Impacts of potential models on calculating the thermal conductivity of graphene using non-equilibrium molecular dynamics simulations

This work employs non-equilibrium molecular dynamics (NEMD) simulations to examine the applicability of four kinds of interatomic potential models: the Tersoff, the REBO, the opt-Tersoff and the AIREBO, which are widely used to model the thermal transport in single- and multi-layer graphene, as well as graphite crystallites. Thermal conductivities of ∼17 × 5 nm2 and ∼50 × 5 nm2 graphene are calculated in the temperature range of 200∼500 K with the four potentials and quantum correction is applied due to an extremely high Debye temperature of about 2100 K for graphene. The predicted thermal conductivities are compared with experimental data and phonon spectrum functions are calculated to quantify the degree of phonon scattering. The results show that two original potentials, the Tersoff and the REBO, as well as the AIREBO significantly underestimate thermal conductivities of single-layer graphene but they can qualitatively describe the trend of thermal conductivities with temperature. The opt-Tersoff is found to be the most suitable potential for modeling the thermal conductivity of both single- and multi-layer graphene because it predicts a larger frequency range and a larger frequency value for the high frequency peak, while appropriately capturing phonon scattering in thicker multi-layer graphene when Lennard-Jones term is added into the opt-Tersoff to describe interlayer atomic interactions.