Convergence of defect energetics calculations

TL;DR

This paper investigates the convergence of defect energetics calculations in hexagonal boron nitride, highlighting the importance of methodological accuracy for reliable defect property predictions.

Contribution

It systematically analyzes convergence issues in defect calculations, revealing intrinsic methodological shortcomings affecting accuracy.

Findings

Convergence depends on sample size, basis set, and electron correlation treatment.

Poor performance of some methods is due to intrinsic methodological limitations.

Establishes guidelines for reliable defect energetics calculations.

Abstract

Determination of the chemical and spectroscopic natures of defects in materials such as hexagonal boron nitride (h-BN) remains a serious challenge for both experiment and theory. To establish basics needs for reliable calculations, we consider a model defect $V_N N_B$ in h-BN in which a boron-for-nitrogen substitution is accompanied by a nitrogen vacancy, examining its lowest-energy transition, (1)2B1 to (1)2A1. This provides a relatively simple test system as open-shell and charge-transfer effects, that are difficult to model and can dominate defect spectroscopy, are believed to be small. We establish calculation convergence with respect to sample size using both cluster and 2D-periodic models, convergence with respect to numerical issues such as use of plane-wave or Gaussian-basis-set expansions, and convergence with respect to the treatment of electron correlation. The results strongly suggest that poor performance of computational methods for defects of other natures arise through intrinsic methodological shortcomings.