Rapid Prediction of RRAM RESET-State Disturb by Ramped Voltage Stress

Abstract

This letter proposes a novel technique for predicting with high confidence the disturbance of the resistive-switching random access memory (RRAM) RESET state based on ramped voltage stress. The technique yields statistical distributions and voltage acceleration parameters equivalent to those of a conventional constant voltage method. Several ramp rates and acceleration models were validated for the accuracy regarding conversion between the two methods. The proposed method not only reduces the time and cost of reliability analysis but also provides a quantitative link between disturbance properties and the widely available RRAM data measured by a linear voltage ramp. Additionally, the non-Poisson area scaling supports the localized filament model.