Initial Oxidation of a Rh(110) - American Chemical Society

Jun 25, 2005 - Science Park, I-34012 BasoVizza-Trieste, Italy, and Department of ... Nanostructured Materials, UniVersity of Trieste, I-34127 Trieste,...
0 downloads 0 Views 3MB Size
Download PDF
J. Phys. Chem. B 2005, 109, 13649-13655

13649

Initial Oxidation of a Rh(110) Surface Using Atomic or Molecular Oxygen and Reduction of the Surface Oxide by Hydrogen P. Dudin,† A. Barinov,† L. Gregoratti,† M. Kiskinova,*,† F. Esch,‡ C. Dri,‡,§ C. Africh,‡,§ and G. Comelli‡,§ Sincrotrone Trieste, Area Science Park, I-34012 BasoVizza-Trieste, Italy, Laboratorio TASC-INFM, Area Science Park, I-34012 BasoVizza-Trieste, Italy, and Department of Physics and Center of Excellence for Nanostructured Materials, UniVersity of Trieste, I-34127 Trieste, Italy ReceiVed: February 15, 2005; In Final Form: May 19, 2005

The formation conditions, morphology, and reactivity of thin oxide films, grown on a Rh(110) surface in the ambient of atomic or molecular oxygen, have been studied by means of laterally resolved core level spectroscopy, scanning tunneling microscopy and low energy electron diffraction. Exposures of Rh(110) to atomic oxygen lead to subsurface incorporation of oxygen even at room temperature and facile formation of an ordered, laterally uniform surface oxide at ∼520 K, with a quasi-hexagonal structure and stoichiometry close to that of RhO2. In the intermediate oxidation stages, the surface oxide coexists with areas of high coverage adsorption phases. After a long induction period, the reduction of the Rh oxide film with H2 is very rapid and independent of the coexisting adsorption phases. The growth of the oxide film by exposure of a Rh(110) surface to molecular oxygen requires higher pressures and temperatures. The important role of the O2 dissociation step in the oxidation process is reflected by the complex morphology of the oxide films grown in O2 ambient, consisting of microscopic patches of different Rh and oxygen atomic density.

1. Introduction Identification of the active chemical state of transition metal surfaces during catalytic oxidation reactions is a key issue in understanding the reaction mechanisms. According to recent theoretical and experimental studies of the initial oxidation stages of several catalytically relevant transition metals (Ru, Pd, Rh, Ag, Pt) the formation of bulk or surface oxide phases is common phenomenon under realistic catalytic conditions.1-3 In the case of Ru, rutile RuO2 patches formed on the catalyst have been considered as the active phase in CO oxidation, providing undersaturated Ru sites, where the CO adsorbs and reacts with oxygen.4-8 The oxide formed on the Rh catalyst does not expose active Rh sites and was supposed to act as a source of the adsorbed oxygen, which reacts with the CO.9 The existence of another direct reaction pathway was suggested for the reduction of the Rh oxide films by H2.10 The recent studies on the oxidation of close-packed Rh(111) and Rh(100) surfaces and the theoretical atomistic description of the microscopic processes yielding a crystalline bulk Rh2O3 phase considered the existence of several precursor states, some of them being already identified by spectroscopic and structural methods.1,11,12 The important steps toward the bulk oxide formation after adsorption are incorporation of oxygen in the subsurface region and formation of the so-called trilayer surface oxide structures, which on some transition metals may kinetically hinder the bulk oxide formation.13 We have extensively studied the variety of oxygen adsorption phases on the Rh(110) surface, most of them involving major surface reconstructions.14-16 The dissociative adsorption of O2 * Corresponding author. E-mail: [email protected]. Telephone: +39-040-3758549. Fax: +39-040-3758565. † Sincrotrone Trieste. ‡ Laboratorio TASC-INFM. § University of Trieste.

at temperatures higher than 400 K induces (1 × n) missing row reconstructions of the Rh(110) surface. Subsequent (2 × 2)p2mg, c(2 × 6), c(2 × 8) and c(2 × 10) LEED structures were observed with increasing oxygen coverage from 0.5 to 0.8 monolayer (ML), involving (1 × 2), (1 × 3), (1 × 4), and (1 × 5) missing-row reconstructions; i.e., each nth [11h0] row is missing. One ML corresponds to the number of Rh atoms on the Rh(110)-(1 × 1) surface, 9.8 × 1014 atoms/cm2. A distinct feature is that at coverage g0.5 ML the O adatoms always occupy 3-fold sites in a zigzag arrangement along the [11h0] rows. The unreconstructed Rh(110)-(1 × 1) surface can accommodate 1 ML of oxygen forming a (2 × 1)p2mg structure, which readily occurs at adsorption temperatures e300 K. Another adsorption phase with oxygen coverage of 0.9 ML is the metastable (10 × 2) structure, where each Rh surface atom is coordinated with four oxygen atoms. It can be obtained by exposing the O-(2 × 2)p2mg surface to O2 at low temperature, followed by mild annealing.16 The [11h0] periodicity of the (10 × 2) structure is due to the absence of two out of each 10 Rh atoms in the [11h0] rows, accompanied by an increase of the Rh-Rh interatomic distance.16,17 The present study is focused on the initial stages of oxide formation on the Rh(110) surface, which follow the formation of the adsorption phases described above. We used laterally resolved core level spectroscopy, scanning tunneling microscopy (STM) and low energy electron diffraction (LEED) to identify the composition and structure of the oxygen-containing phases formed on the Rh(110) surface under different oxidation conditions. In particular, we explored (i) the role of the O2 dissociation step, comparing the oxidation state of the Rh(110) surface and the morphology of the thin oxide films after exposures to atomic or molecular oxygen and (ii) the reactivity of the surface oxide and the adsorption phases during reduction with H2.

