Implementation of sub-filamentary network-based variability model for Ta2O5/TaOx RRAM

This paper presents an analytical model of the cycle to cycle (c2c) variability in Pt/Ta2O5/TaOx/Pt RRAM device. The model is developed by considering the dynamic randomness in physical parameters related to the conduction filament of the device by demonstrating the formation of a sub filamentary network. The reasons behind the variability of the device characteristics are categorized into three possible aspects such as the formation of unstable sub filaments with varying lengths, formation of conduction filament with different radius in multiple cycles, and the barrier height modulations due to change in the number of oxygen vacancies(Vos) present at the interface and bulk. The major observations from the model are; the multiple resistive levels seen in the RRAM device is due to the formation of multiple stable/unstable filaments, the radius of the conduction filament has an impact on the RESET voltage, and finally, the modulations in tunneling barrier height cause randomness in high resistance state (HRS) of the device.