Application of an experimental design approach to optimize aging protocols for lithium-metal batteries

The rapid expansion of the electric vehicle (EV) market has necessitated the use of high-performance battery packs, predominantly lithium-ion batteries (LIBs). Their implementation in devices and adaptation to specific applications can profit of computational models able to predict their functional behaviour and aging. However, the advancement of LIBs is constrained by the chemical and electrochemical limits of their materials, leading to interest in lithium metal batteries (LMBs) due to lithium's superior theoretical specific capacity and redox potential. Despite the potential advantages of LMBs, challenges such as uneven metal deposition leading to continuous side reaction with the electrolyte, active material loss through formation of dead Li, dendrite formation and safety issues hinder their practical application. These critical points limited the developments of reliable predictive models to outline in silico the functional properties of LMBs and aging. This study aims to develop a computational tool to monitor the state-of-health (SOH) of LMBs and predict capacity fading. A D-optimal experimental design approach was employed to systematically investigate the effects of various aging factors, including state of charge (SOC), C-rate, rest time, and depth of discharge (DoD) on LMB performance by selecting 18 compatible experimental cycling conditions. Starting from this dataset a regression framework was utilized to model the SOH, providing key insights into the aging mechanisms. The results indicate that while overall capacity loss correlates with the selected variables, the specific impact on open-circuit voltage changes was less pronounced. This study highlights the effectiveness of combining experimental design and chemometric analysis to enhance our understanding of LMB aging, thereby paving the way for improved battery health monitoring and management strategies.