Distinct Physicochemical Properties of the First Ceria Monolayer on

Article pubs.acs.org/JPCC

Distinct Physicochemical Properties of the First Ceria Monolayer on Cu(111) Lucie Szabová,† Oleksandr Stetsovych,† Filip Dvořaḱ ,† Matteo Farnesi Camellone,‡,§,∥ Stefano Fabris,‡,§ Josef Mysliveček,*,‡ and Vladimír Matolín‡ †

Faculty of Mathematics and Physics, Charles University, V Holešovičkách 2, Praha 8, Czech Republic Theory@Elettra group, Istituto Officina dei Materiali, CNR-IOM DEMOCRITOS, s.s. 14 km 163,5 in AREA Science Park, I-34149 Trieste, Italy § SISSA, Scuola Internazionale Superiore di Studi Avanzati, via Bonomea 265, I-34136 Trieste, Italy ‡

ABSTRACT: Discontinuous ceria layers on Cu(111) represent heterogeneous catalysts with notable activities in water− gas shift and CO oxidation reactions. Ultrathin ceria islands in these catalysts are composed of monolayers of ceria exhibiting CeO2(111) surface ordering and bulklike vertical stacking (O−Ce−O) down to a single ceria monolayer representing the oxide-metal interface. Scanning tunneling microscopy (STM) reveals marked differences in strain buildup and the structure of oxygen vacancies in this first ceria monolayer compared to thicker ceria layers on Cu(111). Ab-initio calculations allow us to trace back the distinct properties of the first ceria monolayer to pronounced finite size effects when the limiting thickness of the oxide monolayer and the proximity of metal substrate cause significant rearrangement of charges and oxygen vacancies compared to thicker and/or bulk ceria.

■

INTRODUCTION In heterogeneous catalysis, oriented thin films of oxides on metal substrates have traditionally been used as model systems mimicking the surfaces of bulk oxides.1 Decreasing the thickness of the oxide films causes emerging of a broad range of effects that significantly influence the catalytic properties.2,3 When the oxide thickness reduces to few monolayers (ML) oxide stoichiometry,4 electronic and crystalline structure,5−7 and charge of molecular adspecies8 start to differ significantly from thicker films or bulk-truncated surfaces. Ultrathin oxide films are often discontinuous representing inverse model catalysts with unique catalytic properties.9,10 Cerium oxide (ceria, CeO2) is a broadly studied oxide with a vast application potential in catalysis and energy conversion.11−13 Continuous thin films of ceria have been prepared on Ru(0001)14,15 and Cu(111).16,17 These films have served numerous model studies evaluating the catalytic properties of bare ceria surfaces18,19 as well as ceria surfaces activated with metal clusters.20−23 Discontinuous ultrathin films of ceria have been prepared on various metal substrates. Some of these ceriabased inverse model catalysts show exceptional activity in technologically relevant reactions such as water−gas shift (WGS)24,25 and CO oxidation.26−28 Particularly, ceria on Cu(111) is a featured inverse model catalyst inspiring design of novel noble-metal-free industrial catalysts.25,29 The high activity of inverse model catalysts is explained by a cooperative action of oxide and metal at the perimeter of oxide islands.24−28 However, as first pointed out by Castellarin-Cudia et al.,30 the surface of ultrathin oxide islands in inverse model catalysts may itself incorporate catalytically active sites, which © 2012 American Chemical Society

are not stable on thicker films and/or bulk oxide surfaces. Indeed, microscopic studies of 1−3 ML thick ceria islands on various metal substrates confirm that ultrathin ceria is strained,30−32 influenced by coincidence effects between substrate and oxide lattices,15,17,30 and shows several characteristic phenomena regarding oxygen vacancies, such as their ordering in coincidence structures30−32 or their segregation at metal-oxide interfaces or at oxide surfaces in islands of various thickness (1 or more ML).32 In the present work, we provide clear experimental evidence of these effects in ultrathin ceria on Cu(111) and identify their origins on the basis of ab-initio calculations. Particularly, we describe differences in structure and morphology between 1-ML-thick and thicker ceria as obtained with atomic-resolution scanning tunneling microscopy (STM). These measurements reveal the distinct strain buildup and behavior of oxygen vacancies in 1-ML thin films of ceria on Cu(111). The ab-initio calculations show that both the strain buildup and the varying nature of oxygen vacancies in 1-ML and thicker layers represent finite size effects caused by restrictions imposed upon charge transfer and charge localization in the oxide by the limiting thickness of the ceria layer and by the close proximity the metal substrate. Our study shows that, in comparison to ceria surfaces and metal−supported thick ceria overlayers, 1-ML thin films of ceria on Cu(111) display distinct physicochemical properties, Received: December 12, 2011 Revised: February 10, 2012 Published: March 13, 2012 6677

dx.doi.org/10.1021/jp211955v | J. Phys. Chem. C 2012, 116, 6677−6684

The Journal of Physical Chemistry C

Article

the free-standing ceria slabs was calculated by relaxing all internal degrees of freedom for a set of lattice parameters a.

which likely contribute to the enhanced reactivity of ceria-based inverse model catalysts.

