Transparent Optoelectronic Synaptic Transistors for Solar-Blind Neuromorphic Perception and Autonomous Monitoring.

Abstract

Developing high-performance optoelectronic neuromorphic hardware with specificity to the deep ultraviolet solar-blind spectrum, combined with transparency, multi-state capability, and persistent photoconductivity, is critical for solar-blind neuromorphic perception and autonomous monitoring. Here, we report a highly transparent optoelectronic synaptic transistor array based on a solution-processed indium oxide (In2O3) and indium tin oxide (ITO) architecture, featuring specific responsiveness to 270 nm light, a 4096-state capability, and persistent photoconductivity exceeding 10 min. The device achieves near-ideal analog weight updates (R2 = 99.95%), attributed to oxygen vacancy-related carrier trapping and interfacial band engineering. Leveraging these properties, we establish a proof-of-concept intelligent power inspection system using a transparent embedded architecture to autonomously detect faint corona discharges under strong solar background. The hardware performs dual-layer filtering, in which steady-state sunlight suppresses noise via trap filling, whereas transient fault pulses induce long-term potentiation, allowing fault severity quantification and priority-map generation for proactive maintenance. Furthermore, a 20 × 20 synaptic array integrated with an in-sensor computing platform confirms the neuromorphic capability of devices by reproducing biological forgetting and achieving 97.82% accuracy on the MNIST benchmark, with over 80% recognition retained after 10 min. This work identifies transparent metal-oxide synapses as a promising route toward noise-resilient, solar-blind neuromorphic perception and edge vision hardware.