J . Chem. SOC., Faraday Trans. I, 1984,80, 1567-1578 Hydrogenation of Nitric Oxide on (0 0 1) and (1 1 10) Surfaces of Ruthenium BY TETSUYA NISHIDA, CHIKASHI EGAWA, SHUICHI NAITO* AND KE~NZI TAMARU Department of Chemistry, Faculty of Science, University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113, Japan Received 15th September, 1983 Hydrogenation of nitric oxide on flat Ru(0 0 1) and stepped Ru(1 1 10) surfaces has been studied by Auger electron spectroscopy, X-ray photoelectron spectroscopy, ultraviolet photo- electron spectroscopy and thermal-desorption studies as well as by investigation of reaction kinetics. The products of the reaction were N, and H,O under the investigated conditions ( 10-s-10-4 Pa of NO pressure and between 300 and 1100 K). A high reaction probability (0.5) was obtained on an oxygen-free clean surface, while the reaction did not take place on an oxygen-covered surface, which depends not only on the reaction temperature but also on the ratio of the partial pressures of NO to H,.Although the rate of formation of N, on an oxygen-free clean surface is almost proportional to the partial pressure of NO above 520 K, the rate of desorption of N(ads) from the surface determines the overall rate of the reaction; N(ads) exhibited heterogeneous behaviour and its activation energy was shown to be 32 kcal mol-l. By comparison with the activity of N(ads) during ammonia decomposition, it was found that the rate of desorption of N(ads) in the reaction between NO and H, was accelerated by the presence of a small amount of oxygen (perhaps < 1 % of a monolayer).The activity of N(ads) above 520 K was the same for the flat and stepped surfaces, although the rate of N, formation on the Ru(1 1 10) stepped surface was 1.5 times as fast as that on the Ru(0 0 1) flat surface at lower temperatures (450 K). It is generally accepted that ruthenium, like iron, is a good catalyst for ammonia synthesis from nitrogen and hydrogen. At room temperature, however, the sticking probability of N, on an Ru surface is considered to be very small, while that on Fe is reported to be ca. lo-’ above room temperature.l Therefore, few investigations of chemisorbed nitrogen on Ru have been carried out under ultrahigh-vacuum conditions. Danielson et aZ.2 studied the adsorption and desorption of ammonia on an Ru(0 0 1) surface by thermal desorption studies and LEED.Ammonia, which was adsorbed at 100 K, was desorbed molecularly at 133 and 186 K, whereas the adsorption of ammonia at temperatures between 300 and 500 K took place via an activated process and gave rise to a (2 x 2) LEED pattern through its dissociation. Klein and Shih3 stud- ied the chemisorption of NO on Ru(l0 T 0) by using FEM and thermal-desorption methods. Nitric oxide showed three and two binding states when it was adsorbed at 120 and 295 K, respectively, while it was dissociatively adsorbed above 250 K. The changes in the state of NO adsorbed on an Ru(0 0 1) surface at low temperature upon subsequent heating have been studied by multi-technique approaches, using methods such as X.P.S., u.P.s., thermal desorption studies and work-function measurements.* Two molecular states of NO below 150 K were distinguished and the dissociation of NO took place at 370 K on an NO-saturated surface.The same behaviour was also observed in EELS s t ~ d i e s . ~ The behaviour of nitrogen adsorbed on an Ru surface, however, has never been 15671568 HYDROGENATION OF NO ON RU SURFACES investigated during the reaction comprising NO hydrogenation and ammonia decomposition.6 In this study it was found that atomic nitrogen accumulates on the Ru surface during NO hydrogenation, as revealed on Pd7 and Rh8 foils, where the formation of N, and NH, proceeds through adsorbed atomic nitrogen. Therefore, the reaction mechanism of NO hydrogenation can be discussed through the behaviour of adsorbed nitrogen and may be compared with that for the decomposition of ammonia.Recently, much attention has been paid to the step sites in a catalyst surface; these are considered to be the active sites for some catalytic reactions. For instance, Blakely and Somorjaig investigated the dehydrogenation and hydrogenolysis ofcyclohexane and cyclohexene on a stepped Pt surface and concluded that the steps play an important role as the active sites for C-H and H-H bond-breaking processes. Since the dissociation of NO takes