10850
J. Phys. Chem. C 2010, 114, 10850–10856
Temperature-Induced Modifications of PdZn Layers on Pd(111) W. Stadlmayr,† Ch. Rameshan,†,‡ Ch. Weilach,§ H. Lorenz,† M. Ha¨vecker,‡ R. Blume,‡ T. Rocha,‡ D. Teschner,‡ A. Knop-Gericke,‡ D. Zemlyanov,| S. Penner,† R. Schlo¨gl,‡ G. Rupprechter,§ B. Klo¨tzer,† and N. Memmel*,† Institute of Physical Chemistry, UniVersity of Innsbruck, Innrain 52a, A-6020 Innsbruck, Austria, Department of Inorganic Chemistry, Fritz-Haber-Institute of the Max-Planck-Society, Faradayweg 4-6, D-14195 Berlin, Germany, Institute of Materials Chemistry, Vienna UniVersity of Technology, Veterina¨rplatz 1, A-1210 Vienna, Austria, and Birck Nanotechnology Center, Purdue UniVersity, 1205 West State Street, West Lafayette, Indiana 47907-2057 ReceiVed: January 29, 2010; ReVised Manuscript ReceiVed: April 21, 2010
Ultrathin PdZn surface alloys on Pd(111) are model systems well-suited for obtaining a microscopic understanding of the mechanisms of Pd/Zn-based catalysis for methanol steam reforming. The temperatureinduced compositional and structural changes of these alloy films are investigated in the catalytically relevant temperature range. Heating of multilayer Zn films to 500 K results in the formation of multilayer PdZn alloy films with surface and near-surface composition close to 1:1. In the temperature regime above 550 K the subsurface layers deplete quickly in Zn due to diffusion of Zn atoms into the Pd bulk. In contrast, the composition of the surface layer changes only slightly, indicating formation of a PdZn film with strong monolayer character. This change in subsurface composition triggers a change of the original Zn-out/Pd-in surface corrugation, leading ultimately to a Pd-out/Zn-in situation for annealing temperatures beyond 700 K. The altered corrugation pattern is also obtained when submonolayer amounts of Zn are heated to ∼500 K. The observed structural changes are in qualitative agreement with predictions by DFT calculations. 1. Introduction Hydrogen technology, although being a promising alternative for future energy supply, still has to overcome many problems. One of the main obstacles is the transport and storage of hydrogen fuel. Methanol can be used as an alternative intermediate energy carrier and source of hydrogen. However, an effective catalyst for synthesis of hydrogen is then needed. Pd/ Zn-based catalysts have recently proven to be highly capable of methanol steam reforming (CH3OH + H2O f CO2 + 3H2), even outrunning Cu-based catalysts because of their better thermal stability.1,2 A particularly important point in this process is to avoid simultaneous CO production, since the presence of the latter in the feed gas of proton-exchange membrane (PEM) fuel cells would be detrimental to the cell anode catalyst. The nature of the surface of this new type of catalysts has been studied experimentally by investigations of ultrathin Zn films on single-crystal Pd(111) surfaces as model systems and was discussed in the literature.3-6 A consensus was reached, that after annealing to temperatures around 500 K a PdZn alloy is formed in the near-surface region with a Pd:Zn ratio of 1:1. This surface exhibits a (3-domain) p(2×1)-structure, resembling a PdZn(111) surface, consisting of alternating rows of Zn and Pd atoms, with the Zn atoms lifted about 0.25 Å above the Pd atoms.6 This finding was made on films thicker than 1 monolayer (ML) and is in agreement with various calculations based on density-functional theory (DFT).7-9 Interestingly, for alloy films with a thickness of only 1 ML, the calculations of Koch et al.9 and Weirum et al.8 predict an inverted corrugation, with the Pd * To whom correspondence should be addressed. † University of Innsbruck. ‡ Fritz-Haber-Institute of the Max-Planck-Society. § Vienna University of Technology. | Purdue University.
