Oxygen Activation and Dissociation on Transition ... - ACS Publications

Nov 20, 2015 - and John Kilner*,†,∥. †. International Institute for Carbon-Neutral Energy Research (WPI-I2CNER), Kyushu University, 744 Motooka,...
0 downloads 0 Views 4MB Size
Download PDF
Article pubs.acs.org/cm

Oxygen Activation and Dissociation on Transition Metal Free Perovskite Surfaces Aleksandar Staykov,*,† Helena Téllez,† Taner Akbay,‡ John Druce,† Tatsumi Ishihara,†,‡,§ and John Kilner*,†,∥ †

International Institute for Carbon-Neutral Energy Research (WPI-I2CNER), Kyushu University, 744 Motooka, Nishi-ku, Fukuoka 819-0395, Japan ‡ Advanced Research Centre for Electric Energy Storage, Kyushu University, 744 Motooka, Nishi-ku, Fukuoka 819-0395, Japan § Department of Applied Chemistry, Faculty of Engineering, Kyushu University, 744 Motooka, Nishi-ku, Fukuoka 819-0395, Japan ∥ Department of Materials, Imperial College London, London SW7 2AZ, United Kingdom S Supporting Information *

ABSTRACT: Density functional theory and low energy ion scattering spectroscopy were applied to study the mechanism of oxygen dissociation on the SrO-terminated surfaces of strontium titanate (SrTiO3) and irondoped strontium titanate (SrTi1−xFexO3−δ). Our study reveals that while O2 dissociation is not favored on the SrO-terminated perovskite surface, oxygen vacancies can act as active sites and catalyze the O−O bond cleavage. Electron transfer from lattice oxygen atoms to the O2 molecule, mediated by the subsurface transition metal cations, plays an important role in the resulting formation of surface superoxo species. The O2 molecule dissociates to produce oxygen ions, which are incorporated into the perovskite lattice, and highly active oxygen radicals on the perovskite surface, which further recombine to O2 molecules. Our focus on the SrO-terminated surface, rather than the TiO2 layer, which is presumed to be more catalytically active, was driven by experimental observation using low energy ion scattering spectroscopy, which reveals that the surface of SrTiO3 after high temperature heat treatment is SrO-terminated, and hence this is the surface that is technologically relevant for devices such as solid oxide fuel cells (SOFCs). Our study demonstrates that although the more active BO2-perovskite layer is not exposed at the gas−solid interface, the SrO-terminated surfaces also actively participate in oxygen exchange reaction.

■

INTRODUCTION Perovskite oxides are inorganic compounds of two or more cations sharing the general chemical formula ABO3 where A is a large cation, usually an alkaline earth or lanthanide metal, and the B cation is a transition metal.1 The prototypical cubic perovskite structure may be viewed as a stack of layers of AO alternating with BO2 layers in the (001) direction. The different ionic radii of the A and B elements determine the lattice structure,2 with distortions from the ideal cubic perovskite lattice occurring depending on the ionic size ratio between the A and B cations, as described by the Goldschmidt3 tolerance factor.4 Typical technologically relevant examples from this family include SrTiO3, BaTiO3, and LaCoO3. This tolerance to accommodate different cations of various ionic radii is one of the aspects that make materials based on the perovskite structure attractive, as it is thus possible to dissolve a wide range of substituent cations into the lattice. Partial substitution of the A and B cations can be used to tailor the electronic and point defect structures of the material for many applications. Ceramic perovskite oxides enjoy a rich variety of applications including electrodes (and electrolytes)5 for solid oxide fuel cells (SOFCs) and electrolyzers (SOECs), in batteries, permeation membranes with high selectivity for chemical conversion © 2015 American Chemical Society

processes, catalysts, electrochemical gas sensors, and materials for memristive switches and in photovoltaic devices. Of particular relevance at present is their important implementation as electrodes for SOFCs and SOECs, due to the mixed oxygen ionic and electronic conduction shown by certain perovskite-based materials at elevated temperatures.1,6,7 At SOFC operating conditions, i.e., ∼700 °C, the mobility of O2− in the perovskite lattice increases and has been investigated as a function of temperature, applied potential bias, or variable oxygen gas pressure.8 Much research on the perovskite materials has addressed the important issues of their oxygen transport properties, oxygen reduction reactions, and water splitting reactions.8−11 Additionally, other chemical reactions were considered such as H2S activation for fuel cell research.12 Various studies have compared the oxygen reduction kinetics and oxygen mobility between perovskites with different (host) chemical compositions, as well as studying the effect of substitution with various substituents (“doping”), lattice strain, and effect of point and extended defects13,14 on the materials properties.15−18 It has Received: August 21, 2015 Revised: November 8, 2015 Published: November 20, 2015 8273

DOI: 10.1021/acs.chemmater.5b03263 Chem. Mater. 2015, 27, 8273−8281

Article

Chemistry of Materials

recently, it has been shown that at SOFC operating conditions the surfaces of perovskite and perovskite-related oxides are AO dominated.31−33 Low energy ion scattering spectroscopy (LEIS)34 was employed to investigate the chemical composition of the outermost surface of various perovskite-related materials at room temperature and after annealing at elevated temperatures representative of those experienced during device processing and operation. Furthermore, it was shown that in the case of A-site doping, the cation with larger ionic radius segregates to the surface.22 In all experimentally investigated cases, no B-site cations were detected at the outer surface layer.31 This discovery represents a significant challenge for computational science, which has to elucidate the chemical mechanisms that take place on that surface termination.35 The interaction of an oxygen molecule with perovskite-type surfaces was previously investigated with first principles methods in the work of Zhou et al.36 They have demonstrated the mechanisms of oxygen adsorption on the (100) surface of La2NiO4. However, all investigated surfaces in their work included Ni atoms. They have shown that the chemisorbed O2 molecule is further activated to superoxo or peroxo state. In the investigated adsorption mechanism, the surface Ni atoms played an important role. They have furthermore elucidated that positively charged oxygen vacancies would significantly increase the oxygen−oxygen distance and might further lead to oxygen dissociation. In this work, our aim is to employ first principles methods to investigate the possibility for oxygen dissociation on AOterminated perovskite surfaces. Our motivation is the experimental data showing that after exposure to temperatures and atmospheres similar to SOFC operating conditions the perovskite surfaces are predominately characterized with AO termination. We have chosen to perform the calculations for the SrTiO3 (STO), owing to the relative computational simplicity of this perovskite material. The bulk defect chemistry of STO is well-understood;37 however, the AO-terminated surface chemical reactivity is yet to be investigated. The tendency for the larger aliovalent dopant to segregate toward the surface means that the surfaces of the commonly used La1−xSrxCoO3 perovskite are also SrO-terminated, so we expect the insights provided by this work to be relevant to these related materials families. Furthermore, in order to investigate the effect of surface oxygen vacancies we have studied Fe-doped SrTiO3, where depending on the oxidation state of Fe, oxygen vacancies could be generated.

