Data retention statistics and modelling in HfO2 resistive switching memories

Resistive switching memory (RRAM) devices are gaining momentum as next generation memory technology for high density and embedded storage. To support technology development by the industry, the scaling and reliability of 1 kb RRAM must be understood and predicted. This work addresses data retention of resistance states in RRAM arrays based on HfO₂. We develop a new method for studying retention statistics and show that retention characteristics are dictated by the cell position in the resistance distribution. Resistance drift and its variation are studied as a function of program time and verify levels. The retention behaviors of set/reset states are finally explained in terms of filament size and mechanical stress for set and reset states, respectively.