atoms displaced outward and the Zn atoms displaced inward relative to a flat surface. Recently, a surprising observation was made by Rameshan et al.10 with respect to methanol steam reforming reactions with PdZn films on Pd-foils as catalysts. For reaction temperatures between 570 and 630 K the catalytic selectivity was found to change drastically from a highly CO2-selective state at 570 K to predominant CO production at 630 K, although the Zn content of the surface layer was only slightly reduced (from 52% to 48%). Thus, to explain this finding, one either has to assume that the catalytic behavior is extremely sensitive to the precise composition of the surface, or one has to consider other reasons to rationalize the selectivity change. Two explanations deserve particular attention: (i) Although the surface composition is basically unaltered upon annealing, the composition of the underlying layers changes considerably. The altered subsurface coordination of the surface atoms affects the electronic structure at the surface, which in turn changes the catalytic behavior. (ii) Although the surface composition varies only slightly, there are marked changes in the geometrical structure of the surface, which has a strong influence on the chemical reactions taking place on the surface. In view of the above-mentioned DFT results,8,9 both scenarios are obviously coupled, leading to the following possible explanation: In the temperature regime from 570 to 630 K the surface composition is only slightly altered, while more drastic changes occur in the second and deeper layers. Zn atoms of the deeper layers of the PdZn alloy start to diffuse into the Pd bulk, finally leaving behind an alloy film where Zn is mainly present in the surface layer with a Zn:Pd ratio close to 1:1, i.e., a situation close to the theoretically described PdZn monolayer film. According to DFT, the corrugation of such an alloy layer should then be inverted or at least reduced. The altered surface
10.1021/jp1008835 2010 American Chemical Society Published on Web 05/28/2010
Modifications of PdZn Layers on Pd(111) geometric and electronic structure induced by the altered subsurface composition are then responsible for the strong change of selectivity in methanol steam reforming. In this report we verify experimentally that after annealing to temperatures around 500 K a homogeneous PdZn alloy with near-surface and surface compositions close to 1:1 is formed, whereas beyond 550 K the scenario as described above sets in. A detailed XPS study of the formation of PdZn surface alloys on Pd(111) was recently performed by Bayer et al.3 To differentiate the topmost alloy layer from deeper layers they recorded Al KR XP spectra at grazing emission (80°) and at normal emission (0°). In the present article the composition and structure of the topmost PdZn surface layer were addressed more directly by applying low-energy ion scattering spectroscopy (LEIS) as well as polarization-modulated infrared reflection absorption spectroscopy (PM-IRAS) of adsorbed CO. Both methods are basically insensitive to subsurface layers. Furthermore, comparison with subsurface-sensitive Auger electron spectroscopy (AES) as well as depth profiling by synchrotronbased XPS with varying photon energy allowed us to probe surface vs subsurface composition. Finally, impact-collision ion scattering spectroscopy (ICISS) was applied to monitor the temperature-dependent evolution of surface corrugation, an effect that has not yet been examined experimentally for PdZn(111). Altogether, the new findings not only corroborate the existing picture of PdZn alloy formation but significantly refine our knowledge on this important model catalyst system. 2. Experimental Section LEIS and AES experiments were carried out in an ultrahigh vacuum (UHV) chamber with a base pressure below 10-10 mbar. Auger electrons were excited with a 3 keV e- beam at normal incidence and detected with a 4-grid retarding-field analyzer with an average detection angle of 26°. For monitoring the Zn and Pd intensities, the peak-to-peak heights of the differentiated MNN signals at 59 and 330 eV, respectively, were evaluated. LEIS data were taken with 5 keV Ne+ ions. The scattering angle was set to 160° and the scattering plane was normal to the surface along the (1j21j) azimuthal direction. LEIS is inherently an extremely surface-sensitive technique due to the high neutralization rate of low-energy ions scattered from deeper layers. This surface sensitivity can be enhanced to the top-layer exclusively if ions are incident at an angle of ψ ) 45° along the chosen azimuthal direction. In this measurement geometry atoms in deeper layers at or close to fcc or hcp lattice sites are hidden in the shadow cones produced by the top-layer atoms and thus are “invisible” to the incoming ion beam.6 Vibrational spectroscopy of adsorbed CO also detects the toplayer only, but