J. Phys. Chem. C 2008, 112, 3751-3758
3751
Photoemission Spectroscopy Study of Cu/CeO2 Systems: Cu/CeO2 Nanosized Catalyst and CeO2(111)/Cu(111) Inverse Model Catalyst Vladimı´r Matolı´n,*,† Libor Sedla´ cˇ ek,† Iva Matolı´nova´ ,† Frantisˇek Sˇ utara,† Toma´ sˇ Ska´ la,‡ Brˇ etislav Sˇ mı´d,† Jirˇ´ı Libra,† Va´ clav Nehasil,† and Kevin C. Prince‡ Charles UniVersity, Faculty of Mathematics and Physics, Department of Surface and Plasma Science, V HolesˇoVicˇ ka´ ch 2, 18000 Prague 8, Czech Republic, and Sincrotrone Trieste, Strada Statale 14, km 163.5, 34012 BasoVizza-Trieste, Italy ReceiVed: September 26, 2007; In Final Form: December 17, 2007
Cerium oxide films equivalent to 2 ML of CeO2 were grown at 520 K in an oxygen atmosphere on a clean Cu(111) substrate in order to prepare a model catalytic system. This “inverse model catalyst” was characterized by low-energy electron diffraction (LEED), X-ray photoelectron spectroscopy (XPS) of core levels, and resonant photoelectron spectroscopy (RPES) of the valence band. Samples annealed at 770 K exhibited a LEED pattern corresponding to the (1.5 × 1.5) CeO2(111)/Cu(111) structure that can be interpreted as formation of a flat, well-ordered cerium oxide overlayer aligned with the principal crystallographic axes of the substrate. The model catalytic system corresponds well to a copper-loaded ceria nanopowder catalyst that exhibits growth of Cu(111) film structure on CeO2(111) planes. Lowering of the CO oxidation temperature due to the Cu loading is explained by CO adsorption on copper in the vicinity of highly active ceria planes providing oxygen for the reaction.
1. Introduction It is known that cerium dioxide (CeO2) is an important catalyst in many chemical reactions, for example, NO reduction under oxidizing conditions or CO oxidation under reducing conditions in automotive exhaust catalysts.1 Several studies indicate that the chemical state of ceria is a critical factor dominating the catalytic behavior. Cu-CeO2 is a highly active catalyst for CO oxidation by oxygen and for water gas shift.2-5 The remarkable redox ability of CuO-CeO2 at lower temperature was found to play an essential role in CO oxidation reactions. It was found that only a small amount of Cu promotes CeO2 catalytic activity by several orders of magnitude,6-12 but the reaction mechanism is not fully understood. Model studies of the metal-cerium oxide catalysts were the principal motivation of cerium oxide growth studies on singlecrystalline transition-metal substrates, Pt(111),13,14 Rh(111),15 Re(0001),16 Au(111).17 Siokou and Nix18 grew cerium oxide on Cu(111) by depositing 10 ML of Ce at room temperature. Freshly deposited cerium was oxidized at room temperature by oxygen exposure giving Ce2O3; annealing in oxygen to 930 K gave Ce4+ oxide and discontinuous layers. In previous studies, it was shown that the reduction of CeO2 resulted in the formation of oxygen vacancies on the cerium oxide surface and a CeO2(111) f Ce2O3(0001) phase transition, which corresponds in general to the crystal structure transition from the cubic fluorite lattice (Fm3m space group) of CeO2 with the layer sequence -O2--Ce4--O2-- to Ce2O3.14-16 Bulk Ce2O3 has the hexagonal crystal structure (P-3m1) characterized by stacking of complete Ce and O layers with -Ce3+-O2-Ce3+-O2--O2-- repeated in the [0001] direction.19 The * Corresponding author. E-mail:
[email protected]; tel: +420 221 912 323; fax: +420 283 072 297. † Charles University. ‡ Sincrotrone Trieste.
