Solar-Blind Optoelectronic Synaptic Transistor Based on a-IGZO/h-BN/MLG Heterostructure

The development of energy-efficient and high-fidelity artificial synapses is crucial for next-generation neuromorphic computing. Here, we present a solar-blind ultraviolet (254 nm) programmable synaptic transistor based on an a-IGZO/h-BN/multilayer graphene (MLG) heterostructure. The atomically smooth h-BN layer provides a defect-suppressed tunneling dielectric, while MLG functions as a charge-trapping and gain layer, enabling enhanced persistent photoconductivity in a-IGZO. The device exhibits a large hysteresis window, high on/off ratio (105), low subthreshold swing (0.24 V/dec), and high carrier mobility (11.478 cm2/(V·s)). Under 254 nm illumination, a single synaptic event consumes only 51.8 pJ, while paired-pulse facilitation reaches 309.55% at ΔT = 1 s, demonstrating low-energy, low-noise optical synaptic modulation. Leveraging these properties, LTP/LTD currents were mapped onto a VGG-8 neural network for CIFAR-10 image recognition, achieving 90.4% accuracy. Furthermore, the high PPF enabled stable, interference-resistant optical information transmission using ASCII-coded pulses, with maximum postsynaptic current deviation below 0.4 nA. These results highlight a versatile, mixed-dimensional oxide/2D-material platform for high-performance, solar-blind optoelectronic neuromorphic systems.