Heteroepitaxy of Cerium Oxide Thin Films on Cu(111)

An important part of fundamental research in catalysis is based on theoretical and modeling foundations which are closely connected with studies of single-crystalline catalyst surfaces. These so-called model catalysts are often prepared in the form of epitaxial thin films, and characterized using advanced material characterization techniques. This concept provides the fundamental understanding and the knowledge base needed to tailor the design of new heterogeneous catalysts with improved catalytic properties. The present contribution is devoted to development of a model catalyst system of CeO2 (ceria) on the Cu(111) substrate. We propose ways to experimentally characterize and control important parameters of the model catalyst—the coverage of the ceria layer, the influence of the Cu substrate, and the density of surface defects on ceria, particularly the density of step edges and the density and the ordering of the oxygen vacancies. The large spectrum of controlled parameters makes ceria on Cu(111) an interesting alternative to a more common model system ceria on Ru(0001) that has served numerous catalysis studies, mainly as a support for metal clusters.

fruitfully include operando experimental studies at realistic pressures of the working atmospheres, and large-scale "computer experiments" carried out using electronic-structure ab-initio theoretical methods. For the effective investigation of the structure-property relationships in Pt-ceria and metal-ceria systems the large portfolio of physical methods for highly defined synthesis and atomic-level characterization of nanostructured model ceria, Pt-ceria and metal-ceria systems allows complex model studies of ceria based systems. It includes control of the surface step density [18], the oxygen vacancy concentration and structure [19] and allows complex model studies of ceria based systems shedding light at various physical aspects of catalysis over ceria [10,20,21].
Model studies represent an efficient physical approach based on preparation of well-defined and at atomic level well-characterized surfaces, the simplest case being typically a single crystal, which can be subsequently used for molecule-surface interaction study in conditions of ultra-high vacuum (UHV) [12][13][14][15][16][17]. To obtain a sufficient level of complexity of the model surfaces, and to bridge the so-called materials gap [22] strategies are sought for controlled nanostructuring of model substrates prepared e.g., in the form of oxide heteroepitaxial thin films on metals [12,15,17,23].
During last years (metal)-cerium oxide model systems were successfully prepared by growing CeO x films on Ru(0001) [24][25][26][27][28][29], Pt(111) [30][31][32][33], Au [34] or Cu(111) [35,36] substrates. Further steps toward more realistic modeling of ceria based nanocatalysts require, however, to go beyond the current state-of-the-art and to develop new bottom-up approaches to achieve new degrees of freedom and increased level of control in preparing model ceria surfaces. In parallel it is necessary to develop new advanced techniques for characterization of the electronic and crystallographic structure, charge transfer, the morphology and molecular interactions on nanostructured metal-systems.

Continuous CeO 2 (111) Films on the Cu(111) Substrate
A basic approach to prepare the epitaxial ultra-thin CeO 2 (111) films on the Cu(111) substrate is deposition of Ce metal on clean Cu(111) substrate kept at the temperature of 520 K in a background pressure of 5ˆ10´5 Pa of O 2 [35]. This approach yields continuous films of ceria as evidenced by the LEED diffraction patterns in Figure 1 showing no contribution of Cu(111) spots for the films with equivalent thickness above 2.5 ML. Resonance Photoelectron Spectroscopy (RPES, see the Experimental Section) measurements confirm a good CeO 2 stoichiometry with 5 ML continuous film having predominantly a Ce 4+ character indicating a negligible concentration of Ce 3+ surface defects. Discontinuous CeO 2 (111) layers as on Figure 1A exhibit, on the other hand, a higher concentration of Ce 3+ and defects than continuous layers grown at the same conditions.
