A generic compact and stochastic model for non-filamentary analog resistive switching devices

This article introduces a new 1D distributed compact model for non-filamentary memristive devices that combines the strengths of 1D concentrated and 3D distributed models into one model. The proposed model is highly appropriate for conducting SPICE-level crossbar simulations as it is both robust and less computationally demanding and can be modified to suit any material stack that exhibits non-filamentary switching. Using the cloud-in-a-cell method, a single model can simulate ion/vacancy transport, trapping/de-trapping, or both mechanisms that cause stochastic resistive switching in these devices. The flexibility of the model is demonstrated using niobium oxide and bismuth ferrite oxide-based analog memristive devices. The current-voltage characteristics of both devices are calculated based on different conduction mechanisms, such as direct trap-assisted tunneling, ohmic transport, or thermionic emission. The voltage drops across different regions of both devices and charge distribution across them are also presented to show the difference in switching mechanisms from a modeling point of view. The synaptic plasticity measurements of both devices in terms of potentiation and depression dynamics are studied using the proposed model.