Comparison of charged-defect finite-size supercell correction methods in a general framework

TL;DR

This paper compares various charged-defect supercell correction methods within a unified DFT framework, identifying two approaches that perform reliably across different defect types and materials.

Contribution

It introduces a unified framework for comparing correction schemes and evaluates their effectiveness on multiple defect cases, guiding future method development.

Findings

Two correction methods perform well across all tested cases.

The framework allows for comparison and development of correction schemes.

Potential alignment and Freysoldt-like methods are effective.

Abstract

Starting from the total energy expressions within density functional theory, we are able to perform a comparison of several currently used charged-defect finite-size supercell correction schemes in a unified manner. This approach also provides a framework for a further development of corrections not only for DFT supercell calculations, but also for more advanced methods and for complex geometries. The comparison is performed for three separate defect cases: a gallium vacancy in GaAs, a beryllium interstitial in GaAs and a vacancy in diamond. We found two methods working sufficiently well for all three cases: a method which is very similar to one presented by Freysoldt, and a slightly altered potential alignment method.