Size Scaling in Phase Change Memory Cells: From Traditional to Emerging Device Structures

Abstract

1. INTRODUCTION AND METHODOLOGY Phase change memory (PCM) based on the reversible phase-transition of chalcogenides (such as Ge2Sb2Te5 (GST)) between a low-resistance crystalline state and high-resistance amorphous state is one of the leading candidates of emerging non-volatile solid-state memories [1]. Scaling is one of the most important aspects for PCM development as it leads to enhanced storage density, reduction in power consumption and improvement in switching speeds [2]. To demonstrate the excellent scalability of PCRAM, switching capability in the sub-10nm region [3-5], programming currents less than 10μA [4], switching speeds in picoseconds [6], and storage densities in Tb/in2 using scanning probe recording and thermal recording [7-8] have all been reported. In this manuscript we combine electro-thermal simulations with the Gillespie Cellular Automata (GCA) phase switching approach to simulate and predict the scaling behaviour (down to sub-10nm dimensions) of three GST-based device structures; (1) mushroom-type PCM cells, (2) trilayer patterned PCM cells, and (3) spherical phase change nanoclusters. The GCA approach is a sophisticated stochastic simulator capable of spatio-temporal modeling in PCM devices, and has previously been described in detail in [9]. This approach is potentially capable of spanning the length scales between atomistic modeling and bulk scale methods such as the JMAK or the classical nucleation and growth methods. Electrical switching is performed by applying trapezoidal Reset and Set pulses of various amplitudes and durations in a test bench consisting of an electrical pulse source, a series load resistance of 10kΩ, and the phase change memory cell itself. [...]