Understanding the Influence of Initial Cluster Size Distribution On Crystallization Dynamics in The Ge2Sb2Te5 Phase‐Change Alloy

• Published in: physica status solidi (b) 2019-07, Vol.256 (7), p.n/a
• Main Authors: Aladool, Azzam, Aziz, Mustafa, Wright, David
• Format: Article
• Language: English
• Subjects: cluster size distribution; crystallization; master rate equation; phase‐change materials
• ISSN: 0370-1972; 1521-3951
• DOI: 10.1002/pssb.201800583

Phase‐change materials are recognized as major contenders to develop fast, low power, and small non‐volatile memories and data processors. Crystallization remains the limiting factor in deciding the operating speeds of phase‐change devices. It has been shown experimentally that pre‐annealing amorphous phase‐change materials leads to shorter crystallization times. This was attributed to the formation of crystalline nano‐clusters embedded in the amorphous phase that reduce the incubation time and enhance crystallization speeds, which was recently confirmed using enhanced transmission microscopy. The correlation between the initial size distribution of nano‐clusters on the crystallization dynamics is still however unclear which is crucial for increasing the speed of crystallization. This theoretical investigation simulates initial cluster size distributions following pre‐annealing and melt‐quenching in as‐deposited amorphous Ge2Sb2Te5 phase‐change materials using the physically realistic Master rate equation approach, and their influence on the crystallization dynamics. This approach is also extended to systematically study assumed Gaussian initial cluster densities in the sub‐nanometer to few tens of nanometers size range on the crystallization dynamics at different annealing rates. The simulations revealed enhancement of crystallization speeds with narrow initial distributions of small cluster sizes of few nanometers (for the same initial crystalline fraction) in agreement with enhanced microscopy observations. Phase‐change materials are major contenders to develop fast, low power, and small non‐volatile memories and data processors. Crystallization limits the operating speed of these devices. Simulations in this article using the Master rate equation revealed potential enhancement of crystallization speeds with narrow initial distribution of small cluster sizes of few nanometers in agreement with enhanced microscopy observations.