Wavelength‐Driven Kinetics in a CrOCl‐Gated Graphene Optoelectronic Synapse for Front‐End Motion Extraction

Creating optoelectronic synapses with spectrally programmable plasticity is crucial for neuromorphic vision but remains challenging. Here, we present a wavelength‐adaptive synaptic transistor using a trilayer graphene channel gated by chromium oxychloride (CrOCl), exploiting dual optical responses for multi‐modal operation. The device exhibits distinct kinetics: visible light (532 nm) triggers fast, high‐gain potentiation (103 A/W responsivity) through band‐edge absorption, while near‐infrared (912 nm) excitation enables gradual, accumulative weight updates via sub‐gap processes, achieving 256 distinguishable states with biexponential dynamics and long‐term retention (∼2 h). Leveraging this contrast, we develop a motion extraction algorithm that computes a differential current map, suppressing static backgrounds while preserving transient signals. Integrated with a compact convolutional neural network, this approach achieves 93% accuracy in vehicle speed estimation using real‐world BrnoCompSpeed dataset, demonstrating optoelectronic synapses in dynamic visual processing. Our work highlights bandgap‐engineered heterostructures for energy‐efficient, in‐sensor neuromorphic vision systems with spectrally tunable temporal processing capabilities.