Angular and velocity distributions of desorbing molecules in steady-state NO+CO reaction on Pd(110)

Surface Science 445 (2000) 472–479 www.elsevier.nl/locate/susc

Angular and velocity distributions of desorbing molecules in steady-state NO+CO reaction on Pd(110) Ivan Kobal a, Kazushi Kimura b, Yuichi Ohno c, *, Tatsuo Matsushima c a J. Stefan Institute, 1000 Ljubljana, Slovenia b Graduate School of Environmental Earth Science, Hokkaido University, Sapporo 060-0811, Japan c Catalysis Research Center, Hokkaido University, Sapporo 060-0811, Japan Received 2 June 1999; accepted for publication 14 October 1999

Abstract A Pd(110) surface was exposed to a constant flow of a (NO+CO) mixture. Rates of N , N O, and CO formation 2 2 2 were determined over the surface temperature range from 350 K to 750 K in angular-resolved form in a plane in the [001] direction. N desorption showed a two-directional desorption approximated by cos28(h±41°), while N O 2 2 followed a simple cos h distribution, showing complete accommodation to the surface temperature. CO desorption 2 was collimated in the surface normal direction, described by cos15 h accompanied by a cos h component. Desorbing N molecules were highly energetic as confirmed by velocity distribution measurements in which two components 2 with kinetic energies of 0.46 eV and 0.23 eV were found. © 2000 Elsevier Science B.V. All rights reserved. Keywords: Catalysis; Nitrogen oxides; Palladium; Single crystal surfaces; Surface chemical reactions

1. Introduction The catalytic removal of nitrogen oxides from the exhaust gases of mobile and stationary sources presents a serious and complex environmental problem [1]. In commercial automotive catalytic converters, the reduction of NO by CO over supported transition and noble metals is one of the key processes in the three-way catalyst designed to reduce NO, and oxidize CO and unburnt hydrocarbons simultaneously. The decomposition of NO over noble metals yields N O concomitantly with 2 N . Knowledge of the relation of both N and 2 2 N O formation is a requisite for improving such 2 environmental catalysts. * Corresponding author. Fax: +81-11-7063695. E-mail address: [email protected] ( Y. Ohno)

The interest in using palladium has recently increased due to its distinguished resistance to sintering at elevated temperatures and high activity for hydrocarbon oxidation [1,2]. It is thus not surprising that a high number of works have been devoted to the NO+CO reaction over supported (mainly alumina and silica) mono- [3–15] and bimetallic [12,16–22] palladium catalysts, as well as over polycrystalline palladium [3,23–25] and well-defined Pd(111) [8,14,26–28], Pd(100) [8,14,22,28–34], Pd(110) [27,29,35], Pd(211) [35], Pd(320) [36 ] and Pd(331) [37]. Generally, the single-crystal palladium is by far more reactive than the supported palladium, and among supported samples those with larger particle size are more reactive [8]. For well-defined surfaces, reactivity increases in the sequence Pd(110) Pd(100)Pd(111), but the same direction is also

0039-6028/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved. PII: S0 0 39 - 6 0 28 ( 99 ) 0 11 1 1 -5

I. Kobal et al. / Surface Science 445 (2000) 472–479

followed by the undesirable selectivity for N O 2 production [8,28,29]. This paper proposes the first use of inclined desorption dynamics for successful analysis of N 2 desorption in the course of catalytic NO decomposition. We first report the high velocity of desorbing N in the steady-state decomposition of NO 2 on Pd(110). The N flux peaked at ±41° off the 2 surface normal in a plane along the [001] direction. This confirms that N O(a) is the intermediate 2 immediately before N emission in catalytic NO 2 decomposition [38]. Ikai et al. found such inclined desorption of N in the thermal programmed decomposition 2 ( TPD) of NO on Pd(110) [39,40]. The desorption was concentrated in a plane in the [001] direction and collimated far from the surface normal direction. They also argued that this inclined desorption process did not play a role in the catalytic reaction cycle because N desorption in NO steady-state 2 decomposition (SSD) in the presence of hydrogen was collimated along the surface normal direction. Ohno et al. renewed their data with a higher angle resolution and added velocity data showing a hyperthermal translational energy [38]. Furthermore, they found N O desorption below 2 150 K concomitantly with its dissociation into N (g) and O(a). The resultant product N showed 2 2 very similar angular and velocity distributions to those in TPD of NO. In fact, N O has been 2 proposed as intermediate to N desorption without 2 ample evidence [1,41].

