Thermal conductivity of polydisperse hexagonal BN/polyimide composites: Iterative EMT model and machine learning based on first principles investigation

Abstract

Demand for thermal management materials (TMMs) with efficient in-plane heat dissipation has grown with the advancement of intelligent wireless communication equipment. Herein, polydisperse hexagonal boron nitride (ae-BN) in the range of micrometers to nanometers via aqueous-assisted exfoliation. First principles investigation revealed that ae-BN possess high intrinsic thermal conductivity. A series of ae-BN/PI composites were then fabricated through facile methods: vacuum-filtration and hot-pressing. The ae-BN/PI composites with 30 vol% ae-BN loading exhibited superior in-plane thermal conductivity (6.57 W/(m·K) compared to pristine h-BN/PI composite (3.92 W/(m·K)). SEM images and structural modeling of ae-BN/PI composites revealed that thermal conduction pathways constructed in the composites continuously increased with ae-BN content, attributing to an increased contact probability in composites with higher content of ae-BN. Reduction in thermal boundary resistance of ae-BN/PI composites was proved by our iterative EMT model. In-plane thermal conductivity of ae-BN/PI composites with different fillers’ contents at variable temperatures were predicted by machine learning technique, viz. artificial neural network (ANN) model. In brief, ae-BN/PI composites with high thermal conductivity, electrical insulation, thermal stability, and mechanical strength were successfully fabricated. The heat conduction mechanism of ae-BN/PI composites was investigated, serving as an important piece of puzzle for the advancement in TMMs of the advanced electronic devices.