10.1021/jp0508002 CCC: $30.25 © 2005 American Chemical Society Published on Web 06/25/2005

13650 J. Phys. Chem. B, Vol. 109, No. 28, 2005

Dudin et al.

2. Experimental Section The laterally resolved core level spectroscopy experiments were carried out using the scanning photoelectron microscope (SPEM) at the ELETTRA light source.18 The SPEM uses zone plate focusing optics to produce the X-ray microprobe onto the sample (0.04 µm2 in the present study) and a hemispherical electron analyzer with large angle acceptance lenses. The SPEM can be operated in two modes: imaging by collecting photoelectrons within a selected kinetic energy window, while scanning the sample with respect to the focused beam (spectroimaging), and conventional energy dispersive spectroscopy from the illuminated area (microspot spectroscopy). When the element under consideration is present in a single chemical state the spatial variation in the contrast of the images reflects the variation of the photoelectron yield, which is a measure of the local concentration of the element. When the element is present in several chemical states the electron level chemical shifts can be used for mapping the lateral variations in the chemical state. Our photoelectron analyzer was equipped with a 48-channel detector, allowing to map different chemical states with a single scan and to reconstruct the spectrum corresponding to the selected energy window (spectroimaging) by applying the necessary processing procedures.18 In the present study the processing of the images always involved subtraction of the secondary electron background signal, thus the resulting contrast corresponds to the density of the atoms in the chemical phase under consideration. Using the spectroimaging mode reliable spectroscopic information can be obtained for features g1 µm2, compared to 0.04 µm2 in microspot spectroscopy mode. The SPEM measurement station consisted of several interconnected vacuum chambers for sample preparation, including a highpressure (10-4 to 10-1 mbar) chamber. The preparation chambers were equipped with LEED, an Auger electron spectrometer (AES), an ion gun, and a gas inlet system. The atomic O flux was provided by a Tectra electron cyclotron resonance plasma source. The STM experiments were carried out with a variable temperature Omicron VT-STM in a separate UHV chamber, also equipped with a LEED-AES system and the same resonance plasma source. The Rh(110) sample was cleaned using the well-established procedure of repeated sputtering-annealing and oxidationreduction cycles.19 The oxidation using atomic oxygen was carried out at pressures in the 10-6 mbar range and different temperatures starting from room temperature. Since the cracking efficiency was not calibrated, only the upper exposure limits were quantified. The oxidation in O2 ambient was performed in the high-pressure cell at pressures in the 10-4 mbar range and temperature ∼750 K, the same conditions used for oxidation of the Rh(111) and Rh(100) surface.11,12 The SPEM measurements were performed with photon energy of 645 eV, lateral resolution of 0.2 µm and an energy resolution of 0.2 eV for Rh 3d spectra and 0.4 eV for the O 1s spectra. According to the universal curve for the electron mean free path, the effective escape depths for the O 1s and Rh 3d photoelectrons for our geometry are ∼2.8 and 4.2 Å, respectively. This limits the probing depth to the top few layers, ∼6 and 12 Å, respectively. 3. Results and Discussion 3.1. Oxidation of Rh(110) by Atomic Oxygen. The oxidation of the Rh(110) surface with atomic oxygen was studied at different substrate temperatures. This allowed us to obtain different oxidation stages and to examine the conversion between them. The O 1s and Rh 3d5/2 SPEM images measured after exposure of Rh(110) to atomic oxygen were featureless,

Figure 1. Surface oxide formed by an exposure of Rh(110) at 520 K to atomic oxygen: (a) c(2 × 4) LEED pattern; (b) 4.0 × 4.0 nm2 STM image showing the quasi-hexagonal arrangement of the top O atoms in the surface oxide. Imaging conditions: Vb ) -0.6 V (sample with respect to the tip), I ) 0.4 nA, T ) 520 K. (c) Deconvoluted O 1s spectrum of the oxide film. The dotted lines indicate the components corresponding to the oxygen atoms at the surface, Os, and “below the surface”, Ob (d) Deconvoluted Rh 3d5/2 spectra of the oxide film (bottom) and of a clean Rh(110) surface (top). The dotted lines indicate the components corresponding to the Rh atoms in the bulk, Rhb, on the surface, Rhs, in the oxide, Rhox and at the metal/oxide interface, Rhi.

