Issues with Density Functional Theory for Describing Defects and Dopants in Metal Oxides

Defects and dopants in metal oxides are absolutely crucial to their performance in
many applications. Modelling of defect formation and doping is an important part of
understanding these processes. The most widely used approach is density functional theory
(DFT). However, due to limitations in present approximate exchange-correlation
functionals, e.g. the local density and generalised gradient approximations, the DFT
description of many defects and dopants in widely used metal oxides
is, in the main, entirely incorrect.

We present some results from our work on cerium dioxide and titanium dioxide, where we
have used DFT corrected for on-site Coulomb interactions, DFT+U, to study oxygen
vacancy formation, metal adsorption and the defects that form upon doping. In general, the
DFT+U results are consistent with
known experimental data and can also be used to predict the cation oxidation states present
upon doping, as well as rationalising experimental findings.

However, DFT+U has its own limitations, principally, that the U parameter is empirical
and that U can only be applied to particular species. To this end, we have begun hybrid
DFT studies of defects and dopants in metal oxides and present some initial results on this
to assess the applicability of DFT+U.