Room‐Temperature Mid‐Infrared Photon‐Triggered Detectors with In‐Sensor Perception and Data Preprocessing Capabilities

Abstract Harnessing the bioinspired architecture of the brain and integrating neuromorphic optoelectronic logic at room temperature represents a transformative step toward energy‐efficient optical computing platforms capable of advanced artificial intelligence functionalities. Although infrared photodetection transcends biological vision constraints, conventional charge transport mechanisms impose fundamental limitations on synaptic devices—including nonlinear conductance modulation and inadequate photocurrent generation under operational biases—that restrict in‐sensor perception and adaptive data preprocessing. These challenges are addressed through a heterostructure that integrates black arsenic phosphorus (b‐AsP) with high‐κ bismuth selenite oxide (Bi 2 SeO 5 ), thereby enabling combinational optoelectronic logic operations. The optimized device exhibits a peak mid‐wave infrared responsivity of 0.63 A W −1 and demonstrates synaptic functionality, as evidenced by 148% paired‐pulse facilitation. A hybrid optical‐electronic neural network is further implemented that supports multimodal learning paradigms, attaining 97% inference accuracy on the Modified National Institute of Standards and Technology (MNIST) dataset for image classification tasks. This network integrates seven distinct logic gates within a single synaptic unit, enabling simultaneous signal perception and encrypted data preprocessing, thereby enhancing system security and data confidentiality. These advances underscore the viability of mid‐infrared photon‐triggered detectors in neuromorphic applications and highlight the potential for significant reductions in transistor resource utilization and spatial footprint in optoelectronic systems.