■

■

METHODS Experiment. The experiments were performed in an ultra high-vacuum system with a base pressure of 1 × 10−8 Pa. Cu(111) substrates were cleaned by cycles of Ar+ sputtering and annealing in vacuum. Discontinuous CeO2 layers on Cu(111) were obtained by evaporation of Ce in a background oxygen atmosphere of 5 × 10−5 Pa, and substrate temperature of 400 °C. Ce was evaporated from a Ta crucible heated by electron bombardment at a rate yielding ∼5 ML/h CeO2 on the substrate. The amount of CeO2 on Cu(111) on the samples was up to 0.5 ML. Oxygen background was established ∼2 min before starting the Ce deposition. Immediately after deposition, sample heating was switched off for best preservation of growth shapes of the ceria islands. Oxygen atmosphere was pumped out when the sample temperature dropped below 100 °C. Samples were investigated at room temperature in a homemade STM using chemically etched tungsten tips. STM images were obtained tunneling the electrons into unoccupied states of the sample. STM data were processed using opensource software Gwyddion.33 Some of the atomically resolved images in this study were obtained on samples prepared alternatively by titrating an oxidized Cu(111) surface with minute amounts of Ce (less than 0.1 ML) without oxygen background. Ce reduces the surface copper oxide, yielding sparse ceria islands and extended areas of unoxidized Cu(111) favorable for the stabilization of the STM tip. Properties of ceria islands discussed in the present study remain the same for both methods of sample preparation. Calculation. The spin-polarized calculations were based on density functional theory (DFT and DFT+U)34 and employed the Perdew−Burke−Ernzerhof exchange and correlation functional.35 They were performed in the plane-wave pseudopotential method as implemented in the PWscf code of the Quantum ESPRESSO distribution.36 Following our previous DFT+U analysis on ceria-based materials,37−44 we set the value of the parameter U acting on the Ce 4f states to 4.5 eV, which is consistent with the literature for this system, reporting values between 4.5 and 5 eV.39,45−49 The plane-wave basis set and the representation of the electron density were limited by the energy cutoffs of 30 and 300 Ry, respectively. The interaction of the valence electrons with the ionic cores was represented by Vanderbilt ultrasoft pseudopotentials. Integrals in the Brillouin zone of the supercell slab described below were performed on a Monkhorst−Pack (4 × 4 × 1) k-point mesh and by using a Methfessel−Paxton smearing of 0.02 Ry. Convergency of the results with respect to the computational parameters was determined in a previous work.42,50 The CeO2(111)/Cu(111) interface was modeled by means of hexagonal and ortorhombic supercells (for stoichiometric and defective films, respectively) comprising 1 or 2 MLs (O−Ce−O trilayer) (2 × 2) CeO2(111) film in contact with a four-layer (3 × 3) Cu(111) slab. The external surfaces were separated in the perpendicular direction by 11 Å of vacuum. Shifts of one slab with respect to the other as well as the Cu−O interfacial distances were established in a previous work.42 The atomic positions of the metal/oxide interface and of the ceria thin film were optimized according to the Hellmann−Feynman forces. During the relaxation, the two bottom Cu layers were kept fixed in their bulk-like position. The lattice parameter of

RESULTS AND DISCUSSION

STM of Ceria Islands on Cu(111). Orientation of Ceria on Cu(111). Inverse model catalysts of ceria on Cu(111) are prepared by dosing Ce on Cu(111) substrate in a background atmosphere of 5 × 10−5 Pa O2 and substrate temperature about 400 °C.25,28 The highest chemical activity of these model catalysts is observed when approximately 20% of the copper substrate is covered by ceria islands.24,25,28 At this coverage and substrate temperature, most of the ceria islands are formed by one complete ceria ML (i.e., a vertical stack of interfacial O, Ce, and surface O).17,25,28 For the purpose of the present study, we prepared samples with an increased amount of ceria of about 0.5 ML deposited at 400 °C. The islands are composed of ceria monolayers stacked one on top of each other [Figure 1a].17 The islands are locally 1, 2, and >2 ML thick and allow for a direct comparison of ceria layers with different thickness in one experiment. Step heights of the islands are 3.1 Å for 1 ML relative to unoxidized Cu substrate (determined in Ce titration experiment), 3.2 Å for 2 ML relative to 1 ML, and 2.9 Å for 3 ML relative to 2 ML, corresponding to a stack of complete ceria monolayers residing on unoxidized Cu substrate.

Figure 1. (a) Azimuthal orientations of ceria islands on Cu(111). M and W (45% islands each) are islands with ceria [112] antiparallel and parallel to Cu [112]. X (10% islands) are islands with random orientation. Image width 140 nm. (b−d) Orientation-dependent textures in the 1-ML ceria: (b) “moiré” on M, (c) “waggon-wheel” on W, (d) “small moiré” on X oriented islands, respectively. The textures reflex a compressive strain in the 1-ML ceria. In panels b and c, the 2 × 2 atom pattern is indicated with red lines, and periodically missing 2 × 2 is indicated with green fill. Image widths are (b,c) 16 nm and (d) 8 nm. Sample bias and tunneling current: (a,b) 3 V, 0.2 nA; (c) 1.5 V, 0.3 nA; (d) 1.2 V, 0.3 nA. 6678



