An integrated broadband spectrometer on thin-film lithium niobate


Optical spectrometry is a tool to investigate wavelength-dependent light–matter interactions and is widely used in astronomy, physics and chemistry. Integration and miniaturization of the currently bulky spectrometers will have an impact on applications where compactness and low complexity are key, such as air- and spaceborne missions. A high-resolution spectroscopy principle based on the near-field detection of a spatial standing wave inside a subwavelength waveguide has shown great promise to accomplish some of the aforementioned demands. However, small-scale devices based on this principle face strong bandwidth limitations due to undersampling of the standing wave. Here, we demonstrate an integrated single-waveguide Fourier transform spectrometer with an operational bandwidth of 500 nm in the near- and short-wavelength infrared, not relying on any moving components. The prototype device, with a footprint of less than 10 mm2, exploits the electro-optic properties of thin-film lithium niobate in order to retrieve the complete spatial interferogram.


The miniaturization of optical spectrometers is an active field of research fuelled by its impact on a vast number of applications. For air- and spaceborne astrophotonic sensing applications1 in particular, reducing spectrometer size, weight and complexity through compact integration is of crucial importance.

Over recent decades, numerous approaches to realize small-scale spectrometers have been investigated, including devices based on microelectromechanical systems2,3, dispersive spectrometers using arrayed waveguide gratings4,5,6 or microring resonators7 and digital planar holograms8,9. Another type of spectrometer, predominantly used in infrared spectroscopy, is the Fourier transform spectrometer (FTS)10. In a FTS, the spectrum is deduced from its corresponding interferogram, which is formed by interference of the optical signal with a temporally or spatially delayed version of itself. Integrated FTSs without any moving components can be roughly categorized as classical FTSs, where a temporal interferogram is measured11,12,13, as spatial heterodyne spectrometers (SHS)14,15,16 or as stationary-wave integrated Fourier-transform spectrometry (SWIFTS) systems17.

In SWIFTS, a spatial interferogram is formed inside a waveguide by counter-propagating modes of the optical signal17. Optical nanosamplers on top of the waveguide probe the evanescent field of the interferogram. This technology enables high-resolution spectrometers18 with compact and robust design. Integrated spectrometers based on the SWIFTS principle have been demonstrated on various material platforms including silicon17, silicon nitride19, quartz18,20 and polymers21,22. However, due to the inevitably extreme undersampling of the standing-wave pattern, the resolvable bandwidth of devices based on SWIFTS is limited to 5–14 nm at a wavelength of 630–1,100 nm (ref. 20) by the Nyquist–Shannon criterion17. Attempts to overcome the undersampling issue have been pursued by multiplexing20, by external optical path delays23,24, by including a moving mirror21 and by using a slab waveguide22. Recently, the concept of an electro-optic spectrometer using titanium-indiffused waveguides in bulk lithium niobate (LN), with a metallic mirror on one end, was studied25. Because the optical mode is weakly confined in indiffused waveguides due to the low refractive index contrast (Δn ≈ 0.02; ref. 26), sampling of the interferogram is challenging27 and the electro-optic efficiency is limited.

Here, we present an integrated, broadband FTS based on a hybrid thin-film lithium niobate–silicon nitride (LN-SiN) platform. The device employs the linear electro-optic effect in LN to retrieve the fully sampled interferogram without any moving or external components. The LN-SiN platform allows for a tight modal confinement, which enhances the electro-optic efficiency, and for compact photonic integration. The prototypical device validates the working principle experimentally over a broad bandwidth of 500 nm in the near- and short-wavelength infrared (NIR and SWIR). The measured spectral resolution of 5.5 nm at a wavelength of 1,550 nm (corresponding to a resolution of 0.69 THz or 22.9 cm−1) originates from the device design and is not restricted by intrinsic limitations of the underlying spectroscopic principle. Extending the sampled length of the interferogram linearly enhances the resolution. Consequently, wavelength resolutions down to the picometre scale are achievable with this technology20.


