EXTENDED ABSTRACT: The electrolyte is a pivotal component of lithium-ion batteries, primarily functioning to transport lithium ions and significantly impacting the battery's actual performance, earning it the vivid metaphor of "the blood of batteries." However, the design and development of advanced electrolytes face challenges such as the complexity of solution chemistry, the vast number of electrolyte molecules, and the difffculty in optimizing the strong correlations between electrolyte components. This report focuses on understanding the solvent chemistry principles of electrolytes and the research on designing advanced electrolytes using machine learning. Speciffcally, it explores the solvent chemistry rules of electrolytes through multi-scale simulation methods combining first-principles calculations and molecular dynamics simulations, discovers that the formation of ion-solvent structures is a crucial factor affecting electrolyte interface stability, and thereby establishes an ion-solvent chemistry model. Furthermore, it develops various prediction methods for electrolyte physical properties such as dielectric constant and viscosity. By integrating multiple physical property calculation methods and highthroughput computations, a large electrolyte database is constructed, encompassing 200,000 molecular structures and over 20 electrolyte properties. Based on this constructed electrolyte database, machine learning models are developed to quantitatively correlate the molecular structure of electrolytes with their physicochemical properties, enabling high-throughput screening and inverse design of electrolyte molecules. Consequently, over a dozen new molecular systems have been obtained, and further experimental validations are underway. This research content constructs a large electrolyte database, develops new methods for designing advanced lithium-ion battery electrolytes using artiffcial intelligence, and achieves rapid and precise design of electrolyte molecules within a space of billions of molecules, advancing the practical application of next-generation high-energy-density batteries and providing critical technological support for achieving carbon neutralizing target.
Keywords：lithium batteries; electrolyte; artiffcial intelligence; molecular machine learning; big database
REFERENCES:
[1] Yao, N.; Chen, X.;* Fu, Z.-H.; Zhang, Q.* Chem. Rev. 2022, 122, 10970–11021.
[2] Chen, X.; Shen, X.; Li, B. et al. Angew. Chem. Int. Ed. 2018, 57, 734–737.
[3] Chen, X.; Liu, X.; Shen, X.; Zhang, Q. Angew. Chem. Int. Ed. 2021, 60, 24354–24366.
[4] Yao, N.; Chen, X.;* Shen, X. et al. Angew. Chem. Int. Ed. 2021, 60, 21473–21478.
[5] Gao, Y.-C.; Yao, N.; Chen, X.* et al. J. Am. Chem. Soc. 2023, 145, 23764–23770.
[6] Yao, N.; Sun, S.-Y.; Chen, X.* et al. Angew. Chem. Int. Ed. 2022, 61, e202210859.
Xiang Chen gained his Bachelor’s and Ph.D. degrees from the Department of Chemical Engineering at Tsinghua University in 2016 and 2021, respectively. Then, he was a Shuimu Tsinghua Scholar at Tsinghua University. Currently, he is an associate research professor at Tsinghua University. His research interests focus on understanding the chemical mechanism and materials science in rechargeable batteries mainly through multi-scale simulations and machine learning. He has published more than forty (co)-ffrst-author papers in Chem. Rev., Acc. Chem. Res., Sci. Adv., Chem, Angew. Chem., J. Am. Chem. Soc., J. Energy Chem., Fundam. Res. et al., with more than 17000 citations and an H-index of 67. He has been rewarded the 2023 MIT TR35 Asia Paciffc and the Clarivate Highly Cited Researchers from 2020 to 2023. He is supported by the National Natural Science Foundation of China for Excellent Young Scholars and the Young Elite Scientists Sponsorship Program by CAST. He has reviewed more than 100 papers for high-impact journals such as Nature, Nat. Catal., Nat. Commun., Angew. Chem., Adv. Mater., and Joule