indicating laterally uniform oxygen distribution at our microscopic length scales. A well-ordered thin oxide film was formed only by exposures at elevated temperatures. The characteristic LEED pattern, STM image, and Rh 3d5/2 and O 1s photoemission spectra of the surface oxide, formed by exposing to atomic oxygen at ∼520 K, are shown in Figure 1. The LEED pattern in Figure 1a can be described as a c(2 × 4) structure with brighter hexagonally arranged spots. The STM image in Figure 1b shows the quasi-hexagonal arrangement of the top oxygen atoms of the oxide film. The structural model of this surface oxide, based on a detailed STM study supported by theoretical calculations, will be reported in ref 20. In brief, the surface oxide grown on the Rh(110) surface resembles the structure of those formed on the Rh(111) and Rh(100) surfaces and described in terms of O-Rh-O trilayers.11,12 This is confirmed by the O 1s and Rh 3d5/2 spectra of the oxide film on the Rh(110) surface (Figure 1c,d), which are very similar to those reported for the surface oxide on the Rh(111) surface.12 The low and high binding energy (BE) components in the O 1s spectrum reflect the photoemission from the oxygen atoms at the surface and “below the surface” (“b”), Os (-528.8 eV) and Ob (-529.9 eV), respectively. Because of the small escape depth of the O 1s photoelectrons, the O from the surface contributes ∼70% to the total O 1s intensity. The high BE component in the Rh 3d5/2 spectrum reflects the highly O coordinated Rh atoms in the surface oxide, Rhox (-308.0 eV). The lower BE components at -307.0 and -306.8 eV are assigned to emission from the bulk Rh atoms, Rhb, and the Rh atoms at the oxide/Rh interface, Rhi, respectively.12 We would like to note that due to the limited spectral resolution of our experimental setup the Rhi and Rhb components overlap, which introduced uncertainty in resolving

Initial Oxidation of a Rh(110) Surface

J. Phys. Chem. B, Vol. 109, No. 28, 2005 13651

Figure 2. Evolution of the Rh 3d5/2 and O 1s photoemission spectra with increasing temperature after exposure of Rh(110) to 300-350 L atomic oxygen at 300 K. The dotted lines indicate the energy positions of the Rh 3d components corresponding to the surface oxide, Rhox and the metal/oxide interface, Rhi, and the O 1s components corresponding to the oxygen atoms at the surface, Os, and “below the surface” , Ob. The Rh 3d5/2 bulk component is plotted with a dashed line. The other components are related to the coexisting adsorption phases.

Figure 3. STM image (40 × 40 nm2) showing the coexistence of adsorption and oxide phases, formed after exposure of Rh(110) to atomic oxygen at 520 K. The arrows indicate areas with surface oxide and adsorbed oxygen. Imaging conditions: Vb ) -0.6 V, I ) 1.0 nA.

the relative contribution of these two components. However, the relative intensity of the well resolved Rhox component is the most important parameter and is used as a fingerprint of the Rh oxidation state in the present study. The top Rh 3d5/2 spectrum in Figure 1d is from the clean Rh(110)-(1 × 1) surface taken before the oxygen exposure. The Rhs and Rhb components used to fit this spectrum account for emission from the surface and bulk Rh atoms, respectively. Comparing the two deconvoluted spectra in Figure 1d one can see that the intensity of the Rhox component is about 1.5 times the intensity of the surface component, Rhs. This indicates that approximately 1.5 ML Rh atoms reside in the formed oxide layer, comparable with the Rh atomic density in the O-Rh-O trilayer of 1.25 ML.20 When the Rh(110) surface is exposed to atomic oxygen at room temperature the (1 × 1) LEED pattern converts into a diffuse background, which indicates disorder of the topmost substrate layers. Upon annealing up to ∼520 K the c(2 × 4) hexagonal pattern appears, although not as sharp as the one in Figure 1a, obtained by exposing to atomic oxygen at 520 K. Figure 2 shows the evolution of the O 1s and Rh 3d5/2 spectra of the Rh(110) surface exposed to atomic oxygen at room temperature followed by annealing to different temperatures. The deconvolution of the Rh 3d5/2 spectra is performed using the established components for a clean Rh(110) surface, the O adsorption phases and the surface oxide.12,19 The initial Rh 3d5/2 spectrum requires three O-related components, which are present in the other spectra as well. Two of the components closely resemble Rhox and Rhi of the c(2 × 4) surface oxide in Figure 1c, whose energy positions are marked by the dotted lines. The presence of these components indicates that the incorporation of oxygen occurs already at room temperature and results in a disordered phase, where many Rh atoms have O coordination similar to that of the Rh atoms in the surface oxide. The third component has an intermediate binding energy (-307.4 eV), which corresponds to the adsorption phases.19 Upon annealing to 520 K, the Rhox component develops further and grows at the expense of the component related to the adsorption phases. Judging from the relative intensity of the Rhox, the Rh atoms involved in the surface oxide phase are ∼0.8 ML, implying that about 60% of the surface is covered by the surface oxide. The

