Ab initio and empirical modeling of lithium atoms penetration into silicon

[Display omitted] •The kinetics of lithium penetration into Si(100) was studied on atomic level.•Li is unlikely to penetrate into silicon at dilute concentrations.•It is favorable for Li to go below Si(100) surface covered by not less than 2ML of Li.•Empirical potential was developed to reproduce DFT forces in MD simulations. A process of lithium atoms penetration into silicon (100) subsurface layers was investigated with the help of DFT method. It was shown that, while the concentration of lithium adatoms on reconstructed (100) silicon surface is low, the bonding energy of lithium atoms in the subsurface layers is smaller than the bonding energy on the surface, so lithium atoms are unlikely to migrate into the crystal. When the (100) silicon surface is covered by 2 layers of lithium, migration into the subsurface layer becomes favorable. In addition to this, the reconstruction of the surface changes to the form with symmetric dimers as the concentration increases. Thus, all possible lithium migration paths become energy-wise equal, so the rate of lithium atom transfer into silicon crystal rises. In addition to the ab initio calculations, an ad-hoc empirical interatomic potential was developed and the kinetics of lithium diffusion into silicon were studied. It was shown that lithium penetration proceeds in a layer-by-layer way with a sharp border between undoped and lithiated silicon. This is accounted for the fact that, once a tetrahedral interstice is occupied by a lithium atom, the migration barriers between the adjacent interstices become lower and the rate of diffusion increases.