2. Experimental Our UHV system consists of three chambers [42]. (i) The reaction chamber is equipped with low energy electron diffraction (LEED)–Auger electron spectroscopy (AES ) optics, an Ar+ gun, a sample handling system, and a quadrupole mass spectrometer. (ii) The middle chamber has a slit at both ends to the other chambers and a pseudorandom chopper blade. (iii) The analyzer chamber has another quadrupole mass spectrometer for angle-resolved analysis. A Pd(110) single crystal (MaTeck, Germany) in a w10 mm×1 mm disk shape was fixed on an

473

Fig. 1. Structure of Pd(110) and orientation of N O intermedi2 ate: (a) top view and (b) side view. The size of N O is referred 2 to a gaseous molecule. The dotted circle indicates the N mole2 cule estimated from the van der Waal’s radius.

L-shaped manipulator. It could be rotated to achieve various angles between the surface normal direction and the analyzer axis (the line-of-sight position of the mass spectrometer). This desorption angle (h) could be adjusted in a plane along the [001] direction with an accuracy of ±1°(see Fig. 1). The acceptance angle was less than 1.1° since the ionizer of the mass spectrometer had an aperture with a w5 mm diameter and the second slit was located 299 mm from the first slit with 1 mm width. The ambiguity in the desorption angle came mostly from setting procedures of the crystal plane. The flight path from the chopper blade to the ionization area measured 304 mm. The crystal was cleaned following standard procedures [42]. To run NO decomposition under steady-state conditions, the crystal was flashed to 1000 K and then kept at a desired temperature (T ) and in a S constant gas flow at 26 mPa of NO and 13 mPa of

474

I. Kobal et al. / Surface Science 445 (2000) 472–479

CO. The formation rate of N , N O or CO was 2 2 2 monitored in angular-resolved form at selected T and h. The signal at 350 K was subtracted from S the observed signals as the background level. This signal was due to ambient gas in the reaction chamber since it agreed with the signal when the crystal was out of the line of slit position. Appropriate time resolutions for time-of-flight ( TOF ) measurements were 15 ms (130.72 Hz) for N and 30 ms (65.36 Hz) for N O. For selected T 2 2 S and h, about 15 min data accumulation was enough to achieve a tolerable TOF spectrum.

3. Results 3.1. Angular distribution Desorbing products of N , N O and CO 2 2 2 showed quite different angular distributions. Fig. 2 shows their formation rates in steady-state NO+CO reaction in the T ranging from 350 K S to 750 K. For all products, the rate was negligible below 480 K, thereupon it abruptly reached its maximum at about 520 K and slowly decreased toward zero at 750 K. For N O ( Fig. 2b), only 2 signals at h=0° are shown, because these were insensitive to h, indicating a broad angular distribution. On the other hand, the N ( Fig. 2a) and 2 CO (Fig. 2c) signals were sensitive to h. The 2 former peaked at an angle far from the normal direction, whereas the latter quickly decreased with increasing h. The angular distributions are summarized in Fig. 3. N desorption showed a sharp two-direc2 tional form as approximated by cos28(h±41°). The power was estimated to be 28±5, and the collimation angle (h ) 41±1° by considering the experiC mental errors. On the other hand, N O desorption 2 followed a simple cosine form. CO desorption is 2 composed of a sharp component collimated along the surface normal direction and a broad cosine component being thus entirely represented by cos15 h+0.30 cos h. The power was estimated to be 15±3. These distributions were insensitive to the surface temperature, as in the case of N . 2

Fig. 2. Mass signals of product molecules desorbing in the steady-state NO+CO reaction on Pd(110) at various T for S selected h. P /P =2, with P =26 mPa. Numbers attached NO CO NO to the curves signify h values. (a) N ; (b) N O; (c) CO . 2 2 2

3.2. Velocity distribution The observed velocity distribution of N desorb2 ing at h=+41° is presented by triangles in Fig. 4a. A Maxwellian distribution at the surface temperature (T =540 K ) would hold the shape and posiS tion as depicted by the left dotted line. This contribution was subtracted from the observed data because it originates from the ambient N 2 molecules in the reaction chamber reflecting from the sample surface. The resultant points are indicated by open circles. The majority of desorbing product N molecules are highly hyperthermal. 2 The broadness and asymmetry of the distribution