Rhox component loses intensity upon further annealing, whereas the Rhb component grows, the final Rh 3d5/2 spectrum resembling those of the adsorption phases. In fact, the LEED pattern after the last annealing is a mixture of the (10 × 2), c(2 × 2n) and (2 × 1)p2mg adsorption structures. The O 1s spectrum measured after room-temperature exposure is rather broad. However, it also contains the oxide-related features, Os and Ob, marked by the dotted lines. During the transition from a disordered (300 K) to an ordered state (520 K), the O 1s spectrum evolves toward development of more distinct oxide features, but, as in the case of the Rh 3d5/2 spectra, the rather prominent “non-oxide” component between the two oxide peaks remains. This component resembles the low energy O 1s peak of adsorbed O bonded along the “1 × 2” rows on the (1 × n) reconstructed surfaces.19 By further increasing the temperature the O 1s spectrum loses intensity, because of the gradual removal of the oxide components. The oxide components partly convert into those corresponding to O adsorbed along the “1 × 2” and “1 × 1” rows, the latter positioned at ∼ -530.0 eV,19 very close to the Ob (-529.9 eV) of the oxide. Very similar O 1s and Rh 3d5/2 spectra were also measured after shorter exposure to atomic oxygen (150-200 L) at 520 K. It should be noted that the changes in the spectra in Figure 2 cannot be related to the loss of oxygen via thermal desorption, because the onset of O2 thermal desorption from saturated adsorption layers and thin Rh oxide films starts above 700 K. 15,21 The adsorption origin of the “non-oxide” components in the O 1s and Rh 3d5/2 spectra was confirmed by the STM images, obtained under comparable oxidation conditions. The STM image in Figure 3 shows the coexisting surface oxide and adsorption regions. The quasi-hexagonal oxide areas appear 0.4 Å higher than the surrounding areas with O-(2 × 1)p2mg, O-c(2 × 2n) and O-(n × 2) adsorption phases. The brightest features are segments of added Rh rows: they are aligned along the close packed [11h0] rows of the adsorption structures and decorate the boundaries of the oxide domains. In brief, the interaction of atomic oxygen with the Rh(110) surface reveals that the incorporation of oxygen subsurface can readily occur at room temperature, but the organization of the accumulated oxygen into an ordered surface oxide needs higher

13652 J. Phys. Chem. B, Vol. 109, No. 28, 2005

Figure 4. Evolution of the O 1s and Rh 3d5/2 photoemission spectra with increasing H2 exposure at 320 K. The initial surface was prepared by exposure of Rh(110) to 300 L atomic oxygen at 450 K followed by annealing to 520 K. The dotted lines indicate the energy positions of the Rh 3d components corresponding to the surface oxide, Rhox, and the metal/oxide interface, Rhi, and the O 1s components corresponding to the oxygen atoms at the surface, Os, and “below the surface” , Ob. The Rh 3d5/2 bulk component is plotted with a dashed line. The other components are related to the coexisting adsorption phases.

temperatures. Since no thermal desorption of oxygen occurred at temperatures below 700 K, the decomposition of the oxide film above 520 K indicates a loss of oxygen via deeper penetration into the bulk, in agreement with the exponential increase of the absorption rate of O in the Rh substrate with increasing substrate temperature, reported in ref 21. 3.2. Reduction of the Rh Oxide Films by Reduction with H2. The reactivity of the surface oxide, formed by exposure to atomic oxygen, has been probed by H2 reduction. Since at the reaction temperature used in the present study the reaction product H2O immediately desorbs from Rh surfaces,15,22 the O 1s and Rh 3d5/2 spectra are reliable fingerprints for the compositional changes in the surface oxide and the adsorption phases during the reduction process. Figure 4 shows the evolution of the O 1s and Rh 3d5/2 spectra of the surface oxide with increasing H2 exposure at 320 K. The initial O 1s and Rh 3d5/2 spectra also contain small components indicating coexistence of adsorption phases. The onset of reduction is preceded by a long induction period (exposures up to ∼30 L), during which the O 1s and Rh 3d5/2 spectra remain practically unchanged. After the onset of reduction the oxide components disappear from the O 1s and Rh 3d5/2 spectra within a very narrow exposure window. The final O 1s spectrum resembles those of the adsorption phases involving the (1 × n) reconstructions with oxygen coverage less than 0.8 ML.19 Accordingly, the Rh 3d5/2 spectrum measured after the final exposure contains only the components corresponding to the single and double O-coordinated Rh atoms in the adsorption phase. Complete reduction of the remaining adsorbed oxygen, leading to extinction of the O 1s peaks, occurred only after increasing the temperature to ∼360 K. The high reactivity of the oxide film is confirmed by the STM results. The STM experiments were carried out at 370 K, since it was easier to obtain atomic resolution. This can be tentatively attributed to temperature effects on the tip stability and/or on the conductivity of the oxide film. The STM image in Figure 5a illustrates the surface morphology before the reaction, with a patch of the 1 ML O-(2 × 1)p2mg adsorption structure in the center surrounded by regions of surface oxide. Exposures of