The LEED diffraction pattern presented in Figure 1 can be interpreted as the formation of a CeO 2 (111)/Cu(111) epitaxial overlayer with the morphological relationship: where a is the surface lattice parameter.  Suitability of the CeO2(111)/Cu(111) very thin films for mimicking the cerium oxide single-crystal surface depends on the substrate-oxide interaction that can strongly influence the chemical properties of the ceria/Cu systems as demonstrated in many studies of Cu-ceria inverse catalysts. DFT + U calculations of systems consisting of Cu atoms supported by stoichiometric and reduced CeO2(111) surfaces show that Ce 3+ species are always present underneath the Cu particles supported by stoichiometric and reduced ceria (111) surfaces [37,38]. The calculations predict a substantial charge transfer across the coherent Cu(111)/CeO2 interface leading to the full reduction of the first ceria monolayer underneath the supported Cu particles. Therefore the emerging question concerning the physicochemical properties of the CeO2(111)/Cu(111) thin films was related to the ceria-copper interaction and the extent to which this interaction determines the properties of ceria/Cu(111). Scanning Tunneling Microscopy (STM) and ab-initio calculations allowed to determine the unusual properties of the first ceria monolayer in contact with the Cu(111) substrate showing finite size effects when the limiting thickness of the oxide monolayer and the proximity of the metal substrate cause significant rearrangement of charges and oxygen vacancies compared to thicker and/or bulk ceria [39]. This rearrangement of charges is also responsible for a slight contraction of the lateral lattice constant of ultrathin ceria films on Cu(111) substrate [40].
The main property of ceria in chemical reactions is the release and the uptake of lattice oxygen to/from the reaction atmosphere. Upon leaving the ceria lattice, the neutral O atom leaves behind two electrons that localize on two Ce atoms occupying the 4f state of Ce [41]. The changes in the occupation of the 4f state result in changes in both valence band spectra and XPS spectra of Ce 3d and Ce 4d core level states due to different final state effects. The stoichiometry of cerium oxide is usually Suitability of the CeO 2 (111)/Cu(111) very thin films for mimicking the cerium oxide single-crystal surface depends on the substrate-oxide interaction that can strongly influence the chemical properties of the ceria/Cu systems as demonstrated in many studies of Cu-ceria inverse catalysts. DFT + U calculations of systems consisting of Cu atoms supported by stoichiometric and reduced CeO 2 (111) surfaces show that Ce 3+ species are always present underneath the Cu particles supported by stoichiometric and reduced ceria (111) surfaces [37,38]. The calculations predict a substantial charge transfer across the coherent Cu(111)/CeO 2 interface leading to the full reduction of the first ceria monolayer underneath the supported Cu particles. Therefore the emerging question concerning the physicochemical properties of the CeO 2 (111)/Cu(111) thin films was related to the ceria-copper interaction and the extent to which this interaction determines the properties of ceria/Cu(111). Scanning Tunneling Microscopy (STM) and ab-initio calculations allowed to determine the unusual properties of the first ceria monolayer in contact with the Cu(111) substrate showing finite size effects when the limiting thickness of the oxide monolayer and the proximity of the metal substrate cause significant rearrangement of charges and oxygen vacancies compared to thicker and/or bulk ceria [39]. This rearrangement of charges is also responsible for a slight contraction of the lateral lattice constant of ultrathin ceria films on Cu(111) substrate [40].
The main property of ceria in chemical reactions is the release and the uptake of lattice oxygen to/from the reaction atmosphere. Upon leaving the ceria lattice, the neutral O atom leaves behind two electrons that localize on two Ce atoms occupying the 4f state of Ce [41]. The changes in the occupation of the 4f state result in changes in both valence band spectra and XPS spectra of Ce 3d and Ce 4d core level states due to different final state effects. The stoichiometry of cerium oxide is usually determined by analyzing the Ce 3d XPS spectra. The spectra consist of three 3d 3/2 -3d 5/2 spin-orbit-split doublets (f 0 , f 1 and f 2 ) representing different 4f configurations in the photoemission final state and arising from 4f hybridization in both the initial and the final states [42]. The appearance of a high f 0 signal at 917 eV, together with an f 1 peak (889 eV) which is less intense than the f 2 peak (882.5 eV), is evidence of the formation of CeO 2 oxide [43,44]. Two spectral components that appear at binding energies BE = 880 and 885 eV correspond to the Ce 3+ state. In order to estimate the Ce 3+ state concentration the spectra must be decomposed to elementary doublets. However this is not a simple task because of the ambiguity of background subtraction (the energy interval of the Ce 3d spectrum is too large for correct Shirley background use), choice of elemental peak shape including asymmetry and insufficient spectrometer resolution in general [45]. A typical Ce 3d spectrum of a partially reduced cerium oxide, and the corresponding decomposition of the spectrum into the elementary doublets and the background are shown in Figure 2.
