- Email: [email protected]
Surface structure and relaxation during the oxidation of carbon monoxide on Pt–Pd bimetallic surfaces
Surface Science 479 (2001) 241±246
www.elsevier.nl/locate/susc
Surface structure and relaxation during the oxidation of carbon monoxide on Pt±Pd bimetallic surfaces C.A. Lucas a,*, N.M. Markovic b, M. Ball a, V. Stamenkovic b, V. Climent b, P.N. Ross b a
b
Oliver Lodge Laboratory, Department of Physics, University of Liverpool, Liverpool, L69 7ZE, UK Materials Sciences Division, Lawrence Berkeley National Laboratory, University of California, Berkeley, CA 94720, USA Received 10 January 2001; accepted for publication 16 February 2001
Abstract The atomic structure and surface relaxation of Pd monolayer on Pt(1 1 1) has been studied by surface X-ray scattering, in an aqueous environment under electrostatic potential control, during the adsorption and oxidation of carbon monoxide. The results show that the Pd±Pt layer spacing contracts at the onset of CO oxidation before the Pd adlayer forms an oxide structure that is incommensurate with the Pt lattice. Both the oxide formation and the lattice contraction are fully reversible over many cycles of the applied electrode potential. Ó 2001 Elsevier Science B.V. All rights reserved. Keywords: Surface structure, morphology, roughness, and topography; Carbon monoxide; Surface relaxation and reconstruction; Platinum; Palladium; X-ray scattering, diraction, and re¯ection; Solid±liquid interfaces; Electrochemical methods
Thin transition metal ®lms supported on single crystal metal surfaces have been studied in great detail in order to establish a link between the physical properties of a surface and its chemical reactivity [1±3]. These results are particularly relevant to the ®eld of electrocatalysis where changes in catalytic activity can be attributed to ensemble eects (or morphological eects), structure eects due to changes in local bond geometry and electronic (or ligand) eects where a metal's reactivity is modi®ed directly [4]. The experimental work has been supported and motivated by theoretical cal-
* Corresponding author. Tel.: +44-151-7943361; fax: +44151-793441. E-mail address: [email protected] (C.A. Lucas).
culations in which, for example, a strong correlation was found between molecular chemisorption energies and the location of the d-band center of the surface metal atoms [5±7]. The experimental, ultra-high-vacuum (UHV) work has also stimulated progress in the ®eld of surface electrochemistry, where similar bimetallic systems can be studied in an aqueous environment under electrostatic potential control. These systems have the advantage that they can be studied under conditions of chemical equilibrium and that catalytic reactions can be monitored by traditional voltammetric methods. By using techniques which can probe the surface structure in situ, such as surface X-ray scattering (SXS) and scanning tunneling microscopy (STM), catalytic reactions can also be probed on the atomic level. Furthermore,
0039-6028/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 3 9 - 6 0 2 8 ( 0 1 ) 0 0 9 8 4 - 0
242
C.A. Lucas et al. / Surface Science 479 (2001) 241±246
recent work on platinum electrode surfaces has shown that eects such as surface relaxation induced by the adsorption of hydrogen and CO [8] are in remarkable agreement with theoretical calculations of the ideal adsorption system [9], i.e. a 2% outward relaxation induced by the adsorption of a monolayer of hydrogen. The adsorption and oxidation of carbon monoxide on transition metal single crystal surfaces has been a prototypical system with which to study molecular adsorption and catalytic reactions. In a recent series of publications, we have shown that CO forms a densely packed p(22) structure on the Pt(1 1 1) surface that undergoes an order±disorder transition induced by partial oxidation of the CO adlayer that is in a Ôweakly-bondedÕ adsorption state [8,10]. In the presence of the p(2 2) structure the Pt(1 1 1) surface was expanded by 4% of the bulk atomic spacing, presumably as a result of the charge transfer in the back-bonding mechanism. It was possible to follow the CO oxidation reaction indirectly by monitoring the Pt surface expansion as a function of the applied electrode potential. In this letter we extend these results to a bimetallic system, namely a monolayer of Pd on the Pt(1 1 1) electrode surface. Given that Pt and Pd are neighboring elements in the periodic table and that their lattice constants dier by <1%, a monolayer of Pd on Pt(1 1 1) represents a catalytic system in which only a ÔpureÕ electronic eect is expected. By utilizing SXS, in combination with voltammetric experiments to provide adsorption isotherms, it is possible to build up a detailed picture of the surface atomic geometry during the CO-oxidation process. Thin Pd ®lms on Pt substrates can be produced by deposition in UHV [11] or by electrochemical deposition [12±14]. In this work the latter method was used. Brie¯y, before each experiment the Pt(1 1 1) electrode was ¯ame annealed and cooled down in a H2 Ar atmosphere. After that, the surface was protected with a droplet of ultrapure water and mounted into the disk position of an insertable rotating ring-disk electrode assembly (Pine Instruments). The Pd ®lm was deposited from a solution containing 10 5 M PdO in 0.05 M H2 SO4 as the potential was swept at 20 mV/s. The amount deposited was controlled by the continu-
