Experimental and Theoretical Study on the Electronic Interaction

Article pubs.acs.org/JPCC

Experimental and Theoretical Study on the Electronic Interaction between Rh Adatoms and CeOx Substrate in Dependence on a Degree of Cerium Oxide Reduction Klára Ševčíková,*,† Lucie Szabová,†,‡ Miroslav Kettner,† Petr Homola,† Nataliya Tsud,† Stefano Fabris,§ Vladimír Matolín,† and Václav Nehasil† †

Department of Surface and Plasma Science, Faculty of Mathematics and Physics, Charles University in Prague, V Holešovičkách 2, 180 00 Prague 8, Czech Republic ‡ International Center for Materials Nanoarchitectonics (WPI-MANA), National Institute for Materials Science, 1-1 Namiki, Tsukuba, Ibaraki 305-0044, Japan § CNR-IOM DEMOCRITOS, Istituto Officina dei Materiali, Consiglio Nazionale delle Ricerche and SISSA, Scuola Internazionale di Studi Superiori Avanzati, Via Bonomea 265, I-34136 Trieste, Italy ABSTRACT: The electronic metal−substrate interaction plays an important role in the surface reactions on supported metal nanoparticles. We study the interaction between rhodium clusters and cerium oxide substrate having various stoichiometries, CeOx (2.00 > x > 1.67), by means of photoelectron spectroscopy. Our results show that rhodium deposited on substrates with stoichiometry of 2.00 > x > 1.93 induces reduction of cerium oxide. On the other hand, cerium oxide with a higher degree of reduction (1.93 > x > 1.67) becomes partially oxidized after the Rh deposition. Density functional theory simulations of Rh adatom adsorbed on the CeO2(111) surface and on an oxygen vacancy at the CeOx(111) surface show that there is a charge transfer between rhodium and cerium oxide substrate. While Rh adsorption on the CeO2(111) surface leads to electron depletion at the Rh adatom and cerium oxide reduction, Rh adsorption on the oxygen vacancy at CeOx(111) leads to electron accumulation about the Rh adatom and partial oxidation of cerium oxide. The results clearly demonstrate that the electronic metal−substrate interaction occurs on Rh/CeOx systems and strongly depends on the stoichiometry of cerium oxide. These findings could be beneficial for designing catalysts with specific properties. materials,4 it is essential to investigate the character of the interaction in particular catalytic systems. Recently, our research has been focused on systems containing rhodium and cerium oxide. Both of these materials are components of commercial automotive exhaust converters.2,7,8 Rhodium is used mainly because of its high selectivity to convert toxic nitrogen oxides to N2.9 It is also efficient in oxidizing unsaturated hydrocarbons.10,11 Cerium oxide, CeOx (2.0 > x > 1.5), is used as a support for metal particles because it can enhance their catalytic activity.2,12 For instance, the synergetic enhancement of the catalytic activity was reported for Rh/CeOx.13 Nevertheless, cerium oxide is mostly valued for its high oxygen storage capacity.7 Cerium can reversibly transform between two stable oxidation states, Ce4+ and Ce3+. This transition is accompanied by the release or uptake of oxygen in

1. INTRODUCTION The chemistry on the catalyst is generally determined by the morphology and electron structure of the surface. These may differ substantially for different systems and materials. Heterogeneous catalysts often consist of metal particles as an active component and oxide acting as a support.1−3 However, the interaction between metal and oxide may significantly modify the properties of the particles.4 If the particles are big, the metal−oxide interaction is restricted to the interfacial area and the modifications of the particles surface properties are very small, thus approaching those of the corresponding metal.5 The situation is different when the particles consist of less than 100 atoms.6 In this case, even small perturbations in the cluster electron structure due to the interaction with the substrate may influence the clusters surface properties and their reactivity. The effects responsible for modifications of the cluster electron structure, e.g., a charge transfer between the particle and oxide, were collectively named as electronic metal support interaction (EMSI). Because EMSI may enhance or suppress the activity of the catalyst depending on the combination and quantity of © 2016 American Chemical Society

Received: November 23, 2015 Revised: February 11, 2016 Published: February 18, 2016 5468

