1: \documentclass[prl,aps,twocolumn,showpacs,home,floats]{revtex4}

2: %\documentclass[prl,superscriptaddress,aps,twocolumn,showpacs,home,floats]{revtex4}

3: %\documentclass[prl,aps,preprint,showpacs,home,floats]{revtex4}

4: 

5: \usepackage{graphics}

6: \bibliographystyle{apsrev}

7: \begin{document}

8: 

9: \title{{\rm\small\hfill (Phys. Rev. Lett., in press)}\\

10: Kinetic hindrance during the initial oxidation of Pd(100) at ambient pressures}

11: 

12: 

13: \author{E. Lundgren}

14: \email{edvin.lundgren@sljus.lu.se}

15: \affiliation{Department of Synchrotron Radiation Research,

16: Institute of Physics, University

17: of Lund, Box 118, S-221 00 Lund, Sweden}

18: \author{A. Stierle}

19: \affiliation{Max-Planck Institut f\"ur Metallforschung,

20: Heisenbergstr.3, D-70569 Stuttgart, Germany}

21: \author{M. Todorova}

22: \affiliation{Fritz-Haber Institut der Max-Planck Gesellschaft,

23: Faradayweg 4-6, D-14195 Berlin, Germany}

24: \author{J. Gustafson}

25: \affiliation{Department of Synchrotron Radiation Research,

26: Institute of Physics, University of Lund, Box 118, S-221 00 Lund,

27: Sweden}

28: \author{A. Mikkelsen}

29: \affiliation{Department of Synchrotron Radiation Research,

30: Institute of Physics, University of Lund, Box 118, S-221 00 Lund,

31: Sweden}

32: \author{J. Rogal}

33: \affiliation{Fritz-Haber Institut der Max-Planck Gesellschaft,

34: Faradayweg 4-6, D-14195 Berlin, Germany}

35: \author{K. Reuter}

36: \affiliation{Fritz-Haber Institut der Max-Planck Gesellschaft,

37: Faradayweg 4-6, D-14195 Berlin, Germany}

38: \author{J.N. Andersen}

39: \affiliation{Department of Synchrotron Radiation Research,

40: Institute of Physics, University of Lund, Box 118, S-221 00 Lund,

41: Sweden}

42: \author{H. Dosch}

43: \affiliation{Max-Planck Institut f\"ur Metallforschung,

44: Heisenbergstr.3, D-70569 Stuttgart, Germany}

45: \author{M. Scheffler}

46: \affiliation{Fritz-Haber Institut der Max-Planck Gesellschaft,

47: Faradayweg 4-6, D-14195 Berlin, Germany}

48: 

49: 

50: \begin{abstract}

51: The oxidation of the Pd(100) surface at oxygen pressures in the

52: 10${}^{-6}$ to 10${}^3$\,mbar range and temperatures up to 1000 K

53: has been studied {\em in-situ} by surface x-ray diffraction (SXRD)

54: The results provide direct structural information on the phases

55: present in the surface region and on the kinetics of the oxide

56: formation. Depending on the $(T,p)$ environmental conditions we

57: either observe a thin $(\sqrt{5} \times \sqrt{5})R27^{\circ}$

58: surface oxide or the growth of a rough, poorly ordered bulk oxide

59: film of PdO predominantly with (001) orientation. By either

60: comparison to the surface phase diagram from first-principles

61: atomistic thermodynamics or by explicit time-resolved measurements

62: we identify a strong kinetic hindrance to the bulk oxide formation

63: even at temperatures as high as 675\,K.

64: \end{abstract}

65: 

66: \date{\today}

67: \pacs{61.10.-i, 81.65.Mq, 68.55.Jk, 68.43.Bc}

68: 

69: %%KR Edvin: Here's two possible PACS suggestions. I don't know,

70: %%KR whether there is a special PACS for high-pressure experiments,

71: %%KR but there is certainly one for SXRD. The 68.43.Bc should come

72: %%KR last - we don't want to get theoretical referees...

