DUAL-FREQUENCY-COMB SPECTROMETER AND SPECTROSCOPY METHOD FOR SPECTROSCOPIC INVESTIGATION OF A SAMPLE

Abstract

A dual-frequency-comb spectrometer and a method for spectroscopic investigation of a sample are described. The spectrometer includes first and second frequency comb devices for emitting laser pulses along first and second light paths, wherein the repetition frequency of the laser pulses emitted by the second device is offset from that of the first device. First and second multi-core waveguides including at least two separate single core waveguides having field-coupling via a coupling gap therebetween are arranged in the first and second light paths. The sample is irradiated by the second frequency comb in the second light path. A detector device is arranged in a third light path where the first and second light paths are combined, for simultaneously sensing the first frequency comb and the second frequency comb after an interaction with the sample. A computing device receives output of the detector device and calculates spectroscopic properties of the sample.

Claims

1. A dual-frequency-comb spectrometer, being configured for a spectroscopic investigation of a sample, comprising a first frequency comb source device being configured for emitting a first sequence of laser pulses with a first repetition frequency along a first light path, a second frequency comb source device being configured for emitting a second sequence of laser pulses with a second repetition frequency along a second light path, wherein the second repetition frequency has a frequency offset relative to the first repetition frequency and the second light path is configured to accommodate the sample to be investigated, a first waveguide device being arranged in the first light path for receiving the first sequence of laser pulses and for creating a first frequency comb by supercontinuum generation, a second waveguide device being arranged in the second light path for receiving the second sequence of laser pulses and for creating a second frequency comb by supercontinuum generation, wherein the second light path is configured for irradiating the sample with the second frequency comb, a detector device being arranged in a third light path where the first and second light paths are combined, for simultaneously sensing the first frequency comb and the second frequency comb after an interaction with the sample, and a computing device being arranged for receiving an output of the detector device and being configured for calculating spectroscopic properties of the sample based on the output of the detector device, wherein each of the first and second waveguide devices is a multi-core waveguide with a multi-core waveguide section comprising at least two separate single core waveguides having field-coupling via a coupling gap therebetween.

2. The dual-frequency-comb spectrometer according to claim 1, wherein: one of the single core waveguides of the first waveguide device is a first input waveguide being arranged for receiving the first sequence of laser pulses, and one of the single core waveguides of the second waveguide device is a second input waveguide being arranged for receiving the second sequence of laser pulses.

3. The dual-frequency-comb spectrometer according to claim 2, wherein: each of the first and second input waveguides has an input section with a double-inverse taper of the waveguide shape, and after the input section is the multi-core waveguide section.

4. The dual-frequency-comb spectrometer according to claim 3, wherein the input sections of the first and second input waveguides have a length being selected such that a pulse broadening point, where pulse broadening by supercontinuum generation starts, is after the input sections, and is within the multi-core waveguide section.

5. The dual-frequency-comb spectrometer according to claim 2, wherein: each single core waveguide in the first and second waveguide devices has a rectangular cross-section shape, and each of the first and second input waveguides has a width or a height which is different than remaining single core waveguides of the first and second waveguide devices.

6. The dual-frequency-comb spectrometer according to claim 1, wherein the first and second waveguide devices are monolithically formed on a common substrate.

7. The dual-frequency-comb spectrometer according to claim 1, wherein each of the first and second waveguide devices comprises two, three or four single core waveguides.

8. The dual-frequency-comb spectrometer according to claim 1, wherein the single core waveguides are made of Si.sub.3N.sub.4, Si, GaP or AlN.

9. The dual-frequency-comb spectrometer according to claim 1, wherein the single core waveguides have a width in a range from 0.01 m to 5 m and a height in a range from 0.01 m to 2.5 m.

10. The dual-frequency-comb spectrometer according to claim 1, wherein each coupling gap has a width in a range from 0.01 m to 5 m.

11. The dual-frequency-comb spectrometer according to claim 1, wherein each coupling gap is filled with SiO.sub.2, Al.sub.2O.sub.3, or air.

12. The dual-frequency-comb spectrometer according to claim 1, wherein: the first and second sequences of laser pulses have a center wavelength in a range from 750 nm to 2400 nm, and the first and second frequency comb cover a spectral region from 4166.67 cm.sup.1 to 13333.33 cm.sup.1.

13. The dual-frequency-comb spectrometer according to claim 1, wherein: the first and second sequences of laser pulses have a repetition frequency in a range from 1 MHz to 100 GHz, and the frequency offset is selected in a range from 1 Hz to 100 MHz.