Principle of the broadband integrated electro-optic FTS

The integrated electro-optic FTS relies on the detection of the stationary wave formed inside a single-mode waveguide. The optical signal is counter-propagating inside the waveguide and the resulting superposition of the standing waves, the interferogram, is sampled by discrete evanescent field samplers (EFSs) on top of the waveguide (see Fig. 1a for a schematic). The EFSs, typically metallic nanowires covering the width of the waveguide, scatter a fraction of the local energy from the evanescent field to the far field where it can be detected. However, to fulfil the Nyquist–Shannon sampling theorem, which guarantees discrete sampling without any loss of spectral information due to aliasing, the spacing between the EFSs needs to be sufficiently small28. For monochromatic light, the period of the standing wave is given by Λ = λ/2neff, where neff is the effective index of the guided mode at wavelength λ. The resulting maximal sampler spacing required to avoid undersampling is 215 nm (for λ = 1,550 nm, neff = 1.8), which is below the optical diffraction limit. Therefore, the equidistant array of EFSs, with a pitch of d = 3 μm, undersamples the interferogram and the recoverable spectral bandwidth is strongly limited. This is illustrated in Fig. 1b, where an analytical interferogram (blue curve; Supplementary Section 1) is compared to the undersampled one (black curve), which would be obtained from a classical stationary wave FTS without the electro-optic phase shifting proposed in this work.

Fig. 1: Electro-optic sampling of the broadband interferogram.

a, Broadband light coupled to a waveguide from both input facets forms a stationary spatial interferogram inside the waveguide. EFSs on top of the waveguide locally probe the near field of the interferogram. The relative phase φ of the counter-propagating modes is controlled by an external voltage via the linear electro-optic effect in the LN waveguide. b, Enlarged view of the broadband interferogram (blue curve) around the zOPD. The black line indicates the undersampled intensity distribution probed by the discrete array of EFSs without any applied external voltage. c, An electro-optically induced relative phase difference shifts the interferogram along the waveguide. The static array of EFSs now samples a different intensity distribution.

Here, we use the strong electro-optic effect in LN to overcome the bandwidth limitations imposed by undersampling. By means of an electro-optically induced change in the refractive index, the phases of the counter-propagating modes can be delayed with respect to each other. Therefore, by continuously increasing the phase difference between the modes, the zero optical path difference (zOPD) is shifted along the waveguide (Fig. 1c). Conversely, this can be seen as moving the interferogram underneath the stationary array of EFSs and the full interferogram is obtained by a combination of spatial and temporal sampling. To draw an analogy to a classical FTS with a Michelson interferometer, by observing the sampled intensity of a single EFS while varying the OPD, a temporal interferogram is obtained at the position of this EFS. Nevertheless, it is beneficial to sample the interferogram with an equidistant array of EFSs covering a total length L of the waveguide. For N EFSs and a fixed total phase difference, the interferogram is sampled over an OPD N times larger than for a single EFS, resulting in an N times increase in the spectral resolution. The spectral resolution in Fourier-transform spectrometry is determined by the length of the sampled interferogram and is defined as R = λλ = neffL/λ (ref. 17), where L is the length of the symmetric interferogram around the zOPD. Accordingly, this spatial sampling principle is not intrinsically limited in resolution. An increase of L linearly translates into an enhancement of the spectral resolution.

Besides spectral resolution, the bandwidth of a spectrometer that is accessible without any intermediate recalibration or change of set-up is a defining parameter. For existing devices based on the sampling of the evanescent standing wave, the bandwidth is strongly limited by undersampling (see Supplementary Section 2 for an experimental demonstration of the aliasing effect in a static stationary wave FTS as well as a comparison of a sampled broadband interferogram with and without electro-optic phase shifting). By using the electro-optic effect in LN, this obstacle can be overcome and the recoverable spectral bandwidth is now limited only by the single-mode condition of the waveguide. For hybrid LN–SiN waveguides in the NIR–SWIR range, the bandwidth, where only the fundamental transverse-electric (TE) mode is guided, is typically several hundred nanometres and it can be matched to the desired spectral window by adjusting the waveguide dimensions (Supplementary Section 3).

