Structural Effects of Electrochemical Oxidation of Formic Acid on

Journal of the American Chemical Society 0 (proofing), ... The high Tafel slope and small potential dependence of activation energy for formic acid ox...
0 downloads 0 Views 107KB Size
Download PDF
12480

J. Phys. Chem. B 2006, 110, 12480-12484

Structural Effects of Electrochemical Oxidation of Formic Acid on Single Crystal Electrodes of Palladium Nagahiro Hoshi,* Kaori Kida, Masashi Nakamura, Miou Nakada, and Kazuhito Osada Department of Applied Chemistry and Biotechnology, Faculty of Engineering, Chiba UniVersity, 1-33, Yayoi-cho, Inage-ku, Chiba 263-8522, Japan ReceiVed: February 9, 2006; In Final Form: May 1, 2006

Structural effects on the rates of formic acid oxidation have been studied on Pd(111), Pd(100), Pd(110), and Pd(S)-[n(100)×(111)] (n ) 2-9) electrodes in 0.1 M HClO4 containing 0.1 M formic acid with use of voltammetry. On the low index planes of Pd, the maximum current density of formic acid oxidation (jP) increases in the positive scan as follows: Pd(110) < Pd(111) < Pd(100). This order differs from that on the low index planes of Pt: Pt(111) < Pt(100) < Pt(110). Pd(S)-[n(100)×(111)] electrodes with terrace atomic rows n g 3 have almost the same jP as Pd(100), except Pd(911) n ) 5. The value of jP on Pd(911) n ) 5 is 20% higher than those of the other surfaces. Pd(311) n ) 2, of which the first layer is composed of only step atoms, has the lowest jP in the Pd(S)-[n(100)×(111)] series. The adsorption geometry of the reaction intermediate (formate ion) is optimized by using density functional theory.

Introduction Surface structure affects the catalytic activity and the selectivity of electrochemical reactions remarkably.1-3 Elucidation of surface structures enhancing the activity for electrochemical reactions will contribute to the development of new electrocatalysts with high efficiency and low energy consumption. A well-defined surface plays a key role to determine the active site of a catalytic reaction. Especially, a stepped surface is important for the estimation of an active site, because the surface structures can be customized systematically. A stepped surface is also regarded as a model of a well-defined nanoparticle that can be used for a practical catalyst. Formic acid is one of the candidates for the fuels of fuel cells. It is also a model material for the anodic reaction of a direct methanol fuel cell (DMFC); methanol is oxidized via adsorbed CO and adsorbed formate ion (HCOO-) on Pt and Pd electrodes (dual path mechanism).4,5 The oxidation current of formic acid on polycrystalline Pd is higher than that on Pt.6,7 Elucidation of Pd surface structures enhancing formic acid oxidation may contribute to the development of direct fuel cells with higher efficiency. Structural effects on the oxidation of formic acid have been studied by using single-crystal electrodes of Pt.2,8-10 Formic acid is oxidized on Pt electrodes in acidic solution via poisonous intermediate and reactive intermediate.7,11,12 Infrared reflection absorption spectroscopy (IRAS) verified that the poisonous intermediate is adsorbed CO.13-15 Surface enhanced infrared reflection absorption spectroscopy (SEIRAS) identified the reactive intermediate as adsorbed formate ion.16 No anodic current is obtained in the voltammogram up to 0.6 V (RHE) on Pt electrodes in the solution containing formic acid, because the surfaces are covered with adsorbed CO. The current of formic acid oxidation is observed at the onset potential of adsorbed CO oxidation. The oxidation rate of formic acid, which is estimated from the anodic peak current of a voltammogram * Address correspondence [email protected].

to

this

author.

E-mail:

