Mechanism of Gradual Reset in Resistive Switching of Metal Oxide Based RRAM

Investigation of the metal oxide-based resistive memory elements is of great interest because they can be used as artificial synapses in neuromorphic systems due to their ability to switch in analog mode and achieve multiple intermediate resistive states. The main challenge for their successful utilization is the current spread in the high resistive state in which multistage switching can be achieved. Different memory cells on the same die can have both abrupt RESET and smooth RESET, which must be learned to control. The goal of our work is to theoretically and experimentally investigate the mechanism of gradual switching from a low to a high resistive state. The paper presents a numerical calculation and experimental validation of the developed model explaining both the current spread in high resistive state and the unpredictable nature of switching within the framework of multifilament switching in hafnium and titanium oxide based resistive memory elements.