been shown that precise control of the substituents on either the A-site cation or B-site cation can be used to tailor the concentration of oxygen vacancies in the perovskite material.19,20 Substituents in the perovskite lattice can have distinctly different effects, arising from the different ionic radii, e.g., steric effects, or differences in valence and electronic structure. Steric effects manifest themselves in variations in the A−O−A angles or substituents segregation to the surface. An example is Srdoped LaCoO3 where a significant amount of Sr segregates to the surface and degrades the material’s performance as an SOFC electrode.21−23 Doping either on the A- or B-sites with elements with different valence states or charges compared to the main perovskite lattice changes the number of oxygen anions necessary to neutralize the charge of the cations. This introduces oxygen vacancies in the perovskite lattice, which are crucial for the oxygen transport properties required for application in SOFC and SOEC. The transport of oxygen through the lattice can become significant at elevated temperatures and can be driven by a chemical gradient or by an electrical bias. Various studies deal with the oxygen transport properties of perovskite-related materials, showing that oxygen can be transported through the lattice or grain boundaries, or as interstitial atoms depending upon the particular type of perovskite.24−26 The oxygen transport through mixed oxide materials (ABOn) significantly depends on the lattice structure. A wide range of oxygen ion diffusivities have been reported for transport through perovskite lattices and perovskite-related oxides, such as the double perovskites and Ruddlesden−Popper phases.27 Double perovskites and Ruddlesden−Popper oxides show also anisotropy in the transport properties which is related to the crystal structure transport direction.28,29 The layered and predominantly ionic structure of perovskites makes them challenging materials for calculations of electronic properties due to an artificial dipole moment in the slabs, charge compensation, and the tendency of these materials toward surface reconstruction. Especially challenging are calculations that include variable numbers of oxygen atoms in the perovskite lattice, i.e., oxygen vacancy formation and annihilation. In reality, oxygen vacancies are rarely observed in unsubstituted ABO3 perovskite lattices, and their presence in any significant concentration is the result of substitution with cations with different charge. Substitution with elements with variable valence state such as iron that would allow for variable concentrations of oxygen ions and vacancies in the perovskite lattice (i.e., incorpation and excorporation of oxygen in response to changes in oxygen activity of the surrounding environment) is of great importance for both experimental and theoretical research. Besides doping, recent studies show that perovskite materials after extreme plastic deformation under high pressure torsion are also characterized by a high concentration of oxygen vacancies.30 An important issue for theoretical studies of these perovskite materials is the modeling of their surfaces. While at low temperatures different surfaces are thermodynamically stable, most of the reactions in which we are interested for applications such as SOFCs occur at elevated temperatures where the surface composition can be significantly different.31 Until very recently the experimental information for the perovskite outmost surfaces layer composition at operating temperatures of SOFC and steam electrolyzers was unknown. The majority of theoretical works have investigated BO2-terminated and mixed (AO and B2O) perovskite surfaces owing to the assumed high catalytic activity of the transition metal cation. Only

■

COMPUTATIONAL METHODS

The calculations in this study were performed with the plane wave density functional theory (DFT) implemented in the Vienna ab initio simulation package (VASP).38−41 The Perdew−Burke−Ernzerhof exchange-correlation functional (PBE) was employed using projector augmented wave pseudopotentials. Throughout this study, we have used the graphical visualization package VESTA to analyze and visualize the computed DFT electron density distribution.42 The bulk SrTiO3 crystal lattice was optimized with 400 eV cutoff energy and 4 × 4 × 4 k-points sampling. Spin polarized calculations were performed throughout the study for all investigated systems. The surface model was constructed after cleavage of the bulk SrTiO3 at the crystal plane with (001) Miller indexes and vacuum slab of 12 Å. The thickness of the vacuum layer was sufficient to avoid the physical interaction between neighboring supercells. The top surfaces are SrO-terminated and the bottom surface are TiO2-terminated. The slabs contain eight layers. The atoms in the bottom four layers are with fixed coordinates, and the atoms in the top four layers are fully relaxed. The AO layers 8274

DOI: 10.1021/acs.chemmater.5b03263 Chem. Mater. 2015, 27, 8273−8281

Article

Chemistry of Materials

Figure 1. LEIS spectra and depth profile for a SrTiO3 single crystal after annealing in 1 atm O2 at 1000 °C: (A) 6 keV Ne+ LEIS surface spectra obtained for the outer surface (black solid line) and the bulk composition (blue dashed line); (B) depth profile normalized to the stoichiometry (red spheres denote Sr, while white squares denote Ti). are the top layer and third, fifth, and seventh layers. The BO2 layers are the second, fourth, sixth, and eighth layers. Three surface models are considered: 1 × 1 surface unit cell of SrTiO3; 1 × 1 surface unit cell of SrTi0.75Fe0.25O3‑δ, where the iron atoms are located in the fourth and sixth layers beneath the surface and the surface is characterized with a single oxygen vacancy; and 2 × 1 surface unit cell of SrTi0.75Fe0.25O3‑δ, where the iron atoms are located in the fourth and sixth layers beneath the surface and the surface is characterized with two oxygen vacancies. The position of the iron atoms is selected far from the surface in order to have minimal effect on the surface reactions. One of the iron atoms is within the layers with fixed coordinates, and the second iron atom is in the layers with fully relaxed coordinates. The 1 × 1 surfaces are modeled with 4 × 4 × 1 k-point sampling and the 2 × 1 surfaces are modeled by 2 × 4 × 1 k-point sampling. Bader population analysis was performed to determine the atomic charges.43 Activation barriers for various reaction mechanisms were obtained using the nudged elastic band method (NEB) combined with the climbing nudged elastic band method (cNEB). In the process of NEB calculations five images were used between the starting and ending geometries.

■

the etching rate of the Ar+ beam using laser interferometry (as reference sputter rate, the sputter rate obtained on other perovskite epitaxial thin films was used as an approximate value). Due to the extreme surface specificity of the technique, the surface required in situ cleaning by oxidation at room temperature with neutral atomic oxygen in order to remove any adventitious contaminants from the previous exposure to air.