Design of the hybrid LN–SiN spectrometer

For the integrated electro-optic FTS, a hybrid LN–SiN platform was chosen. SiN ridge waveguides (1,300 nm wide, 200 nm thick) are patterned on top of a 300 nm x-cut LN thin film by electron-beam lithography (EBL) and reactive ion etching (RIE). An optical microscope image of the closed-loop waveguide structure is shown in Fig. 2a. From a single input (Fig. 2b), the light is split and propagates through two parallel arms, each closely surrounded by gold electrodes. The gold electrodes are fabricated in close vicinity to the ridge waveguides (3 μm electrode–waveguide separation) via a standard metal liftoff process. The crystal orientation of the LN thin film is chosen so that the crystal z axis, which experiences the largest electro-optic effect, aligns with the external electric field29. The electrodes are connected to a voltage source in a push–pull configuration (see + and − signs in Fig. 2a). In this configuration, an applied voltage induces an opposite phase difference in the two arms of the closed-loop circuit due to the reversed direction of the electric field. In Fig. 2c, the simulated fundamental TE mode at 1,550 nm (neff = 1.8) and the external electric field are superimposed with the cross-section of the waveguide–electrode structure. After this phase-control region, the waveguides are recombined in a straight section (L = 250 μm) where the interferogram forms symmetrically around the centre, which is the zOPD. Along this interference region, 87 EFSs with a pitch of d = 3 μm are deposited on top of the waveguide by a focused ion beam (FIB) tool in platinum deposition mode (Fig. 2d). The width of these 15- to 20-nm-thick metallic stripes is 60 nm, which is smaller than the standing wave period for wavelengths in the SWIR (for a coupled wavelength of 1,550 nm, the standing wave period is around 400 nm). Experimentally, the scattering efficiency of a single EFS at a wavelength of 1,550 nm was measured to be 0.36% of the local power in the waveguide (Supplementary Section 6).

Fig. 2: Electro-optic LN waveguide spectrometer.

a, Optical microscope image of the closed-loop LN–SiN waveguide structure with gold electrodes. The intensity sampled by the array of EFSs (black dashed rectangle) is monitored from the top with a focal plane array detector (FPA). For the results presented in this work, a device with 1-cm-long electrodes was used. However, for illustration purposes, in this figure a shorter device is shown. b, Scanning electron microscope (SEM) image of the input of a SiN–LN ridge waveguide. c, Cross-sectional schematic of the waveguide–electrode configuration. The fundamental TE mode and the electric field distribution (blue arrows) are superimposed. d, SEM image of four platinum EFSs on top of the waveguide. e,f, Images of the undersampled interferogram at 1,550 nm for different applied voltages. The intensity pattern probed by the array of 87 EFSs is shifted along the waveguide with changing voltage.

For broadband applications, the relevant spectral information of the optical signal—that is, the main lobe of the interferogram—typically lies within a few micrometres around the zOPD (Fig. 1b). Therefore, the closed-loop design offers an advantage over devices using reflective20,21,24 or co-propagative spectrometers18, where the region close to the zOPD is not accessed by nanosamplers. Furthermore, the chosen hybrid LN–SiN waveguide platform offers multiple advantages. Due to the LN thin film and the sub-wavelength dimensions of the SiN ridge, the optical mode is tightly confined, which allows for closely spaced electrodes and increases the electro-optic efficiency29,30,31. Another advantage arising from the strong modal confinement is that the footprint of curved circuits can be severely reduced compared to titanium-indiffused, proton-exchanged or femtosecond laser-written LN waveguides25,26,32. Additionally, the mature fabrication techniques for SiN enable relatively low-loss waveguides, resulting in propagation losses of 1.39 dB cm−1 (Supplementary Section 4). These moderate propagation losses are expected to be reducible by improving the SiN deposition process and the EBL33. In summary, the hybridization of SiN ridge waveguides on thin-film LN is a promising platform for active on-chip applications exploiting the nonlinear properties of LN34,35,36.

