Computational study of impact of composition, density, and temperature on thermal conductivity of amorphous silicon boron nitride

We study thermal conductivity (κ) of amorphous silicon boron nitride (a‐ S i BN ) for different compositions and densities as a function of temperature using density functional theory ( DFT ) calculations and equilibrium molecular dynamic ( MD ) simulations. Our library of amorphous structures consists of network models comprising 100‐200 atoms and large‐scale models with up to 57 000 atoms generated using the empirical Marian‐Gastreich two‐body potential. Crystalline structures within the Si 3 N 4 ‐ BN system are considered as well. We use 2 distinct approaches to compute thermal conductivity of a‐Si BN . To estimate κ in the high‐temperature limit we feed Clarke's phenomenological model with elasticity data obtained by DFT calculations. We further perform equilibrium MD simulations and apply the Green‐Kubo method. This approach shows decrease of κ with increasing temperature and provides results at high temperatures that agree with results derived within Clarke's model. We find that κ of a‐Si BN depends on composition and increases as the BN content in the structure increases. The effect is pronounced at low temperature but almost vanishes at high temperature. Furthermore, thermal conductivity depends on density and porosity, with a linear relation between κ and density.