Development of a semi-empirical potential suitable for molecular dynamics simulation of vitrification in Cu-Zr alloys.

The fast increase in available computation power allowed us to decrease the cooling rate in molecular dynamics (MD) simulation of vitrification by several orders of magnitude. While the reliability of the MD simulation should obviously benefit from this increase in the computational power, in some cases, it led to unexpected results. In particular, Ryltsev et al. [J. Chem. Phys. 149, 164502 (2018)] found that the most popular potentials for the Cu-Zr and Cu-Zr-Al alloys from Mendelev et al. [Philos. Mag. 89, 967 (2009)] and Cheng et al. [Phys. Rev. Lett. 102, 245501 (2009)] do not actually describe good glass forming systems but in contradiction with experiment predict rather fast crystallization of the Cu64.5Zr35.5 alloy which is the well-known example of bulk metallic glasses. In this paper, we present a new Cu-Zr semiempirical potential suitable to simulate vitrification. No crystal nucleation was observed in MD simulation using this potential in the concentration range from 75% to 5% of Zr. Since the new potential leads to about the same liquid structure and viscosity as the Cu-Zr potential from Mendelev et al. [Philos. Mag. 89, 967 (2009)] which failed to describe the good glass formability, our study clearly shows that no reliable conclusions about the glass formability can be deduced based solely on the analysis of the liquid properties and a nucleation/crystal growth study should be performed to address this question.