Article

localization of oxygen vacancies that are very specific for the first ceria monolayer on Cu(111). 1-ML Ceria: Strain. Morphological patterns on the lateral scale of nanometers can be traced back to the coincidence effects between ceria and the metal substrate.15,17,28,30,31 We observe that these patterns in CeO2/Cu(111) are strongly dependent on the azimuthal orientation of oxide islands. Typical morphologies observed by STM in the first monolayer ceria in M-, W-, and X-oriented islands are shown in Figure 1b−d. The atomically resolved images were low pass-filtered to remove noise induced by frequent tip−sample interactions. The change of the morphological pattern between M and W islands reveals the extent of the metal−oxide interaction in CeO2/ Cu(111). For obtaining a change in the geometry of in CeO2/ Cu(111) upon 180° rotation of the oxide, at least two monolayers of Cu must interact with the oxide (cf. Figure 5). The STM pattern on the 1-ML ceria in M-oriented islands (Figure 1b) corresponds to the moiré-like coincidence of substrate and adsorbate (111) planes with dissimilar lattice constant observed in many substrate−adsorbate systems.54 In spite of the fact that ceria is expected to grow on Cu(111) pseudomorphically in an almost ideal 2:3 relation with a negligible 0.2% expansion of ceria, electron diffraction experiments on ceria on Pt(111)32 and on Cu(111)55 reveal contraction of the ultrathin ceria films. On the basis of the periodicity of the moiré pattern, we estimate the contraction of the 1-ML ceria on Cu(111) to 3.2 ± 0.3%. Atomic resolution obtained within the moiré pattern allows for determining minute deviations in the azimuthal orientations between the substrate and the oxide.56 Within the precision of our measurement, we make an upper estimate of the rotation between the lattices of Cu and ceria of 0.5°. The morphology of the first ceria monolayer in W-oriented islands (Figure 1c) corresponds to a “waggon-wheel” superstructure, which is a generalized moiré superstructure.57 The atomically resolved detail in Figure 1c shows a larger periodicity than the moiré structure on M islands (Figure 1b). However, on most W areas, the waggon-wheel superstructure features a periodicity matching the moiré pattern on M islands. Thus, we do not detect major differences in the ceria contraction and/or rotation with respect to substrate in M and W islands, respectively. The differences between superstructures in M and W islands can be discussed in terms of different relaxations of oxide and metal atoms. Indeed, changes in the registry of substrate and oxide atoms by a fraction of the interatomic distance significantly influence the STM contrast in ultrathin oxide films on metals.30,54 We propose that the relaxations of oxide atoms in M islands are minimal because the moiré pattern is already obtained for isotropic (111) lattices. In W islands, the relaxations of oxide atoms may be more pronounced to induce a departure from the moiré pattern toward the waggon-wheel pattern. The morphology of the 1-ML ceria in X islands shows many variations strongly depending on the actual azimuthal orientation of the islands. An example of a moiré-like pattern with a periodicity of 1.2 nm is shown in Figure 1d. However, with comparable probability, the first ceria monolayer in X islands shows stripe patterns, random height modulation, or no height modulation, all on different islands within a single sample. 1-ML Ceria: Oxygen Vacancies. Apart from patterns on the nanometer scale, 1-ML ceria in M and W islands shows modification to the 1 × 1 atom pattern of CeO2(111). We identify areas of 2 × 2 reconstruction (Figure 1b,c, red lines) that are locally, and in registry with the moiré pattern, interleaved

The triangular shape of the islands (Figure 1a) differs from the hexagonal shape of islands and pits on the surfaces of thermally equilibrated bulk CeO2(111).51 Islands on equilibrated ceria surfaces are delimited by the equal proportion of straight monolayer-high steps in [112̅] and [112] directions, indicating that the free energy of the (112̅) and (112) steps on ceria are very comparable.51 The triangular shape of islands in our experiment is rather a result of the different propagation speeds of these two steps under kinetic conditions, delimited by a slow-propagating step.52,53 The triangular shape and orientation of the island is transferred from the first monolayer to higher monolayers. This indicates that the islands are coherent single crystals of ceria. Ceria islands exhibit different azimuthal orientations with respect to the copper surface. Two major populations of islands (45% islands each) are represented by islands that are mutually rotated by 180°. These islands are marked “M” and “W” in Figure 1 (a). Previous investigations of CeO2(111)/Cu(111) showed that the main crystallographic directions of ceria layer correspond to that of the copper substrate.16 Thus, the ceria [1̅12] direction must be parallel and antiparallel to the Cu [1̅12] direction in M and W islands. A small number of ceria islands (10%) show a random azimuthal orientation. An example of such island is marked “X” in Figure 1a. In a more detailed view, STM on the 1-ML ceria film reveals two characteristic patterns: one extending, on the lateral scale, for several nanometers (cf. Figure 1b−d), and the other ranging in the order of ångstroms (cf. Figure 2a,b). These patterns are largely suppressed or absent on the 2-ML or thicker ceria films (cf. Figure 2c). These patterns allow us to draw conclusions about the metal−oxide interaction, the strain buildup, and the

Figure 2. (a,b) 2 × 2 atom pattern in the 1-ML ceria mutually rotated by 180° between M and W islands (cf. Figure 1). 2 ×2 reconstruction on ceria surface is indicative of subsurface oxygen vacancies. (c) In 2ML ceria, the surface is unreconstructed. 1 ×1 and 2 × 2 unit cells are highlighted red. Brighter protrusion in 2 × 2 unit cell is marked with cross. Image width for a−c is 3 nm. Sample bias and tunneling current: (a) 1.5 V, 0.3 nA; (b) 1 V, 0.3 nA; (c) 2.3 V, 0.15 nA. 6679



