Accelerated Cathode Material Discovery Via Laboratory Automation and Machine Learning Tool

Lithium-ion batteries have revolutionized energy storage across a wide range of applications, from electric vehicles to medical devices and grid energy storage. Since it’s commercialization in 1991, scientific innovations such as nickel-rich cathodes have significantly improved the cost, energy density, and sustainability associated with this technology. However, the development of next-generation lithium-ion batteries via traditional lab scale research methods remains a slow and labor-intensive process, requiring optimization of a wide range of variables such as material concentrations, synthesis parameters, electrolyte choice, and battery architectures. To accelerate the material discovery workflow, the National Research Council (NRC) Canada, has established the Critical Battery Materials Initiative (CBMI), with a key focus on novel battery materials discovery, processing as well as recycling. This is done by integrating lab scale automation and machine learning (ML) into battery research. Currently, we have implemented automated systems for both the synthesis and testing of novel and optimized high-nickel concentration cathode materials, improving experimental throughput, accuracy, and repeatability. As well, ML is being used for predictive discovery and data-driven analysis to learn from previously conducted experiments, optimize key variables, greatly reduce the trial and error related to experimentation, and predict promising materials with a higher degree of efficiency. The most promising materials chemistries from this workflow will be adopted by NRC-Ottawa for refining scale-up processes, which are then subsequently utilized by NRC-Boucherville for device prototyping to offer a complete solution to the electric vehicle’s battery development sectors. 1,2 Our materials discovery workflow starts with an automated high-throughput synthesis platform, capable of producing up to 40 unique advanced cathode materials (doping and reagent) per batch. These synthesized materials are then subject to high-throughput X-ray diffraction (XRD) and scanning electron microscopy (SEM) characterizations, with the resulting data analyzed by our ML tools to identify phase compositions and predict novel chemistries with enhanced electrochemical performances using trained large language models (LLM), enabling high-efficiency filtering of promising cathodes. The down-selected materials are then integrated into a combinatorial cell (combi cell), 3 a custom printed circuit board, capable of simultaneously testing up to 64 in-house batteries. These batteries are then subject to electrochemical impedance spectroscopy (EIS) measurements, with the collected data analyzed by ML models to predict high-performance cathode chemistries in terms of charge transfer, diffusion properties, and interfacial stability. Through leveraging this highly automated and repeatable workflow, we have significantly accelerated our research cycle and synthesized high-performance cathodes for next-generation lithium-ion batteries. References You Y., Celio H., Li J. Li, Dolocan A., and Manthiram A., Angew. Chem. Int. Ed. 2018, 57, 6480 –6485. Duffner F., Kronemeyer N., Tübke J., Leker J., Winter M. and Schmuch R., Nat. Energy, 2021, 6, 123. Potts K. P., Grignon E., and McCalla E., ACS Appl. Energy Mater. 2019, 2, 8388−8393. Figure 1

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.