DOI: 10.1021/acs.jpcc.5b11431 J. Phys. Chem. C 2016, 120, 5468−5476

Article

The Journal of Physical Chemistry C

Model cerium oxide thin films can be easily prepared by evaporating metallic cerium in oxygen atmosphere. According to Matoliń et al.,25 an epitaxial thin film of CeO2 grows on the copper single crystal with the surface orientation (111). A desired stoichiometry of cerium oxide can be achieved by an additional deposition of cerium (in UHV) which results in a partial reduction of the surface cerium oxide layers.26 The Cu(111) crystal (MaTecK, 99.999%) was cleaned by Ar ion sputtering and annealing to 750 K prior to cerium oxide deposition. The surface cleanliness was checked by monitoring the C 1s photoelectron signal. The cycles of sputtering and annealing were repeated until there were no traces of carbon left and the crystal gave a sharp LEED pattern typical for the Cu single crystal. Thin layers of CeO2 were deposited onto the Cu(111) single crystal by evaporating cerium (Goodfellow, 99.99%) in 5 × 10−7 mbar of O2 at 520 K. The thickness of the cerium oxide film was estimated through the attenuation of the Cu 2p3/2 photoelectron signal. An average thickness of the CeO2 buffer layers was approximately 6 ML (monolayers), considering a thickness of the O−Ce−O trilayer (defined as 1 ML of CeO2) to be 3.12 Å.27 Afterward, the sample was annealed 10 min at 520 K and cooled down in 5 × 10−7 mbar of O2 to obtain stoichiometric CeO2. LEED measurements showed a sharp (1.5 × 1.5) pattern, relative to Cu(111). Reduction of cerium oxide thin films was achieved by an additional deposition of cerium in UHV. The temperature of the substrate was kept at 520 K in order to enhance the migration of oxygen toward the surface and moderate the gradient of the degree of reduction through the cerium oxide film. By varying the time of the Ce deposition, we were able to prepare the CeOx substrates with stoichiometries ranged between 1.98 and 1.67. Total thickness of the film varied from 20 to 25 Å. LEED measurements of our films did not show any of the surface reconstructions reported by Duchoň et al.26 but only the (1.5 × 1.5) pattern. This pattern became more diffusive for the films with higher degrees of cerium reduction. Rhodium (Goodfellow, 99.9%, wire of 0.5 mm diameter) was evaporated onto the model CeOx thin films in several consecutive steps with the deposition rate of 0.1 nm·min−1.The temperature of the substrate was kept at 300 K. The pressure during the deposition did not exceed 4 × 10−9 mbar. The samples were not annealed after the Rh deposition in order to prevent forming large Rh particles and breaking the CeOx thin film. Therefore, we were not able to completely avoid the surface contamination. The samples were free of carbon until depositing approximately 0.15 ML of Rh. After that CO and C could be found at every sample and their amount increased with growing amount of rhodium. Considering these contaminants, we did not find any correlation between them and the effects described in section 4. Therefore, the carbon contaminants may be neglected from further discussions. The photoelectron spectra were acquired using energy 1486.6 eV (Al Kα). Additionally, energies of 630.0 and 420.0 eV were used to measure O 1s and Rh 3d (and C 1s) core levels, respectively. The valence band region was acquired at four different photon energies, 124.8, 121.4, 115.0, and 45 eV. The photons with an energy of 45.0 eV have a 10 times higher cross-section for exciting photoelectrons from Rh 4d states compared to Ce 4f. Because the signal from cerium is significantly suppressed, a photon energy of 45.0 eV is suitable for monitoring changes in rhodium valence band structure.

dependence on the environment. Because of the high mobility of oxygen atoms through cerium oxide, it may supply oxygen to surface reactions in oxygen-lean conditions or it may absorb excess oxygen from among the reactants.12 Transport of oxygen between the particle and the substrate is therefore very important. In our previous experiments with CO oxidation on Rh/CeOx systems, we observed that the degree of cerium oxide reduction before the deposition of rhodium has a great impact on the reactivity of the systems.14 The transport of oxygen between the particles and the oxide substrate was very intense if rhodium was deposited onto almost fully oxidized ceria (e.g., CeO 1.94 ). On the contrary, the oxygen transport was significantly suppressed at samples with an initially high degree of reduction (CeO1.85). We suggested that the interaction between rhodium particles and the cerium oxide substrate, EMSI, might be responsible for these phenomena and that there is a charge transfer between rhodium and cerium oxide depending on a degree of cerium oxide reduction.14 It is well known that the reactivity of CeOx-based systems is determined by the stoichiometry of cerium oxide (represented by the parameter x).15−17 According to the model experiments performed by Zhou et al.,18 rhodium deposited onto cerium oxide substrates forms three-dimensional particles regardless on the CeOx stoichiometry. They concluded that the distribution, size, and structure of Rh particles are independent of a degree of cerium oxide reduction. This supports our previous suggestion that the EMSI may play an important role in the case of Rh/CeOx. This idea was also proposed by Mullins et al.,16 who explained much easier CO dissociation on Rh supported by reduced CeOx, compared to Rh on stoichiometric CeO2, by an accumulation of negative charge on rhodium due to the Rh−Ce3+ interaction. In the case of reducible oxide, such as ceria, an ionic bond can be formed between the oxide surface and metal due to the exchange of electrons.4 The charge transfer from the metal particles to the ceria support was reported for Pt/CeO2,19 Pd/CeO2,20 Au/CeO2,21 Rh/CeO2,22 or Cu/CeO223 systems. However, only a few studies exist which deal with the effects of the metal deposition on reduced CeOx surfaces.23,24 In the presented work, we focus on investigating the interaction between rhodium clusters and cerium oxide substrates with different stoichiometries (1.98 > x > 1.67) by means of the photoelectron spectroscopy. Additionally, we support our findings by DFT simulations of rhodium adatom adsorbed on the stoichiometric CeO2(111) surface and at the CeOx oxygen vacancy site.

2. EXPERIMENTAL SECTION The experiments were carried out at the Materials Science Beamline at the synchrotron Elettra (Italy). The beamline is based on a plane grating monochromator providing a narrow band synchrotron light in the energy range 21−1000 eV. The end station consists of an ultrahigh vacuum chamber (UHV) with a base pressure of ∼5 × 10−10 mbar. The UHV chamber was equipped with a hemispherical analyzer (Specs Phoibos 150), a dual Al/Mg Kα X-ray source, an ion gun, LEED optics, a quadrupole mass spectrometer, Ce and Rh evaporators, a gas inlet system, and a sample holder enabling temperature regulation. The sample was heated by heat transmission from supporting Ta wires. The temperature of the sample was checked using a K-type thermocouple positioned at the rear side of the sample. 5469


Article


Figure 1. (a) Ce 3d, (b) O 1s, and (c) valence band spectra for various CeOx substrates.

