Engineered spectral response in photodetectors combined with advanced signal processing and deep learning-based image reconstruction enables widespread applications of hyperspectral imaging. These advancements in spectral imaging eliminate the need for complex filters and dispersion lenses, benefiting various fields such as remote sensing, astronomy, agriculture, healthcare, forensics, food quality assessment, environmental monitoring, and cultural heritage preservation. We present a spectral response design method using photon-trapping surface textures (PTSTs) to enable system miniaturization by eliminating the need for external diffraction optics and employing detector-only spectral sensors. We additionally demonstrate the fabrication of cost-effective, high-performance silicon photodetectors with unique spectral responses by integrating PTSTs. These CMOS-compatible photodetectors are ultra-fast, highly sensitive, and suitable for wideband multi/hyperspectral imaging systems. Our investigation uncovers a prominent linear correlation between the PTST periods and the peak coupling wavelengths while observing a weaker relationship with the PTST diameters. Furthermore, we establish a significant association between inter-PTST spacing and wave propagation patterns. In a proof-of-principle demonstration, we effectively employ these photodetectors with distinct spectral responses to capture visible and near-infrared wavelengths for multispectral imaging. These findings support the feasibility of integrating high-performance on-chip spectrometers, offering compact form factors, extensive applicability, and real-time data acquisition and manipulation capabilities.
Keywords: Avalanche photodiodes; hyperspectral imaging; multispectral imaging; photon-trapping features; spectral response engineering.