I. Kobal et al. / Surface Science 445 (2000) 472–479

Fig. 3. Angular distributions of product molecules desorbing in the steady-state NO+CO reaction on Pd(110) shown in Fig. 1. (a) N : experimental points at temperatures of 500 (6), 510 2 (%), 520 ()), 540 (#), 550 (( ), 570 (+), 600 (,), 620 K ($). All the data are normalized to the signal at h=41°. Full lobular lines represent cos28(h±41°) distribution. (b) N O and CO at 2 2 540 K. Data are normalized at h=0°. Solid lines describe cos h and cos15 h+0.30 cos h distributions, respectively, whilst both broken lines represent the two components of the latter. The radial bars indicate experimental errors.

475

Fig. 4. Velocity distributions of product molecules desorbing in the steady-state NO+CO reaction on Pd(110) at T =540 K. S (a) N at h=+41°: triangles denote raw experimental data, 2 while circles were obtained after subtraction of the Maxwellian component at 540 K (dotted line). The full line is a sum of broken lines representing deconvoluted F (1300 K ) and F 2 1 (2660 K ) components. (b) N O at h=0°: experimental points 2 are well described by a Maxwellian distribution at 540 K (solid line).

as obtained from their fluxes. The velocity distribution of N O shown in Fig. 4b obeyed a Maxwellian 2 form at the surface temperature.

4. Discussion encouraged us to deconvolute it into two components, F and F , by applying modified Maxwellian 1 2 forms [43]:

4.1. Surface structure

f (v)dv3v4 exp[−(v−v )2/a2]dv, 0 where f(v) is the velocity distribution function and v and v are the velocity and stream velocity, 0 respectively, while a is defined by the width parameter, T =ma2/2k, m is the mass of a molecule, W and k is the Boltzmann constant. The resultant distributions of these two components are drawn in Fig. 4a by broken lines. The mean energy is 0.46 eV (T =2660 K, with T = E/2k) for

E

E F and 0.23 eV (T =1300 K ) for F . The angular 1

E 2 distribution of N was mostly determined by the 2 F component since the F shared only below 30% 1 2

Possible surface species in the course of catalyzed NO decomposition above 500 K are adatoms of oxygen and nitrogen, and admolecules of CO, NO and N O [1,41]. Products of N and CO are 2 2 2 desorbed immediately after formation as indicated by sharp angular distributions, whereas the other product N O stays on the surface long enough to 2 be accommodated. This is because the latter showed a cosine distribution and a Maxwellian velocity form at the surface temperature. The amounts of CO(a) and NO(a) are not high above 600 K because of their fast desorption [41]. The amount of N O(a) is negligibly small because of a 2

476

I. Kobal et al. / Surface Science 445 (2000) 472–479

small adsorption energy [38]. However, this species plays a key role in N desorption as described 2 below. N(a) may also be present in a small amount because of its high reactivity toward NO(a) [44]. Surface nitrogen atoms implanted with an ion gun seem to be less reactive than those produced from NO dissociation [35]. Thus, the surface structure is mostly affected by oxygen adatoms which are effectively removed as CO in the presence of CO. 2 Under our experimental conditions, CO 2 desorption collimated sharply along the surface normal direction as cos15 h (Fig. 3b). This is identical to that of CO produced in the active region 2 of the steady-state CO+O reaction and also in 2 TDS on the same surface [42,45]. This distribution clearly identifies a non-reconstructed Pd(110) surface. This is reasonable because the reconstruction into a missing-row structure occurs at high oxygen coverages [46 ]. On the missing-row surface, the CO desorption would be split into two directions 2 in a plane in the [001] direction [47]. Thus we may discuss our observation on a non-reconstructed surface drawn in Fig. 1. 4.2. Reaction pathway For NO decomposition, the following mechanism was proposed [38]. NO(a) molecules may desorb or dissociate into N(a) and O(a). N(a) atoms recombine with each other to form N or 2 interact with NO(a) to form N O(a). N O(a) 2 2 molecules may either desorb or decompose into O(a) and N (g). In the present experiments, this 2 N was repulsively desorbed into inclined direc2 tions. On the other hand, the N produced by 2 recombination of N atoms is repulsively desorbed along the surface normal direction [35]. Ikai et al. reported a broad cosine N desorption with a 2 maximum at about 600 K in the presence of hydrogen [39]. 4.3. Inclined desorption Intermediate N O(a) has a high reactivity 2 toward metal surfaces releasing oxygen to the surface, as for instance: Rh(110) and (111) [48], Ag/Rh(100) [49], Ni(111) [50] and Ni(755) [51], Pt(111) and polycrystalline [52,53], Ru(1010)

