The Disappearing Quartz-Coesite Path: the Phase Transition Mechanism of Silicon Dioxide from Machine Learning Simulations

This study systematically investigates the structural phase transition mechanisms of silicon dioxide under high pressure by using the high-dimensional neural network potential model combined with the stochastic surface walking algorithm. First, a global potential energy surface of quartz, coesite, stishovite, and amorphous states was constructed, and the thermodynamic phase diagram reveals the thermodynamic stability advantage of stishovite in high-pressure regions. Further analysis demonstrated that the energy barrier for the quartz-to-stishovite transition path shows a significant decrease under high pressure, which exhibits strong kinetic feasibility; while for the coesite-to-stishovite pathway, it follows a single transition state mechanism and the energy barrier displays a slightly increase under high pressure. Regarding the amorphization transition, the low symmetry structure group plays a key role in the high-pressure amorphization of quartz based on sampling and identification, and the “short-range order−middle range-disorder−topological order” structure was unveiled as a defining characteristic of the amorphous state. Notably, we did not observe effective quartz-coesite transition path during the study and further confirmed that the advantage of kinetic dynamics in the amorphization transition inhibits this transformation pathway, revealing the origin of the absence of the quartz-coesite transition. This work systematically explores the mechanisms of crystalline and amorphous phase transitions in silicon dioxide under high pressure and provides theoretical foundations and methodological paradigms for high-pressure simulation studies of complex oxides.