Battery-on-chip characterization and modeling by machine learning-assisted techniques

Job description: The physical mechanisms involved in the operation of a microbattery are still poorly understood and modelled. To study them, the CEA has a manufacturing and characterisation platform dedicated to lithium components. The aim of this thesis is to develop a physical model to describe the performance of batteries (voltage, power output) depending on the conditions of use. The proposed methodology consists of : 1. Using 15,000 batteries per wafer as test vehicles. Measurements taken during battery cycling are compiled in a database. 2. Participating in the development of data processing programmes based on machine learning and Bayesian inference methods to highlight optimal cycling protocol parameters. the results are feed back into the physical and electrochemical model (validation/understanding/exploration). 3. Iteration with the manufacture and electrical testing of new battery architectures/designs.