Four-phonon scattering reduces intrinsic thermal conductivity of graphene and the contributions from flexural phonons

Abstract

We have developed a formalism of the exact solution to linearized phonon Boltzmann transport equation (BTE) for thermal conductivity calculation including three- and four-phonon scattering. We find strikingly high four-phonon scattering rates in single-layer graphene (SLG) based on the optimized Tersoff potential. The reflection symmetry in graphene, which forbids the three-ZA (out-of-plane acoustic) scattering, allows the four-ZA processes $\mathrm{ZA}+\mathrm{ZA}\ensuremath{\rightleftharpoons}\mathrm{ZA}+\mathrm{ZA}$ and $\mathrm{ZA}\ensuremath{\rightleftharpoons}\mathrm{ZA}+\mathrm{ZA}$ + ZA. As a result, the large phonon population of the low-energy ZA branch originated from the quadratic phonon dispersion leads to high four-phonon scattering rates, even much higher than the three-phonon scattering rates at room temperature. These four-phonon processes are dominated by the normal processes, which lead to a failure of the single mode relaxation time approximation. Therefore, we have solved the exact phonon BTE using an iterative scheme and then calculated the length- and temperature-dependent thermal conductivities. We find that the predicted thermal conductivity of SLG is lower than the previously predicted value from the three-phonon scattering only. The relative contribution of the ZA branch is reduced from 70% to 30% when four-phonon scattering is included. Furthermore, we have demonstrated that the four-phonon scattering in multilayer graphene and graphite is not strong due to the ZA splitting by interlayer van der Waals interaction. We also demonstrate that the five-phonon process in SLG is not strong due to the restriction of reflection symmetry.