Analysis of RTN and cycling variability in HfO2 RRAM devices in LRS

In this work, we present a thorough statistical characterization of cycling variability and Random Telegraph Noise (RTN) in HfO2-based Resistive Random Access Memory (RRAM) cells in Low Resistive State (LRS). Devices are tested under a variety of operational conditions. A Factorial Hidden Markov Model (FHMM) analysis is exploited to extrapolate the properties of the traps causing multi-level RTN in LRS. The trapping and de-trapping of charge carriers into/out of defects located in the proximity of the conductive filament results in a shielding effect on a portion of the conductive filament, leading to the observed RTN current fluctuations. The variations of the current observed at subsequent set/reset cycles are instead attributed to the stochastic variations in the filament due to oxidation/reduction processes during reset and set operations, respectively. The statistical characterization of RTN and cycling variability does not show correlation between these phenomena.