Molecular Cocrystal-Based Neuromorphic Vision System With Near-Infrared Responsivity and High Electronic Performance for Facial Recognition in Scattering Media.

Abstract

Neuromorphic visual systems are core technologies enabling round-the-clock perception for the Internet of Things (IoT). While reliable sensing in complex lighting conditions requires robust near-infrared (NIR) capabilities, existing optoelectronic devices struggle to simultaneously achieve a broad NIR spectral response and high electronic performance. This limitation severely hinders their practical applications in anti-interference imaging and intelligent recognition. Here, high-performance organic photonic synaptic transistors (OPSTs) are reported, which employ a C8-BTBT channel layer and a perylene-TCNQ cocrystal NIR photosensitive layer, with polystyrene (PS) incorporated to enable efficient interfacial charge modulation and assist charge transport. The OPSTs exhibit a high mobility of 2.65 cm²·V^-1·s^-1 and an on/off ratio exceeding 10⁶, while extending the spectral response range to the NIR region up to 1200 nm, breaking the inherent trade-off between spectral response bandwidth and charge mobility that limits most existing NIR optoelectronic synapses. Benefiting from its excellent NIR response and electrical robustness, the device successfully emulates a series of retinal-like optical synapse plasticity behaviors. It achieves high-contrast imaging in simulated scattering media and attains a 94% face recognition accuracy in neural network simulations. This work offers a promising strategy for anti-interference NIR neuromorphic vision in complex illumination and all-weather IoT applications.