Optimization of laser-patterned electrode architectures for fast charging of Li-ion batteries using simulations parameterized by machine learning

In this work, we employ continuum-scale modeling to optimize Highly Ordered Laser-patterned Electrode (HOLE) architectures for fast-charging (4C and 6C) of Li-ion batteries. First, we describe the workflow for parameterizing the model, which includes an automated parameterization procedure based on the particle swarm optimization algorithm. We then use the parameterized model to optimize the HOLE architecture in terms of channel size and spacing for a given volume retention value. Our results show that while closer (and smaller) channels generally result in improved fast-charging performance compared to those with larger spacings and diameters, there exists an optimal spacing below which the marginal gain in the performance falls rapidly. We also define the second Damko center dot hler number, DaII, as a metric to quantify the effect of the channel size/spacing on the electrode performance and to provide a metric for optimizing the HOLE design. Our results show that the optimal configuration has DaII approximate to 1 throughout charging. Based on this finding, we develop a semi-analytical method to obtain a time-averaged value of DaII, which can be used for high-throughput screening of various candidate electrode architectures, thereby reducing the computational cost of the overall optimization process.