14. The dual-frequency-comb spectroscopy method for spectroscopic investigation of a sample, wherein the dual-frequency-comb spectrometer of claim 1 is used, comprising the steps of: providing the sample in the second light path, creating the first frequency comb by supercontinuum generation with the first waveguide device, simultaneously creating the second frequency comb by supercontinuum generation with the second waveguide device, interaction of the second frequency comb with the sample, so that a spectral content of the second frequency comb is changed, simultaneous detection of the first and second frequency combs with the detector device, and calculating spectroscopic properties of the sample based on the output of the detector device.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Further details and advantages of the invention are described in the following with reference to the attached drawings, which schematically show in:

[0070] FIG. 1: features of preferred embodiments of the inventive dual-frequency-comb spectrometer;

[0071] FIG. 2: further details of preferred embodiments of the inventive dual-frequency-comb spectrometer;

[0072] FIG. 3: an illustration of designing the dispersion landscape of a waveguide device;

[0073] FIG. 4A: a schematic illustration of the electric field distribution of the hybrid modes at different wavelengths;

[0074] FIG. 4B: a graph of calculated dispersion (curves C10, C11 and C12) of the asymmetric mode in dual-core waveguides with three gap distances of 0.6 m (C10), 0.7 m (C11) and 0.8 m (C12), which evolves from the pumped mode (curve C13);

[0075] FIG. 4C: a graph of the calculated dispersion landscape in a Si.sub.3N.sub.4 dual-core waveguide device (curve C14, geometric parameters as indicated in FIG. 4A), compared with that in a single-core waveguide (curve C15, geometric parameters as indicated in FIG. 4A);

[0076] FIGS. 4D, 4E and 4F: infrared spectra;

[0077] FIG. 5A: an infrared spectrum;

[0078] FIGS. 5B and 5C: temporal interferogram traces recorded by a detector device;

[0079] FIG. 5D: an infrared spectrum of measured gas absorbance for methane (C27) compared with HITRAN database curves inverted for clarity (C28);

[0080] FIGS. 5E and 5F: infrared spectra showing details of gas absorbance (curves C27), including both the intensity and the phase information, in comparison with the HITRAN database (curves C28);

[0081] FIG. 5G: an infrared spectrum of measured gas absorbance for acetylene (C27) compared with HITRAN database curves inverted for clarity (C28);

[0082] FIGS. 5H and 5I: infrared spectra showing details of gas absorbance (curves C27), including both the intensity and the phase information, in comparison with the HITRAN database (curves C28); and

[0083] FIGS. 6A, 6B, 6C, 6D and 6E: infrared spectra showing experimental results demonstrating the spectral resolution of the spectroscopic detection with the example of detecting methane isotopologues.

PREFERRED EMBODIMENTS OF THE INVENTION

[0084] Features of preferred embodiments of the invention are described in the following with reference to the configuration of the waveguide devices each with two single core waveguides and the operation thereof for creating the frequency combs. It is emphasized that the implementation of the invention is not restricted to the details of the illustrated examples, but correspondingly possible with changed parameters, in particular with changed number and/or geometry of the single core waveguides. Furthermore, the implementation of the invention is not restricted to the use of the fiber lasers as frequency comb source devices, but correspondingly possible with other pulse lasers. Details of selecting operation conditions of the spectrometer, in particular features of selecting the frequency offset of the frequency combs or analyzing the detector output, are not described as they are known per se from conventional DCS.

[0085] Exemplary reference is made to mid-IR dual-comb gas-phase spectroscopy, in which dual mid-IR frequency combs are generated via the coherent supercontinuum process in nanophotonic chip-based Si.sub.3N.sub.4 waveguides, seeded by a mutually locked dual-frequency-comb source at the telecom-band (i.e. about 1550 nm). The application of the invention is not restricted to these materials and this wavelength range but also possible with other wavelength ranges.

Features of Embodiments of the Dual-Frequency-Comb Spectrometer

[0086] FIG. 1 schematically illustrates features of embodiments of the dual-frequency-comb spectrometer 100 with a first frequency comb source device 10, a second frequency comb source device 20, a first waveguide device 30, a second waveguide device 40, a beam combiner device 50, a detector device 60 and a computing device 70. A first light path I is spanned from the first frequency comb source device 10 via incoupling optics, comprising a half wave plate 11 and a focusing lens 12, and the first waveguide device 30 to the beam combiner device 50. A second light path II is spanned from the second frequency comb source device 20 via incoupling optics, comprising a half wave plate 21 and a focusing lens 22, the second waveguide device 40 and a sample receptacle 1 to the beam combiner device 50, where the first and second light paths I, II are combined into a third light path III towards the detector device 60. The beam combiner device 50 comprises a semi-transmissive mirror having a transparency in the mid-infrared range. The detector device 60 is a mid-infrared photodetector. The computing device 70 comprises a data acquisition unit 71 (DAQ) receiving an output of the detector device 60 and a computer unit 72 running a software for calculating spectroscopic properties of the sample based on the output of the detector device 60.