73: 

74: %61.10.-i %X-ray diffraction and scattering

75: %81.65.Mq  %Oxidation

76: %68.43.Bc % Ab initio calculations of adsorbate structure and

77: %reactions

78: 

79: \maketitle

80: 

81: Many technologically important materials containing transition

82: metals are intended for use under oxygen pressures much higher

83: than those of the high or ultra high vacuum (UHV) environment

84: typically employed in surface science related investigations of

85: the structural, electronic, and chemical properties of these

86: materials. As the surface properties of such materials may be

87: significantly altered by the oxidation or corrosion

88: \cite{over00,hendriksen02,stampfl02} occurring at oxygen pressures

89: difficult to achieve in conventional UHV experiments, it is

90: disconcerting that most of our present atomic-scale knowledge

91: derives from such experiments or from theoretical treatments which

92: neglect the surrounding gas-phase. Despite the pressure

93: limitations a few such experiments and theoretical simulations

94: have in recent years significantly advanced our atomic-scale

95: understanding of the initial oxidation of metal surfaces,

96: demonstrating e.g. how radically a surface may change its

97: structure and functionality under conditions appropriate for high

98: pressure oxidation catalysis \cite{over00,hendriksen02} and how

99: the growth of bulk oxide films is often preceded by the formation

100: of few-atomic-layer-thin so-called surface oxides of complex

101: geometrical structure and with properties often unknown from the

102: bulk oxides \cite{carlisle00,lundgren02,todorova03}. The

103: unexpected and complex behavior revealed by these experiments

104: emphasizes the need for further {\em in-situ} investigations of

105: the structural, electronic and chemical surface properties at

106: higher oxygen pressures but maintaining the accuracy known from

107: UHV studies - in particular to also address the kinetics of

108: oxidation and corrosion processes.

109: 

110: A main reason for the lack of {\em in-situ} investigations has

111: been the scarcity of appropriate experimental techniques that

112: provide the aspired atomically-resolved information at high

113: pressure {\em in situ}, or theories that explicitly include the

114: effect of the surrounding gas-phase. Recently enforced attempts to

115: overcome this limitation have on the theoretical side led to the

116: development of first-principles atomistic thermodynamics, where

117: electronic structure theory calculations are combined with

118: thermodynamic considerations to address the surface structure and

119: composition of a metal surface in {\em equilibrium} with arbitrary

120: environments (See Refs. \cite{reuter02,li03} and references

121: therein). On the experimental side the state-of-the-art is,

122: however, currently still characterized by either traditional {\em

123: ex-situ} atomic-scale investigations \cite{over00}, or new high

124: pressure techniques like high pressure scanning tunneling

125: microscopy (STM) that still lack atomic resolution

126: \cite{hendriksen02}.

127: 

128: 

129: In this Letter we demonstrate how precise knowledge of the

130: $(T,p)$-conditions under which such various oxides, -bulk and

131: surface-, exist can be obtained by means of {\em in-situ} surface

132: X-ray diffraction (SXRD) measurements. We have chosen to study Pd

133: as the metal is a highly active oxidation catalyst of hydrocarbons

134: under oxygen-rich conditions \cite{ziauddin97}. Still it is

135: unknown whether Pd or PdO is the active phase \cite{veser99}.

136: Monitoring the oxidation of the Pd(100) surface over the pressure

137: range from 10${}^{-6}$ to 10${}^3$\,mbar and up to sample

138: temperatures of 1000\,K, we observe the formation of both the

139: previously characterized $(\sqrt{5} \times \sqrt{5})R27^{\circ}$

140: surface oxide \cite{todorova03} and the transform to

141: three-dimensional bulk oxide films. Framing our experimental data

142: with the equilibrium results from atomistic thermodynamics

143: calculations we can identify kinetically inhibited bulk oxide

144: growth even at temperatures as high as 675\,K. Under suitable

145: conditions this hindered transformation to the bulk oxide can be

146: followed on a time scale currently accessible to the experiment,

147: opening the door to time-resolved atomic-scale studies of the

148: initial oxidation process of metal surfaces at ambient pressures.