(DFT) based calculations, similarly to the previously studied system of Cu/CeOx.23,30 According to recent theoretical works,31−34 a reliable description of both stoichiometric and reduced cerium oxide-based materials can be achieved by employing the generalized gradient approximation (GGA) and adding a Hubbard U term to the GGA exchange functional to compensate for the strong Coulomb repulsion among the localized Ce 4f electrons. The value of the parameter U used for the Rh/CeOx systems was 4.5 eV.23,31,32,35−37 More computational details can be found in the work of Szabová et al.30 The (111) ceria surface (with or without the oxygen vacancy) was modeled by supercells consisting of three CeO2 trilayers with the following sequence: O−Ce−O−O−Ce−O− O−Ce−O. Several adsorption configurations of rhodium adatoms were considered for stoichiometric and reduced cerium oxide surfaces. The bonding charge analysis was performed for the most stable configuration at the CeO2(111) surface and the surface containing the oxygen vacancy. The bonding charge analysis was performed by calculating the charge difference between the densities of the complete system, ρRh/CeOx, and the isolated noninteracting systems of a rhodium atom, ρRh, and cerium oxide surface, ρCeOx

The other three energies were used for so-called resonance photoelectron spectroscopy (RPES). The valence band spectra measured at photon energies of 124.8 and 121.4 eV exhibit distinct maxima at about 4.0 and 1.5 eV, respectively. These correspond to the resonant enhancements of the Ce 4f photoemission from Ce4+ and Ce3+ states, respectively.28 The valence band spectrum acquired at a photon energy of 115.0 eV corresponds to the off-resonance conditions. This spectrum is used as a background to calculate the resonance enhancements D(Ce4+) and D(Ce3+) as intensity differences between the onresonance and the off-resonance spectra at 4.0 and 1.5 eV. The resonant enhancement ratio D(Ce3+)/D(Ce4+) can be used as a direct indicator of a degree of cerium oxide reduction. The higher the value, the more reduced cerium oxide is. The degree of reduction was also estimated from concentrations of Ce4+ and Ce3+ states. The Ce 3d spectra were fitted by the procedure presented by Skála et al.29 A combination of linear and Shirley background was used. The satellites originating from Cu 2p due to a nonmonochromatized X-ray source were subtracted before fitting. The concentrations of Ce4+ and Ce3+ ions were derived from fitting the Ce 3d region as total areas of Ce4+ and Ce3+ fitting components, A(Ce4+) and A(Ce3+). The degree of CeOx reduction is evident from its stoichiometry which is represented by parameter x x=2−

A(Ce3 +) 1 · 2 A(Ce 4 +) + A(Ce3 +)

Δρ = ρRh / CeO − (ρCeO + ρRh )

(1)

x

The photoelectron spectra were taken at a normal emission geometry (using synchrotron radiation) and 20° to normal (using Al Kα). Total resolutions at energies 1486.6, 630.0, 420.0, and 124.8−115.0 eV were 1000, 690, 420, and 170 meV, respectively. The evolution of the degree of cerium oxide reduction in time (exposed and unexposed to the radiation) was checked, and the system is stable within the time necessary for the measurement (∼1 h).

x

(2)

The adsorption energy of an Rh adatom on the CeOx surface was calculated in terms of total energies of individual systems according to Eads = E Rh / CeOx − (ECeOx + ERh)

(3)

where ERh/CeOx is the total energy of the complete system (rhodium adatom interacting with the cerium oxide surface), ECeOx is the total energy of the corresponding cerium oxide surface, and ERh is the total energy of a rhodium atom isolated in vacuum.

3. THEORETICAL CALCULATIONS The interaction between the rhodium adatom and the cerium oxide substrate was investigated by density functional theory 5470


Article


reduced CeO1.67 is approximately 0.5 eV. At first glance, these shifts could be attributed to different charging of the CeOx layers due to their different thicknesses. However, we did not observe any relation between the layer thickness and the binding energy shift. Moreover, the Ce 4f state close to the Fermi energy remains at the same energy for various x, which goes against the charging of the CeOx layer. Therefore, the shift might be caused by different pinning of the Fermi energy due to changes in the occupancy of Ce 4f states which was proposed by Yang et al.34 It is interesting that the energy difference between the Ce 3d and the O 1s peaks remains the same for all of our CeOx films. This result differs from the work of Duchoň et al.,26 who observed lowering of the energy difference between the Ce 3d and the O 1s peaks for cerium oxide with higher degrees of reduction. While Duchoň et al. worked with highly ordered CeOx thin films with four different surface reconstructions,26 all our systems retained the surface reconstruction of (1.5 × 1.5) relative to the Cu(111) substrate. Therefore, the discrepancy between our studied films and the published results might be caused by different morphologies of the films. 4.2. Rh Deposition. The interaction of rhodium and cerium oxide was studied by several consecutive depositions of rhodium onto the model CeOx thin films. Figure 3 shows the evolution of the degree of reduction during the deposition of rhodium onto various CeOx substrates. The parameter x estimated from the XPS Ce 3d spectra according to eq 1 is plotted in Figure 3a. The resonance enhancement ratio derived from RPES measurements is shown in Figure 3b. Both of these parameters can be used independently for expressing the degree of reduction of cerium oxide. We can see that the degree of reduction develops in two different ways. The films of nearly stoichiometric CeO2 are reduced by deposited rhodium. On the contrary, CeOx with lower x become more oxidized after adding rhodium. Similar behavior of cerium oxide was observed by Wilson et al.24 for Pd/CeOx and by Szabová et al.23 for Cu/CeOx. In the case of Rh/CeOx, the stoichiometry at which the deposited rhodium induces neither reduction nor oxidation of cerium oxide is CeO1.93 (marked by turquoise stars in Figure 3). Both oxidation