Dudin et al. this surface up to ∼18 L of H2 did not cause visible changes in the STM image, which is in fair agreement with the photoemission data in Figure 4. Figure 5b shows the first new features which appear as “bright dashes” homogeneously distributed over the oxide regions (the apparent height of these features is up to 1.0 Å), whereas the O-(2 × 1)p2mg adsorption structure remains still intact. Similar “dashlike” features, observed during the O-induced reconstructions of the Rh(110) surface, were attributed to mobile Rh-O species.23 The mechanism of creation of similar mobile Rh-O species on top of the oxide phase is not clear. However, their delayed appearance may be attributed to the onset of H2 dissociation on some active sites followed by diffusion and penetration of the highly mobile H atoms subsurface, destabilizing the lattice of the thin oxide film, as also suggested in ref 10. It is interesting to note that after a few images (∼7 min) these “dashlike” features disappear. Even though most of the surface still preserves the initial quasihexagonal structure, new dark features appear inside the oxide region. They resemble lattice vacancies observed in the oxide film and appear as a c(2 × 4) ordered array of defects (missing O and/or Rh atoms) as illustrated in Figure 5c. Such organization of the created defects points to a correlation between their formation energy and their registry with respect to the underlying Rh(110) lattice. The subsequent image 5d shows how a reduction front passes within the acquisition time (32 s) of a single frame. It appears as a diagonal line, indicating that the velocity of the front propagation is slower than the horizontal line scan velocity of the STM instrument. The sharp, atomically resolved edge in the bottom part of the figure clearly divides still nonreduced oxide region ahead of the front from the structurally changed reduced area behind. Apparently the Rh atoms building the surface oxide have rearranged into a complete Rh(110)-(1 × 1) layer with added [11h0] Rh rows above, resulting from the 25% higher density of Rh in the surface oxide with respect to the Rh(110) surface.20 The Rh atoms behind the front exhibit a transient mobility for a few seconds (corresponding to several scan lines) before reorganizing in added rows, which are stabilized by the remaining oxygen. The reduction front also penetrates inside the coexisting the O-(2 × 1)p2mg adsorption phase. The dashlike features appearing within the O-(2 × 1)p2mg island immediately behind the reaction front are attributed to the rearrangement of the oxygen atoms into O-O pairs, after partial reduction of the oxygen amount by the ongoing reaction, as previously observed in refs 17 and 24. The appearance of O pairs as dashes and not as defined units is related to the higher reaction temperature used in the present study. The reduction of oxygen from the added-row adsorption phase (a product of the oxide destruction) takes place with a slower rate (see images 5(d,e)), following the same pathway observed for the O-(2 × 2)p2mg surface in ref 25. In summary, the STM images evidence the accumulation of defects on the surface oxide during the induction period followed by a fast propagating reduction front. It reduces the oxide to oxygen adsorption phases, which also undergo reduction in a following stage. The facile and effective reduction of the Rh oxide film with H2 supposes that the surface oxide interacts directly with H and is not only a source of O to the active adsorption phase, as suggested in the case of CO oxidation.9 The first H atoms are produced by H2 dissociation on some defects resulting in formation of nuclei of reduced Rh which allows further H2 dissociation and triggers the ignition and propagation of reduction fronts, as was observed for removal of oxygen by H2 from saturated adsorption layers on Rh(110).26 Nearly identical

Initial Oxidation of a Rh(110) Surface

J. Phys. Chem. B, Vol. 109, No. 28, 2005 13653

Figure 5. STM images (15.5 × 16.0 nm2) taken during H2 reduction at 370 K (pH2 ) 1 × 10-8 mbar, corrected for sensitivity and the screening factor of the tip of 5, as estimated in ref 23). The insert in (c) is a zoom of the area, where the created c(2 × 4) ordered defect structure has developed. The white ellipse in (d) indicates the reaction front (propagating downward) and behind it the diffusing Rh atoms, which rearrange in a few scan lines to form the (1 × 2) reconstruction. H2 exposures: (a) 17.7 L, (b) 19.3 L, (c) 23.1 L, (d) 23.4 L, (e) 23.7 L, (f) 28.6 L. Imaging conditions: 32 s/image, Vb ) -0.6 V, I ) 0.4 nA; fast scan in x direction, slow scan upward.