place at temperatures as low as 200 K on an Ru(1 1 10) stepped surface,l0 the effect of steps in the Ru surface in NO hydrogenation has been investigated in this study by comparing the activity on flat Ru(O0 1) and stepped Ru(1 1 10) surfaces. EXPERIMENTAL The experiments were carried out using two different stainless-steel ultrahigh-vacuum systems, one equipped with 4-grid LEED-A.e.s.optics and mass filter as described pre- viously,ll and the other an ESCALAB5 (Vacuum Generators) apparatus used in the X.P.S. and U.P.S. measurements. The sample used in the experiment was an Ru single crystal (99.999% purity) obtained from Metal Research Ltd. The (0 0 1) and (1 1 10) planes of the crystal were prepared by spark erosion and standard polishing after the normal to the desired plane had been determined by an X-ray diffraction method. The sample was 8.0 mm in diameter and 0.9 mm thick for the (0 0 1) plane and 7.7 mm in diameter and 0.5 mm thick for the (1 1 10) plane. The sample was mounted on a sample holder made of spot-welded Ta wire (0.3 mm in diameter), through which the sample was resistively heated by a d.c.source. This facilitated the use of the thermal desorption technique. The sample temperature was measured using a Pt-Pt/Rh (13 %) thermocouple spotwelded at the centre edge of the sample. Gas pressures were measured by an ion gauge and a mass filter. At the beginning the sample surface was contaminated by carbon and sulphur which were detected by A.e.s. These impurities were burned off under 2 x Pa of oxygen at 900 K and then the sample was reduced by 1 x Pa of H, at 800 K. After the repetition of this procedure for 2 h, a clean surface was obtained and no impurities were detected by A.e.s. during the reaction. Edge effects were eliminated by the sulphidation of the surface by the adsorption of 5 langmuir of H,S, followed by Ar-ion bombardment (1.3 kV and 4 pA for 30 min) on the defined surface only.Following the cleaning procedure a hexagonal LEED pattern of (1 x 1) structure was observed in the case of the (0 0 1) plane. The (1 1 10) plane of the stepped surface was identified by the splitting of the LEED spots. The direction and width of this splitting gave information about the terraces and steps on the surface. Further, the occurrence of splitting in (0 0) and (1 0) spots for various beam voltages makes it possible to determine the height of the step. From an analysis of these LEED patterns a surface structure of five terrace widths and one step height was determined, as is depicted in fig. 1 ; this means that the step density is 20% on the surface.The surface density of the (1 1 10) plane was calculated to be 1.51 x 1015 atom ern-,, which is a little smaller than that of the (0 0 1) plane (1.58 x lOI5 atom ern-,). The rates of the reaction in the steady state were determined using the following equation for a flow reactor: AP. S . r . = - BAP, a RT,An,T. NISHIDA, C. EGAWA, S. NAITO AND K. TAMARU 1569 17.6' Fig. 1. Surface atomic structure of the stepped Ru(1 1 10) surface. where ri is the specific rate of formation of product i in the unit of turnover frequency, Api is the change in partial pressure of species i during the reaction, S, is its pumping speed, R is the gas constant, q is the gas temperature, A is the sample area and no is the surface density of Ru atoms per cm2.The amount of species adsorbed on the surface during the reaction was measured by the thermal desorption technique, calibrated by A.e.s. This amount, nads, was calculated by the following equation : where S,, Tp, R, A and no are defined above, a is the number of atoms of a specific type per desorbed molecule and the integral gives the intensity of the thermal desorption peak measured by a mass filter, where nads is expressed as a coverage. Ta was used as the support in this study because its nitride and oxide are very stable and had no effect on the measurements. A preliminary study showed that nitrogen adsorbed on Ta never desorbed to the gas phase and diffused into the bulk when heated to higher temperatures (ca. 2000 K). The 15N0 isotope was employed in the experiments in order to distinguish N, produced from CO, a residual gas, during the reaction.One of the gas inlet systems was improved so that gas could be leaked through a stainless-steel capillary (0.3 mm in diameter) on top of the sample. As a result, the gas pressure near the sample