Mimicking excitatory and inhibitory synaptic behaviors with optical-absorption and electrical-switch heterostructures

Abstract

Inspired by human brain neural networks, two-terminal optoelectronic synaptic devices could process large amounts of information and promise to pave the way for overcoming the Von Neumann bottleneck in non-traditional computing. In this study, an optoelectronic synaptic device based on a tin oxide/cerium lead bromide quantum dot heterojunction with a two-terminal vertical structure was designed. Optical potentiation behaviors with the bias-free mode were mimicked due to carrier trapping effects in cerium lead bromide quantum dots, whereas electrical inhibition functions were achieved ascribed to the vacancy migration in tin oxide. Based on the current optoelectronic neuromorphic devices, for the simulation of classification for recognizing handwritten digits in the modified National Institute of Standards and Technology dataset, the artificial neural network achieved a high precision rate (97.94%), indicating its great potential for future application in neuromorphic computing.