the CO binding energy-and thus the C-O resonance frequency-may be affected by the subsurface composition/coordination. Surface vibrational spectra were acquired by polarization-modulation infrared reflection absorption spectroscopy (PM-IRAS), which provides surface-specific information about adsorbed molecules in the full range from UHV to ambient pressure. Because the effective surface intensity of s-polarized infrared light on a metal surface is basically zero, no surface absorption occurs and s-polarization spectra are thus gas phase spectra. In contrast, IR spectra acquired in ppolarization contain absorption contributions of both surface and gas phase species. Consequently, (p-s) IR spectra represent the vibrational characteristics of the adsorbed surface species. PMIRAS experiments were performed in a UHV-compatible highpressure cell, coupled to a UHV surface preparation/analysis system, equipped with LEED and XPS.11-13 CO (purity
J. Phys. Chem. C, Vol. 114, No. 24, 2010 10851 >99.997%) was passed over a carbonyl absorber cartridge and then introduced via a cold trap filled with liquid nitrogen in order to remove potential Ni- and Fe-carbonyl impurities. XPS data were acquired at beamline U49/2-PGM1 at BESSY II (Helmholtz-Zentrum Berlin) for photon energies between 120 and 1235 eV and normal electron emission. The setup there also allowed us to perform in situ photoelectron spectroscopy during methanol steam reforming at elevated gas pressures up to 1 mbar.10,14 The Pd(111) single crystal was cleaned by repeated cycles of Ar+ sputtering, annealing up to 950 K in UHV, exposure to O2, and final annealing to 950-1000 K. Surface cleanliness was either checked directly by XPS or by the presence of a strong oxygen desorption signal (rather than CO or CO2) during the final annealing procedure.15 Zn was evaporated from a home-build Knudsen-type evaporator (LEIS, AES, XPS) or by electron-beam evaporation (PMIRAS). In the AES and LEIS studies the coverage calibration was obtained from the LEIS Zn-uptake curves at T ) 150 K as presented in ref 6, whereas in the XPS work the coverage was estimated from the damping of the Pd3d XPS signal. In the PM-IRAS studies, the Zn coverage was monitored by a quartz crystal microbalance and by XPS. Coverages are given in monolayers, with one monolayer (ML) referring to the atom density of the Pd(111) surface, i.e., 1.53 × 1015 atoms/cm2. During Zn deposition the sample was kept at 150 K in the LEIS and AES experiments, and at 90 or 300 K in the PM-IRAS experiments. As the Zn layers are subsequently annealed to temperatures above 300 K, no pronounced effect of the deposition temperature was observed. For the XPS experiments Zn was deposited around room temperature, as the XPS setup did not allow for LN2 cooling. 3. Results and Discussion 3.1. Elemental Composition. The top part of Figure 1 depicts the evolution of the normalized Zn intensities, i.e. IZn/ (IZn + IPd), as obtained from LEIS and AES measurements when a 2.2 ML Zn film (deposited at 150 K) was annealed for 10 min to various temperatures. Whereas AES probes the composition of several layers, LEIS data were taken in the surface sensitive measurement geometry (angle of incidence 45° along ([1j21j])-azimuth). Thus, due to equal scattering efficiencies for Zn and Pd,6 the Zn portion in the topmost layer is directly given by IZn/(IZn + IPd). The evolution of the LEIS data with temperature was already discussed in ref 6. Four characteristic regimes have to be distinguished: In region I (up to 300 K) the surface consists basically of Zn, only small variations in surface composition take place upon annealing. In region II (300-400 K), alloy formation sets in, leading to an increase in Pd and a decrease in Zn atoms at the surface. Our temperature-programmed desorption data show that desorption of Zn is negligibly small (up to 800 K), in agreement with the findings of ref 8 that low-temperature desorption is only important for coverages above 2 ML. In region III between 400 and 550 K a plateau is observed in both the AES and the LEIS data. This is the stability regime of the (2×1) alloy film discussed in the introduction. The LEIS data-which directly yield the surface composition-show a Pd: Zn ratio close to 1:1 in this region. To obtain information on the composition of deeper layers, XPS depth profiling was performed for a multilayer Zn film (J10 ML) deposited at room temperature and annealed to 500 K. Figure 2 shows the evolution of the Pd3d and Zn3d signal as a function of photoelectron kinetic energy in the range 100 to 900 eV. XPS intensities were
10852
J. Phys. Chem. C, Vol. 114, No. 24, 2010
Stadlmayr et al.
Figure 2. XPS intensities of Pd3d and Zn3d signals versus kinetic energies of the photoelectrons of a multilayer Zn film (J10 ML) annealed to 500 K. Intensities are corrected for elemental sensitivities and normalized to the incident photon flux.
