Atomic-Scale Insights Into Graphene/Fullerene Tribological Mechanisms and Machine Learning Prediction of Properties

Abstract

Abstract Graphene/fullerene carbon–based nanoparticles exhibit excellent tribological properties in solid–liquid two-phase lubrication systems. However, the tribological mechanism still lacks profound insights into dynamic friction processes at the atomic scale. In this paper, the friction reduction and anti-wear mechanism of graphene/fullerene nanoparticles and the synergistic lubrication effect of the binary additive system were investigated by molecular dynamics simulations and tribological experiments. The friction performance was predicted based on six machine learning algorithms. The results indicated that in fluid lubrication, graphene promoted “liquid–liquid” interlayer sliding, whereas fullerene facilitated “solid–liquid” interface sliding, resulting in a decrease or increase in friction force. Under boundary lubrication, graphene/fullerene nanoparticles were adsorbed and anchored at the metal interface to form a physical protective film, which improved the bearing capacity of the lubricating oil film, transformed the direct contact between asperities into interlayer sliding of graphene and roll–slide polishing, filling, and repairing of fullerene, thus improving the frictional wear of the lubrication system as well as the friction temperature rise and stress concentration of the asperities. Furthermore, six machine learning algorithms showed low error and high precision, and the coefficient of determination was greater than 0.9, indicating that all models had good prediction and generalization capabilities, fully demonstrating the feasibility of combining molecular simulation and machine learning applications in the field of tribology.