Self-consistent Automated Parameter Extraction of RRAM Physics-Based Compact Model

RRAM physics-based compact models are an essential tool for studying and designing novel RRAM-based circuits. Ideally, the compact model parameters should be easily calibrated on different RRAM technologies to enable the adoption of device-circuit co-optimization strategies. Still, most models in the literature lack a simple and self-consistent parameter extraction procedure. In this work, we devise a self-consistent automated parameter extraction procedure for the UniMORE RRAM compact model. The proposed procedure requires the execution of a few experiments that are commonly performed during device characterization. The procedure is validated on data collected experimentally on a TiN/Ti/HfOx/TiN RRAM technology, and on data from three RRAM technologies from the literature. The results show that the proposed automated parameter extraction procedure enables correctly calibrating the model parameters on all four considered RRAM technologies, enabling the simulation of the device characteristic in different operating conditions using a single set of parameters, and the implementation of device-circuit co-optimization strategies.