A systematic molecular dynamics study of the thermal conductivities of polyetherimide/graphene nanocomposites using response surface methodology

Abstract In this work, an I‐optimal response surface model was used to systematically investigate the effects of graphene (Gr) content (Factor A; 0–10 wt%), temperature (Factor B; 0–200°C), and Gr layer structure (Factor C; monolayer versus five‐layer) on the thermal conductivities of PEI/Gr nanocomposites, which were determined using reverse non‐equilibrium molecular dynamics (RNEMD) simulation with the Müller‐Plathe algorithm. Based on a reduced quadratic model that was fit to the data, the effect of Factor A on thermal conductivity was found to be more pronounced for the PEI/Gr nanocomposite with the five‐layer Gr structure. Moreover, Factor B had expectedly the largest effect on thermal conductivity, followed by Factor C. However, these two factors were involved in significant interactions with Factor A. Based on numerical optimizations, the predicted thermal conductivities of the PEI/Gr nanocomposites varied from 0.057 (minimum) to 0.174 W m⁻¹ K⁻¹ (maximum). Overall, the maximum thermal conductivity of the PEI/Gr nanocomposite may be obtained at any given temperature in the range of 0–200°C by the addition of multi‐layer Gr (a cheaper alternative to monolayer Gr) at a content of 10 wt%. For example, the addition of 10 wt% five‐layer Gr to PEI at room temperature (25°) results yields an increase in its thermal conductivity of about 30%. Also, going from 0 to 100°C, an increase of about 76% is predicted for the thermal conductivity of the PEI/Gr nanocomposite containing 10 wt% five‐layer Gr. The results of this study shed light on the interactions between the three investigated factors. Highlights Thermal conductivities of polyetherimide/graphene (PEI/Gr) nanocomposites simulated. Graphene (Gr) content, temperature, and Gr layer structure affected the thermal behavior. Response surface methodology (RSM) aided in the identification of factorial interactions. Numerical optimization of the thermal conductivities enabled by a predictive model.