▌Rational electrolyte design for Li-metal batteries operated under extreme conditions: a combined DFT, COSMO-RS, and machine learning study
運用COSMO-RS與機器學習篩選鋰金屬電池二元溶劑混合物之計算研究
L.T. Wu, Y.T. Zhan, Z.L. Li, P.T. Chen, B.J. Hwang, J.C. Jiang*
https://doi.org/10.1039/d4ta03026e
SEED Member: B.J. Hwang, J.C. Jiang
Major Contributions
1. Developed a comprehensive computational framework combining COSMO-RS and machine learning to evaluate binary solvent mixtures for lithium metal batteries, successfully predicting phase diagrams, salt solubility, and ion binding energies.
2. Identified optimal binary solvent mixtures with wide liquid ranges and high LiTFSI solubility, particularly highlighting PC/TMS, FEC/TMS, and GBL/TMS as promising candidates for extreme temperature applications.
3. Created an efficient machine learning model based on σ-descriptors that accurately predicts Li+ and TFSI- binding energies, providing a rapid screening tool for electrolyte design without requiring extensive DFT calculations.
主要貢獻
1. 開發了結合COSMO-RS和機器學習的全面性計算框架，成功預測二元溶劑混合物的相圖、鹽溶解度和離子結合能，為鋰金屬電池電解質設計提供有效工具。
2. 發現具有寬廣液態範圍和高LiTFSI溶解度的最佳二元溶劑混合物，特別指出PC/TMS、FEC/TMS和GBL/TMS作為極端溫度應用的理想候選物。
3. 建立了基於σ-描述符的高效機器學習模型，能夠準確預測鋰離子和TFSI離子的結合能，無需進行大量密度泛函理論計算即可快速篩選電解質。