4. RESULTS AND DISCUSSION 4.1. CeOx Substrate. Figure 1a, 1b, and 1c shows the Ce 3d, O 1s, and valence band photoelectron spectra for the CeOx thin films with various stoichiometries, respectively. The features of the Ce 3d spectra correspond to a mixture of Ce4+ and Ce3+ states. Their ratio defines the stoichiometry of CeOx. The O 1s spectra consist of two peaks. The peak at the lower binding energy, OLat, can be assigned to oxygen from the ceria lattice.26,38−42 The peak at the higher binding energy is usually associated with OH groups adsorbed on the surface oxygen vacancies close to Ce3+ sites.17,38,41,43 The peaks corresponding to hydroxyl groups are also apparent from the valence band spectra (Figure 1c). Water is a common component of the residual atmosphere in the UHV system. It does not adsorb on the stoichiometric CeO2 at 300 K, but it readily fills the oxygen vacancies on the reduced cerium oxide surface. The increasing intensity of the OH peaks at CeOx with lower x is consistent with adsorption of hydroxyl groups in oxygen vacancies. We can notice from Figure 1 that most of the peaks shift to higher binding energies for lower x (indicated by blue lines). This shift is more apparent from Figure 2, which shows the

Figure 2. Binding energies of (a) the Ce 3d3/2 (Ce4+ f 0) peak and (b) the OLat and OH peaks of the O 1s region for various CeOx substrates.

binding energies of the Ce 3d3/2 (Ce4+ f 0), OLat, and OH peaks designated in Figure 1a and 1b. The maximum difference between the OLat binding energies of oxidized CeO1.98 and

Figure 3. Evolution of the degree of CeOx reduction for various cerium oxide stoichiometries (1.98 > x > 1.67) during the consequent depositions of rhodium estimated from (a) XPS and (b) SRPES measurements. 5471


Article


plotted in Figure 4b as solid and dashed lines, respectively. We can see that the contribution of the Rh 4d electrons to the total photoelectron yield is very small at an energy of 121.4 eV. Therefore, using energies of 45.0 and 121.4 eV helps disentangling the mingling cerium and rhodium contributions to the valence band spectra. The main feature of the valence band spectra in Figure 4a and 4b is stretching from 2.5 to 7.5 eV. It corresponds to the O 2p Ce 5d6s hybridized valence orbitals. The Rh 4d and Ce 4f states are located around 1.5 eV. These states are absent at the CeO1.98 substrate (gray lines) but become prominent after the Rh deposition. Both states overlap, but we can state that the maximum of the Rh 4d structure is at a slightly higher binding energy than that of Ce 4f, which is consistent with ref 44. The most apparent change between the curves of consecutive deposition is the attenuation of the signal from O 2p Ce 5d6s caused by an increasing Rh surface coverage. Besides, the maximum of the Rh 4d structure moves to lower binding energies with an increasing amount of deposited rhodium. The same shift was observed in the Rh 3d spectra (data not shown) and can be attributed to the particle-size effect. The particles exhibit a metallic character from approximately 1.7 ML, judging from the formation of the Fermi edge in the valence band structure. The substantial increment of the Ce 4f intensity in the first few steps of rhodium deposition can be seen in Figure 4b. It can be interpreted as Ce4+ ions undergoing the reduction by accepting electrons from rhodium adatoms. The reduction of cerium is apparent also from the growth of the resonance enhancement ratio plotted in Figure 3. It should be noted that the reduction caused by a higher amount of deposited rhodium is less pronounced compared to the reduction induced by the first few steps of the deposition. This is probably caused by a combination of two effects. First, the less effective reduction of cerium can result from a relatively decreasing amount of rhodium atoms adsorbing directly onto the CeOx substrate. According to Zhou et al.,18 rhodium grows three dimensionally on the cerium oxide support. It means that the relative amount of rhodium adatoms interacting directly with the substrate is decreasing with further addition of rhodium. Second, some of the Ce4+ ions close to deposited rhodium are reduced to Ce3+. The accumulation of negative charge in the substrate near Rh adatoms may hinder the charge transfer from additionally deposited rhodium (when attached to the already deposited Rh). Our experimental results are supported by the density functional theory calculations. The results derived from the simulation of the rhodium adatom adsorbed on the stoichiometric CeO2(111) surface are displayed in Figure 5. The most stable adsorption configuration is displayed from a top view and a side view in Figure 5d and 5e, respectively. The rhodium adatom adsorbs into the hollow between three surface oxygen ions, which agrees well with the results of Lu et al.22 The bonding charge analysis displayed in Figure 5c shows the spatial distribution of the difference in charge density upon Rh adsorption, which was calculated as described by eq 2. The bonding charge analysis exhibits charge difference along the Rh−O bond, suggesting the interaction between rhodium and surface oxygen. The red curve in Figure 5a demonstrates the bonding charge integrated on planes parallel to the surface and plotted as a function of the distance from the surface (along the direction perpendicular to the surface). It shows the depletion of the charge density on rhodium adatom and the accumulation

and reduction during Rh deposition can be ascribed to the electronic metal−support interaction as is discussed in ref 14. Because of the apparently different interactions between rhodium and oxidized or reduced cerium oxide, these two systems will be examined separately. 4.3. 2.00 > x > 1.93. First, we will examine the interaction between rhodium and nearly stoichiometric CeO2. The reduction of ceria by a deposited metal is a well-known effect which has been proved earlier for, e.g., Pt/CeO2,19 Pd/CeO2,20 Au/CeO2,21 and Cu/CeO223 systems. In all cases, the reduction of cerium oxide was caused by an electron transfer from the deposited metal to the support, leading to oxidation of the metal adatoms and reduction of Ce4+ ions. According to Vayssilov et al.,19 the electron transfer from the metal cluster and the formation of Ce3+ is facilitated by a small energy difference between the highest occupied levels in the metal and the empty Ce 4f states of ceria. This view is corroborated for the Rh/CeO2(111) system by the work of Pfau et al.,44 who experimentally derived a nonzero overlap between the Rh 4d and the Ce 4f orbitals. The charge transfer from Rh to CeO2 has been also predicted by the density functional calculations of Lu et al.22 The changes in the electron structure of the Rh/CeOx system due to the interaction between rhodium and nearly stoichiometric ceria are demonstrated on the CeO1.98 sample. The total amount of rhodium deposited onto CeO1.98 was approximately 1.7 ML. The valence band spectra for every step of the deposition measured at two photon energies are plotted in Figure 4. At a photon energy of 45.0 eV, the photoionization

