Efficient and accurate approach to modeling the microstructure and defect properties of LaCoO3

Complex perovskite oxides are promising materials for cathode layers in solid oxide fuel cells. Such materials have intricate electronic, magnetic, and crystalline structures that prove challenging to model accurately. We analyze a wide range of standard density functional theory approaches to modeling a highly promising system, the perovskite LaCoO3, focusing on optimizing the Hubbard U parameter to treat the self-interaction of the B-site cation's d states, in order to determine the most appropriate method to study defect formation and the effect of spin on local structure. By calculating structural and electronic properties for different magnetic states we determine that U=4 eV for Co in LaCoO3 agrees best with available experiments. We demonstrate that the generalized gradient approximation (PBEsol+U) is most appropriate for studying structure versus spin state, while the local density approximation (LDA+U) is most appropriate for determining accurate energetics for defect properties.