was ca. one order of magnitude greater than the background pressure. By using such an improved gas inlet system the reaction-gas pressure decreased (or increased) to about one-tenth (or ten-fold) at the instant of stopping (or introducing) the gas stream. The binding energy in X-ray and ultraviolet photoelectron spectra is referred to the Fermi level of the Ru sample.1570 rr a 0 4-l E 0 WDROGENATION OF NO ON RU SURFACE? i \ I &+ I t Fig. 2. Rate of formation of N, and nitrogen coverage plotted as a function of temperature during the reaction of NO and H, on Ru(0 0 1).pNO = 2.8 x low5 Pa, pHz = 1.2 x Pa. 0, rate of formation of N,; 0, nitrogen coverage. RESULTS AND DISCUSSION STEADY-STATE REACTION RATE AND ITS SURFACE CONDITIONS Fig. 2 shows a typical result of the reduction of nitric oxide with hydrogen on an Ru(0 0 1) surface under pNO = 2.8 x Pa. The reduction took place above 500 K, and N, and H,O were formed. Neither N,O nor NH, was produced under the conditions investigated, which is characteristic behaviour over Ru catalysts.12 The rate of the reaction increased with temperature rapidly up to 600 K and then remained almost constant, at which point the reaction probability (the ratio of N, formation to NO collision frequency) attained a high value, 0.5.During the reaction only nitrogen was adsorbed on the Ru surface, and its coverage decreased with increasing temperature. Other adsorbed species were not detected by A.e.s. and thermal-desorption methods. On the other hand, the reaction did not occur at temperatures below 500K, for which adsorbed oxygen was observed on the Ru surface by A.e.s. Once oxygen covered the Ru surface, the NO+H, reaction did not occur unless adsorbed oxygen was first removed from the surface; this required heating the sample to above 1000 K. Surface species formed during the reaction were also identified by X.P.S. and U.P.S. measurements as shown in fig. 3-5. As shown by the A.e.s. technique, atomic nitrogen was only observed at 490 K under pNO = 5 x loa6 Pa and pHz = 5 x Pa, which was identified by N 1s core-level emission at 397.3 eV [fig.3(a)] and a 5.6 eV peak in He I1 valence region [fig. 4(b)]. This assignment is supported by the fact that both peaks are also observed during NH, decomposition above 500 K.l0 By lowering the temperature from 500 to 350 K [fig. 4(c) and (41 the 5.6 eV peak shifted to the higher-binding-energy side of 6.5 eV and increased in intensity, and a small new peak appeared at 9.5 eV at 350 K. This spectral change indicates that at lower temperatures the surface was mainly covered by the dissociated oxygen and a small amount of NO was adsorbed in molecular form, because, as shown in fig. 6, molecularly adsorbed NO gives an intense peak at 9.5 eV (5a+ In) together with the peak at 13.9 eV (k), Pa and pH, = 1.2 xT.NISHIDA, C. EGAWA, S. NAITO AND K. TAMARU 1571 0 1s N 1s A at (c) 500K (b ) at 350K - (a at 490K 535 530 525 400 3 95 P C " " " ' . . L binding energylev Fig. 3. N 1s and 0 1s X-ray photoelectron spectra during the reaction of NO and H, on the Ru(1 1 10) surface.pNo = 5 x Pa,pHz = 5 x lop5 Pa. (a) At 490 K; (b) temperature lowered to 350 K; (c) at 500 K (deactivated surface). 15 10 5 0 15 10 5 O = E , binding energy/eV Fig. 4. He I1 ultraviolet photoelectron spectra and difference spectra during the reaction of NO and H, on the Ru(1 1 10) surface. pNO = 5 x Pa. (a) Clean surface; (6) at 490 K; (c) temperature lowered to 420 K; (d) temperature lowered to 350 K. Pa, pHe = 5 x and the dissociated oxygen formed by heating following the adsorption of NO displays the peak at 6.5 eV.This is also confirmed by 0 1s core-level emission at 530.3 eV in the X-ray photoelectron spectra [fig. 3(6)] under the same conditions, where the shoulder on the higher-binding-energy side (53 1.8 eV) is due to molecular NO.l0 Fig. 5 gives the ultraviolet photoelectron spectra for surface species on the oxygen-dissociated surface, where the reaction did not proceed at higher temperatures. No atomic nitrogen was detected at 505 K on this surface [fig. 3(c)], although molecularly adsorbed NO1572 HYDROGENATION OF NO ON RU SURFACES 15 10 5 O = E , binding energy/eV Fig. 5. He I1 ultraviolet photoelectron spectra for a mixture of NO+H, on deactivated Ru(1 1 1O).pN0 = 5 x Pa. (a) At 380 K; (b) heated to 420 K; (c) heated to 505 K.Pa,p,, = 5 x could be observed at 420 K [fig. 5(b)]. Accordingly, the