Reactive potential for the study of phase-change materials: GeTe

Abstract

We developed a classical potential to model phase-change materials based on the binary chalcogenide alloy of GeTe that are currently exploited for memory applications. Our potential is based on the recently proposed extension of the Tersoff potential plus additional terms to better reproduce the structure of the amorphous and the crystalline phases of GeTe. The parameters defining the potential reported in this work were fitted to reproduce the energies and forces of a database of reference structures obtained via density-functional theory molecular-dynamics simulations. This paper reports on the method used to construct the potential and on its validation against first-principles calculations either available in literature or part of this work. We found that the structural properties of amorphous GeTe were well reproduced. The advantage of the current implementation toward more flexible neural network-based methods is that most of the parameters can be reconnected to physical properties. Moreover, the relatively small number of parameters results in a simple implementation and facilitates the introductions of further interactions among additional species.