Figure 4. Evolution of the valence band spectra of Rh/CeO1.98 during consecutive deposition of rhodium (from black to red) measured with energies (a) 45.0 and (b) 121.4 eV. Gray lines represent CeO1.98 before deposition. Thin dashed lines in the b were measured at a photon energy of 115.0 eV after the first and last deposition steps. The total amount of Rh was approximately 1.7 ML.

cross-section of the Rh 4d electrons is 10 times higher compared to that of the Ce 4f electrons. Therefore, the signal from rhodium is more pronounced than that of cerium at this photon energy. On the other hand, the photon energy of 121.4 eV corresponds to the resonant electron emission from the 4f states of Ce3+, thus substantially enhancing the signal from Ce 4f. The Ce 4f resonance enhancement added to the very low photoionization cross-section for Rh 4d electrons (it is 100 times lower at photon energies of 115−125 eV than at 45 eV) results in highlighting the Ce 4f state and suppressing the signal from Rh 4d. This may be clear from comparing the valence band spectra in the near edge region measured using the “on-resonance” and “off-resonance” photon energies 5472


Article


Figure 5. Results from the DFT simulations of an Rh adatom on the stoichiometric CeO2(111) surface. (a) Change of bonding charge integrated in planes parallel to the surface plotted as a function of the height from the surface, (b) atom-resolved projected density of states, and (c) an arrangement of the bonding charge. Electron accumulation and depletion are represented by red and blue areas, respectively. (d) Top view and (e) side view of the most stable adsorption configuration of the Rh adatom on the CeO2(111) surface (Rh, gray; Ce, green, O, orange; O in the first surface layer, red). Adsorption energy was calculated according to eq 3.

of the charge density on cerium and oxygen ions in the first monolayer. It means that the charge transferred from the rhodium atom to the surface oxygen and cerium layers. Projected density of states analysis (pDOS, Figure 5b) provides further details on the charge transfer. Compared to standard pDOS characteristics of stoichiometric undoped CeO2,32 the pDOS of ceria supporting the Rh atom exhibits a peak in the occupied states of one of the Ce atoms (blue curve, peak at ca. −1 eV), suggesting a reduction of this Ce atom as a result of the charge transferred from Rh. 4.4. 1.93 > x > 1.67. The situation is different for rhodium adsorbed on reduced cerium oxide. Figure 3 shows that rhodium induces the oxidation of Ce3+ ions if it is deposited onto oxide with the high degree of reduction. Because the oxidation of the support upon the deposition of a metal is rather unusual, we first considered the interaction of the reduced cerium oxide with the residual atmosphere in the UHV system. The atmosphere in the MSB chamber was composed mainly from water, argon (originated from ion sputtering), hydrogen, and carbon dioxide. Molecular oxygen was not present in the residual atmosphere of the UHV system because of its high reactivity. Argon is completely inert, and hydrogen causes the reduction of ceria in Rh/CeOx systems.45 As for CO2, it was reported that the exposure of the reduced cerium oxide to CO2 leads to CeOx oxidation.46 However, the decomposition of CO2 molecules accompanied by the oxidation of cerium ions also leads to the formation of the surface carbonates and carboxylates. There were no traces of these species in the C 1s spectra at our samples. Finally, water would not cause any detectable oxidation of cerium oxide in our system at a temperature of 300 K.17,47 Therefore, the oxidation induced by adsorbates from the residual atmosphere can be ruled out.

Another way how the surface of reduced cerium oxide can be oxidized is a transport of oxygen from deeper layers toward the surface. It should be noted that the degree of reduction of our model thin films is not uniform. The surface contains more Ce3+ ions compared to the deeper layer.26 The migration of oxygen through cerium oxide is a well-known effect.9 However, it does not play an important role during Rh deposition at room temperature because the extent of oxygen migration at 300 K is very low. Besides, we did not observe any significant changes in the O 1s spectra (data not shown). Moreover, we observed the oxidation at both, the near surface area (RPES) and the deeper layers (XPS), which goes against the transport of oxygen ions from deeper layers to the surface. Therefore, the oxidation of reduced cerium oxide is likely to be induced by the interaction with rhodium, similarly to Cu/CeO223 and Au/TiO248,49 systems. The nature of this interaction would be also the electronic metal−support interaction but different from that for the Rh/CeOx samples with the lower degree of reduction. The nearly stoichiometric cerium oxide thin films are terminated with O2− layer (see Figure 5), while the partially reduced cerium oxide surface contains exposed Ce3+ ions. Rhodium can thus interact directly with cerium ions (Rh−Ce), unlike in the case of stoichiometric cerium oxide where Rh and Ce are separated by an oxygen layer (Rh−O−Ce). The Rh−Ce interaction is evident from our valence band region measurements of the Rh/CeOx samples with the high degree of reduction. The surfaces of such samples contain high amounts of oxygen vacancies and the electron-rich Ce3+ sites in their vicinity. The valence band spectra for every deposition step onto the sample CeO1.67 are plotted in Figure 6. The total amount of deposited rhodium reached approximately 1.2 ML. The decreasing occupancy of Ce 4f state can be clearly seen 5473