removal of oxygen formed by NO dissociation from the surface is essential for the catalytic cycle of NO reduction with high activities. EFFECT OF PARTIAL PRESSURES ON THE STEADY-STATE REACTION RATE The temperature at which dissociated oxygen starts to cover the surface as demonstrated above depends greatly on the ratio of pressures of NO to H,, as reported for Rha and Ir(1 1 O).13 When a 1 : 1 mixture of NO and H, was introduced, the dissociated oxygen began to spread over the surface above 600 K and the rate of N, formation became negligible. On the other hand, oxygen was not adsorbed near 400 K under the conditions of pNO/pH, = 0.03, where the reaction proceeded readily even at lower temperatures. The partial pressure of hydrogen which is sufficient to keep the surface free of oxygen is therefore determined by the partial pressure of NO as well as the reaction temperature.Accordingly the reaction mechanism of NO hydrogenation on an Ru surface not covered by dissociated oxygen was investigated in more detail as follows. The dependence of the rate of N, formation upon NO pressure under conditions of an excess of H, is given in fig. 7, where the rate of N, formation is nearly pro- portional to NO pressure in the temperature range 520-750 K on both the Ru(O0 1) and Ru( 1 1 10) planes. However, even when the hydrogen pressure was far higher than that defined above, both the rate of N, formation and the product distribution were independent of the partial pressure of H,, i.e.no ammonia formation was detected; this is unlike the reaction on Rh,8 Pd,' Ir13 and Pt.14 The rate of N, formation was independent of H, pressure and a high reaction probability was always attained.T. NISHIDA, C. EGAWA, S. NAITO AND K. TAMARU 1573 , ‘ a a . - ‘ * . . . ” . . I 15 10 5 O=E, binding energy /eV Fig. 6. He I1 ultraviolet photoelectron spectra for the adsorption of NO on Ru(1 1 10) and subsequent heating. (a) 2.3 langmuir exposure at 210 K; (b) 4.3 langmuir exposure at 300 K; (c) heated to 370 K; (d) heated to 450 K. Fig. 7. NO pressure dependence of the rate of formation of N, during the reaction of NO and H, on Ru(O0 1) and Ru(1 1 10) surfaces. Open symbols Ru(1 1 10); filled symbols Ru(O0 1).0, 520; 0, 562; A, 598; 0, 633 K.1574 HYDROGENATION OF NO ON R U SURFACES Fig. 8. Desorption of N(ads) on Ru(0 0 1). 0 , 5 2 0 ; 0 , 5 6 2 ; V, 578; A, 598; 0 , 6 3 3 ; c), 663 K. Table 1. Activation energy for the desorption of N(ads) on Ru(0 0 1). peak heating initial activation Tm/K P/K s-l no(&) E,/kcal mold' temperature, rate, coverage, energy, 665 12 9.2 x 31.8 700 1 1 5.1 31.5 715 10 4.2 32.1 735 9.5 2.8 31.8 DESORPTION OF N(ads) ON Ru SURFACES As the nitrogen-containing product is only molecular nitrogen, the process of desorption of adsorbed nitrogen that had accumulated during NO hydrogenation was investigated according to the following procedure. A given amount of N(ads) was prepared on the Ru surface at various temperatures in the range 520-633 K at steady state in the NO+H, reaction, i.e. pNO = 5 x lo-' Pa andpH2 = 4 x lod5 Pa.At t = 0 both the NO and H, gases were evacuated and after a given period for desorption the amount of N(ads) remaining on the surface was measured by thermal desorption. The change in amount of N(ads) on the Ru surface is plotted as a function of time in fig. 8, and it is seen that nitrogen coverage decreases linearly with the logarithm of time for nitrogen coverages above 0.0 1. Accordingly, it is seen that the process of desorption of N(ads) obeys neither first-order kinetics, i.e. ud = ke,, nor second-order desorption kinetics, i.e. vd = kek, but instead may be expressed by the Zeldovich- Roginsky equation, i.e. ud = k exp (he,); this shows the heterogeneous behaviour of N(ads) on the Ru surface.From Arrhenius plots of the rate constant, k, of N(ads) desorption the activationT. NISHIDA, C. EGAWA, S. NAITO AND& TAMARU 1575 energy for N(ads) desorption on an Ru surface at near-zero coverage can be estimated as 32.1 & 1 kcal mol-l, with the following parameters h = 1.2-1.5 x 10, k = 7.3 x lo5 exp (- 32.1 kcal mol-l/RT) (s-l) where vd = k exp (he,) and h is a heterogeneous parameter. Since the heterogeneity of N(ads) on the surface was revealed by the isothermic desorption method, the activation energy for the desorption of N(ads) was also estimated from the temperature of the thermal-desorption peak as