Efficient SPICE Modeling of Ta2O5-Based Bipolar RRAM Device Including Monte Carlo Simulation

In this paper, implementation of efficient SPICE modeling with the improved Verilog-A model based on the Stanford-PKU RRAM model (v2.0.0 beta) for Ta
2
O
5
-based bipolar Resistive Random Access Memory (RRAM) devices is presented. The improved model has been calibrated to the experimental data. The simulation results show that the improved model is robust and accurate in simulating resistive switching (RS) behaviors of Ta
2
O
5
-based bipolar RRAM devices. With Monte Carlo (MC) model being added in the model library for simulating device-to-device random variations due to process fluctuations, the calibrated model implemented in Cadence Virtuoso process design kit (PDK) has been applied in the large-scale memory array simulations by our design team.