electronic structure of the CeO2 oxide is characterized by unoccupied 4f states of Ce4+ (4f 0) while the Ce2O3 oxide has a Ce3+ (4f1) configuration.20 Different 4f configurations for Ce4+ and Ce3+ result in different core-level and valence-band (VB) structures.21,22 Photoelectron spectroscopy is a powerful tool for Ce 4f state investigation. There are many spectroscopic data showing different 4f configurations using Ce 3d and Ce 4d corelevel and Ce VB spectra1,13,14,16,23,24 including resonant techniques in the Ce 4d-4f photoabsorption region.4,25-31 One of the important properties of ceria is its oxygen storage capacity, which can provide oxygen to the gas mixture in catalytic contexts. The key factor for this property is the reversible transformation from Ce4+ to Ce3+. The interaction at the interface between the ceria and the added metal may promote this behavior and consequently the catalytic activity for CO oxidation, perhaps via the creation of active sites at the oxide-metal boundary.32,33 This is conventionally described as strong metal-support interaction (SMSI). In order to understand these interactions, we investigated the valence-band states of nanosized ceria powder doped with copper by means of resonant photoelectron spectroscopy in the Ce 4d-4f photoabsorption region. The obtained results were compared with those of a similar study on the model inverse catalyst prepared by growing CeO2(111) islands on a Cu(111) substrate. 2. Experimental Details Metal-loaded ceria powder was prepared by a conventional impregnation technique. The CeO2 powder of submicrometer particles (Alfa Aesar, 99.5% purity) was added to a toluene solution of Cu(O2C2H3)2‚H2O, containing 8 wt % of metal with respect to ceria. The mixture was stirred and then evaporated under vacuum at 333 K to remove toluene and dry the sample. In order to decompose copper acetate, we reduced the powder
10.1021/jp077739g CCC: $40.75 © 2008 American Chemical Society Published on Web 02/14/2008
3752 J. Phys. Chem. C, Vol. 112, No. 10, 2008
Matolı´n et al.
Figure 1. HRTEM image of (a) Cu-loaded ceria powder with (b) elemental mapping of the area; white indicates cerium, and red indicates copper.
for 2 h at 300 °C in 50 sccm of H2 gas flow. The reference powder sample was prepared by the identical treatment of the CeO2 powder in hydrogen as the Cu-loaded ceria sample. The particle morphology and nanostructure of the powder catalysts was observed using a high-resolution transmission electron microscope (HRTEM), Tecnai F30-FEG, equipped with an electron energy-loss spectroscopy (EELS) detector for elemental mapping. TEM observation was performed with a gun voltage of 300 kV. Crystal phases of the powder were analyzed by X-ray diffraction (XRD); the Rigaku RINT-2500HF X-ray diffractometer was operated at 40 kV and 300 mA using Cu KR radiation. The CO oxidation activity was determined by means of a flow chip reactor. For this treatment, 2 mg of the catalyst was deposited on the 10 × 10 mm2 SiO2/Si wafer heated by a temperature-programmed heater. We used a 2:1 stoichiometric ratio of CO/O2; the gas flow was composed of 3.0 sccm of CO, 1.5 sccm of O2, and the rest of the total flow rate of 40 sccm was Ar. The gas mixture flowed over the catalyst through a 100-µm-thick interspace. The reaction product was analyzed by means of a differentially pumped quadrupole mass spectrometer. The reactor construction ensured good thermal contact between the catalyst and its support, which was necessary for reliable catalyst temperature control. The photoemission experiments were performed at the Materials Science Beamline at the Elettra synchrotron light source in Trieste. It is a bending magnet beamline with a plane grating monochromator. The UHV experimental chamber was equipped with a 150 mm mean radius electron energy analyzer, rear view LEED optics, a dual Mg/Al X-ray source, Ce evaporation source, and an ion gun. The base pressure of the vacuum chamber was 1 × 10-10 mbar. The photoelectron spectra were recorded at different photon energies: Al KR (hν ) 1486.6 eV) for Ce 3d and Cu 2p core levels and variable excitation energy in the interval 115-130 eV for the resonant photoelectron spectroscopy (RPES) of the Ce 4f states. The powder catalyst samples were prepared for photoemission measurements by pressing into an indium foil that was mounted on the sample holder. This technique of sample preparation minimizes particle charging problems due to the photoemission process.