Figure 1. Top: Normalized intensity of the Zn signal as measured with low-energy ion scattering (LEIS, left axis) and Auger-electron spectroscopy (AES, right axis) for 2.2 ML of Zn/Pd(111). Note the faster decrease in the AES signal in region IV beyond 550 K. Middle: Difference in critical angles for backscattering from Pd and Zn, respectively, versus annealing temperature for 2.2 ML of Zn/Pd(111). Bottom: Peak position of the Pd 3d5/2 XPS signal versus annealing temperature for J10 ML of Zn/Pd(111).
corrected for photon-energy-dependent elemental sensitivities, using the photoionization cross sections calculated by Band et al.16 and normalized to the incident photon flux. It is clearly evident that in the energy range up to about 700 eV the Pd and Zn signals follow the same straight line, indicating a 1:1 composition throughout the film in the accessible depth range. As the data are already corrected for elemental sensitivities, the linear increase in intensity directly reflects the almost linear increase in information depth (or effective local attenuation length) with increasing kinetic energy from ∼3 Å at 100 eV to ∼11 Å at 1000 eV.17,18 Only above ∼700 eV do the Zn data points seem to level off due to the finite thickness of the PdZn alloy film. Region IV at temperatures higher than 550 K is the regime that is most relevant with respect to the strong changes observed in catalytic selectivity and which is the focus of this work. The LEIS signal (Figure 1a) shows a slow decrease due to Zn diffusion into the bulk of the Pd(111) substrate.6 The AES signal decreases too, but it is clearly evident that this decrease is much stronger. Above 700 K zinc can hardly be detected by our AES setup, whereas LEIS still shows non-negligible amounts of Zn
in the surface layer. These data clearly demonstrate that for temperatures exceeding 550 K the subsurface layers deplete faster in Zn than the topmost surface layer, leading to a situation that around 630 K can be described as a PdZn film with strong monolayer character on a Pd substrate. This can be inferred from a quantitative analysis of the Auger intensities depicted in Figure 1a: For simplification we neglect backscattering effects of the electron beam and make the usual assumption that the damping of outgoing Auger electrons follows an exponential LambertBeer-type law with material-independent attenuation lengths of 2.8 Å for the Zn-MNN (59 eV) and 5.3 Å for the Pd-MNN (330 eV) Auger electrons.17,18 The Auger cross sections of Zn and Pd were chosen with a ratio of 64:100 in order to reproduce the experimental Zn/(Zn + Pd) ratio of 0.28 at 500 K, assuming that this PdZn film contains a total of 2.2 ML of Zn with Zn: Pd compositions of 1:1 in subsurface layers (as derived from XPS, Figure 2) and a surface composition of 55:45 (as derived by LEIS, Figure 1a). For cross-checking of this calibration we also calculated the Zn AES intensity expected directly after deposition of 2.2 ML. At the growth temperature of 150 K simultaneous multilayer growth with a poisson height distribution of the Zn layer was proposed.6 For such a growth model our calculation reveals an AES intensity of 0.41, slightly higher than the experimental value of 0.38 in region I of Figure 1a (note that for AES the right-hand axis applies). Repeating the calculation for a monolayer PdZn with a surface composition of Zn:Pd ) 45:55 (as deduced from LEIS at T ) 630 K) an AES ratio of 0.12 is obtained, slightly lower than the experimental value of 0.14. The calculated value can be matched with the experimental value by assuming that on average each subsurface layer contains 0.1 ML of Zn. Increasing the Zn content of each subsurface layer to 0.2 ML would already yield an AES ratio of 0.17. Thus this analysis strongly supports the notion that after annealing to 630 K a state coming close to a “monolayer” of PdZn on Pd(111) is formed. However, it should also be clearly pointed out that this is a somewhat simplified terminus, since-as is evident from our AES analysis-Zn atoms are also present in the subsurface layers, although with lower concentrations as compared to the surface layer. Furthermore, also Bayer et al. found convincing evidence (by comparison of XPS data at normal and grazing electron) that Pd and Zn exist in a variety of stoichiometries due to concentration variations
Modifications of PdZn Layers on Pd(111)
J. Phys. Chem. C, Vol. 114, No. 24, 2010 10853
Figure 3. XPS spectra of the Pd 3d5/2, Zn 3d, and valence band (VB) region, taken after successive anneals (5 min each) of a multilayer Zn film (J10 ML) to various temperatures. Photon energy was 650 eV. Spectra are unsmoothed data, normalized to the same photon flux. A Shirley background has been subtracted from the Pd 3d signal. To avoid confusing crossing of curves, the 500, 540, and 580 K spectra in the VB region were reduced by a factor of 0.9.
