Pd(111

Article pubs.acs.org/JPCC

Geometric Arrangement of Components in Bimetallic PdZn/Pd(111) Surfaces Modified by CO Adsorption: A Combined Study by Density Functional Calculations, Polarization-Modulated Infrared Reflection Absorption Spectroscopy, and Temperature-Programmed Desorption Christian Weilach,*,† Sergey M. Kozlov,‡ Harald H. Holzapfel,† Karin Föttinger,† Konstantin M. Neyman,*,‡,§ and Günther Rupprechter† †

Institute of Materials Chemistry, Vienna University of Technology, Getreidemarkt 9/BC, 1060 Vienna, Austria Departament de Química Física and Institut de Química Teòrica i Computacional, Universitat de Barcelona, Martí i Franquès, 1, 08028 Barcelona, Spain § Institució Catalana de Recerca i Estudis Avançats, 08010 Barcelona, Spain ‡

ABSTRACT: Combining density functional theory (DFT) calculations with experimental data from temperature-programmed desorption (TPD) and polarization-modulated infrared reflection absorption spectroscopy (PM-IRAS), we have provided evidence for the rearrangement of Pd and Zn atoms in PdZn surface alloys on Pd(111) in the presence of CO. Such systems represent valuable models for novel methanol steam reforming catalysts, with CO being an undesired byproduct of this reaction. The reconstructed surface was calculated to be more stable in the presence of CO and capable of adsorbing CO on top of Pd atoms up to a saturation coverage of 1/2 monolayer (ML). To the contrary, on a surface with conventional, bulklike geometric arrangement of Pd and Zn atoms in rows, CO adsorption on bridge sites with a saturation coverage of ∼1/3 ML was predicted. Adsorption of CO on top sites was confirmed with PM-IRAS and the saturation coverage of 1/2 ML was derived from TPD experiments, supporting the suggested reconstruction of PdZn/Pd(111). The rearrangement of the atomic components on the surface was calculated to affect structural and adsorptive properties but kept essentially unchanged the electronic structure, reflecting the intermetallic character of PdZn on Pd(111). This study sheds light on a new degree of complexity of bimetallic systems, that is, changes of component arrangement on the surface in the presence of certain adsorbates, which is not accompanied by surface segregation. Such rearrangement is expected to notably affect the properties of a bimetallic catalyst. surface. It recently attracted a lot of attention11−18 and in certain cases was shown to be promoted or hindered by certain adsorbates.4,6,19−21 At the same time, component or elemental arrangement, which is related to the change in relative order of components without changing the relative amount of each element on the surface, was neither sufficiently studied experimentally22,23 nor theoretically.24−26 However, it will be illustrated in this work that elemental arrangement is as important for catalytic properties of a surface as segregation.23 Indeed, the arrangement affects (1) elemental composition of adsorption sites present on the surface, (2) distances between atoms in the surface, (3) surface corrugation, and (4) relative stability of an adsorbate on various sites and its saturation coverage. It is shown in the present study that the magnitude

1. INTRODUCTION Structure and reactivity of a catalyst are well-known to change in the presence of certain adsorbates.1−4 However, this phenomenon seldom becomes a subject of thorough investigation,5,6 and often even very chemically potent adsorbates are regarded just as spectator species.7,8 Nevertheless, experiments suggest that reconstruction of bimetallic Co/Re/γAl2O39 and LaNi10 catalysts may occur even in the presence of CO, a common probe molecule, which should not modify the properties of the investigated surface. Thus, especially on bi- or multimetallic alloys, the validity of the spectator species “approximation” should be reassessed. The reason for these complications is the structural complexity of the alloys, which yields additional degrees of freedom sensitive to the environment, such as surface segregation and geometric arrangement of metal components on the surface. The former process, segregation, is related to changes in the amount of one or the other component on the © 2012 American Chemical Society

Received: May 10, 2012 Revised: July 17, 2012 Published: August 28, 2012 18768

dx.doi.org/10.1021/jp304556s | J. Phys. Chem. C 2012, 116, 18768−18778

The Journal of Physical Chemistry C

Article

atoms, while PdZn monolayers on Pd(111) are suggested to be flat or with opposite Pd-out/Zn-in corrugation.33,34,51,53 The density of states (DOS) of PdZn surface alloys near Fermi level was found to be closer to DOS of Cu(111) than to that of Pd(111) surface27,53 (in agreement with the PdZn catalytic activity reminiscent of that of Cu). Studies of PdZn surface alloys with DFT methods corroborated the geometrical and electronic structure of PdZn surface34,53 and its stability with respect to possible segregation of Zn or Pd.54 Detailed theoretical investigations of various adsorbed molecules55 and intermediates56 relevant to MSR as well as relevant elementary steps57−64 have been undertaken at low or moderate coverage of the adsorbates on PdZn/Pd(111) with various content of Zn. In addition, recent studies of bulk PdZn suggest that it has intermetallic nature with significant calculated charges on atoms (Pd−0.4Zn+0.4).65 The issue of atomic arrangement on the surface of PdZn/ Pd(111) in the absence of adsorbates has been recently studied with the help of finite temperature Monte Carlo simulations, based on DFT-derived nearest-neighbor interaction energy.26 These studies suggested that surface reconstruction occurs already at 400 K due to an entropic factor, even if multibody interactions and pair interactions with atoms from farther coordination spheres are not included into the simulation. In the present study, however, we perform DFT calculations at 0 K and the modified elemental arrangement is suggested on the basis of its energetic stability and comparison with experimental results. The structure of the PdZn/Pd(111) alloy in the presence of CO and other gases has not been extensively studied. However, recent studies31−33 reveal that the presence of CO shifts the onset of the alloy formation temperature by 75 K to higher T, suggesting a complex nature of the interaction between CO and the substrate.

and importance of the induced changes are significantly amplified by high coverage of an adsorbate. The scarcity of the available studies on the arrangement of elements on a bimetallic surface may be partially attributed to its possible aperiodicity, which makes very difficult to draw definitive conclusions from studies by certain surface science and highlevel theoretical techniques essentially limited to periodic systems. However, the combination of experimental and theoretical efforts has proven to be useful to treat this problem.25 In the present work, we study the surface reconstruction in the form of elemental rearrangement in thin films of PdZn alloys grown on Pd(111), induced by CO adsorption. Modification of the elemental arrangement and resulting surface reconstruction are suggested by density functional theory (DFT) calculations and corroborated by temperatureprogrammed desorption (TPD) and polarization-modulated infrared reflection absorption spectroscopy (PM-IRAS) experiments on well-defined model 1:1 PdZn/Pd(111) surface alloys. To the best of our knowledge, modification of elemental arrangement in the presence of adsorbates and its influence on adsorptive properties have not been reported before. The considered system, PdZn surface alloys grown on Pd(111), has been recently intensively studied,27−35 because it is envisaged as a possible replacement for presently industrially used pyrophoric36 and prone to sintering37 Cu/ZnO catalysts38 for methanol steam reforming (MSR), CH3OH + H2O → CO2 + 3H2.39,40 This reaction is crucial for future environmentally friendly hydrogen-based energy technologies, because it allows on-board generation of hydrogen from a (bio)renewable source. However, to be used in energy production, MSR has to be performed with high selectivity to avoid CO formation, which is a poison for Pt electrodes in polymer electrolyte membrane fuel cells.41 In contrast to presently used Cu/ZnO, PdZn-based MSR catalysts are stable at elevated operation temperatures, and unlike pure Pd they exhibit good conversion and high selectivity toward CO2 and hydrogen in MSR.42−46 However, after detailed analysis of model studies using PdZn surface alloys, the selectivity was found to depend critically on preparation conditions and, in particular, on elemental composition of subsurface layers and alloy thickness.31 Thus, we focused our attention only on the stoichiometric PdZn alloys, which are known to have good catalytic properties. Moreover, PdZn-based catalysts prepared by Pd deposition on ZnO were recently shown to be very selective in semihydrogenation of alkynes.47 In ultrahigh vacuum (UHV) conditions, surface alloys of PdZn have been produced either by evaporation of both Pd and Zn onto a Ru(001) single crystal48 or by Zn deposition on a Pd(111) substrate. The latter approach is more commonly used, and surface alloys prepared by this method were characterized by means of ultraviolet photoelectron spectroscopy (UPS),27 low-energy electron diffraction (LEED),27,49 scanning tunneling microscopy (STM)34,50 and low-energy ion scattering (LEIS).33,51,52 The active phase of the catalyst was found to be PdZn surface alloy of 1:1 stoichiometry with a p(2 × 1) surface periodicity,27,34,49 that is, Pd and Zn atoms aligned in rows. However, a bigger surface cell of 16.52 Å × 6.35 Å was suggested in some LEED experiments and was attributed to surface reconstruction at high temperature and specific preparation conditions.49 LEIS studies showed that a surface of multilayer PdZn is buckled with Zn atoms on the surface being located approximately 25 pm above surface Pd

