Knowledge of transformation mechanisms and related free-energy barriers in pressure-induced solid-solid first-order phase transitions is of fundamental interest but also of major importance for technological applications (e.g., synthesis of the diamond preventing titanium alloys from forming the brittle ω phase, etc.).
In simulations of smaller systems with < 10,000 atoms these transitions typically proceed via a collective mechanism, which is, however, not representative of the true one in the thermodynamic limit. It is known that the true mechanism of a first-order phase transition involves a nucleation of the emerging phase and therefore realistic simulation must reach beyond the collective regime, requiring typically > 100,000 atoms.
Due to the recent advent of machine-learning potentials and the development of enhanced sampling methods such simulations are within reach now.
Besides studying the transitions in idealized conditions under hydrostatic pressure, it is now possible to get closer to realistic experimental conditions, allowing simulations of heterogeneous nucleation under the presence of defects such as dislocations or grain boundaries as well as non-hydrostatic stress. Therefore, the simulations can provide valuable insight into how the structural defects and non-hydrostatic stress affect the formation of novel phases, including the metastable ones.
This goal is beyond reach of the hitherto commonly used methods.
Faculty of Mathematics, Physics and Informatics, Comenius University in Bratislava, Slovakia;
SISSA, Trieste, Italy.