An optimized smearing scheming for first Brillouin zone sampling and its application on thermal conductivity prediction of graphite

We propose an optimized scheme to determine the smearing parameter in the Gaussian function that is used to replace the Dirac δ function in the first Brillouin zone sampling. The broadening width is derived by analyzing the difference of the results from the phase-space method and Gaussian broadening method. As a demonstration, using the present approach,we investigate the phonon transport in a typical layered material, graphite. Our scheme is benchmarked by comparing with other zone sampling methods. Both the three-phonon phonon scattering rates and thermal conductivity are consistent with the prediction from the widely used tetrahedron method and adaptive broadening method. The computational efficiency of our scheme is more than one order of magnitude higher than the two other methods. Furthermore, the effect of fourphonon scattering in phonon transport in graphite is also investigated. It is found that four-phonon scattering reduces the through-plane thermal conductivity by 10%. Our methods could be a reference for the prediction of thermal conductivity of anisotropic material in the future.