Spectroscopy serves as an indispensable foundation for modern scientific research and industrial applications, ranging from environmental monitoring to food quality control. As demand grows for portable integrated optical sensing platforms, the research community has sought to migrate traditional benchtop spectrometers to chip-scale platforms. However, integrated spectrometers have long faced a persistent contradiction among device footprint, spectral resolution, and bandwidth. A research team led by Prof. Zhenhua Ni and Prof. Junpeng Lu from the School of Electronic Science and Engineering at Southeast University has achieved a major milestone in addressing this challenge by developing a computational spectrometer based on a silicon photonic "Vernier Caliper" concept.
The primary obstacle in miniaturizing spectrometers is that high resolution typically necessitates extended optical paths, which inevitably expands the device footprint to the millimeter range. While interference-based resonators allow for finer resolution in smaller dimensions, they introduce resonant wavelength periodicity that severely restricts the operational bandwidth. Furthermore, many current computational approaches rely on ill-conditioned matrix inversion, which is highly susceptible to numerical instability from noise and often requires extensive detector arrays or complex thermal isolation, compromising compactness.
To overcome these physical and mathematical barriers, the team introduced a hybrid approach that synergizes widely tunable narrowband filtering with computational spectrometry. The core hardware utilizes two cascaded Trapezoidal Subwavelength Grating Microring Resonators (TSWG-MRRs) with radii of only 10 µm. By precisely engineering the dispersion through trapezoidal silicon pillars, the team achieved near-uniform resonator responses across a broad wavelength range. This architecture creates an optical Vernier filter with a 160 nm working window, effectively bypassing the free spectral range limitations of individual microrings.
The system architecture facilitates a distinctive scanning strategy where Vernier resonances are continuously tuned through simultaneous thermo-optic actuation. This process provides adequate orthogonal bases for computational reconstruction at each wavelength step, significantly simplifying the mathematical complexity. The entire detection loop is closed by an edge computing unit, specifically an Nvidia Jetson AI development board, which utilizes GPU-accelerated algorithms to de-convolve the system response and enhance resolution.
Experimental results confirm the effectiveness of this synergistic co-design, with the spectrometer resolving fine spectral peaks separated by as little as 0.74 pm (the average resolution over the 160 nm bandwidth is 1.35 pm). In a practical demonstration of gas-phase molecular spectroscopy, the chip-scale system identified 49 absorption lines of hydrogen cyanide with accuracy exceeding that of a commercial benchtop spectrometer. Occupying a footprint of less than 0.002 mm2, the device establishes a record bandwidth-to-resolution-to-footprint ratio demonstrated to date. This work paves the way for high-resolution, broadband on-chip spectroscopy, maintaining the cost-efficiency and compactness required for future lab-on-a-chip and mobile sensing technologies.
Reference: "Miniaturized computational dispersion-engineered silicon photonic vernier caliper spectrometer" by Hao Deng, Tong Lin, et al., 26 March 2026, PhotoniX.