in the positive scan, gives the following order in 0.1 M HClO4: Pt(111) < Pt(100) < Pt(110).2,8 The structural effect on the formic acid oxidation is attributed to the difference of the onset potential of adsorbed CO oxidation.8 On Pd electrodes, no adsorbed CO is produced during the formic acid oxidation according to IRAS study.16 Adsorbed formate ion is detected by using SEIRAS on Pd and Pd/Au(111) electrodes.17,18 Kolb et al. reported the voltammograms of formic acid oxidation on the low index planes of Pd in 0.5 M H2SO4sthe maximum current density of the oxidation gives the following order in the positive scan: Pd(111) < Pd(110) < Pd(100).19 They did not discuss, however, the activity on Pd single crystals in detail, because their scope was the catalytic activity of Pd adlayer on Au single-crystal surfaces. In this paper, we report voltammograms of formic acid oxidation on the low index planes of Pd and Pd(S)-[n(100)×(111)] surfaces that are composed of n atomic rows of (100) terrace and (111) step. We adopted 0.1 M HClO4 as an electrolytic solution, in which no anion is strongly adsorbed on the electrode. The structural effects on the oxidation rates are discussed according to the adsorption geometry of formate ion that is optimized by using the density functional theory. Experimental Section A single-crystal bead of Pd in about 3 mm diameter was prepared from Pd wire (1 mm diameter, 99.99% purity, Furuuchi Kagaku) according to the method described previously.20-22 The crystal was oriented by using the reflection beam of a He-Ne laser from (111) and (100) facets on the crystal,23 and then mechanically polished with diamond slurries. The polished surface was annealed in an H2/O2 flame at about 1300 °C for removing distortions due to the mechanical polishing, and then cooled to room temperature in an Ar atmosphere. The annealed surface was protected with ultrapure water, and transferred to the electrochemical cell. The surfaces examined were Pd(111), Pd(100), Pd(110), and Pd(S)-[n(100)×(111)] (n ) 2, 3, 4, 5, 6, 9). The cross sectional areas of the electrodes were between

10.1021/jp0608372 CCC: $33.50 © 2006 American Chemical Society Published on Web 06/02/2006

Structural Effects of Electrochemical Oxidation

Figure 1. (a) Voltammograms of the oxidation of formic acid on the low index planes of Pd in 0.1 M HClO4 containing 0.1 M formic acid. (b) Voltammograms in 0.1 M HClO4 without formic acid. Scanning rate: 0.020 V s-1.

0.065 and 0.075 cm2. The scanning rate of a voltammogram was 0.020 V s-1. Electrolytic solutions were prepared from ultrapure water treated with Milli Q plus low TOC (Millipore) and suprapur grade chemicals (Merck). The purity of Ar is higher than 99.9999%. All the potentials were referred to RHE. IRAS spectra were measured by using a JIR-6000M spectrometer (JEOL) with p-polarized light with a resolution of 4 cm-1. Infrared light was incident to the electrode surface through a CaF2 prism with the angle of incidence 65°. The reflected light was detected with an MCT detector cooled with liquid nitrogen. IRAS spectra were measured during the oxidation of formic acid according to the following procedure: (1) An annealed Pd electrode was immersed into 0.1 M HClO4 containing 0.1 M formic acid at a sample potential (ES) at which formic acid is oxidized, and kept in the bulk solution for 60 s, then pushed to the CaF2 window. (2) Sample spectra RS were collected at ES. The spectra were averaged over 100 scans. (3) The potential was stepped to a reference potential at 1.0 V (ER) where the oxidation of formic acid stops, and reference spectra RR were collected. (4) The subtracted spectrum is calculated with the following equation: absorbance ) -log(RS/RR). Results and Discussion Pd single-crystal electrodes give voltammograms characteristic of their orientation in 0.5 M H2SO4 solution.21,22 The voltammograms of the prepared electrodes agreed with those reported previously, thus we judged that the surfaces were welldefined. Low Index Planes. Figure 1a shows the voltammograms of formic acid oxidation in 0.1 M HClO4 containing 0.1 M formic acid. All the voltammograms are the first scan after the annealing. We scan the potential up to the oxide film formation region to examine the correlation of the formic acid oxidation with the oxide film formation. The surfaces of Pd are roughened by the formation and reduction of the oxide film; the surfaces are not well-defined in the negative scans. Thus we estimated the structural effects using the curves in the first positive scans. The currents of formic acid oxidation are observed above 0.2 V on all the electrodes. The onset potentials of the anodic current are much lower than those of Pt electrodes,2,8-10 showing

