Event-Driven Stochastic Compact Model for Resistive Switching Devices

Abstract

A stochastic compact model for resistive switching (RS) devices is presented. The motivation is twofold: first, introducing variability in a natural way, and second, accounting for the discrete jumps of conductance observed during set and reset transitions. The model is based on an event generation rate, and it is an 'on-the-fly' procedure because events are randomly generated as the simulation proceeds in time. For the generation of events, we assume a mixed nonhomogeneous Poisson process (NHPP). Before considering RS, we deal with the generation of successive breakdown (BD) events in metal-insulator-semiconductor structures. This confirms the validity of the approach by comparing it with experimental data in which discrete events are evident. To deal with RS, we transform a previous compact model into a stochastic model. Comparison with experiments in TiN/Ti/HfO2/W devices shows the validity of the approach. Current-voltage loops and potentiation-depression transients in pulsed experiments are captured with a single set of parameters. Moreover, the model is an adequate framework to deal with both cycle-to-cycle and device-to-device variability.