[0087] Each of the first and second frequency comb source devices 10, 20 comprises e. g. an ultralow noise femtosecond fiber laser with sub-MHz individual linewidth (e. g. Menlo M-combs, a femtosecond Erbium-fiber laser, wavelength about 1550 nm, repetition rate f.sub.rep about 250 MHz). The first and second frequency comb source devices 10, 20 have the carrier-offset frequency locked via self-referencing, and one comb mode optically locked to a shared reference laser (not shown) (at about 1541 nm, the laser is free-running and a daily shift in frequency is of the order of 10 MHz). Optical locking is implemented based on a control loop employing a beating signal of the frequency comb source device and the reference laser as it is known per se in pulse laser techniques. The locked mode index is different by one, which leads to a small difference in the repetition rate of the initial frequency combs, i.e. f.sub.rep320 Hz, and in principle allows the dual-comb spectrometer 100 to cover a large span in the optical window, i.e. about 100 THz.

[0088] The mid-IR frequency comb generated by the supercontinuum process in the first and second waveguide devices 30, 40 can be considered as the spectral extension of the initial frequency comb structure of the original pump source (first and second frequency comb source devices 10, 20), therefore it inherits the full properties of the initial frequency combs. Based on such a configuration, a phase-resolved mid-IR dual-comb spectrometer 100 is provided, with one mid-IR frequency comb passing along the second light path II through the sample receptacle 1, like a gas cell for gas-phase detection or a container or a liquid flow cell for liquid-phase detection, and the other frequency comb travelling along the first light path I serves as the reference.

[0089] After the interaction with the sample, the two frequency combs are interfered on the mid-IR photodetector 60 (e. g. VIGO PV-4TE, mercury cadmium telluride (HgCdTe) detector). Due to the difference in the repetition rate f.sub.rep, a radio frequency (RF) comb is generated, which is composed of distinguishable heterodyne beats between pairs of optical comb teeth. In the time domain, it corresponds to a periodic interferogram pattern that can be directly recorded by the detector device 60.

[0090] The data acquisition is implemented by a field programmable gate array (FPGA). Realtime coherent averaging process is preferably provided by the computer unit 72 for multiple sets of signal. For the averaging process, triggered by the pulse repetition rate of one of the near-IR pump frequency comb source devices 10, 20, the FPGA data acquisition unit 71 can record the output voltage level of the mid-IR detector device 60 and continuously save up to e. g. 84 interferograms. The data is then read out on the computer unit 72, which in the meantime co-adds these interferograms and performs averaging to get a single averaged interferogram (so-called realtime coherent averaging). After each set of e. g. 84 interferograms record in the FPGA data acquisition unit 71, the data communicating and saving on the computer unit 72 will introduce a dead time of about 8 s. Then tens to hundreds of such saved and coherent-averaged interferograms are post-processed with phase calibration, and thus a single interferogram is obtained by averaging the phase-corrected interferograms (so-called offline averaging). To calculate the time-normalized DCS SNR and the figure of merit (DCS quality factor), for offline averaging only the effective data acquisition time is taken into account (dead time excluded).

[0091] The normalized signal-to-noise ratio of the spectrometer 100 has a peak value of 25/s at the region of 3400 cm.sup.1 (where the spectral intensity is strongest). The averaged signal-to-noise ratio is estimated as 10/s. Therefore, a figure of merit of 1.010.sup.6/s can be concluded for the inventive dual-comb spectrometer 100, which is mostly limited by the RIN on the mid-IR frequency combs. Although this figure of merit does not reach the shot-noise limit, it is comparable to reported results in other works, which mostly for DFG based dual-comb spectrometers is 1-610.sup.6/s.

Details of the First and Second Waveguide Devices

[0092] The first and second waveguide devices 30, 40 are manufactured on a common substrate or on separate substrates with lithographic techniques which are known per se. For increasing the durability of the waveguides, they can be coated with a protective cover layer. The single core waveguides of the waveguide devices 30, 40 can be made of the same or different materials.

[0093] Details of the first and second waveguide devices 30, 40 are shown in FIG. 2. FIG. 2A shows a single core waveguide of a conventional technique for supercontinuum generation of a mid IR frequency comb (prior art). One single core waveguide has two geometric degrees of freedom only, comprising the height h and the width w, for influencing the dispersion landscape during frequency comb generation. On the contrary, the multi core structure of the waveguide devices 30, 40 as used by the invention and shown in FIG. 2B, has more geometric degrees of freedom in controlling the cross section geometry and thus controlling the dispersion landscape, in particular, comprising the height h, the widths w.sub.1, w.sub.2 and the gap g of/between the waveguides 31, 41, 32, 42 (examples see below). With increasing the number of waveguides, the number of degrees of freedom even is increased.

