Microscopic origins of velocity and temperature dependent friction in unpassivated amorphous carbon

Interfacial friction is an extremely common form of interaction between materials in the field of material science. It is well known that hydrogenation or environmental passivation can influence the frictional performance of amorphous carbon coatings. However, the atomic-scale friction mechanisms at clean, unpassivated amorphous carbon interfaces remain insufficiently understood. The primary reason is the experimental observation and conventional friction theories have difficulty in describing the atomic-scale dynamics in non-periodic amorphous systems. Here, we employ a machine learning-based Deep Potential model and Quantum Thermal Bath method to achieve first-principles accuracy in studying the atomic-scale frictional behavior of amorphous carbon interfaces. The simulated friction is qualitatively similar to some experimental findings. By analyzing atomic-scale bond dynamics, we develop a phenomenological friction model that quantitatively describes the microscopic sliding friction at the unpassivated amorphous carbon interface. We further demonstrate that increasing temperature suppresses sliding friction in this system. And an ultralow-friction state can be approached at extreme temperatures (~2000 ℃). This work establishes a universal, structure-independent model bridging atomic simulations with friction model, offering insights into designing low-friction materials.