ous change of the voltammetric features characteristic of a pseudomorphic monolayer ®lm of palladium, for details see Ref. [15]. Finally, the Pdmodi®ed Pt(1 1 1) electrode was rinsed with water and transferred either to a second electrochemical cell or into the X-ray cell containing a solution free of Pd2 . The X-ray experiments were performed on beamline 7-2 at the Stanford Synchrotron Radiation Laboratory (SSRL) using a monochromatic incident X-ray beam of energy 10 keV de®ned by slits to be a 1 1 mm2 spot at the sample. In order to study the adsorption of CO, the outer shell of the X-ray electrochemical cell was purged with CO (99.99% purity), instead of nitrogen which is usually used to prevent interaction of the surface with the atmosphere. The CO is able to diuse through the polypropylene ®lm that traps the electrolyte and, therefore, saturate the solution. The reciprocal space notation is for the conventional hexagonal unit cell for the (1 1 1) surface such that the surface normal is along the 0; 0; lhex direction and the h; 0; 0hex and 0; k; 0hex vectors lie in the plane of the surface and subtend 60°. Full experimental details for the SXS and rotating disk electrode (RDE) experiments are given in Ref. [16]. All potentials are quoted versus the reversible hydrogen electrode to an accuracy of 0.05 V. A precise description of the Pt(1 1 1)±Pd surface structure can be obtained by measuring and modeling crystal truncation rods (CTRs) [17]. This is particularly true for Pd/Pt as Pd forms a pseudomorphic overlayer, i.e. it is fully commensurate with the Pt lattice [18]. Fig. 1 shows the specular 0; 0; l and ®rst-order, non-specular 1; 0; l and 0; 1; l CTRs measured at 0.05 V and after the solution had been saturated with CO. The calculated CTRs for an ideally terminated Pt(1 1 1) surface are shown by the dashed curves and this is close to what is measured for the clean Pt(1 1 1) surface in electrolyte [8]. The data in Fig. 1 dier signi®cantly from the model calculations at the Ôanti-BraggÕ positions, midway between the Bragg re¯ections. At these positions, the scattering from the bulk Pt lattice is eectively canceled and the scattered intensity is due to the topmost Pt atomic layer and the Pd overlayer. Due to the dierence in atomic form factor for Pt and Pd this
C.A. Lucas et al. / Surface Science 479 (2001) 241±246
Fig. 1. The specular and ®rst-order non-specular CTRs for the Pd/Pt(1 1 1) surface measured at an electrode potential of 0.05 V in CO-saturated solution. The dashed lines are calculated for an ideally terminated Pt(1 1 1) surface. The solid lines are a ®t according to the structural model described in the text. The indicated positions (a,b,c) are those at which the data shown in Fig. 3 were measured.
results in a signi®cant decrease in intensity compared to the ideal Pt surface calculation [19]. The solid lines in Fig. 1 are a calculated best ®t to the data using a structural model in which the Pd coverage (hPd ), Pt±Pd distance (dPt±Pd ), relaxation of the Pt surface (ePt ) and surface roughness parameters (rPd , rPt ) were varied. The Pd atoms were located at Pt three-fold hollow sites to continue the ABC stacking of the bulk lattice. In agreement with the previous study of the Pt(1 1 1)±CO system [8,10], the CTR data was insensitive to the
243
inclusion of CO in the structural model due to its relatively low scattering power. The structural parameters obtained are hPd 1:0 0:1 mono expansion of layers (ML), dPt±Pd 2:32 0:02 A, the Pt lattice ePt 0:02 0:02 A and static Debye± and Waller-type roughness, rPd 0:12 0:04 A rPt 0:09 0:04 A. Including a second Pd layer in the structural model did not improve the ®t to the data. The analysis of the CTRs indicates that the adsorption of CO on the Pd/Pt surface causes no changes in the atomic structure or surface relaxation [18]. This result was indicated during the experiment when CO was added to solution whilst monitoring the X-ray intensity at a particular reciprocal lattice point and no changes were observed. In contrast, for the Pt(1 1 1) surface, saturation of the solution with CO led to a change in asymmetry around the bulk Bragg re¯ections; for example an increase of intensity at (1,0,3.6) and a corresponding decrease at (1,0,4.4) [8]. These changes were consistent with an increase in the surface expansion of the topmost Pt atomic layer from 2% of the bulk lattice spacing (in the presence of adsorbed hydrogen) to 4% of the spacing (in the presence of adsorbed CO). For the Pd/Pt(1 1 1) surface, however, it appears that CO adsorption has no eect on the lattice expansion at the surface. Compared to the bulk Pt±Pt spacing in the (1 1 1) it can be stated that the Pt±Pd direction (2.266 A), spacing is expanded by 2.3% both in the presence of adsorbed hydrogen and adsorbed CO. For the Pt(1 1 1)/CO system previous measurements have shown that a p(2 2)-CO structure is formed at 0.0 V with a coverage of 0.75 CO molecules per surface Pt atom [8,20,21]. We have performed a detailed SXS study of this structure which is observed both in pure acid solutions and in the presence of other metallic species where displacement of the metal from the Pt surface occurs to allow formation of the close-packed CO adlayer [8,10,16]. For the Pd/Pt(1 1 1) system, a careful reciprocal space search was made in an attempt to locate scattering from an ordered CO adlayer. This consisted of comparative scans (i.e. both with and without CO present in solution) along high symmetry surface reciprocal lattice directions and at positions where scattering would
244
C.A. Lucas et al. / Surface Science 479 (2001) 241±246
arise if structures were present that has previously been p observed p on Pt(1 1 1) p or Pd(1 p 1 1) in UHV, e.g. ( p19 p 19)R23.4°, ( 7 7)R19.1°, c(42) and ( 3 3)R30° structures [20±24]. No scattering due to the CO adlayer was observed and we conclude that either the CO is disordered or forms an incommensurate structure. In the absence of SXS data regarding the CO adlayer, information regarding the surface coverage by CO and the oxidation of CO can be obtained from RDE measurements (as described in Ref. [25]). Fig. 2 compares the polarization curves obtained for the Pt(1 1 1) surface (solid lines) and Pd/Pt(1 1 1) surface (dashed lines) as the potential