2. EXPERIMENTAL DETAILS All experiments described below were carried out in an UHV chamber with a base pressure of 5 × 10−10 mbar. The system is equipped with a Specs ERLEED 150 and a Specs X-ray photoelectron spectroscopy (XPS) system consisting of a nonmonochromatic Al/Mg dual anode and a Phoibos EA 150 hemispherical analyzer. TPD spectra were collected by a differentially pumped MKS eVision+ quadrupole mass spectrometer, and temperature ramping was performed by a Eurotherm 3208 PID controller. Unwanted contributions of residual gases and molecules desorbing from the sample holder were minimized by placing a nozzle in front of the single crystal (∼1 mm distance). For PM-IRAS, a UHV-compatible high-pressure cell coupled to the UHV system has been used.66 The IR spectra were acquired on a Bruker IFS66v/S spectrometer, and polarization modulation was performed by a Hinds-PEM-90 ZnSe photoelastic modulator. Based on the accurate subtraction of gas-phase contributions, PM-IRAS provides surface vibrational spectra of adsorbed species and can be applied from UHV to ambient pressure.66−72 To clean the Pd(111) substrate, the Pd single crystal was sputtered in 5 × 10−6 mbar Ar at 1 kV for 45 min at 300 K, followed by annealing for 15 min at 1200 K. The sample was then cooled in 5 × 10−7 mbar O2 to 650 K to remove residual carbon. Finally, the single crystal was flashed to 1200 K in UHV. Sample cleanliness was checked by XPS and LEED. 18769



Article

PdZn surface alloys were prepared by physical vapor deposition of Zn (high-purity Zn wire, 99.98%, GoodFellow) out of a Ta crucible mounted in an Omicron ebeam evaporator. The deposited amount of Zn was calibrated by use of a quartz microbalance with the Zn flux during evaporation being additionally controlled via an internal flux monitor. In this respect, one monolayer (ML) is always referred to 1.53 × 1015 atoms/cm2, that is, the value of a close-packed (111) crystal plane in bulk Pd. It has been shown by Kratzer et al.35 that when a quartz microbalance is used to calibrate the amount of Zn deposited, great care has to be taken due to the lowered sticking coefficient of Zn on the contaminated surface of the quartz crystal. To overcome this problem we have deposited several ML of Zn on the quartz surface immediately before actual flux calibration. Upon annealing of the deposited Zn layers, a well-ordered LEED pattern was obtained under UHV, suggesting the coexistence of three rotational domains of a (2 × 1) structure consisting of alternating rows of Pd and Zn atoms, in agreement with recent STM studies.34,35 In general, deposition of 1 ML Zn results in the formation of two layers of a PdZn 1:1 surface alloy. However, for submonolayer amounts of Zn, bilayer growth of the surface alloy was observed by STM34 in agreement with simulations.26 Thus, 0.5 ML Zn deposition resulted in the formation of two-monolayer-high islands of PdZn layers, rather than one complete PdZn layer covering the whole Pd substrate. To ensure the formation of complete alloy surface layers, we have experimentally studied model systems after deposition of 1 ML Zn (=2 ML PdZn) and 3 ML Zn (>4 ML PdZn).

Eads(θ ) = {E[PdZn/Pd] + NE[COgas ] − E[(CO) N /PdZn/Pd]}/N

according to which positive values of Eads mean stabilization. Harmonic vibrational frequencies were calculated by displacing C and O atoms by 2 pm in all three Cartesian directions. Estimated contributions of harmonic vibrations to adsorption energy are 0.06 eV per CO molecule at 0 K (zero point energy corrections) and only 0.014 eV per CO molecule at 200 K. Because of their rather small magnitude, these corrections are not included into consideration. The calculated harmonic vibrational frequency of a gas-phase CO molecule, 2120 cm−1, underestimates experimental harmonic frequency of CO stretching by 50 cm−1 and is by 23 cm−1 lower than experimental anharmonic frequency.82 Thus, for better comparison of (anharmonic) frequencies yielded by PM-IRAS experiments with calculated frequencies, the latter are shifted by 23 cm−1 to higher values. We have performed test calculations employing (1) PBE exchange−correlation functional,74 (2) denser k-point grid of 5 × 5 × 1, and (3) supercell constructed with bulk-optimized Pd−Pd distance of 282 pm. Adsorption energies of CO and relative stabilities of the surface alloys obtained in these tests were by less than 0.1 eV different from those discussed in the following.

4. RESULTS AND DISCUSSION We have found that the results of TPD and PM-IRAS experiments performed in the present study are not compatible with the CO adsorption calculated on conventional structure of PdZn/Pd(111). To resolve this discrepancy between experimental and theoretical results, we propose a revised surface structure with Pd and Zn atoms arranged in zigzags. In this section we first describe the model of PdZn/Pd(111) surface with modified elemental arrangement and discuss the properties of reconstructed surface in the absence and in the presence of CO. Then we compare PM-IRAS and TPD experimental results with observables calculated for the conventional and reconstructed surface structures. 4.1. Stability of Modified Component Arrangement on the Surface of PdZn. Under UHV conditions (in the absence of CO), the elemental arrangement on the surface of PdZn/Pd(111) was determined to be alternating rows of Pd and Zn atoms (Figure 1, left panel)27,49 and was even visualized by STM.34 Here and further on in the text we refer to this structure as “row” structure. The modified elemental arrange-

3. COMPUTATIONAL DETAILS Spin-restricted periodic DFT calculations were performed with the VASP73 package with a revised Perdew−Burke−Ernzerhof (RPBE)74 exchange−correlation functional, which yields adsorption energies for CO on Pd75−77 close to the experimental values.78 Plane-wave basis set with cutoff energy of 415 eV was used to calculate eigenstates of valence electrons; core electrons were modeled with projector augmented wave (PAW) technique.79 The first Brillouin zone was sampled by Monkhorst−Pack 3 × 3 × 1 mesh of k-points,80 and the Methfessel−Paxton smearing81 of 0.1 eV was applied to Kohn− Sham eigenstates during geometry optimization. The total energies, however, were extrapolated to 0 smearing. DOS were calculated by use of a denser 5 × 7 × 1 k-points mesh and the smearing of 0.3 eV. PdZn surface alloys on Pd(111) were modeled by a six-layerthick p(4 × 3) supercell, which allowed us to consider CO coverage as low as 1/12 ML. Depending on the thickness of surface alloy under consideration, the top four layers of the slab consisted of either stoichiometric PdZn alloy or pure Pd and were relaxed (together with adsorbed CO molecules). The two bottom layers of the slab consisted only of Pd atoms and were fixed at bulk-derived positions with the experimental interatomic Pd−Pd distance of 275 pm. Geometry optimization was performed until all forces acting on atoms became less than 0.1 eV·nm−1. Average adsorption energies of CO at the coverage θ = N/12 ML, where N is the number of CO molecules present in the supercell, were calculated from the total energies E of systems X, E[X], in the following way:

