Estimating thermal conductivity of amorphous silica nanoparticles and nanowires using molecular dynamics simulations

In several recent applications, including those aimed at developing thermal interface materials, nanoparticulate systems have been proposed to improve the effective behavior of the system. While nanoparticles by themselves may have low conductivities relative to larger particles owing to interfacial resistance, their use along with larger particles is believed to enhance the percolation threshold leading to better effective behavior overall. One critical challenge in using nanoparticulate systems is the lack of knowledge regarding their thermal conductivity. In this paper, the thermal conductivity of silica clusters (or nanoparticles) as well as nanowires is determined using molecular dynamics (MD) simulations. The equilibrium MD simulations of nanoparticles using Green-Kubo relations are demonstrated to be computationally very expensive and unsuitable for such nanoscaled systems. A nonequilibrium MD method adapted from the study of Muller-Plathe is shown to be faster and more accurate. The method is first demonstrated on bulk amorphous silica (using both cubic and orthorhombic simulation cells) and silica nanowires. The thermal conductivity values are compared to those reported in the literature. The mean thermal conductivity values for bulk silica and silica nanowire were estimated to be 1.2 W/mK and 1.435 W/mK, respectively. To model nanoparticles, the Muller-Plathe technique is adapted by dividing the cluster into concentric shells so as to capture the naturally radial mode of heat transfer. The mean thermal conductivity value of a 600-atom silica nanoparticle obtained using this approach was 0.589 W/mK. This value is similar to 50-60 % lower than those of bulk silica or silica nanowire.