Microscopic KMC Modeling of Oxide RRAMs

Abstract

We investigate the microscopic behavior of oxide-based resistive random-access memory (RRAM) cells by using a unique three-dimensional (3D) physical simulator. RRAMs are attracting substantial attention and are considered as the next generation of non-volatile memory technologies. We study the operation of RRAM cells based on silica-rich silicon (SiO$_x$) and hafnia (HfO$_x$), by employing the stochastic kinetic Monte Carlo (KMC) approach for charge transport. The simulator self-consistently couples electron and ionic transport to the heat generation and diffusion phenomena and includes carefully the physics and random nature of vacancy generation and recombination, and trapping mechanisms. It models the dynamics of conductive filaments (CFs) in the 3D real space and captures correctly resistance switching regimes, including the CF formation (electroforming), set and reset processes. We describe the stochastic simulation process used for device analysis. We discuss the differences in the origin of switching between silica and hafnia based devices, and address the influence of the initial vacancy population on resistance switching in silicon rich silica RRAMs. We also emphasis the need for using 3D models and including thermal self-consistency to capture accurately the memristive nature of device switching.