Materials 2015, 8 5 determined by analyzing the Ce 3d XPS spectra. The spectra consist of three 3d3/2-3d5/2 spin-orbit-split doublets (f 0 , f 1 and f 2 ) representing different 4f configurations in the photoemission final state and arising from 4f hybridization in both the initial and the final states [42]. The appearance of a high f 0 signal at 917 eV, together with an f 1 peak (889 eV) which is less intense than the f 2 peak (882.5 eV), is evidence of the formation of CeO2 oxide [43,44]. Two spectral components that appear at binding energies BE = 880 and 885 eV correspond to the Ce 3+ state. In order to estimate the Ce 3+ state concentration the spectra must be decomposed to elementary doublets. However this is not a simple task because of the ambiguity of background subtraction (the energy interval of the Ce 3d spectrum is too large for correct Shirley background use), choice of elemental peak shape including asymmetry and insufficient spectrometer resolution in general [45]. A typical Ce 3d spectrum of a partially reduced cerium oxide, and the corresponding decomposition of the spectrum into the elementary doublets and the background are shown in Figure 2.
XPS Ce 3d spectrum of partially reduced cerium oxide obtained using Al Kα laboratory X-ray source (1486 eV). Decomposition of the Ce 3d spectrum yields background contribution (monotonously decreasing curve), spectral peak belonging to Ce 4+ doublets (thin lines) and to Ce 3+ doublets (thick lines).
Employing tunable radiation of a soft X-ray synchrotron photoemission beamline, resonance effects in the Ce 4d-4f photoabsorption region can be used to distinguish between Ce 3+ and Ce 4+ contributions with very high sensitivity using so called Resonance Photoelectron Spectroscopy (RPES) [44,46]. This method is based on tuning the photon energy in the proximity of the resonant energy where a resonant enhancement of the Ce 4f photoemission can be observed [47]. A series of resonant valence band Ce 4f photoelectron spectra of a partially reduced CeOx is shown in Figure 3 at photon energies hν = 115-130 eV. Two resonances appear for photon energies hν = 121.5 eV and 124.5 eV corresponding to Ce 3+ (4f 1 ) and Ce 4+ (4f 0 ) valence states. At 115 eV there is no resonance. As we proposed in [35] the density of the 4f states can be obtained by subtracting the off-resonance spectrum from the on-resonance spectrum, i.e., by obtaining so called resonance enhancement DCe 3+ or DCe 4+ , Figure 2. XPS Ce 3d spectrum of partially reduced cerium oxide obtained using Al Kα laboratory X-ray source (1486 eV). Decomposition of the Ce 3d spectrum yields background contribution (monotonously decreasing curve), spectral peak belonging to Ce 4+ doublets (thin lines) and to Ce 3+ doublets (thick lines).
Employing tunable radiation of a soft X-ray synchrotron photoemission beamline, resonance effects in the Ce 4d-4f photoabsorption region can be used to distinguish between Ce 3+ and Ce 4+ contributions with very high sensitivity using so called Resonance Photoelectron Spectroscopy (RPES) [44,46]. This method is based on tuning the photon energy in the proximity of the resonant energy where a resonant enhancement of the Ce 4f photoemission can be observed [47]. A series of resonant valence band Ce 4f photoelectron spectra of a partially reduced CeO x is shown in Figure 3 at photon energies hν = 115-130 eV. Two resonances appear for photon energies hν = 121.5 eV and 124.5 eV corresponding to Ce 3+ (4f 1 ) and Ce 4+ (4f 0 ) valence states. At 115 eV there is no resonance. As we proposed in [35] the density of the 4f states can be obtained by subtracting the off-resonance spectrum from the on-resonance spectrum, i.e., by obtaining so called resonance enhancement DCe 3+ or DCe 4+ , see Figure 4. The resonant enhancement ratio (RER) DCe 3+ /DCe 4+ was proposed and is used as a parameter sensitively indicating the degree of reduction of cerium oxide surface. Besides the high sensitivity to small concentrations of Ce 3+ the energy of detected photoelectrons in the range of 100 eV guarantees the highest surface sensitivity of the RPES the Ce 4d-4f signal.