in the near-surface layers.3 Our XPS data discussed below also support this picture. At 700 K an apparent discrepancy exists in our AES data. While our calculation (for 0.3 ML of Zn in the top layer and no subsurface Zn) yields an AES ratio of 0.08, in the measurements the Zn signal is only 0.03. However, it has to be noted that the 59 eV Zn MNN signal resides on a steeply rising background, which-for low Zn contents-makes detection of the Zn signal difficult and results in an underestimation of the true peak-to-peak height of the differentiated AES signal. Further support for relatively strong changes taking place between 550 and 630 K is obtained from a detailed analysis of the Pd3d XPS signal upon annealing of a multilayer Zn film (J 10 ML) as shown in Figure 3, left panel. At annealing temperatures between 400 and 500 K the Pd3d intensity increases steeply due to the simultaneous actions of alloy formation and Zn desorption. Note that for Zn coverages beyond 2 ML pronounced Zn desorption occurs in the temperature window from 450 to 500 K.4,8 As a consequence no clear intensity plateau is observed for temperatures between 400 and 550 K-in contrast to the LEIS and AES data (which were taken for only 2.2 ML of Zn, where desorption is essentially not taking place). Only in the small temperature interval from 500 to 550 K does the peak height stay constant, see dashed line in Figure 3. Beyond 550 K again an intensity increase is observed. On the basis of the previous results and in agreement with literature3 this increase is attributed to diffusion of Zn into the Pd bulk. The diffusing Zn atoms mainly stem from subsurface layers, resulting in the gradual formation of a PdZn film with increasing monolayer character. The increase of intensity above 550 K is paralleled by a stronger energy shift of the Pd 3d5/2 peak maximum toward the binding energy of 335 eV of clean Pd(111), see Figure 1c. Again this indicates that above 550 K the chemical surrounding of the probed Pd atoms changes considerably. Note that the Pd3d peak shows a continuous energy shift without clearly separable components (except at the lowest annealing temperatures). Following the conclusions of Bayer et al.3 we attribute this observation to the simultaneous
existence of Pd atoms in a variety of Zn coordination states, the differences of which are too small to be experimentally resolved. With increasing annealing temperature the relative abundances of the various species change, producing the observed continuous shift of the peak maximum. Note that in the temperature range around 630 K, where a PdZn monolayerlike film has been shown to exist, all our spectroscopic results (LEIS, AES, XPS) show a rather smooth behavior with respect to temperature increase, indicating that the monolayer-like film is not a particularly pronounced and stable structure, but just an intermediate state during Zn migration into the Pd bulk. In summary, the XPS data indicate a conversion from the asdeposited film, with an alloyed interface and Zn-layers on top (having characteristic Pd3d binding energies around 336.5 eV), via a 1:1 PdZn multilayer alloy film at 500 K (Ebind ≈ 336.1 eV) and a PdZn film with strong monolayer character at temperatures around 630 K (Ebind ≈ 335.4 eV) toward the clean Pd(111) surface (Ebind ≈ 335.1 eV). Our findings are in full agreement with the XPS data of Bayer et al.3 Similar to the experiments shown in the present paper, they prepared a 3 ML Zn film by deposition at low temperature (∼105 K) and subsequently annealed for 30 s to various temperatures up to 700 K. Annealing the as-deposited films to 500 K resulted in the appearance of a high-binding energy component (336.1 eV) of the Pd3d5/2 core level. This component was attributed by the authors to Pd atoms in the environment of an 1:1 PdZn bulk-like alloy. In addition, a second smaller component was observed at 335.5 eV, i.e., closer to the bulk feature at 335.04 eV. This second component was assigned to Pd atoms in a “Pd-rich” phase (as compared to a PdZn 1:1 alloy), i.e., to the Pd atoms at the PdZn/Pd-bulk interface. Annealing to 700 K caused the high-binding energy PdZn “bulk” feature to disappear. Only one component could be resolved (besides the signal from the Pd bulk substrate), whose energy was virtually identical with that of the former “Pd-rich” component. On the basis of the present experiments we attribute the dominating contribution to this signal to Pd atoms of the PdZn “monolayer”, which are in a Pd-rich environment due to the presence of the underlying Zn-depleted Pd(111) substrate.
