Reservoir Computing for Temporal Data Processing Using Resistive Switching Memory Devices Based on ITO Treated With O2 Plasma

Abstract

In this work, we investigate the effects of short-term memory (STM) on resistive switching (RS) memory resistive random access memory (RRAM) device, which is based on indium tin oxide (ITO) treated with O<inline-formula> <tex-math notation=LaTeX>$_{\text{2}}$</tex-math> </inline-formula> plasma. The STM characteristics of the fabricated Ag/ITO(O<inline-formula> <tex-math notation=LaTeX>$_{\text{2}}$</tex-math> </inline-formula> plasma)/TiN RRAM devices are caused by rupturing of the spontaneous Ag conductive filament due to the surface energy minimization effect. Importantly, this volatility characteristic can be controlled according to the compliance current (CC). The conductance change in the low resistance state over time is initially rapid and then converges to the initial high resistance state, and this relaxation phenomenon is well-fit by the stretched exponential (SE) model. Furthermore, the relaxation time (extracted through quantitative analysis) increases with increasing CC, indicating that our RRAM device can be controlled to accurately mimic the STM characteristics of biological synapses. To emulate the paired-pulse facilitation (PPF) of a biological synapse, we confirm the response of the device after implementing PPF according to the time interval. Finally, we experimentally demonstrate the feasibility for use in reservoir computing (RC) systems by implementing a binary 4-bit code ranging from 0000 to 1111 in Ag/ITO(O<inline-formula> <tex-math notation=LaTeX>$_{\text{2}}$</tex-math> </inline-formula> plasma)/TiN RRAM devices. IEEE