Compact Modeling of I-V Characteristics, Temperature Dependency, Variations, and Noise of Integrated, Reproducible Metal-Oxide Memristors

Abstract

We present a comprehensive phenomenological model for the crossbar integrated metal-oxide continuous-state memristors. The model consists of static and dynamic equations, which are obtained by fitting a large amount of experimental data, collected on several hundred devices. The static equation describes the device current, at non-disturbing voltages, as a sum of three components, representing the average behavior and its device-to-device and temporal variations. All three components are expressed as functions of the applied voltage, ambient temperature, and the current memory state. The dynamic equation models the change in the memory state as a function of the applied voltage stress and the current memory state, and is also expressed as a sum of the average and the device-to-device variation terms. Both equations are explicit, computationally inexpensive, and suitable for SPICE modeling. At the same time, the model has good predictive power, which is supported by the validation results. The presented model is useful for realistic simulations of various mixed-signal computing circuits, such as image compression and image classification applications, discussed in this paper.