# First Principles Modeling of Topological Insulators: Structural Optimization and Exchange Correlation Functionals

## TL;DR

This paper evaluates various density functional theory methods for modeling topological insulators, focusing on structural optimization and electronic properties, and recommends GGA with van der Waals interactions as the most suitable approach.

## Contribution

It provides a comprehensive assessment of DFT formalisms for TIs and identifies the most accurate and computationally efficient method for structural and electronic property predictions.

## Key findings

- GGA with vdW interactions best predicts atomic structures
- MetaGGA offers slightly higher accuracy in electronic structure
- LDA tends to underestimate TI slab thickness

## Abstract

Topological insulators (TIs) are materials that are insulating in the bulk but have zero band gap surface states with linear dispersion and are protected by time reversal symmetry. These unique characteristics could pave the way for many promising applications that include spintronic devices and quantum computations. It is important to understand and theoretically describe TIs as accurately as possible in order to predict properties. Quantum mechanical approaches, specifically first principles density functional theory (DFT) based methods, have been used extensively to model electronic properties of TIs. Here, we provide a comprehensive assessment of a variety of DFT formalisms and how these capture the electronic structure of TIs. We concentrate on Bi$_2$Se$_3$ and Bi$_2$Te$_3$ as examples of prototypical TI materials. We find that the generalized gradient (GGA) and kinetic density functional (metaGGA) produce displacements increasing the thickness of the TI slab, whereas we see an opposite behavior in DFT computations using LDA. Accounting for van der Waals (vdW) interactions overcomes the apparent over-relaxations and retraces the atomic positions towards the bulk. Based on an intensive computational study, we show that GGA with vdW treatment is the most appropriate method for structural optimization. Electronic structures derived from GGA or metaGGA employing experimental lattice parameters are also acceptable. In this regard, we express a slight preference for metaGGA in terms of accuracy, but an overall preference for GGA due to compensatory improvements in computability in capturing TI behavior.