Impact and Quantization of Short-Term Relaxation effect in Analog RRAM

Analog RRAM is considered as a promising emerging device technology for the future storage and neuromorphic computing. Different from long-term retention degradation, the usually overlooked relaxation effect shows more significant impact on computing applications, which is manifested in the low energy efficiency of data mapping and in the high accuracy loss of application functions. In this work, we have statistically studied the relaxation effect on analog RRAM arrays. The random conductance fluctuation behaviors due to relaxation effect is captured and quantified with a newly developed fast characterization platform. The intermediate conductance state (ICS) shows more severe relaxation than the high and low conductance states (HCS and LCS). A multi-filament-formation-and-rupture model is established to explain the underlying mechanism of relaxation effect and the reason for differences between different conductance states.