Recently, a team led by Professor Ke Xu and Professor Qinghai Song from Harbin Institute of Technology (Shenzhen) proposed a silicon-based on-chip speckle spectrometer based on cascaded disordered metasurfaces. By generating speckles rich in spectral fingerprint information through subwavelength phase modulation units with high transmittance, it ultimately achieves high-resolution and broadband spectral reconstruction simultaneously within a sub-square-millimeter chip size. This achievement was published in *Light: Science & Applications* under the title "Scalable on-chip diffractive speckle spectrometer with high spectral channel density". The first author is Zhang Zimeng, a doctoral student at Harbin Institute of Technology (Shenzhen), and Professor Ke Xu and Professor Qinghai Song from Harbin Institute of Technology (Shenzhen) are the corresponding authors. ### Research Background Traditional benchtop spectrometers have obvious limitations in design and application. They rely heavily on complex discrete optical components and precise mechanical structures, which not only increase the volume of the instrument but also significantly increase its weight. In practical use, such spectrometers are difficult to miniaturize and lightweight, failing to meet the development needs of current mobile and portable spectral detection technologies. In sharp contrast, chip-level micro-spectrometers have significant advantages such as small size, light weight, and low power consumption, making them more in line with the development trend of mobile and portable spectral detection technologies. Whether in outdoor environmental monitoring or rapid on-site medical diagnosis, chip-level micro-spectrometers can play a huge role, bringing more possibilities to spectral detection. Silicon-based photonic integration technology plays an important role in the development of chip-level micro-spectrometers. Due to its high reliability and compatibility with microelectronic processes, it has become an ideal carrier for realizing high-performance on-chip spectrometers. With the help of silicon-based photonic integration technology, various optical functions can be integrated on a tiny chip, improving the performance and stability of spectrometers. However, the development of chip-level spectrometers also faces key challenges. Currently, how to achieve high spectral resolution and broadband spectral reconstruction simultaneously in a tiny chip size is an urgent problem to be solved. ### Innovative Scheme The spectrometer proposed by the research team is designed based on a silicon-on-insulator (SOI) platform. Its core architecture includes several key components: an input single-mode waveguide, a wavefront collimating metasurface lens, cascaded multi-layer disordered metasurfaces, and a multi-mode output diffraction grating. These components work together to form the basic structure of the spectrometer, laying the foundation for the processing of spectral information. The mode field output by the single-mode waveguide will undergo wavefront shaping through the metasurface lens. The metasurface lens plays an important role here: it can effectively adjust the mode field to form a collimated beam. This collimated beam then enters the cascaded metasurfaces, providing an appropriate optical signal for subsequent spectral information processing. The cascaded metasurfaces are a key component of the spectrometer, consisting of subwavelength units with fully etched rectangular grooves. By precisely controlling the length of the grooves, these subwavelength units can introduce a phase shift from 0 to 2π while maintaining high transmittance. This characteristic enables the metasurfaces to perform special processing on optical signals, creating conditions for the encoding of spectral information. According to the Huygens-Fresnel principle, the randomly arranged groove arrays on the metasurfaces generate disordered wavefronts. These disordered wavefronts are converted into intensity distributions in the waveguide cross-section through interference and diffraction effects. In this way, spectral information is encoded into speckle patterns. This process is the core step for the spectrometer to process spectral information, storing and transmitting the spectral information in optical signals in the form of speckle patterns. Finally, the spectral speckles are output by grating diffraction. The grating serves as the output part of the entire spectrometer system, outputting the encoded spectral speckle information for subsequent analysis and processing. The design of the entire spectrometer, through the close cooperation of various components, realizes the effective processing and output of spectral information, providing a feasible solution for applications such as spectral detection. ### Performance Evaluation To comprehensively evaluate the universality of the spectrometer, the research team carefully designed experiments. In the experiments, a representative set of test samples was selected, including single-peak spectra, multi-peak composite spectra, and single-peak spectra superimposed with a Gaussian background. These samples cover different spectral types from simple to complex, which can fully test the performance of the spectrometer under various conditions. To ensure the accuracy of the evaluation, the experiment adopted a comparative analysis method. The measurement results of the fabricated spectrometer were compared with those of a commercial spectral analyzer, and the performance of the spectrometer was evaluated through relative error analysis. Commercial spectral analyzers have high accuracy and reliability in the industry, and using them as a reference can more objectively evaluate the performance of the new spectrometer. From the experimental results, the fabricated spectrometer showed excellent performance for simple single-peak spectra. It can accurately capture the characteristics of the spectra and achieve high-precision reconstruction. This indicates that the spectrometer has good adaptability and accuracy in processing basic spectral types, providing strong support for its reliability in practical applications. When faced with complex mixed spectra, such as multi-peak composite spectra and single-peak spectra superimposed with a Gaussian background, the spectrometer also performed excellently. Although these spectra have more complex structures and contain more information and interference factors, the spectrometer can still accurately parse the spectral information and achieve high-precision reconstruction. This fully proves that the spectrometer has strong processing capabilities and can cope with various complex spectral situations. Based on the comprehensive experimental results, it can be concluded that the fabricated spectrometer can achieve high-precision reconstruction in both simple and complex spectral measurements. This result proves that the device has universal spectral measurement capabilities and high accuracy, laying a solid foundation for its wide application in the field of spectral analysis and is expected to play an important role in scientific research, industrial testing, and other fields.