149: 

150: The SXRD measurements were performed at the Angstr{\o}m Quelle

151: Karlsruhe (ANKA) beamline in Germany \cite{beamline}. A photon

152: energy of 10.5\,keV was used and the experiments were conducted in

153: a six-circle diffraction mode. The crystal basis used to describe

154: the (H K L) diffraction is a tetragonal basis set \

155: ({\bf{a}$_{1}$},{\bf{a}$_{2}$},{\bf{a}$_{3}$}), with

156: {\bf{a}$_{1}$} and {\bf{a}$_{2}$} lying in the surface plane and

157: of length equal to the nearest neighbor surface distance

158: a$_{\circ}$/$\sqrt{2}$, and a$_{3}$ out-of-plane with length

159: a$_{\circ}$ (a$_{\circ}$(Pd) = 3.89{\AA}). The UHV x-ray

160: diffraction chamber allowed partial oxygen pressures of up to

161: 10${}^3$ mbar, and the temperature was estimated from a

162: thermocouple mounted behind the transferable sample holder,

163: resulting in an uncertainty of the sample temperature of $\pm

164: 25$\,K. The sample and the cleaning procedure are identical to the

165: one described in an earlier UHV study \cite{todorova03}. The oxide

166: films grown were found to be metastable on the time scale of hours

167: under UHV conditions, and could readily be desorbed at 1175\,K.

168: 

169: The atomistic thermodynamics results are based on

170: density-functional theory (DFT) calculations performed within the

171: Full-Potential Linear Augmented Plane Wave (FP-LAPW) scheme

172: \cite{blaha99} using the generalized gradient approximation (GGA)

173: \cite{perdew96} for the exchange-correlation functional. The

174: supercell setup and the highly converged basis-set \cite{basis}

175: have been detailed in our preceding study \cite{todorova03}. To

176: determine the range of $(T,p)$-conditions in which this surface

177: oxide would represent the thermodynamically most stable state, we

178: evaluate the Gibbs free energy of adsorption \cite{reuter02,li03}

179: and compare it to other possible states of the system, like the

180: reported $p(2 \times 2)$ on-surface phase with O in fcc sites

181: \cite{zheng00,todorova03} or the tetragonal PdO bulk oxide

182: \cite{rogers71}.

183: 

184: \begin{figure}[t!]

185: \scalebox{0.28}{\includegraphics{fig1.ps}}

186: \caption{\label{fig1} SXRD in-plane K-scan in reciprocal lattice

187: units (r.l.u) along (0.6 K 0.3) at $T=575$\,K and

188: $p=10^{-3}$\,mbar. The location of this scan line in reciprocal

189: space is indicated in the inset showing the diffraction from the

190: $(\sqrt{5} \times \sqrt{5})R27^{\circ}$ surface oxide as observed

191: in UHV-LEED. The direction of the in-plane (H K) basis vectors and

192: the unit-cell of one of the surface oxide domains are also plotted

193: in the inset.}

194: \end{figure}

195: 

196: Figure \ref{fig1} depicts a SXRD K-scan at a sample temperature of

197: 575\,K and partial oxygen pressure of $10^{-3}$\,mbar. From the

198: inset showing the reciprocal space as observed with UHV low-energy

199: electron diffraction (LEED) it is obvious that the two observed

200: peaks in the SXRD scan arise from the $(\sqrt{5} \times

201: \sqrt{5})R27^{\circ}$ surface oxide. The width of the rocking scan

202: at the (0.6 0.8 0.3) reflection is (0.13${}^{\circ}$) equal to the

203: substrate rocking scan in the minimum of the crystal truncation

204: rod (CTR) before dosing, allowing us to conclude that the surface

205: oxide domains extend over complete substrate terraces -- in

206: agreement with our previous STM results \cite{todorova03}.

