Infrared tunable diode-laser absorption spectroscopy (IR-TDLAS) is an enabling technology for trace-gas detection, with applications ranging from air-quality monitoring to medical diagnostics. However, such sensors typically utilize discrete optical components that pose practical cost limits for large-scale network deployments. Here, we leverage silicon photonics technology to demonstrate IR-TDLAS on an integrated CMOS-compatible platform for methane () spectroscopy. Using near-IR (1650 nm) light from a distributed-feedback laser and an uncooled InGaAs detector, the evanescent optical field of a high-index contrast nanoscale silicon waveguide is used to probe ambient , yielding Gaussian-noise-limited sub-100 parts-per-million by volume detection limits. Our results demonstrate the feasibility of chip-scale photonic integration for realizing compact, cost-effective, and versatile gas sensors capable of tackling diverse energy and environmental challenges, such as natural-gas leak quantification and localization for fugitive-emissions monitoring.
© 2017 Optical Society of America
Optical detection of trace gases has been demonstrated in a wide variety of sensing configurations [1–6], ranging from open-path free-space fiber-based point spectrometers  to surface-enhanced on-chip optical sensors [3,4]. In particular, by targeting rovibrational transitions of molecular compounds in the near- to mid-IR spectral “fingerprint” region , infrared tunable diode-laser absorption spectroscopy (IR-TDLAS) has been demonstrated as a tool unrivalled in both sensitivity and species selectivity for a wide variety of spectroscopic applications [1,5]. Recent sensor embodiments for signal enhancement and noise reduction [2,6] have allowed IR-TDLAS systems to reach parts-per-trillion-by-volume-level detection limits [1,6], thus demonstrating absorption-based sensors as an alternative to nonselective chemical sensors and more costly mass-spectrometry systems.
A particularly important application space involves environmental and greenhouse-gas sensing, which have recently undergone significant development in light of increasing emissions control regulations. A pertinent example involves the reduction of methane (), a main constituent of natural gas, as climate impact studies have indicated its severe radiative forcing and global warming potential . Mitigation of fugitive emissions is thus a key requirement toward the viability of natural-gas harvesting as a clean energy alternative. Ideally, wide-area deployment of autonomous and low-power sensors in a network configuration will provide both temporal and spatial resolution for emissions quantification and localization. However, conventional IR-TDLAS implementations for leak detection are in the mobile and/or handheld configuration, which limits their utility for continuous, spatially resolved monitoring in unattended and remote locations. The lack of manufacturing scalability is presently a limiting factor to wide-area deployment of absorption-based sensors; typical IR-TDLAS systems utilize discrete optical components [1,2,6] with tight alignment tolerances, resulting in size, complexity, and cost limitations to large-scale manufacturing.
Here, we leverage recent developments in silicon photonics  to demonstrate an on-chip silicon photonic waveguide absorption spectrometer (SPWAS) for in the telecommunications U-band (1650 nm) . The high-index contrast achievable within silicon (Si) waveguides on silicon dioxide () buried oxide (BOx) provides an ideal platform for the design and integration of high-density nanoscale photonic and electronic circuitry for optical sensing applications [9,10]. Various integrated optical devices, such as waveguides , modulators , phase shifters , and switches , have been previously demonstrated and applied for high-data-rate optical interconnect applications. Moreover, numerous foundries have demonstrated silicon photonics manufacturing within their existing commercial CMOS infrastructure, illustrating the transition to volume wafer-scale processing. We exploit these technical and manufacturing advances to demonstrate the first scalable on-chip IR-TDLAS sensor with sub-100 parts-per-million by volume (ppmv) sensitivity. Our results indicate the potential for integrated photonic chip sensors in large-scale, cost-effective sensor deployments, providing sensitivity and molecular specificity for measuring atmospheric gas constituents and pollutants.