[54], Re polycrystalline [55], Cs/Pt(111) [56 ], W(100) [57], Al(100) [58] and Cu(110) [59]. It is in a highly tilted form via its terminal N atom on Pt(111) [60], Ni(111) [50] and Ru(001) [61], while it lies horizontally on W(110) and Ru(001) at low coverages [62]. No ample evidence was found for a vertical form irrespective of some proposals [62]. Even when the former is major, it must move its O atom toward the surface, being eventually almost lying on the surface prior to decomposition. To explain the high excess energy and inclined desorption, Ohno et al. proposed the model [38] in which nascent N can partly receive 2 the heat of reaction for the process of N O(a)N (g)+O(a) and also Pauli’s repulsion 2 2 from the surface. This is because the process is highly exothermic by about −190 kJ mol−1 [38], and the formation of N is in close proximity to 2 the surface. The repulsive force originated in the former process may be operative along the ruptured bond, and that in the latter closely toward the surface normal direction, since the product N , without being bonded, would behave toward 2 the surface as a large size molecule, as shown by large dotted circles in Fig. 1. This mechanism is reminiscent of the hot-atom collision mechanism proposed in photo-induced oxygen desorption [63]. Nascent oxygen atoms in N O(a) dissociation 2 must be highly energetic [64] because the PdMO bond formation is highly exothermic [38]. In steady-state NO+CO reaction, we may expect CO with energies higher than those in CO+O 2 2 reaction, although the internal energy was reported not to be excited more than in CO oxidation on Pt [65]. The above model can be further examined in two extreme cases, namely the short (a-type) and long (b-type) lifetime of activated N O(a). The 2 geometry of N O on the surface is drawn in Fig. 1, 2 where the bond length of gaseous N O and non2 distorted Pd(110) is used. The size of N O mole2 cule in a physical adsorption state is shown by the ˚ . Oxygen adadotted circle with a radius of 2.0 A toms are stabilized on suitable adsorption sites, probably on three-fold hollow sites in the troughs [66,67]. At the moment when this dissociation proceeds, the resultant N molecule receives the 2 repulsive force from the surface mostly along the

I. Kobal et al. / Surface Science 445 (2000) 472–479

normal direction. As described above, this is because the position of the NMN part in the N O chemisorption state is too close to the surface 2 compared with the equilibrium position for the physical adsorption or the weakly-bonded chemisorption state [68,69]. Thus, Pauli’s repulsion must push the molecule in the normal direction. Eventually the molecule will leave into the angle inclined from the normal direction. Both the energy partition of energetic oxygen (before being stabilized on suitable sites) and Pauli’s repulsion are important in determining the desorption angle. The collimation angle is sensitive to the position of N molecules immediately after the dissociation. 2 In the desorption channel with b-type intermediate, N would collimate along the NMNMO bond 2 axis irrespective of the translational and internal energies. The collimation angle can be predicted when its molecular orientation with the b-type intermediate is analyzed (Fig. 1b). However, no reports have been found for N O(a) bound by its 2 oxygen to palladium surfaces. 4.4. Collimation angle and energy partition The two desorption components with different translational energies might be assigned to vibrationally excited molecules at different levels. In fact, the difference in the kinetic energy of around 0.23 eV is close to the first vibrational excitation level of N , 0.28 eV [70]. This excitation model 2 might be examined by preferable ionization of vibrationally excited molecules at lower electron energy by using electron-beam-induced fluorescence technique [71]. It is difficult to predict the inclined collimation angle in the desorption channel with the a-type intermediate because a large amount of energy may be dissipated to the surface. The collimation angle (h ) should be described by a simple relation C as tan h =v /v with v =v sin h and v = C x z x rms C z v cos h where v is the root-mean-square rms C rms velocity derived from the mean translational energy: E=mv2 /2. In the present case rms v =1.72 km s−1, v =1.13 km s−1, and v = rms x z 1.30 km s−1 since E=0.46 eV. Thus, the energy transferred to N from the heat of N O decomposi2 2 tion is e =mv2 /2=0.20 eV, and the energy transExo x