10854
J. Phys. Chem. C, Vol. 114, No. 24, 2010
The Zn3d region (Figure 3, middle panel) shows a complementary behavior to that of the Pd3d signal: At annealing temperatures beyond 400 K the Zn signal decreases drastically due to Zn desorption and multilayer alloy formation. Only small shifts in energy are observed up to 550 K. From 550 to 620 K strong peak shifts toward lower binding energies indicate formation of the monolayer-like PdZn alloy. Finally, the intensity decreases due to further Zn dissolution in Pd bulk, while the peak energy stays almost constant. Similarly, the valence band region (Figure 3, right panel) shows only minor changes from 500 to 550 K, while at lower and higher annealing temperatures more distinct changes are observed. In particular above 540 K the Pd4d level shows a pronounced shift from ca. -2 eV toward EF, leading to a marked increase of the density of states at the Fermi level. The shift of the Pd4d peak intensity toward EF upon annealing beyond 500 K was already reported in ref 3. The temperature region close to and above 800 K deserves another comment. As is evident from the LEIS, AES, and XPS data, there are at most a few tenths of a monolayer of Zn present in the near-surface region at temperatures above 750 K. On the other hand, using temperature-programmed desorption, pronounced Zn desorption is observed beyond 800 K,4,8 with peak areas corresponding to Zn amounts up to and even beyond 2 ML. Obviously, zinc desorbing above 800 K is due to Zn atoms which were dissolved in the Pd bulk at lower temperatures.8 To explain the different methanol steam reforming selectivities of the multilayer and monolayer PdZn surface alloys, observed by Rameshan et al., it is clearly required that the model catalyst surfaces are structurally very homogeneous. The coexistence of several different active sites/ensembles would otherwise lead to less pronounced differences in CO2 selectivity. To characterize the adsorption sites of the 1:1 PdZn surface alloys and, particularly, to monitor their thermal evolution we have carried out PM-IRAS spectroscopy using CO as the probe molecule. Figure 4a shows a series of PM-IRAS spectra, acquired at a CO pressure of 10-6 mbar at 150 K, after annealing (∼10 min) a PdZn multilayer alloy (2 ML Zn deposited at 300 K) to the indicated temperatures. After annealing to 473 K, which is in the center of the alloy stability window discussed above (region III), a single sharp peak at ∼2068 cm-1 (fwhm of 9 cm-1) was observed. In light of the desorption temperature of CO from PdZn (∼220 K) the surface is saturated with CO (Θsat ) 0.5 ML), with each Pd atom occupied by a CO molecule.19 This vibrational resonance is characteristic of CO adsorbed on-top of individual Pd atoms in the Pd rows of the PdZn surface alloy. The vibrational frequency agrees well with the on-top CO peak observed for the selective state of PdZnZnO powder catalysts, obtained upon hydrogen reduction of Pd nanoparticles supported on ZnO.20 This clearly demonstrates the structural homogeneity of the entire multilayer alloy surface (the diameter of the IR beam is several mm), because in the case of the remaining Pd patches bridge-bonded CO (at ∼1955 cm-1) and on-top CO (at ≈2085 cm-1) would additionally occur at these p/T conditions.12 The lower wavenumber (ca. -15 cm-1) of on-top CO on PdZn, as compared to Pd(111), is due to a charge transfer from Zn to Pd (according to DFT9,21), leading to better backbonding of Pd to the CO molecules. This increases the CO adsorption energy but weakens the internal C-O bond, thus leading to lower wavenumbers, in agreement with DFT.22 Figure 4a also shows the effect of higher annealing temperatures. In agreement with the LEIS/AES/XPS measurements, there is hardly any change up to 573 K. Annealing to 623 K induced a small shift of the on-top CO band to 2073 cm-1,
Stadlmayr et al.