A schematic of the integrated SPWAS sensor is shown in Fig. 1(a), consisting of a 10 cm long nanoscale Si waveguide defined in the device layer of a silicon-on-insulator (SOI) substrate, using conventional deep-UV (DUV) photolithography within a 200 mm silicon photonics process flow. The total footprint of the 10 cm waveguide device is , yielding thousands of sensors in a single fabricated wafer, even without an optimized layout. Despite the larger spatial footprint than alternative configurations (e.g., microring resonators  or photonic-crystal structures ), our nonresonant sensor design approach presents the advantage of robustness to fabrication tolerances and does not require active temperature stabilization. The top of the waveguide is exposed to the ambient and provides the interaction region, in which the evanescent field of the optical mode probes the rovibrational transitions of atmospheric . A scanning-electron microscope image of the sensor waveguide cross section is shown in Fig. 1(b), and the simulated transverse electric field component of the transverse magnetic (TM) mode is shown in Fig. 1(c). The long Si waveguide with low optical confinement forms the integrated counterpart of a free-space optical path in conventional TDLAS sensors and represents the central building block for extending CMOS-compatible photonic technologies to optical trace-gas sensing.
The optical interaction within the spatial vicinity of the Si waveguide induces propagation loss in accordance with the Beer–Lambert law and provides ideal narrow spectral signatures for high-resolution spectroscopy. Examples of absorption spectra of and the other main constituents of natural gas, extracted from high-resolution transmission molecular database (HITRAN)  and Pacific Northwest National Laboratory (PNNL)  databases, are shown in Fig. 1(d). Our SPWAS sensor probes the R(4) line in the overtone band of at 1650.96 nm (), where spectral cross talk with water vapor and other long-chain hydrocarbons is negligible.
The evanescent field of the waveguide is designed for low optical confinement to obtain a large modal overlap with the ambient atmosphere. Here, is defined as the modal power fraction residing in the air cladding and must be as high as possible to maximize the effective optical path length and therefore the absorption contrast through the waveguide. This is achieved when the waveguide is designed with a low effective index () close to cutoff, i.e., where of the propagating mode approaches the refractive index of the BOx () and the light is no longer guided. The unique high index contrast of the platform allows waveguides with large to be designed [9,10]; lower index contrast waveguides, such as silicon nitride (SiN) , are typically limited to smaller and therefore weaker absorption signals. Numerical simulations of the SPWAS waveguide and are performed for various structural parameters, with a predefined substrate Si thickness of 250 nm to utilize existing fabrication processes developed for optical interconnect applications. Based upon modal simulations, a waveguide width of 430 nm is chosen to guide a TM mode with and , which is approximately 15% above to provide a margin for fabrication tolerances. For a physical waveguide length , this corresponds to free-space equivalent optical path length of 2.83 cm. It is important to note that this simulated accounts for group index enhancement of the evanescent field overlap, with recent studies further extending the use of “slow light” for signal enhancement in sensing applications .
The sensing cross section (the area in the air cladding where the ratio of the normalized electric field to the maximum value at the upper interface is greater than ) is , which represents a volume of over the 10 cm long waveguide. At standard temperature pressure conditions and for a concentration of 10 ppm, this means that molecules are probed within the evanescent field region of the propagating optical mode.
The SPWAS sensor performance is tested and validated in an environmental chamber with controlled concentration [Fig. 2(a)]. Input light from a distributed-feedback (DFB) laser chip is current-ramped at 100 Hz around the R(4) transition and coupled into the SPWAS, which is mounted within the chamber with piezo-aligned lensed fibers on each side for coupling light into and out of the waveguide. A fiber-coupled free-space open-path sensor with a 10 cm optical path length is also placed within the test chamber as a concurrently measured reference for direct validation of the SPWAS concentration retrievals. Upon propagating through the waveguide and interacting with the chamber gas sample, the light is collected by an uncooled InGaAs detector. From an initial optical power of , the measured waveguide loss of and 8 dB coupling loss per facet yields an output optical power of incident on the photodiode. During testing, the chamber is sealed and purged with the desired gas atmosphere; Fig. 2(b) shows concurrently acquired transmission spectra of both the SPWAS and free-space reference sensor using 1.5% in the chamber, along with the corresponding Voigt profile least-mean squares spectral fit for concentration retrieval with an adaptive fringe removal algorithm (Supplement 1). To extract of the SPWAS, note that 6.3% of the light intensity is absorbed at the transition peak in the reference sensor, while a peak absorption of 1.6% is measured through the SPWAS sensor. Since the SPWAS sensor and free-space reference path have the same 10 cm physical length, the ratio of their relative absorptions can be used to extract the overlap factor of the waveguide, yielding .
