Ferroelectric Quantum Dots for Retinomorphic In-Sensor Computing.

Quantum dots (QDs) offer significant potential for neuromorphic machine vision, owing to their high absorption coefficients, and to absorption that spans the ultraviolet-to-visible range. However, their practical application faces critical challenges in achieving accurate target recognition and tracking in low-light and dynamically-changing environments. A fundamental limitation is a result of the exciton-confinement effect of QDs, which impedes efficient exciton dissociation. To overcome this problem, we synthesized ferroelectric QDs (FE-QDs) that are functionalized with thiol-terminated polyvinylidene fluoride (PVDF-SH) ligands, and empolyed them as the photo-sensitive floating gate in an organic synaptic transistor. When a polarization voltage is applied to the organic synaptic transistors, the FE-QD film generates an electric field that counteracts exciton confinement. The process substantially facilitates exciton dissociation in QDs, and regulates charge accumulation in the channel layer. Integrated with machine learning algorithms, the QD-based device achieved 100% accuracy in detecting simulated car motion in low-light environments, highlighting the potential of adaptive, dynamic sensing technologies for applications in night vision, autonomous driving, and intelligent transportation systems.