Fig. 2. Polarization curves for the unmodi®ed Pt(1 1 1) surface (dashed lines) and the Pd/Pt(1 1 1) surface (solid lines) measured in CO-saturated solution. (inset) The CO stripping voltammetry measured after CO was adsorbed on the surface at 0.05 V and the solution was purged with nitrogen to remove CO present in solution before the potential sweep was performed. The area under the peak in each case is proportional to the CO coverage.1
was cycled from 0.0 to 1.1 V in CO-saturated solution. The inset in Fig. 2 shows the CO stripping voltammetry, i.e. recorded after the solution was saturated with CO at 0.0 V and then purged with nitrogen to remove the solution CO. This latter measurement gives the CO surface coverage by integration of the charge under the CO stripping peak. Three main results can be deduced from these experiments: (i) The CO coverage on the Pd/Pt surface is identical to that observed on the Pt surface. In both cases integration of the stripping curves gives a total charge of 400 lC/cm2 (after correction for other processes 1) which is equivalent to 0.65 CO molecules per surface Pt (or Pd) atom. (ii) The electrooxidation of CO is shifted anodically on the Pd/Pt(1 1 1) surface by 0.1 V, either by inhibition of OH adsorption on the Pd/Pt surface caused by stronger bonding of Pd to sulfate molecules or due to the Pd±CO bond being stronger than that of Pt±CO. If the Pd±CO bond is stronger than Pt±CO then it is surprising that no change in the Pt±Pd bondlength was observed upon the adsorption of CO. (iii) There is no signi®cant pre-oxidation peak on the Pd/Pt surface. The pre-oxidation peak has been attributed to oxidation of a Ôweakly-adsorbedÕ state of CO on Pt(1 1 1) the presence of which was crucial in the ordering of the p(22)-CO adlayer [8,10]. The absence of the pre-oxidation wave may be due to the blocking of active Pt sites for OH adsorption by strongly-adsorbed bisulfate anions. Potentiodynamic SXS measurements (or X-ray voltammetry) give considerable insight into the behavior of the Pd/Pt surface during the oxidation and adsorption of CO. Fig. 3 shows the results for three reciprocal lattice positions as the electrode potential was cycled over the range 0.05±1.15 V in CO-saturated solution. The (1,0,3.6) and (1,0,4.4) positions are either side of the Pt(1,0,4) Bragg re¯ections and are principally sensitive to changes in the Pt±Pd spacing, provided that the Pd atoms
1 After correction for adsorption of bisulfate anions, adsorption of OH and charging of the double layer, see Ref. [7] for details.
C.A. Lucas et al. / Surface Science 479 (2001) 241±246
Fig. 3. The X-ray intensity measured at three reciprocal lattice positions (indicated in Fig. 1) as the potential was cycled over the range 0.05±1.15 V (sweep rate 2 mV/s). At each position the intensities are normalized to the intensity measured at 0.05 V. The vertical line marker denotes the two potential regions of interest (see text).
remain commensurate with the Pt lattice. Thus an increase in intensity at (1,0,3.6) and a corresponding decrease at (1,0,4.4) implies an expansion of the Pd±Pt surface and vice-versa. The (0,1,0.5) position is an Ôanti-BraggÕ point, midway between the (0,1, 1) and (0,1,2) Pt Bragg re¯ections and is primarily sensitive to the Pd coverage (as can be seen from the model calculations in Fig. 1). The data in Fig. 3 essentially show two structural eects which are separated in the ®gure by the vertical marker line at 0.85 V. Considering
245
only the anodic sweep, it is clear from the (0,1,0.5) data that no changes in the Pd surface coverage occur over the range 0.0±0.85 V, the upper limit being the potential at which CO oxidation occurs (see Fig. 2). The intensities at (1,0,3.6) and (1,0,4.4), however, show dramatic changes over this potential region consistent with a signi®cant contraction in the Pd±Pt layer spacing. Such changes are not observed in CO-free solution [18]. As the intensities shown are normalized to the scattered intensity measured at 0.05 V, the change in the Pt±Pd spacing necessary to reproduce the intensity ratios, R, (R 0:8 and R 1:5 for l 3:6 and l 4:4 respectively) can be calculated. For a the calculated ratios are Pd±Pt spacing of 2.25 A R 0:80 and R 1:42 which are in reasonable agreement with the experimental results. 2 It is apparent, therefore, that the CO oxidation reaction causes a large contraction ( 1% compared to the Pt±Pt bulk spacing) of the Pd±Pt surface. Given that the adsorbed CO molecules are reacting with OH species [presumably in a Langmuir± Hinshelwood reaction as observed for Pt(1 1 1)], it may be that the lattice contraction re¯ects the expected Pd±Pt bondlength in the absence of strong chemisorption on the surface. In fact this bondlength is in good agreement with a calculation of the Pt±Pd distance according to a hard sphere model using the atomic radii of Pt and Pd. Previous calculations have shown excellent agreement with measurements of surface expansion induced by hydrogen adsorption on Pt(1 1 1) [9] and it would be an interesting test case if similar agreement could be found with the results for the Pd/Pt surface. The second structural eect illustrated in Fig. 3 occurs when the electrode potential is scanned positive of 0.85 V where continuous oxidation of CO is occurring. As can be seen from the data at (0,1,0.5), oxidation of CO is accompanied by simultaneous rearrangement of the adsorbed Pd monolayer which is manifested by a large increase
2 It is dicult to stabilize the intensity changes by holding at a ®xed electrode potential as the oxidation of CO is a dynamic process which can be in¯uenced by changes in the local solution-CO concentration.
