Mechanisms of liquid state evolution during phase change memory RESET melting

Matthew Lewis,Lucien N. Brush
DOI: https://doi.org/10.1016/j.commatsci.2024.113066
IF: 3.572
2024-05-27
Computational Materials Science
Abstract:Previous research on phase change memory (PCM) devices shows that correct modeling of the RESET melting process requires including the Gibbs–Thomson melting temperature for the solid–liquid interface. This additional physical effect results in substantially different predicted scaling properties and melting behavior than shown in previous research on conventional PCM devices scaled to nanometer dimensions. The computations also reveal that the thermodynamic Gibbs–Thomson effect gives rise to two steady state classes of liquid pools over a range of applied voltages. At any given voltage both states appear. This research characterizes these two classes of pool shapes by analyzing the linear stability of each pool shape class to elucidate the time-dependent, dynamic properties of the liquid pools. The stability analysis uses a novel numerical approach, required due to the large thermal and potential gradients within the system and the lack of geometrical symmetry of the steady base states. Results of the analysis show that at all given applied voltages, the smaller liquid pool shape class is linearly unstable and the larger pool shape class is linearly stable. The stability characteristics of the classes of liquid pools justify the designation of the nucleus state for the smaller liquid pool and the melting limit state for the larger liquid pool. The results imply that a system with a liquid pool larger than the nucleus state will eventually grow to the melting limit state which is a stable fixed point. Comparison with a spherically symmetric PCM model system reveals that in the isothermal limit the nucleus state exhibits precisely classical nucleation behavior found for a thermodynamically unstable critical nucleus. The theory used to identify the class of critical nuclei steady state shapes in this analysis is also novel, because it avoids the need to identify the extremum of a thermodynamic potential that is used in most treatments that attempt to identify the initial formation of new phases in inhomogeneous, evolving systems.
materials science, multidisciplinary
What problem does this paper attempt to address?