Article

with stripes showing rather 3 × 2 reconstruction (Figure 1b,c, green fill). On M islands, regular 2 × 2 or 3 × 2 patterns cover the entire surface of the 1-ML ceria. On W islands, the 2 × 2 or 3 × 2 is localized among less ordered areas. A detailed view of the 2 × 2 reconstruction in M and W islands is shown in Figure 2a,b, respectively. The images were band-pass-filtered to remove noise and the nanometer-scale modulation. A unit cell of the 2 × 2 reconstruction is marked in both images. The unit cell includes one bright and three darker protrusions as well as one darker and three brighter depressions. The unit cell is rotated by 180° between M and W islands in accordance with the orientation of the ceria layer in these islands. The 2 × 2 reconstructions on the surface of bulk ceria58 as well as on thin ceria films32 have been interpreted as a fingerprint of subsurface oxygen vacancies. This allows us to propose that the first ceria monolayer on Cu(111) includes oxygen vacancies at the ceria/Cu interface. Given the long-range periodicity of the 2 × 2 pattern, the vacancies form a regular array perturbed locally (in 3 × 2 areas) by coincidence effects between substrate and oxide layer. 2-ML Ceria. In 2 ML thick areas of both M and W islands, the 2 × 2 surface reconstruction cannot be observed (cf. Figure 2c for the W island). This indicates a different behavior of oxygen vacancies in 2-ML and thicker ceria similar to observations of ultrathin ceria on Pt.32 From the patterns on the nanometer scale observed in 1-ML ceria (Figure 1) only moiré on M islands becomes partly visible in 2-ML ceria. This indicates a significantly reduced influence of the substrate. The moiré in 2-ML ceria reveals a reduced contraction of 2.4 ± 0.3%. DFT Simulations. Strain Analysis and Defects Energetics. The experimentally observed moiré pattern on the 1-ML ceria film points to a contraction of the ceria layer with respect to its equilibrium bulk lattice parameter. Our DFT calculations show that this is a size effect stemming from the confined dimensions of the slab in the perpendicular direction. Indeed, the lattice parameter a of an unsupported 1-ML ceria slab turns out to be smaller than the bulk CeO2 value (a0) by 6.7% (Table 1). On

than 1 ML can therefore yield small strain build-up at the interfaces with Cu(111), even in the absence of crystal and electron defects. The 1-ML ceria thin films present a completely different behavior than thicker ceria films with respect to the location of O vacancies. The calculated relative energetics of an O vacancy in the symmetry-inequivalent sites of 1- and 2-ML CeO2/Cu systems is reported in Table 2. The symmetry-inequivalent sites Table 2. Energy Difference (ΔE) between Structures Containing O Vacancies at Symmetry-Inequivalent O Sites of the First Interface Layer (OI) with Respect to the Outermost Surface Sites (OIII and OVI in the 1 ML and 2 ML Ceria Slabs, Respectively)a 1 ML CeO1.75/Cu

2 ML CeO1.875/Cu

site

ΔE [eV]

OTI OBI OBIII OTI VI

−1.20 −0.82 0 +1.08 0

O

a

Labeling of O sites is defined in Figure 3.

are defined in the top and side views of the supported ceria films displayed in Figure 3. The calculations show that O vacancies tend to segregate at the metal-oxide interface in the case of 1-ML ceria film with a driving force of −1.2 eV. Among the two inequivalent O sites at the interface, the vacancy at the O site on top of a Cu atom (OTI ) is more favored than at the O site bridging between two Cu atoms (OBI ) by 0.38 eV. In the case of thicker ceria films, an oxygen vacancy in the same (2 × 2) lateral supercell would instead segregate at the outermost surface layer (OVI), where it is more stable by 1.08 eV with respect to the interface position (OTI ). These structural and energetics analyses thus suggest that O vacancies in the 1-ML ceria film

Table 1. Equilibrium Lattice Parameter (a) of Unsupported Stoichiometric and Defective Ceria Layers, and Corresponding Difference with Respect to the Bulk Value of CeO2 (a0) bulk CeO2 1 ML CeO2 1 ML CeO(1.75) 1 ML Ce2O3 2 ML CeO2

a(CeO2) [Å]

(a−a0)/a0 [%]

3.91 3.65 3.75 3.80 3.76

6.7 4.1 2.8 3.7

one hand, this contraction in the lattice parameter of thin ceria films quickly decreases as a function of the number of layers: the difference is reduced to 3.7% already for a 2 ML ceria slab. On the other hand, the strain is considerably reduced by the presence of O defects and reduced Ce ions as shown by the smaller contractions displayed by unsupported defective CeO(1.75) (4.1%) and Ce2O3 (2.8%) monolayers. Our calculations therefore provide evidence of the lateral contraction of thin stoichiometric ceria films with respect to bulk CeO2 and demonstrate that oxygen vacancies together with reduced Ce3+ ions in the slab allow for reducing the lattice mismatch between 1-ML ceria films and the copper crystal. Thicker stoichiometric 2 ML CeO2 slabs display a much smaller contraction than 1 ML CeO2. Ceria films thicker

Figure 3. (a) Top view of the ceria/Cu interface. Each atomic layer is indicated by a different color, while the labels refer to the symmetryinequivalent atomic sites. (b,c) Side view of the supercells used to model the stoichiometric and defective 1-ML and 2-ML ceria/Cu. 6680



Article

Figure 4. Spin density (left panel), projected density of states (pDOS, middle panel), and simulated STM images (right panel) of (a) 1 ML CeO2/ Cu(111), (b) 1 ML CeO1.75/Cu(111), and (c) 2 ML CeO2/Cu(111). In the pDOS analysis, red, blue, and thin-black lines indicate the p, d, and f contributions, respectively. For obtaining an STM image of 1-ML ceria comparable to experiment, interface oxygen vacancies must be considered (compare b, right panel and Figure 2b).