246
C.A. Lucas et al. / Surface Science 479 (2001) 241±246
in the scattered intensity, presumably due to a reduction in the number of Pd atoms that reside in commensurate Pt sites. This structural change also aects the intensities at (1,0,3.6) and (1,0,4.4) where the changes can no longer be understood in terms of the simple surface contraction/expansion model. These results imply that CO oxidation is accompanied by oxidation of the Pd monolayer, although the process is fully reversible over the measured potential range. This is indicated by the cathodic sweeps in Fig. 3 which mirror the anodic sweeps with some hysteresis and also by the fact that the data in Fig. 3 were unaected by the number of potential cycles performed. No de®nitive model for the surface structure at 1.15 V is currently available. Specular CTR data could be modeled by the inclusion of a Pd±O bilayer structure and evidence for a disordered PdO adlayer was also given by the appearance of diuse rings of scattering [26]. Regardless of the exact nature of the Pd±O structure, however, the key result is that the reaction is fully reversible and the Pd monolayer is recovered at 0.05 V. In summary, we have shown that a model catalytic system in which the modi®cations in the catalytic activity can be attributed to a ÔpureÕ electronic eect can be probed, in situ, by SXS and voltammetric techniques to correlate changes in the atomic structure and surface relaxation with the surface chemical reactions. Further experimental and theoretical work, to develop models which include surface relaxation eects, is anticipated. Acknowledgements We would like to thank the sta and user administration at SSRL for their continued support in providing an excellent synchrotron radiation facility. This work was supported by the Director, Oce of Energy Research, Materials Sciences Division (MSD) of the US Department of Energy (DOE) under contract no. DE-AC03-76SF00098. Research was carried out in part at SSRL, which is funded by the Division of Chemical Sciences (DCS), US DOE. CAL and MB were partially supported by the University of Liverpool and by
NATO grant reference CRG 973107. CAL acknowledges the support of an EPSRC Advanced Research Fellowship.
References [1] J.A. Rodriguez, D.W. Goodman, Science 257 (1992) 897. [2] J.A. Rodriguez, D.W. Goodman, J. Phys. Chem. 95 (1991) 4196. [3] R.A. Campbell, J.A. Rodriguez, D.W. Goodman, Phys. Rev. B 46 (1992) 7077. [4] B. Hammer, J.K. Norskov, in: R.M. Lambert, G. Pacchioni (Eds.), Chemisorption and Reactivity on Supported Clusters and Thin Films, Kluwer Academic Publishers, Dordrecht, 1997, p. 285. [5] J.A. Rodriguez, M. Kuhn, J. Phys. Chem. 98 (1994) 11251. [6] M. Mavrikakis, B. Hammer, J.K. Norskov, Phys. Rev. Lett. 81 (1998) 2819. [7] V. Pallassana, M. Neurock, L.B. Hansen, B. Hammer, J.K. Norskov, Phys. Rev. B 60 (1999) 6146. [8] C.A. Lucas, N.M. Markovic, P.N. Ross, Surf. Sci. Lett. 425 (1999) L381. [9] P.J. Feibelman, Phys. Rev. B 56 (1997) 2175. [10] N.M. Markovic, B.N. Grgur, C.A. Lucas, P.N. Ross, J. Phys. Chem. 103 (1999) 487. [11] G.A. Attard, A. Bannister, J. Electroanal. Chem. 300 (1991) 467. [12] J. Clavilier, M.J. Llorca, J.M. Feliu, A. Aldaz, J. Electroanal. Chem. 310 (1991) 429. [13] M. Baldauf, D.M. Kolb, J. Phys. Chem. 100 (1996) 11375. [14] V. Climent, N.M. Markovic, P.N. Ross, J. Phys. Chem. 104 (2000) 3116. [15] M.P. Soriaga, Y.-G. Kim, J.E. Soto, in: A. Wieckowski (Ed.), Interfacial Electrochemistry; Theory, Experiment and Applications, Marcel Dekker, New York, 1999 (Chapter 15). [16] C.A. Lucas, N.M. Markovic, P.N. Ross, Surf. Sci. 448 (2000) 77. [17] I.K. Robinson, D.J. Tweet, Rep. Prog. Phys. 55 (1992) 599. [18] N.M. Markovic, C.A. Lucas, V. Climent, V. Stamenkovic, P.N. Ross, Surf. Sci. 465 (2000) 103. [19] M. Takahasi, Y. Hayashi, J. Mizuki, K. Tamura, T. Kondo, H. Naohara, K. Uosaki, Surf. Sci. 461 (2000) 213. [20] I. Villegas, M.J. Weaver, J. Chem. Phys. 101 (1994) 1648. [21] M.W. Severson, C. Stuhlmann, I. Villegas, M.J. Weaver, J. Chem. Phys. 103 (1995) 9832. [22] G. Ertl, M. Neumann, K. Streit, Surf. Sci. 64 (1977) 393. [23] H. Steininger, S. Lehwald, H. Ibach, Surf. Sci. 123 (1982) 264. [24] D.F. Ogletree, M.A. Van Hove, G.A. Somorjai, Surf. Sci. 173 (1986) 351. [25] N.M. Markovic, H.A. Gasteiger, P.N. Ross, Langmuir 11 (1995) 4098. [26] C.A. Lucas, N.M. Markovic, unpublished.