Figure 4. (a) PM-IRAS spectroscopy on 2 ML of Zn/Pd(111). Spectra were acquired at 150 K in 10-6 mbar CO, after successive annealing to the indicated temperatures. The IR intensity in the frequency range of multiple coordinated CO (below 2000 cm-1) was enlarged by a factor of 3 in the lowest three spectra. (b) Elevated pressure spectra of CO adsorption on 2 ML of Zn/Pd(111) annealed to 550 K, and on Pd(111); see text.
paralleled by the evolution of a broad weak peak around 1900 cm-1. These changes indicate the formation of the Zn-lean monolayer-like surface alloy. Considering DFT calculations of CO adsorbed on different Pd-Zn ensembles,22 the peak around 1900 cm-1 originates from multiple-coordinated (hollow) CO in a Pd-rich environment. Annealing to 773 K led to an IR spectrum that is identical to a corresponding spectrum on Pd(111), with bridge and on-top bonded CO at 1955 and 2085 cm-1, respectively. The spectrum acquired after annealing to 673 K represents an intermediate state between a Zn-lean surface alloy (623 K) and a Pd surface (773 K). The IR spectra thus not only corroborate the surface characterization data but also indicate (although PM-IRAS is not spatially resolved) that the compositional changes occur homogeneously on the entire model catalyst surface. If not, a superposition of spectra characterizing different surface states would have been observed, which was not the case (except maybe at 673 K). Taking advantage of the elevated pressure capability of PMIRAS, Figure 4b displays CO adsorption spectra at pressures of 1 and 10 mbar, contrasting the well-established PdZn surface
Modifications of PdZn Layers on Pd(111)
Figure 5. Angular dependence of the backscattering from Pd and Zn atoms respectively after (a) annealing 2.2 ML of zinc to 500 K, (b) annealing 2.2 ML of Zn to 700 K, (c) annealing 1.2 ML of Zn to 475 K, and (d) annealing 0.6 ML of Zn to 475 K. All spectra are normalized to the same height. The data in panel a are relatively noisy due to a shorter data accumulation time. The insets in the figures depict schematically the situations at the critical angles for backscattering from Pd and Zn atoms respectively for a corrugated (Zn-up/Pd-down) and a flat PdZn surface. For illustration purposes the corrugation is drawn largely exaggerated.
alloy (obtained after annealing to 550 K) with Pd(111).12,23 The spectra were taken at 195 K, close to the desorption temperature of CO from PdZn.19 The 1 mbar spectrum on PdZn is again characteristic for the CO saturated surface (0.5 ML of CO). Increasing the pressure to 100 mbar had no effect (not shown). Consequently, even at elevated CO pressure the PdZn 1:1 surface remained well-defined with no adsorption sites other than on-top on individual Pd atoms being populated. For ease of comparison, and to demonstrate the differences to a pure Pd surface, a corresponding spectrum on Pd(111) is included, with the spectrum at 10 mbar CO being representative for pressures up to 100 mbar. 3.2. Surface Corrugation. Information about the structural changes associated with the subsurface Zn depletion above 550 K was obtained from the angular dependence of the ionscattering intensities (ICISS).24,25 Figure 5 depicts the intensities of both the Zn and the Pd backscattering signals recorded as a function of the angle of incidence ψ for 2.2 ML Zn films annealed to 500 (Figure 5a) and 700 K (Figure 5b), respectively. Close to gracing incidence (ψ ) 0°) the backscattering intensities are low, since each surface atom is located in the shadow cone cast by its immediate “sinistral” neighbor (assuming that the ion beam impinges from the left). Accordingly, it cannot be hit by the incoming ions in an almost head-on collision as required for a backscattering angle of 160°. Upon increasing the angle of incidence, at some point a situation is reached where the edge of the shadow cone is swept over the neighboring “rightward” atoms of the shadow-casting atoms. Now these atoms become visible to the ion beam, causing a steep increase in the signal intensity at the so-called critical angle ψc. The critical angle ψc is defined as the angle at 50% of the maximum intensity increase. Its value depends on the interatomic vector between the shadow-cone-producing atom and the final backscatterer relative to the incoming ion beam. ψc is particularly sensitive to the atomic coordinates in the scattering plane perpendicular to the incoming beam, i.e., to atomic
J. Phys. Chem. C, Vol. 114, No. 24, 2010 10855 displacements perpendicular to the surface when working close to grazing incidence. It has to be mentioned that the width of the shadow cone and in turn the critical angle depend also on the nuclear charge of the shadow-casting atom.25 However, the difference between Zn and Pd is small (
