Effects of Strain, d-Band Filling, and Oxidation State on the Surface Electronic Structure and Reactivity of 3d Perovskite Surfaces
Sneha A. Akhade and John R. Kitchin, “Effects of strain, d-band filling, and oxidation state on the surface electronic structure and reactivity of 3d perovskite surfaces”, The Journal of Chemical Physics 137(8), 084703 (2012). https://doi.org/10.1063/1.4746117
This summary was written by a language model from the paper and has not been checked line by line. Rely on the published paper for details.
Question
Perovskite oxides (ABO3) are used as solid oxide fuel cell cathodes and in many other catalytic and electronic applications, but the link between their composition and their reactivity toward oxygen was poorly understood. For metals, electronic structure descriptors such as the d-band center have organized trends in adsorption energies, and strain is known to shift the d band. Building on an earlier study of the bulk electronic structure of the same materials, this paper asks whether strain, filling of the B-cation d band, and the B-cation oxidation state systematically control the surface electronic structure and the oxygen reactivity of cubic 3d perovskite surfaces, and whether the d-band center works as a reactivity descriptor for these oxides (Akhade 2012).
Approach
The authors studied cubic LaBO3 and SrBO3 with B = Sc, Ti, V, Cr, Mn, Fe, Co, Ni and Cu, so the B cation is formally +3 in the La series and +4 in the Sr series. They used the DACAPO plane-wave code with the PW91 GGA functional (no Hubbard U), Vanderbilt ultrasoft pseudopotentials, a 450 eV cutoff, an 8 x 8 x 1 k-point grid and non-spin-polarized calculations, arguing that magnetism matters little at high operating temperature and had a small effect in their bulk work. Surfaces were BO2-terminated (001) slabs in a 1 x 1 cell with four stoichiometric layers (eight planes), the bottom two layers held at bulk positions, 15 Å of vacuum and a dipole correction. Uniform strains of -5, -2.5, 0, +2.5 and +5% were applied relative to the DFT lattice constant. Reactivity was measured by the dissociative adsorption energy of atomic O atop the B atom (relative to half an O2 molecule) and the energy to form a surface oxygen vacancy next to the B atom. The surface d-band center and width were computed as the first and square root of the second moments of the B-projected d density of states, after separating the localized d band from the O s,p region that contains hybridized B d states.
Key findings
Strain modified the surface d band as in the bulk: compressive strain broadened it and tensile strain narrowed it, with the effect clearest on unrelaxed surfaces because relaxation of the top layer partly removes the strain normal to the surface. Unrelaxed surfaces showed a roughly linear width versus center correlation whose slope changed sign with filling, as in a rectangular band model, but the sign change occurred at fractional d fillings near 0.8 (La series) and 0.6 (Sr series) rather than the ideal 0.5, which the authors attribute partly to difficulty separating hybridized states. Relaxation degraded these correlations.
Strain changed the O adsorption and vacancy formation energies, but much less than d-band filling and oxidation state did. On relaxed LaBO3 surfaces compressive strain made O adsorption stronger and tensile strain weaker, opposite to the usual behavior on metals; on SrBO3 both signs of strain made adsorption somewhat more favorable over a smaller range. Compressive strain generally lowered and tensile strain raised the vacancy formation energy. The authors note that some extreme-strain points reflect relaxation effects and that unrelaxed and relaxed trends differ.
The two reactivity measures were approximately linearly correlated: as B-site d filling decreases, O adsorption gets stronger and vacancy formation gets more expensive, so early transition metal perovskites should have few vacancies and late ones should form them readily. Strain effects were largest for low d filling. Both reaction energies correlated with the surface d-band center, except for LaScO3 and SrTiO3, which have no free d electrons, and the Cu compounds, whose filled d bands were hard to separate from the O bands. Reaction energies scaled roughly linearly with the number of B d electrons, with slopes set by oxidation state; O adsorption was more favorable on LaBO3 surfaces while vacancies formed more easily on SrBO3 surfaces. (The extracted text gives the Sr series as +IV in the body but +II in one figure caption; the body value is used here.) Compared with published DFT+U results for B = Mn, Fe, Co and Ni, the PW91 vacancy trends match those at U = 0 and differ for material-specific U values, which the authors weigh against experimental uncertainties of about 0.5 eV.
Significance
The study gives a consistent screen of 18 perovskite surfaces and identifies d-band filling and oxidation state as coarse handles on reactivity, with strain as a fine-tuning handle. It extends the d-band center from metals to oxides, provided the d band is deconvoluted from hybridized O states, and shows that O adsorption and vacancy formation cannot be tuned independently in these materials (Akhade 2012).