Table 3. Difference in the Löwdin Charge of the Cu Atoms in the I and II Layers (CuI, and CuII, cf. Figure 3) with Respect to the Bulk-like Value

reduce the lateral strain and that they are located at the metal/ oxide interface. As a consequence of the strong preferential interface segregation, O vacancies are predicted to be subsurface in the 1-ML ceria films. Electronic and Atomistic Structures. The analysis of the electronic structure allows one to explain the fundamental difference in the O-vacancy segregation at the interface between 1- and 2-ML ceria films reported above. The interface between the stoichiometric thin ceria film and a copper surface displays a charge transfer from the metal to the oxide, yielding electron localization and reduction of all the interfacial Ce ions to Ce3+. This interfacial effect, observed also in the case of thicker ceria films,42 is clearly displayed by the spin polarization of the cerium atoms (see spin density in Figure 4) and by the charge population analysis of the Cu atoms reported in Table 3. The latter shows the difference of the Löwdin charge of the Cu atoms with respect to the bulk value. For the case of the stoichiometric 1-ML CeO2/Cu system, the data clearly support the charge depletion (oxidation) of the Cu atoms in the first

CuI(OTI ) [×1] CuI(CeTII) [×1] CuI(OTIII) [×1] CuI(OBI ) [×6] CuII [×9]

1 ML

1 ML

2 ML

CeO2/Cu

CeO1.75/Cu

CeO2/Cu

−0.3 −0.1 −0.2 −0.3 −0.1

0.0 0.0 −0.2 −0.2 −0.1

−0.3 −0.1 −0.2 −0.2 −0.1

interfacial layer (CuI) with respect to the deeper and bulk-like metal atoms (CuII and CuIII). The Cu atoms displaying the largest variations are the closest to the interfacial O atoms, namely, the Cu atom with an O atom on top labeled as CuI(OTI ), and the six connected to a bridging O labeled as CuI(OBI ). The Cu atoms further from and in hollow position with respect to the interfacial O atoms are also reduced but by a minor extent. 6681



Article

this case, the surface of the thin film has an effective (2 × 2) structural periodicity. However, for the case of thicker films, the structural effects of the interface do not propagate up to the surface atoms. The stoichiometric thicker (2-ML) ceria film displays the same interface structure of the 1-ML case, but the structural differentiation of one atom in each atomic layer is negligible for the outermost O−Ce−O trilayer, and the surface of the supported film has an effective (1 × 1) periodicity, both in terms of local atomistic and electronic structures. Simulated STM Images. This different periodicity of 1-ML and 2-ML supported ceria films is reflected in the simulated STM images reported in Figure 4. In order to allow for a direct comparison with the experiment (measured at positive bias), we report in the figure the STM images calculated for the unoccupied states. In these conditions, the bright spots are associated with the Ce ions, as clearly shown by the pDOS reported in the middle panels of Figure 4 for the Ce and O atoms. In the energy window relevant for the comparison with experiment, the contribution of the oxygen states is negligible. A representative simulated STM image for the 1 ML CeO2/ Cu system is displayed in Figure 4a, and presents important differences with the experimental micrograph of Figure 2a,b. In particular, the simulated STM image displays a large darker area around the symmetry-inequivalent CeTII site, which is not observed in the experimental data. A better agreement is not obtained by calculating the STM image at different biases or by plotting the integrated charge on planes at different distances from the surface. On the other hand, for a wide range of biases and sample-plane distances (0−3 eV and 0.35−2.0 Å, respectively), the simulated STM image of the defective 1 ML CeO1.75/Cu system presents a much better agreement with the experiment (Figure 4b). The calculated pattern displays a (2 × 2) periodicity of four spots (one of which is brighter than the others) and of two darker areas (one of which is darker than the other). By superimposing the atomic positions on the simulated STM image, it is possible to associate the brightest spot and darkest areas with the CeTII and OT atoms, respectively. This similarity between the simulated STM image for the defective 1 ML CeO1.75/Cu system and experimental images further supports the presence of vacancies at the interface between the 1-ML ceria film and copper. Quite differently, the simulated STM image of the 2 ML CeO2/Cu system (Figure 4c) displays a 1 × 1 pattern of brighter and darker spots, which is in very good agreement with the images of the 2-ML ceria observed in STM. This therefore confirms that the 2-ML ceria film is stoichiometric. Paralell versus Antiparalell Orientation of the 1-ML Film. In the following, we investigate the possible origins of the two types of islands experimentally observed. We have so far calculated and described the most stable structure of the 1 ML CeO1.75/Cu with the oxygen vacancy at the OI1 site where the [112] directions of ceria and copper are antiparallel (“A” in Figure 5). There is another configuration of this 1 ML CeO1.75/ Cu interface, in which the two [112] directions are parallel (“P” in Figure 5). Differences between the two orientations arise when considering the subsurface copper atoms (brown circles). Calculated atomic relaxation in the P and A orientations are graphically represented by colored arrows in Figure 5 with magnitude in ångstroms indicated in the insets. Interface P induces larger horizontal relaxations than interface A by 0.13 Å for the interface oxygen atoms and by 0.02 Å for the cerium atoms. This can be explained by the different interactions of

