Quaternary PdPS_0.55Se_0.45 for SBUV-to-SWIR Broadband Photodetection and Tri-Band In-Sensor Processing

Area of Science:

• Materials Science
• Optoelectronics
• Remote Sensing

Background:

• Accurate post-fire damage assessment requires multi-band remote sensing and effective in-sensor image pre-processing.
• Existing photodetectors have limited spectral responses, hindering unified sensing and computing.
• Broadband perception from solar-blind ultraviolet (SBUV) to short-wave infrared (SWIR) is crucial for advanced remote sensing.

Purpose of the Study:

• To develop a stable quaternary vdW semiconductor photodetector with broadband spectral response.
• To demonstrate in-sensor convolutional processing capabilities for remote sensing image pre-processing.
• To enhance post-fire target recognition accuracy in complex remote sensing scenarios.

Main Methods:

• Fabrication of a stable quaternary vdW semiconductor, PdPS_0.55Se_0.45.
• Demonstration of broadband sensing from SBUV (266 nm) to SWIR (1550 nm).
• Exploitation of power-density-dependent responsivity for programming band-specific convolution kernels and analog multiply-accumulate weights.

Main Results:

• The photodetector achieved peak responsivity (R) of 98.13 A W^-1 and specific detectivity (D*) > 10¹³ Jones at 266 nm.
• Intrinsic responsivity differences were mapped into analog weights for tri-band in-sensor pre-processing.
• Convolutional neural network (CNN) integration achieved ~96% accuracy for post-fire scene recognition on noisy images.

Conclusions:

• A single photodetector enabling broadband SBUV-to-SWIR sensing and in-sensor computing was successfully developed.
• The device offers a practical route for perception-computation integration in remote sensing.
• This advancement creates new opportunities for remote sensing vision in complex environments.