Monochromatic characterization

As a first characterization of the fabricated device, we coupled monochromatic light at a wavelength of 1,550 nm to the waveguide with a lensed fibre. The sampled intensity pattern of the EFSs is monitored from the top with an objective and an FPA. Figure 2e shows an image of the sampled intensity pattern at a voltage of 5 V. The extreme undersampling of the standing wave manifests in the appearance of a periodic pattern with a period of about seven EFSs (~20 μm), which largely deviates from the real standing-wave period, which is expected to be around 400 nm. By incrementally changing the applied voltage and simultaneously observing the EFSs, the sampled interferogram is recorded at progressing positions of the zOPD. Consequently, the spatial sampling rate is no longer given by the pitch of the EFSs, but by the voltage increment. The image taken at 6.4 V, shown in Fig. 2f, illustrates this interferogram shift. By extracting the sampled intensity of an individual EFS as a function of the voltage, the standing wave is partially retrieved around the position of the EFS. The covered OPD is linearly dependent on the voltage and therefore a single EFS would be sufficient to sample the interferogram. However, because moderate voltages are desired, an equidistant array of samplers is used. Experimentally, we found that for the given electrode design and a sampler spacing of 3 μm, a voltage difference of ~20 V is needed to shift a specific point of the interferogram from one sampler to the adjacent one. By normalizing each sampler with its scattering efficiency, the standing waves of individual EFSs are stitched together to obtain the interferogram covering a total OPD of 87 × 3 μm (Fig. 3a). The oversampled standing wave is shown in a magnified plot around the zOPD in Fig. 3b. The same procedure is repeated for a monochromatic source at 1,040 nm and the orange curve in Fig. 3b depicts the corresponding measured standing wave. This result demonstrates that the device is capable of completely sampling the standing waves from signals over a bandwidth of 500 nm. By Fourier transforming the sampled interferogram, the spatial frequency of the standing wave is recovered, which is then translated to wavelength by taking the waveguide dispersion into account (Fig. 3c; for a discussion of the dispersion correction see Supplementary Section 5). The side lobes in the spectra in Fig. 3c, which are equidistant in spatial frequency, arise from imperfections of the stitching procedure. This introduces periodic distortions with a frequency of ~1/3 μm−1, resulting in the side lobes due to mixing terms with the optical frequency. The monochromatic measurements are used to calibrate the device by extracting the exact sampler spacing and hence by fine-tuning the voltage–OPD relation.

Fig. 3: Monochromatic and dual-wavelength measurements.

a, Experimentally sampled interferogram of a monochromatic source at 1,550 nm over a total OPD of 250 µm. The full interferogram is obtained by stitching the sampled intensity from 87 EFSs. b, Magnified views of the measured interferograms around the zOPD for monochromatic illumination at 1,550 nm and 1,040 nm. The parts of the interferogram sampled by different EFSs are highlighted in grey and white. c, Reconstructed spectra for the measurements at 1,550 nm and 1,040 nm. FT, Fourier transform. d,e, Beating interferograms of two simultaneously coupled sources, 1,480 nm and 1,550 nm (d) and 1,540 nm and 1,550 nm (e), with magnified plots around the zOPD. For all measured interferograms, a faulty EFS at an OPD of −120 µm can be seen. f,g, Reconstructed Fourier transform spectra from the dual-wavelength interferograms in d and e.

To test the calibrated device and to demonstrate simultaneous detection of wavelengths over a broad spectral range, two different monochromatic sources were combined in the lensed fibre and coupled to the waveguide. The spatial interferogram of distinct wavelengths displays a beating envelope, depending on the wavelength difference. Figure 3d,e shows the retrieved interferograms for 1,480 nm and 1,550 nm and for 1,540 nm and 1,550 nm. The corresponding spectra are plotted in Fig. 3f,g. The sampled interferograms have only been processed by normalizing with the scattering efficiencies of the individual EFSs, by interpolation of the interferogram and by correcting for the waveguide dispersion (Supplementary Section 5). No further digital signal processing techniques like zero-padding, apodization or correction of phase errors were performed10,37.

Broadband spectra

We also tested our device with a broadband incoherent superluminescent light emitting diode (SLED) with a centre wavelength of 1,555 nm and a 3 dB bandwidth of 50 nm. The analytical interferogram is calculated by superposition of the monochromatic standing waves, each weighted with its respective proportion of the optical signal. Generally, it can be established that a broader spectrum causes a narrower spatial interferogram, making it crucial to sample the region around the zOPD. The experimentally obtained interferogram of the bare SLED is shown in Fig. 4a and the main lobe of the interferogram only spans ~10 μm to each side of the zOPD. A slight shift of the zOPD in the detected interferogram is most likely due to a minor imbalance of the closed-loop spectrometer caused by fabrication imperfections. The analytically calculated interferogram related to the SLED spectrum is shown in black as an inset of Fig. 4a. A Fourier transform, with the monochromatic calibration from the previous paragraph taken into account, extracts the spectrum of the optical signal, which is shown in Fig. 4b together with a reference measurement recorded with an optical spectrum analyser. We demonstrate the inverse relation between the spectral bandwidth and the expansion of the interferogram by narrowing the spectrum of the SLED with a bandpass filter (centre, 1,550 nm; bandwidth, 12 nm). Figure 4c,d displays the measured interferogram and the corresponding spectrum. Although the experimental spectrum fits in wavelength and overall shape to the reference spectrum, slight deviations are visible. These are most likely an artefact of the limited length of the detected interferogram. In Fig. 4c, the probed OPD is not large enough for the interferogram to decay to zero, which leads to side lobes in the spectrum.

