PNNL

Abstract

We demonstrate that the practice of using formal ionic charges to interpret core-level (CL) x-ray photoemission binding energies measured for oxides containing transition metal (TM) cations leads to a physically inconsistent picture of the electronic structure of these materials. Strong O 2p – TM 3d hybridization in the valence band is well known to result in significant covalency in TM – O bonds. However, the fully ionic bonding approach remains the standard paradigm for assigning metal CL photoemission features. A more physically meaningful way to extract information on charge distribution within TM – O bonds from x-ray photoemission involves utilizing Dirac-Hartree-Fock theory to calculate CL spectra from first principles, and then use the resulting wave functions to extract charges on atoms based on orbital electron occupancies. TM cation charges can also be determined using density functional theory and Bader population analysis. We illustrate these two methods using the Ti 2p spectrum for SrTiO3(001) and show that agreement between them is excellent. Significantly, the atomic charge determined in this way is considerably lower than the formal charge. The high degree of similarity between the Ti 2p spectrum for SrTiO3 and those for the rutile and anatase polymorphs of TiO2 suggests that the charge densities surrounding Ti in the latter materials are similar to those in SrTiO3. Taking a broader perspective, oxides containing other first-row transition metals at the B-sites also exhibit covalent character, leading to TM cation charges lower than the analogous fully ionic values.