Simulations of Volatile Memristors Based on Vanadium Dioxide—Linking Material Properties to the Dynamics of Neuromorphic Circuits

Volatile memristors have been gathering a lot of attention as devices suitable for technological applications, such as neuromorphic devices and random number generators, to name a few. Their operating mechanism is a sharp and volatile resistivity variation, switching between high and low resistance states. A thorough understanding and modeling of this resistive switching are essential to fully exploiting volatile memristors. In this respect, multi-physics simulations are crucial. This is especially the case when dealing with devices fabricated with innovative materials, where both material growth and device processing are still to be completely controlled. Then, simulations allow to investigate the interplay between different material properties and/or device features that would be otherwise obscured due to the inherent variability of devices.<br/>In this work, we show results of 3D technology computer-aided design (TCAD) electrothermal simulations of volatile memristors based on vanadium dioxide (VO₂). We avail of a dedicated TCAD approach [1], [2], developed to model the resistive switching of VO₂ volatile memristors as induced by the Joule effect. We combine the electrothermal device simulations with SPICE circuit simulations to simulate the dynamics of VO₂ oscillators. VO₂ oscillators have recently attracted a large interest [3] as the main elements for realization of systems of coupled oscillators, or oscillatory neural networks (ONNs), which are circuits whose dynamics resemble that of networks of neurons in the human brain [4]. By our mixed-mode TCAD-SPICE approach, we are able to link realistically VO₂ material properties to the behavior of the oscillator and the dynamics of simple ONN systems, thus being able to simulate realistically the working of ONN circuits as analogue computing engines. Our findings shed light on the coupled thermal and electrical behavior of VO₂ oscillator as well as the behavior of networks of VO₂ oscillators, providing some physical insights into the successful implementation of ONN technology.<br/><b>Acknowledgments.</b> Authors wish to thank Dr. S. Karg, IBM Research Europe, Zurich, Switzerland, for providing the experimental data used for calibrating the TCAD model and the valuable discussions about the experimental devices. Authors also wish to thank Dr. A. Nejim and Dr. A. Plews, of Silvaco Europe Ltd., Cambridgeshire, United Kingdom, for providing the customized version of PCM model [1] used to simulate the VO₂ material as well as for the useful discussions about the TCAD and mixed-mode simulations.<br/>[1] ”Victory Device User Manual”, version 1.19.1.C, Silvaco Inc<br/>[2] S. Carapezzi et al., IEEE J. Emerg. Sel. Topics Circuits Syst. 11, 4, 2021. DOI: 10.1109/JETCAS.2021.3128756.<br/>[3] E. Corti et al., Front. Neurosci., vol. 15, 2021. DOI: 10.3389/fnins.2021.628254<br/>[4] A. Todri-Sanial et al., IEEE Trans. Neural Netw. Learn. Syst., 2021. DOI: 10.1109/TNNLS.2021.3107771