behavior with a long induction period is reported for the reduction of Rh2O3 films under ambient pressure flow-reactor conditions at temperatures 300-470 K.10 The induction period was attributed to the low number of sites available for dissociative H2 adsorption on the initial oxide surface; once dissociation occurs, the reaction is ignited and the surface oxide is reduced by very fast propagating fronts. 3.3. Oxidation of Rh(110) with Molecular Oxygen. A distinct feature of the initial oxidation of the Rh(110) surface exposed to molecular oxygen is the laterally nonuniform oxide growth. This is evidenced by the lateral variations in the O 1s and Rh 3d5/2 intensities illustrated by the SPEM images in Figure 6a, measured after exposure of Rh(110) to O2 (PO2 ) 10-4 mbar) at 750 K, the oxidation conditions used for Rh(111) and Rh(100) oxidation, as well.11,12 The contrast levels in the O 1s image reflect the different local oxygen concentrations. The lateral variations in the intensity of the Rh 3d5/2 oxide component, illustrated by the Rhox image, reflect the local density of the Rh atoms in the oxide film. The Rhb+i image outlines the emission from the bulk and interfacial Rh atoms below the oxide film. Comparison of the darker and brighter regions in the O, Rhox and Rhb+i images reveals that the contrast cannot simply be related to a different local oxide thickness. If this was the case, the O 1s and Rhox maps should be identical, whereas the contrast of the Rhb+i map should be reversed, as observed during the initial oxidation of Ru(0001).27 On the contrary, most of the brighter areas in the O 1s image appear darker in the Rhox image and there is no correlation with the contrast of the Rhb+i image. There are only a few exceptions; for example, region “3” appears darker in the Rhox and O images and brighter in the Rhb+i image. This supposes that, to a great extent, the inhomogeneity is due to spatial variations in the local density of the Rh and O atoms in the oxide film. From the intensity profiles across the darker and brighter regions of the Rhox and O images, we evaluated the following Rh and O atomic

density ratios: Rhox(dark)/Rhox(bright) ) 0.8 ( 0.05 and O(bright)/O(dark) ) 1.1 ( 0.2. These numbers are remarkably similar to the Rh and O atomic density ratios between the two bulk oxides, the rutile RhO2 and the corundum Rh2O3, 0.83 and 1.1, respectively.28 The difference in the local stoichiometry of the darker and brighter areas is also clearly manifested by the deconvoluted Rh 3d5/2 and O 1s spectra from the regions “1”, “2”, and “3” (Figure 6b), and by the high-resolution Rh 3d5/2 and O 1s microspot spectra taken inside regions representing the three types of the oxide film (Figure 6c). The major contribution to the O 1s signal comes from the surface O atoms, in accordance with the very limited probe depth of 2.8 Å. All Rh 3d5/2 spectra contain the bulk and the two oxide-related components. The oxide components in the Rh 3d5/2 spectra, taken from regions “1”, “2”, and “3”, have the same binding energy within our spectral resolution. The Rhox/O 1s intensity ratios for the oxide films from the dominant “1” and “2” regions are close to the RhO2 and Rh2O3 stoichiometries, respectively. The Rh 3d5/2 spectrum from region “3”, which appears darker in both O 1s and Rhox maps and brighter in the Rhb+i map, has a more pronounced bulk component and a weaker Rhox component. The O 1s spectrum is similar to those from the regions “1” and “2”. The Rhox/O 1s ratio in the areas “3” is comparable to that measured for the region “1”, i.e., close to the RhO2 stoichiometry. The Rh 3d5/2 spectrum also requires a small “non-oxide” component, indicating that the formation of the surface “RhO2” film in “3” is not completed. From the relative intensity of the Rhb component, we evaluated the thickness of the oxide films in regions “1”, “2”, and “3” as ∼4, ∼4.6, and ∼3 Å, respectively. In the frame of the theoretical models, the surface oxide with a RhO2 stoichiometry should represent the O-Rh-O trilayers, the precursor to the formation of bulk Rh2O3.11,20 The comparison of the initial oxidation with atomic and molecular

13654 J. Phys. Chem. B, Vol. 109, No. 28, 2005

Dudin et al.

