Accelerating Battery Organic Materials Discovery

The increasing energy storage capacity of next-generation materials for Li-ion batteries underscores the crucial role of organic electrolyte materials in suppressing degradation and enhancing performance of the systems, highlighting the criticality of electrolyte materiel development. Recently, the material development paradigm has been changing into employing automation, data, and deep learning. However, in the case of organic electrolyte materials, it is arduous to define and model multifaceted behaviors, let alone computational automation to extract features, due to complex interplay of various physicochemical properties of electrolyte materials such as ion transport in a solvation state, interaction between electrode and electrolyte components, and formation and evolution of interfaces. Organic molecules have inherently enormous chemical diversity, making it more difficult to develop hypotheses related to the function of electrolyte materials. While there have been attempts to apply data-driven methodologies for developing organic electrolyte materials in batteries, most previous studies have described models for the complex behavior of electrolytes with simple chemical properties in a single molecular state, and there have been no reports of building intermolecular relational databases.<br/> This presentation provides a pioneering framework for accelerating the discovery of battery organic materials by establishing the first-ever intermolecular relational database. For organic electrolyte materials that are difficult to understand due to their multifaceted behavior, more than 10 material design factors based on systematic operating mechanisms were quantified using density functional theory (DFT) modeling. These design factors define crucial functions of electrolyte materials such as behavior at the interface and electrolyte stabilization, and were verified using representative electrolyte materials with known functionalities, including commercially used electrolyte materials. We have developed python-based automation program AUTOMOL that encompasses the entire process from SMILES (Simplified Molecular Input Line Entry System) input to database construction, including molecular modeling of intermolecular interactions. The AUTOMOL-DB of thousands of organic material has been established by high-performance computing whereby conducting practical material development. This study is worthwhile due to the scarcity of automated modeling processes for intermolecular interactions, and thus, the exclusive database that includes molecular interactions with hydrogen, proton, cation, anion, radical, and small molecule, is invaluable as a source that can be extended to applications in chemical, biological, medical, functional molecular and nanomaterial research. It also developed a web-based system that integrates all design factors of the database so that material data can be searched, selected, and analyzed. The AUTOMOL-DB web system provides all researchers at our R&D center with a new approach to enhancing research horizons with insight into development while understanding the location of electrolyte materials in the big data space of organic matter. This system serves as a navigator for electrolyte material excavation as well as accelerate material discovery. In addition, in order to overcome the inherent disadvantage in computational cost and minimize the intervention of human intuition, state-of-the-art machine learning technology that predicts design factors and generates target materials in real time is being applied to our workflow.