[0094] The top-view layout of the waveguide devices 30, 40 in FIG. 2C illustrates that an input section 33, 43 is a single-core waveguide with a double-inverse taper at the beginning, for pumping a selected mode, while the mid-IR continuum will be generated and propagated in both cores in a subsequent multi-core waveguide section 34, 44. An exemplary microscopic picture of the waveguide input section 33, 43 and a scanning electron microscopic (SEM) picture of the multi-core waveguide section 34, 44 are presented in the lower portion of FIG. 2C.

[0095] FIG. 3 illustrates the principle of enhanced mid-IR continuum generation serving as the frequency comb, which corresponds to a dispersion landscape that is engineered and flattened in the mid-IR, particularly by the anomalous GVD produced in the coupled waveguide of the multi-core waveguide section 34, 44, as described with further details in the following. In FIG. 3, the dashed curves C1 and C2 represent the optical power and relative phase constant of a frequency comb generated by the pump pulse (curve C3) in the conventional single core waveguide, while drawn curves C4 and C5 represent the optical power and relative phase constant of a frequency comb generated with the multi-core waveguide of the invention.

[0096] The geometry (width and height) of the single waveguide, e. g. according to FIG. 2A, can be controlled to achieve dispersion engineering, leading to mid-IR dispersive waves [17]. However, this control poses limitations to achieve high power conversion efficiency and ultra-broadband mid-IR continuum. Therefore, the inventors employ an advanced technique based on coupled structures consisting of multiple single core waveguides as shown in FIG. 2B. Such a structure is designed based on the knowledge on a directional waveguide coupler [29], where strong supermode dispersion is induced as a result of spatial mode coupling ([30] to [34]). FIG. 2B shows utilizing a dual-core waveguide device, though this technique can be extended to more parallel single core waveguides.

[0097] When two single core waveguides are in close proximity, the optical mode propagating in one core is coupled to the other core, which effectively changes its phase, i.e. the propagation constant of the mode ((), is the angular frequency of the light). In this way, the group velocity dispersion (GVD) is also changed as it corresponds to the frequency-dependent phase change induced by the mode coupling (GVD=2/2). Physically, mode coupling leads to the hybridization of mode-field distributions, resulting in a pair of hybrid modes that are symmetric and anti-symmetric superpositions of the original waveguide modes (supermodes) [35]. The coupling induced dispersion is then reflected by the phase profile of hybrid modes, which is curved to bridge that of original waveguide modes, and feature avoided crossing between each other.

[0098] Deterministically, anomalous GVD is produced by the anti-symmetric mode, while normal GVD is in the symmetric mode. In principle, such mode coupling (formally referred as mode hybridization) can be engineered at arbitrary wavelength region, particularly in the mid-IR where anomalous GVD is essential for tailoring a flattened dispersion landscape (FIG. 3), but is hardly accessible in conventional single-core waveguides. The dispersion landscape can be considered as relative propagation constant compared with pump pulses (assumed as solitons):

[00001]$\begin{array}{cc}.Math..Math.\left(\right)=\left(\right)-{}_s\left(\right)& \left(1\right)\\ .Math.=\left(\right)-\left(\left({}_s\right)+{}_g^{-1}\left(-{}_s\right)\right)& \left(2\right)\end{array}$

where .sub.s() indicates the dispersionless phase profile of the soliton pulse, .sub.s is the angular frequency of the pump and .sub.g is the soliton group velocity.

Tuning the Dispersion Landscape

[0099] The inventors have tested the effect of tuning the waveguide geometry parameters on the mid-IR spectral structure via the supercontinuum process with the results summarized below and shown in FIG. 4. During the test, the pumping condition for all spectra is similar (pulse energy 1 nJ) and the waveguide length is always 5 mm. Generally, it has been found that the mid-IR continuum is built up by two parts: a long wavelength edge is raised by the mid-IR dispersive wave that is determined by the phase matching; a moderate-wavelength part (in between the dispersive wave and the edge of the broadened soliton sideband) that is by the mode hybridization where the dispersion landscape is changed such that the overall phase mismatch is reduced. As a result, it can be observed that by decreasing the gap distance between the two Si.sub.3N.sub.4 cores, the mid-IR dispersive wave is slightly shifted to longer wavelengths, and in the meantime, the effect of mode hybridization on raising the moderate-wavelength part is reduced. The latter is counter-intuitive as a closer gap usually means stronger mode coupling effect. However, this increased mode coupling would also involve a large wavelength span such that the relative phase-change over wavelength (which is directly linked to the induced group velocity dispersion) is reduced.

[0100] The fingerprint of the mode hybridization is also revealed by measured orthogonal polarized light generation with respect to the pump. While the pump beam is coupled to the horizontally polarized TE00 mode in the narrow core, the moderate-wavelength part is supported by a narrow-band vertically polarized beam with its wavelength in accordance to the designed mode coupling region. This orthogonal polarized beam is found almost independent on the change of the geometry, i.e. the gap distance as well as the width of the wide waveguide core.

