Comprehensive Study of Lithium Adsorptionand Diffusion on Janus Mo/WXY (X, Y= S,Se, Te) using First Principles and MachineLearning Approaches

The structural asymmetry of two-dimensional (2D) Janus transition metal dichalcogenides (TMDs) produces internal dipole moments that result in interesting electronic properties. These properties differ from the regular (symmetric) TMD structures that the Janus structures are derived from. In this study, we, at first, examine adsorption and diffusion of a single Li atom on regular MX2and Janus MXY (M = Mo, W; XY =S, Se, Te) TMD structures at various concentrations using first principles calculations within density functional theory. To gain more physical insight and prepare for future investigations of regular TMD and Janus materials, we applied a supervised machine learning (ML) model that uses cluster-wise linear regression to predict the adsorption energies of Li on top of 2D TMDs. We developed a universal representation with few descriptors that take into account the intrinsic dipole moment and the electronic structure of regular and Janus 2D layers, the side where the adsorption takes place and the concentration dependence of adatom doping. This representation can easily be generalized to be used for other impurities and 2D layer combinations, including alloys as well. At last, we focus on analyzing these structures as possible anodes in battery applications. We conducted Li diffusion, open-circuit-voltage and storage capacity simulations. We report that Lithium atoms are found to easily migrate between transition metal (Mo, W) top sites for each considered case, and in these respects many of the examined Janus materials are comparable or superior to graphene and to regular TMDs. The results imply that theexamined Janus structures should perform well as electrodes in Li-ion batteries.