Study of the microstructure of amorphous silicon and its effect on Li transportation with neural network potential

The machine learning-based simulation methods have attracted much attention recently. In this talk, neural network (NN) potential is used to study Li diffusion mechanism in amorphous silicon (a-Si).

The structure and property of the experimental a-Si sample are significantly affected by the experimental fabrication method. In this work, the NN potential was used to generate a series of atomic structures of a-Si with different degrees of disorder. By systematically comparing various structural and vibrational properties with experiments, we can determine the corresponding theoretical model for experimental samples prepared with a certain method.[1]

The kinetics of Li diffusion in a-Si is one of the most important issues for its performance as the anode of lithium-ion battery. The effect of structural order on Li diffusion behavior is investigated with NN potential. We found that Li transportation needs higher activation energy in the highly disordered a-Si matrix. The result can be explained with the “trap” mechanism.

[1] W. Li and Y. Ando, J. Chem. Phys. 151, 114101 (2019).