[0101] FIG. 4 further illustrates the considerations for designing the Si.sub.3N.sub.4 dual-core waveguide device 30 or 40. FIG. 4A represents the calculated effective refractive indices of (curve C6) the fundamental TE00 mode in a narrow Si.sub.3N.sub.4 core (width w=1.3 m); (curve C7) TE10 mode in a wide core (=3.4 m; and (curves C8 and C9) two hybrid modes in the coupled dual-core waveguide (gap distance g=0.8 m). The height of Si.sub.3N.sub.4 cores is h=0.85 m). Such refractive indices are shown as relative difference to the index of TE00 mode in the narrow core (n1()), which is also the selected pump mode. The insets of FIG. 4A schematically illustrate the electric field distribution of the hybrid modes at different wavelengths. FIG. 4B represents the calculated dispersion (curves C10, C11 and C12) of the asymmetric mode in dual-core waveguides with three gap distances of 0.6 m (C10), 0.7 m (C11) and 0.8 m (C12), which evolves from the pumped mode (curve C13) and produce anomalous GVD in the mid-IR. FIG. 4C represents the calculated dispersion landscape in a Si.sub.3N.sub.4 dual-core waveguide device (curve C14, geometric parameters as indicated in FIG. 4A), compared with that in a single-core waveguide (curve C15, geometric parameters as indicated in FIG. 4A). Both feature the phase matching (zero point) in the mid-IR (about 3500 nm).

[0102] With more details, the cross-section of the two Si.sub.3N.sub.4 single core waveguides are separately selected, in which two modes (one from each core) feature hybridization in the mid-IR region, by matching their propagation constants, (or equivalently by matching the effective refractive index (n.sub.eff), since 8=n.sub.eff /c, c indicates the speed of light in vacuum). For a choice of Si.sub.3N.sub.4 core widths of w.sub.1=1.3 m and w.sub.2=3.4 m, respectively, and for an identical core height of h=0.85 m, the fundamental TE00 mode in the narrow core and the TE10 mode in the wide core have the same n.sub.eff at the wavelength of 3200 nm (FIG. 4A). The propagation constant and the mode-field distribution of hybrid modes are calculated for a coupling gap of g=0.8 m, which exhibit strongly curved phase profile in the mid-IR compared with the original waveguide mode (the TE00 mode in the narrow core is selected as the reference). Therefore, strong and dominant anomalous GVD is produced in the mid-IR (centered at 3200 nm, see FIG. 4B). Counterintuitively, the larger change in dispersion is not achieved by closer proximity of the two single core waveguides, since the latter causes the hybridization region to occur over a larger spectral bandwidth such that the frequency-dependent phase change is reduced.

[0103] Moreover, to exploit anomalous GVD in the mid-IR for engineering the supercontinuum, it is preferred to selectively excite the anti-symmetric mode only. This is accomplished by designing the waveguide input section 33, 43 (see FIG. 2C) to be a single-core waveguide (with an inverse taper in the beginning [28], which excites the TE00 mode, followed by the dual-core section 34, 44. The length of the input section 33, 43 is chosen such that the pulse propagation enters into the dual-core section 34, 44 before significant broadening occurs. In this way, the spectral broadening will extend into the coupling region in the dual-core waveguide, and excite the anti-symmetric mode in the mid-IR. The designed dispersion landscape is shown in FIG. 4C, in which the mid-IR portion is particularly tailored and flattened.

[0104] FIG. 4D represents experimentally observed supercontinuum generation in a Si.sub.3N.sub.4 dual-core waveguide device (e. g. according to FIG. 2B), which features ultra-broadband mid-IR continuum from 2000-3800 nm (OSA: optical spectral analyzers; FTIR: Fourier-transform infrared spectrometer. FIG. 4E represents experimentally observed supercontinuum generation in a conventional single-core waveguide (=1.85 m), which leads to a mid-IR dispersive wave generated at 3800 nm. FIG. 4F shows spectrally overlapped two mid-IR continua from two separate Si.sub.3N.sub.4 dual-core waveguide devices, with similar cross-section geometry.

[0105] With more details, the experiments presented in FIGS. 4D to 4F for mid-IR supercontinuum generation in Si.sub.3N.sub.4 coupled waveguide devices have been carried out as follows. The single core waveguides of the waveguide devices are fabricated using the photonic Damascene process with a crucial reflow step to largely reduce the scattering losses caused by sidewall roughness (see e. g. [36]). Thermal annealing steps can be used, in order to reduce the hydrogen content and the related absorption losses [37]. The waveguide length is 5 mm including the inverse taper input section 33, 43 at the beginning.