www.elsevier.nl/locate/susc
Surface structure and relaxation during the oxidation of carbon monoxide on Pt±Pd bimetallic surfaces C.A. Lucas a,*, N.M. Markovic b, M. Ball a, V. Stamenkovic b, V. Climent b, P.N. Ross b a
b
Oliver Lodge Laboratory, Department of Physics, University of Liverpool, Liverpool, L69 7ZE, UK Materials Sciences Division, Lawrence Berkeley National Laboratory, University of California, Berkeley, CA 94720, USA Received 10 January 2001; accepted for publication 16 February 2001
Abstract The atomic structure and surface relaxation of Pd monolayer on Pt(1 1 1) has been studied by surface X-ray scattering, in an aqueous environment under electrostatic potential control, during the adsorption and oxidation of carbon monoxide. The results show that the Pd±Pt layer spacing contracts at the onset of CO oxidation before the Pd adlayer forms an oxide structure that is incommensurate with the Pt lattice. Both the oxide formation and the lattice contraction are fully reversible over many cycles of the applied electrode potential. Ó 2001 Elsevier Science B.V. All rights reserved. Keywords: Surface structure, morphology, roughness, and topography; Carbon monoxide; Surface relaxation and reconstruction; Platinum; Palladium; X-ray scattering, diraction, and re¯ection; Solid±liquid interfaces; Electrochemical methods
Thin transition metal ®lms supported on single crystal metal surfaces have been studied in great detail in order to establish a link between the physical properties of a surface and its chemical reactivity [1±3]. These results are particularly relevant to the ®eld of electrocatalysis where changes in catalytic activity can be attributed to ensemble eects (or morphological eects), structure eects due to changes in local bond geometry and electronic (or ligand) eects where a metal's reactivity is modi®ed directly [4]. The experimental work has been supported and motivated by theoretical cal-
* Corresponding author. Tel.: +44-151-7943361; fax: +44151-793441. E-mail address: [email protected] (C.A. Lucas).
culations in which, for example, a strong correlation was found between molecular chemisorption energies and the location of the d-band center of the surface metal atoms [5±7]. The experimental, ultra-high-vacuum (UHV) work has also stimulated progress in the ®eld of surface electrochemistry, where similar bimetallic systems can be studied in an aqueous environment under electrostatic potential control. These systems have the advantage that they can be studied under conditions of chemical equilibrium and that catalytic reactions can be monitored by traditional voltammetric methods. By using techniques which can probe the surface structure in situ, such as surface X-ray scattering (SXS) and scanning tunneling microscopy (STM), catalytic reactions can also be probed on the atomic level. Furthermore,
0039-6028/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 3 9 - 6 0 2 8 ( 0 1 ) 0 0 9 8 4 - 0
242
C.A. Lucas et al. / Surface Science 479 (2001) 241±246
recent work on platinum electrode surfaces has shown that eects such as surface relaxation induced by the adsorption of hydrogen and CO [8] are in remarkable agreement with theoretical calculations of the ideal adsorption system [9], i.e. a 2% outward relaxation induced by the adsorption of a monolayer of hydrogen. The adsorption and oxidation of carbon monoxide on transition metal single crystal surfaces has been a prototypical system with which to study molecular adsorption and catalytic reactions. In a recent series of publications, we have shown that CO forms a densely packed p(22) structure on the Pt(1 1 1) surface that undergoes an order±disorder transition induced by partial oxidation of the CO adlayer that is in a Ôweakly-bondedÕ adsorption state [8,10]. In the presence of the p(2 2) structure the Pt(1 1 1) surface was expanded by 4% of the bulk atomic spacing, presumably as a result of the charge transfer in the back-bonding mechanism. It was possible to follow the CO oxidation reaction indirectly by monitoring the Pt surface expansion as a function of the applied electrode potential. In this letter we extend these results to a bimetallic system, namely a monolayer of Pd on the Pt(1 1 1) electrode surface. Given that Pt and Pd are neighboring elements in the periodic table and that their lattice constants dier by <1%, a monolayer of Pd on Pt(1 1 1) represents a catalytic system in which only a ÔpureÕ electronic eect is expected. By utilizing SXS, in combination with voltammetric experiments to provide adsorption isotherms, it is possible to build up a detailed picture of the surface atomic geometry during the CO-oxidation process. Thin Pd ®lms on Pt substrates can be produced by deposition in UHV [11] or by electrochemical deposition [12±14]. In this work the latter method was used. Brie¯y, before each experiment the Pt(1 1 1) electrode was ¯ame annealed and cooled down in a H2 Ar atmosphere. After that, the surface was protected with a droplet of ultrapure water and mounted into the disk position of an insertable rotating ring-disk electrode assembly (Pine Instruments). The Pd ®lm was deposited from a solution containing 10 5 M PdO in 0.05 M H2 SO4 as the potential was swept at 20 mV/s. The amount deposited was controlled by the continu-
