Key points are not available for this paper at this time.
Abstract Non‐volatile memory‐based neuromorphic systems are compelling to build energy‐efficient computer hardware tailored for artificial intelligence. Solution‐processed earth‐abundant metal oxide memristor capable of adaptive learning holds the potential of a cost‐effective neuromorphic computing system that withstands harsh environmental conditions. Here, multilevel resistive switching at TiO 2 /NiFe 2 O 4 heterointerface prepared by spin‐coating is reported that shows a maximum ON−OFF ratio of ≈10 3 , endurance of up to 10 5 iterations, and retention stability of over 10 5 s with narrower SET‐RESET voltage scattering. The heterostructure device shows typical synaptic behaviors like long‐term potentiation/depression, short‐term/long‐term memory, paired‐pulse facilitation, and spike time‐dependent plasticity under electrical stimuli, highlighting its potential applications in associative learning and Morse code detection. This device mimics hypoxia in the biological nervous system by limiting its memristor functionalities in the absence of an oxygen environment and recovering normal operation with the resumption of oxygen. Moreover, the all‐oxide architecture of our device enables maintaining its functionalities after dripping water, followed by drying at 100 °C and retaining synaptic behavior of gradual potentiation up to 150 °C. Overall, these results widen the applicability of solution‐processed oxide‐heterostructure replicating the biological nervous system.
Kaith et al. (Fri,) studied this question.