Fig. 4: Broadband spectra.

a, Measured interferogram for a broadband optical signal of a SLED (centre wavelength, 1,555 nm; full-width at half-maximum (FWHM), 50 nm). The inset, highlighted in orange, is a magnification of the region around the zOPD. The black curve is the analytically computed interferogram. b, Reconstructed spectrum (red curve) of the interferogram in a compared with a reference spectrum from an optical spectrum analyser (black curve). c, Spatial interferogram of the same broadband source as in a, but spectrally narrowed with a bandpass filter (FWHM, 12 nm). The insets show the highlighted regions in more detail and the black curve is again the analytic interferogram. d, Reconstructed narrowband spectrum of the bandpass-filtered SLED (blue curve) together with a reference measurement (black curve).


The principle of spectroscopy based on the near-field sampling of a standing wave inside a waveguide has proved to be valuable for the miniaturization of high-resolution spectrometers17. Nevertheless, integrated devices lack the ability to probe light over a broad spectral range due to technological limitations. In this work, we present the development of an integrated waveguide spectrometer without any external or moving parts and which is not limited by the Shannon–Nyquist sampling theorem. An on-chip interferogram scanning technique is implemented by a controllable phase shift using the Pockels effect in LN, and the ability to simultaneously detect light over a broad spectral range of 500 nm has been demonstrated experimentally. The spectral resolution of the presented device with a footprint of 10 mm2 is R = 290. However, by increasing the length of the sampled interferogram the resolution can be readily improved without any increase in power consumption. Furthermore, the use of an inherently imbalanced interference circuit would make it possible to sample only one side of the symmetric interferogram, leading to another doubling of the resolution.

Another key parameter is the required voltage to stitch the sampled intensity of adjacent EFSs. The current device design results in stitching voltages of ~20 V. The most straightforward path to reduce the power consumption would be to extend the electrode length, which translates linearly to the acquired phase difference of the propagating modes. Compared to integrated traditional FTSs, such as the thermo-optic approach on silicon13 that offers large-scale and cost-efficient production, the resolution of the electro-optic spectrometer presented here can be enhanced without any increase in power consumption.

Further efforts should be made to develop the fabrication of the EFSs. For prototyping and demonstrating the spectrometer principle, deposition with FIB proved to be reliable, but this approach is not suitable for wafer-scale fabrication and so substitution with techniques like EBL or nanoimprint lithography is envisaged38. Also, by engineering the scattering properties of a single EFS, the efficiency of the sampling and ultimately the dynamic range of the device could be enhanced (Supplementary Section 6). Besides, a better signal-to-noise ratio of the recorded images would facilitate the retrieval and stitching of the interferogram, which should be optimized to reduce the side lobes in the measured spectra.

Finally, integration of the detection scheme is crucial to obtain more compact spectrometers, for example by direct bonding of a pixel array on top of the waveguide20. Moreover, developments could be focused on efficient spectral reconstruction using compressed sensing39 or relaxing the sampling constraints by dispersing the signal into narrow spectral channels40. In conclusion, this concept of a broadband high-resolution spectrometer can be an essential component for applications where size and complexity are key, such as spaceborne spectroscopy and remote sensing or integration in mobile devices.


