Machine learning-driven molecular dynamics decodes thermal tuning in graphene foam composites

Abstract

Graphene foam (GF), synthesized via Chemical Vapor Deposition (CVD), has been proven to be the ideal bulk porous material. The addition of poly(dimethylsiloxane) (PDMS) within the porous structure enables enhancement of mechanical strength and alteration of heat transfer behavior. This study focuses on the thermodynamic behavior of GF/PDMS composites during deformation, and employs stochastic modeling and neuroevolution potential (NEP) for complex material modeling with precise prediction of microscopic mechanisms governing thermal property variations. The results demonstrate that the composite with a 5% doping rate of PDMS achieves the optimal mechanical performance and shows a 7.13-fold modulation in thermal resistance during the deformation from 40% stretching to 50% compression. Findings indicate PDMS fortifies structural stability while enabling dynamic thermal conductivity modulation in GF. This research provides critical insights into the micro-mechanisms of GF/PDMS composites and offers a theoretical foundation for applications in dynamic thermal management and self-powered sensor networks.