[0106] In waveguides similar to the design, supercontinuum generation was observed (FIG. 4D), which is seeded by frequency comb source device 10 (see FIG. 1) comprising an amplified femtosecond fiber laser in the telecom-band (pulse duration <70 fs, maximum averaged power >350 mW, pulse energy >1 nJ, repetition rate about 250 MHz). Significantly, with respect to the pumped wavelength (i.e. 1550 nm) the supercontinuum is mostly extended to the long wavelength side, leading to ultra-broadband mid-IR continuum ranging from 2000-3700 nm, while at the short wavelength side, it features a sharp edge (stopping at 1000 nm) followed by a dispersive wave in the visible range (at 600 nm). The spectral envelope exactly reflects the designed dispersion landscape, i.e. the spectrum amplitude is inverse proportional to the relative phase constant. The dispersive wave is generated by its phase matching to the pump pulse (cf. the zero-valued relative phase constant in the dispersion landscape), and with the group velocity exceeding the pump wave, as it behaves analogous to Cherenkov radiation. Significantly, filtering out the mid-IR continuum by an edge-filter (cut-on at 2500 nm), measured the net power in the mid-IR is measured to be 1-3 mW, depending on the intensity of the pump. Such a power level comprises already the insertion loss of the waveguide and the loss in the light collecting component (e.g. a mid-IR collecting length that has a transparency of 70%). The conversion efficiency is then estimated 1-5%, producing thereby sufficient power to implement dual-comb spectroscopy. The inventors also observed a narrow-band mid-IR wave generation in the opposite polarization direction to the pump wave, at mode coupling wavelength 3200 nm. This radiation is understood as the result of mode hybridization such that the polarization of the mode in the coupling region is rotated as well.

[0107] For comparison, the inventors also performed the supercontinuum experiment in a conventional single-core waveguide (FIG. 4E). This waveguide, with a selected core width (w=1.85 m), can feature the dispersive wave generation at 3800 nm. Comparing the mid-IR continuum in the dual-core and the dispersive wave in the single-core waveguide demonstrates an enhancement in terms of both the intensity and the spectral coverage. In particular, while the dispersive wave in the single-core waveguide is detached spectrally from the main pulse, due to significant phase mismatch at intermediate wavelengths (FIG. 4C), the latter is strongly reduced in the case of dual-core waveguides such that the mid-IR continuum can exhibit a broader spectrum that covers the important spectroscopic window 2500-3500 nm.

[0108] From separate Si.sub.3N.sub.4 waveguide chips, similar mid-IR continuum can be generated with similar dual-core waveguide devices (FIG. 4F). A high level of spectral overlap is implemented, which is essential for building up the dual-comb spectrometer 100 of FIG. 1.

Modelling the Dispersion Landscape

[0109] In theory, nonlinear wave propagating dynamics in a waveguide can be described by the following wave equation: (it is sufficient to consider the spontaneous response in the cubic nonlinearity)

[00002]$\begin{array}{cc}\frac{\tilde{E}\left(,r\right)}{z}=-i.Math..Math.\left(\right).Math.\tilde{E}\left(,r\right)-i.Math.\frac{.Math..Math.{}^{\left(3\right)}}{2.Math.\mathrm{cn}}.Math.{\left[{E\left(t,t\right)}^3\right]}_{}& \left(3\right)\end{array}$

where E(t, r) indicates the electric field of the light wave in the time domain (t-axis), and its amplitude spectral density is E.sup. (, r) in the frequency domain (-axis), namely via the Fourier transform (operator F) there has: E.sup. (, r)=dtE(t, r)e.sup.it= F [E(t, r)].sub.; r={x, y, z} indicates the space frame and the light propagation direction in the waveguide is defined as the z-axis; () indicates the propagation constant of the light wave in a waveguide, which is frequency dependent reflecting dispersion properties; n is the effective refractive index of the waveguide; (3) is the cubic nonlinear susceptibility of the waveguide material; c is the speed of light in vacuum. The electric-field can be further expressed as:

{tilde over (E)}(,r)={tilde over (B)}(,x,y)(,z)(4)

with B.sup. the normalized mode distribution such that: dxdy B.sup.2=1. Thus the propagation dynamics of the light field is enfolded in A.sup. and the Eq. 3 can be modified to: (if only considering the nonlinear phase modulation effect, i.e. the Kerr nonlinearity)

[00003]$\begin{array}{cc}\frac{\tilde{A}\left(,z\right)}{z}=-i.Math..Math.\left(\right).Math.\tilde{A}\left(,z\right)-i.Math.\frac{}{c}.Math.\frac{{}^{\left(3\right)}}{2.Math..Math.n.Math..Math.A_{\mathrm{eff}}}.Math.{\left[{.Math.A\left(t,z\right).Math.}^2.Math.A\left(t,z\right)\right]}_{}& \left(5\right)\end{array}$

where information of the mode confinement in the waveguide is reflected on the parameter of the effective mode area, A.sub.eff.