Materials 2015, 8 6 see Figure 4. The resonant enhancement ratio (RER) DCe 3+ /DCe 4+ was proposed and is used as a parameter sensitively indicating the degree of reduction of cerium oxide surface. Besides the high sensitivity to small concentrations of Ce 3+ the energy of detected photoelectrons in the range of 100 eV guarantees the highest surface sensitivity of the RPES the Ce 4d-4f signal.  The resonance spectra shown in Figure 4 were obtained for the sample analyzed by LEED in Figure 1c showing that highly sensitive Ce 4d-4f RPES yields RER close to zero, i.e., perfect stoichiometry of the ceria film.

Adjusting the Morphology of CeO2(111) Nanostructured Thin Films on Cu(111)
Scanning Tunneling Microscopy (STM) represents a primary research tool for investigating morphology of nanostructured ceria and metal-ceria samples yielding an indispensable input for the see Figure 4. The resonant enhancement ratio (RER) DCe 3+ /DCe 4+ was proposed and is used as a parameter sensitively indicating the degree of reduction of cerium oxide surface. Besides the high sensitivity to small concentrations of Ce 3+ the energy of detected photoelectrons in the range of 100 eV guarantees the highest surface sensitivity of the RPES the Ce 4d-4f signal.  The resonance spectra shown in Figure 4 were obtained for the sample analyzed by LEED in Figure 1c showing that highly sensitive Ce 4d-4f RPES yields RER close to zero, i.e., perfect stoichiometry of the ceria film.

Adjusting the Morphology of CeO2(111) Nanostructured Thin Films on Cu(111)
Scanning Tunneling Microscopy (STM) represents a primary research tool for investigating morphology of nanostructured ceria and metal-ceria samples yielding an indispensable input for the The resonance spectra shown in Figure 4 were obtained for the sample analyzed by LEED in Figure 1c showing that highly sensitive Ce 4d-4f RPES yields RER close to zero, i.e., perfect stoichiometry of the ceria film.

Adjusting the Morphology of CeO 2 (111) Nanostructured Thin Films on Cu(111)
Scanning Tunneling Microscopy (STM) represents a primary research tool for investigating morphology of nanostructured ceria and metal-ceria samples yielding an indispensable input for the advanced structure-property studies. Local information of the morphology of the model catalysts combines favorably with the information on their electron and chemical state obtained by space-averaging experimental techniques. STM imaging provides information on densities of atomic step edges [18], densities and sizes of metal nanoclusters on ceria [48] or densities of surface oxygen vacancies [20]. Atomically resolved STM imaging provides information on surface reconstructions that in turn represents a complementary information on the charge state of ceria surfaces [39,49].
Adjustable morphology and degree of reduction represent desirable properties of model oxide substrates for heterogeneous catalysis [18] prepared in form of single crystals or heteroepitaxial single-crystalline thin films. The density of atomic steps in the ceria layer determines the dispersion and the electronic structure of the ceria-supported metal clusters [50], because the atomic steps on ceria serve as preferential nucleation sites for many metals [50][51][52][53].
A range of bottom-up experimental approaches that allow preparation of oriented thin films of CeO 2 (111) on Cu(111) with deterministically controlled density of atomic steps [18] and the density and spatial ordering of oxygen vacancies [19,49] has been developed. These approaches rely on self-organization properties of cerium and oxygen atoms on the Cu substrate and in ceria and utilize careful control of deposition parameters of the ceria layers. Varying the substrate temperature during layer growth the density of atomic steps can be changed between approximately 5% and 20% ( Figure 5A-C) [18]. advanced structure-property studies. Local information of the morphology of the model catalysts combines favorably with the information on their electron and chemical state obtained by space-averaging experimental techniques. STM imaging provides information on densities of atomic step edges [18], densities and sizes of metal nanoclusters on ceria [48] or densities of surface oxygen vacancies [20]. Atomically resolved STM imaging provides information on surface reconstructions that in turn represents a complementary information on the charge state of ceria surfaces [39,49]. Adjustable morphology and degree of reduction represent desirable properties of model oxide substrates for heterogeneous catalysis [18] prepared in form of single crystals or heteroepitaxial single-crystalline thin films. The density of atomic steps in the ceria layer determines the dispersion and the electronic structure of the ceria-supported metal clusters [50], because the atomic steps on ceria serve as preferential nucleation sites for many metals [50][51][52][53].
