A refined SEI model for lithium-ion batteries

Summary:

This poster talks about adding a new sub-model, solvent consumption, into the physic-based battery life model with SEI layer growth. Solvent consumption is one important side effect of SEI layer growth, has been almost completely overlooked in models. Without solvent consumption, battery models are unable to capture lots of important phenomena such as electrolyte dry-out and effects of initial electrolyte volume on the battery life.
We first introduce our assumptions, based on these assumptions, we define three ratios to update the active electrode area, electrolyte concentration, and EC concentration in the model.
We compare the results between cells with different initial excessive electrolyte and a cell not considering solvent consumption. we find that all the cells, either with or without solvent consumption, show decelerating capacity fade as a result of the self-limiting SEI growth, but with different degradation rates.
We also find that the cell with 6% more electrolyte deviates with the cell with 9% electrolyte (which never dries out during the whole cycling period) at about 400 cycles, corresponding to the dry-out point.
Cells with solvent (in this case, EC) consumption exhibit reduced EC concentration and therefore lower SEI current density. That will lead to less loss of lithium inventory (LLI) to SEI. But these cells will have more loss of active material when dry-out occurs. The cell with 9% more electrolyte keeps ~7% more capacity than those without any extra electrolyte after ~1200 ageing cycles.
Therefore, our take away information are, battery life can be extended by adding more electrolyte at BOL. A battery life model without solvent consumption underestimates LAM but overestimates LLI. Solvent consumption may induce positive or negative feedback with lithium plating