Interference and k-point sampling in the supercell approach to phase-coherent transport

TL;DR

This paper systematically examines how interference effects and k-point sampling influence phase-coherent electron transport calculations, highlighting the importance of adequate k-point sampling to avoid unphysical results in supercell approaches.

Contribution

It demonstrates that k-point sampling significantly affects transmission calculations and provides guidelines for accurate supercell modeling in phase-coherent transport.

Findings

Interference between supercell images is a minor effect compared to k-point sampling errors.

Insufficient k-point sampling can cause unphysical features in transmission functions.

First-principles calculations confirm the importance of proper k-point sampling in realistic systems.

Abstract

We present a systematic study of interference and k-point sampling effects in the supercell approach to phase-coherent electron transport. We use a representative tight-binding model to show that interference between the repeated images is a small effect compared to the error introduced by using only the Gamma-point for a supercell containing (3,3) sites in the transverse plane. An insufficient k-point sampling can introduce strong but unphysical features in the transmission function which can be traced to the presence of van Hove singularities in the lead. We present a first-principles calculation of the transmission through a Pt contact which shows that the k-point sampling is also important for realistic systems.