Figure 1. PdZn surface alloy on Pd(111) with bulk terminated row structure and reconstructed zigzag structure. Pd atoms are displayed in cyan, and Zn atoms are displayed in blue. Lateral dimensions of calculated supercells are marked with solid lines; the zigzag arrangement formed by Pd atoms is marked with a dashed line. 18770



Article

ment on the surface was modeled by “zigzag” structure (Figure 1, right panel), which is constructed from the row structure by interchanging Pd and Zn atoms in one horizontal line, thus, transforming rows of Pd and Zn atoms into zigzags. This model may not match exactly the actual elemental arrangement, which may have an intractably big periodically repeated cell or even be aperiodic. Nevertheless, in this study we refrain from considering other plausible arrangements for two main reasons. First, in the absence of experimental input (i.e., supercell size) it is impossible to reliably predict the exact arrangement of atoms by periodic DFT methods. Second (unlike the conventional row structure), the present model is sufficient to reconcile results of PM-IRAS and TPD experiments with those obtained by DFT. Before discussing other properties of the reconstructed surface with the zigzag arrangement of components, we will address the principal issue of its stability in the absence of adsorbates and in the presence of CO adsorbed on the most stable adsorption sites. Relative stability of the PdZn/Pd(111) surface alloys of various thickness, which have one or more layers arranged in zigzags, with respect to the alloy of the same thickness but structured in rows is displayed in Figure 2. One

energy of the actual arrangement should be even lower than that calculated for zigzag structure. 4.2. Geometric Structure of the Surface Alloys. Structural parameters of PdZn arranged in zigzags are markedly different from those of PdZn with row structure (Figure 3).

Figure 3. Color-coded interatomic distances, d, between atoms in the surface of one-, two-, and four-layer-thick PdZn surface alloys on Pd(111) with row and zigzag structure, in the absence of CO and at θ(CO) = 1/2 ML. Numbers shown in italic type indicate surface corrugation in picometers; positive values correspond to Pd-out/Zn-in type corrugation.

Interatomic distances in the alloy with row structure are within 273−278 pm in the absence of CO and in the range of 274− 283 pm at θ(CO) = 1/2 ML. The longest distances are calculated between Pd and Zn atoms, while interatomic Pd−Pd and Zn−Zn distance stay close to 275 pm, the value for Pd(111). The slight elongation of the average bond length at high CO coverage accompanied with negligible lateral displacements is due to substantial increase of surface corrugation. The latter is measured as average difference between vertical coordinates of Pd and Zn atoms located on the surface, ⟨z(Pd)⟩ − ⟨z(Zn)⟩. In line with previous DFT and LEIS studies,33,34,51,53 in the present work surface corrugation of PdZn/Pd(111) is calculated to be 15, −28, and −19 pm for monolayer, bilayer, and multilayer surface alloys, respectively. Upon CO adsorption, corrugation increases to 58, 49, and 46 pm on one-, two-, and four-layer-thick alloys, respectively, and becomes of Pd-out/Zn-in type for any alloy thickness. The distribution of bond lengths is much wider in PdZn surface alloys with the zigzag structure. The lateral reconstruction makes possible elongation of Pd−Pd (284− 300 pm) and Zn−Zn (281−297 pm) distances and simultaneous contraction of the majority of Pd−Zn bonds (262−282 pm). These changes in bond lengths, not possible in the highly symmetric row structure, are in line with elongation of Pd−Pd and Zn−Zn bonds (289 pm) and contraction of Pd− Zn bonds (264 pm) measured in PdZn bulk65 in comparison with interatomic distance in Pd bulk (275 pm). Adsorption of mutually repulsing CO molecules at high coverage drives Pd atoms apart from each other and results in increased Pd−Pd distances of 290−315 pm. This in turn increases distances between adjacent Zn atoms (280−308 pm) and contracts or expands Pd−Zn bonds (259−290 pm). In the absence of CO, the corrugation of PdZn with zigzag structure is not much

Figure 2. Calculated relative stability of various structures of PdZn/ Pd(111) surface alloys with respect to the conventional structure (all layers arranged in rows), depending on the number of layers arranged in zigzags and on the CO coverage. Values are given per p(4 × 3) supercell, which has 12 metal atoms in each layer.

can see that, in the absence of adsorbates, a monolayer PdZn surface alloy is by 0.4 eV per p(4 × 3) supercell more stable with the zigzag structure. However, for multilayer surface alloys the row structure is the most stable one. The second most stable structure is that with two top layers arranged in zigzags in the case of bilayer alloy, and only one zigzag layer in the case of four-layer alloy. Later in the text we discuss only the most stable configurations of zigzag structures, that is, structures with only the top layer arranged in zigzags for the monolayer and four-layer-thick alloy and a structure with two PdZn layers arranged in zigzags for bilayer alloy. Adsorption of CO at experimentally observed (see section 4.5) surface coverage of 1/2 ML (one CO molecule per Pd atom on the surface), however, stabilizes zigzag structures of multilayer alloys by ∼1 eV, turning them into the most stable configuration. The relative stability of monolayer PdZn surface alloy with zigzag structure is further increased by ∼0.5 eV. One should recall here that the suggested zigzag structure is only a model for the most thermodynamically stable elemental arrangement present in the experiment. Thus, the relative 18771



Article

Calculated Bader charges83 reveal that, irrespective of the elemental arrangement, the PdZn surface alloys on Pd(111) have peculiar intermetallic nature in agreement with results calculated for bulk (Pd−0.4Zn+0.4).65 That is, charges of (+0.34) − (+0.43) (herein charges are given in electron units) are induced on the majority of Zn atoms; atoms on the interface of multilayer PdZn and Pd(111) bear charges of (+0.46) − (+0.5). In turn, the charges on most Pd atoms are (−0.4) − (−0.37), and on those located on the interface with Pd substrate, (−0.29) − (−0.21) Thus, calculated charges on Pd atoms located on the surface of monolayer PdZn/Pd(111) are almost 2 times smaller in magnitude than those located on the multilayer alloys, which is again in line with their different catalytic properties.31 The difference between electronic properties of surface Pd atoms in monolayer and in multilayer alloys may be rationalized by different numbers of adjacent Zn atoms (4 and 6, respectively). The fewer number of Zn neighbors and higher number of Pd neighbors renders Pd atoms in the monolayer PdZn/Pd with DOS and Bader charges closer to those of pristine Pd(111). This explains the phenomenon of subsurfacecontrolled selectivity, that is, reported dependency of PdZn/ Pd(111) catalyst properties on the surface alloy thickness.31 4.4. Adsorption of CO on Different Sites and PM-IRAS Studies. As mentioned above, the modification of elemental arrangement may change the relative stability of different adsorption sites. Because CO was shown not to adsorb on sites including Zn atoms (at temperatures ≥200 K),31−33 there are only two types of adsorption sites available for it on the surface: on top of Pd atoms and on Pd−Pd bridge sites. In general, three conclusions can be drawn from calculated adsorption energies of CO on PdZn/Pd(111) summarized in Table 1: (i) On the four-layer-thick surface alloy, CO adsorbs by 0.2−0.4 eV more weakly than on the monolayer alloy, illustrating the