Figure 6. (a) O, Rhox and Rhb images (12.8 × 12.8 µm2) taken after exposure of Rh(110) to 1.2 × 105 L O2 at 750 K. (b) Rh 3d5/2 and O 1s spectra reconstructed from the regions marked as “1”, “2”, and “3” in the images displayed in (a). (c) Deconvoluted high-resolution Rh 3d5/2 and O 1s microspot taken inside regions representing the “1”, “2”, and “3” regions. The dotted lines indicate the energy positions of the Rh 3d components corresponding to the surface oxide, Rhox, and the metal/oxide interface, Rhi, and the O 1s components corresponding to the oxygen atoms at the surface, Os, and “below the surface” , Ob. The Rh 3d5/2 bulk component is plotted with a dashed line. The other components are related to the coexisting adsorption phases. HPads indicates the spectrum of the adsorption phase after exposure at 1 × 10-4 mbar.

oxygen shows that in the former case the formation of a homogeneous surface oxide with RhO2 stoichiometry is favored, whereas the inhomogeneous oxide growth in O2 ambient leads to the formation of an oxygen deficient surface oxide as well. Since the inhomogeneous oxide film also shows a c(2 × 4) LEED pattern, a possible explanation is that in the areas with a lower O supply the oxide film grows with high concentration of O vacancies and is closer to the Rh2O3 stoichiometry. The major reason for the lower rate of oxygen incorporation and the following inhomogeneous oxide growth in O2 ambient should be the O2 dissociation step, which is hindered on the O-covered Rh surface. The shape of regions “1”, most clearly seen in the Rhox map, resembles the step bunches observed on the Rh(110) surface with low energy electron microscopy (LEEM).29 This indicates that structural defects and irregular structures may act as more effective centers for the O2 dissociation and stabilize the surface oxide with RhO2 stoichiometry. Similar nonuniform oxide growth was observed during the initial oxidation of Ru(0001) carried out under comparable oxidation conditions.27 The autocatalytic mechanism of Ru oxide growth suggested in ref 30 was attributed to a higher O2 dissociation rate on the coordinatevely unsaturated (cus) sites of the RuO2 islands, nucleating inside the surrounding saturated adsorption phase. This mechanism cannot be applied to the present system, because the structure of the Rh oxide does not exhibit “active” cus-sites as RuO2. Even for the case of RuO2 growth the autocatalytic mechanism is disputable, because according to the SPEM results in the initial oxidation stages

surface space was distributed between ordered oxide and disordered O-rich precursor phases, the latter nucleating inside the adsorption phases.27 The bottom high pressure adsorption (HPads) spectra in Figure 6c correspond to a highly O loaded adsorption phase produced by exposure of a Rh(110) surface to 10-4 mbar of O2 at room temperature. They have very different line shapes and strongly resemble the spectra of the mixed high oxygen coverage (2 × 1)p2mg and segmented (10 × 2) phases.16,19 This result indicates that the interaction of the Rh(110) surface with molecular oxygen at room temperature is limited to the formation of high coverage adsorption phases, whereas using atomic oxygen sensible subsurface incorporation readily occurs. 4. Concluding Remarks From a combination of XPS microscopy, LEED and STM insights have been gained regarding the structure morphology and reactivity of the surface oxide, formed by exposures of a Rh(110) surface to atomic and molecular oxygen. The ordered thin oxide film, formed both in atomic and molecular oxygen ambient, is characterized by the same c(2 × 4) structure, but there are significant differences in the oxidation rate and in the morphology of the surface oxide phase. When using atomic oxygen, high concentrations of adsorbed and incorporated oxygen can be accumulated at temperatures far below those required for the formation of an ordered oxide film, resulting in strong enhancement of the oxidation rate at lower tempera-

Initial Oxidation of a Rh(110) Surface tures. The incorporation of oxygen and the following oxide formation in O2 ambient require higher pressures and temperatures, because the reaction is kinetically hindered by the O2 dissociation step. When atomic oxygen is used, such constraint does not exist, even though an energy barrier has still to be overcome in order to reorganize the accumulated oxygen into an ordered thin oxide film. Using atomic oxygen also leads to laterally uniform oxide growth, because the defects and irregularities at atomic and nanoscales, which act as more active sites for O2 dissociation, do not affect substantially the accumulation of oxygen if the O2 dissociation step is excluded from the reaction pathway. Furthermore, the surface Rh oxide is highly reactive with respect to H2. As in the case of Rh covered with a saturated oxygen adsorption layer, the onset of the reduction front is preceded by an induction period due to the very low H2 dissociative adsorption rate on the “oxide” surface. After the ignition of the reduction front the reduction occurs in two steps: fast partial reduction of the oxidic oxygen and conversion of the surface oxide to a reconstructed oxygen adsorption phase, followed by slower reduction of the adsorbed oxygen and deconstruction of the surface. Acknowledgment. This work was partially supported by Regione FVG (L.R.3-98) and by the MIUR under the programs PRIN2003 and FIRB2001. The EU is acknowledged for financial support under Contract No. NMP3-CT-2003-505670 (NANO2) and I3 project IA-SFS. References and Notes (1) Reuter, K.; Scheffler, M. Appl. Phys. A: Mater. Sci. Process. 2004, 78, 793 and references therein. (2) Gong, X.-Q.; Liu, Z.-P.; Raval, R.; Hu, P. J. Am. Chem. Soc. 2004, 126, 8 and references therein. (3) Hendriksen, B. L. M.; Frenken, J. W. M. Phys. ReV. Lett. 2002, 89, 46101. (4) Over, H.; Kim, Y. D.; Seitsonen, A. P.; Lundgren, E.; Schmid, M.; Varga, P.; Morgante, A.; Ertl, G. Science 2000, 287, 1474.

