(392aq) Active Learning-Based Optimal Experimental Design of Non-Flammable Electrolytes for Durable Li-Ion Batteries

Lithium-ion batteries (LIBs) have become a focal point of research and development in both academia and industry, primarily due to the rising global demand for electric vehicles. As the secondary battery market continues to grow, the need for enhanced energy density and improved safety features in LIBs has become critical. To meet these demands, extensive efforts, such as exploring various advanced materials, are being made to develop next-generation materials that can significantly enhance the long-term stability and overall safety of LIBs ^[1-4]. However, optimizing the composition of battery materials through repetitive experiments is a highly time-consuming and resource-intensive process. This is primarily because evaluating battery performance requires extensive sampling and rigorous testing involving long cycle tests to assess stability, capacity retention, and overall efficiency ^[5-7]. Each cycle test can take weeks or even months to complete, making it challenging to iterate and optimize material compositions rapidly.

To address these challenges, we employ multi-objective constrained batch Bayesian optimization (MCB-BO) to develop an active learning-driven design of experiments (DoEs). This approach efficiently optimizes a non-flammable electrolyte with superior durability for LIBs. In addition, we suggest new metrics (involved in the objective function) to predict the long-term cyclability that requires only the early cycle data. This approach facilitates the efficient planning of the experiments, minimizing the number of required trials through a probabilistic framework to achieve high battery performance while ensuring non-flammability. Applying this methodology to electrolyte development, we demonstrate that our approach efficiently identifies optimal compositions with minimal experimental data. Furthermore, the developed electrolyte exhibits exceptional long-term performance in LIBs while adhering to safety constraints, highlighting its practical applicability beyond the experimental design strategy itself.

References

[1] Yuan, K., Lin, Y., Li, X., Ding, Y., Yu, P., Peng, J., Wang, J., Liu, H., & Dou, S. (2024). High‐Safety anode materials for advanced Lithium‐Ion batteries. Energy & Environment Materials, 7(5). https://doi.org/10.1002/eem2.12759

[2] Sun, Y. (2020). Promising All-Solid-State batteries for future electric vehicles. ACS Energy Letters, 5(10), 3221–3223. https://doi.org/10.1021/acsenergylett.0c01977

[3] Hemavathi, S., Srirama, S., & Prakash, A. S. (2023). Present and Future generation of Secondary Batteries: A review. ChemBioEng Reviews, 10(6), 1123–1145. https://doi.org/10.1002/cben.202200040

[4] Li, Y. (2024). Safety hazards and solutions of lithium-ion batteries. Highlights in Science Engineering and Technology, 121, 158–163. https://doi.org/10.54097/app0nv17

[5] Dave, A., Mitchell, J., Kandasamy, K., Wang, H., Burke, S., Paria, B., Póczos, B., Whitacre, J., & Viswanathan, V. (2020). Autonomous Discovery of Battery Electrolytes with Robotic Experimentation and Machine Learning. Cell Reports Physical Science, 1(12), 100264. https://doi.org/10.1016/j.xcrp.2020.100264

[6] Folch, J. P., Lee, R. M., Shafei, B., Walz, D., Tsay, C., Van Der Wilk, M., & Misener, R. (2023). Combining multi-fidelity modelling and asynchronous batch Bayesian Optimization. Computers & Chemical Engineering, 172, 108194. https://doi.org/10.1016/j.compchemeng.2023.108194

[7] Noh, J., Doan, H. A., Job, H., Robertson, L. A., Zhang, L., Assary, R. S., Mueller, K., Murugesan, V., & Liang, Y. (2024). An integrated high-throughput robotic platform and active learning approach for accelerated discovery of optimal electrolyte formulations. Nature Communications, 15(1). https://doi.org/10.1038/s41467-024-47070-5