Neural network molecular dynamics simulation on friction-induced chemical reactions of Si3N4 in water and ethylene glycol environments

Silicon nitride (Si3N4) exhibits low friction in aqueous environments due to a tribolayer that is formed through tribochemical reactions. However, the low friction state is not maintained in high contact pressure conditions, where surface-surface contact is dominant at the sliding interface, i.e. the load carrying capacity is low. Recently, it was reported that an ethylene glycol (EG) additive improves the load carrying capacity of Si3N4 in aqueous environments, though their mechanism is still in debate. In this study, we performed friction simulations to analyze the tribochemical reactions of water and an EG additive using a neural network molecular dynamics method which enables large-scale simulation with high accuracy comparable with ab initio molecular dynamics calculations. We found that tribochemical reactions of water produce SiO2 particles. On the other hand, tribochemical reactions of EG produce compounds which consist of carbon, nitrogen, and hydrogen atoms on the Si3N4 surface and the Si3N4 surface is covered by the compounds. Based on this finding, we propose that the compounds covering the Si3N4 surface can improve its load carrying capacity.