follows. The rate of desorption of N(ads) can be described by the following equation: v(t) = -- dn(t) = v exp [hn(t)] exp (- E,/RT) dt where n is the concentration of adsorbed species. When the sample temperature was raised linearly at a rate of B K s-l the following equation is valid at the thermal- desorption peak temperature, T, : dv(t) dt - = 0.If the nitrogen coverage at T, is assumed to be in,, from the thermal-desorption curve, where no is the initial coverage, the activation energy can be derived from the following equation : The activation energy obtained from this equation for several different initial coverages and B is summarized in table 1, and it is found to be in good agreement with that obtained from isothermic desorption. Note that there exists a general trend in activation-energy change for N, desorption on Ru, Rha and Pd7 surfaces, i.e. 32, 25 and 16 kcal mol-l, respectively.The homogeneous-desorption treatment was also applied to this system, employing the following equation : where rn is the kinetic order of the desorption. After various amounts of N(ads) were accumulated on the surface, the sample temperature was lowered to 435 K and then thermal-desorption spectra were measured at a low heating rate (ca. 5 K s-l). Plots of reciprocal temperature against log (desorption rate) at a certain coverage or of log(nitrogen coverage) against log(N, desorption rate) did not have a linear relationship. It is therefore confirmed that the homogeneous treatment cannot be employed for the desorption of N(ads) in this case. By using these parameters it is possible to calculate the rate of desorption of N(ads) at any coverage and temperature, this is plotted as the filled points in fig.9. The open points in fig. 9 give the rate of N, formation and nitrogen coverage at a steady state of NO hydrogenation, indicating that the rate of the NO+H, reaction in the steady state is also proportional to the exponential of nitrogen coverage, i.e. rNg cc exp(&), and equal to the rate of desorption of N(ads). Accordingly it is concluded that the process of desorption of N(ads) from the surface is the rate-limiting step in NO hydrogenation on the Ru surface.1576 HYDROGENATION OF NO ON RU SURFACES 0 0.5 1 nitrogen coverage (X 10-1) Fig. 9. Rate of N, formation plotted against nitrogen coverage during the reaction of NO and H, on Ru(0 0 1). 0, 520; 0, 562; A, 598; 0, 633 K. St 640 750 K Fig.10. Thermal desorption spectra of N, on Ru(0 0 1): (a) during NH, decomposition (pNHa = 1 x Pa); (b) during NH, decomposition after 0.32 langmuir exposure of 0,; (c) dunng NH, decomposition after 0.97 langmuir exposure of 0,; ( d ) during the reaction of NO and H,. Thermal-desorption spectra from 520 K.T. NISHIDA, C. EGAWA, S. NAITO AND K. TAMARU 1577 COMPARISON OF THE BEHAVIOUR OF N(ads) ON THE RU SURFACE BETWEEN NO HYDROGENATION AND NH, DECOMPOSITION The behaviour of N(ads) formed during NO hydrogenation was compared with that formed during NH, decomposition. For the decomposition of ammonia6 relationships similar to those of fig. 9 could be obtained between the rate of N, formation and nitrogen coverage in the steady state under various ammonia pressures at each temperature. However, the rate of formation of N, in the hydrogenation of NO is about ten times faster than that of ammonia decomposition at the same nitrogen coverage between 520 and 630K, which corresponds well to the change of peak temperatures in the thermal desorption of N,, i.e.the peak temperature of the desorption of N(ads) during NO hydrogenation appeared at ca. 650 K, while that formed during ammonia decomposition was observed to be ca. 750 K. It is therefore considered that the desorption of N(ads) in NO hydrogenation is affected by a small amount of oxygen on the surface. To make this point clearer, the effect of a small amount of adsorbed oxygen on ammonia decomposition was examined. When a small exposure of 0, (below 1 langmuir) was introduced, temperature peaks corresponding to N, desorption appeared at both 640 and 750 K, in accordance with the increase in the rate of formation of N,, as shown in fig.10. Consequently it may be inferred that the desorption of N(ads) during NO hydrogenation is accelerated by the presence of a small amount of adsorbed oxygen which cannot be detected by electron spectroscopic methods and is therefore estimated to be below 1 % of a monolayer. DIFFERENCES