Bayesian optimization with experimental failure for high-throughput materials growth

TL;DR

This paper introduces a Bayesian optimization algorithm that effectively handles missing data due to experimental failures, enabling efficient high-throughput materials growth with successful application to SrRuO3 film production.

Contribution

The paper presents a novel Bayesian optimization method that manages missing data, improving the search efficiency in high-dimensional materials growth parameter spaces.

Findings

Achieved the highest residual resistivity ratio of 80.1 in SrRuO3 films.

Successfully applied the method to real MBE growth of SrRuO3.

Optimized materials growth with only 35 experimental runs.

Abstract

A crucial problem in achieving innovative high-throughput materials growth with machine learning and automation techniques, such as Bayesian optimization (BO) and robotic experimentation, has been a lack of an appropriate way to handle missing data due to experimental failures. Here, we propose a new BO algorithm that complements the missing data in the optimization of materials growth parameters. The proposed method provides a flexible optimization algorithm capable of searching a wide multi-dimensional parameter space. We demonstrate the effectiveness of the method with simulated data as well as in its implementation for actual materials growth, namely machine-learning-assisted molecular beam epitaxy (ML-MBE) of SrRuO3, which is widely used as a metallic electrode in oxide electronics. Through the exploitation and exploration in a wide three-dimensional parameter space, while complementing the missing data, we attained tensile-strained SrRuO3 film with a high residual resistivity ratio of 80.1, the highest among tensile-strained SrRuO3 films ever reported, in only 35 MBE growth runs.