High-throughput chemistry at the interface with machine learning for advanced lithium-ion batteries

• Applicant: Dr. Janine Richter
• Subject Area: Solid State and Surface Chemistry, Material Synthesis
• Term: since 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 555840337

The challenges of climate change force a rapid transition of the global energy supply from fossil to renewable feedstocks, involving the electrification of numerous areas of societal life. The associated increase in demand for electrical energy requires enormous electrochemical storage capacities. The development of advanced battery materials is, therefore, an urging topic in current research. Lithium-ion batteries (LIBs) are at the forefront of these investigations since they are already in commercial use, but still bear great potential for optimization. This project concerns the development of advanced cathode materials for LIBs by the application of modern high-throughput chemistry at the interface with machine learning. In two promising systems, namely LiCoPO4 and compounds of disordered rock-salt structure, LIB cathode materials with improved properties shall be found. This will be realized by doping with impurity metal atoms, which has shown to be effective for the optimization of material properties. An unknown number of unknown dopants in unknown ratios spans a large search space of at least tens of thousands of materials to be synthesized. Therefore, in the course of these investigations, semi-automated high-throughput methods will be applied and developed, allowing for the resource-efficient synthesis of a large number of compounds in a short timeframe. The resulting datasets from characterization methods, such as powder X-ray diffraction and different electrochemical techniques, will serve the training of machine learning (ML) algorithms. Two ML application modes are envisioned within this project. Firstly, explainable ML shall uncover chemical coherencies between material composition and electrochemical properties. Secondly, predicting ML shall be used for the targeted synthesis of promising cathode materials. This is expected to contribute to chemical understanding as well as the efficient design of advanced LIBs. The application of ML to chemistry is only at its beginning. Fundamental developments are necessary to evolve its complete potential. Semi-automated high-throughput chemistry appears to be a promising match that could allow for the widespread implementation in research laboratories in the future. This project will contribute to the development of modern digital and automated methods in chemical research that are urgently needed for more sustainable, time- and resource-efficient laboratory procedures. Applying this to LIB cathode materials represents a highly relevant field where new materials could be vital components in global actions against climate change.

• DFG Programme: WBP Fellowship
• International Connection: Canada
• Host: Professor Eric McCalla, Ph.D.