Constructing first-principles phase diagrams of amorphous LixSi using machine-learning-assisted sampling with an evolutionary algorithm

TL;DR

This paper introduces a machine-learning-assisted sampling method combining genetic algorithms and neural network potentials to efficiently construct phase diagrams of amorphous LixSi, significantly reducing the need for extensive first-principles calculations.

Contribution

It presents a novel methodology that accelerates the sampling of amorphous materials' phase space using a small number of first-principles calculations combined with machine learning and evolutionary algorithms.

Findings

The ANN potential requires only ~1,000 first-principles calculations for sampling.

The phase diagram matches results from extensive molecular dynamics simulations.

The approach is effective for first-principles modeling of amorphous materials.

Abstract

The atomistic modeling of amorphous materials requires structure sizes and sampling statistics that are challenging to achieve with first-principles methods. Here, we propose a methodology to speed up the sampling of amorphous and disordered materials using a combination of a genetic algorithm and a specialized machine-learning potential based on artificial neural networks (ANN). We show for the example of the amorphous LiSi alloy that around 1,000 first-principles calculations are sufficient for the ANN potential assisted sampling of low-energy atomic configurations in the entire amorphous LixSi phase space. The obtained phase diagram is validated by comparison with the results from an extensive sampling of LixSi configurations using molecular dynamics simulations and a general ANN potential trained to ~45,000 first-principles calculations. This demonstrates the utility of the approach for the first-principles modeling of amorphous materials.