Recursive Green's functions optimized for atomistic modelling of large superlattice-based devices

TL;DR

This paper presents optimized recursive Green's function algorithms that enable efficient atomistic modeling of large superlattice-based devices, such as twisted bilayer graphene, overcoming computational challenges associated with large supercells.

Contribution

The paper introduces improvements to recursive Green's function methods, making large superlattice device simulations computationally feasible and more efficient.

Findings

Enhanced algorithms successfully model large superlattices.

Demonstrated efficiency with twisted bilayer graphene calculations.

Overcame numerical difficulties in atomistic supercell modeling.

Abstract

The Green's function method is recognized to be a very powerful tool for modelling quantum transport in nanoscale electronic devices. As atomistic calculations are generally expensive, numerical methods and related algorithms have been developed accordingly to optimize their computation cost. In particular, recursive techniques have been efficiently applied within the Green's function calculation approach. Recently, with the discovery of Moir\'e materials, several attractive superlattices have been explored using these recursive Green's function techniques. However, numerical difficulty issues were reported as most of these superlattices have relatively large supercells, and consequently a huge number of atoms to be considered. In this article, improvements to solve these issues are proposed in order to keep optimizing the recursive Green's function calculations. These improvements make the electronic structure calculations feasible and efficient in modelling large superlattice-based devices. As an illustrative example, twisted bilayer graphene superlattices are computed and presented to demonstrate the efficiency of the method.