BETWEEN THE FLAT Ru(0 0 1) AND STEPPED Ru( 1 1 10) SURFACES IN NO HYDROGENATION As reported by Umbach et aZ.,4 the NO molecule is dissociatively adsorbed on the Ru(0 0 1) surface at temperatures above 370 K; this is observed on both terrace and stepped surface.l0 It is therefore likely that the dissociation of NO readily takes place on both surfaces under reaction conditions.In addition, taking the high reaction probability (0.5) into consideration, a difference in the rate of N, formation between the two surfaces would not be expected. In fact, the rate of N, formation above 520 K in the NO + H, reaction was the same for both the flat and stepped surfaces. However, when the reaction temperture was lowered to 450 K under conditions of an excess of H,, the rate of N, formation on the stepped Ru( 1 1 10) surface was observed to be 1.5 times as fast as that on the Ru(0 0 1) surface. At similar temperatures (below 500 K) the rate of ammonia decomposition on the stepped Ru(1 1 10) surface is initially one order of magnitude faster than that on the flat Ru(0 0 1) surface, where a second peak, N*(ads), was observed at 570 K in thermal-desorption spectra.This was considered to be an activated nitrogen atom formed selectively on the stepped sites and has been confirmed to have an important role in N, formation.6 We therefore consider that the difference in the rate of N, formation at 450 K during the NO+H, reaction is related to the formation of N*(ads) on the stepped surface. CONCLUSIONS NO reduction with H, on flat Ru(0 0 1) and stepped Ru(1 1 10) surfaces has been investigated in this study. The products of the reaction were N, and H,O; neither NH, nor N,O was formed under the conditions investigated. The rate of N, formation exhibited a high reaction probability (0.5) almost irrespective of the reaction temperature on an oxygen-free clean surface, while the reaction did not take place when oxygen atoms covered the surface.The surface conditions during the reaction 52 FAR 11578 HYDROGENATION OF NO ON RU SURFACES described above are dependent on the reaction temperature as well as the ratio of partial pressures of NO and H,. The rate of N, formation on an oxygen-free surface is proportional to NO pressure and independent of H, pressure. The overall rate of NO reduction is determined by that of N(ads) desorption on the surface, which is heterogeneous in nature, and the activation energy is found to be 32 kcal mol-I. By a comparison with the decomposition of ammonia, where surface nitrogen also exhibits heterogeneous behaviour, it is found that the desorption of N(ads) is accelerated by the presence of a small amount of oxygen on the surface (perhaps below 1 % of a monolayer).The rate of N, formation in NO hydrogenation is an order of magnitude faster than that during ammonia decomposition at the same nitrogen coverage. Although the rate of the reaction of NO and H, above 520 K was the same for both the flat and stepped surfaces, the rate of N, formation on the Ru(1 1 10) surface was observed to be 1.5 times as fast as that on the Ru(0 0 1) surface at temperatures as low as 450 K. This work was supported by a Grant-in-Aid for Scientific Research from the Japanese Ministry of Education, Science and Culture. F. Bozso, G. Ertl, M. Grunze and M. Weiss, J. Catal., 1977, 49, 18. L. R. Danielson, M. J. Dresser, E. E. Donaldson and J. T. Dickinson, Surf. Sci., 1978, 71, 599. R. Klein and A. Shih, Surf. Sci., 1977, 69, 403. E. Umbach, S. Kulkami, P. Feulner and D. Menzel, Surf. Sci., 1979,88,65; P. Feulner, S. Kulkami, E. Umbach and D. Menzel, Surf. Sci., 1980,99,489. G. E. Thomas and W. H. Weinberg, Phys. Rev. Lett., 1978, 41, 1181; P. A. Thiel, W. H. Weinberg and J. T. Yates Jr, Chem. Phys. Lett., 1979, 67, 403. C. Egawa, T. Nishida, S. Naito and K. Tamaru, to be published. ' A. Obuchi, S. Naito, T. Onishi and K. Tamaru, Surf. Sci., 1982, 122, 235. * A. Obuchi, S. Naito, T. Onishi and K. Tamaru, Sur- Sci., 1983, 130, 29. D. W. Blakely and G. A. Somorjai, J. Catal., 1976, 42, 181. lo C. Egawa, S. Naito and K. Tamaru, Surf. Sci., submitted for publication. l1 C. Egawa, S. Naito and K. Tamaru, Surf. Sci., 1983, 125, 605. l2 K. C. Taylor and R. L. Klimisch, J. Catal., 1973, 30, 478. l3 D. E. Ibbotson, T. S. Wittrig and W. H. Weinberg, Surf. Sci., 1981, 111, 149. l4 G. Pirug and H. P. Bonzel, J. Catal., 1977, 50, 64. (PAPER 3/ 1625)