■

RESULTS AND DISCUSSION Our LEIS analysis shows that after the annealing treatment, the investigated perovskite surface is exclusively AO-terminated, within the limit of detection of the technique, with the presence of a single surface peak at a scattering energy of ∼2570 eV (Figure 1 A, solid line), corresponding to Sr. As observed in the Figure 1A, using 6 keV Ne+ scattering, there are no transition metal cations detected at the outer surface in the SrTiO3 single crystal after the annealing at 1000 °C. These results are in agreement with previous work on perovskite and perovskiterelated materials that showed a predominant AO-terminated surface after oxidation at high temperatures.31,44 In comparison, the cation composition of the single crystal after low energy sputtering to reach the bulk material (Figure 1 A, dashed line) showed the presence of both Ti and Sr cations. Figure 1B shows the depth profile of the SrTiO3 sample measured with LEIS and normalized to the bulk stoichiometry. The depth profile verifies that, for the outermost layer, the composition is SrO, i.e., the surface is SrO-terminated, while the subsurface layer contains Ti cations. The bulk ratio is reached approximately 5 nm below the surface. The aim of the remainder of this study is to provide an understanding of the mechanisms for oxygen dissociation on such a SrO-terminated surface and the subsequent oxygen incorporation into the perovskite lattice. To achieve this, we have applied the DFT-based theoretical approach detailed previously. The bulk SrTiO3 crystal lattice was fully relaxed with resulting lattice parameters 5.516, 5.506, and 7.796 Å. The bulk SrTiO3 crystal was cleaved at the (001) plane with the SrO-terminated top surface. In the surface calculations, eight alternating layers were considered, four SrO layers and four TiO2 layers. The bottom surface is TiO2-terminated. The charge of each layer is neutral, and the slabs can be considered as type one Tasker surfaces with zero dipole moment.45 However, recent theoretical studies have shown weak polarization between the SrTiO3 layers, although it does not require

EXPERIMENTAL METHODS

Low energy ion scattering provides information about the chemical composition of the outer layer of the materials by probing the surface using a noble gas ion beam (typically He+, Ne+, or Ar+, accelerated to 1−8 keV). The kinetic energy distribution of the backscattered primary ions is then related to the elemental composition of the surface according to the kinematics of the scattering event, which depends on the masses of the ions involved in the collision and the initial and final energies of the primary species. The depth resolution of the technique is specifically limited to the first atomic layer at the surface. This atomic layer resolution is given by the high neutralization probability of the noble gas primary ions when they penetrate into the inner atomic layers of the material, since those ions that are neutralized cannot be detected by the analyzer. In this work, a (100) SrTiO3 single crystal surface was analyzed after being ultrasonically cleaned in aqua regia solution (1 mL of 70% HNO3 and 3 mL of 70% HCl) at room temperature for 10 min and then annealed for 2 h at 1000 °C in 1 atm of O2. The analysis was performed using a Qtac100 LEIS spectrometer (Ion-ToF GmbH, Münster, Germany) fitted with a high brightness ion source to provide the analytical beam and a second sputter gun for depth profiling in dual mode. In this case, the samples were analyzed using 6 keV Ne+ impinging normal to the sample surface. The cation distribution from the outer surface to the bulk was also analyzed by depth profiling using a 500 eV Ar+ beam impinging at 59° to sputter the material between the analysis cycles with 6 keV Ne+ ion beam. In order to avoid crater-wall effects, the analysis was performed in a 1 × 1 mm2 crater, while the sputtering area was significantly larger (1.3 × 1.3 mm2). The estimated depth of the profiles was obtained by calibrating 8275

DOI: 10.1021/acs.chemmater.5b03263 Chem. Mater. 2015, 27, 8273−8281

Article

Chemistry of Materials dipole correction similar to the strong polar surfaces of other polar perovskites.46,47 Figure 2 shows the optimized geometry

Figure 3. O2 adsorbed on SrTiO3 (001) SrO-terminated surface and adsorption sites for dissociated O atoms: (A) O2 adsorbed on topbridge-top site; (B) O2 adsorbed on a slip position over a Sr atom; (C) dissociated oxygen on the SrO surface. Sr is denoted by green, Ti is denoted by blue, and O is denoted by red. Distances are given in angstroms, and energies are given in electronvolts.

Figure 2. SrTiO3 crystal lattice and (001) SrO-terminated surface: (A) bulk crystal lattice; (B) SrO-terminated surface. Sr is denoted by green, Ti is denoted by blue, and O is denoted with red.

of the SrTiO3 bulk crystal and the SrO-terminated slab model. In the bulk SrTiO3 crystal lattice the Ti−O−Ti angle is 168.37°, which is the reason for the tilting of the TiOoctahedra. The Sr−Sr distance along the c-axis for the bulk lattice is 3.90 Å, while the Sr−Sr distance at the surface is slightly contracted to 3.85 Å. The distance between the second and third layer Sr atoms is slightly expanded to 3.94 Å. The geometry parameters follow the well-established trend for cleaved surfaces with surface layer contraction and subsurface layer expansion. Different O2 adsorption geometries were investigated on the SrO surface, and we have determined two minima, the first corresponding to O2 adsorbed on a top-bridge-top site (a common notation in catalysis for active site between two atoms on the surface) and the second corresponding to O2 adsorbed at a slip position over a Sr atom. Both geometries are shown in Figure 3. The total energy of the top-bridge-top O2 adsorption site (shown in Figure 3A) is −324.64 eV, and the total energy of the slipped on top of a Sr atom O2 adsorption site (shown in Figure 3B) is −324.54 eV. Both geometries have similar energy although the top-bridge-top O2 adsorption site is slightly stabilized by 0.1 eV. In both adsorption geometries the O−O bond is 1.28 Å, close to the bond length in the isolated gas phase O2. The distance between the Sr and O atoms is approximately 2.8 Å. The magnetization shows 1.6 unpaired electrons per unit cell for both surface adsorption sites, which is close to the triplet state of the isolated gas O2 molecule. The Bader population analysis shows that there is no charge transfer between the surface and the O2 molecule. In the gas phase, the oxygen molecule is stable in its triplet ground state with two unpaired electrons occupying degenerate π-molecular orbitals with an orthogonal spatial distribution. Any dissociation of the oxygen molecule progresses through its activated states, i.e., singlet oxygen, superoxo state (O2−), or peroxo state (O22−). Such activation requires significant interaction with the catalytic surface and possible electron transfer. The activated states of the oxygen molecule are characterized by different spin states and bond lengths. The superoxo state has radical character and O−O bond of 1.3−1.4 Å, while the peroxo state is singlet and is characterized with an