A range of bottom-up experimental approaches that allow preparation of oriented thin films of CeO2(111) on Cu(111) with deterministically controlled density of atomic steps [18] and the density and spatial ordering of oxygen vacancies [19,49] has been developed. These approaches rely on self-organization properties of cerium and oxygen atoms on the Cu substrate and in ceria and utilize careful control of deposition parameters of the ceria layers. Varying the substrate temperature during layer growth the density of atomic steps can be changed between approximately 5% and 20% ( Figure 5A-C) [18].

Adjusting the Stoichiometry of CeO2(111) Nanostructured Thin Films on Cu(111)
For obtaining a broad range of reduction of ceria layers on Cu(111) we developed a method based on physical vapor deposition of metallic Ce onto a stoichiometric CeO2(111) film, i.e., on using metallic Ce as a homotypical reducing species [19,49]. We demonstrated that following the reactive interaction of the two components it is possible to obtain highly ordered films of Ce2O3 on Cu(111) [19,49] as well as on the Ru(0001) [54] substrate. Ce2O3 represents the ultimate reduction of ceria that is difficult to obtain by other methods practically used for reducing ceria samples. The extremely sensitive RPES reveals no Ce 4+ contribution (at 3.56 eV) in the Ce2O3 layer after reaction (the uppermost spectrum in Figure 6B). The XPS Ce 3d spectra in Figure 6C point that the reaction of ceria layers with metallic Ce yields bulk-reduced samples of Ce2O3 [49].

Adjusting the Stoichiometry of CeO 2 (111) Nanostructured Thin Films on Cu(111)
For obtaining a broad range of reduction of ceria layers on Cu(111) we developed a method based on physical vapor deposition of metallic Ce onto a stoichiometric CeO 2 (111) film, i.e., on using metallic Ce as a homotypical reducing species [19,49]. We demonstrated that following the reactive interaction of the two components it is possible to obtain highly ordered films of Ce 2 O 3 on Cu(111) [19,49] as well as on the Ru(0001) [54] substrate. Ce 2 O 3 represents the ultimate reduction of ceria that is difficult to obtain by other methods practically used for reducing ceria samples. The extremely sensitive RPES reveals no Ce 4+ contribution (at 3.56 eV) in the Ce 2 O 3 layer after reaction (the uppermost spectrum in Figure 6B). The XPS Ce 3d spectra in Figure 6C point that the reaction of ceria layers with metallic Ce yields bulk-reduced samples of Ce 2 O 3 [49].
Upon stepwise reduction of CeO 2 (111) by metallic Ce, LEED measurements reveal surface reconstructions in the reduced ceria that can be characterized as 1ˆ1, ( ' 7ˆ'7) R19.1˝, (3ˆ3), and 4ˆ4 ( Figure 6A), [19]. Photoemission data analysis of the Ce 3d core-level spectra reveals relative concentrations of Ce 3+ corresponding to surface terminations of ordered bulk phases of reduced ceria, the ι-Ce 7 O 12 or CeO 1.71 phase for the ( ' 7ˆ'7) R19.1˝reconstruction, the CeO 1.67 phase for the (3ˆ3) reconstruction, and the cubic bixbyite c-Ce 2 O 3 (111) phase for the (4ˆ4) reconstruction ( Figure 6B). Besides the varying concentration of oxygen vacancies, these bulk reduced phases also represent distinct regular arrangements of oxygen vacancies in cubic ceria.