Article


stable adsorption configuration. We can see that the rhodium atom sits in the oxygen vacancy site, slightly above the position of the original oxygen ion. It is bonded to the nearest oxygen and two closest cerium ions as shown in the spatial distribution of the bonding charge in Figure 7c. The integrated bonding charge curve plotted in Figure 7a exhibits a charge transfer from the cerium layer to the rhodium adatom, opposite to the case of Rh/CeO2. The ceria surface with the oxygen vacancy contains two reduced Ce3+ ions.23,32,34 Rh adsorption in the oxygen vacancy is accompanied by the charge transfer from ceria to the Rh atom, causing reoxidation of one of the reduced Ce3+ ions, as is evidenced by the presence of only one peak in the occupied Ce 4f states (Figure 7b, blue curve, peak at ca. −1 eV). These calculations support the hypothesis about the electronic metal−support interaction on the Rh/CeOx system, which was suggested in ref 14. An interesting effect which can be noticed in Figure 3b for reduced CeOx is an initial increase of the D(Ce3+)/D(Ce4+) ratio, which means further reduction. This effect can be observed only for extremely low coverages of rhodium ( x > 1.67) were prepared on the Cu(111) substrate. Rhodium was deposited in several consecutive steps on these systems. The changes in electron structure of the system induced by the metal−oxide interaction were monitored by means of the photoelectron spectroscopy. If cerium oxide stoichiometry was close to CeO2, the substrate was reduced by deposited rhodium (Ce4+ → Ce3+). On the contrary, the rhodium deposition induced the oxidation of Ce3+ ions in CeOx with a high degree of reduction. A boundary CeOx stoichiometry between these two effects is CeO1.93. DFT simulations of rhodium adatom on cerium oxide surfaces showed that there is a charge transfer between rhodium and cerium oxide substrate. In the case of stoichiometric CeO2(111), a charge transfers from rhodium to cerium oxide, reducing Ce4+ ion and leaving Rh adatom slightly positive. On the other hand, the rhodium adsorption on the oxygen vacancy site at the CeOx(111) surface results in the depletion of the charge density in cerium oxide and its accumulation on rhodium. Rhodium adatom is therefore slightly negative. Therefore, the oxidation and reduction of cerium oxide substrate induced by deposited rhodium were interpreted by the electronic metal support interaction. There are two types of EMSI for Rh/CeOx systems, depending on the stoichiometry of cerium oxide. For CeOx with x higher than 1.93, the EMSI results in the charge transfer from rhodium to cerium oxide. In the case of CeOx with x lower than 1.93, the charge transfers from cerium oxide to rhodium. These effects could explain the strong dependence of the reactivity of the Rh/CeOx systems on the cerium oxide stoichiometry. It is clear that the EMSI has a great influence on the properties of Rh/CeOx. Therefore, our results could be beneficial for designing catalysts with specific properties.

■

■

ABBREVIATIONS DFT, density functional theory; EMSI, electronic metal substrate interaction; GGA, generalized gradient approximation; RPES, resonance photoelectron spectroscopy; UHV, ultrahigh vacuum

■

REFERENCES

(1) Henry, C. R. Surface Studies of Supported Model Catalysts. Surf. Sci. Rep. 1998, 31, 231−325. (2) Gandhi, H. S.; Graham, G. W.; McCabe, R. W. Automotive Exhaust Catalysis. J. Catal. 2003, 216, 433−442. (3) Hirata, H. Recent Research Progress in Automotive Exhaust Gas Purification Catalyst. Catal. Surv. Asia 2014, 18, 128−133. (4) Pacchioni, G. Electronic Interactions and Charge Transfers of Metal Atoms and Clusters on Oxide Surfaces. Phys. Chem. Chem. Phys. 2013, 15, 1737−1757. (5) Gates, B. C. Supported Metal Clusters: Synthesis, Structure, and Catalysis. Chem. Rev. 1995, 95, 511−522. (6) Halperin, W. P. Quantum Size Effects in Metal Particles. Rev. Mod. Phys. 1986, 58, 533. (7) Kašpar, J.; Fornasiero, P.; Graziani, M. Use of CeO2-Based Oxides in the Three-Way Catalysis. Catal. Today 1999, 50, 285−298. (8) Muraki, H.; Zhang, G. Design of Advanced Automotive Exhaust Catalysts. Catal. Today 2000, 63, 337−345. (9) Zafiris, G.; Gorte, R. J. Evidence for Low-Temperature Oxygen Migration from Ceria to Rh. J. Catal. 1993, 139, 561−567. (10) Miyazawa, T.; Okumura, K.; Kunimori, K.; Tomishige, K. Promotion of Oxidation and Reduction of Rh Species by Interaction of Rh and CeO2 over Rh/CeO2/SiO2. J. Phys. Chem. C 2008, 112, 2574− 2583. (11) Ferrizz, R. M.; Egami, T.; Vohs, J. M. Temperature Programmed Desorption Study of the Reaction of C2H4 and CO on Rh Supported on α-Al2O3(0001), YSZ(100) and CeO2 Thin Films. Surf. Sci. 2000, 465, 127−137. (12) Trovarelli, A.; de Leitenburg, C.; Boaro, M.; Dolcetti, G. The Utilization of Ceria in Industrial Catalysis. Catal. Today 1999, 50, 353−367. (13) Cunningham, J.; Hickey, N. J.; Cataluna, R.; Conesa, J. C.; Soria, J.; Martinez-Arias, A. Interfacial RhOx/CeO2 Sites as Locations for Low Temperature N2O Dissociation. Stud. Surf. Sci. Catal. 1996, 101, 681−690. (14) Ševčíková, K.; Kolárǒ vá, T.; Skála, T.; Tsud, N.; Václavů, M.; Lykhach, Y.; Matolín, V.; Nehasil, V. Impact of Rh−CeOx Interaction on CO Oxidation Mechanisms. Appl. Surf. Sci. 2015, 332, 747−755. (15) Cordatos, H.; Bunluesin, T.; Stubenrauch, J.; Vohs, J. M.; Gorte, R. J. Effect of Ceria Structure on Oxygen Migration for Rh/Ceria Catalysts. J. Phys. Chem. 1996, 100, 785−789. (16) Mullins, D.; Overbury, S. H. CO Dissociation on Rh Deposited on Reduced Cerium Oxide Thin Films. J. Catal. 1999, 188, 340−345. (17) Lykhach, Y.; Johánek, V.; Aleksandrov, H. A.; Kozlov, S. M.; Happel, M.; Skála, T.; Petkov, P. St.; Tsud, N.; Vayssilov, G. N.; Prince, K. C.; et al. Water Chemistry on Model Ceria and Pt/Ceria Catalysts. J. Phys. Chem. C 2012, 116, 12103−12113.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: +420 733 514 434. Fax: +420 284 685 095. Author Contributions