O−O bond length of 1.4−1.6 Å. These can be distinguished computationally, as well as experimentally, using EPR and IR spectroscopy. Different positions for the dissociated oxygen atoms were tested on the surface, and a minimum was found corresponding to oxygen atoms on a bridge site between two Sr atoms as shown in Figure 3 C. The spin state of each O atom on the surface is close to a triplet, and the Bader analysis shows that only minimal electron density is transferred from the surface to the adsorbed O atoms; this makes them unstable and highly reactive. The energy of the slab shown in Figure 3C is −321.30 eV, which is 3.4 eV higher than the energy of the chemisorbed O2 on the surface. We performed an NEB calculation for the transition from the chemisorbed O2 at the bridge position shown in Figure 3A to the dissociated oxygen shown in Figure 3C. The results of the NEB analysis show that there is virtually no barrier to the formation of O2 on the SrO surface. Thus, we can conclude that the SrO surface alone could not catalyze the O2 dissociation reaction. In order to explore further possibilities for the O 2 dissociation reaction on the SrO-terminated surface, we investigated the catalytic activity of surface oxygen vacancies. Oxygen vacancies typically result from the presence of transition metal substituents with varying valence states (often deliberately introduced as “dopants”), as well as from electrochemical potential, or even high pressure torsion, applied to the materials. In the case of SrTiO3, a typical such dopant at the B-site is Fe. The oxidation number of Ti in perovskites is +4, and one Ti ion is compensated for by two O ions. The oxidation number of the Fe impurity in the perovskite lattice can be +4, identical to Ti, or +3. In the case of the oxidation number of +3 to each Fe impurity correspond 1.5 O atoms; i.e., the addition of two Fe3+ atoms would lead to the elimination of one O atom from the perovskite lattice or, in other words, the formation of one O-vacancy. The vacancy concentration in the investigated slab corresponds to the maximum number of vacancies in the SrTi0.75Fe0.25O3‑δ which is a realistic doping level. In the real material the actual number of oxygen vacancies would depend on the oxidation state of the Fe atoms, i.e., Fe3+/ 8276

DOI: 10.1021/acs.chemmater.5b03263 Chem. Mater. 2015, 27, 8273−8281

Article

Chemistry of Materials Fe4+. On the other hand, that ratio will depend on the experimental conditions, such as electrochemical reactions, oxygen pressure, and temperature. The actual number of vacancies would be less or equal to the maximum theoretical number. We have investigated the energy of the oxygen vacancies in the surface and the first subsurface layers, which are respectively −295.635 and −295.510 eV. The energy difference between them is 0.152 eV and predicts a slightly higher stability of the O-vacancy within the surface layer (i.e., a tendency for the surface layer of the STO to be slightly more reduced than the bulk). A similar tendency was previously reported in the work of Lee and Morgan46 LEIS measurements of Sr-doped SmCoO3 (Sm1−xSrxCoO3) verify that, with increasing dopant concentration, the surface oxygen concentration is decreased, implying the oxygen surface vacancy concentration is increased.48 We have performed the NEB transition state search for an oxygen atom diffusing between surface- and subsurface-vacancy sites, respectively, shown in Figure 4A,C. The transition state geometry is shown in Figure

Figure 5. Oxygen molecule adsorbed in surface vacancy of SrTi0.75Fe0.25O3−δ: (A) slab model; (B) surface geometry parameters; (C) electron density difference plot. Blue designates positive electron density while yellow designates negative electron density. Sr is denoted with green, Ti is denoted with blue, O is denoted with red, and Fe is denoted with brown. Distances are given in angstroms.

the O2 adsorption into the surface vacancy site. Electron density is donated from the dz2 orbital of the Ti-atom underneath the vacancy to the π* orbital of O2. Additionally, electron density re-distribution occurs in the O2 molecule as electron density is transferred from the O−O internuclei region to the π* orbitals. The π* orbitals of O2 are the result of out-ofphase combination between the oxygen 2p atomic orbitals and are characterized by an antibonding interaction, with their occupancy leading to O−O bond elongation. We have further investigated the O2 dissociation from the superoxo activated state shown in Figure 5 to an oxygen ion incorporated into the vacancy and an oxygen atom at a bridge site between two Sr-atoms. The calculations were performed using the NEB method with five images between the starting and the ending geometries. The starting, transition state, and ending geometries are shown in Figure 6 with the corresponding energies. The O2 dissociation proceeds with a barrier of 1.023 eV and is endothermic by 1.00 eV. The high energy of the resulting product, i.e., the oxygen atom incorporated into the lattice and the adsorbed O atom on the surface, is due to the small electron transfer from the surface to the adsorbed O atom and its high spin state. The Bader analysis reveals that the charge of

Figure 4. Oxygen surface and subsurface vacancies in SrTi0.75Fe0.25O3−δ: (A) vacancy in the surface layer; (B) transition state of O atom diffusion between the surface and the subsurface layers; (C) vacancy in the subsurface layer. Sr is denoted with green, Ti is denoted with blue, and O is denoted with red. Energies are given in electronvolts.

4B, and its energy is 0.598 eV higher than the energy of the slab with a surface vacancy, and 0.473 eV higher than that of the slab with a subsurface O-vacancy. The low activation barrier for Odiffusion between the two vacancy sites suggests that this process would not be rate-determining during the O 2 dissociation and lattice incorporation. We have further investigated the O2 adsorption on a SrOterminated surface with an oxygen vacancy. The optimized geometries of different initial positions of the O2 molecule converge in an end-on O2 configuration with one oxygen atom occupying the vacancy and the second at a bridge position between two Sr-atoms. The optimized geometry of the O2 molecule adsorbed at the O-vacancy site is shown in Figure 5. Part A of this figure depicts the entire slab model, showing that one of the oxygen atoms has occupied the vacant lattice position. A closer view of the surface geometry is given in Figure 5B. The Ti−O bond length at the surface is 1.87 Å. However, the distance between the exposed Ti-atom at the vacancy site and the O-atom from the adsorbed O2 molecule is 2.21 Å. This suggests that the O-atom has not yet been incorporated into the perovskite lattice. The O−O distance in the O2 molecule is increased to 1.47 Å, which suggests that the oxygen molecule has been activated either to superoxo or peroxo species. We have performed Bader population analysis, which shows identical charges of −0.65 electrons at each O atom, or −1.3 electrons for the total O2 molecule. The geometry and charge analyses suggest that the O2 molecule was activated to its superoxo state. To analyze the mechanism of O2 activation to superoxo species, we have studied the electron density difference map plotted in Figure 5C, which is a result of

Figure 6. Oxygen dissociation into vacancy site on SrTi0.75Fe0.25O3−δ surface: (A) starting geometry; (B) transition state; (C) end geometry. Sr is denoted by green, Ti is denoted by blue, O is denoted by red, and Fe is denoted by brown. Energies are given in electronvolts and are relative to the energy of the starting geometry. Distances are given in angstroms. 8277

