Towards Self-Driving Labs for Better Batteries: Accelerating Electrolyte Discovery with Automation and Bayesian Optimization

Abstract

The integration of automation and data-driven methodologies offer a promising approach to accelerating materials discovery in energy storage research. Thus far, in battery research, coin-cell assembly has advanced to become near fully-automated but remains largely disconnected from data-driven methods, which have been primarily developed for computational or multi-fidelity datasets. To bridge the disconnect, this work presents a self-driving laboratory framework designed to accelerate electrolyte discovery by integrating automated coin-cell assembly, galvanostatic cycling of LiFePO4||Li4Ti5O12 organic-aqueous full-cells, and Bayesian optimization for selecting subsequent experiments based on prior results. The integration of Bayesian optimization highlights machine-intelligent decision-making, enabling closed-loop experimentation-analysis workflow. The study focuses on an organic-aqueous hybrid electrolyte system comprising four co-solvents—dimethyl sulfoxide, trimethyl phosphate, acetonitrile, and water—and two salts, lithium perchlorate and lithium bis(trifluoromethanesulfonyl)imide (LiTFSI). Using this framework, electrolyte formulations with at least 94% Coulombic efficiency were identified. Additionally, quantification of hydrogen evolution by online electrochemical mass spectrometry revealed a direct correlation between the electrolyte water content and the hydrogen evolution kinetics, irrespective of the electrolyte co-solvent compositions. The results highlight the potential of combining Bayesian optimization with autonomous experimentation, while contributing new insights into electrolyte design for next-generation sustainable aqueous batteries.