Volatility characterization for RRAM devices

Emerging technologies, such as resistive random access memory (RRAM), are being actively researched for its potential applications in developing new technologies inspired by brainlike neuromorphic computing. However, developing automated characterization algorithms for the metastable resistive state (RS) transitions, i.e., volatility in the myriad RRAM configurations fabricated for achieving desired performance remains a major bottleneck. Here, we propose a novel algorithm for extracting the volatility parameters for the utilized metal-oxide TiO_x-based devices. The module applies an appropriate stimulus and then estimates the RS changes of the device under test (DUT) using the standard two mean t-test method over a fixed interval of time. The module halts when an equilibrium state has been detected and is followed by a retention condition that checks the DUT for the achieved equilibrium state. The output of the proposed module determines the time the t-test lasted for and the voltage range under which the DUT can be safely operated in purely volatile region. In conclusion, the proposed characterization protocol ensures a methodical process development for automated characterization of RRAM arrays for operating them in the volatile region.