Title : Modeling phase crystallization in ge-rich ge-sb-te pcrams
Abstract:
Ge-rich Ge-Sb-Te (GGST) alloys are already integrated into the complementary-metal-oxide-semiconductor (CMOS) technology for industrial phase-change random access memory (PCRAM) production. This new type of memories allowing low-power system on-chip production and in-memory computing are expected to support emerging technologies for automotive and artificial intelligence applications. However, structure and chemical evolution of the material in the memory cell, during repeated crystallization/amorphyzation (i.e. SET/RESET) cycles, is complex and depend on Ge concentration in the GGST alloy. Memory cell electrical properties being dependent on atomic distribution in the GGST alloy, it is important to be able to simulate cell aging at the material level in order to design the best GGST PCRAMs [1-2].
Simulation of atomic redistribution in GGST memory cells is challenging since the Ge-Sb-Te ternary system involves semiconductor, metal and semi-metal elements of different structures, forming four different binary compounds as well as several possible ternary compounds. Furthermore, the simulation of SET/RESET cycles requires the three states amorphous, crystalline, and liquid to be simulated.
We propose to use a simplified Ge-Sb-Te system, able to reproduce the main phenomena occurring during atomic redistribution in GGST memory cells, using atomistic kinetic Monte Carlo (KMC) simulations based on the tight-biding Ising model and direct exchanges between first-neighbor lattice sites. Order-disorder transitions on a rigid fcc lattice are used to model amorphous-crystalline transitions at low temperature and solid-liquid transitions at high temperature. Simulations of Ge-rich GST film crystallization show good agreements with experiments according to crystallization kinetics and phase formation sequence versus Ge excess. Simulations of the cycling of a 50 nm-wide mushroom-type PCRAM cell show strong atomic redistribution, which suggests significant electrical property variations with cell ageing.