207: 

208: \begin{figure}[t!]

209: \scalebox{0.32}{\includegraphics{fig2.ps}}

210: \caption{\label{fig2} Top: SXRD out-of-plane L-scan along (0.6 0.8

211: L). Bottom: corresponding schematic of the observed out-of-plane

212: diffraction (full lines - rods due to both Pd(100) and the surface

213: oxide, dashed lines - rods only due to the surface oxide. Bragg

214: reflections from bulk Pd and bulk PdO are marked as black and grey

215: circles respectively). a) $p= 10^{-3}$\,mbar and T=575 K, only the

216: surface oxide diffraction is apparent, b) p= 10${}^3$\,mbar and

217: T=675 K, diffraction is now due to a bulk-like PdO film.}

218: \end{figure}

219: 

220: 

221: 

222: The  out-of-plane diffraction from the (0.6 0.8 L) reflection

223: under the same temperature and pressure conditions is shown as an

224: L-scan in the top section of Fig \ref{fig2}a. The smooth decrease

225: of the intensity and the absence of sharper peaks with increasing

226: L is a clear fingerprint of a single diffracting layer, in

227: agreement with the recent finding \cite{todorova03} that the

228: $(\sqrt{5} \times \sqrt{5})R27^{\circ}$ surface oxide consists of

229: a single PdO(101) layer adsorbed on the Pd(001) surface. The

230: observed diffraction changes significantly as the oxygen pressure

231: and temperature is increased to 10${}^3$\,mbar and 675 K,

232: respectively, as shown in Fig. \ref{fig2}b. Instead of a smoothly

233: decreasing diffraction intensity with increasing L, a peak is now

234: observed at L=0.74. Since the reciprocal lattice is defined by the

235: lattice constant of Pd, this peak corresponds to a lattice

236: distance of a$_{\circ}$(Pd)/0.74 = 5.26\,{\AA}, which is very

237: close to the $c$-lattice constant of bulk PdO, namely 5.33\,{\AA}.

238: Thus, we observe bulk-like diffraction from PdO, indicating the

239: formation of several layers of PdO on the Pd(001) surface.

240: Interestingly, we also note that no diffracted intensity is

241: observed at L=0 anymore demonstrating that the $(\sqrt{5} \times

242: \sqrt{5})R27^{\circ}$ surface oxide has completely disappeared

243: from the surface, i.e. the initially formed PdO(101) plane does

244: not continue to grow but instead restructures. This is in

245: agreement with our DFT calculations identifying the (101)

246: orientation as a higher-energy facet of PdO \cite{todorova03}, and

247: also with experimental observations on the preferred growth

248: direction of PdO \cite{mcbride91}. Further, since no finite

249: thickness oscillations are observed along the rod, the observed

250: oxide film must be rough, as has also been reported in a recent

251: high-pressure STM study of this surface \cite{hendriksen03}.

252: 

253: A more detailed analysis of our diffraction data allows us to even

254: draw some more quantitative conclusions about the properties of

255: the grown oxide. We observe no change in the overall shape of the

256: (1 1 L) CTR upon oxidation. Without apparent relaxation, we

257: therefore attribute the change in integrated intensity at the

258: minimum (1 1 1) to a change in surface roughness. Estimating the

259: latter within the $\beta$-model \cite{robinson88} yields

260: approximately root-mean-square roughness of 10\,{\AA}, whereas the

261: measured width of $\sim 0.1$ of the L=0.74 peak indicates an

262: average film thickness of around $\sim 3.89$\,{\AA}/0.1 = $\sim

263: 40$\,{\AA}. The poor order of the formed oxide fringe is further

264: reflected in the 1${}^{\circ}$ width of the rocking scan at the

265: L=0.74 peak, indicating either an enhanced mosaic spread, or a

266: small domain size of the PdO at the surface, which would comply

267: with the previous STM observations by Hendriksen {\em et al.}

268: \cite{hendriksen03}. This poor order is probably also reflected by

269: our inability to detect reflections from the O sublattice. In

270: fact, the (0.6 0.8 0.74) peak corresponds to the (1 0 1)

