Lithium-ion batteries (LIBs) are the principal energy storage technology for electric vehicles and portable electronics, which can release our worldwide dependency on fossil fuels. Therefore, LIBs with high power density and high energy, along with the potential to resist extreme cycling environments, are required to compete with the flexibility of the fossil fuel-dependent combustion engine. It is important to study the reversibility and the efficiency of the redox reactions at both electrodes. Degradation in reversibility and efficiency occur for numerous reasons: reactions in the electrode bulk, high charging rate, electrolyte consumption in secondary reactions, and overcharging the cell past its peak capacity. The functioning of LIBs is also linked to the morphology of the constituent materials. To develop solutions to these problems, a greater understanding of these unwanted behaviors and the chemical processes behind them is required.<br/><br/>In the following study hard X-ray nanocomputed tomography was performed on NCM523 (cathode) and graphite (anode) samples which were subjected to multiple fast charged cycles. Deep learning-based U-Net convolutional neural network (CNN) is used for semantic segmentation to investigate the morphological degradation of cathode and anode electrodes. The following models can be trained and applied to segment electrodes into its components (which are active particle, cracks and carbon binder). Particle damage is tracked across the depth of the electrodes. Thus, the following segmentations are utilized to look into evolution of microstructures of the electrodes in relation to electrochemical cycling.

Deep Learning Based Segmentation of Li-ion battery electrodes to probe inhomogeneity of microstructures

Fall 2022 abstracts, posters, and presentations will launch in mid-September.

ACS Fall 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.