Device fabrication

The device was fabricated on a commercially available x-cut lithium niobate-on-insulator thin film chip (NANOLN). The LN thin film with a thickness of 300 nm was separated from the bulk LN substrate by a 2-μm-thick cladding layer of silicon dioxide (SiO2). A 200 nm layer of Si3N4 was deposited on top by plasma-enhanced chemical vapour deposition followed by 100 nm of chromium by physical vapour deposition. The waveguide structures were patterned via EBL with a negative EBL resist (AR-N 7520.073). The waveguides were then transferred to the chromium hard mask with a chlorine-based recipe in an inductively coupled plasma RIE tool (ICP-RIE). Afterwards, the sample was etched by RIE and fluorine chemistry to define the SiN ridge waveguides. After removal of the remaining chromium mask, the electrode design was aligned and written by means of a direct write laser machine into a positive photoresist (AZ6612, 1.2 μm thick). Deposition of 10 nm of chromium and 300 nm of gold and a subsequent liftoff process provided the electrodes along the waveguides. The chip (typically 2 × 2 cm2) was then diced and the electrodes were wire-bonded in a push–pull configuration and connected to larger contact pads. Finally, the EFSs were deposited on the interference waveguide. This was done by FIB operated in deposition mode with a platinum precursor outgassed inside the chamber. During the same FIB process, the ridge waveguides were polished to obtain smooth facets for improved fibre-to-chip coupling.

Monochromatic and broadband measurements

Monochromatic characterization was carried out with a tunable laser source in the telecommunications range. The optical signal was end-fire coupled to the waveguide with a lensed fibre. For dual-wavelength measurements, a second laser source at a fixed wavelength of 1,550 nm was combined with the tunable source into the lensed fibre using a fibre wavelength division multiplexer. The scattered intensity pattern of the array of EFSs on top of the interference waveguide was recorded with an infrared FPA and a ×50 magnification objective (NA 0.65). The electrodes were contacted via probe needles and connected to a d.c. voltage source. Both the camera and the voltage source were controlled digitally to perform the electro-optical sampling of the interferogram. Broadband measurements were performed using a fibre-coupled SLED and a bandpass filter. The reference spectra were measured with an optical spectrum analyser.

Data availability

The data that support the plots within this paper and other findings of this study are available from the corresponding author upon reasonable request.


  1. 1.

    Blind, N., Le Coarer, E., Kern, P. & Gousset, S. Spectrographs for astrophotonics. Opt. Express 25, 27341–27369 (2017).

    ADS  Article  Google Scholar 

  2. 2.

    Manzardo, O., Herzig, H. P., Marxer, C. R. & de Rooij, N. F. Miniaturized time-scanning Fourier transform spectrometer based on silicon technology. Opt. Lett. 24, 1705–1707 (1999).

    ADS  Article  Google Scholar 

  3. 3.

    Knipp, D. et al. Silicon-based micro-Fourier spectrometer. IEEE Trans. Electron. Devices 52, 419–426 (2005).

    ADS  Article  Google Scholar 

  4. 4.

    Cheben, P. et al. A high-resolution silicon-in-insulator arrayed waveguide grating microspectrometer with sub-micrometer aperture waveguides. Opt. Express 15, 2299–2306 (2007).

    ADS  Article  Google Scholar 

  5. 5.

    Delâge, A. et al. Static Fourier-transform waveguide spectrometers. In 2009 11th International Conference on Transparent Optical Networks (ICTON) We.D1.1 (IEEE, 2009).

  6. 6.

    Ma, X., Li, M. & He, J.-J. CMOS-compatible integrated spectrometer based on echelle diffraction grating and MSM photodetector array. IEEE Photon. J. 5, 6600807 (2013).

    ADS  Article  Google Scholar 

  7. 7.

    Kyotoku, B. B. C., Chen, L. & Lipson, M. Sub-nm resolution cavity enhanced micro-spectrometer. Opt. Express 18, 102–107 (2009).

    ADS  Article  Google Scholar 

  8. 8.

    Babin, S. et al. Digital optical spectrometer-on-chip. Appl. Phys. Lett. 95, 041105 (2009).

    ADS  Article  Google Scholar 

  9. 9.

    Peroz, C. et al. Multiband wavelength demultiplexer based on digital planar holography for on-chip spectroscopy applications. Opt. Lett. 37, 695–697 (2012).

    ADS  Article  Google Scholar 

  10. 10.

    Griffiths, P. R. & de Haseth, J. A. Fourier Transform Infrared Spectrometry (Wiley, 2007).

  11. 11.

    Li, J., Lu, D.-f & Qi, Z.-m. Miniature Fourier transform spectrometer based on wavelength dependence of half-wave voltage of a LiNbO3 waveguide interferometer. Opt. Lett. 39, 3923–3926 (2014).