ous change of the voltammetric features characteristic of a pseudomorphic monolayer ®lm of palladium, for details see Ref. [15]. Finally, the Pdmodi®ed Pt(1 1 1) electrode was rinsed with water and transferred either to a second electrochemical cell or into the X-ray cell containing a solution free of Pd2 . The X-ray experiments were performed on beamline 7-2 at the Stanford Synchrotron Radiation Laboratory (SSRL) using a monochromatic incident X-ray beam of energy 10 keV de®ned by slits to be a 1 1 mm2 spot at the sample. In order to study the adsorption of CO, the outer shell of the X-ray electrochemical cell was purged with CO (99.99% purity), instead of nitrogen which is usually used to prevent interaction of the surface with the atmosphere. The CO is able to diuse through the polypropylene ®lm that traps the electrolyte and, therefore, saturate the solution. The reciprocal space notation is for the conventional hexagonal unit cell for the (1 1 1) surface such that the surface normal is along the 0; 0; lhex direction and the h; 0; 0hex and 0; k; 0hex vectors lie in the plane of the surface and subtend 60°. Full experimental details for the SXS and rotating disk electrode (RDE) experiments are given in Ref. [16]. All potentials are quoted versus the reversible hydrogen electrode to an accuracy of 0.05 V. A precise description of the Pt(1 1 1)±Pd surface structure can be obtained by measuring and modeling crystal truncation rods (CTRs) [17]. This is particularly true for Pd/Pt as Pd forms a pseudomorphic overlayer, i.e. it is fully commensurate with the Pt lattice [18]. Fig. 1 shows the specular 0; 0; l and ®rst-order, non-specular 1; 0; l and 0; 1; l CTRs measured at 0.05 V and after the solution had been saturated with CO. The calculated CTRs for an ideally terminated Pt(1 1 1) surface are shown by the dashed curves and this is close to what is measured for the clean Pt(1 1 1) surface in electrolyte [8]. The data in Fig. 1 dier signi®cantly from the model calculations at the Ôanti-BraggÕ positions, midway between the Bragg re¯ections. At these positions, the scattering from the bulk Pt lattice is eectively canceled and the scattered intensity is due to the topmost Pt atomic layer and the Pd overlayer. Due to the dierence in atomic form factor for Pt and Pd this
C.A. Lucas et al. / Surface Science 479 (2001) 241±246
Fig. 1. The specular and ®rst-order non-specular CTRs for the Pd/Pt(1 1 1) surface measured at an electrode potential of 0.05 V in CO-saturated solution. The dashed lines are calculated for an ideally terminated Pt(1 1 1) surface. The solid lines are a ®t according to the structural model described in the text. The indicated positions (a,b,c) are those at which the data shown in Fig. 3 were measured.
results in a signi®cant decrease in intensity compared to the ideal Pt surface calculation [19]. The solid lines in Fig. 1 are a calculated best ®t to the data using a structural model in which the Pd coverage (hPd ), Pt±Pd distance (dPt±Pd ), relaxation of the Pt surface (ePt ) and surface roughness parameters (rPd , rPt ) were varied. The Pd atoms were located at Pt three-fold hollow sites to continue the ABC stacking of the bulk lattice. In agreement with the previous study of the Pt(1 1 1)±CO system [8,10], the CTR data was insensitive to the
243
inclusion of CO in the structural model due to its relatively low scattering power. The structural parameters obtained are hPd 1:0 0:1 mono expansion of layers (ML), dPt±Pd 2:32 0:02 A, the Pt lattice ePt 0:02 0:02 A and static Debye± and Waller-type roughness, rPd 0:12 0:04 A rPt 0:09 0:04 A. Including a second Pd layer in the structural model did not improve the ®t to the data. The analysis of the CTRs indicates that the adsorption of CO on the Pd/Pt surface causes no changes in the atomic structure or surface relaxation [18]. This result was indicated during the experiment when CO was added to solution whilst monitoring the X-ray intensity at a particular reciprocal lattice point and no changes were observed. In contrast, for the Pt(1 1 1) surface, saturation of the solution with CO led to a change in asymmetry around the bulk Bragg re¯ections; for example an increase of intensity at (1,0,3.6) and a corresponding decrease at (1,0,4.4) [8]. These changes were consistent with an increase in the surface expansion of the topmost Pt atomic layer from 2% of the bulk lattice spacing (in the presence of adsorbed hydrogen) to 4% of the spacing (in the presence of adsorbed CO). For the Pd/Pt(1 1 1) surface, however, it appears that CO adsorption has no eect on the lattice expansion at the surface. Compared to the bulk Pt±Pt spacing in the (1 1 1) it can be stated that the Pt±Pd direction (2.266 A), spacing is expanded by 2.3% both in the presence of adsorbed hydrogen and adsorbed CO. For the Pt(1 1 1)/CO system previous measurements have shown that a p(2 2)-CO structure is formed at 0.0 V with a coverage of 0.75 CO molecules per surface Pt atom [8,20,21]. We have performed a detailed SXS study of this structure which is observed both in pure acid solutions and in the presence of other metallic species where displacement of the metal from the Pt surface occurs to allow formation of the close-packed CO adlayer [8,10,16]. For the Pd/Pt(1 1 1) system, a careful reciprocal space search was made in an attempt to locate scattering from an ordered CO adlayer. This consisted of comparative scans (i.e. both with and without CO present in solution) along high symmetry surface reciprocal lattice directions and at positions where scattering would
244
C.A. Lucas et al. / Surface Science 479 (2001) 241±246
arise if structures were present that has previously been p observed p on Pt(1 1 1) p or Pd(1 p 1 1) in UHV, e.g. ( p19 p 19)R23.4°, ( 7 7)R19.1°, c(42) and ( 3 3)R30° structures [20±24]. No scattering due to the CO adlayer was observed and we conclude that either the CO is disordered or forms an incommensurate structure. In the absence of SXS data regarding the CO adlayer, information regarding the surface coverage by CO and the oxidation of CO can be obtained from RDE measurements (as described in Ref. [25]). Fig. 2 compares the polarization curves obtained for the Pt(1 1 1) surface (solid lines) and Pd/Pt(1 1 1) surface (dashed lines) as the potential