[0110] Furthermore, the case is considered wherein the light wave consists of both a primary wave packet (seeded by the pump wave and assumed as solitons) A.sup..sub.s and a nonlocal small wave .sup. (i.e. the dispersive wave). Therefore, in the frequency range w>0, it is defined:

(>0,z)=.sub.s(,z)e.sup.i.sup.s.sup.z+{tilde over ()}e.sup.i(.sup.d.sup.)z(6)

where =.sub.s defines a relative frequency frame with respect to the pumping frequency s; s() indicates the phase constant of the soliton, which is dispersionless, i.e.:

.sub.s()=(.sub.s)+.sup.(1)(.sub.s)(7)

[00004]$\left(m\right).Math.\left(\right)=\frac{{}^m}{.Math.m}.Math.\left(\right)$

indicates the m-th order of dispersion with respect to .sub.s; v.sub.g=1/(1)() is also known as the group velocity of the soliton; .sub.d indicates the central frequency of the small wave and .sub.d.sub.s.

[0111] Using Eq. 6 in Eq. 5, we obtain the following equations:

[00005]$\begin{array}{cc}\frac{{\tilde{A}}_s}{z}=-i.Math..Math..Math..Math.\left(\right).Math.{\tilde{A}}_s-i.Math.\frac{}{c}.Math.\frac{3.Math.{}^{\left(3\right)}}{8.Math..Math.n.Math..Math.A_{\mathrm{eff}}}.Math.{\left[{.Math.A_s.Math.}^2.Math.A_s\right]}_{}& \left(8\right)\\ \stackrel{~}{}{\tilde{A}}_s.Math.e^{i.Math..Math..Math..Math.\left({}_d\right).Math.z}& \left(9\right)\end{array}$

where A.sup..sub.s=F [As].sub.. Equation 8 is written in the frequency domain (-axis), and its form in the time domain is the well-known nonlinear Schrdinger equation with full dispersion (i.e. the dispersion landscape), namely

[00006]$.Math.\left(\right)=\left(\right)-{}_s={}_{m2}.Math.\frac{{}^{\left(m\right)}}{m!}.Math.{}^m.$

In particular, with anomalous group velocity dispersion (GVD), i.e. .sup.(2)<0, solitons are supported in the waveguide. Equation 9 is derived at the sideband of A.sup.s, i.e. when =.sub.d, and the nonlinear effect is also neglected.

[0112] Significantly, from Eq. 9, the phase matching condition between A.sup..sub.s and .sup. is: (.sub.d)=0. This is also understood as the phase matching between the soliton and the dispersive wave (Note: soliton would have an extra nonlinear induced phase constant (q) enfolded in A.sup..sub.s, which however is usually small valued and is neglected). Moreover, the conversion efficiency of the dispersive wave depends on the intensity of the soliton sideband. Described by Eq. 8, A.sub.s would experience the nonlinear self-phase modulation (second term on the right hand side) that in the frequency domain, results in its spectral broadening (i.e. raising the sideband power). Conventionally, this effect will be counterbalanced by certain dispersion in the system, i.e. ()0. Nevertheless, the maximum sideband power comes where there is the lowest dispersion, i.e.: ().fwdarw.0.

[0113] Therefore, apart from the phase matching condition, the overall landscape of actually determines the conversion efficiency of the dispersive wave, which in the range [.sub.s, .sub.d] can be assimilated to a spectral barrier between the soliton and the dispersive wave. The purpose of our design is indeed to implement a flattened and reduced dispersion landscape, in the mid-IR, such that the supercontinuum generation can be broadband with enhanced efficiency.

Experimental Results

[0114] FIG. 5 illustrates experimental results of gas-phase detection enabled by the ultra-broadband mid-IR dual-frequency-comb spectrometer 100 of FIG. 1. FIG. 5A represents a retrieved mid-infrared spectrum from the detected and coherent averaged interferogram trace, after the averaging of 52 s (net data acquisition time), which covers a large span from 2800-3600 cm.sup.1.

[0115] The gas species in the gas cell, i.e. methane and acetylene, as well as water vapor in the circumstance, are featured as sharp absorbance in the spectrum. FIGS. 5B and 5C show the temporal interferogram trace recorded by the detector device 60. The encircled area C26 marks the mode hybridization region which exhibits a big dip in the retrieved spectrum. In FIGS. 5D and 5G, the measured gas absorbance (curves C27), methane (FIG. 5D) and acetylene (FIG. 5G) are compared with the HITRAN database (curves C28, inverted for clarity) in a large span of wavenumbers. The gas cell of the sample receptacle 1 (see FIG. 1) has a total pressure of e. g. 1 atm, with methane at 136.4 ppm, acetylene at 406.5 ppm, and nitrogen as the buffer gas. FIGS. 5E, 5F, 5H and 5I show details of gas absorbance (curves C27), including both the intensity and the phase information, in comparison with the HITRAN database (curves C28). Residual data from the fitting is also presented (curves C29), an offset value is artificially imposed merely for plotting purpose, and the dashed lines assist to know the range of the residual data.

