表題番号：2025C-514
日付：2026/03/25
研究課題Development of Machine Learning-Based Reactive Force Fields for Screening of Non-Toxic Metals in Sodium and Lithium Ion Batteries
| 研究者所属（当時） | 資格 | 氏名 | |
|---|---|---|---|
| （代表者） | 理工学術院 国際理工学センター（理工学術院） | 准教授 | サクティ アディチャ ウィバワ |
| （連携研究者） | Waseda University, Japan | Professor | Hiromi Nakai |
| （連携研究者） | Mirror Physics Company, USA | Doctor | Sam Walton Norwood |
| （連携研究者） | Mirror Physics Company, USA | Doctor | Sabrina Shen |
| （連携研究者） | Mirror Physics Company, USA | Doctor | Kayvon Tabrizi |
| （連携研究者） | National Research and Innovation Agency, Indonesia | Doctor | Sun Theo Constant Lotebulo Ndruru |
| （連携研究者） | University of Trento, Italy | Professor | Narges Ataollahi |
| （連携研究者） | Bandung Institute of Technology, Indonesia | Associate Professor | Rino Rakhmata Mukti |
- 研究成果概要
- Our first outcome was published in the Journal of Molecular Catalysis, as a continuation of the research project in AY2024 to AY2025, where our research focuses gradually changed from homogeneous catalytic chemistry to the battery field. The first study investigates the catalytic conversion of furfural to furfuryl alcohol, a key biomass-derived platform chemical. We performed DFT calculations to unravel the possible reaction pathway of furfural hydrogenation to furfuryl alcohol catalyzed by cobalt(II) phenoxyimine. The study concluded that the analog Meerwein-Ponndorf-Verley reaction is energetically more favorable than stepwise pathways. Additionally, the role of ligand design is clarified, showing that electron-withdrawing bidentate ligands stabilize key transition states and improve catalytic efficiency. These findings provide fundamental insights into designing safer and more sustainable catalytic systems that avoid high-pressure hydrogen gas while maintaining high selectivity toward furfuryl alcohol production. The second outcome is more related to our research journey towards the machine learning force field development. As an initial development, we published the outcomes in the Journal of Physical Chemistry C, namely, the surface chemistry between liquid electrolytes and functionalized graphene oxides. The research was done by using the DFTB method as a basis to further develop our own machine learning interatomic potential. Our research highlights the importance of interfacial structuring in enhancing ion transport and stabilizing electrode-electrolyte interactions. The study advances understanding of sodium-ion battery systems, which are promising alternatives to lithium-ion batteries due to resource abundance and lower cost.Aside from those two research outcomes, we are also further developing the graph neural network-based machine learning force field for the Li-dendrite formation in Li-S battery. The Li-dendrite formation in many battery systems, is generally depends on the presence of sulfur and the type of electrolytes. For the liquid electrolytes, the presence of organic solvents promotes the lithium anode oxidation, leading to a further solvent decomposition into smaller chemical species, e,g., propene and carbonate ion. These chemical species are also chemical ingredients of the infamous solid-electrolyte interface. The research outcomes are submitted to the Journal of Materials Chemistry A, RSC publisher.