J. Phys. Chem. B, Vol. 110, No. 25, 2006 12481 that no adsorbed CO is formed on Pd surfaces. We measured IRAS spectra on Pd(100) at ES between 0.1 and 0.7 V. No band of adsorbed CO was obtained. Pd single crystals do not provide stable adsorbed CO during formic acid oxidation, as is the case of the polycrystalline Pd electrode. Other featuring bands including adsorbed formate ion were not detected. It is difficult for the conventional IRAS method to detect a reactive intermediate such as formate ion. The supply of formic acid to the electrode surface is heavily inhibited in the IRAS configuration: the electrode is pushed to the window strongly for the decrease of IR absorption by the solvent. Thus, formic acid and reactive intermediate disappear too fast during the collection of IR spectra. On the other hand, SEIRAS measurement successfully detected adsorbed formate ion on the Pd/Au(111) electrode.18 Electrode surfaces are open to the bulk solution in the SEIRAS configuration; formic acid is supplied to the surface in the same way as for voltammogram measurements. A SEIRAS method needs, however, surface roughness of 20-30 nm for the enhancement of IR absorptivity.24 “Well-defined roughness” of (111) orientation can be obtained by the evaporation of Au onto the base plane of a silicon prism.25 It is, however, impossible to prepare “well-defined” roughened surfaces other than (111) orientation. Thus the SEIRAS method cannot be applied to stepped surfaces. We estimate the maximum oxidation rate of formic acid using the anodic peak current density in the positive scan. In the case of formic acid oxidation on Pd, the voltammogram will give a peak current according to the following mechanisms: (1) the reaction is diffusion controlled and (2) the rate of the oxidation reaction gives a maximum at a certain potential. If a reaction is diffusion controlled, the maximum current is determined by the concentration gradient in a diffusion layer. In such a case, the maximum current density does not depend on crystal orientation. Remarkable orientation dependence of the voltammogram shows that the oxidation of formic acid is not a diffusion-controlled reaction (Figure 1). In addition, the rate of a diffusion-controlled reaction depends on the stirring rate of a solution, because the thickness of a diffusion layer decreases with the increase of the stirring rate. The voltammogram of Pd(100) in the solution stirred at 400 rpm gave the same curve as the one in the static solution in Figure 1. These facts strongly support that the oxidation of formic acid on Pd electrodes is not a diffusioncontrolled reaction. The maximum current in the voltammogram shows the maximum rate of formic acid oxidation in Figure 1. The maximum oxidation rate depends on the surface structure remarkably: Pd(110) < Pd(111) < Pd(100). Pd(100) has the highest oxidation rate of formic acid. The order differs from that on Pt, which is assessed by using the maximum anodic current density in the positive scan in 0.1 M HClO4: Pt(111) < Pt(100) < Pt(110).2,8 The difference in the structural effects between Pd and Pt may be attributed to the difference of the mechanism of formic acid oxidation: adsorbed CO is produced on Pt, whereas no CO is formed on Pd. The oxidation rate on Pd(100) was higher than that on Pt(110). The oxidation rates of formic acid in HClO4 are higher than those in H2SO4.19 That is because the (bi)sulfate anion is strongly adsorbed on Pd surfaces above 0.2 V26-28 to inhibit the formic acid oxidation in H2SO4. The order of the oxidation rates in HClO4 differs from that in H2SO4: Pd(111) < Pd(110) < Pd(100). IRAS study shows the coverage of adsorbed (bi)sulfate anion on Pd(111) is much higher than that on Pd(110).26 The higher coverage of (bi)sulfate anion may prevent the formic acid oxidation on Pd(111) more significantly.

12482 J. Phys. Chem. B, Vol. 110, No. 25, 2006

Hoshi et al.

Figure 2. Selected voltammograms of the oxidation of formic acid on Pd(S)-[n(100)×(111)] electrodes in 0.1 M HClO4 containing 0.1 M formic acid. Scanning rate: 0.020 V s-1.

When the oxidation rate of formic acid is assessed at lower potential in 0.1 M HClO4, a different order of reactivity is obtained in Figure 1: e.g., at 0.5 V, Pd(100) < Pd(110) < Pd(111). The Pd(111) electrode is the most active for formic acid oxidation at negative potentials. The potential of the anodic peak depends on the crystal orientation in Figure 1a: 0.4 V on Pd(110), 0.5 V on Pt(111), and 0.8 V on Pd(100). On Pt electrodes, oxide film formation affects the oxidation of methanol and formic acid, because adsorbed CO (poisonous intermediate) is oxidized by the oxygen donor on the surface such as PtOH: COad + 2PtOHad f CO2 + H2O + 2Pt.8 Figure 1b shows voltammograms in 0.1 M HClO4 without formic acid. Comparison of Figure 1a with Figure 1b shows that formic acid oxidation needs no oxygen donor on Pd electrodes. On Pd(100), PdOH (Pd2O) is formed at 0.7 V, but the onset potential of formic acid oxidation is far more negative (0.2 V). The current of formic acid oxidation drops at 0.83 V, and becomes zero at 1.0 V. Anodic charge after the double layer correction supports that the first layer of Pd(100) is fully covered with PdOH (Pd2O) at 1.0 V in Figure 1b. These results suggest that oxygen donor on the surface inhibits the oxidation of formic acid on Pd(100). The oxidation may proceed according to the following mechanism: HCOOH f HCOOad- + H+ f CO2 + 2H+ + 2e.16,29 Above 1.0 V, PdOH (Pd2O) will be further oxidized to PdO.30,31 When the potential is scanned to the negative direction, the oxidized species of Pd is reduced as the following: PdO f PdOH (Pd2O) f Pd. When the bare surface of Pd is regenerated, formic acid begins to be oxidized abruptly and a sharp anodic current is obtained in Figure 1a. On Pd(110) and Pd(111), the anodic currents give maxima at 0.4 and 0.5 V in the positive scan, respectively. The onset potentials of the oxide film formation, however, do not coincide with the anodic peak of formic acid oxidation. These facts suggest that the oxide film has nothing to do with the anodic peak on Pd(110) and Pd(111). On Pd(110), the current of formic acid oxidation decreases abruptly at 0.75 V. This potential agrees with the onset of PdOH (Pd2O) formation (Figure 1b). The oxidation of formic acid is inhibited completely at 1.0 V where the first layer of Pd(110) is covered with PdOH (Pd2O), as is the case of Pd(100). Oxidation of formic acid is also prevented at 1.0 V on Pd(111). However, the surface is partially covered with PdOH (Pd2O) at this potential. Not only the oxide film but also the interaction of adsorbed formate ion with the surface may determine the curve of formic acid oxidation. A scan to the negative direction gives a sharp anodic peak in Figure 1a. This is not due to the surface roughness caused by the oxide film formation, because the second scan to the positive direction gave a lower current than the first cycle. A previous study reports that the reduction of the oxide film of Pd activates the oxidation of formic acid on Pd polycrystalline electrodes.32 Low index planes of Pd give sharp anodic peaks at the potentials where most of the oxide film is reduced in the negative scan

Figure 3. Maximum anodic current density (jP) plotted against step atom density of Pd(S)-[n(100)×(111)] electrodes in 0.1 M HClO4 containing 0.1 M formic acid.

(Figure 1). Pd atoms, not at equilibrium with the crystal lattice shortly after the oxide film reduction, may have high activity for formic acid oxidation.32 The Pd(S)-[n(100)×(111)] Series. Because Pd(100) has the highest rate for the formic acid oxidation in the low index planes, we incorporate the (111) step onto the (100) terrace to find out surface structures that have higher activity for formic acid oxidation. Figure 2 shows representative voltammograms of Pd(S)-[n(100)×(111)] electrodes in 0.1 M HClO4 containing 0.1 M formic acid. The potential of oxide film formation did not depend on the crystal orientation in this series: PdOH begins to be formed at 0.75 V and reached full coverage at 1.0 V in 0.1 M HClO4 without formic acid. The anodic current drops around the onset potential of PdOH formation, becoming zero at the potential where the oxide film fully covers the first layer, as is the case for Pd(100). The maximum rate of the formic acid oxidation is again assessed with the anodic peak current density (jP) in the positive scan. The potential of the anodic peak depends little on the orientation of Pd(S)-[n(100)×(111)]: all the anodic peaks give maxima around 0.8 V in the positive scan in Figure 2. The values of jP are plotted against the density of step atoms in Figure 3. The surfaces with terrace atomic rows n g 3 have almost the same value of jP as Pd(100), except Pd(911) n ) 5. This result indicates that the (100) terrace with n g 3 is suitable for formic acid oxidation. The value of jP on Pd(911) n ) 5 is 20% higher than that of the other surfaces with n g 3. This result is definitely reproducible, because two sets of Pd(911) n ) 5 electrodes give the same values. Pd(311) n ) 2, of which the first layer is composed of only step atoms, has a low rate of formic acid oxidation, as is the case for Pd(110) of which the first layer also consists of only step atoms (Figure 1). These facts support that step site is not appropriate for the fast formic acid oxidation. This is contrary to the case of adsorbed CO oxidation on the Pt(S)-[n(111)×(111)] series on which step sites activates CO oxidation.33,34 Calculation with Density Functional Theory. There is no way to detect the reaction intermediate (formate ion) on the

Structural Effects of Electrochemical Oxidation

J. Phys. Chem. B, Vol. 110, No. 25, 2006 12483

Figure 4. Adsorption of formate ion on Pd(S)-[n(100)×(111)] electrodes optimized using density functional theory.