Creating an oxygen vacancy in the bulk or at surfaces of ceria is usually accompanied by the localization of two excess electrons at two Ce4+ ions, leading to their reduction to Ce3+.39,49 This is not possible in the case of vacancy formation at the 1 ML CeO2/Cu(111) interface, since all the cerium atoms in the film are already in the 3+ state. The excess charge originating from the presence of the O vacancy turns out to be localized at the metallic Cu/oxide interface. This is supported by the calculated charge analysis for the defective 1 ML CeO(1.75) film (Table 3), which indicates an excess of charge at the interfacial Cu atoms with respect to the stoichiometric case. In particular, the Cu atom displaying the largerst variation is the one directly below the O vacancy, CuI(OTI ), while the charge on the other Cu atoms is only marginally affected by the presence of the vacancy. A further support of this effect is given by the calculated density of electronic states of the 1 ML CeO2/Cu and CeO1.75/Cu systems (Figure 4), displaying the same degree of reduction of the four Ce ions, but a different position of the Fermi level crossing the Cu s-band. A thicker 2-ML ceria film displays behavior opposite to that of the 1-ML case with respect to the position of O vacancies. Our calculations predict that this difference has an electronic origin. The spin density (Figure 4c, left panel) shows that the effect of the interface on the reduction of the cerium atoms extends up to the first monolayer, while the second monolayer is formed exclusively by Ce4+ atoms. Reduced thicker ceria films can therefore easily localize the excess electrons on the Ce4+ sites of its second monolayer. The difference between supported 1- and 2-ML ceria films is also apparent in the local atomistic structures of the surface atoms, whose electronic states are probed by the STM tip. The local structure of the stoichiometric 1-ML ceria film is schematically shown in Figure 3a. The OTI site has a different local environment than the other three interfacial OBI atoms, which are at the bridge position between two copper atoms. The same applies also to the other atomic layers in ceria slab causing one of the four atomic sites, CeTII and OTIII, to be symmetry inequivalent from the other three, CeBII and OBIII, respectively. The equilibrium bondlengths at the interface are six 2.05 Å Cu−OBI bonds and one 1.91 Å Cu−OTI bond. Due to the presence and periodicity of the underlying copper atoms, the one symmetry-inequivalent O atom (OTI ) relaxes away from the interface by 0.35 Å. This displaces outward the three neighboring cerium atoms CeBII by 0.11 Å with respect to the other cerium atom in the supercell (CeTII). The coherent and stoichiometric 1 ML CeO2/Cu interface therefore induces structural changes that differentiate one of the four O and Ce atoms in each atomic layer up to the surface plane, thus resulting in an effective (2 × 2) periodicity at the surface of the thin film. The presence of the O vacancy at the OTI interfacial site of the 1 ML CeO(1.75)/Cu(111) system has a qualitatively similar but more pronounced effect on the outermost surface atoms. The lowest-energy site for vacancy formation is in fact the most symmetrical OTI site. The O vacancy induces a shift of the three neighboring cerium atoms away (both in the vertical (by 0.31 Å) and horizontal (by 0.12 Å) directions). The other CeTII atom is not only geometrically distinct from the other CeBII three, but is the only one that is 6-fold coordinated. At the interface, this atom repulses the neighboring interface oxygen atoms by 0.23 Å, leading to an increase of the six interface Cu−OBI bonds to 2.09 Å. At the surface, one OTIII atom protrudes by 0.43 Å with respect to the other three surface OBIII oxygen atoms. Therefore, also in 6682



■

Article

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Present Address ∥

Lehrstuhl für Theoretische Chemie, Ruhr Universität Bochum, 44780, Bochum, Germany. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the Ministry of Education of the Czech Republic (LH11017, ME08056) and by the Grant Agency of the Czech Republic (GACR 204/11/1183, GACR 202/09/H041). L.S, O.S., and F.D. acknowledge the support from the Grant Agency of the Charles University (GAUK 92209, GAUK 103410). Computer resources were allocated by the Coupled Cluster computing system of Faculty of Mathematics and Physics, Charles University in Prague. We would like to thank Pavel Sobotk, Tomás ̌ Skála, and Viktor Johánek for many helpful discussions.

Figure 5. Paralell (P) and antiparalell (A) orientations of the 1-ML CeO1.75 with respect to Cu(111) substrate. Difference between P and A is due to Cu atoms in the subsurface monolayer (CuII, brown circles). Arrows indicate the horizontal components of the calculated atomic relaxation.

interface oxygen atoms with the copper atoms in the subsurface layer, considering that in the A supercell, the oxygen atoms are forced to move horizontally closer to the subsurface copper atoms due to the presence of the vacancy. As a result, the atoms in all atomic layers in the ceria layer relax vertically to a higher level in the A supercell than in the P case. In addition, the A orientation turns out to be energetically more favorable than the P orientation by 0.12 eV. This small energy difference is consistent with the comparable frequency of occurrence of two types of islands in the experimental data. Establishing a correspondence between calculated P and A orientations and M and W islands in experiment is difficult, because the supercell size used in our calculations is too small to capture the long-range moiré (which develops over more than 14 unit cells) or waggon-wheel structures. However, on the basis of the calculated energy preference of the A orientation as well as its smaller relaxation magnitudes and larger vertical accommodation, we tentatively propose that the A orientation corresponds to the locally better-accommodated M islands.