DOI: 10.1021/acs.chemmater.5b03263 Chem. Mater. 2015, 27, 8273−8281

Article

Chemistry of Materials the surface oxygen in the product is −0.7 e−. In the O2 adsorbed in the vacancy site, shown in Figure 6A, the O−O bond is 1.47 Å and the Ti−O distance is 2.22 Å. In the transition state, shown in Figure 6B, the O−O bond elongates to 2.10 Å and the Ti−O distance is 1.92 Å. This shows that, while the reacting structure is a surface with a vacancy site and adsorbed O2 molecule, in the transition state, the O−O bond is cleaved and a new Ti−O bond is formed. In the final product, shown in Figure 6C, the O−O distance has increased to 2.70 Å and the Ti−O bond is formed with a length of 1.86 Å. One of the oxygen atoms from the adsorbed O2 molecule is fully incorporated in the lattice. The high energy of the product is due to the highly unstable remaining oxygen atom on the surface. The surface oxygen atom can further recombine back with a lattice oxygen atom which will produce an oxygen molecule and a surface oxygen vacancy, e.g., the reverse process of that shown in Figure 6. It could combine either with the same oxygen from the original oxygen molecule or with an oxygen atom from a neighboring lattice site. In the first case, the original oxygen molecule would be restored, while in the second, an oxygen molecule including a lattice oxygen atom would be created. The barrier in both cases would be the one estimated in Figure 6, e.g., 1.023 eV. This calculation corresponds to the energetics of a tracer isotope surface exchange reaction where, for example, gas phase oxygen molecules would consist of 18O and the lattice oxygen would be 16O.49 The activation barrier for the surface exchange formation of mixed molecule 18O16O would be 1.023 eV. Yoo and Bouwmeester have estimated experimentally the barrier for 16 O/18O exchange on the SrTi0.65Fe0.35O3−δ surface to be 1.15 eV, which is slightly higher compared to our results, however, Yoo and Bouwmeester used a higher Fe-doping concentration.50 The gas phase/lattice oxygen exchange on perovskite surfaces is an important equilibrium process. An understanding of this process is central to a further understanding of ionic transport phenomena in these materials.51 We believe that an additional reaction pathway can be the further recombination of two oxygen atoms on the surface (situated over the Sr−Sr bridges) to the significantly more stable O2 molecule. Such a reaction will occur rapidly. To study the entire process, we have doubled the surface area of the previously studied supercell and have investigated the dissociation of two O2 molecules on nearby located vacancy sites to lattice incorporated oxygen atoms and surface oxygen atoms and the subsequent recombination of the latter to a single oxygen molecule. The mechanism is summarized in Figure 7.

The reaction is exothermic leading to the incorporation of two lattice oxygen atoms and the formation of a stable O2 molecule. The energy of the product is 0.47 eV lower than the energy of the reactant. The reaction proceeds through a transition state with an activation barrier of 2.39 eV (approximately 1.2 eV per oxygen molecule). The oxygen reduction reaction passes through four important steps. Those are the oxygen activation by surface to molecule electron transfer (step 1), oxygen molecule dissociation reaction (step 2), oxygen atoms incorporation into the perovskite lattice (step 3), and migration and recombination of the remaining surface oxygen atoms to oxygen molecules (step 4). That activation barrier includes the dissociation of two O2 molecules, O atom lattice incorporation, and the surface O atoms migration and O2 formation. The rather high activation barrier suggests that elevated temperatures are indeed needed for this reaction, which is in agreement with the fact that high temperatures are required experimentally to observe oxygen exchange. The exothermic nature of the process suggests that the equilibrium will be shifted toward O2 dissociation and O atoms incorporation into the lattice. An important destabilizing factor is the formation of atomistic oxygen on the surface, which is not stabilized by surface electron transfer. These oxygen atoms will tend to rapidly recombine to the more stable O2 molecule. Thus, the oxygen vacancies on the SrO-terminated surface appear to be the catalytically active sites where oxygen activation occurs and might lead to oxygen dissociation. If we compare the reaction mechanisms for the oxygen surface exchange (shown in Figure 6) and oxygen lattice incorporation and recombination of the remaining surface oxygen atoms in the oxygen molecule (shown in Figure 7), we can conclude that the exchange reaction is characterized with ∼0.2 eV lower activation barrier. Thus, although both mechanisms would be possible, the exchange reaction would be the proffered one. Favorable positioning of oxygen vacancies on the AOterminated perovskite surface might lead to further clustering of catalytic sites which could reduce the activation barrier for oxygen dissociation. We have investigated the relative energetics of two oxygen vacancies separated by a lattice site occupied by an oxygen atom and the migration of this oxygen atom into one of the vacancies. This would give rise to clustering of both vacancies next to each other leading to a surface geometry that could adsorb an entire oxygen molecule. The oxygen surface migration is summarized in Figure 8. Figure 8A shows the geometry of both vacancies away from each other, Figure 8B shows the transition state for oxygen migration, and Figure 8C shows the surface geometry of both

Figure 8. Migration of oxygen vacancies on the AO-terminated SrTi0.75Fe0.25O3−δ surface: (A) two vacancies far from each other; (B) transition state of O atom diffusion between two vacancies; (C) two neighboring O-vacancies. Sr is denoted with green, Ti is denoted with blue, O is denoted with red, and Fe is denoted with brown. Energies are given in electronvolts and are relative to the energy of the starting geometry.

Figure 7. Two oxygen molecules dissociation into vacancy sites on the SrTi0.75Fe0.25O3−δ surface and their sequential recombination: (A) starting geometry; (B) transition state; (C) end geometry. Sr is denoted with green, Ti is denoted with blue, O is denoted with red, and Fe is denoted with brown. Energies are given in electronvolts and are relative to the energy of the starting geometry. 8278

DOI: 10.1021/acs.chemmater.5b03263 Chem. Mater. 2015, 27, 8273−8281

Article

Chemistry of Materials vacancies next to each other. Intuitively, one would expect that oxygen surface migration occurs at lower activation barriers compared to the subsurface to surface layer oxygen migration because the surface lacks the steric hindrance of atoms situated above the migrating oxygen. However, our calculations have shown that the energy barrier for the surface migration, 0.98 eV, is almost twice the energy barrier required for subsurface to surface layer oxygen migration. To understand that result, we considered the transition state geometries in Figure 4B and Figure 8B. In the case of subsurface to surface oxygen migration (shown in Figure 4B), the oxygen atom is bound to a Ti atom through the entire reaction path being stabilized by electron transfer from Ti to O. In the case of oxygen surface migration (shown in Figure 8B), the O−Ti bond is completely cleaved and this process requires additional energy. The oxygen atom must pass over a bridge site between two Sr-atoms, where it is not stabilized by Ti to O electron transfer. The resulting vacancy clustering (shown in Figure 8C) is stabilized by −0.73 eV leading to the conclusion that multiple oxygen vacancy islands might be formed on the AO-terminated surface. The catalytic activity of such vacancy islands might differ from the catalytic activity of a single surface oxygen vacancy. The decreased energy is a result of lattice relaxation of several vacancies clustered together. In bulk SrTiO3 the Ti−O−Ti angle is 168°. In the case of single surface oxygen vacancy, the Ti-atom below the vacancy “sinks” toward the bulk which decreases the Ti−O−Ti angle next to the vacancy site to 156°. However, this geometry gives rise to a steric tension caused by the neighboring Ti-atom compensated by a surface lattice oxygen atom. That Ti remains at its position and decreases the lattice deformation and the vacancy stabilization. In the case of neighboring vacancies, the Ti−O−Ti angle is reduced further to 136°. Both Ti atoms sink in the direction of the bulk leading to smaller Ti−O−Ti angle and lattice relaxation. Figure 9

