Integrated Single-Resonator Spectrometer beyond the Free-Spectral-Range Limit

Abstract

The chip-scale miniaturization of optical spectrometers may enable many potential applications, such as wearable health monitoring, field deployable biochemical sensing, dense hyperspectral imaging, and portable optical coherence tomography. However, the widespread use of integrated spectrometers is hampered by an inherent trade-off between resolutions and bandwidths. Here, a ground-breaking design strategy is proposed to overcome the bottleneck. The most noteworthy finding in this work is that, by simultaneously leveraging temporal and spatial decorrelations, a single micro-ring resonator (MRR) can serve as a spectrometer with an ultra-high-resolution across an ultrabroad bandwidth far exceeding the narrow free spectral range (FSR). The structure is based on a tunable MRR that supports TE 0 and TE 1 transverse modes. When tuning the MRR, the unknown spectrum is scanned by dual-mode resonances in a synchronized manner. The recorded signal is formed by "splicing" the responses of TE 0 and TE 1 . Due to the intermodal dispersion, all of the wavelength channels in the transmission matrix are sufficiently decorrelated beyond the FSR limit by cross-referring two matrix halves; thus, any arbitrary spectra can be retrieved by solving a linear inverse problem with preconditioned iterative optimizations. Experimental results demonstrate an ultrahigh resolution of <80 pm across an ultrabroad working bandwidth of >100 nm, which yields an ultralarge-wavelength channel capacity of 1250. The device footprint is also as compact as 20 × 35 μm 2 . These results represent the smallest-resolution-footprint product (≈0.056 pm•mm 2 ), the highest bandwidth-to-footprint ratio (≈0.14 nm μm −2 ), and the highest channel-to-footprint ratio (≈1.79 μm −2 ) ever demonstrated.

Integrated Microring Spectrometer with In‐Hardware Compressed Sensing to Break the Resolution‐Bandwidth Limit for General Continuous Spectrum Analysis

“…On the other hand, we also utilized the singular value decomposition to valuate the matrix orthogonality of the response matrix of CDGFs and PDGFs, which has been widely applied in solving ill-posed inverse problems. [34,42,43] The SV curve of our proposed PDGFs is smoother and has a slower decay than that of CDGFs, and the kink is not clearly visible (Note S1, Supporting Information), which agrees well with the results in Figure 2. Therefore, we can achieve high-performance random filters with fast and different response spectra by adopting PDGFs.…”

“…The utilization of an MRR with a higher quality factor can enhance spectral resolution to a level of a picometer without diminishing the FSR. [43,45] To increase the spectral bandwidth, the most straightforward method is to employ an MRR with a larger FSR. In this regard, we have used an MRR with a 9.4 μm radius to increase the FSR to 10.13 nm, which led to a spectral bandwidth of about 60 nm (Supplementary Note 2).…”

Miniature computational spectrometer with a plasmonic nanoparticles-in-cavity microfilter array

“…With sophisticated spectrum reconstruction algorithms, the spectral resolving ability of spectrometers can be greatly improved, e.g., by enabling a commercial RGB image sensor with only three broadband color filters for effective spectrum measurement 10 , 11 and decreasing the orthogonal requirement of wavelength-selective components in spectrometer construction. Therefore, many types of photonic technologies, such as thin-film optical filters 12 , perovskite film 13 , single nanowire and superconducting nanowire 14 , 15 , tunable van der Waals junction 16 , folded digital meta-lenses 17 , integrated photonic chips 18 , 19 , and wavelength-selective photodetector 13 , 20 , have been used as computational spectrometers for integrated chip-size spectrometers.…”

Scalable integrated two-dimensional Fourier-transform spectrometry

“…Fourier-transform spectrometers (FTS) sample an unknown spectrum with sinusoidal responses at varying periods and reconstruct it in the Fourier domain 5 . FTSs have two principal advantages, compared to the schemes based on diffraction gratings 6 , 7 or tunable filters 8 , 9 . First, the spectral information at all wavelengths is simultaneously captured, resulting in a noise reduction proportional to the square root of the number of channels (known as Fellgett’s advantage 10 ).…”