An improved multifilamentary conduction model for multiphysics analysis of reset process in resistive random access memory

Abstract

An improved multifilamentary conduction model for the reset process in resistive random access memory (RRAM) is constructed by considering the stochastic distribution of oxygen vacancies (Vo). In this context, conduction filaments (CFs) have different Vo densities and diffusion barriers. Fully coupled multiphysics simulations of RRAM with three CFs are performed using the time domain finite difference method to self-consistently solve the current transport, heat conduction, and Vo transport equations. The simulated I-V characteristics agree well with experiment. Since the three CFs have different thermal diffusion barriers, as in previous studies, they rupture at different applied voltages. Evolution of each CF in the reset process is investigated: from when its electrical conductivity is linearly dependent to when it is exponentially dependent on Vo density.