Materials 2015, 8 8 Upon stepwise reduction of CeO2(111) by metallic Ce, LEED measurements reveal surface reconstructions in the reduced ceria that can be characterized as 1 × 1, (√7 × √7) R19.1°, (3 × 3), and 4 × 4 ( Figure 6A), [19]. Photoemission data analysis of the Ce 3d core-level spectra reveals relative concentrations of Ce 3+ corresponding to surface terminations of ordered bulk phases of reduced ceria, the ι-Ce7O12 or CeO1.71 phase for the (√7 × √7) R19.1° reconstruction, the CeO1.67 phase for the (3 × 3) reconstruction, and the cubic bixbyite c-Ce2O3(111) phase for the (4 × 4) reconstruction ( Figure 6B). Besides the varying concentration of oxygen vacancies, these bulk reduced phases also represent distinct regular arrangements of oxygen vacancies in cubic ceria. Reduction of ceria with metallic Ce allows adjusting the concentration of oxygen vacancies between 0% and 25% [19]. Width of STM images 60 nm; (B) RPES spectra of the valence band of the ordered phases of stoichiometric and reduced ceria on Cu(111). The spectra are measured off-resonance (photon energy 115 eV, dotted lines), in the Ce 4+ resonance (124.8 eV, dashed lines), and in the Ce 3+ resonance (121.4 eV, full lines). The resonance enhancements D (Ce 3+ ) and D (Ce 4+ ) are indicated by arrows [19]; (C) XPS of cerium oxide before (thin lines) and after (thick lines) the reaction between Ce and CeO2 buffer layer. The Ce 3d core-level spectra were taken at two different emission angles indicating no changes in the stoichiometry of cerium oxide with information depth [49].
By combining ceria reduction by Ce and oxidation by annealing in oxygen it was shown that both processes are fully reversible. Annealing of the reduced ceria layers in oxygen preserves the morphology of the reduced ceria layer; in particular, the low step density of the Ce2O3 thin films  [19]. Width of STM images 60 nm; (B) RPES spectra of the valence band of the ordered phases of stoichiometric and reduced ceria on Cu(111). The spectra are measured off-resonance (photon energy 115 eV, dotted lines), in the Ce 4+ resonance (124.8 eV, dashed lines), and in the Ce 3+ resonance (121.4 eV, full lines). The resonance enhancements D (Ce 3+ ) and D (Ce 4+ ) are indicated by arrows [19]; (C) XPS of cerium oxide before (thin lines) and after (thick lines) the reaction between Ce and CeO 2 buffer layer. The Ce 3d core-level spectra were taken at two different emission angles indicating no changes in the stoichiometry of cerium oxide with information depth [49].
By combining ceria reduction by Ce and oxidation by annealing in oxygen it was shown that both processes are fully reversible. Annealing of the reduced ceria layers in oxygen preserves the morphology of the reduced ceria layer; in particular, the low step density of the Ce 2 O 3 thin films shown in Figure 7A is preserved upon oxidation to CeO 2 ( Figure 7B) [19]. The CeO 2 layers obtained by oxidation of Ce 2 O 3 exhibit the smallest step density and the highest degree of oxidation from the above described model systems CeO 2 /Cu(111). However, the contraction of the lattice constant of ceria upon oxidation causes cracking of the ceria layer revealing up to 2% of the Cu substrate on reoxidized Ce 2 O 3 /Cu(111) samples ( Figure 7C). Still, the highly ordered ceria surface represents a suitable substrate for STM experiments that can be further modified e.g., by homoepitaxy and high-temperature annealing of CeO 2 for increasing the density of steps on the ceria surface without destabilizing the surface thermally ( Figure 7D).
shown in Figure 7A is preserved upon oxidation to CeO2 ( Figure 7B) [19]. The CeO2 layers obtained by oxidation of Ce2O3 exhibit the smallest step density and the highest degree of oxidation from the above described model systems CeO2/Cu(111). However, the contraction of the lattice constant of ceria upon oxidation causes cracking of the ceria layer revealing up to 2% of the Cu substrate on reoxidized Ce2O3/Cu(111) samples ( Figure 7C). Still, the highly ordered ceria surface represents a suitable substrate for STM experiments that can be further modified e.g., by homoepitaxy and high-temperature annealing of CeO2 for increasing the density of steps on the ceria surface without destabilizing the surface thermally ( Figure 7D). Practically, ceria layers reduced by the interface reaction with metallic Ce represent a realization of the ideal scenario of reduction and reoxidation of ceria by removing/adding oxygen from/to the fluorite CeO2 lattice without largely modifying the structure of the Ce sub-lattice. This, accompanied by the preference of oxygen vacancies to arrange in regular structures, makes the ceria layers reduced by interface reaction with metallic Ce a unique experimental playground for studying the influence of the oxygen vacancy concentration and coordination on the physico-chemical properties of nanostructured ceria.

