Inhibition of lithium dendrite growth with highly concentrated ions: cellular automaton simulation and surrogate model with ensemble neural networks
Document Type
Article
Publication Date
12-13-2021
Department
Department of Physics
Abstract
We have developed a lattice Monte Carlo (MC) simulation based on the diffusion-limited aggregation model that accounts for the effect of the physical properties of small ions such as inorganic ions and large salt ions that mimic ionic liquids (ILs) on lithium dendrite growth. In our cellular automaton model, molecular and atomistic details are largely coarse-grained to reduce the number of model parameters. During lithium deposition, the cations of the salt and ILs form positively charged electrostatic shields around the tip of the dendrites, and the anions of the salt and ILs form negative local potential lumps in adjacent areas to the dendrite. Both of the effects change the distribution of the electrostatic potential and notably inhibit dendrite formation between electrodes. The applied voltage and the physical properties of the salt ions and ILs, such as the size of the ions, the size asymmetry between the cation and anion, the dielectric constant, the excluded volume of the ions, and the model parameter η, notably affect electric-field screening and hence the variation in the local potential, resulting in substantial changes in the aspect ratio and the average height of the dendrites. Our present results suggest that the large salts such as ILs more significantly inhibit the dendrite growth than the small ions, mainly because the ions highly dissociated in ILs can participate in electrostatic shielding to a greater degree. To reduce the computational complexity and burden of the MC simulation, we also constructed a surrogate model with ensemble neural networks.
Publication Title
Molecular Systems
Recommended Citation
Gao, T.,
Qian, Z.,
Chen, H.,
Shahbazian-Yassar, R.,
&
Nakamura, I.
(2021).
Inhibition of lithium dendrite growth with highly concentrated ions: cellular automaton simulation and surrogate model with ensemble neural networks.
Molecular Systems,
7(3), 260-272.
http://doi.org/10.1039/D1ME00150G
Retrieved from: https://digitalcommons.mtu.edu/michigantech-p/16101