271: reflection in PdO bulk coordinates. In addition to this reflection

272: we also found the (103), (200) and the (202), and equivalents when

273: rotating by 90 degrees. By observing that this indexing is similar

274: to the selection rule for a bcc lattice for which the sum of all

275: indices must be even, we conclude that what we observe is the Pd

276: sublattice in PdO, which forms a distorted bcc lattice. The

277: orientation and geometry of the ordered domains are then

278: predominantly PdO(001)$\|$Pd(100).

279: 

280: 

281: \begin{figure}[t!]

282: \scalebox{0.32}{\includegraphics{fig3.ps}} \caption{\label{fig3}

283: Left: $(T,p)$-diagram showing the measured phases in the whole

284: range of experimentally accessible conditions from UHV to ambient

285: pressure. The ``phase boundaries'' (see text) are rough estimates

286: to guide the eye. Right: Corresponding surface phase diagram, as

287: calculated by first-principles atomistic thermodynamics (see

288: text). The dashed line indicates the thermodynamic stability range

289: of the $p(2 \times 2)$ adphase, if formation of the surface oxide

290: was kinetically inhibited.}

291: \end{figure}

292: 

293: Having thus established the means to distinguish and characterize

294: the formation of either surface or thicker oxide films at higher O

295: partial pressures from the diffraction signals, we may in a

296: straightforward way construct a diagram showing which phase we

297: measure under which $(T,p)$-conditions. The result is shown in

298: Fig. \ref{fig3}, covering the whole range of now experimentally

299: accessible gas pressures from 10${}^{-6}$\,mbar to ambient

300: pressure. Most strikingly, the $(\sqrt{5} \times

301: \sqrt{5})R27^{\circ}$ structure is found under a wide variety of

302: conditions -- even at an oxygen pressure of 10${}^3$\,mbar and a

303: sample temperature of 575\,K we still observe only the formation

304: of this surface oxide and no indications for the growth of a

305: thicker oxide film on the time scale of 1\,hr currently accessible

306: to the experiment.

307: 

308: As a first step to understand this data we proceed by comparing

309: it with the surface phase diagram obtained by DFT and atomistic

310: thermodynamics. Figure \ref{fig3}b gives the corresponding

311: $(T,p)$-diagram, showing which phase would be most stable on the

312: basis of our DFT data, if the surface were in full thermodynamic

313: equilibrium with the surrounding oxygen gas phase. We immediately stress

314: that although state-of-the-art, some approximations like e.g. the

315: present treatment of vibrational contributions to the free energies,

316: as well as the uncertainty introduced by finite basis set and the

317: employed exchange-correlation functional may well allow for errors

318: in the phase boundaries of the order of $\pm 100$\,K and (depending on

319: temperature) of up to several orders in pressure \cite{reuter02,li03}.

320: 

321: Taking this into account we notice a gratifying overall agreement

322: between theory and experiment in this wide range of environmental

323: conditions. At a closer look there is, however, a notable

324: difference that is beyond the uncertainties underlying both the

325: experimental and theoretical approach: the experimental

326: observation of the $(\sqrt{5} \times \sqrt{5})R27^{\circ}$ surface

327: oxide in the top left corner of the drawn diagram, i.e. at lower

328: temperatures and high pressures. Since the central assumption of

329: theory, which predicts the stability of the bulk oxide under such

330: conditions, is the full thermodynamic equilibrium between surface

331: and gas phase, we interpret this difference as reflecting kinetic

332: limitations to the growth of the bulk oxide under such conditions.

333: This is also apparent even within the experimental data set alone.

334: If the surface was fully equilibrated with the environment, the

335: evaluated phase boundaries would have to follow lines of constant

336: oxygen chemical potential, which is the single determining

337: quantity if thermodynamics applies. In the drawn $(T,p)$-plots

338: such lines of constant chemical potential would always be parallel

339: to the phase boundaries as drawn in the theoretical diagram, cf.

