Reduced thermal conductivity of Si/Ge random layer nanowires: A comparative study against superlattice counterparts

Si/Ge nanowires are considered to be promising candidates as efficient thermoelectric materials due to their remarkable thermal insulating performance over bulk counterparts. In this study, thermal insulating performance of Si/Ge nanowires of randomly organized layer thickness, called random layer nanowires (RLNWs), is systematically investigated and compared against superlattice nanowires (SLNWs).The thermal conductivity (TC) of these structures is evaluated via non-equilibrium molecular dynamic simulations, and more informative insight is gained by normal mode decomposition and lattice dynamics calculations. It is demonstrated that the modes in random layer structures, in general, exhibit similar characteristics except the degree of localization to the corresponding superlattice counterparts by comparing the mode spectral energy densities, relaxation times, density of states, and participation ratios. For all physical and geometrical conditions investigated here, RLNWs show improved thermal insulating performance over corresponding SLNWs. More importantly, a RLNW of low mean layer thickness attains even lower TC than the corresponding Si/Ge alloy nanowire indicating the effectiveness of the random layer arrangements. An anomalous trend in TC of RLNWs (larger than the bulk counterpart) is observed at higher cross-sectional widths, and it is explained as a competing effect of phonon localization and wall scattering. Moreover, it is illustrated that the effectiveness of thermal insulating performance of RLNW depends on the fraction of coherent phonons that exist and how effectively those phonons are subject to localization under different cases. Published by AIP Publishing.