Figure 10. Oxygen molecule dissociation into neighboring vacancy sites on the SrTi0.75Fe0.25O3−‑δ surface: (A) starting geometry; (B) transition state; (C) end geometry. Sr is denoted by green, Ti is denoted by blue, and O is denoted by red. Energies are given in electronvolts and are relative to the energy of the starting geometry. Distances are given in angstroms.

elongated to 1.51 Å, which suggests that the oxygen molecule has been activated by surface to molecule electron transfer. The elongated O−O bond is the result of electron density transfer to the π* molecular orbital of O2. The geometry and charge analyses suggest that the O2 molecule was activated to its superoxo state. The distance between the Ti-atom and the closest O-atom from the oxygen molecule is 2.31 Å, showing that it has not been incorporated into the lattice structure. The transition state geometry, estimated with the NEB method, is shown in Figure 10B. The oxygen molecule was split over a Sr− Sr bridge site, and the activation barrier was estimated to be 0.5 eV. The reaction proceeds further as the second oxygen atom fills the second lattice oxygen vacancy, leading to a vacancy free AO-terminated perovskite surface. The energy of the product (Figure 10C) relative to the energy of the starting structure (Figure 10A) is −2.38 eV. The reaction is strongly exothermic, and the equilibrium is shifted toward O2 dissociation and oxygen atoms incorporation into the neighboring surface vacancies. The activation barrier of the O2 dissociation is lower compared to the O atom surface migration, which would make the later the rate-determining step for the proposed reaction mechanism. The reaction mechanism includes only three of the four possible steps for oxygen reduction reaction. Those are the molecular oxygen activation (step 1), molecular oxygen dissociation (step 2), and atomic oxygen incorporation in the perovskite lattice (step 3). The fourth step, surface atomic oxygen migration and recombination to oxygen molecules, is omitted due to the favorable location of the oxygen vacancies. We have also investigated the oxygen dissociation reactions for the SrTi0.75Fe0.25O3−δ surface, where the Fe impurities were located beneath the oxygen surface vacancies. In this way the Fe atoms would have maximal effect on the oxygen dissociation reaction. The results are summarized in Figure S1 and Figure S2 of the Supporting Information. While the Fe atoms beneath the oxygen vacancy indeed have small effect and slightly reduce the required activation barrier from 0.5 eV (in the case of Ti) to 0.4 eV (in the case of Fe), the change does not alter the qualitative conclusions for the reaction mechanism and reaction energetics.

Figure 9. Surface geometry and Ti−O−Ti angle for various slabs: (A) SrTiO 3 surface without surface oxygen vacancies; (B) SrTi0.75Fe0.25O3−δ surface with single surface oxygen vacancy; (C) SrTi0.75Fe0.25O3−δ surface with two neighboring surface oxygen vacancies. Sr is denoted by green, Ti is denoted by blue, O is denoted by red, and Fe is denoted by brown. Distances are given in angstroms.

summarizes the results for the surface geometries. In Figure 9A is shown the geometry of the SrTiO3 surface without surface oxygen vacancies. Figure 9B shows the geometry of the SrTi0.75Fe0.25O3−δ surface with single surface oxygen vacancy, and Figure 9C shows the geometry of the SrTi0.75Fe0.25O3−δ surface with two neighboring surface oxygen vacancies. We have also investigated the mechanism of O2 dissociation and oxygen lattice incorporation catalyzed by two neighboring oxygen vacancy sites. The results are summarized in Figure 10, starting from the geometry of the lattice vacancy activated oxygen molecule (Figure 10A). The oxygen−oxygen bond is

■

CONCLUSION We have investigated the oxygen reduction reaction on the SrO-terminated surfaces of SrTiO3 and SrTi0.75Fe0.25O3−δ. We have shown using LEIS measurements that at SOFC operating temperatures those surfaces are SrO-terminated. As those materials are exposed to such conditions for long times when applied in solid oxide fuel cells and electrolyzers, it can be expected that the oxygen molecule dissociation and further O 8279

DOI: 10.1021/acs.chemmater.5b03263 Chem. Mater. 2015, 27, 8273−8281

Article

Chemistry of Materials

P13770). We thank Dr. José Santiso (ICN2, Barcelona, Spain) for providing the SrTiO3 single crystal sample.

atom incorporation into the perovskite surface should take place on the SrO-terminated surface and not on the presumed catalytically active TiO2-terminated surface. The latter is often assumed when describing the oxygen exchange reactions, as it exposes the transition metals to the gas. However, our calculations for the oxygen dissociation on a pristine SrOterminated surface have shown that the O2 molecule is not readily activated by the surface and the reaction is strongly endothermic by 3.4 eV. We have further considered the possibility for O2 dissociation catalyzed by oxygen vacancies on the SrO-terminated surface, arising from substitution of Tiatoms with Fe-atoms. We have shown that O-vacancies in the surface layer are stabilized by ∼0.15 eV compared to subsurface vacancies. We have estimated that the activation barrier for Ovacancy migration between the surface and the subsurface layer to be ∼0.5 eV. We have also investigated the oxygen vacancy mobility on the SrO-terminated surface showing that they tend to form vacancy clusters where the activation barrier for the clustering was estimated to be 0.98 eV. Furthermore, we have shown that O-vacancies indeed activate O2 molecules to superoxo state, which is a precursor for further O2 dissociation. The activation is a result of interaction between the dz2 atomic orbitals of the subsurface Ti-atom and the π* molecular orbital of O2. We have suggested reaction mechanisms for the O2 dissociation catalyzed by two adjacent vacancies and two clustered vacancies. In the case of adjacent vacancies two oxygen molecules are split into the vacancies leading to the incorporation of two O atoms into the perovskite lattice. The remaining O atoms recombine to form O2 molecule which desorbs into the gas phase. This reaction is exothermic and is characterized with activation barrier of 2.39 eV (1.2 eV per oxygen molecule). In the case of clustered vacancies one oxygen molecule is split and incorporated into the vacancy sites. The reaction is exothermic and is characterized with activation barrier of 0.5 eV. For the later process the ratedetermining step would be the surface vacancy clustering with activation barrier of 0.98 eV. The oxygen surface exchange reaction, e.g., O2 dissociation and formation of new O2 molecule, has an activation barrier of 1.02 eV.