    ADS  Article  Google Scholar 

  12. 12.

    Dong, B. et al. Nano-silicon-photonic Fourier transform infrared (FTIR) spectrometer-on-a-chip. In Conference on Lasers and Electro-Optics (CLEO) STu4I.1 (OSA, 2015).

  13. 13.

    Souza, M. C. M. M., Grieco, A., Frateschi, N. C. & Fainman, Y. Fourier transform spectrometer on silicon with thermo-optic non-linearity and dispersion correction. Nat. Commun. 9, 665 (2018).

    ADS  Article  Google Scholar 

  14. 14.

    Florjańczyk, M. et al. Planar waveguide spatial heterodyne spectrometer. Proc. SPIE 6796, 67963J (2007).

    Article  Google Scholar 

  15. 15.

    Velasco, A. V. et al. High-resolution Fourier-transform spectrometer chip with microphotonic silicon spiral waveguides. Opt. Lett. 38, 706–708 (2013).

    ADS  Article  Google Scholar 

  16. 16.

    Nedeljkovic, M. et al. Mid-infrared silicon-on-insulator Fourier-transform spectrometer chip. IEEE Photon. Technol. Lett. 28, 528–531 (2015).

    ADS  Article  Google Scholar 

  17. 17.

    Le Coarer, E. et al. Wavelength-scale stationary-wave integrated Fourier-transform spectrometry. Nat. Photon. 1, 473–478 (2007).

    ADS  Article  Google Scholar 

  18. 18.

    Bonneville, C. et al. SWIFTS: a groundbreaking integrated technology for high-performance spectroscopy and optical sensors. Proc. SPIE 8616, 86160M (2013).

    Article  Google Scholar 

  19. 19.

    Nie, X., Ryckeboer, E., Roelkens, G. & Baets, R. CMOS-compatible broadband co-propagative stationary Fourier transform spectrometer integrated on a silicon nitride photonics platform. Opt. Express 25, A409–A418 (2017).

    ADS  Article  Google Scholar 

  20. 20.

    Ferrand, J. et al. A SWIFTS operating in visible and near-infrared. Proc. SPIE 7010, 701046 (2008).

    Article  Google Scholar 

  21. 21.

    Osowiecki, G. D. et al. Standing wave integrated Fourier transform spectrometer for imaging spectrometry in the near infrared. Proc. SPIE 9611, 96110P (2015).

    Article  Google Scholar 

  22. 22.

    Madi, M. et al. Lippmann waveguide spectrometer with enhanced throughput and bandwidth for space and commercial applications. Opt. Express 26, 2682–2707 (2018).

    ADS  Article  Google Scholar 

  23. 23.

    Ferrand, J. et al. Stationary wave integrated Fourier transform spectrometer (SWIFTS). Proc. SPIE 7604, 760414 (2010).

    Article  Google Scholar 

  24. 24.

    Thomas, F. et al. Expanding sampling in a SWIFTS-Lippmann spectrometer using an electro-optic Mach–Zehnder modulator. Proc. SPIE 9516, 95160B (2015).

    Google Scholar 

  25. 25.

    Loridat, J. et al. All integrated lithium niobate standing wave Fourier transform electro-optic spectrometer. J. Lightwave Technol. 36, 4900–4907 (2018).

    ADS  Article  Google Scholar 

  26. 26.

    Schmidt, R. & Kaminow, I. Metal-diffused optical waveguides in LiNbO3. Appl. Phys. Lett. 25, 458–460 (1974).

    ADS  Article  Google Scholar 

  27. 27.

    Thomas, F. et al. First results in near and mid IR lithium niobate-based integrated optics interferometer based on SWIFTS-Lippmann concept. J. Lightwave Technol. 32, 4338–4344 (2014).

    Article  Google Scholar 

  28. 28.

    Lyons, R. G. Understanding Digital Signal Processing. (Pearson Education, 2011).

  29. 29.

    Reig Escalé, M., Pohl, D., Sergeyev, A. & Grange, R. Extreme electro-optic tuning of Bragg mirrors integrated in lithium niobate nanowaveguides. Opt. Lett. 43, 1515–1518 (2018).

    ADS  Article  Google Scholar 

  30. 30.