Figure 5. (a) Formate ion adsorbed on Pd(S)-[n(100)×(111)] electrodes with n g 3. (b) Formate ion adsorbed on Pd(311) n ) 2. The viewpoint is parallel to the molecular plane.

stepped surfaces of Pd; the adsorption geometry of formate ion is calculated on the basis of density functional theory. Gaussian 03 is used for the calculation. Basis sets for C, H, and O atoms are BLYP/3-21G, and the one for Pd is LANL2DZ ECP. The adsorption geometry of the adsorbed formate ion is optimized with the structure of the Pd cluster fixed. Figure 4 shows the optimized adsorption geometry of formate ions. On the surfaces with n g 3, the formate ion is adsorbed on the bridge site of the (100) terrace with its molecular plane perpendicular to the step line. Pd(911) n ) 5, which has the highest rate for fomic acid oxidation, adsorbs two formate ions on the (100) terrace. One of the formate ions locates at the terrace edge, and the other is adsorbed near the foot of the step. Two formate ions are also adsorbed on the (100) terrace of Pd(11 1 1) n ) 6, but the intermolecular distance is longer than that on Pd(911) n ) 5. Densely packed adsorbed formate ions adjacent to step sites may promote the formic acid oxidation on Pd(911) n ) 5. On Pd(311) n ) 2, which has the lowest rate of formic acid oxidation in the Pd(S)-[n(100)×(111)] series, the formate ion is adsorbed on the step with its molecular plane parallel to the step line. Study in ultrahigh vacuum shows that a bending motion of the molecular plane promotes the decomposition of the methoxy ion.35 A vacant site adjacent to the adsorbed methoxy ion is necessary for this bending motion (Figure 5a). Chemical oscillation of formic acid oxidation on Pt electrodes was simulated successfully by using the vacant site.36 Formate ions adsorbed on the surfaces with n g 3 have vacant sites adjacent to the molecular plane as shown in Figure 4. On Pd(311) n ) 2, however, step atoms occupy the site adjacent to the molecular plane; the bending vibration of the molecular plane is inhibited by the step atoms (Figure 5b). That is why the rates of formic acid oxidation are low on Pd(311). Conclusions (1) Pd(100) has the highest rate of formic acid oxidation in the low index planes of Pd according to the current peak in the voltammogram in the positive scan: Pd(110) < Pd(111) < Pd(100). This activity series differs from that on Pt: Pt(111) < Pt(100) < Pt(110). The oxidation rate of formic acid on Pd(100) is higher than that on Pt(110).