different from corrugation of respective alloys arranged in rows, with deviations up to 7 pm. However, the corrugation of PdZn with zigzag structure in the presence of CO is ∼40 pm for monolayer and ∼25 pm for multilayer alloy. As in the case of PdZn/Pd(111) arranged in rows, Pd-out/Zn-in type corrugation takes place in all cases, but the amplitude of corrugation is ∼20 pm smaller in the case of laterally reconstructed surface. Significant differences between interatomic distances on the surface and surface corrugation of the PdZn alloys arranged in rows and zigzags will affect adsorptive properties of the system. In the present study, this statement is exemplified by considering adsorption of CO, a ubiquitous probe molecule. However, one should note here that on metal surfaces CO usually adsorbs on top, bridge, or 3-fold hollow sites. These sites may be sensitive only to very short-range structural changes and largely insensitive to changes in second and farther coordination spheres of adjacent metal atoms. Also, the considered modification of component arrangement does not introduce new types of the mentioned sites, that is, 3-fold hollow sites made exclusively of Pd or Zn atoms. Thus, one can speculate that adsorption of bigger molecules that interact with higher number of metal atoms on the surface will be much more strongly affected by structural changes induced by the rearrangement of components. 4.3. Electronic Structure of the Alloys. Despite significant structural changes of the surface, the electronic structure of PdZn/Pd(111) with zigzag arrangement is very similar to that of the alloy arranged in rows (Figure 4). Notably, DOS of the monolayer alloys is markedly different from that of the multilayer alloys, being closer to DOS of Pd(111).

Table 1. Adsorption Energies and Vibrational Frequencies of the CO Stretching Mode for CO Adsorbed on PdZn Surface Alloys of Various Thickness and Structure on Pd(111) at θ(CO) = 1/12 and 1/2 ML top 1

/12 ML

bridge 1

/2 ML

Rows, One-Layer-Thick PdZn adsorption energies (eV) 1.11 0.71 vibrational frequencies (cm−1) 2035 2083 Rows, Two-Layer-Thick PdZn adsorption energies (eV) 0.81 0.36 vibrational frequencies (cm−1) 2027 2068 Rows, Four-Layer-Thick PdZn adsorption energies (eV) 0.81 0.48 vibrational frequencies (cm−1) 2024 2059 Zigzags, One-Layer-Thick PdZn adsorption energies (eV) 1.11 0.95 vibrational frequencies (cm−1) 2035 2072 Zigzags, Two-Layer-Thick PdZn adsorption energies (eV) 0.94 0.64 vibrational frequencies (cm−1) 2031 2066 Zigzags, Four-Layer-Thick PdZn adsorption energies (eV) 0.91 0.67 vibrational frequencies (cm−1) 2034 2067

Figure 4. Calculated DOS of one-, two-, and four-layer-thick PdZn surface alloys on Pd(111) in the absence of CO projected on Pd atoms located on the surface. Energies of electronic Kohn−Sham states are given with respect to the Fermi energy, EF.

1

/12 ML

1

/2 ML

1.25 1893

0.87 1995

0.95 1884

0.47 1970

0.88 1894

0.5 1971

1.13 1883

0.73 1966

a

a

0.89 1889

0.44 1967

a

Bridge sites are not stable for these structures; CO moves from bridge to top sites during geometry optimization.

18772



Article

stability, one can conclude that PM-IRAS studies provide evidence for the suggested surface reconstruction. We have also acquired coverage-dependent IR spectra, like those presented in Figure 5 for the system with four PdZn

importance of subsurface layers composition on properties of PdZn catalysts. (ii) The adsorption energies of CO on top and on bridge sites are different by up to 0.14 eV at low CO coverage and by up to 0.23 eV at high coverage. (iii) Most importantly, data in Table 1 show that on monolayer and bilayer PdZn arranged in rows, CO adsorption on bridge sites is preferred, whereas on PdZn arranged in zigzags, on-top sites are clearly preferred at high CO coverage. Indeed, on the alloy arranged in rows, bridge sites are more stable than on-top sites by 0.11−0.16 eV for monolayer and bilayer structures but only by 0.02−0.07 eV for a four-layer-thick alloy. On the contrary, on the reconstructed alloys with zigzag structure, on-top sites are more stable than bridge sites by ∼0.2 eV at high CO coverage, while at the low coverage of 1/12 ML the adsorption sites have essentially the same stability. Also, on the bilayer alloys with zigzag structure, bridge sites are not stable at all, because the energy difference between adjacent top and bridge sites is sufficient to overcome the diffusion barrier between them. One can also note that at high CO coverage the adsorption is always stronger on the reconstructed surface. Preference of bridge sites for CO adsorption on unreconstructed surface can be understood in analogy with CO adsorption on pristine Pd(111). Indeed, at low coverage CO is known to adsorb on 3-fold hollow sites on Pd(111)84 and shows a clear preference for sites with higher coordination by Pd atoms.85 On the defect-free PdZn surface alloys, 3-fold hollow sites entirely made of Pd atoms are not present; thus the sites with the highest possible coordination are Pd−Pd bridge sites, and CO prefers to adsorb on them. At high coverage a similar reasoning would apply to CO adsorption on the alloy with row structure, because in this case distance between CO molecules does not depend on the type of occupied adsorption sites. However, on the reconstructed zigzag alloy the distance between coadsobates at high coverage depends on where they adsorb, being 317−327 pm for on-top sites and only 279−294 pm for bridge sites on a four-layer-thick alloy. Thus, it is not surprising that on-top sites are most stable in this case at high coverage, allowing reduction of the lateral repulsion between CO adsorbates, while at low coverage the stability of the sites is similar. By use of PM-IRAS, it could be shown31−33 that on welldefined PdZn/Pd(111) surface alloys exclusively on-top sites were occupied by adsorbed CO molecules, irrespective of alloy thickness (2−6 PdZn layers). PdZn surface alloys could be characterized by a sharp, characteristic band for on-top CO at ≈2070 cm−1 in IR experiments performed in a CO background pressure, that is, at high CO coverage. Similar spectra are obtained on oxide-supported PdZn nanoparticles.45,86 Only upon annealing of such alloys to temperatures above 623 K, when partial alloy decomposition started, were multiple bound CO species observed by PM-IRAS. DFT-calculated stretching frequencies of CO on top of Pd atoms are 2024−2035 cm−1 at θ(CO) = 1/12 ML and 2059−2083 cm−1 at θ(CO) = 1/2 ML (cf. Table 1) and support the assignment of experimentally observed ≈2070 cm−1 peak. In contrast, CO adsorbed on bridge sites was calculated to vibrate with frequencies of 1883− 1894 cm−1 at low coverage and 1966−1995 cm−1 at high CO coverage and was experimentally observed only at high temperatures, when decomposition of the alloy took place. Notably, calculated CO stretching frequencies do not significantly depend on the alloy thickness. Taking into account that some bridge-adsorbed species should be observed on unreconstructed surface, because of their calculated higher

Figure 5. Coverage-dependent IR spectra of CO adsorption on >4 ML PdZn/Pd(111) at 100 K.

