Flexible Photonic Synaptic Transistors with UV Responsivity via Graphene Quantum Dots for Neuromorphic Vision Systems

Abstract

The ever-growing demands of the Internet of Things and artificial intelligence are pushing conventional von Neumann architectures beyond their limits, largely due to the physical separation of memory and processing units. Neuromorphic computing, engineered to mimic the massively parallel, ultralow-power operation of the human brain, relies critically on devices that faithfully emulate synaptic function. Here, we report a flexible, ultraviolet (UV)-responsive optoelectronic synaptic transistor that integrates light sensing, memory, and signal processing within a single element. Our device employs graphene quantum dots derived from a functionalized hexa-peri-hexabenzocoronene (HBC-PF6) dispersed in a poly(4-vinylphenol) matrix as a floating gate with an ultrathin Al2O3 tunneling layer and a high-mobility organic semiconductor channel (2,7-dioctyl[1]benzothieno[3,2-b][1]benzothiophene (C8-BTBT)) on an indium tin oxide (ITO) gate electrode. Under UV illumination, photogenerated holes tunnel into the channel while electrons are trapped in the floating gate, inducing a nonvolatile shift in threshold voltage and mimicking synaptic potentiation; a reverse bias erases the memory. The transistors retain stable operation under extreme mechanical bending (+/- 1 mm radius) and exhibit robust synaptic behaviors-excitatory postsynaptic currents, paired-pulse facilitation, short- to long-term plasticity transitions, and reversible long-term potentiation/depression-at energy consumptions as low as 1.2 fJ per event. Finally, an array of these devices implemented in a simple artificial neural network achieves >91% accuracy on handwritten-digit recognition, demonstrating their promise for wearable neuromorphic vision systems.