(2) Pd(S)-[n(100)×(111)] electrodes have almost the same oxidation rates of formic acid as Pd(100) on surfaces with n g 3, except Pd(911) n ) 5. The oxidation rate of formic acid on Pd(911) n ) 5 is 20% higher than that of Pd(100). (3) Pd(110) ()2(111)-(111)) and Pd(311) ()2(100)-(111)), of which the first layers are composed of only step atoms, have low rates of formic acid oxidation. (4) Density functional theory predicts that formate ion is adsorbed on the (100) terrace with two oxygen atoms combined to bridge the site of (100) on surfaces with n g 3. Pd(311) n ) 2 adsorbs formate ion along the step with the molecular plane parallel to the step line. Acknowledgment. This work was supported by a grant from the New Energy and Industrial Technology Development Organization, and partially by the Asahi Glass Foundation. References and Notes (1) Adzic, R. Modern Aspects of Electrochemistry; White, R. E., Bockris, J. O’M.; Conway, B. E., Eds.; Plenum Press: New York, 1990; Vol. 21, Chapter 5. (2) Lamy C.; Leger, J. M. J. Chim. Phys. 1991, 88, 1649. (3) Markoviæ, N. M.; Ross, P. N., Jr. Surf. Sci. Rep. 2002, 45, 117. (4) Miki, A.; Ye, S.; Osawa, M. Chem. Commun. 2002, 1500. (5) Chen, Y. X.; Miki, A.; Ye, S.; Sakai, H.; Osawa, M. J. Am. Chem. Soc. 2003, 125, 3680. (6) Watanabe, M.; Suzuki, T.; Motoo, S. Electrochemistry 1971, 39, 394. (7) Capon, A.; Parsons, R. J. Electroanal. Chem. 1973, 44, 239. (8) Adzˇic´, R. R.; Tripkoviæ, A. V.; O’Grady, W. E. Nature 1982, 296, 137. (9) Motoo, S.; Furuya, N. J. Electroanal. Chem. 1985, 184, 303. (10) Sun, S. G.; Clavilier, J. J. Electroanal. Chem. 1986, 199, 471. (11) Capon, A.; Parsons, R. J. Electroanal. Chem. 1973, 44, 1. (12) Capon, A.; Parsons, R. J. Electroanal. Chem. 1973, 45, 205. (13) Beden, B.; Bewick, A.; Lamy, C. J. Electroanal. Chem. 1983, 148, 147. (14) Kunimatsu, K. J. Electroanal. Chem. 1986, 213, 149. (15) Sun, S. G.; Clavilier, J. J. Electroanal. Chem. 1988, 240, 147. (16) Nishimura, K.; Kunimatsu, K.; Machida, K.; Enyo, M. J. Electroanal. Chem. 1989, 260, 181. (17) Iwasaki, S.; Miki, A.; Ye, S.; Osawa, M. Abstract of the 2002 Fall Meeting of the Electrochemical Society of Japan, 2002, 2M26. (18) Pronkin, S.; Hara, M.; Wandlowski, Th. Russ. J. Electrochem. In press. (19) Baldauf, M.; Kolb, D. M. J. Phys. Chem. 1996, 100, 11375. (20) Clavilier, J.; Faure, R.; Guinet, G.; Durand, R. J. Electroanal. Chem. 1980, 107, 205. (21) Hoshi, N.; Kagaya, K.; Hori, Y. J. Electroanal. Chem. 2000, 485, 55. (22) Hoshi, N.; Kuroda, M.; Hori, Y. J. Electroanal. Chem. 2002, 521, 155. (23) Furuya, N.; Koide, S. Surf. Sci. 1989, 220, 18. (24) Osawa, M.; Ataka, K.; Yoshii, K.; Yotsuyanagi, K. J. Electron Spectrosc. Relat. Phenom. 1993, 64/65, 371. (25) Sun, S.-G.; Cai, W.-B.; Wan, L.-J.; Osawa, M. J. Phys. Chem. B 1999, 103, 2460. (26) Hoshi, N.; Kuroda, M.; Koga, O.; Hori, Y. J. Phys. Chem. B 2002, 106, 9107.

12484 J. Phys. Chem. B, Vol. 110, No. 25, 2006 (27) Hoshi, N.; Kuroda, M.; Ogawa, T.; Koga, O.; Hori, Y. Langmuir 2004, 20, 5066. (28) Pronkin, S.; Wandlowski, Th. Surf. Sci. 2004, 573, 109. (29) Solis, V.; Iwasita, T.; Pavese, A.; Vielstich, W. J. Electroanal. Chem. 1988, 255, 155. (30) Chiechie, T.; Mayer, C.; Lorenz, W. J. Electroanal. Chem. 1982, 135, 211. (31) Sashikata, K.; Matsui, Y.; Itaya, K. J. Phys. Chem. 1996, 100, 20027.

Hoshi et al. (32) Manzanares, M. I.; Pavese, A. G.; Solis, V. M. J. Electroanal. Chem. 1991, 310, 159. (33) Lebedeva, N. P.; Koper, M. T. M.; Herrero, R.; Feliu, J. M.; van Santen, R. A. J. Electroanal. Chem. 2000, 487, 37. (34) Hoshi, N.; Tanizaki, M.; Koga, O.; Hori, Y. Chem. Phys. Lett. 2001, 336, 13. (35) Li, Y.; Bowker, M. Surf. Sci. 1993, 285, 219. (36) Samjeske´, G.; Miki, A.; Ye, S.; Yamakata, A.; Mukouyama, Y.; Okamoto, H.; Osawa, M. J. Phys. Chem. B 2005, 109, 23509.