Lateral Asymmetrical Graphene Heterojunction for In‐Sensor Broadband Convolutional Processing

ABSTRACT In‐sensor broadband convolutional processing can achieve efficient and high‐precision sensing for informative multi‐band images, which is crucial for the development of high‐efficiency edge intelligent perception systems. However, traditional photodetectors suffer from constrained tunability of photoresponse, confronting an insurmountable bottleneck for implementing in‐sensor broadband convolution processing. Herein, a lateral asymmetric graphene heterojunction is proposed to achieve simultaneous broadband image sensing and convolutional processing. The device achieves highly tunable broadband photoresponse through bandgap engineering and carrier regulation. Specifically, the device exhibits a broadband detection spanning from visible light to extended short‐wave infrared and possesses a linear bias dependence for both positive and negative photoresponses, with goodness of fit ( R ² ) exceeding 0.99. Moreover, broadband image sensing, preprocessing, and recognition functions are integrated within a single device, demonstrating high‐quality image processing results. These results demonstrate the potential of in‐sensor broadband convolutional processing to advance neuromorphic computing, offering new opportunities for designing intelligent and compact hardware systems.