477

ferred to N in Pauli repulsion is e = 2 Pauli mv2 /2=0.26 eV (Fig. 1b). These values are not z unreasonable, because the excess energy estimated from angular distribution of desorbing N is only 2 0.18 eV on W(110) [72], although the translational energy required for N dissociation is around 2 0.83 eV [73]. The magnitude of e is only about Exo 10% of the released energy 2.0 eV [38]. This fraction does not seem to be too small in this kind of energy partition since, for example, in the reactive CO desorption on palladium surfaces, about 50% 2 of the released energy was transferred to the surface, and only about 25% of the energy into the translational mode [74].

4.5. Other surfaces Ikai et al. reported the collimation of N thermal 2 desorption at around 15° off the normal direction downward steps on Rh(335)=[(S)4(111)×(001)] and at 25–30° on Pd(112)=[(S)3(111)×(001)] after adsorption of NO [35,75]. The normal direction of (001) facets is at 40.3° on Rh(335) and 35.2° on Pd(112). The collimation angle on Rh(335) shifted by about 25° toward the bulk surface normal direction, and that on Pd(112) slightly less shifted from the facet normal direction. This suggests that N O(a) is decomposed on (001) 2 facets. In fact, four-fold hollow sites on these facets are able to accept oxygen with a high binding energy. The above desorption channel with b-type intermediate is suitable to explain the results on Pd(112). The N moiety would be distant enough 2 from (111) facets to avoid Pauli’s repulsion and would receive only a momentum along its NMNMO tilted axis as proposed on Si(100) [76 ]. On the other hand, the results on Rh(335) should invoke a significant repulsive force due to Pauli’s repulsion from (111) facets. This is not unreasonable because the normal direction of (001) facets on this surface is close to the (111) plane. Furthermore, the repulsive force along the bond axis should be higher because the binding energy of RhMO is higher than that of PdMO [70,77]. This consideration predicts desorbing N on 2 Rh(335) holding a kinetic energy higher than that on Pd(112).

478

I. Kobal et al. / Surface Science 445 (2000) 472–479

5. Summary The angular and velocity distributions of desorbing products N , N O and CO were studied 2 2 2 in steady-state NO+CO reaction on Pd(110). From the characteristic angular and velocity distributions of desorbing N , the decomposition of 2 intermediate N O(a) was concluded to be the 2 process yielding highly energetic N emission. 2 Acknowledgements The authors thank Ms. Atsuko Hiratsuka for her drawings. Ivan Kobal acknowledges the support he received from the Ministry of Education, Science and Sports of Japan through the foreignresearcher (COE ) invitation program in 1998. This work was partly supported by a 1996 COE special equipment program of the above Ministry.

References [1] V.I. Paˆrvulescu, P. Grange, B. Delmon, Catal. Today 46 (1998) 233. [2] S.M. Vesecky, J. Paul, D.W. Goodman, J. Phys. Chem. 100 (1996) 15 242. [3] S. Moriki, Y. Inoue, E. Myazaki, I. Yausmori, J. Chem. Soc., Faraday Trans. 78 (1982) 171. [4] F. Schu¨th, E. Wicke, Ber. Bunsenges. Phys. Chem. 93 (1989) 491. [5] M. Valden, J. Aaltonen, E. Kuusisto, M. Pessa, C.J. Barnes, Surf. Sci. 307–309 (1994) 193. [6 ] M. Valden, R.L. Keiski, N. Xiang, J. Pere, J. Aaltonen, M. Pessa, T. Maunula, A. Savima¨ki, A. Lahti, M. Ha¨rko¨nen, J. Catal. 161 (1996) 614. [7] H. Cordatos, T. Bunluesin, R.J. Gorte, Surf. Sci. 323 (1995) 219. [8] D.R. Rainer, S.M. Vesecky, M. Koranne, W.S. Oh, D.W. Goodman, J. Catal. 167 (1997) 234. [9] D. Rainer, M. Koranne, S.M. Vesecky, D.W. Goodman, J. Phys. Chem. B 101 (1997) 10 769. [10] K. Almusaiteer, S.S.C. Chuang, J. Catal. 180 (1998) 161. [11] A.M. Pisanu, C.E. Gigola, Appl. Catal. B: Environ. 20 (1999) 179. [12] C.M. Grill, R.D. Gonzales, J. Phys. Chem. 84 (1980) 878. [13] X. Xu, D.W. Goodman, Catal. Lett. 24 (1994) 31. [14] X. Xu, P. Chen, D.W. Goodman, J. Phys. Chem. 98 (1994) 9242. [15] A.M. Venezia, L.F. Liotta, G. Deganello, P. Terreros, M.A. Pena, J.L. Fierro, Langmuir 15 (1999) 1176.

