Galvanic corrosion underlies coulombic efficiency differences in high-performing lithium metal battery electrolytes

Abstract

Current guidelines for electrolyte engineering in lithium metal batteries are based on design metrics such as lithium morphology, electrolyte transport properties, solid electrolyte interphase (SEI) characteristics, and lithium-electrolyte reactivity. In our work, we show that those design metrics fail to account for performance differences in new high-performing electrolytes whereas galvanic corrosion does. This insight regarding the importance of galvanic corrosion is enabled by the combination of machine learning with rigorous experimental characterization. First, we partition our electrolyte data into low and high coulombic efficiency (CE) segments to obtain an interpretable machine learning model which informs the design of high-performing (high CE) electrolytes. We test new model-guided, high-performing electrolytes and use spectroscopy and electroanalytical methods to demonstrate only a weak correlation between common design metrics and performance in the high-performing electrolytes. Our work results in the design of a high-performing electrolyte with a coulombic efficiency (CE) of 99.6%, a new understanding that common performance indicators are not sufficient for informing the development of high-performing electrolytes, and the identification of galvanic corrosion as an important performance driver in high-performing electrolytes.