Rational search for optimal synthesis process conditions for storage battery materials using a combination of experiments and Bayesian optimisation

2023 Fiscal Year Final Research Report

• Project/Area Number: 21K14715
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 36020:Energy-related chemistry
• Research Institution: Nagoya Institute of Technology
• Principal Investigator: Takeda Hayami 名古屋工業大学, 工学(系)研究科(研究院), 准教授 (70599000)
• Project Period (FY): 2021-04-01 – 2024-03-31
• Keywords: 蓄電池材料 / ベイズ最適化 / 実験プロセス

Outline of Final Research Achievements

Experiments and Bayesian optimisation were used to improve the conductivity of solid electrolyte materials. The target material was double-doped LiZr2P3O12 with partial substitution of Zr and Si, and the search for optimum conditions was conducted using 102 known data points as initial data in a search space of 576 points combining composition and heating conditions. As a result, it was confirmed that efficient search for synthesis conditions was possible by using Bayesian optimisation, and the search could be completed with about a quarter of the experimental volume of an exhaustive search experiment. I also worked on improving the conductivity of LiTa2PO8. A search for factors affecting the conductivity revealed that the mixing and grinding conditions of the raw materials affected the Li-ion conductivity. By optimising the mixing and grinding conditions, LiTa2PO8 samples with a Li ion conductivity of around １m S/cm at room temperature could be synthesised.

Free Research Field

無機材料

Academic Significance and Societal Importance of the Research Achievements

本研究での目的は、「低コスト・省エネルギー（少ない実験回数）で高性能材料を合成する最適プロセス条件を見出し、汎用性の高い合理的な最適条探索方法を確立する」ことであった。ものづくりの現場では、高性能材料を合成するためのプロセスを最適化するために膨大な試作を繰り返す必要がある。これには多大な時間的、金銭的コストが必要となる。本研究では、ベイズ最適化を用いてこのような問題の解決を目指した。本研究ではLiイオン蓄電池材料の最適プロセスの探索を行い、網羅的に探索した場合と比較し、約1/4のコストで発見できることを実証した。