A cell-based clustering model for the reset statistics in RRAM

摘要

In this letter, the experimental reset voltage and current statistics in a Cu/HfO_2/Pt RRAM device are found to deviate from the Weibull model in the high percentiles. A clustering model is developed for the reset statistics based on the cell-based model. The relationship between the parameters (Weibull slope, scale factor, and clustering factor) of the clustering model and on-state resistance is established. The modeling results are in good agreement with the experimental data of reset voltage and reset current statistics. Our model explains well why the tail bits of experimental data appear in the high percentiles. The implicit meaning of the scale factor of the clustering model is explored, which represents the threshold point that defects emerge to cluster.