This manuscript was written through contributions of all authors. K.Š. lead the experiments at MSB (Elettra), analyzed the data, and prepared the manuscript. L.S. performed the DFT simulations and prepared the manuscript. M.K., P.H., and N.T. participated in the experiments at MSB (Elettra). S.F. supervised the DFT calculations. V.N. supervised the experiments and participated in the measurements and data analyzing. The results were consulted with V.M. All authors have given approval to the final version of the manuscript. 5475


Article

The Journal of Physical Chemistry C (18) Zhou, J.; Baddorf, A. P.; Mullins, D. R.; Overbury, S. H. Growth and Characterization of Rh and Pd Nanoparticles on Oxidized and Reduced CeOx(111) Thin Films by Scanning Tunneling Microscopy. J. Phys. Chem. C 2008, 112, 9336−9345. (19) Vayssilov, G. N.; Lykhach, Y.; Migani, A.; Staudt, T.; Petrova, G. P.; Tsud, N.; Skála, T.; Bruix, A.; Illas, F.; Prince, K. C.; et al. J. Support Nanostructure Boosts Oxygen Transfer to Catalytically Active Platinum Nanoparticles. Nat. Mater. 2011, 10, 310−315. (20) Skála, T.; Šutara, F.; Škoda, M.; Prince, K. C.; Matolín, V. Palladium Interaction with CeO2, Sn-Ce-O and Ga-Ce-O Layers. J. Phys.: Condens. Matter 2009, 21, 055005. (21) Škoda, M.; Cabala, M.; Matolínová, I.; Prince, K. C.; Skála, T.; Šutara, F.; Veltruská, K.; Matolín, V. Interaction of Au with CeO2(111): A Photoemission Study. J. Chem. Phys. 2009, 130, 034703. (22) Lu, Z.; Yang, Z. Interfacial Properties of NM/CeO2(111) (NM = Noble Metal Atoms or Clusters of Pd, Pt and Rh): A First Principles Study. J. Phys.: Condens. Matter 2010, 22, 475003. (23) Szabová, L.; Skála, T.; Matolínová, I.; Fabris, S.; Farnesi Camellone, M.; Matolín, V. Copper-Ceria Interaction: A Combined Photoemission and DFT Study. Appl. Surf. Sci. 2013, 267, 12−16. (24) Wilson, E. L.; Chen, Q.; Brown, W. A.; Thornton, G. CO Adsorption on the Model Catalyst Pd/CeO2‑x(111)/Rh(111). J. Phys. Chem. C 2007, 111, 14215−14222. (25) Matolín, V.; Libra, J.; Matolínová, I.; Nehasil, V.; Sedlácě k, L.; Šutara, F. Growth of Ultra-Thin Cerium Oxide Layers on Cu(111). Appl. Surf. Sci. 2007, 254, 153−155. (26) Duchoň, T.; Dvořaḱ , F.; Aulická, M.; Stetsovych, V.; Vorokhta, M.; Mazur, D.; Veltruská, K.; Skála, T.; Mysliveček, J.; Matolínová, I.; et al. Ordered Phases of Reduced Ceria As Epitaxial Films on Cu(111). J. Phys. Chem. C 2014, 118, 357−365. (27) Bernal, S.; Calvino, J. J.; Cauqui, M. A.; Pérez Omil, J. A.; Pintado, J. M.; Rodríguez-Iziquirdo, J. M. Image Simulation and Experimental HREM Study of the Metal Dispersion in Rh/CeO2 Catalysts. Influence of the Reduction/reoxidation Conditions. Appl. Catal., B 1998, 16, 127−138. (28) Matolín, V.; Matolínová, I.; Sedlácě k, L.; Prince, K. C.; Skála, T. A Resonant Photoemission Applied to Cerium Oxide Based Nanocrystals. Nanotechnology 2009, 20, 215706. (29) Skála, T.; Šutara, F.; Prince, K. C.; Matolín, V. Cerium Oxide Stoichiometry Alteration via Sn Deposition: Influence of Temperature. J. Electron Spectrosc. Relat. Phenom. 2009, 169, 20−25. (30) Szabová, L.; Camellone, M. F.; Huang, M.; Matolín, V.; Fabris, S. Thermodynamic, Electronic and Structural Properties of Cu/CeO2 Surfaces and Interfaces from First-Principles DFT+U Calculations. J. Chem. Phys. 2010, 133, 234705. (31) Camellone, M. F.; Fabris, S. Reaction Mechanisms for the CO Oxidation on Au/CeO2 Catalysts: Activity of Substitutional Au3+/Au+ Cations and Deactivation of Supported Au+ Adatoms. J. Am. Chem. Soc. 2009, 131, 10473−10483. (32) Fabris, S.; de Gironcoli, S.; Baroni, S.; Vicario, G.; Balducci, G. Taming Multiple Valency with Density Functionals. Phys. Rev. B: Condens. Matter Mater. Phys. 2005, 71, 041102. (33) Da Silva, J. L. F.; Ganduglia-Pirovano, M. V.; Sauer, J.; Bayer, V.; Kresse, G. Hybrid Functionals Applied to Rare-Earth Oxides: The Example of Ceria. Phys. Rev. B: Condens. Matter Mater. Phys. 2007, 75, 045121. (34) Yang, Z.; Luo, G.; Lu, Z.; Woo, T. K.; Hermansson, K. Structural and Electronic Properties of NM-Doped Ceria (NM = Pt, Rh): A First-Principles Study. J. Phys.: Condens. Matter 2008, 20, 035210. (35) Szabová, L.; Tateyama, Y.; Matolín, V.; Fabris, S. Water Adsorption and Dissociation at Metal-Supported Ceria Thin Films: Thickness and Interface-Proximity Effects Studied with DFT+U Calculations. J. Phys. Chem. C 2015, 119, 2537−2544. (36) Negreiros, F. R.; Fabris, S. Role of Cluster Morphology in the Dynamics and Reactivity of Subnanometer Pt Clusters Supported on Ceria Surfaces. J. Phys. Chem. C 2014, 118, 21014−21020.

