Graphene is now considered as a possible alternative anode to the widely used graphite for Lithium ion
batteries (LIBs) due to its surface to mass ratio.
Recently, it was demonstrated experimentally  that a battery whose anode is made
of graphene flakes can perform a charge-discharge cycles exceeding 80 with a high reversible capacity.
Remarkably, a stoichiometry corresponding to 1Li:2C
was observed by X-ray photoelectron spectroscopy (XPS) before the anode was inserted in the battery.
Understanding of this remarkable observation requires
an understanding of the nature of interaction of lithium with graphene. Here, we studied Lithium adsorption on
graphene and its fragments by means of
Density Functional Theory calculations. Several van der Waals functionals
were compared based on structures and energetics of Lithium metal and
Carbon in Diamond, Graphite and Graphene and rVV10 was found to be a well
balanced functional in all scenarios. The binding energies of Li adsorbed
on graphene and graphene nanoribbons were therefore computed with the
rVV10 functional and within PBE for comparison. Li binds in the hollow
site of graphene and the binding becomes stronger as the Li concentration
decreases. Another unexpected observation was the increase in the height of Li as the strength of the binding energy increases.
This can be understood on the basis of an electrostatic model.
In all cases results indicate an instability of Li-graphene system with
respect to Li bulk. The zigzag and armchair graphene nanoribbons, both with unsaturated and Hydrogen-saturated edges were examined.
In all cases, and especially for unsaturated edges, Lithium binding at the edge
is significantly increased. But in no case Li binding energy terrace
sites becomes favorable compared to the Bulk suggesting that presence of
edges in graphene nanoflakes cannot explain the experimentally reported
 extremely high Li capacity. Our calculations suggest that formation
of Li clusters on graphene is energetically favoured even for very small
cluster-size. This could account for the reported high capacity provided
entropic factors and/or spatial constraints prevent the Li dendrite
formation but appear at variance with the experimental evidence of no
cluster formation based on XPS measurement. In order to clarify this
issue we have undertaken a detailed study of C 1s and Li 1s core-level
shifts as a function of the chemical environment.
 Hassoun et. al., Nano letters. 2014 , 14, 4901-4906.