SKM 2023 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
MM: Fachverband Metall- und Materialphysik
MM 14: Development of Computational Methods: Thermodynamics and Local Chemistry, Electronic Structure
MM 14.10: Talk
Tuesday, March 28, 2023, 12:45–13:00, SCH A 251
Understanding the success of mGGAs for band gaps. Is it the orbital dependence? — •Péter Kovács, Peter Blaha, and Georg K. H. Madsen — Institute of Materials Chemistry, Technical University of Vienna
Density functional theory has shown remarkable success in predicting various properties of solids, such as lattice parameters and cohesive energies, yet with most functionals it is known to heavily underestimate band gaps. mGGA functionals tend to result in better band gap predictions than LDA or GGAs, but the best results are still achieved at the cost of accuracy for the other properties. Recently using a systematic search in the space of mGGA functionals we were able to find functionals, where this tradeoff is small.[Péter Kovács et al., J. Chem. Phys. 157, 094110 (2022)]
While the failure of LDA and GGA functionals for gaps are often attributed to their lack of the derivative discontiunity, the success of mGGAs can not be explained solely based on their ability to exhibit discontinuous behaviour. On a database of 440 solids, we analyze how two specialized functionals, TASK and our own mGGA23, are able to better predict gaps. We aim to understand what parts of the functional shape is responsible for their success and how these affect their prediction of structural properties, such as lattice parameter. These findings can be used in functional design and also to understand failure cases of already existing functionals.