Retina-inspired visible color discrimination optoelectronic synapse via local surface plasmon-assisted reduction of graphene oxide
摘要
Hardware mimicking human vision attracts extensive interest in the field of the neuromorphic domain. An optoelectronic synapse with visible multi-wavelength modulated characteristics is essential to emulate the human retina for color image recognition, which remains a challenge. Herein, an optoelectronic synaptic memristor with visible multiwavelength-modulated plasticity via local surface plasmon-assisted reduction (LSPR) of graphene oxide (GO) is first demonstrated. The memristor is fabricated with GO and Au nanoparticle composite and the size- and aspect ratio-dependent absorption wavelength of Au LSPR endows the device with red, green, and blue light distinguishable synaptic plasticity. The synaptic strength can be dynamically modulated through electrically driven oxygen redistribution, mimicking the advanced visual adaptation function to enable optimal visual perception across different light conditions. Color image sensing, memory and color-filtering preprocessing functions are realized with the optoelectronic synapse array. Moreover, the color-filtering remarkably improves the rate of recognizing colorful images based on an artificial neural network of the optoelectronic synapse. This work provides an effective way to design an optoelectronic synapse for a neuromorphic vision application.