Kinetic and Thermodynamic Transition Pathways of Silica by Machine Learning: Implication for Meteorite Impacts

Abstract

AbstractRocks falling to Earth from space may generate pressure and temperature approaching Earth's deep mantle, but such meteorite impact only persists for a very short period. Under these extreme conditions, kinetical factors largely control mineral phase transitions, in which the resultant phase may deviate from those at thermal equilibrium. Here, we focus on the phase transitions of silica during meteorite impact, and have elucidated multiple pathways from low‐coordinated silica to seifertite, the densest known silica found in meteorite samples. Utilizing a high‐dimensional neuro‐network potential specifically designed for silica, we exhaustively map the potential energy landscape through stochastic surface walking and uncover low‐barrier transition pathways toward seifertite at pressures far away from thermal equilibrium. These kinetic‐driven transitions are then characterized by first‐principles simulations, revealing narrow transition windows of pressure, with seifertite becoming more kinetically favored over stishovite at pressures in the vicinity of 10 and 25 GPa. Our results suggest that meteorite impacts should have reached such target pressures to overcome the thermodynamic limit of forming seifertite. The presence of seifertite may provide key information in constraining the relevant dynamic compression conditions.