Conclusions
In the present contribution we review methods of preparation and characterization of ceria-based model systems in the form of thin films layers of cerium oxide with different surface stoichiometry, structure, the density of surface steps and oxygen vacancies epitaxially grown on Cu(111).
Heteroepitaxial ultra-thin CeO2(111) films of different thickness were grown on the Cu(111) substrate, typically at 520 K in 5 × 10 −5 Pa of O2. The LEED diffraction pattern shows no contribution of Cu(111) spots for the film equivalent thickness above 2.5 ML. The resonance photoelectron spectroscopy at the Ce 3d→4f resonance has been used as an efficient tool for determining the surface stoichiometry of cerium oxide. Discontinuous CeO2(111) layers exhibit a higher concentration of defects than continuous layers grown at the same conditions. The 5 ML continuous films can be prepared with practically Ce 3+ free surfaces; on the other hand concentration of Ce 3+ site and oxygen vacancies can be tailored by combining growth at constant and variable temperature. We can obtain Practically, ceria layers reduced by the interface reaction with metallic Ce represent a realization of the ideal scenario of reduction and reoxidation of ceria by removing/adding oxygen from/to the fluorite CeO 2 lattice without largely modifying the structure of the Ce sub-lattice. This, accompanied by the preference of oxygen vacancies to arrange in regular structures, makes the ceria layers reduced by interface reaction with metallic Ce a unique experimental playground for studying the influence of the oxygen vacancy concentration and coordination on the physico-chemical properties of nanostructured ceria.

Conclusions
In the present contribution we review methods of preparation and characterization of ceria-based model systems in the form of thin films layers of cerium oxide with different surface stoichiometry, structure, the density of surface steps and oxygen vacancies epitaxially grown on Cu(111).
Heteroepitaxial ultra-thin CeO 2 (111) films of different thickness were grown on the Cu(111) substrate, typically at 520 K in 5ˆ10´5 Pa of O 2 . The LEED diffraction pattern shows no contribution of Cu(111) spots for the film equivalent thickness above 2.5 ML. The resonance photoelectron spectroscopy at the Ce 3dÑ4f resonance has been used as an efficient tool for determining the surface stoichiometry of cerium oxide. Discontinuous CeO 2 (111) layers exhibit a higher concentration of defects than continuous layers grown at the same conditions. The 5 ML continuous films can be prepared with practically Ce 3+ free surfaces; on the other hand concentration of Ce 3+ site and oxygen vacancies can be tailored by combining growth at constant and variable temperature. We can obtain independent control of coverage and step density of the ceria layers on Cu(111) and prepare ceria layers with adjustable density of the surface steps.
It was shown that interfacial reaction of a stoichiometric CeO 2 thin film on Cu(111) with deposited metallic Ce yields a highly ordered layer of cubic bixbyite c-Ce 2 O 3 (111). The surface structure of the layer corresponds to bulk-terminated c-Ce 2 O 3 (111). It contains ordered vacancy clusters, each consisting of four oxygen vacancies. The surface exhibits a very characteristic and sharp (4ˆ4) LEED pattern relative to CeO 2 (111), allowing easy experimental identification. We suggest that the c-Ce 2 O 3 (111) film is a unique model experimental system for highly reduced ceria surfaces. It provides an atomically well-defined surface exposing exclusively Ce 3+ ions and a high density of oxygen vacancies with a precisely defined environment. A stepwise reduction of CeO 2 (111) by metallic Ce leads to different surface reconstructions in the reduced ceria that can be characterized as ( ' 7ˆ'7) R19.1˝and (3ˆ3) surface structures corresponding to samples with ceria stoichiometry CeO 1.71 and CeO 1.67 .
The high degree of control of the basic morphological and chemical properties of thin film ceria on Cu(111) presented in this article make them versatile substrates for present and future model catalytic experiments revealing the most important information about the relations between the morphology, stoichiometry, electronic structure and chemical reactivity in ceria based catalysts.