The linearity of the SPWAS sensor is validated by comparing its relative absorption to the free-space reference for concentrations ranging from 0.2% to 1.4% [rising step in Fig. 3(a)]. The chamber is first flushed with nitrogen and gradually filled with , demonstrating that the steady-state concentration extracted from the SPWAS is consistent with the reference and therefore reliably reflects the true concentration value. Indeed, the response of the SPWAS peak absorption variation shown in Fig. 3(b) shows excellent linearity () with respect to the free-space reference. In contrast to Fig. 2(b) where was calculated from a single measurement, here it can be extracted from the slope of the correlation over multiple absorption values and concentrations, yielding an average overlap factor of 25.5%. While this is slightly lower than the simulated value of , both experimental assessments of presented here are highly consistent. The difference between simulated and experimental values is attributable to fabrication variations and uncertainty in the exact waveguide dimensions over the 10 cm SPWAS length.
The minimum detectable concentration using our waveguide sensor is evaluated by characterizing the SPWAS noise in the absence of in the test chamber. The inset of Fig. 4 shows a time series of the retrieved concentration over 2400 s (40 min) with a temporal resolution of 5.7 s (5 s raw data, 0.7 s acquisition processing time). The extracted concentration is centered on zero and dominated by white noise. Indeed, Allan-variance analysis indicates performance, with a bandwidth-normalized sensitivity of , corresponding to a minimum fractional absorption of , which is consistent with conventional IR-TDLAS systems . Detection limits below 100 ppmv can be achieved for integration times , reaching at , thus validating the SPWAS as a promising integrated alternative for environmental trace-gas sensing. Spectral density analysis indicates significant noise contributions from mechanical vibrations at the lensed fiber coupling interfaces, and it is expected that full integration of the laser and detector will reduce sensor noise and target fractional absorptions , thus enabling sensitivities of the order of .
It is important to highlight the fact that our SPWAS architecture is ideal for integration using photonic packaging techniques from the telecommunications industry. The DFB laser chip does not require moving mechanical parts, and work is presently underway for source and detector integration via self-aligned flip-chip III-V die assembly on a Si chip substrate . In conjunction with an integrated hybrid III-V/Si external laser cavity configuration for wide tunable single-mode emission , we anticipate a fully integrated TDLAS sensor without the need for active cooling of the laser chip, resulting in significant size, weight, and power (SWaP) advantages.
To conclude, we have demonstrated absorption spectroscopy using an on-chip platform by leveraging silicon photonic technologies available for telecommunications and optical interconnect applications. Migration of IR-TDLAS sensors onto a CMOS-compatible platform is expected to provide substantial SWaP and cost benefits, thus enabling a next generation of highly sensitive and selective on-chip sensors manufactured in a scalable process for large-volume applications. Additionally, our SPWAS is extensible beyond the near-IR; indeed, with novel waveguide architectures to avoid absorption from the buried oxide, the transparency of silicon up to 8 μm  allows our SPWAS to be applied to molecules in the mid-IR range, thus presenting on-chip silicon photonic integrated sensors as an enabling platform for broadband, multispecies detection.
We thank Dr. Josephine Chang (IBM Research, currently at Northrop Grumman) for assistance with the acquisition of the scanning electron micrographs, Joe Green for generation of the rendered images, and Gerard Wysocki (Princeton University) for helpful discussions.
See Supplement 1 for supporting content.
1. J. Hodgkinson and R. P. Tatam, Meas. Sci. Technol. 24, 012004 (2013). [CrossRef]
2. B. Bernhardt, A. Ozawa, P. Jacquet, M. Jacquey, Y. Kobayashi, T. Udem, R. Holzwarth, G. Guelachvili, T. W. Hansch, and N. Picque, Nat. Photonics 4, 55 (2010). [CrossRef]
3. F. T. Dullo, S. Lindecrantz, J. Jágerská, J. H. Hansen, M. Engqvist, S. A. Solbø, and O. G. Hellesø, Opt. Express 23, 31564 (2015). [CrossRef]
4. T. H. Stievater, M. W. Pruessner, D. Park, W. S. Rabinovich, R. A. McGill, D. A. Kozak, R. Furstenberg, S. A. Holmstrom, and J. B. Khurgin, Opt. Lett. 39, 969 (2014). [CrossRef]
5. E. J. Zhang, L. Tombez, J. S. Orcutt, S. Kamlapurkar, G. Wysocki, and W. M. J. Green, in Conference on Lasers and Electro-Optics, San Jose, California, 2016, paper SF2H.1.