[16 ] K. Teruuchi, H. Habazaki, A. Kawashima, K. Asami, K. Hashimoto, Appl. Catal. 76 (1991) 79. [17] I. Halasz, A. Brenner, M. Shelef, K.Y.S. Ng, Appl. Catal. A: Gen. 82 (1992) 51. [18] I. Halasz, A. Brenner, M. Shelef, Catal. Lett. 22 (1993) 147. [19] I. Halasz, A. Brenner, M. Shelef, Catal. Lett. 16 (1992) 311. [20] P. Araya, C. Ferreda, J. Corte´s, Catal. Lett. 35 (1995) 175. [21] T.E. Hoost, G.W. Graham, M. Shelef, O. Alexeev, B.C. Gates, Catal. Lett. 38 (1996) 57. [22] A. Sandell, A.J. Jaworowski, A. Beutler, M. Wiklund, Surf. Sci. 421 (1999) 116. [23] A. Obuchi, S. Naito, T. Ohnishi, K. Tamaru, Surf. Sci. 122 (1982) 235. [24] L.M. Carballo, T. Hahn, H.-G. Lintz, Appl. Surf. Sci. 40 (1989) 53. [25] G. Xi, J. Bao, S. Shao, S. Li, J. Vac. Sci. Technol. A 10 (1992) 2351. [26 ] C.T. Williams, A.A. Tolia, H.Y.H. Chan, C.G. Takoudis, M.J. Weaver, J. Catal. 163 (1996) 63. [27] B.E. Nieuwenhuys, Surf. Sci. 126 (1983) 307. [28] S.M. Vesecky, P. Chen, X. Xu, D.W. Goodman, J. Vac. Sci. Technol. A 13 (1995) 1539. [29] S.M. Vesecky, D.R. Rainer, D.W. Goodman, J. Vac. Sci. Technol. A 14 (1996) 1457. [30] S.W. Jorgensen, N.D.S. Canning, R.J. Madix, Surf. Sci. 179 (1987) 322. [31] T. Yamada, I. Matsuo, J. Nakamura, M. Xie, H. Hirano, Y. Matsumoto, K. Tanaka, Surf. Sci. 231 (1990) 304. [32] G.W. Graham, A.D. Logan, M. Shelef, J. Phys. Chem. 97 (1993) 5445. [33] M. Date´, H. Okuyama, N. Takagi, M. Nishijima, T. Aruga, Surf. Sci. 341 (1995) L1096. [34] M. Date´, H. Okuyama, N. Takagi, M. Nishijima, T. Aruga, Surf. Sci. 350 (1996) 79. [35] M. Ikai, K. Tanaka, J. Chem. Phys. 110 (1999) 7031. [36 ] M. Hirsima¨ki, S. Suhonen, J. Pere, M. Valden, M. Pessa, Surf. Sci. 402–404 (1998) 187. [37] P.W. Davies, R.M. Lambert, Surf. Sci. 110 (1981) 227. [38] Y. Ohno, K. Kimura, M. Bi, T. Matsushima, J. Chem. Phys. 110 (1999) 8221. [39] M. Ikai, H. He, C.E. Borroni-Bird, H. Hirano, K. Tanaka, Surf. Sci. 315 (1996) L976. [40] M. Ikai, K. Tanaka, Surf. Sci. 357/358 (1996) 781. [41] R.G. Sharpe, M. Bowker, Surf. Sci. 360 (1996) 21. [42] T. Matsushima, K. Shobatake, Y. Ohno, K. Tabayashi, J. Chem. Phys. 94 (1992) 2783. [43] G. Comsa, R. David, Surf. Sci. Rep. 5 (1985) 145. [44] H. Wang, R.G. Tobin, C.G. DiMaggio, G.B. Fisher, D.K. Lambert, J. Chem. Phys. 107 (1997) 9569. [45] K. Kimura, Y. Ohno, T. Matsushima, Surf. Sci. 429 (1999) L455. [46 ] R.A. Bennett, S. Poulston, I.Z. Jones, M. Bowker, Surf. Sci. 401 (1998) 72. [47] T. Matsushima, Heterogen. Chem. Rev. 2 (1995) 51. [48] Y. Li, M. Bowker, Surf. Sci. 348 (1996) 67.