■

■

REFERENCES

(1) Freund, H.-J. Surf. Sci. 2002, 500, 271. (2) Freund, H.-J. Surf. Sci. 2007, 601, 1438. (3) Netzer, F. P.; Allegretti, F.; Surnev, S. J. Vac. Sci. Technol. B 2010, 28, 1. (4) Weiss, W.; Ritter, M. Phys. Rev. B 1999, 59, 5201. (5) König, T.; Simon, G. H.; Rust, H.-P.; Heyde, M. J. Phys. Chem. C 2009, 113, 11301. (6) Kresse, G.; Schmid, M.; Napetschnig, E.; Shishkin, M.; Köhler, L.; Varga, P. Science 2005, 308, 1441. (7) Netzer, F. P. Surf. Sci. 2010, 604, 485. (8) Sterrer, M.; Risse, T.; Pozzoni, U. M.; Giordano, L.; Heyde, M.; Rust, H.-P.; Pacchioni, G.; Freund, H.-J. Phys. Rev. Lett. 2007, 98, 096107. (9) Sock, M.; Surnev, S.; Ramsey, M. G.; Netzer, F. P. Top. Catal. 2001, 14, 15. (10) Rodríguez, J. A.; Hrbek, J. Surf. Sci. 2010, 604, 241. (11) Trovarelli, A. Catalysis by Ceria and Related Materials; Imperial College Press: London, 2002. (12) Fu, Q.; Saltsburg, H.; Flytzani-Stephanopoulos, M. Science 2003, 301, 935. (13) Flytzani-Stephanopoulos, M.; Sakbodin, M.; Wang, Z. Science 2006, 312, 1508. (14) Mullins, D. R.; Radulovic, P. V.; Overbury, S. H. Surf. Sci. 1999, 429, 186. (15) Lu, D.-L.; Gao, H.-J.; Shaikhutdinov, S.; Freund, H.-J. Surf. Sci. 2006, 600, 5004. (16) Šutara, F.; Cabala, M.; Sedlácě k, L.; Skála, T.; Škoda, M.; Matolı ́n, V.; Prince, K. C.; Cháb, V. Thin Sol. Films 2008, 516, 6120. (17) Dvořaḱ , F.; Stetsovych, O.; Steger, M.; Cherradi, E.; Matolı ́nová, I.; Tsud, N.; Škoda, M.; Skála, T.; Mysliveček, J.; Matolı ́n, V. J. Phys. Chem. C 2011, 115, 7496. (18) Mullins, D. R.; Robbins, M. D.; Zhou, J. Surf. Sci. 2006, 600, 1547. (19) Staudt, T.; Lykhach, Y.; Tsud, N.; Skála, T.; Prince, K. C.; Matolín, V.; Libuda, J. J. Catal. 2010, 275, 181. (20) Zhou, J.; Mullins, D. R. J. Phys. Chem. B 2006, 110, 15994. (21) Baron, M.; Bondarchuk, O.; Stacchiola, D.; Shaikhutdinov, S.; Freund, H.-J. J. Phys. Chem. C 2009, 113, 6042. (22) Lykhach, Y.; Staudt, T.; Lorenz, M. P. A; Streber, R.; Bayer, A.; Steinrück, H.-P.; Libuda, J. ChemPhysChem 2010, 11, 1496. (23) Vayssilov, G. N.; Lykhach, Y.; Migani, A.; Staudt, T.; Petrova, G. P.; Tsud, N.; Skála, T.; Bruix, A.; Illas, F.; Prince, K. C.; Matolín, V.; Neyman, K. M.; Libuda, J. Nat. Mater. 2011, 10, 310.

CONCLUSION

Using STM and ab-initio calulations, we have investigated physical and chemical properties of the 1st monolayer of cerium oxide on Cu(111). Nanometer-scale coincidence patterns on 1ML ceria reveal a compressive strain with respect to bulk ceria that is identified as a size effect in ultrathin ceria layers by the calculations. The 2 × 2 surface reconstruction on the 1-ML ceria indicates the presence of oxygen vacancies in the ceria− copper interface, which is confirmed by evaluating vacancy energies and simulating STM images in the calculation. The extraordinary behavior of oxygen vacancies is a consequence of a proximity of the metal substrate when all Ce atoms in 1-ML ceria are Ce3+. Electrons left after creation of oxygen vacancies must then localize in the Cu substrate. Both Ce3+ and oxygen vacancies relieve the intrinsic compressive strain in ultrathin ceria. Metal−oxide interactions determining the properties of ultrathin ceria reach as far as 2 ML in the Cu substrate but not further than 1 ML in the oxide. Indeed, already 2-ML ceria shows a bulklike termination in experiment and a standard behavior of oxygen vacancies in the calculation. The distinct physicochemical properties of the first ceria monolayer likely play a significant role in the enhanced reactivity of inverse model ceria/Cu catalysts where 1-ML ceria represents the most extended surface phase. 6683



