Computational Chemistry, Short talk

CC-015

**Self-consistent DFT+****U****+****V**** study of oxygen vacancies in SrTiO**_{3}

Neutral oxygen vacancies are known to determine the physical properties and in particular the conductivity of SrTiO_{3} (STO). Despite a great deal of effort, the nature of these defects in STO is still widely debated, and results are contradictory, especially in regards to the nature of the associated electronic states. This ambiguity has often been related to STO’s peculiar electronic properties: STO is a d^{0} transition metal oxide with a mixture of ionic and covalent interactions, leading to a competition between trapping electrons associated with the defect in the vacancy or to localize them on Ti-3d orbitals. This particular situation may not be properly described by standard density functional theory (DFT) approaches or even by Hubbard corrected DFT+*U*, both providing largely underestimated bandgaps and inaccurate crystal field splitting compared to experiments. Here, we apply, for the first time, a novel DFT+*U*+*V* method in which the Hubbard *U* and *V *parameters are computed self-consistently to oxygen deficient STO. This extended Hubbard model combines the conventional on-site Coulomb interaction for *d* electrons on Ti sites (*U*) together with inter-site electronic interactions between Ti-d and O-p states (*V*). Compared to DFT+*U*, this approach increases the predictive accuracy for weakly correlated transition metal oxides and systems where hybridization plays a major role, as is the case in STO. Results show that self-consistent DFT+*U*+*V* is able to provide a picture of V_{O} defects in STO similar to the one given by hybrid DFT functionals but at a much lower computational cost.