Modeling disorder effect of the oxygen vacancy distribution in filamentary analog RRAM for neuromorphic computing

Abstract

Although bi-directional analog switching capability is crucial for neuromorphic computing application, it is still difficult to be realized in filamentary RRAM cells. This work investigates the physical mechanism of the abrupt switching to the analog switching transition using Kinetic Monte Carlo simulation method. A disorder-related model for oxygen vacancy distribution is proposed with an order parameter Oy to quantify the analog behaviors of different RRAM devices. The simulation results and model predictions are verified by experiments performed on 1kb RRAM array. It is suggested that disordered oxygen vacancy distribution is desired for analog switching. Optimization guideline for improving the analog performance of filamentary RRAM is provided.