image: Figure | Schematic of the working principle of computational spectrometers. a, Learning process (also known as the calibration process). Various optical elements act as encoders. The response matrix R of the encoders is measured via diverse known spectra. b, Measurement process. The output signal I is detected under an unknown incident light. c, Reconstructed spectral information S by computational methods with the data obtained in the above two processes.
Credit: Yiru Zhang et al.
Spectrometers analyze light-matter interactions across a range of wavelengths, serving as vital tools in scientific research and industries. Historically constrained by large size, complexity, and high cost, traditional instruments faced limited accessibility. Now, advances in computational spectrometers have enabled compact, portable, and cost-effective devices while achieving competitive performance. This enables versatile, real-time spectral analysis across diverse environments beyond laboratory settings.
In a comprehensive review published in eLight, an international team led by Professor Weiwei Cai (Shanghai Jiao Tong University), Professor Tawfique Hasan (University of Cambridge), and Professor Zhipei Sun (Aalto University) details a transformative approach to spectral analysis. The work thoroughly examines the theoretical foundations of spectral encoding/decoding, explores innovations in encoder design, and critically reviews leading-edge decoding techniques. Uniquely, the review emphasizes hardware-software co-design as a paradigm-shifting approach, enabling high-fidelity spectral reconstruction within miniaturized systems. By integrating insights from inverse design, nanofabrication robustness, and lightweight algorithm deployment, this work provides a roadmap for developing field-portable, high-resolution spectrometers with applications ranging from healthcare and environmental monitoring to consumer electronics.
Reconstructive spectrometers are based on a synergistic fusion of miniaturized encoding hardware and computational reconstruction algorithms. Unlike traditional bulky instruments, these systems use nanostructured photonic components (e.g., metasurfaces, microrings, or quantum materials) to encode spectral information into compact measurable signals, while advanced algorithms decode this data to reconstruct high-fidelity spectra. These scientists summarize the design innovations of reconstructive spectrometers:
“The performance of a reconstructive spectrometer depends not only on the spectral response matrix (determined by the hardware encoder) but also on the inversion algorithm. This dual dependency significantly increases the complexity and workload of designing a spectrometer system compared to focusing on just one of these components.”
“End-to-end design offers an intuitive and powerful solution for future spectrometer design, offering significant guidance for bridging the gap between simulation and real-world scenarios” they added.
Scientists stated that “At the design level, significant technical challenges remain.” “Addressing these limitations calls for data-efficient acquisition combined with advanced on-device computation. Neuromorphic vision sensors, inspired by biological vision, use event-driven sampling and in-sensor computation to minimize redundant data and power consumption, enabling real-time spectral analysis in compact, energy-constrained devices. Additional strategies, such as near-sensor and in-sensor computing with memristive implementations, further reduce latency and energy consumption by minimizing data transfer between memory and processing units. Direct photocurrent-based material fingerprinting without full spectral reconstruction offers another pathway to significantly cut computational complexity and power demand.” In summary, this field is continuously bursting with vitality and innovation. Scientists predict that “As research continues to evolve, miniaturized spectrometers are expected to become increasingly sophisticated, efficient, and application-focused”.
Journal
eLight
Article Title
Reconstructive spectrometers: hardware miniaturization and computational reconstruction