The model catalyst samples were prepared by evaporating metallic Ce from a Mo Knudsen cell onto the clean Cu(111) surface at 520 K in oxygen at 5 × 10-7 mbar.34 The copper single crystal supplied by MaTeck GmbH was a 1.5-mm-thick disc of 10 mm in diameter oriented to within 0.1° of the (111) plane. Cleaning was performed by cycles of sputtering and flashing to 723 K in order to obtain a sharp LEED pattern, indicating good crystalline surface quality. The surface cleanliness was checked by XPS by monitoring the C 1s and O 1s levels. Quantitative XPS analysis of the Cu 2p and Ce 3d intensity permitted us to calibrate the deposition rate, which was kept constant at 0.15 CeO2 monolayers (ML) per minute. The equivalent of 2 ML of ceria was deposited in 13 min, after which the sample was annealed at 770 K for 10 min. 3. Results 3.1. Cu-Cerium Oxide Powder Catalyst. The Cu-loaded ceria powder morphology was investigated by means of HRTEM. In Figure 1a we present a TEM image of Cu/ceria particles, and Figure 1b represents an elemental map of cerium (white) and copper (red). The image shows a large ceria particle with a faceted surface and small ceria particles with diameters around 10 nm that are attached to it. On the ceria surface, copper is present in the form of a fragmented thin film covering facets of the large particle and some facets of the small particles. The estimated Cu film thickness is around 2 nm. Copper wets the ceria surface, and there is no evidence of three-dimensional growth of Cu. Beside cerium oxide particle edges the Cu signal is detected with low intensity, probably because of the higher sensitivity in the case of the grazing incidence of the electron beam. The XRD pattern in Figure 2 shows the typical diffraction intensity of CeO2 and diffraction intensity due to copper. The most intense copper peak corresponds to Cu(111), the lower ones to Cu2O(111) and Cu(200). The Cu(111)/Cu(200) intensity ratio in the inset of Figure 2 shows the Cu(200) peak is much smaller than it should be in the case of bulk copper. A relatively high Cu(111)/Cu(200) intensity ratio together with the broad shape of the Cu(200) peak signify preferential growth of Cu(111) planes, that is, Cu(111) films. The small Cu2O(111) intensity in Figure 2 showed that copper was partially oxidized.
Photoemission Spectroscopy Study
Figure 2. XRD diffraction pattern of copper-loaded ceria powder. Copper diffraction peaks are enlarged in the inset.
Figure 3. Temperature dependence of CO2 production during treatment of reduced powder samples in a flow chip reactor in the presence of CO and O2 for (squares) pure and (circles) copper-loaded ceria.
The catalyst was tested for CO oxidation using the flow reactor. In Figure 3 we compare the CO conversion curve obtained for the reference and Cu-doped ceria powder (Figure 1a), and we see that copper considerably enhanced the catalyst activity at low temperature. The stoichiometry of cerium oxide particles was checked by means of XPS, and Figure 4a shows the Ce 3d spectra of pure (reference) and Cu-doped ceria powder. The spectra consisted of three 3d3/2-3d5/2 spin-orbitsplit doublets (f 0, f1, and f2) representing different 4f configurations in the photoemission final state and arising from 4f hybridization in both the initial and final states.23 The appearance of a high f 0 signal at 917 eV, together with an f1 peak (889 eV), which is less intense than the f2 peak (882.5 eV), is evidence of the formation of CeO2 oxide1,16,23 A shoulder that appears at binding energy BE ) 886 eV for Cu/CeO2 corresponds to the Ce3+ state. The spectrum is decomposed to elementary doublets in order to estimate the Ce3+ state concentration. By comparing f 0, f1, and f2 peak areas with that of the Ce3+ state, we obtain the ion concentration ratio Ce3+/ Ce4+ ) 0.16. It shows that in the presence of copper ceria is partially reduced to Ce2O3, probably because of the Cu-ceria interaction as reported in previous studies where it has been found that Cu facilitated the reduction of ceria powders35 or CeO2(111) surfaces.36 Alternatively, the reference ceria revealed a good CeO2 stoichiometry even after annealing of the powder in hydrogen. In Figure 4b, we can see a relatively broad Cu 2p3/2 peak showing that copper is present in a mixture of metallic