I. Kobal et al. / Surface Science 445 (2000) 472–479 [49] W.M. Daniel, Y. Kim, H.C. Peebles, J.M. White, Surf. Sci. 111 (1981) 189. [50] P. Va¨terlein, T. Krause, M. Ba¨ssler, R. Fink, E. Umbach, J. Taborski, V. Wu¨stenhagen, W. Wurth, Phys. Rev. Lett. 76 (1996) 4749. [51] C. Kodama, H. Orita, H. Nozoye, Appl. Surf. Sci. 121/ 122 (1997) 579. [52] C.G. Takoudis, L.D. Schmidt, J. Catal. 80 (1983) 274. [53] K. Sawabe, Y. Matsumoto, Surf. Sci. 283 (1993) 126. [54] R. Klein, R. Siegel, Surf. Sci. 92 (1980) 337. [55] R.P.H. Gasser, C.J. Marsay, Surf. Sci. 20 (1970) 116. [56 ] M. Brandt, T. Greber, F. Kuhlmann, N. Bo¨wering, U. Heinzmann, Surf. Sci. 402–404 (1998) 160. [57] W.H. Weinberg, R.P. Merrill, Surf. Sci. 32 (1972) 317. [58] A. Pashutski, M. Folman, Surf. Sci. 216 (1989) 395. [59] F.H.P.M. Habraken, G.A. Bootsma, Surf. Sci. 87 (1979) 333. [60] N.R. Avery, Surf. Sci. 131 (1983) 501. [61] T.E. Madey, N.R. Avery, A.B. Anton, B.H. Toby, W.H. Weinberg, J. Vac. Sci. Technol. A 1 (1983) 1220. [62] E. Umbach, D. Menzel, Chem. Phys. Lett. 84 (1981) 491. [63] M. Sano, Y. Ohno, T. Yamanaka, T. Matsushima, E.B. Quiney, K. Jacobi, J. Chem. Phys. 108 (1998) 10 231. [64] K. Sawabe, J. Lee, Y. Matsumoto, J. Chem. Phys. 99 (1993) 3143.

479

[65] D.J. Bald, S. Bernasek, J. Chem. Phys. 109 (1998) 746. [66 ] T. Matsushima, T. Yamanaka, C. Moise, Trends Chem. Phys. 4 (1996) 1. [67] K. Yagi, K. Higashiyama, H. Fukutani, Surf. Sci. 295 (1993) 230. [68] Y. Kuwahara, M. Jo, H. Tsuda, M. Onchi, M. Nishijima, Surf. Sci. 180 (1987) 421. [69] M. Bertolo, K. Jacobi, Surf. Sci. 265 (1992) 1. [70] A.A. Radzig, B.M. Smirnov, Reference Data on Atoms, Molecules and Ions, Springer, Berlin, 1985. [71] R.P. Thorman, D. Anderson, S.L. Bernasek, Phys. Rev. Lett. 11 (1980) 743. [72] R.C. Cosser, S.R. Bare, S.M. Francis, D.A. King, Vacuum 31 (1981) 503. [73] D.J. Auerbach, H.E. Pfnu¨r, C.T. Rettner, J.E. Schlaegel, J. Lee, R.J. Madix, J. Chem. Phys. 81 (1984) 2515. [74] G. Cao, Y. Seimiya, Y. Ohno, T. Matsushima, Chem. Phys. Lett. 294 (1998) 419. [75] M. Ikai, N.M.H. Janssen, B.E. Nieuwenhuys, K. Tanaka, J. Chem. Phys. 106 (1997) 311. [76 ] J. Lee, H. Kato, K. Sawabe, Y. Matsumoto, Chem. Phys. Lett. 240 (1995) 417. [77] D. Loffreda, D. Simon, P. Sautet, J. Chem. Phys. 108 (1998) 6447.