[0116] With more details, FIG. 5 shows the result of mid-IR dual-comb gas-phase spectroscopy, where the spectral coverage is 2800-3600 cm.sup.1 (about 25 THz) corresponding to the number of comb teeth of 100,000. The spectrum shown in FIG. 5A is retrieved from a coherent averaged temporal interferogram pattern which in the Fourier domain corresponds to the RF comb constructed by the interference of the two optical frequency combs. The sample receptacle 1 (see FIG. 1) is a gas cell with a length of 104 cm and with wedged sapphire windows to avoid etalon effects. The sample receptacle 1 is filled with low concentration of methane and acetylene as targeted gas species, and nitrogen as the buffer gas.

[0117] To extract the absorption spectrum, first the spectrum through the sample gas T is measured, then the cell is purged and filled back to the original pressure with pure nitrogen. Subsequently, the reference spectrum T0 is measured. The spectral absorbance is then determined via-ln(T/T0). The gas concentrations are extracted from the absorption spectrum via a nonlinear least square fitting with data (line center frequency, line intensity, pressure broadening and shift coefficients) from HITRAN 2016 database. The HITRAN phase spectra are calculated from the Kramers-Kronig transformation of the absorption spectra. The retrieved gas absorbance and phase spectra agree very well with HITRAN database (FIGS. 5D to 5I). In addition the water vapor in the circumstance was also detected as about 1.8%.

[0118] The performance of the inventive supercontinuum-based mid-IR dual-comb spectroscopy is also benchmarked by successful detection of natural isotopologues of methane (FIG. 6), i.e. .sup.12CH.sub.4, .sup.13CH.sub.4 and .sup.12CH.sub.3D. Measured absorbance of methane (curve C30) is compared with the HITRAN database (curve C31, inverted for clarity), in the case of high concentration, i.e.12.5% (FIG. 5A). The total pressure of the gas cell is 0.1 atm. FIGS. 5B to 5E show the absorbance of .sup.12CH.sub.4, .sup.13CH.sub.4 and .sup.12CH.sub.3D, at selected regions B to E.

[0119] For this measurement, the gas cell is operated at low pressure such that the collisional broadening of spectral lines is reduced and those corresponding to isotopologues (i.e. .sup.13CH.sub.4, natural abundance 1.11%; .sup.12CH.sub.3D, 0.06%) can be resolved as separated from traditional elements (.sup.12CH.sub.4). In experiments, the pressure of the gas cell is set to be about 0.1 atm and the methane concentration is about 12.5%. At this pressure, the full-width-at-half-maximum spectral linewidth of methane is reduced to about 480 MHz, which is both sufficient for separating isotopes, and resolvable by our sub-Doppler resolution (i.e., 250 MHz, determined by the mode spacing) of the mid-IR frequency comb. The capability of identifying natural abundance of isotopes is of high importance as it provides signatures in earth science as well as in cosmology.

[0120] In conclusion, the inventors have demonstrated a high-performance mid-IR dual-comb spectrometer based on the supercontinuum process in nano-fabricated non-linear Si.sub.3N.sub.4 waveguides. The inventive use of coupled single core waveguides has enabled strong and dominant supermode dispersion in the mid-IR, particularly the anomalous GVD that is essential for generating flat-envelope, ultra-broadband mid-IR frequency combs. Based on this regime, the dual-comb spectrometer 100 can cover a large wavelength span, covering most of the functional groups window (2800-3600 cm.sup.1) and allowing for parallel gas-phase detection of trace gas. Such supercontinuum based dual-comb spectrometer 100 has not only detected traditional chemical species, e.g. methane and acetylene, but can also trace their isotopologues, including methane-C13 and mono-deuterated methane. This is enabled by the high-SNR of the spectrometer, which has a peak value of about 25/{square root over (s)}. With the bandwidth consisting of more than 100,000 comb modes, the figure of merit is >10.sup.6/{square root over (s)}, which is comparable with those DFG-based mid-IR dual-comb spectrometer.

[0121] The long wavelength edge of the spectrometer 100 can be limited, e. g. to 4.0 m by a SiO.sub.2 cladding in the fabrication of Si.sub.3N.sub.4 waveguides. In particular, although Si.sub.3N.sub.4 shows a much larger transparency window reaching the beginning of the backbone region (5 m), mode coupling will expose the light propagation mostly in the cladding and therefore feature strong loss in SiO.sub.2. Such a problem can be solved with air-cladding waveguides or substrates that are mid-IR transparent (e.g. using sapphire substrate).

[0122] The features of the invention disclosed in the above description, the drawings and the claims can be of significance individually, in combination or sub-combination for the implementation of the invention in its different embodiments.