Innovative SPICE model for VO₂ threshold-switching devices: Bridging insulator-metal transitions and thermal-electrical feedback for neuromorphic applications

This paper presents a novel SPICE-compatible compact model for VO₂-based Insulator-Metal Transition (IMT) devices, skilfully integrating thermal dynamics with electrical behaviour through advanced circuit modelling techniques. The thermal characteristics of the VO₂ layer are represented utilizing an RC Cauer model, which effectively captures the processes of heat generation and dissipation. A nonlinear relationship between the temperature derived from the thermal circuit and the VO₂ resistance is modelled using a sigmoid function combined with exponential terms. This approach accurately depicts the phase transition from a high-resistance insulating state to a low-resistance metallic state, along with the associated hysteresis effects. An artificial neuron based on IMT has been developed and simulated using this compact model, which faithfully represents both the electrical and thermal dynamics of IMT behaviour. Implemented and validated in the LTspice environment, this model incorporates a realistic RC thermal network alongside a temperature-dependent resistance formulation, reflecting the physical switching characteristics of VO₂ devices. Validation against experimental data confirms the model's accuracy and reliability. Furthermore, several biologically plausible spiking behaviors previously observed in studies of single VO₂ IMT-based neurons have been successfully replicated and validated with this model. The simulation results align with theoretical predictions and are consistent with earlier implementations of IMT-based neurons. In conclusion, this work offers circuit designers a robust and experimentally validated platform for exploring and simulating IMT-based neuromorphic systems, thereby paving the way for further developing and integrating these innovative devices into next-generation computing architectures.