6. P. Weibring, D. Richter, A. Fried, J. G. Walega, and C. Dyroff, Appl. Phys. B 85, 207 (2006). [CrossRef]
7. R. A. Alvarez, S. W. Pacala, J. J. Winebrake, W. L. Chameides, and S. P. Hamburg, Proc. Natl. Acad. Sci. USA 109, 6435 (2012). [CrossRef]
8. R. Soref, IEEE J. Quantum Electron. 12, 1678 (2006). [CrossRef]
9. A. Densmore, D. X. Xu, P. Waldron, S. Janz, P. Cheben, J. Lapointe, A. Delage, B. Lamontagne, J. H. Schmid, and E. Post, IEEE Photon. Technol. Lett. 18, 2520 (2006). [CrossRef]
10. J. T. Robinson, K. Preston, O. Painter, and M. Lipson, Opt. Express 16, 16659 (2008). [CrossRef]
11. V. R. Almeida, C. A. Barrios, R. R. Panepucci, and M. Lipson, Nature 431, 1081 (2004). [CrossRef]
12. G. T. Reed, G. Mashanovich, F. Y. Gardes, and D. J. Thompson, Nat. Photonics 4, 518 (2010). [CrossRef]
13. A. Liu, R. Jones, L. Liao, D. Samara-Rubio, D. Rubin, O. Cohen, R. Nicolaescu, and M. Paniccia, Nature 427, 615 (2004). [CrossRef]
14. N. Dupuis, A. V. Rylyakov, C. L. Schow, D. M. Kuchta, C. W. Baks, J. S. Orcutt, D. M. Gill, W. M. J. Green, and B. G. Lee, J. Lightwave Technol. 35, 615 (2017). [CrossRef]
15. A. Nitkowski, L. Chen, and M. Lipson, Opt. Express 16, 11930 (2008). [CrossRef]
16. W. Lai, S. Chakravarty, X. Wang, C. Lin, and R. T. Chen, Opt. Lett. 36, 984 (2011). [CrossRef]
17. L. S. Rothman, R. R. Gamache, A. Goldman, L. R. Brown, R. A. Toth, H. M. Pickett, R. L. Poynter, J.-M. Flaud, C. Camy-Peyret, A. Barbe, N. Husson, C. P. Rinsland, and M. A. H. Smith, Appl. Opt. 26, 4058 (1987). [CrossRef]
18. S. W. Sharpe, T. J. Johnson, R. L. Sams, P. M. Chu, G. C. Rhoderick, and P. A. Johnson, Appl. Spectrosc. 58, 1452 (2004). [CrossRef]
19. L. Stern, B. Desiatov, I. Goykhman, and U. Levy, Nat. Commun. 4, 1548 (2013). [CrossRef]
20. T. Barwicz, Y. Taira, T. W. Lichoulas, N. Boyer, Y. Martin, H. Numata, J.-W. Nah, S. Takenobu, A. Janta-Polczynski, E. L. Kimbrell, R. Leidy, M. H. Khater, S. Kamlapurkar, S. Engelmann, Y. A. Vlasov, and P. Fortier, IEEE J. Sel. Top. Quantum Electron. 22, 455 (2016). [CrossRef]
21. A. J. Zilkie, P. Seddighian, B. J. Bijlani, W. Qian, D. C. Lee, S. Fathololoumi, J. Fong, R. Shafiiha, D. Feng, B. J. Luff, X. Zheng, J. E. Cunningham, A. V. Krishnamoorthy, and M. Asghari, Opt. Express 20, 23456 (2012). [CrossRef]
22. R. A. Soref, S. J. Emelett, and W. R. Buchwald, J. Opt. A 8, 840 (2006). [CrossRef]