MULTI-OBJECTIVE OPTIMIZATION AND ROBUST SCREENING MODELING OF AQUEOUS ELECTROLYTE FORMULATIONS
Keywords:
Aqueous electrolyte, Formulation optimization, Random forest, Gaussian process regression, Robustness analysisAbstract
Aqueous electrolyte formulation optimization involves coupled conductivity, acid-base compatibility, electrochemical stability, and preparation robustness under limited experimental budgets. Based on 251 experimental records, this study constructs an integrated modeling framework for performance evaluation, predictive modeling, mechanism interpretation, reliability assessment, and candidate formulation screening. The original stock-solution volumes, molalities, densities, component existence indicators, ionic load, mass load, Li/Na ratio, anion-family proportions, water volume fraction, and component entropy are organized into a unified feature space. Conductivity, pH, and the electrochemical stability window measured at the 1 mA/cm2 threshold are then combined with robustness indicators to support multi-objective formulation ranking. Ridge regression, random forest, and Gaussian process regression are compared under random and structural validation. The Gaussian process model performs best for conductivity prediction, while random forest provides the strongest pH and stability-window predictions. Feature importance and response analysis show that total ion load dominates conductivity, Li/Na-related descriptors influence pH, and NaBr-related descriptors strongly affect the electrochemical window. Candidate formulations are further screened through Bayesian optimization and local perturbation analysis, identifying NaClO4-NaNO3 systems as high-potential and robust candidates for subsequent experiments.References
[1] Ji D, Kim J. Trend of developing aqueous liquid and gel electrolytes for sustainable, safe, and high-performance Li-ion batteries. Nano-Micro Letters, 2024, 16(1): 2.
[2] Xu J J, Ji X, Zhang J X, et al. Aqueous electrolyte design for super-stable 2.5 V LiMn2O4||Li4Ti5O12 pouch cells. Nature Energy, 2022, 7(2): 186-193.
[3] Xu G S, Jiang M X, Li J L, et al. Machine learning-accelerated discovery and design of electrode materials and electrolytes for lithium ion batteries. Energy Storage Materials, 2024, 72: 103710.
[4] Yik J T, Hvarfner C, Sjolund J, et al. Accelerating aqueous electrolyte design with automated full-cell battery experimentation and Bayesian optimization. Cell Reports Physical Science, 2025, 6(5): 102548.
[5] Dave A, Mitchell J, Burke S, et al. Autonomous optimization of non-aqueous Li-ion battery electrolytes via robotic experimentation and machine learning coupling. Nature Communications, 2022, 13: 5454.
[6] Xu G S, Zhang Y J, Jiang M X, et al. A machine learning-assisted study on organic solvents in electrolytes for expanding the electrochemical stable window of zinc-ion batteries. Chemical Engineering Journal, 2023, 476: 146676.
[7] Huang G C, Huang F Q, Dong W J. Machine learning in energy storage material discovery and performance prediction. Chemical Engineering Journal, 2024, 492: 152294.
[8] Wang Y. Application-oriented design of machine learning paradigms for battery science. NPJ Computational Materials, 2025, 11: 89.
[9] Wang F, Tang Y H, Ma Z B, et al. Domain oriented universal machine learning potential enables fast exploration of chemical space of battery electrolytes. Nature Communications, 2026, 17: 1226.
[10] Nguyen T M, Biressaw G M, Lee M H, et al. Hybrid aqueous electrolyte design for interfacial stabilization in high-energy-density and long-life LiNi0.8Mn0.1Co0.1O2-Li4Ti5O12 lithium-ion batteries. Journal of Energy Storage, 2025, 139: 118915.