Empirical modeling of degradation mode evolution during lithium-ion battery aging

Lithium-ion battery degradation leads to the fade of capacity and power capability over lifetime. Thus, degradation is a key concern during the design of lithium-ion batteries and lithium-ion battery systems. To estimate the effect of battery degradation during the design process, lithium-ion battery aging models are employed. These models are commonly parameterized on capacity fade curves measured either in a laboratory setup or collected during operation and these models aim to identify the correlation between capacity fade and the operating strategy.

This correlation is complex, since a myriad of degradation mechanism are triggered inside of the cells in dependence of e.g. temperature, mechanical stress, charge and discharge currents and state of charge. Thus, on the one hand, complex mathematical models, such as algebraic equations with many parameters or machine learning models, are commonly used to model this correlation. However, such high model complexity or black-box approaches come at the cost of reduced interpretability and transferability of these models.

On the other hand, during post-mortem aging studies or in-depth analyses of degradation behavior, the large number of degradation mechanisms are often categorized into a smaller number of degradation modes, which carry additional insights about the cause of capacity. Three commonly estimated degradation modes are loss of active anode material, loss of active cathode material and loss of lithium inventory.

In this study, we propose a method to bridge the gap between the empirical capacity fade models and degradation mode analysis through modeling the correlation between degradation mode evolution and operating strategy. By modeling the degradation modes, instead of capacity fade directly, we can employ less complex mathematic models, while increasing the interpretability and accuracy of modeled capacity fade. We parameterize a model based on aging data obtained with a commercial 18650 lithium-ion battery with an NCA cathode and a Si-doped graphite anode. To estimate the degradation modes, we use the electrode potential curves to reconstruct the full-cell open-circuit-voltage curve in an optimization algorithm.