layers. Upon a CO dose of 0.1 Langmuir, a broad band at 2048 cm−1 and a sharper, less intense band at 2068 cm−1 could be distinguished in the spectrum. Subsequently increasing the CO dose resulted in an intensity increase of the band at 2068 cm−1 accompanied by a decrease of the 2048 cm−1 band until the latter finally vanished from the spectrum after a CO dose of 10 Langmuir. Of course, the appearance of two IR bands corresponds to two different CO species on the surface. In our opinion, we can differentiate between CO that does not have other CO molecules at adjacent adsorption sites (dominates at low θ) and those that have (dominates at high θ). Their frequencies are different because increasing dipole−dipole interactions and reduced back bonding from the metal at higher coverage result in a spectral blue shift of the C−O stretching vibration. Positions of experimental bands, 2048 and 2068 cm−1, are in agreement with calculated frequencies of CO adsorbed on the four-layer-thick alloy with zigzag structure at θ(CO) = 1/12 and 1 /2 ML, 2034 and 2067 cm−1, respectively. Vibrational frequencies of CO adsorbed on top sites of the unreconstructed PdZn with four-layer thickness (2024 and 2059 cm−1, respectively) were somewhat further away from the experimental band positions. 4.5. Adsorption of CO at Intermediate Coverage and TPD Studies. To model TPD spectra, we have calculated differential adsorption energies of CO, that is, energies required for only one CO molecule to desorb at a given coverage, θ = N/12, where N is the number of CO molecules in the supercell, using the following definition: Ediff (θ ) = E[(CO) N − 1/PdZn/Pd] + E[COgas ] − E[(CO) N /PdZn/Pd] 18773



Article

We have calculated Ediff for CO adsorbed on the on-top sites of the alloys with zigzag structure and both on-top and bridge sites on the alloys with row structure at θ = 1/2 ML (N = 6). In some cases calculations were also done at an intermediate coverage (Figure 6).

reconstructed surface alloy were positive and higher than 0.34 eV even at θ(CO) = 1/2 ML. In order to determine experimentally the saturation CO coverage on PdZn/Pd(111) and investigate alloy formation and decomposition in more detail, we have performed TPD measurements monitoring CO adsorption and desorption, depending on the PdZn surface alloy thickness, annealing temperature, and CO dose (Figure 7). In these experiments we have started from unalloyed Zn layers on the Pd substrate (1 and 3 ML Zn deposited at liquid N2 temperature), with the sample first heated to 273 K and cooled down in 1 × 10−6 mbar CO to 90 K to saturate the surface with CO. After that, repeated TPD runs were performed to consecutively higher end temperatures (each of which was kept for 5 min to anneal the layers at this temperature). The values given in Figure 7 indicate the annealing temperature before the TPD run. After one run was finished, the sample was saturated again with CO during cooling before the next TPD spectrum was acquired. Mainly three regimes can be differentiated in Figure 7: (i) Up to the annealing temperature of 373 K, basically no CO desorption was observed, in good agreement with the previously reported absence of CO adsorption in PM-IRAS.33 In this regime, the deposited Zn layers are not yet alloyed with the Pd substrate. Only trace amounts of CO were detected by mass spectrometry, most probably due to CO desorbing from the heating wires holding the crystal or from the crystal edges. (ii) Alloy formation and stability regime: after heating to 473 K, a clear CO desorption peak could be observed, which we attribute to CO desorbing from alloyed PdZn layers. After annealing to 573 K, this peak reached its maximum intensity (“ideal” PdZn alloy) and was made up from two desorption

Figure 6. Dependence of differential adsorption energy, Ediff, of CO on its coverage on various PdZn/Pd(111) surface alloys. Solid symbols, adsorption on top; open symbols, adsorption on bridge site. Adsorption on the alloys with row structures is presented in black and on those with zigzag structure in gray. Circles, squares, and triangles label the one-, two-, and four-layer-thick alloys, respectively.

It turned out that, on nonreconstructed surfaces arranged in rows, CO is noticeably stable only up to coverage of θ ≤ 0.33 ML, with a sharp decrease of stability at higher coverage. This makes full coverage of the surface Pd atoms by CO molecules strongly unfavorable at room temperature for the row structure. At the same time, differential adsorption energies of CO on the

Figure 7. CO TPD spectra from (a) 2 ML and (b) >4 ML PdZn/Pd(111) depending on annealing temperature Tanneal. 18774



Article

Figure 8. CO coverage as deduced from the TPD spectra in Figure 7 versus annealing temperature. The dashed line indicates saturation coverage on the “perfect” alloy.

meaning that lower temperatures were required for the thinner alloy to reach the same surface state. Additionally, there seemed to be some effect of subsurface Zn on the CO adsorption on Pd, because when the TPD spectra acquired after annealing to 900 K are compared, the CO desorption maximum was shifted by about 15 K to lower temperatures for the system with higher Zn content. This finding is in agreement with results of recent DFT studies that showed lowering of CO adsorption energies on Pd(111) if Zn atoms are present in the subsurface region.55 By integrating the peak areas and taking the well-known 0.75 ML coverage of the (2 × 2) CO saturation structure on Pd(111) as a reference (see ref 87 and references therein), we have calculated the CO coverage for each TPD spectrum. The obtained data are plotted in Figure 8; error bars were derived from three repeated measurements. After annealing of both systems to 573 K, that is, when the alloy-related TPD and PMIRAS features were most pronounced (dashed lines in Figure 8), the CO coverage was 0.50 ± 0.02 ML for the 2 ML PdZn/ Pd(111) system and only slightly lower, 0.47 ± 0.02 ML, for the >4 ML PdZn system, in good agreement with similarly prepared systems.89 When it is kept in mind that the Pd:Zn ratio is 1:1 in a PdZn alloy surface layer, these values indicate that every surface Pd atom is occupied by a CO molecule. Along the Pd rows/zigzags the CO molecules are even more densely packed than in the (2 × 2) CO super structure on Pd(111), which could represent a possible driving force for rearrangement of the surface. Altogether, the experimentally determined saturation coverage of 0.5 ML CO, which was computed to be achievable only for the alloy with zigzag structure, gives another strong indication for a surface reconstruction. To investigate this in more detail, a number of TPD spectra were acquired upon increasing the initial CO dose from 0.5 L to saturation coverage, similar to the PM-IRAS spectra of Figure 5. The TPD spectra obtained after low dosage should be comparable with the calculations at 1/12 coverage, whereas the spectra upon higher doses and cooling in CO (i.e., surface saturated with CO) correspond to the calculations at high (1/2 ML) coverage. The obtained spectra are plotted in Figure 9, again for the systems of 2 ML and >4 ML PdZn/Pd(111) (panels a and b, respectively). Two desorption regimes can be distinguished in