340: Fig. \ref{fig3}b, which the bulk/surface oxide boundary drawn in

341: Fig. \ref{fig3}a (even considering all uncertainties) is clearly

342: not. Similarly we also understand the experimental observation of

343: the $p(2 \times 2)$ adsorbate phase, which according to theory is

344: never a thermodynamically stable phase, as a sign of kinetic

345: hindrance to the formation of the $(\sqrt{5} \times

346: \sqrt{5})R27^{\circ}$ phase. Correspondingly, we have also marked

347: in the theoretical plot the area, where the $p(2 \times 2)$ would

348: turn out more stable than the clean surface, if the surface oxide

349: can not form.

350: 

351: \begin{figure}[t!]

352: \scalebox{0.29}{\includegraphics{fig4.ps}}

353: \caption{\label{fig4} Six consecutive L-scans along (0.6 0.8 L) at

354: 675\,K and 50\,mbar. Each scan took 240 sec to record, showing the

355: gradual transformation from the diffraction signal characteristic

356: of the surface oxide to the one of the bulk oxide, cf. Fig.

357: \ref{fig2}. Solid circles: first spectrum; empty circles: last

358: spectrum taken.}

359: \end{figure}

360: 

361: Under suitable conditions we can even follow the kinetically

362: limited transition from surface to bulk oxide within the time

363: resolution of our current experiment. This is illustrated in Fig.

364: \ref{fig4} showing how the diffraction signal at 675\,K and $p =

365: 50$\,mbar slowly transforms at every consecutive scan from the one

366: characteristic for the surface oxide phase, cf. Fig. \ref{fig2}a,

367: to the one of the bulk oxide film, cf. Fig. \ref{fig2}b.

368: Obviously, this transformation will be faster for higher

369: temperatures, so that we immediately observe the bulk oxide, or

370: eventually so slow at lower temperatures that we can no longer

371: measure the transition within the time frame open to our

372: experiment. In passing we finally note that these clear kinetic

373: limitations at time scales of 1\,hr even at technologically

374: relevant temperatures as high as 675\,K, are somewhat at variance

375: with the well-known theoretical notion by King and coworkers

376: suggesting that oxide growth should immediately set in as soon as

377: it is thermodynamically possible \cite{carlisle00b}.

378: 

379: 

380: 

381: In conclusion we have studied the oxidation of the Pd(100) surface

382: from $10^{-6}$\,mbar to ambient pressure by {\em in-situ} SXRD.

383: Depending on the environmental conditions we observe either the

384: formation of the $(\sqrt{5} \times \sqrt{5})R27^{\circ}$ surface

385: oxide (essentially a well-ordered layer of PdO(101)), or the

386: growth of  $\sim 40$\,{\AA} poorly ordered and rough PdO bulk

387: oxide, predominantly with PdO(001) orientation. The range of

388: $(T,p)$-conditions where we measure the surface oxide is

389: surprisingly large, and goes for $T < 600$\,K even up to ambient

390: pressures. Comparing with the theoretical surface phase diagram

391: from first-principles atomistic thermodynamics we interpret this

392: as reflecting a kinetic hindrance to the formation of the bulk

393: oxide, which is clearly an activated process due to the involved

394: massive restructuring at the surface. Such kinetic limitations in

395: the initial oxide formation process have hitherto barely been

396: addressed, but could be crucial for understanding high pressure

397: applications like e.g. oxidation catalysis. Our measurements

398: demonstrate the usefulness of SXRD for the study of the oxidation

399: process of almost any material in such high pressure environments,

400: providing the aspired atomically resolved structural information

401: of any thin film or nano-based structure exposing its surface to

402: an ambient working atmosphere.

403: 

404: This work was financially supported by the

405: Swedish Natural Science Council and the DFG-priority

406: program ``Realkatalyse''.

