Multifunctional organic artificial optoelectronic synapses for neuromorphic computing and a weak-light-sensitive visual system

Abstract

Artificial optoelectronic synapses (AOSs) with electrical and photonic responsiveness are the prerequisite for the realization of a neuromorphic visual system that emulates the biological retina. These retina-inspired devices enable efficiently perceiving, processing, and storing information through synaptic weight modulation via synergistic photonic and electrical stimulation. However, challenges persist in AOS research, particularly regarding two-terminal vertical architectures capable of achieving intraneuronal sensory integration through synergistic electrical and photonic stimulation. Herein, we fabricated a two-terminal AOS using an organic donor:acceptor blend (PBDB-T-2F:BTP-eC9) that exhibited electrical- and photonic-driven synaptic behaviours. Our device exhibited voltage-driven analog resistive switching behaviours and synaptic plasticity. Meanwhile, an artificial neural network could be simulated based on the as-fabricated AOS, demonstrating 93.7% recognition accuracy for a handwritten small digit image. Additionally, our device demonstrated light-driven synaptic plasticity and a low-density (20 μW cm⁻²) light-induced “learning-experience” behaviour. Significantly, our device also demonstrated the capacity to process complex electrical and optical inputs, which was validated by a photonic programming–electrical erasing function and in experiments involving associative learning. Our work shows that the proposed AOS offers great potential for functional extension in artificial visual systems.