features: a shoulder at around 170 K and the peak maximum at 210 K. (iii) Further annealing above 623 K resulted in the appearance of additional desorption features at higher desorption temperature, indicating the onset of alloy degradation via Zn dissolution in the bulk and the reappearance of Pd patches at the surface. Finally, after annealing to 900 K, the PdZn alloy was completely decomposed and the surface had transformed back to clean Pd(111). The acquired TPD (and PM-IRAS) spectra closely resemble the spectra obtained upon CO desorption from Pd(111).87 In the TPD spectra shown in Figure 7, no major difference was seen between 2 ML and >4 ML PdZn/Pd(111) (at temperatures when alloys are stable), indicating that CO desorption was not really affected by the thickness of the underlying PdZn layer. When the small differences in calculated CO adsorption energies on bilayer and multilayer alloys are considered (cf. Table 1), this observation seems reasonable. For the “ideal” PdZn alloys (annealing to 573 K), we have estimated the adsorption energy by applying the Redhead equation.88 A pre-exponential factor for desorption of 1015.3 s−1 was used, which is in good agreement with available data for desorption energies reported for CO on pristine Pd(111).78 When CO desorption from the 2 ML PdZn system is considered, the maximum of 212 K corresponds to a desorption energy of ETPD = 0.66 eV per molecule, which is in remarkable agreement with the calculated value of CO adsorption energy Eads = 0.64 eV at high coverage on PdZn with zigzag structure. Also, for the desorption maximum of the 4 ML system at 207 K, the experimentally estimated value of ETPD = 0.64 eV/ molecule matches nicely the calculated Eads = 0.67 eV for CO adsorption on the reconstructed surface. This may also be taken as a hint toward CO-induced surface reconstruction, although one has to keep in mind that the experimental numbers change if another pre-exponential factor is chosen. While the general trend was the same for both PdZn systems investigated, there were some subtle differences in the TPD spectra of Figure 7, especially in the “transformation regime”, that is, above 573 K. The experimental results suggest that the transformation back to Pd(111) proceeded faster and more easily on the thinner alloy. For example, the 623 K spectrum from the 2 ML PdZn system (Figure 7 a) closely resembles the 673 K spectrum of the >4 ML PdZn system (Figure 7 b), 18775



Article

As for the TPD spectra presented in Figure 7, we have again estimated the desorption energy from the temperature of the peak maximum by applying the Redhead equation. The numbers obtained are summarized in Table 2. Table 2. CO Desorption Maxima and Respective Desorption Energies for 2 and >4 ML PdZn Surface Alloysa 2 ML PdZn/Pd(111)

a

>4 ML PdZn/Pd(111)

CO dose (L)

Tdes (K)

ETPD (eV)

Tdes (K)

ETPD (eV)]

0.5 1 2 5 sat.

295 285 185/227 183/226 183/224

0.92 0.89 0.58/0.70 0.57/0.70 0.57/0.69

256 236 194/231 193/218 180/213

0.80 0.74 0.60/0.72 0.60/0.68 0.56/0.67

According to the TPD spectra in Figure 9.

The experimentally measured values of desorption energy ETPD obtained after low dose fit quite well the DFT numbers (compare Table 1). However, because calculated CO adsorption energies on PdZn with row and zigzag structures are very similar at low coverage, it is not possible to differentiate the two structures on the basis of the experimental results. At higher CO dosage, the main peaks at 210−230 K seem to be reproduced well by average adsorption energy of CO on top sites of the respective alloys with zigzag structure (cf. Table 1). At low CO coverage, the calculated differential adsorption energies of the reconstructed structure are also in line with experimental values. The agreement is worse if one compares experimental values to the Ediff values at θ(CO) = 1/2 ML, calculated to be 0.71 and 0.34 eV for two- and four-layer-thick alloys, respectively. At the same time, average adsorption energies for on-top sites in the alloys with row structure are less than 0.5 eV at high coverage and differential adsorption energies are even negative in this case, contradicting the experimental observations. Adsorption of CO on bridge site of the unreconstructed alloys yields somewhat higher adsorption energies. Nevertheless, it does not affect significantly the results at high CO coverage, with Eads still being less than 0.5 eV and Ediff still being negative. Thus, the adsorption energies derived from TPD experiments definitely favor the modification of elemental arrangement in the surface layer.

Figure 9. Coverage-dependent CO TPD spectra for (a) 2 ML and (b) >4 ML PdZn/Pd(111). For each series the CO dose was subsequently increased from 0.5 to 5 Langmuir. Saturation coverage (sat.) was achieved by cooling the system in 1 × 10−6 mbar CO to 90 K.

the spectra: CO desorbing from the PdZn surface alloy below 300 K, and CO desorption from remaining Pd(111) areas above 400 K (cf. Figure 7). For interpreting the coveragedependent adsorption properties of CO on PdZn, we may disregard the latter here and focus on the low temperature desorption features. After the smallest dosage of 0.5 L CO onto the 2 ML PdZn system, a small desorption feature was observed with a maximum at 295 K, attributed to desorption from the alloy. Increasing the dose gradually shifted the desorption maximum to lower temperature, and it reached a value of 227 K after dosing of 2 L CO. Upon this dose, another desorption feature appeared as a low-temperature shoulder near 183 K. Further CO dosing then increased the signal intensity but basically did not alter the position of the desorption maximum. The third feature around 295 K was also detected in the spectra after elevated doses, which might tentatively be attributed to CO desorbing from imperfect areas of the surface. The general trend was similar for the thicker PdZn system (Figure 9 b); however, the alloy-related features were more pronounced. CO desorption from PdZn led to a small peak at 256 K, which also subsequently increased in intensity and shifted to lower temperature with increasing CO dose, reaching a maximum at 213 K upon desorption from the saturated surface. Analogous to the thinner PdZn system, after a CO dose of 2 L a low-temperature shoulder appeared at 195 K, which in this case was finally shifted to 180 K.

5. CONCLUSION In the present study we have investigated in detail the interaction between adsorbed CO molecules and a few layers thick PdZn surface alloys on Pd(111) by means of TPD and PM-IRAS spectroscopies and DFT modeling. We have addressed the issue of possible modification of elemental arrangement on the surface by considering alloys of PdZn/ Pd(111) with model “zigzag” structure on the surface. PdZn alloys with the zigzag arrangement on the surface are found to be energetically more stable in the cases of high CO coverage and one-monolayer alloy thickness. The reconstructed surface exhibits markedly different adsorptive properties, adsorbing CO on top of Pd atoms, while adsorption is predicted to preferentially occur on bridge sites on alloys with the conventional “row” structure. CO adsorbed on bridge sites, however, is never observed in PM-IRAS experiments (at moderate annealing temperature), providing evidence for similar surface reconstruction in the experiment. Another line of evidence comes from TPD experiments, where 1/2 ML 18776