J. Phys. Chem. B, Vol. 109, No. 28, 2005 13655 (5) Madhavaram, H.; Idriss, H.; Wendt, S.; Kim, Y. D.; Knapp, M.; Over, H.; Assmann, J.; Lo¨ffler, E.; Muhler, M. J. Catal. 2001, 202, 296. (6) Over, H.; Muhler, M. Prog. Surf. Sci. 2003, 72, 3. (7) Kim, S. H.; Wintterlin, J. J. Chem. Phys. 2004, 108, 14565. (8) Reuter, K.; Frenkel, D.; Scheffler, M. Phys. ReV. Lett. 2004, 93, 116105. (9) Lundgren, E.; Gustafson, J.; Resta, A.; Weissenrieder, J.; Andersen, J. N.; Ko¨hler, L.; Kresse, G.; Klikovits, J.; Biederman, A.; Schmidt, M.; Varga, P. J. Electron. Spectrosc. Relat. Phenom. 2005, 144-147, 367. (10) Williams, Ch. T.; Chen, E. K.-Y.; Takoudis, Ch. G.; Weaver, M. J. J. Phys. Chem. B 1998, 102, 4785. (11) Gustafson, J.; Mikkelsen, A.; Borg, M.; Andersen, J. N.; Lundgren, E.; Klein, C.; Schmidt, M.; Varga, P.; Ko¨hler, L.; Kresse, G.; Kaspar, N.; Stierle, A.; Dosch, H. Phys. ReV. B 2005, 71, 115442. (12) Gustafson, J.; Mikkelsen, A.; Borg, M.; Lundgren, E.; Ko¨hler, L.; Kresse, G.; Schmidt, M.; Varga, P.; Yuhara, J.; Torrelles, X.; Quiros, C.; Andersen, J. N. Phys. ReV. Lett. 2004, 92, 126102. (13) Todorova, M.; Li, W. X.; Ganduglia-Pirovano, M. V.; Stampfl, C.; Reuter, K.; Scheffler, M. Phys. ReV. Lett. 2002, 89, 96103. (14) Kiskinova, M. Chem. ReV. 1996, 96, 1431. (15) Comelli, G.; Dhanak, V. R.; Kiskinova, M.; Prince, K. C.; Rosei, R. Surf. Sci. Rep. 1998, 32, 165. (16) Vesselli, E.; Africh, C.; Baraldi, A.; Comelli, G.; Esch, F.; Rosei, R. J. Chem. Phys. 2001, 114, 4221. (17) Africh, C.; Esch, F.; Li, W. X.; Corso, M.; Hammer, B.; Rosei, R.; Comelli, G. Phys. ReV. Lett. 2004, 93, 126104. (18) Kiskinova, M. Surf. Interface Anal. 2000, 30, 464; http://www.elettra.trieste.it/experiments/ beamlines/esca/index.html. (19) Gu¨nther, S.; Marbach, H.; Imbihl, R.; Baraldi, A.; Lizzit, S.; Kiskinova, M. J. Chem. Phys. 2003, 119, 12503. (20) Esch, F.; Kresse, G.; et al. Manuscript in preparation. (21) Gibson, K. D.; Viste, M.; Sanchez, E. C.; Sibener, S. J. J. Chem. Phys. 1999, 110, 2757. (22) Gregoratti, L.; Baraldi, A.; Dhanak, V. R.; Comelli, G.; Kiskinova, M.; Rosei, R. Surf. Sci. 1995, 340, 205. (23) Africh, C.; Esch, F.; Comelli, G.; Rosei, R. J. Chem. Phys. 2001, 115, 477. (24) Africh, C.; et al. Submitted to J. Am. Chem. Soc. (25) Africh, C.; Esch, F.; Comelli, G.; Rosei, R. J. Chem. Phys. 2002, 116, 7200. (26) Mertens, F.; Imbihl, R. Chem. Phys. Lett. 1995, 242, 221. (27) Bo¨ttcher, A.; Starke, U.; Conrad, H.; Blume, R.; Gregoratti, L.; Kaulich, B.; Barinov, A.; Kiskinova, M. J. Chem. Phys. 2002, 117, 8104. (28) Abe, Y.; Kato, K.; Kawamura, M.; Sasaki, K. Surf. Sci. Spectra 2001, 8, 117. (29) Dudin, P.; Locatelli, A.; et al. Manuscript in preparation. (30) Over, H.; Seitsonen, A. P. Science 2003, 297, 2003.