■

■

(1) Ishihara, T., Ed. Perovskite Oxide for Solid Oxide Fuel Cells; Springer Science and Business Media: Dordrecht, The Netherlands, 2009; DOI: 10.1007/978-0-387-77708-5. (2) Jacobson, A. J.; Tofield, B. C.; Fender, B. E. F. The structures of BaCeO3, BaPrO3 and BaTbO3 by neutron diffraction: Lattice parameter relations and ionic radii in O-perovskites. Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem. 1972, 28, 956−961. (3) Goldschmidt, V. M. Die Gesetze der Krystallochemie. Naturwissenschaften 1926, 14, 477−485. (4) Abbes, L.; Noura, H. Perovskite oxides MRuO3 (M = Sr, Ca and Ba): Structural distortion, electronic and magnetic properties with GGA and GGA-modified Becke−Johnson approaches. Results Phys. 2015, 5, 38−52. (5) Ishihara, T.; Furutani, H.; Honda, M.; Yamada, T.; Shibayama, T.; Akbay, T.; Sakai, N.; Yokokawa, H.; Takita, Y. Improved Oxide Ion Conductivity in La0.8Sr0.2Ga0.8Mg0.2O3 by Doping Co. Chem. Mater. 1999, 11, 2081−2088. (6) Zhu, J.; Li, H.; Zhong, L.; Xiao, P.; Xu, X.; Yang, X.; Zhao, Z.; Li, J. Perovskite Oxides: Preparation, Characterizations, and Applications in Heterogeneous Catalysis. ACS Catal. 2014, 4, 2917−2940. (7) Adler, S. B. Factors governing oxygen reduction in solid oxide fuel cell cathodes. Chem. Rev. 2004, 104, 4791−4843. (8) Royer, S.; Duprez, D.; Kaliaguine, S. Oxygen mobility in LaCoO3 perovskites. Catal. Today 2006, 112, 99−102. (9) Suntivich, J.; Gasteiger, H. A.; Yabuuchi, N.; Nakanishi, H.; Goodenough, J. B.; Shao-Horn, Y. Design principles for oxygenreduction activity on perovskite oxide catalysts for fuel cells and metalair batteries. Nat. Chem. 2011, 3, 546−550. (10) Rincon, R. A.; Masa, J.; Mehrpour, S.; Tietz, F.; Schuhmann, W. Activation of oxygen evolving perovskites for oxygen reduction by functionalization with Fe-Nx/C groups. Chem. Commun. 2014, 50, 14760−14762. (11) May, K. J.; Carlton, C. E.; Stoerzinger, K. A.; Risch, M.; Suntivich, J.; Lee, Y.-L.; Grimaud, A.; Shao-Horn, Y. Influence of Oxygen Evolution during Water Oxidation on the Surface of Perovskite Oxide Catalysts. J. Phys. Chem. Lett. 2012, 3, 3264−3270. (12) Li, J.-H.; Fu, X.-Z.; Luo, J.-L.; Chuang, K. T.; Sanger, A. R. Application of BaTiO3 as anode materials for H2S-containing CH4 fueled solid oxide fuel cells. J. Power Sources 2012, 213, 69−77. (13) Metlenko, V.; Ramadan, A. H.; Gunkel, F.; Du, H.; Schraknepper, H.; Hoffmann-Eifert, S.; Dittmann, R.; Waser, R.; De Souza, R. A. Do Dislocations Act as Atomic Autobahns for Oxygen in the Perovskite Oxide SrTiO3. Nanoscale 2014, 6, 12864−12876. (14) Marrocchelli, D.; Sun, L.; Yildiz, B. Dislocations in SrTiO3: easy to reduce, but not so Fast for Oxygen Transport. J. Am. Chem. Soc. 2015, 137, 4735−4748. (15) Inoue, I. H. Electrostatic carrier doping to perovskite transitionmetal oxides. Semicond. Sci. Technol. 2005, 20, S112. (16) Xie, Z.; Zhao, H.; Du, Z.; Chen, T.; Chen, N.; Liu, X.; Skinner, S. J. Effects of Co Doping on the Electrochemical Performance of Double Perovskite Oxide Sr2MgMoO6−δ as an Anode Material for Solid Oxide Fuel Cells. J. Phys. Chem. C 2012, 116, 9734−9743. (17) Kim, C. H.; Qi, G.; Dahlberg, K.; Li, W. Strontium-Doped Perovskites Rival Platinum Catalysts for Treating NOx in Simulated Diesel Exhaust. Science 2010, 327, 1624−1627. (18) Bévillon, E.; Dezanneau, G.; Geneste, G. Oxygen incorporation in acceptor-doped perovskites. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 83, 174101. (19) Eichel, R.-A. Structural and dynamic properties of oxygen vacancies in perovskite oxides-analysis of defect chemistry by modern multi-frequency and pulsed EPR techniques. Phys. Chem. Chem. Phys. 2011, 13, 368−384. (20) Zhang, J.; Xie, K.; Wei, H.; Qin, Q.; Qi, W.; Yang, L.; Ruan, C.; Wu, Y. In situ formation of oxygen vacancy in perovskite

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.chemmater.5b03263. Reaction paths, geometries, and images of all calculated NEBs (PDF)

■

REFERENCES

AUTHOR INFORMATION

Corresponding Authors

*(A.S.) E-mail: [email protected]. *(J.K.) E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by World Premier International Research Center Initiative (WPI), Ministry of Education, Culture, Sports, Science, and Technology of Japan (MEXT), Japan. H.T. acknowledges the Japanese Society for the Promotion of Science (JSPS) for her fellowship (Grant 8280