    Wang, C. et al. Integrated lithium niobate electro-optic modulators operating at CMOS-compatible voltages. Nature 562, 101–104 (2018).

    ADS  Article  Google Scholar 

  31. 31.

    He, M. et al. High-performance hybrid silicon and lithium niobate Mach–Zehnder modulators for 100 Gbit s−1 and beyond. Nat. Photon. 13, 359–364 (2019).

    ADS  Article  Google Scholar 

  32. 32.

    Bazzan, M. & Sada, C. Optical waveguides in lithium niobate: recent developments and applications. Appl. Phys. Rev. 2, 040603 (2015).

    ADS  Article  Google Scholar 

  33. 33.

    Ji, X. et al. Ultra-low-loss on-chip resonators with sub-milliwatt parametric oscillation threshold. Optica 4, 619–624 (2017).

    ADS  Article  Google Scholar 

  34. 34.

    Chang, L. et al. Thin film wavelength converters for photonic integrated circuits. Optica 3, 531–535 (2016).

    ADS  Article  Google Scholar 

  35. 35.

    Ahmed, A. N. R., Shi, S., Zablocki, M., Yao, P. & Prather, D. W. Tunable hybrid silicon nitride and thin-film lithium niobate electro-optic microresonator. Opt. Lett. 44, 618–621 (2019).

    ADS  Article  Google Scholar 

  36. 36.

    Chang, L. et al. Heterogeneous integration of lithium niobate and silicon nitride waveguides for wafer-scale photonic integrated circuits on silicon. Opt. Lett. 42, 803–806 (2017).

    ADS  Article  Google Scholar 

  37. 37.

    Porter, C. D. & Tanner, D. B. Correction of phase errors in Fourier spectroscopy. Int. J. Infrared Millim. Waves 4, 273–298 (1983).

    ADS  Article  Google Scholar 

  38. 38.

    Zgraggen, E. et al. Optical properties of waveguide-coupled nanowires for sub-wavelength detection in microspectrometer applications. J. Opt. 17, 025801 (2015).

    ADS  Article  Google Scholar 

  39. 39.

    August, I., Oiknine, Y., AbuLeil, M., Abdulhalim, I. & Stern, A. Miniature compressive ultra-spectral imaging system utilizing a single liquid crystal phase retarder. Sci. Rep. 6, 23524 (2016).

    ADS  Article  Google Scholar 

  40. 40.

    Hong, B., Monifi, F. & Fainman, Y. Channel dispersed Fourier transform spectrometer. Commun. Phys. 1, 34 (2018).

    Article  Google Scholar 

Download references


We acknowledge support for nanofabrication from the Scientific Center of Optical and Electron Microscopy ScopeM and from the cleanroom facilities BRNC and FIRST of ETH Zürich. This project was initially funded by the Swiss Space Office at the State Secretariat for Education, Research and Innovation, in the frame of the Mesures de Positionnement MdP2016 funding scheme. This project has received funding from the European Union’s Horizon 2020 research and innovation programme from the European Research Council under grant no. 714837 (Chi2-nano-oxides). This work was also supported by Swiss National Science Foundation grant no. 150609. We are also grateful to the Swiss Space Center for thoughtful inputs during the project progress review meetings. We thank I. Shorubalko for contributing to the prototype preparation and G. Scalari and J. Faist for helpful discussions.

Author information




B.G. and M.M. developed the original idea of directly combining electro-optic actuators with the waveguide spectrometers to carry out integrated interferogram scanning. M.M., A.S., U.M., E.A. and R.G. conceived the project. D.P., M.R.E., F.K. and A.S. developed the fabrication process. D.P., M.R.E., M.M., F.K., P.B. and A.S. designed the set-up and performed the experiments. P.G. and B.G. provided technical advice in the course of the project. D.P., M.R.E. and R.G. wrote the manuscript. M.M. and R.G. supervised the project. U.M. managed the administrative and financial aspects of the project.

Corresponding author

Correspondence to David Pohl.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

This file contains more information about the work and Supplementary Figs. 1–5.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Pohl, D., Reig Escalé, M., Madi, M. et al. An integrated broadband spectrometer on thin-film lithium niobate. Nat. Photonics 14, 24–29 (2020).

Download citation

Further reading


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing