for di- -C 2 H 4 desorption from Pt 111 was then calculated to be kcal/mol ev. The TPD result is supported by extended-huckle calculations 39 which suggested that the di- -C 2 H 4 adsorption energy on Pt 111 is 15 kcal/mol 0.65 ev. However, Anderson 40 determined the di- -C 2 H 4 Pt binding energy on Pt 111 using atom superposition and electron delocalization molecular orbital ASED-MO calculation method to be 2.04 ev. Using the CID Ref. 6 experiment the di- -C 2 H 4 Pt 111 binding energy has been measured at 100 K to be 2.1 ev. The CID result is then in good agreement with the ASED-MO calculation. 40 The discrepancy between the TPD and CID results was attributed to a local minimum of intermediate. For systems where no local minima are expected, the CID and TPD energetics agree. For instance, the TPD activation energy and CID binding energy for physisorbed CH 4 on Ni 111 are and ev, respectively. 3,4 A comparison of the activation energy for thermal desorption for -C 2 H 4 /O/Pt 111 measured using TPD with the binding energy measured using CID again suggests that these measurements probe different portions of the PES. To begin we note that the activation energy for thermal desorption has not been yet calculated for the C 2 H 4 /O/Pt 111 system although the data for such a calculation is available. This activation energy for -C 2 H 4 desorption from O/Pt 111 is calculated using the Redhead equation 41 for a linear temperature sweep. In this calculation we input the -C 2 H 4 TPD Ref. 7 peak temperature of 185 K, a heating rate of 15.5 K/sec, and an assumed frequency factor of The activation energy for thermal desorption is then calculated to be 0.42 ev. This value is consistent with the 8 kcal/mol 0.35 ev value obtained from extended-huckle calculations 39 for the -C 2 H 4 Pt 111 adsorption energy. In this paper, the C 2 H 4 /O/Pt 111 system was investigated using CID and the -C 2 H 4 Pt binding energy was determined to be 0.95 ev. The TPD activation energy for -C 2 H 4 desorption from O/Pt 111 is then approximately 50% of the binding energy determined by the CID technique, suggesting that TPD measurements probe only a portion of -C 2 H 4 PES. 38 We can compare the C 2 H 4 /Pt binding energy, as measured by microcalorimetry, 28 with the C 2 H 4 /O/Pt 111 binding energy, as measured by CID, by assuming that the TPD peak temperature is proportional to the binding energy. Using this assumption we can determine whether the binding energies measured by CID and microcalorimetry are consistent. Comparing the -C 2 H 4 TPD peak temperature of 220 K in C 2 H 4 /Pt Ref. 14 and 185 K in C 2 H 4 /O/Pt 111 Ref. 7 we expect that the -C 2 H 4 Pt binding should be weaker in the case of C 2 H 4 /O/Pt 111. The -C 2 H 4 TPD peak temperature of 185 KinC 2 H 4 /O/Pt 111 is approximately 20% lower than the 220 K peak temperature measured for C 2 H 4 /Pt Assuming that the TPD peak desorption temperature is proportional to the binding energy, the -C 2 H 4 Pt 111 binding energy should be proportionally weaker than the C 2 H 4 Pt binding energy. Microcalorimetry has been used to measure the -C 2 H 4 Pt heat of reaction binding energy equal to 1.24 ev. 28 We can then estimate the -C 2 H 4 Pt 111 binding energy which we would expect to measure if microcalorimetry was used on

10 9638 D. Velic and R. J. Levis: Selective collision-induced desorption TABLE I. Binding energies from TPD, CID, microcalorimetry, and calculations. C 2 H 4 /O/Pt 111 C 2 H 4 /Pt C 2 H 4 /Pt 111 Binding Form TPD temperature K 185 Ref Ref Ref. 7 Activation energy for 0.42 a 0.52 Ref. 26 desorption by TPD ev Binding energy 0.95 a 2.1 Ref. 6 by CID ev Heat of reaction by 1.0 a 1.24 Ref. 28 microcalorimetry ev Adsorption energy by 0.35 Ref Ref. 39 extended-huckel calcualtion ev Binding Energy by ASED-MO calculation ev 2.04 Ref. 40 a This work. the C 2 H 4 /O/Pt 111 system. Using the TPD peak desorption temperature ratio we calculate that a value of approximately 1.0 ev would be measured using microcalorimetry. This estimate is in good agreement with the CID determined -C 2 H 4 Pt 111 binding energy of 0.95 ev suggesting a consistency in the CID and microcalorimetric measurements. The energies from TPD, CID, microcalorimetry, and calculations are summarized in Table I. The thermal reaction of C 2 H 4 /O/Pt 111 system is also a simple model for the catalytic combustion of hydrocarbons. The previous reaction studies come mainly from TPD Ref. 26 supported by theoretical calculations Refs. 19 and 39. Thermal activation of this system gives rise to H 2,C 2 H 4, C 2 H 6, CO, CO 2, and H 2 O, but the exact mechanisms of reaction, presumably involving precursor states, are not well resolved. 7,9 Knowledge of the -C 2 H 4 Pt binding energy provides insight into the C 2 H 4 /O/Pt 111 system. This paper also suggests that the selectivity of CID can be a potential tool to probe such precursors by determining absolute binding energies. VI. CONCLUSION At the exposure of 1.3 L C 2 H 4 adsorbs on 0.25 ML atomic oxygen precovered Pt 111 at 100 K in two forms: -C 2 H 4 and di- -C 2 H 4. In this paper both forms of C 2 H 4 were characterized by XPS and C(1s) XP binding energies were assigned to be ev for -C 2 H 4 and ev for di- -C 2 H 4. The C 2 H 4 /O/Pt 111 system was chosen to investigate a selectivity of CID. To our knowledge we have demonstrated for first time the selective CID. The -C 2 H 4 adsorbate was selectively desorbed from O/Pt 111 by using a Xe translational energy which leaves the more strongly bound di- -C 2 H 4 intact on the surface. The minimum Xe translational energy of 2.4 ev required to eject -C 2 H 4 from O/Pt 111 is related to the binding energy of chemisorbed -C 2 H 4 on O/Pt 111. Using two classical binary collisions, the -C 2 H 4 Pt chemical binding energy of 0.95 ev was calculated. ACKNOWLEDGMENTS The support of the National Science Foundation through a Young Investigator Award R.J.L. is gratefully acknowledged. The authors also acknowledge valuable discussions with Gregory Szulczewski. 1 R. D. Levine and R. B. Bernstein, Molecular Reaction Dynamics and Chemical Reactivity Oxford University Press, New York, Y. Zeiri, J. J. Low, and W. A. Goddard III, J. Chem. Phys. 84, J. D. Beckerle, A. D. Johnson, and S. T. Ceyer, Phys. Rev. Lett. 62, ; J. Chem. Phys. 93, G. Szulczewski and R. J. Levis, J. Chem. Phys. 98, G. Szulczewski and R. J. Levis, J. Chem. Phys. 101, ; 103, G. Szulczewski and R. J. Levis unpublished. 7 H. Steininger, H. Ibach, and S. Lehwald, Surf. Sci. 117, A. Cassuto, M. Touffaire, M. Hugenschmidt, P. Dolle, and J. Jupille, Vacuum 41, ; A. Cassuto, M. Mane, M. Hugenschmidt, P. Dolle, and J. Jupille, Surf. Sci. 237, P. Berlowitz, C. Megiris, J. B. Butt, and H. H. Kung, Langmuir 1, A. Cassuto, M. Mane, J. Jupille, G. Tourillon, and Ph. Parent, J. Phys. Chem. 96, B. E. Bent, C. M. Mate, C.-T. Kao, A. J. Slavin, and G. A. Somorjai, J. Phys. Chem. 92, E. M. Stuve and R. J. Madix, J. Phys. Chem. 89, E. Yagasaki and R. I. Masel, Surf. Sci. 222, E. Yagasaki, A. L. Backman, and R. I. Masel, J. Phys. Chem. 94, A. L. Backman and R. I. Masel, J. Phys. Chem. 94, J.-F. Paul and P. Sautet, J. Phys. Chem. 98, M. B. Hugenschmidt, P. Dolle, J. Jupille, and A. Cassuto, J. Vac. Sci. Technol. A 7, A. Cassuto, J. Kiss, and J. M. White, Surf. Sci. 255, P. D. Ditlevsen, M. A. Van Hove, and G. A. Somorjai, Surf. Sci. 292, A. Cassuto, M. Mane, V. Kronneberg, and J. Jupille, Surf. Sci. 251/252, X.-L. Zhou, X.-Y. Zhu, and J. M. White, Surf. Sci. 193, R. G. Windham, M. E. Bartram, and B. E. Koel, J. Phys. Chem. 92, A. Cassuto, S. Schmidt, and M. Mane, Surf. Sci. 284, J. L. Gland, B. A. Sexton, and G. B. Fisher, Surf. Sci. 95, ; J. L. Gland, ibid. 93, H. Steininger, S. Lehwald, and H. Ibach, Surf. Sci. 123, M. Salmeron and G. A. Somorjai, J. Phys. Chem. 86,