Optically Controlled P–CuxO-Based Artificial Synaptic Device for Neuromorphic Applications

Abstract

Memristor-based optoelectronic artificial synapses have a great potential to enhance the efficiency of future neuromorphic computing. Like neurons of the retina, they have the potential to enable real-time visual preprocessing. This highlights the growing importance of improving optoelectronic artificial synapses for next-generation neuromorphic computing and neuromorphic visual systems. These artificial synapses can enhance neuromorphic visual systems, extending their capabilities beyond visible light. This study introduces a P-type copper oxide-based optical memristor device that exhibits fundamental biosynaptic characteristics like long-term potentiation (LTP) and long-term depression (LTD), which can be tuned using optical stimuli. These LTP/LTD characteristics were used as weights in a single-layer perceptron neural network to classify the MNIST data set using an off-chip training algorithm. We also demonstrated light-induced short-term plasticity and optical paired-pulse facilitation, which are the two important characteristics of neurons of the human retina that help in image preprocessing. We also implemented Pavlovian conditioning on the device using a combination of electrical and optical stimuli. These results indicate the possibility of using this device as an optically controlled artificial synaptic device for neuromorphic vision sensor applications.