The interaction potential of a $\mathrm{Li}$ atom with a graphene layer is calculated using the local density approximation to the density functional theory. Two configurations corresponding to different separations between $\mathrm{Li}$ in neighboring supercells were considered to determine the effect of $\mathrm{Li}\text{−}\mathrm{Li}$ interaction on the binding of $\mathrm{Li}$ to graphene. The equilibrium position of $\mathrm{Li}$ is not affected by the $\mathrm{Li}\text{−}\mathrm{Li}$ interaction and remains the same in both cases. It is equal to $3.1\mathrm{a.u.}$ above graphene with the $\mathrm{Li}$ at the center of a hexagonal ring formed by the carbon atoms. However, the binding energies differ substantially in the two configurations. The binding energy of $\mathrm{Li}$ is $0.934\mathrm{eV}$ in configuration A when the $\mathrm{Li}\text{−}\mathrm{Li}$ separation in adjacent supercells is $9.22\mathrm{a.u.}$ The binding energy is $1.598\mathrm{eV}$ in configuration B corresponding to a separation $18.45\mathrm{a.u.}$ between adjacent $\mathrm{Li}$ atoms. There is substantial charge transfer from both lithium and carbon atoms (including those that do not surround the lithium) to a region between the $\mathrm{Li}$ and graphene at the minimum energy configuration. The interaction potential for both configurations can be fitted to a sum of a screened Yukawa potential and a linear superposition of power-law functions of type $r^{-n}$, where $n$ is an integer. The density functional theory underestimates the attractive contribution of dispersion forces for large separations. The attractive interaction potential calculated for $\mathrm{Li}$ positions much greater than the equilibrium distance from graphene may therefore need to be corrected.

• Received 6 February 2004

DOI:https://doi.org/10.1103/PhysRevB.70.125422