J. Phys. Chem. C, Vol. 112, No. 10, 2008 3753 and CuOx states, in agreement with the X-ray diffraction measurements that revealed diffraction patterns of metallic Cu and Cu2O. A set of resonant photoelectron spectra in the Ce 4d-4f photoabsorption region at photon energies between 115 and 130 eV obtained for the Cu/CeO2 catalyst is presented in Figure 5. At photon energy 115.0 eV, there is no resonance. We recognize two resonant features located at BE ) 1.5 and 4.5 eV that reach their maxima at hν ) 122.0 and 124.5 eV, respectively. Valenceband spectra obtained at these on-resonance photon energies together with hν ) 115.0 eV off-resonance excitation are presented in Figure 6a. The resonant features in Figure 6 can be associated with two valence states Ce4+ (BE ) 4.5 eV) and Ce3+ (BE ) 1.5 eV) corresponding to two different 4f electronic configurations in the ground state20 labeled f0O and f1O, respectively. The superscript stands for the number of 4f electrons and the subscript “O” means “oxide”. The origin of the resonant features is explained in more detail in the discussion part of this paper. The ratios Df1O/Df0O of resonance-induced increase of intensities f1O and f0O was 1.0 for Cu/CeO2 and 1.1 for the reference sample. In our previous work, we showed that this ratio was directly related to the surface concentration ratio Ce3+/ Ce4+.37 A relatively high resonance intensity of Ce3+ states observed in the case of the pure ceria sample relative to a low Ce3+ Ce 3d XPS signal can be explained by a high sensitivity of the resonance spectroscopy.37 The Ce3+ resonant feature shows occurrence of low coordination ceria states and oxygen vacancies at small ceria particle surfaces. By comparing Df1O/D f0O values for both samples in Figure 4 and 6, we can see that a relatively high increase of the Ce3+ concentration in the case of the copper-loaded sample observed by XPS in Figure 4 is not observed in the case of the resonant spectroscopy. This apparent discrepancy can be explained by the different surface sensitivity of XPS and RPES; see the Discussion section. 3.2. Cu-Cerium Oxide Model Catalyst. In order to understand the copper-ceria interaction mechanism, we investigated a model, so-called inverse, catalyst prepared by growing cerium oxide on a Cu(111) substrate. Figure 7a is a LEED pattern of the (1 × 1) Cu(111) surface taken at an electron energy of 98 eV. In Figure 7b, we present a reference diffraction pattern corresponding to saturation CO coverage obtained after exposing the Cu(111) surface at CO pressure of 5 × 10-7 mbar for 200 s (75 langmuirs, L), at a sample temperature of 120 K. The CO extra spots correspond to the (x3 × x3)R30° surface structure. After the deposition of 2 ML of ceria, the LEED pattern showed weak extra spots corresponding to the CeO2(111)/Cu(111) structure that appeared together with Cu(111) (1 × 1) spots, Figure 7c. The amount of deposited cerium oxide was chosen as small as possible in order to obtain the smallest particles showing detectable LEED diffraction pattern intensity. The CeO2(111)/Cu(111) LEED structure can be interpreted in terms of formation of an ordered cerium oxide overlayer aligned with the principal crystallographic axes of the substrate. The simultaneous appearance of the substrate and deposit diffraction spots is due to formation of an island structure as in ref 18. The diffuse background and weak character of the LEED pattern in Figure 7c signified that the ceria deposit had many defects. Therefore, the prepared ceria layer was annealed at 770 and 970 K, in both cases for 10 min, which resulted in relatively more intense cerium oxide diffraction spots as can be seen in Figure 7d and e, respectively. The discontinuous character of the ceria deposit can be seen from the CO adsorption at 120 K. CO adsorbs at low temperature on the copper substrate and not
3754 J. Phys. Chem. C, Vol. 112, No. 10, 2008
Matolı´n et al.
Figure 4. XPS (a) Ce 3d and (b) Cu 2p3/2 core-level features of pure and copper-loaded ceria powder.
Figure 5. Resonant photoelectron spectra of the Cu/CeO2 powder sample collected at photon energies between 115 and 130 eV with a step of 0.5 eV.
on the ceria surface and causes a significant decrease of the XPS Cu 2p peak area, while the Ce 3d intensity change is negligible, Table 1. The 75 L CO exposure at 120 K of the CeO2 layer annealed at 770 K gave the LEED pattern in Figure 7f. The CO extra spots correspond to the (x3 × x3)R30° structure of CO adsorption on the Cu(111) substrate. The diffraction pattern can be interpreted as indicating formation of a CeO2(111)/Cu(111) epitaxial overlayer with the morphological relationship
CeO2(111) || Cu(111), CeO2 [0 1] || Cu [0 1] The lattice parameter of copper is 0.360 nm so that the length of the [101h] lattice vector in the Cu(111) plane is aCu ) 0.255 nm. The (cubic) cerium dioxide lattice parameter is 0.54 nm, which corresponds to a [101h ] lattice vector length in the CeO2(111) plane of aCeO2 ) 0.382 nm. Thus, the aCeO2/aCu ratio is 1.50 so that we obtain very good lattice matching for the observed (1.5 × 1.5) commensurate superstructure and expect negligible strain (
