Plasmon-enhanced Sb2Te3/FTO heterojunction optoelectronic synapses for image compression encoding and high-accuracy recognition

Abstract

Broadband, energy-efficient optoelectronic synaptic devices are essential for neuromorphic computing systems and embodied intelligent perception. Here, we present a plasmon-enhanced optoelectronic synapse based on an Ag nanoparticle-Sb2Te3/FTO (fluorine-doped tin oxide) heterojunction. The incorporation of Ag and FTO enhances light absorption in Sb2Te3 across a broad spectral range, while the localized surface plasmon resonance effect of the Ag nanoparticles and the p-Sb2Te3/n-FTO heterojunction improves the generation efficiency and subsequent separation and transport of photocarriers, resulting in a significantly enhanced photocurrent response. The device exhibits tunable photo-responsive memristive behavior and multilevel optoelectronic synaptic plasticity even at a low voltage of 1 V. Furthermore, we demonstrate compressed image encoding and recognition using two artificial neural networks, achieving an average accuracy of ∼95%. This work not only provides a strategy for designing high-performance, broadband, low-power optoelectronic synapses but also offers a viable device foundation for image compression encoding and intelligent visual perception.