First-principles electronic structure calculations (e.g. with density functional theory or DFT) provide unique insights into computational materials and catalyst design. However, the approximations made in semi-local DFT produce delocalization errors that limit predictive accuracy on the most compelling materials, such as correlated transition metal oxides. We carry out a systematic study to identify commonalities in how diverse approaches for approximate delocalization error correction (i.e., DFT+U and hybrid functionals) perform on representative bulk and slab models of catalytic transition metal oxides. In the solid state, DFT+U is often employed as an efficient alternative to hybrid functionals, owing to the comparable cost of the former to semi-local DFT and the high cost of the latter within periodic, plane wave electronic structure codes. We compare these observations to our previous work on molecular inorganic complexes and extended, open-framework solids that resemble transition metal complexes. For these methods, we describe the effect these approaches have on properties relevant for heterogeneous catalysis, including the nature of localization of the density, adsorption energies, and widely applied linear scaling relations. The results show divergent behavior on density localization for DFT+U, which is absent from molecules and only specific to the solid state. This work highlights the limited applicability of trends established for functional tuning on transition-metal complexes even to equivalent coordination environments in the solid state. Our work also provides a caution to the solid-state community in the interchangeability of DFT+U and hybrids and motivates further method development in approximate DFT.

Density localization and scaling relations in the solid state: Understanding divergent behavior for hybrids and DFT+U

256th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.