Anomalous Diffusion of Lithium-Anion Clusters in Ionic Liquids using MD simulations and Deep Learning analysis

To understand the lithium ion transport dissolved in the ionic liquids, we conducted molecular dynamics (MD) simulations consist of [PYR₁₄][TFSI] with added [Li][TFSI] salt. It is well known that the transport of lithium ions involves local shell exchanges of TFSI^- in the medium. We found shell exchanges of TFSI^- undergo power law which is a scale free, or non-Poissonian. We analyzed the non-Poissonian processes of lithium ions transport as two-state (soft and hard) model. We analytically calculated transition probability of two-state model, and this well explains the lifetime autocorrelation functions of Li-TFSI shells. To closely find the fractions of the two states, we utilized the graph neutral network and calculated the diffusion coefficients of each and total states. We can see that soft state mostly contribute to the transport of the lithium ions. Hence, it is necessary to incorporate the idea of increasing the fractions of the soft state to get better lithium ion transport properties under ionic liquids medium.