Article

(6) Gonzalez, S.; Neyman, K. M.; Shaikhutdinov, S.; Freund, H.-J.; Illas, F. J. Phys. Chem. C 2007, 111, 6852−6856. (7) Stamenkovic, V. R.; Mun, B. S.; Arenz, M.; Mayrhofer, K. J. J.; Lucas, C. A.; Wang, G. F.; Ross, P. N.; Markovic, N. M. Nat. Mater. 2007, 6, 241−247. (8) Gokhale, A. A.; Dumesic, J. A.; Mavrikakis, M. J. Am. Chem. Soc. 2008, 130, 1402−1414. (9) Rygh, L. E. S.; Nielsen, C. J. J. Catal. 2000, 194, 401−409. (10) Damyanova, S.; Daza, L.; Fierro, J. L. G. J. Catal. 1996, 159, 150−161. (11) Cheng, D.; Atanasov, I. S.; Hou, M. Eur. Phys. J. D 2011, 64, 37−44. (12) Demirci, U. B. J. Power Sources 2007, 173, 11−18. (13) Chen, S.; Sheng, W. C.; Yabuuchi, N.; Ferreira, P. J.; Allard, L. F.; Shao-Horn, Y. J. Phys. Chem. C 2009, 113, 1109−1125. (14) Dai, Y.; Ou, L. H.; Liang, W.; Yang, F.; Liu, Y. W.; Chen, S. L. J. Phys. Chem. C 2011, 115, 2162−2168. (15) Wei, Y. C.; Liu, C. W.; Chang, W. J.; Wang, K. W. J. Alloys Compd. 2011, 509, 535−541. (16) Wanjala, B. N.; Luo, J.; Fang, B.; Mott, D.; Zhong, C. J. J. Mater. Chem. 2011, 21, 4012−4020. (17) Ferrando, R.; Jellinek, J.; Johnston, R. L. Chem. Rev. 2008, 108, 845−910. (18) Yudanov, I. V.; Neyman, K. M. Phys. Chem. Chem. Phys. 2010, 12, 5094−5100. (19) Baddeley, C. J.; Bloxham, L. H.; Laroze, S. C.; Raval, R.; Noakes, T. C. Q.; Bailey, P. Surf. Sci. 1999, 433, 827−832. (20) Owens, T. G.; Jones, T. E.; Noakes, T. C. Q.; Bailey, P.; Baddeley, C. J. J. Phys. Chem. B 2006, 110, 21152−21160. (21) van den Oetelaar, L. C. A.; Nooij, O. W.; Oerlemans, S.; van der Gon, A. W. D.; Brongersma, H. H.; Lefferts, L.; Roosenbrand, A. G.; van Veen, J. A. R. J. Phys. Chem. B 1998, 102, 3445−3455. (22) Yuhara, J.; Schmid, M.; Varga, P. Phys. Rev. B 2003, 67, No. 195407. (23) Maroun, F.; Ozanam, F.; Magnussen, O. M.; Behm, R. J. Science 2001, 293, 1811−1814. (24) Stephens, J. A.; Ham, H. C.; Hwang, G. S. J. Phys. Chem. C 2010, 114, 21516−21523. (25) Demirci, E.; Carbogno, C.; Gross, A.; Winkler, A. Phys. Rev. B 2009, 80, No. 085421. (26) He, X. A.; Huang, Y. C.; Chen, Z. X. Phys. Chem. Chem. Phys. 2011, 13, 107−109. (27) Bayer, A.; Flechtner, K.; Denecke, R.; Steinrück, H.-P.; Neyman, K. M.; Rösch, N. Surf. Sci. 2006, 600, 78−94. (28) Neyman, K. M.; Lim, K. H.; Chen, Z. X.; Moskaleva, L. V.; Bayer, A.; Reindl, A.; Borgmann, D.; Denecke, R.; Steinrück, H.-P.; Rösch, N. Phys. Chem. Chem. Phys. 2007, 9, 3470−3482. (29) Jeroro, E.; Lebarbier, V.; Datye, A.; Wang, Y.; Vohs, J. M. Surf. Sci. 2007, 601, 5546−5554. (30) Jeroro, E.; Vohs, J. M. J. Am. Chem. Soc. 2008, 130, 10199− 10207. (31) Rameshan, C.; Stadlmayr, W.; Weilach, C.; Penner, S.; Lorenz, H.; Hävecker, M.; Blume, R.; Rocha, T.; Teschner, D.; Knop-Gericke, A.; Schlögl, R.; Memmel, N.; Zemlyanov, D.; Rupprechter, G.; Klötzer, B. Angew. Chem., Int. Ed. 2010, 49, 3224−3227. (32) Rameshan, C.; Weilach, C.; Stadlmayr, W.; Penner, S.; Lorenz, H.; Hävecker, M.; Blume, R.; Rocha, T.; Teschner, D.; Knop-Gericke, A.; Schlögl, R.; Zemlyanov, D.; Memmel, N.; Rupprechter, G.; Klötzer, B. J. Catal. 2010, 276, 101−113. (33) Stadlmayr, W.; Rameshan, C.; Weilach, C.; Lorenz, H.; Hävecker, M.; Blume, R.; Rocha, T.; Teschner, D.; Knop-Gericke, A.; Zemlyanov, D.; Penner, S.; Schlögl, R.; Rupprechter, G.; Klötzer, B.; Memmel, N. J. Phys. Chem. C 2010, 114, 10850−10856. (34) Weirum, G.; Kratzer, M.; Koch, H. P.; Tamtögl, A.; Killmann, J.; Bako, I.; Winkler, A.; Surnev, S.; Netzer, F. P.; Schennach, R. J. Phys. Chem. C 2009, 113, 9788−9796. (35) Kratzer, M.; Tamtögl, A.; Killmann, J.; Schennach, R.; Winkler, A. Appl. Surf. Sci. 2009, 255, 5755−5759.

saturation coverage of CO (one molecule per Pd atom on the surface) on PdZn/Pd(111) was measured. At the same time, calculations of differential adsorption energy of CO molecules suggest that, at θ = 1/2 ML, CO is stable only on alloys with zigzag structure. On multilayer unreconstructed PdZn, the saturation coverage is calculated to be around 1/3 ML. A closer look at the calculated properties of PdZn/Pd(111) alloys with modified elemental arrangement reveals that they feature longer Pd−Pd and Zn−Zn interatomic distances with Pd−Zn bonds being, in general, somewhat shorter. Also the row to zigzag rearrangement significantly affects surface corrugation of the alloy at high CO coverage. However, the electronic structure of the alloys seems to be rather insensitive to the elemental arrangement of the surface. Indeed, calculated DOS projected on the Pd atoms on alloy surface depend mainly on the alloy thickness, explaining different reactivity of monolayer PdZn surface alloys, compared to thicker alloys. Notably, the PdZn surface alloys were found to have intermetallic nature and feature significant polarization of atoms, being roughly Pd−0.4Zn+0.4. In summary, these studies for the first time addressed the issue of elemental rearrangement on the surface of a bimetallic system in the presence of an adsorbate. Experimental evidence of the elemental rearrangement were collected and the influence of the rearrangement on the surface properties was analyzed. This study reveals a new degree of complexity in bimetallic systems under reaction conditions and calls for additional studies of elemental arrangement on the surface of PdZn/Pd(111) and other bimetallic systems.

■

AUTHOR INFORMATION

Corresponding Author

*(C.W.) E-mail [email protected], tel +43 1 58801 165 115, fax +43 1 58801 16599. (K.M.N.) E-mail konstantin. [email protected], tel +34 93 40 37 212, fax +34 93 40 21 231. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work has been performed within activities of the COST Actions MP0903 and CM0904 and was in part supported by the Austrian Science Fund (FWF) [SFB-F4502-N16 FOXSI]. Financial support from the Spanish MICINN (Grant FIS200802238), the Generalitat de Catalunya (Grant 2009SGR1041), and XRQTC is gratefully acknowledged. S.M.K. is grateful to the Spanish Ministerio de Eucación, Cultura y Deporte for predoctoral FPU Grant AP2009-3379. S.M.K. also acknowledges support by COST action MP0903 by granting a shortterm scientific mission and by HPC-Europa2 by financing his visit and providing computer time in Barcelona Supercomputer Center. We thank C. Rameshan, W. Stadlmayer, N. Memmel, and B. Klötzer for valuable discussions.

