Comparison between two numerical methods for the computation of thermal conductivities of particulate composites: FEM and GeoDict

The thermal interface material in direct contact with chip, termed as TIM1, is a kind of composite material with thermally conductive particulate fillers. To better understand the thermal conduction mechanism and promote the design of TIM1, numerical studies are very essential. In this work, we compare two numerical methods, finite element method (FEM) and GeoDict, to compute the effective thermal conductivity of the particulate composites, in which the interface thermal resistance is considered. Firstly, we employed the Monte-Carlo algorithm as well as energy minimization method to generate the representative volume elements (RVEs), and the particulate fillers are dispersed randomly in the RVEs. The interface thermal resistance is introduced in the numerical modeling by an imperfect interface between various components. The effective thermal conductivity is evaluated over RVEs by FEM and GeoDict, respectively, with the efficiency and accuracy of the two algorithms compared. The drawback of FEM is that its pre-treatment is quite difficult for particulate composites of high filler content because of the requirements of fine mesh, which also calls for a large demand of memory. In contrast, GeoDict solved the problem using fast Fourier transform algorithm based on pixels of the RVEs, showing high efficiency and can be extended to any kind of additives. Finally, we show that GeoDict could be an efficient tool for high-throughput screening of formulas of TIM1 for advanced electronic packaging.