(37) Negreiros, F. R.; Camellone, M. F.; Fabris, S. Effects of Thermal Fluctuations on the Hydroxylation and Reduction of Ceria Surfaces by Molecular H2. J. Phys. Chem. C 2015, 119, 21567−21573. (38) Mullins, D.; Overbury, S.; Huntley, D. Electron Spectroscopy of Single Crystal and Polycrystalline Cerium Oxide Surfaces. Surf. Sci. 1998, 409, 307−319. (39) Holgado, J.; Munuera, G.; Espinós, J.; González-Elipe, A. XPS Study of Oxidation Processes of CeOx Defective Layers. Appl. Surf. Sci. 2000, 158, 164−171. (40) Tsud, N.; Skála, T.; Mašek, K.; Hanyš, P.; Takahashi, M.; Suga, H.; Mori, T.; Yoshikawa, H.; Yoshitake, M.; Kobayashi, K.; et al. Photoemission Study of the Tin Doped Cerium Oxide Thin Films Prepared by RF Magnetron Sputtering. Thin Solid Films 2010, 518, 2206−2209. (41) Matolín, V.; Matolínová, I.; Dvořaḱ , F.; Johánek, V.; Mysliveček, J.; Prince, K. C.; Skála, T.; Stetsovych, O.; Tsud, N.; Václavů, M.; et al. Water Interaction with CeO2(111)/Cu(111) Model Catalyst Surface. Catal. Today 2012, 181, 124−132. (42) Skála, T.; Šutara, F.; Cabala, M.; Škoda, M.; Prince, K. C.; Matolín, V. A Photoemission Study of the Interaction of Ga with CeO2(111) Thin Films. Appl. Surf. Sci. 2008, 254, 6860−6864. (43) Force, C.; Román, E.; Guil, J. M.; Sanz, J. XPS and 1H NMR Study of Thermally Stabilized Rh/CeO2 Catalysts Submitted to Reduction/oxidation Treatments. Langmuir 2007, 23, 4569−4574. (44) Pfau, A.; Schierbaum, K. D.; Göpel, W. The Electronic Structure of CeO2 Thin Films: The Influence of Rh Surface Dopants. Surf. Sci. 1995, 331−333, 1479−1485. (45) Cunningham, J.; Hickey, N. J.; Farrell, F.; Bowker, M.; Weeks, C. NO + CO – > 1/2 N2 + CO2 Differentiated from 2NO + CO – > N2O + CO2 over Rhodia/Ceria Catalysts Using 15N180 and 13C160 Reactants or Time- Resolution of Products. Stud. Surf. Sci. Catal. 1998, 116, 409−418. (46) Staudt, T.; Lykhach, Y.; Tsud, N.; Skála, T.; Prince, K. C.; Matolín, V.; Libuda, J. Ceria Reoxidation by CO2: A Model Study. J. Catal. 2010, 275, 181−185. (47) Sadi, F.; Duprez, D.; Gérard, F.; Miloudi, A. Hydrogen Formation in the Reaction of Steam with Rh/CeO2 Catalysts: A Tool for Characterising Reduced Centres of Ceria. J. Catal. 2003, 213, 226− 234. (48) Wang, Y.-G.; Yoon, Y.; Glezakou, V.-A.; Li, J.; Rousseau, R. The Role of Reducible Oxide-Metal Cluster Charge Transfer in Catalytic Processes: New Insights on the Catalytic Mechanism of CO Oxidation on Au/TiO2 from Ab Initio Molecular Dynamics. J. Am. Chem. Soc. 2013, 135, 10673−10683. (49) Jiang, Z.; Zhang, W.; Jin, L.; Yang, X.; Xu, F.; Zhu, J.; Huang, W. Direct XPS Evidence for Charge Transfer from a Reduced Rutile TiO2(110) Surface to Au Clusters. J. Phys. Chem. C 2007, 111, 12434− 12439.

5476


Experimental and Theoretical Study on the Electronic Interaction

Recommend Documents