■

REFERENCES

(1) Lambert, R. M.; Williams, F. J.; Cropley, R. L.; Palermo, A. J. Mol. Catal. A: Chem. 2005, 228, 27−33. (2) Somorjai, G. A. Catal. Lett. 1992, 12, 17−34. (3) Maetz, P.; Touroude, R. Appl. Catal., A 1997, 149, 189−206. (4) Dupont, C.; Delbecq, F.; Loffreda, D.; Jugnet, Y. J. Catal. 2011, 278, 239−245. (5) Grabow, L. C.; Gokhale, A. A.; Evans, S. T.; Dumesic, J. A.; Mavrikakis, M. J. Phys. Chem. C 2008, 112, 4608−4617. 18777



Article

(36) Schuyten, S.; Guerrero, S.; Miller, J. T.; Shibata, T.; Wolf, E. E. Appl. Catal., A 2009, 352, 133−144. (37) Kurtz, M.; Wilmer, H.; Genger, T.; Hinrichsen, O.; Muhler, M. Catal. Lett. 2003, 86, 77−80. (38) Gunter, M. M.; Ressler, T.; Jentoft, R. E.; Bems, B. J. Catal. 2001, 203, 133−149. (39) Sa, S.; Silva, H.; Brandao, L.; Sousa, J. M.; Mendes, A. Appl. Catal., B 2010, 99, 43−57. (40) Palo, D. R.; Dagle, R. A.; Holladay, J. D. Chem. Rev. 2007, 107, 3992−4021. (41) Song, C. S. Catal. Today 2002, 77, 17−49. (42) Iwasa, N.; Mayanagi, T.; Ogawa, N.; Sakata, K.; Takezawa, N. Catal. Lett. 1998, 54, 119−123. (43) Iwasa, N.; Takezawa, N. Top. Catal. 2003, 22, 215−224. (44) Conant, T.; Karim, A. M.; Lebarbier, V.; Wang, Y.; Girgsdies, F.; Schlögl, R.; Datye, A. J. Catal. 2008, 257, 64−70. (45) Föttinger, K.; van Bokhoven, J. A.; Nachtegaal, M.; Rupprechter, G. J. Phys. Chem. Lett. 2011, 2, 428−433. (46) Zhang, Q. L.; Farrauto, R. J. Appl. Catal., A 2011, 395, 64−70. (47) Tew, M. W.; Emerich, H.; van Bokhoven, J. A. J. Phys. Chem. C 2011, 115, 8457−8465. (48) Rodriguez, J. A. J. Phys. Chem. 1994, 98, 5758−5764. (49) Gabasch, H.; Knop-Gericke, A.; Schlögl, R.; Penner, S.; Jenewein, B.; Hayek, K.; Klötzer, B. J. Phys. Chem. B 2006, 110, 11391−11398. (50) Weirum, G.; Barcaro, G.; Fortunelli, A.; Weber, F.; Schennach, R.; Surnev, S.; Netzer, F. P. J. Phys. Chem. C 2010, 114, 15432−15439. (51) Stadlmayr, W.; Penner, S.; Klötzer, B.; Memmel, N. Surf. Sci. 2009, 603, 251−255. (52) Stadlmayr, W.; Klötzer, B.; Penner, S.; Memmel, N. J. Phys. Chem. C 2012, 116, 3635−3644. (53) Chen, Z. X.; Neyman, K. M.; Gordienko, A. B.; Rösch, N. Phys. Rev. B 2003, 68, No. 075417. (54) Chen, Z. X.; Neyman, K. M.; Rösch, N. Surf. Sci. 2004, 548, 291−300. (55) Huang, Y. C.; Ding, W. P.; Chen, Z. X. J. Chem. Phys. 2010, 133, No. 214702. (56) Koch, H. P.; Bako, I.; Schennach, R. Surf. Sci. 2010, 604, 596− 608. (57) Huang, Y. C.; He, X.; Chen, Z. X. J. Chem. Phys. 2011, 134, No. 184702. (58) Huang, Y. C.; Chen, Z. X. J. Phys. Chem. C 2011, 115, 18752− 18760. (59) Lin, S.; Xie, D. Q.; Guo, H. J. Phys. Chem. C 2011, 115, 20583− 20589. (60) Smith, G. K.; Lin, S.; Lai, W. Z.; Datye, A.; Xie, D. Q.; Guo, H. Surf. Sci. 2011, 605, 750−759. (61) Lim, K. H.; Chen, Z. X.; Neyman, K. M.; Rösch, N. J. Phys. Chem. B 2006, 110, 14890−14897. (62) Chen, Z. X.; Lim, K. H.; Neyman, K. M.; Rösch, N. J. Phys. Chem. B 2005, 109, 4568−4574. (63) Chen, Z. X.; Neyman, K. M.; Lim, K. H.; Rösch, N. Langmuir 2004, 20, 8068−8077. (64) Chen, Z. X.; Lim, K. H.; Neyman, K. M.; Rösch, N. Phys. Chem. Chem. Phys. 2004, 6, 4499−4504. (65) Friedrich, M.; Ormeci, A.; Grin, Y.; Armbrüster, M. Z. Anorg. Allg. Chem. 2010, 636, 1735−1739. (66) Rupprechter, G. Adv. Catal. 2007, 51, 133−263. Rupprechter, G. Catal. Today 2007, 126, 2−17. (67) Rupprechter, G.; Weilach, C. Nano Today 2007, 2, 20−29. (68) Borasio, M.; de la Fuente, O. R.; Rupprechter, G.; Freund, H.-J. J. Phys. Chem. B 2005, 109, 17791−17794. (69) Ozensoy, E.; Meier, D. C.; Goodman, D. W. J. Phys. Chem. B 2002, 106, 9367−9371. (70) Gao, F.; McClure, S. M.; Cai, Y.; Gath, K. K.; Wang, Y.; Chen, M. S.; Guo, Q. L.; Goodman, D. W. Surf. Sci. 2009, 603, 65−70. (71) Diemant, T.; Zhao, Z.; Rauscher, H.; Bansmann, J.; Behm, R. J. Surf. Sci. 2007, 601, 3801−3804.

(72) Artiglia, L.; Diemant, T.; Hartmann, H.; Bansmann, J.; Behm, R. J.; Gavioli, L.; Cavaliere, E.; Granozzi, G. Phys. Chem. Chem. Phys. 2010, 12, 6864−6874. (73) Kresse, G.; Furthmuller, J. Phys. Rev. B 1996, 54, 11169−11186. (74) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865−3868. (75) Hammer, B.; Hansen, L. B.; Nørskov, J. K. Phys. Rev. B 1999, 59, 7413−7421. (76) Lopez, N.; Nørskov, J. K. Surf. Sci. 2001, 477, 59−75. (77) Yudanov, I. V.; Sahnoun, R.; Neyman, K. M.; Rösch, N. J. Chem. Phys. 2002, 117, 9887−9896. (78) Guo, X. C.; Yates, J. T. J. Chem. Phys. 1989, 90, 6761−6766. (79) Kresse, G.; Joubert, D. Phys. Rev. B 1999, 59, 1758−1775. (80) Monkhorst, H. J.; Pack, J. D. Phys. Rev. B 1976, 13, 5188−5192. (81) Methfessel, M.; Paxton, A. T. Phys. Rev. B 1989, 40, 3616−3621. (82) Huber, K.-P.; Herzberg, G. Molecular Spectra and Molecular Structure, Vol 4: Constants of diatomic molecules; Van Nostrand: New York, 1979. (83) Bader, R. F. W. Atoms in Molecules; Clarendon Press: Oxford, U.K., 1990. (84) Rose, M. K.; Mitsui, T.; Dunphy, J.; Borg, A.; Ogletree, D. F.; Salmeron, M.; Sautet, P. Surf. Sci. 2002, 512, 48−60. (85) Loffreda, D.; Simon, D.; Sautet, P. Surf. Sci. 1999, 425, 68−80. (86) Lebarbier, V.; Dagle, R.; Conant, T.; Vohs, J. M.; Datye, A. K.; Wang, Y. Catal. Lett. 2008, 122, 223−227. (87) Morkel, M.; Rupprechter, G.; Freund, H.-J. J. Chem. Phys. 2003, 119, 10853−10866. (88) Redhead, P. A. Vacuum 1962, 12, 203−211. (89) Tamtögl, A.; Kratzer, M.; Killman, J.; Winkler, A. J. Chem. Phys. 2008, 129, No. 224706.

18778


Pd(111

Recommend Documents