Fig. 2. Polarization curves for the unmodi®ed Pt(1 1 1) surface (dashed lines) and the Pd/Pt(1 1 1) surface (solid lines) measured in CO-saturated solution. (inset) The CO stripping voltammetry measured after CO was adsorbed on the surface at 0.05 V and the solution was purged with nitrogen to remove CO present in solution before the potential sweep was performed. The area under the peak in each case is proportional to the CO coverage.1
was cycled from 0.0 to 1.1 V in CO-saturated solution. The inset in Fig. 2 shows the CO stripping voltammetry, i.e. recorded after the solution was saturated with CO at 0.0 V and then purged with nitrogen to remove the solution CO. This latter measurement gives the CO surface coverage by integration of the charge under the CO stripping peak. Three main results can be deduced from these experiments: (i) The CO coverage on the Pd/Pt surface is identical to that observed on the Pt surface. In both cases integration of the stripping curves gives a total charge of 400 lC/cm2 (after correction for other processes 1) which is equivalent to 0.65 CO molecules per surface Pt (or Pd) atom. (ii) The electrooxidation of CO is shifted anodically on the Pd/Pt(1 1 1) surface by 0.1 V, either by inhibition of OH adsorption on the Pd/Pt surface caused by stronger bonding of Pd to sulfate molecules or due to the Pd±CO bond being stronger than that of Pt±CO. If the Pd±CO bond is stronger than Pt±CO then it is surprising that no change in the Pt±Pd bondlength was observed upon the adsorption of CO. (iii) There is no signi®cant pre-oxidation peak on the Pd/Pt surface. The pre-oxidation peak has been attributed to oxidation of a Ôweakly-adsorbedÕ state of CO on Pt(1 1 1) the presence of which was crucial in the ordering of the p(22)-CO adlayer [8,10]. The absence of the pre-oxidation wave may be due to the blocking of active Pt sites for OH adsorption by strongly-adsorbed bisulfate anions. Potentiodynamic SXS measurements (or X-ray voltammetry) give considerable insight into the behavior of the Pd/Pt surface during the oxidation and adsorption of CO. Fig. 3 shows the results for three reciprocal lattice positions as the electrode potential was cycled over the range 0.05±1.15 V in CO-saturated solution. The (1,0,3.6) and (1,0,4.4) positions are either side of the Pt(1,0,4) Bragg re¯ections and are principally sensitive to changes in the Pt±Pd spacing, provided that the Pd atoms
1 After correction for adsorption of bisulfate anions, adsorption of OH and charging of the double layer, see Ref. [7] for details.
C.A. Lucas et al. / Surface Science 479 (2001) 241±246
Fig. 3. The X-ray intensity measured at three reciprocal lattice positions (indicated in Fig. 1) as the potential was cycled over the range 0.05±1.15 V (sweep rate 2 mV/s). At each position the intensities are normalized to the intensity measured at 0.05 V. The vertical line marker denotes the two potential regions of interest (see text).
remain commensurate with the Pt lattice. Thus an increase in intensity at (1,0,3.6) and a corresponding decrease at (1,0,4.4) implies an expansion of the Pd±Pt surface and vice-versa. The (0,1,0.5) position is an Ôanti-BraggÕ point, midway between the (0,1, 1) and (0,1,2) Pt Bragg re¯ections and is primarily sensitive to the Pd coverage (as can be seen from the model calculations in Fig. 1). The data in Fig. 3 essentially show two structural eects which are separated in the ®gure by the vertical marker line at 0.85 V. Considering
245
only the anodic sweep, it is clear from the (0,1,0.5) data that no changes in the Pd surface coverage occur over the range 0.0±0.85 V, the upper limit being the potential at which CO oxidation occurs (see Fig. 2). The intensities at (1,0,3.6) and (1,0,4.4), however, show dramatic changes over this potential region consistent with a signi®cant contraction in the Pd±Pt layer spacing. Such changes are not observed in CO-free solution [18]. As the intensities shown are normalized to the scattered intensity measured at 0.05 V, the change in the Pt±Pd spacing necessary to reproduce the intensity ratios, R, (R 0:8 and R 1:5 for l 3:6 and l 4:4 respectively) can be calculated. For a the calculated ratios are Pd±Pt spacing of 2.25 A R 0:80 and R 1:42 which are in reasonable agreement with the experimental results. 2 It is apparent, therefore, that the CO oxidation reaction causes a large contraction ( 1% compared to the Pt±Pt bulk spacing) of the Pd±Pt surface. Given that the adsorbed CO molecules are reacting with OH species [presumably in a Langmuir± Hinshelwood reaction as observed for Pt(1 1 1)], it may be that the lattice contraction re¯ects the expected Pd±Pt bondlength in the absence of strong chemisorption on the surface. In fact this bondlength is in good agreement with a calculation of the Pt±Pd distance according to a hard sphere model using the atomic radii of Pt and Pd. Previous calculations have shown excellent agreement with measurements of surface expansion induced by hydrogen adsorption on Pt(1 1 1) [9] and it would be an interesting test case if similar agreement could be found with the results for the Pd/Pt surface. The second structural eect illustrated in Fig. 3 occurs when the electrode potential is scanned positive of 0.85 V where continuous oxidation of CO is occurring. As can be seen from the data at (0,1,0.5), oxidation of CO is accompanied by simultaneous rearrangement of the adsorbed Pd monolayer which is manifested by a large increase
2 It is dicult to stabilize the intensity changes by holding at a ®xed electrode potential as the oxidation of CO is a dynamic process which can be in¯uenced by changes in the local solution-CO concentration.