DOI: 10.1021/acs.chemmater.5b03263 Chem. Mater. 2015, 27, 8273−8281

Article

Chemistry of Materials Sr0.95Ti0.8Nb0.1M0.1O3 (M = Mn, Cr) toward efficient carbon dioxide electrolysis. Sci. Rep. 2014, 4, 7082. (21) Chen, Y.; Jung, W.; Cai, Z.; Kim, J. J.; Tuller, H. L.; Yildiz, B. Impact of Sr segregation on the electronic structure and oxygen reduction activity of SrTi1−xFexO3 surfaces. Energy Environ. Sci. 2012, 5, 7979−7988. (22) Lee, W.; Han, J. W.; Chen, Y.; Cai, Z.; Yildiz, B. Cation Size Mismatch and Charge Interactions Drive Dopant Segregation at the Surfaces of Manganite Perovskites. J. Am. Chem. Soc. 2013, 135, 7909− 7925. (23) Jalili, H.; Chen, Y.; Yildiz, B. Structural, Chemical, and Electronic State on La0.7Sr0.3MnO3 Dense Thin-Film Surfaces at High Temperature: Surface Segregation. ECS Trans. 2010, 28, 235− 240. (24) Chroneos, A.; Vovk, R. V.; Goulatis, I. L.; Goulatis, L. I. Oxygen transport in perovskite and related oxides: A brief review. J. Alloys Compd. 2010, 494, 190−195. (25) Muñoz-García, A. B.; Ritzmann, A. M.; Pavone, M.; Keith, J. A.; Carter, E. A. Oxygen Transport in Perovskite-Type Solid Oxide Fuel Cell Materials: Insights from Quantum Mechanics. Acc. Chem. Res. 2014, 47, 3340−3348. (26) Tealdi, C.; Mustarelli, P. Improving Oxygen Transport in Perovskite-Type LaGaO3 Solid Electrolyte through Strain. J. Phys. Chem. C 2014, 118, 29574−29582. (27) Shilova, Y. A.; Patrakeev, M. V.; Mitberg, E. B.; Leonidov, I. A.; Kozhevnikov, V. L.; Poeppelmeier, K. R. Order-Disorder Enhanced Oxygen Conductivity and Electron Transport in Ruddlesden-Popper Ferrite-Titanate Sr3Fe2‑xTixO6+δ. J. Solid State Chem. 2002, 168, 275− 283. (28) Burriel, M.; Peña-Martínez, J.; Chater, R. J.; Fearn, S.; Berenov, A. V.; Skinner, S. J.; Kilner, J. A. Anisotropic Oxygen Ion Diffusion in Layered PrBaCo2O5+δ. Chem. Mater. 2012, 24, 613−621. (29) Tarancon, A.; Burriel, M.; Santiso, J.; Skinner, S. J.; Kilner, J. A. Advances in layered oxide cathodes for intermediate temperature solid oxide fuel cells. J. Mater. Chem. 2010, 20, 3799−3813. (30) Edalati, K.; Arimura, M.; Ikoma, Y.; Daio, T.; Miyata, M.; Smith, D. J.; Horita, Z. Plastic Deformation of BaTiO3 Ceramics by HighPressure Torsion and Changes in Phase Transformations, Optical and Dielectric Properties. Mater. Res. Lett. 2015, 3, 216−221. (31) Druce, J.; Téllez, H.; Burriel, M.; Sharp, M. D.; Fawcett, L. J.; Cook, S. N.; McPhail, D. S.; Ishihara, T.; Brongersma, H. H.; Kilner, J. A. Surface termination and subsurface restructuring of perovskitebased solid oxide electrode materials. Energy Environ. Sci. 2014, 7, 3593−3599. (32) Druce, J.; Ishihara, T.; Kilner, J. Surface composition of perovskite-type materials studied by Low Energy Ion Scattering (LEIS). Solid State Ionics 2014, 262, 893−896. (33) Téllez, H.; Druce, J.; Ju, Y.-W.; Kilner, J.; Ishihara, T. Surface chemistry evolution in LnBaCo2O5 + δ double perovskites for oxygen electrodes. Int. J. Hydrogen Energy 2014, 39, 20856−20863. (34) Brongersma, H.; Draxler, M.; Deridder, M.; Bauer, P. Surface composition analysis by low-energy ion scattering. Surf. Sci. Rep. 2007, 62, 63−109. (35) Lee, D.; Lee, Y.-L.; Grimaud, A.; Hong, W. T.; Biegalski, M. D.; Morgan, D.; Shao-Horn, Y. Strontium influence on the oxygen electrocatalysis of La2‑x Srx NiO4±δ (0.0 < x < 1.0) thin films. J. Mater. Chem. A 2014, 2, 6480−6487. (36) Zhou, J.; Chen, G.; Wu, K.; Cheng, Y. Interaction of La2NiO4 (100) Surface with Oxygen Molecule: A First Principles Study. J. Phys. Chem. C 2013, 117, 12991−12999. (37) De Souza, R. A. Oxygen diffusion in SrTiO3 and related perovskite oxides. Adv. Funct. Mater. 2015, 25, 6326−6342. (38) Kresse, G.; Hafner, J. Ab initio molecular dynamics for liquid metals. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 47, 558. (39) Kresse, G.; Furthmüller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 11169.

(40) Kresse, G.; Furthmüller, J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set. Comput. Mater. Sci. 1996, 6, 15−50. (41) Blöchl, P. E. Projector augmented-wave method. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 50, 17953. (42) Momma, K.; Izumi, F. VESTA 3 for three-dimensional visualization of crystal, volumetric and morphology data. J. Appl. Crystallogr. 2011, 44, 1272−1276. (43) Tang, W.; Sanville, E.; Henkelman, G. A grid-based Bader analysis algorithm without lattice bias. J. Phys.: Condens. Matter 2009, 21, 084204. (44) Jung, W.; Tuller, H. Investigation of surface Sr segregation in model thin film solid oxide fuel cell perovskite electrodes. Energy Environ. Sci. 2012, 5, 5370−5378. (45) Tasker, P. W. The stability of ionic crystal surfaces. J. Phys. C: Solid State Phys. 1979, 12, 4977. (46) Lee, Y.-L.; Morgan, D. Ab initio defect energetics of perovskite (001) surfaces for solid oxide fuel cells: A comparative study of LaMnO3 versus SrTiO3 and LaAlO3. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 91, 195430. (47) Goniakowski, J.; Finocchi, F.; Noguera, C. Polarity of oxide surfaces and nanostructures. Rep. Prog. Phys. 2008, 71, 016501. (48) Fullarton, I. C.; Jacobs, J.-P.; van Benthem, H. E.; Kilner, J. A.; Brongersma, H. H.; Scanlon, P. J.; Steele, B. C. H. Study of oxygen ion transport in acceptor doped samarium cobalt oxide. Ionics 1995, 1, 51−58. (49) Kilner, J. A.; Steele, B. C. H.; Ilkov, L. Oxygen self-diffusion studies using negative-ion secondary ion mass spectrometry (SIMS). Solid State Ionics 1984, 12, 89−97. (50) Yoo, C.-Y.; Bouwmeester, H. J. M. Oxygen surface exchange kinetics of SrTi1‑xFexO3-δ mixed conducting oxides. Phys. Chem. Chem. Phys. 2012, 14, 11759−11765. (51) Carter, S.; Selcuk, A.; Chater, R. J.; Kajda, J.; Kilner, J. A.; Steele, B. C. H. Oxygen transport in selected nonstoichiometric perovskitestructure oxides. Solid State Ionics 1992, 53−56, 597−605.

8281

DOI: 10.1021/acs.chemmater.5b03263 Chem. Mater. 2015, 27, 8273−8281