Article

(24) Rodriguez, J. A.; Ma, S.; Liu, P.; Hrbek, J.; Evans, J.; Pérez, M. Science 2007, 318, 1757. (25) Rodriguez, J. A.; Graciani, J.; Evans, J.; Park, J. B.; Yang, F.; Stacchiola, D.; Senanayake, S. D.; Ma, S.; Pérez, M.; Liu, P.; Sanz, J. F.; Hrbek, J. Angew. Chem., Int. Ed. 2009, 48, 8047. (26) Eck, S.; Castellarin-Cudia, C.; Surnev, S.; Prince, K. C.; Ramsey, M. G.; Netzer, F. P. Surf. Sci. 2003, 536, 166. (27) Wrobel, R.; Suchorski, Y.; Becker, S.; Weiss, H. Surf. Sci. 2008, 602, 436. (28) Yang, F.; Graciani, J.; Evans, J.; Liu, P.; Hrbek, J.; Sanz, J. F.; Rodriguez, J. A. J. Am. Chem. Soc. 2011, 133, 3444. (29) Hornés, A.; Hungría, A. B.; Bera, P.; Cámara, A. L.; FernándezGarca, M.; Martínez-Arias, A.; Barrio, L.; Estrella, M.; Zhou, G.; Fonseca, J. J.; Hanson, J. C.; Rodriguez, J. A. J. Am. Chem. Soc. 2010, 132, 34. (30) Castellarin-Cudia, C.; Surnev, S.; Schneider, G.; Podlucky, R.; Ramsey, M. G.; Netzer, F. P. Surf. Sci. 2004, 554, L120. (31) Eck, S.; Castellarin-Cudia, C.; Surnev, S.; Ramsey, M. G.; Netzer, F. P. Surf. Sci. 2002, 520, 173. (32) Grinter, D. C.; Ithnin, R.; Pang, C. L.; Thornton, G. J. Phys. Chem. C 2010, 114, 17036. (33) http://gwyddion.net/, accessed on Mar 1, 2012. (34) Cococcioni, M.; de Gironcoli, S. Phys. Rev. B 2005, 71, 035105. (35) Perdew, J. P.; Burke, K.; Ernzherof, M. Phys. Rev. Lett. 1996, 77, 3865. (36) Giannozzi, P.; Baroni, S.; Bonini, N.; Calandra, M.; Car, R.; Cavazzoni, C.; Ceresoli, D.; Chiarotti, G. L.; Cococcioni, M.; Dabo, I.; et al. J. Phys.: Condens. Matter 2009, 21, 395502. (37) Fabris, S.; de Gironcoli, S.; Baroni, S.; Vicario, G.; Balducci, G. Phys. Rev. B 2005, 71, 041102. (38) Farnesi Camellone, M.; Fabris, S. J. Am. Chem. Soc. 2009, 131, 10473. (39) Fabris, S.; Vicario, G.; Balducci, G.; de Gironcoli, S.; Baroni, S. J. Phys. Chem. B 2005, 109, 22860−22867. (40) Huang, M.; Fabris, S. Phys. Rev. B 2007, 75, 081404. (41) Huang, M.; Fabris, S. J. Phys. Chem. C 2008, 112, 8643−8648. (42) Szabová, L.; Farnesi Camellone, M.; Huang, M.; Matolín, V.; Fabris, S. J. Chem. Phys. 2010, 133, 234705. (43) Colussi, S.; Gayen, A.; Farnesi Camellone, M.; Boaro, M.; Llorca, J.; Fabris, S.; Trovarelli, A. Angew. Chem. 2009, 48, 8481−8484. (44) Wang, X.; Shen, M.; Wang, J.; Fabris, S. J. Phys. Chem. C 2010, 114, 10221−10228. (45) Nolan, M.; Grigoleit, S.; Sayle, D. C.; Parker, S. C.; Watson, G. W. Surf. Sci. 2005, 576, 217−229. (46) Loschen, C.; Carrasco, J.; Neyman, K. M.; Illas, F. Phys. Rev. B 2007, 75, 035115. (47) Nolan, M.; Watson, G. J. Phys. Chem. B 2006, 110, 16600. (48) Zhang, C.; Michaelides, A.; King, D.; Jenkins, S. J. J. Chem. Phys. 2008, 129, 194708. (49) Ganduglia-Pirovano, M. V.; Da Silva, J. L. F.; Sauer, J. Phys. Rev. Lett. 2009, 102, 026101. (50) Szabová, L. M.Sc. Thesis, Faculty of Mathematics and Physics, Charles University, Prague, 2009 (51) Torbrügge, S.; Cranney, M.; Reichling, M. Appl. Phys. Lett. 2008, 93, 073112. (52) Michely, T.; Hohage, M.; Bott, M.; Comsa, G. Phys. Rev. Lett. 1993, 70, 3943. (53) Voigtländer, B.; Kästner, M.; Šmilauer, P. Phys. Rev. Lett. 1998, 81, 858. (54) Galloway, H. C.; Sautet, P.; Salmeron, M. Phys. Rev. B 1996, 54, R11145. (55) Mašek, K.; Beran, J.; Matolı ́n, V. Submitted to Appl. Surf. Sci. (56) Galloway, H. C.; Benítez, J. J.; Salmeron, M. Surf. Sci. 1993, 298, 127. (57) Sedona, F.; Agnoli, S.; Granozzi, G. J. Phys. Chem. B 2006, 110, 15359. (58) Torbrügge, S.; Reichling, M.; Ishiyama, A.; Morita, S.; Custance, O. Phys. Rev. Lett. 2007, 99, 056101.

6684


Distinct Physicochemical Properties of the First Ceria Monolayer on

Recommend Documents