246
C.A. Lucas et al. / Surface Science 479 (2001) 241±246
in the scattered intensity, presumably due to a reduction in the number of Pd atoms that reside in commensurate Pt sites. This structural change also aects the intensities at (1,0,3.6) and (1,0,4.4) where the changes can no longer be understood in terms of the simple surface contraction/expansion model. These results imply that CO oxidation is accompanied by oxidation of the Pd monolayer, although the process is fully reversible over the measured potential range. This is indicated by the cathodic sweeps in Fig. 3 which mirror the anodic sweeps with some hysteresis and also by the fact that the data in Fig. 3 were unaected by the number of potential cycles performed. No de®nitive model for the surface structure at 1.15 V is currently available. Specular CTR data could be modeled by the inclusion of a Pd±O bilayer structure and evidence for a disordered PdO adlayer was also given by the appearance of diuse rings of scattering [26]. Regardless of the exact nature of the Pd±O structure, however, the key result is that the reaction is fully reversible and the Pd monolayer is recovered at 0.05 V. In summary, we have shown that a model catalytic system in which the modi®cations in the catalytic activity can be attributed to a ÔpureÕ electronic eect can be probed, in situ, by SXS and voltammetric techniques to correlate changes in the atomic structure and surface relaxation with the surface chemical reactions. Further experimental and theoretical work, to develop models which include surface relaxation eects, is anticipated. Acknowledgements We would like to thank the sta and user administration at SSRL for their continued support in providing an excellent synchrotron radiation facility. This work was supported by the Director, Oce of Energy Research, Materials Sciences Division (MSD) of the US Department of Energy (DOE) under contract no. DE-AC03-76SF00098. Research was carried out in part at SSRL, which is funded by the Division of Chemical Sciences (DCS), US DOE. CAL and MB were partially supported by the University of Liverpool and by
NATO grant reference CRG 973107. CAL acknowledges the support of an EPSRC Advanced Research Fellowship.
References [1] J.A. Rodriguez, D.W. Goodman, Science 257 (1992) 897. [2] J.A. Rodriguez, D.W. Goodman, J. Phys. Chem. 95 (1991) 4196. [3] R.A. Campbell, J.A. Rodriguez, D.W. Goodman, Phys. Rev. B 46 (1992) 7077. [4] B. Hammer, J.K. Norskov, in: R.M. Lambert, G. Pacchioni (Eds.), Chemisorption and Reactivity on Supported Clusters and Thin Films, Kluwer Academic Publishers, Dordrecht, 1997, p. 285. [5] J.A. Rodriguez, M. Kuhn, J. Phys. Chem. 98 (1994) 11251. [6] M. Mavrikakis, B. Hammer, J.K. Norskov, Phys. Rev. Lett. 81 (1998) 2819. [7] V. Pallassana, M. Neurock, L.B. Hansen, B. Hammer, J.K. Norskov, Phys. Rev. B 60 (1999) 6146. [8] C.A. Lucas, N.M. Markovic, P.N. Ross, Surf. Sci. Lett. 425 (1999) L381. [9] P.J. Feibelman, Phys. Rev. B 56 (1997) 2175. [10] N.M. Markovic, B.N. Grgur, C.A. Lucas, P.N. Ross, J. Phys. Chem. 103 (1999) 487. [11] G.A. Attard, A. Bannister, J. Electroanal. Chem. 300 (1991) 467. [12] J. Clavilier, M.J. Llorca, J.M. Feliu, A. Aldaz, J. Electroanal. Chem. 310 (1991) 429. [13] M. Baldauf, D.M. Kolb, J. Phys. Chem. 100 (1996) 11375. [14] V. Climent, N.M. Markovic, P.N. Ross, J. Phys. Chem. 104 (2000) 3116. [15] M.P. Soriaga, Y.-G. Kim, J.E. Soto, in: A. Wieckowski (Ed.), Interfacial Electrochemistry; Theory, Experiment and Applications, Marcel Dekker, New York, 1999 (Chapter 15). [16] C.A. Lucas, N.M. Markovic, P.N. Ross, Surf. Sci. 448 (2000) 77. [17] I.K. Robinson, D.J. Tweet, Rep. Prog. Phys. 55 (1992) 599. [18] N.M. Markovic, C.A. Lucas, V. Climent, V. Stamenkovic, P.N. Ross, Surf. Sci. 465 (2000) 103. [19] M. Takahasi, Y. Hayashi, J. Mizuki, K. Tamura, T. Kondo, H. Naohara, K. Uosaki, Surf. Sci. 461 (2000) 213. [20] I. Villegas, M.J. Weaver, J. Chem. Phys. 101 (1994) 1648. [21] M.W. Severson, C. Stuhlmann, I. Villegas, M.J. Weaver, J. Chem. Phys. 103 (1995) 9832. [22] G. Ertl, M. Neumann, K. Streit, Surf. Sci. 64 (1977) 393. [23] H. Steininger, S. Lehwald, H. Ibach, Surf. Sci. 123 (1982) 264. [24] D.F. Ogletree, M.A. Van Hove, G.A. Somorjai, Surf. Sci. 173 (1986) 351. [25] N.M. Markovic, H.A. Gasteiger, P.N. Ross, Langmuir 11 (1995) 4098. [26] C.A. Lucas, N.M. Markovic, unpublished.