Large Scale molecular dynamics simulations of the crystallization of GeTe at the crystal-amorphous interface

We simulated the crystallization of amorphous GeTe in contact with the crystalline phase by means of molecular dynamics and a classical interatomic potential generated by fitting a huge ab-initio database with a Neural Network method. The potential has a close-to-ab-initio accuracy and allows simulating thousands of atoms for tens of ns at an affordable computational cost. The simulations provided an estimate of the speed of crystal growth of 5.5 m/s at 500 K in the most favorable conditions under which the density is dynamically adjusted during the crystallization process. The crystallization speed is reduced by a factor three by constraining the density of the amorphous phase to that of the crystal, mimicking the operation conditions of the memory cells.