With rapid industrial upgrading in autonomous driving, semiconductor precision inspection, portable intelligent sensing and space remote sensing, hyperspectral sensors are required to feature broadband coverage, high imaging accuracy, fast dynamic response and monolithic integration. Restricted by spectroscopic principles and chip fabrication processes, conventional hyperspectral devices cannot optimize spectral range, spatial resolution and temporal resolution simultaneously. Broadband detection usually reduces frame rate, and bulky external optical systems increase size and power consumption, limiting large‑scale application in consumer and automotive terminals.
The team led by Bian Le at Zhejiang University achieved original hardware innovation through co‑design of micro‑nano spectroscopic structures and photoelectric detection arrays, fabricating a monolithically integrated hyperspectral sensor. Without complex external optical components, the chip integrates spectral splitting, photoelectric conversion and signal acquisition, realizing continuous broadband spectral response from ultraviolet to near‑infrared. It maintains pixel‑level high spatial resolution and greatly improves temporal response for fast and accurate hyperspectral imaging of dynamic scenes.
Compact, monolithically integrated, fast‑responding and broadband, the sensor can be applied to autonomous driving perception, semiconductor defect inspection, resource remote sensing, in‑vivo bio‑imaging and portable material identification devices. This work resolves the conflict between performance and integration of traditional hyperspectral devices, promoting miniaturization, localization and mass production of high‑performance multispectral sensing hardware.
