Multi-ion controllable metal halide ionic structure for selective short- and long-term memorable synaptic devices

Abstract

Development of memristors based on artificial synapses is a significant advancement in modeling biological synapses that utilize short-term plasticity (STP) and long-term plasticity (LTP). Herein, we present a novel multimode mechanism memristor based on Cs2AgI3 with 1D [AgI4]3− tetrahedral nanowire that allows for the simultaneous manipulation of temporally expanded plasticity through controllable multi-ions. The ion-transport mechanisms are controlled by the energy barriers of vacancy transportation and metal-ion injection, depending on the bias voltage. Finally, the conductance states of the multimode memristors are determined through the distinct rate-determining steps of the dominant ion migrations. To realize a multimode conductance state, the LTP is further expanded to short-lived (SL) and long-lasting (LL) plasticity by adjusting the length of the memory timescale through bias voltage. SL-LTP maintains a relatively low value (≈0.4 mS) for ∼750 ms, whereas the conductance of LL-LTP gradually decreases from ≈1.5 mS to ≈0.5 mS over 23 h. The accuracies achieved for each respective mode simulation in the perception rate of an artificial neural network based on multimode memristors are ≈91% and ≈88%. Therefore, the Cs2AgI3 memristor is capable of a new breakthrough in the development of next-generation neuromorphic computing as a multimode perceptron by simultaneously utilizing temporal plasticity.