407: 

408: \begin{references}

409: 

410: \bibitem{over00}

411: H. Over {\em et al.}, Science {\bf 287}, 1474 (2000).

412: 

413: \bibitem{hendriksen02}

414: B.L.M. Hendriksen and J.W.M. Frenken, Phys. Rev. Lett. {\bf 89},

415: 046101 (2002).

416: 

417: \bibitem{stampfl02}

418: C. Stampfl {\em et al.}, Surf. Sci. {\bf 500}, 368 (2002).

419: 

420: \bibitem{carlisle00}

421: C.I. Carlisle {\em et al.}, Phys. Rev. Lett. {\bf 84}, 3899

422: (2000).

423: 

424: \bibitem{lundgren02}

425: E. Lundgren {\em et. al.}, Phys. Rev. Lett. {\bf 88}, 246103

426: (2002).

427: 

428: \bibitem{todorova03}

429: M. Todorova {\em et al.}, Surf. Sci. {\bf 541},101 (2003).

430: 

431: \bibitem{reuter02}

432: K. Reuter and M. Scheffler, Phys. Rev. B {\bf 65}, 035406 (2002).

433: 

434: \bibitem{li03}

435: W.-X. Li, C. Stampfl and M. Scheffler, Phys. Rev. Lett. {\bf 90}, 256102 (2003).

436: 

437: \bibitem{ziauddin97}

438: M. Ziauddin, G. Veser and L. D. Schmidt, Catal. Lett. {\bf 46},

439: 159 (1997).

440: 

441: \bibitem{veser99}

442: G. Veser, A. Wright, and R. Caretta, Catal. Lett {\bf 58}, 199

443: (1999).

444: 

445: \bibitem{beamline}

446: http://hikwww1.fzk.de/iss/beamlinebook.html

447: 

448: \bibitem{blaha99}

449: P. Blaha, K. Schwarz and J. Luitz, {\bf WIEN97}, Techn.

450: Universit\"at Wien, Austria, (1999). ISBN 3-9501031-0-4.

451: 

452: \bibitem{perdew96}

453: J.P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett. {\bf 77},

454: 3865 (1996).

455: 

456: \bibitem{basis}

457: FP-LAPW basis set parameters: $R_{\rm{MT}}^{\rm{Pd}}=$1.8\,bohr,

458: $R_{\rm{MT}}^{\rm{O}}=$1.3\,bohr, $l_{\rm{max}}^{\rm{wf}} = 12$,

459: $l_{\rm{max}}^{\rm{pot}} = 4$, $E^{\rm max}_{\rm wf} = 20$\,Ry,

460: $E^{\rm max}_{\rm pot} = 169$\,Ry, and a $(4 \times 4 \times 1)$

461: Monkhorst-Pack grid with 28 {\bf k}-points in the full Brillouin

462: zone.

463: 

464: \bibitem{zheng00}

465: G. Zheng and E.I. Altman, Surf. Sci. {\bf 462}, 151 (2000).

466: 

467: \bibitem{rogers71}

468: D. Rogers, R. Shannon and J. Gillson, J. Solid State Chem. {\bf 3}, 314 (1971).

469: 

470: \bibitem{mcbride91}

471: J. McBride, K. Hass and W. Weber, Phys. Rev. B {\bf 44}, 5016 (1991).

472: 

473: \bibitem{hendriksen03}

474: B.L.M. Hendriksen {\em et al.}, Surf. Sci. ({\em submitted}).

475: %, {\em Model Catalysts in Action: High-Pressure Scanning

476: %Tunneling Microscopy}, Ph.D. thesis, Universiteit Leiden (2003).

477: 

478: \bibitem{robinson88}

479: I.K. Robinson, Phys. Rev. B {\bf 33}, 3830 (1986).

480: 

481: \bibitem{carlisle00b}

482: C.I. Carlisle {\em et al.}, Surf. Sci. {\bf 470}, 15 (2000), and

483: references therein.

484: 

485: 

486: \end{references}

487: \end{document}

488: