4420
OPTICS LETTERS / Vol. 39, No. 15 / August 1, 2014
Chirped laser dispersion spectroscopy with differential frequency generation source Michal Nikodem,1,* Karol Krzempek,2 Renata Karwat,1 Grzegorz Dudzik,2 Krzysztof Abramski,2 and Gerard Wysocki3 1 2
Wroclaw Research Center EIT+, ul. Stablowicka 147, 54-066 Wroclaw, Poland
Laser&Fiber Electronics Group, Wroclaw University of Technology, Wybrzeze Wyspianskiego 27, 50-370 Wroclaw, Poland 3 Princeton University, Engineering Quadrangle, 08544 Princeton, New Jersey, USA *Corresponding author:
[email protected] Received May 20, 2014; accepted June 23, 2014; posted June 25, 2014 (Doc. ID 212367); published July 23, 2014 A feasibility study of open-path methane detection at 3.4 μm using chirped laser dispersion spectroscopy (CLaDS) based on nonlinear differential frequency generation (DFG) laser source is performed. Application of a DFG source based on telecom laser sources and modulators allows mid-infrared CLaDS system to be optimized for measurements of gases at atmospheric conditions for which modulation in the GHz range is required. Excellent agreement between observed CLaDS signals and spectroscopic models has been observed. © 2014 Optical Society of America OCIS codes: (300.6390) Spectroscopy, molecular; (300.6310) Spectroscopy, heterodyne; (230.7405) Wavelength conversion devices; (280.3420) Laser sensors. http://dx.doi.org/10.1364/OL.39.004420
Laser spectroscopy is a powerful tool for quantitative chemical detection [1–3]. Particularly interesting is the spectral region located between 3 and 3.5 μm where multiple hydrocarbons can be detected, including methane, propane, ethane, acetylene, and others [4–6]. Historically, these ro-vibrational transitions were accessed using laser sources based on differential frequency generation (DFG) [7,8], but recently, a rapid development of semiconductor lasers in this spectral region was observed [9–13]. Laser diodes, or intraband cascade lasers, can provide high optical powers, compact size, and low energy consumption. Nonetheless, DFGbased sources can still offer many valuable properties and leverage well-established reliable technologies to provide mid-IR radiation. Waves in the 3–3.5 μm range can be generated using near-IR sources that enable using commercially available Er- and Yb-doped fiber amplifiers and periodically poled lithium niobate (PPLN) crystals for nonlinear optical conversion to obtain up to hundreds of microwatt, optical power in the mid-IR [14]. For many spectroscopic systems this level is sufficient to perform spectroscopic measurements using highly sensitive photodetectors available for this spectral range (including thermoelectrically cooled detector elements). Most importantly the DFG-process that utilizes near-IR DFB or DBR sources is very flexible in terms of wavelength selection and wavelength modulation/tuning [15,16]. Telecom laser diodes are commercially available at any wavelength in the 1530–1570 nm range and their lifetime, reliability, and ease of operation are truly outstanding. It should also be noted that broad availability of other optical components in the telecom spectral range such as optical fibers, modulators, and polarization optics is also of importance for building reliable spectroscopic systems. Recently we have introduced the chirped laser dispersion spectroscopy (CLaDS), technique that in contrast to traditional absorption-sensing methods relies on sensing of optical dispersion caused by molecular transitions [17]. In CLaDS multiple waves (usually two 0146-9592/14/154420-04$15.00/0
or three) separated by Ω are simultaneously frequency-chirped and, after traveling through the gas sample, are focused onto the photodetector where an RF beatnote signal at Ω is created. The molecular dispersion affects the propagation of those waves resulting in deviation of the RF beatnote from Ω as the optical waves are chirped across molecular resonance. Because spectroscopic information is encoded into the frequency (not amplitude) of the RF beatnote, CLaDS provides not only increased immunity to optical power fluctuations but also baseline-free detection (no molecular dispersion = no signal). This is a major advantage with respect to absorption-based techniques, including wavelength modulation spectroscopy (WMS). For example, if WMS is used with DFG systems due to nonflat and nonlinear PPLN efficiency characteristics higher order harmonic signals can be detected even without molecular absorption. In contrast, the signal amplitude in CLaDS (retrieved though frequency demodulation of the photodetector signal) is not affected by changes in the optical power level; therefore, no power normalization is necessary and thus less frequent system calibrations are needed. In this respect, application of CLaDS for spectroscopic detection with DFG-based sources can mitigate some of the issues observed if absorption-based sensing is utilized. At the same time application of DFG sources allows us to overcome many drawbacks observed previously in mid-IR CLaDS systems based on quantum cascade lasers. DFG-based CLaDS system can easily operate with Ω in the GHz range by simply using highspeed telecom modulators in the ∼1.55 μm branch [18,19], while this was difficult in previously demonstrated mid-IR instruments based on acousto-optical modulators providing Ω in the tens of MHz range [20]. Larger Ω effectively increases the CLaDS signal amplitude for transitions at atmospheric conditions exhibiting linewidths of >1 GHz [21]. In this Letter we demonstrate feasibility of DFG-based CLaDS detection in open-path that combines the strengths of both approaches presented previously: © 2014 Optical Society of America
August 1, 2014 / Vol. 39, No. 15 / OPTICS LETTERS
access to the fundamental mid-IR transitions [20], and an advantage of optimum performance achieved with telecom-based components [18]. System performance is analyzed with the 2948 cm−1 radiation (approximately 3.4 μm) targeting methane (CH4 ) transitions at atmospheric conditions. This proof-of-concept study demonstrates the baseline-free nature of CLaDS and the immunity of CLaDS signal amplitude to changes in light transmission, and discusses main limitations of the presented DFG-based CLaDS system. A schematic diagram of the optical setup is shown in Fig. 1. The signal beam is produced by a distributed feedback laser diode (DFB LD) operating around 1.55 μm, which is frequency-chirped through sinusoidal current modulation at f m . An additional ramp signal is used to enable recording full chirp-modulated (CM)-CLaDS spectrum (more details about CM-CLaDS and its advantages with respect to direct-CLaDS approach can be found in Ref. [22]). A high-speed modulation at frequency Ω is performed using an electro-optical intensity modulator, which creates a multicolor beam (optical carrier with sidebands separated by Ω) that is further amplified using Er-Yb doped double-clad (DC) fiber amplifier (EYDFA). The active medium is a 5 m long Nufern fiber SM-EYDF-6/125-HE, with high power isolators both at the input and output to prevent unwanted back-reflections. The pump beam used for DFG is produced with a single frequency solid state laser (SSL; previously described in Ref. [23]), coupled into the single mode fiber and amplified with an Yb-doped fiber amplifier (YDFA). The YDFA was a 5 m long DC Nufern fiber SM-YDF-5/130-VIII. Both EYDFA and EDFA were pumped with a 10 W multimode laser diodes. The two amplified beams were collimated using fiber collimators, combined with a dichroic mirror and focused onto 40 mm long PPLN crystal using 150 mm focal length lenses. Crystal was placed in the temperature-stabilized oven. The midinfrared radiation (∼100 μW) was transmitted through a Ge filter (to block residual 1.06 and 1.55 μm light), collimated using CaF2 lens and focused onto a fast TEC-cooled MCT photodetector (Vigo, model PVI-4TE-8-0.5 × 0.5). The total sensing path of 450 cm in the laboratory air was obtained using two 2-inch aluminum-coated mirrors. The heterodyne beatnote at Ω was frequency demodulated using an RF spectrum analyzer (Rohde&Schwarz) and
Fig. 1. Diagram of a DFG-based CM-CLaDS instrument for methane sensing. (MZM, Mach–Zehnder modulator; PD, photodetector; DM, dichroic mirror).
4421
Fig. 2. Absorption spectrum of 1.8 ppmv of methane and 1% of water vapor based on the HITRAN database (pressure 760 Torr, path length 450 cm).
LabVIEW program was used to retrieve the 2f component of the frequency demodulated CM-CLaDS signal. Shown in Fig. 2 is the simulated absorption spectrum of methane and water vapor between 2947 and 2949 cm−1 (based on HITRAN database; CH4 concentration of 1.8 ppmv, 450 cm path length, 760 Torr pressure). The target methane spectral feature has five closely spaced and relatively strong peaks. More importantly, these transitions only slightly overlap with weak water vapor resonances. (The two most useful CH4 lines between 2948 and 2948.5 cm−1 are virtually free of any interference.) Therefore, we decided to use this spectral region to demonstrate feasibility of CH4 open-path detection with DFG-based CLaDS system. Essentially, CLaDS is a technique in which molecular dispersion is detected through measurement of the phase change between optical waves traveling through the sample. As a result, there is a relation between the measured CM-CLaDS signal amplitude and the optical sideband separation Ω that must be matched to the transition linewidth. (More details can be found in Ref. [21]). Additionally since the CM-CLaDS signal amplitude is chirp dependent, the chirp modulation frequency and/ or the optical frequency modulation depth f d should also be optimized [22]. To optimize this system we placed a 5cm-long gas cell within the optical path and CM-CLaDS spectra were collected at different setup settings (f d and Ω). With the cell filled with an air∕CH4 mixture at atmospheric pressure, an indication of the presence of methane was clearly detected and recorded. (Peak absorption was above 10%.) Figure 3 shows three 2f CM-CLaDS spectra recorded for different Ω and f d . To record full spectra laser, wavelength was scanned by changing the current of the 1.55 μm laser from 130 to 210 mA. Other parameters were chirp modulation frequency was f m 40 kHz, optical frequency modulation depth f d was set to 1.2 or 2.4 GHz, averaging time for each point was 10 ms, demodulation bandwidth was set to 200 kHz, and highspeed modulation (equal to the sidebands’ separation) at frequency Ω was set to 600 or 1200 MHz.
4422
OPTICS LETTERS / Vol. 39, No. 15 / August 1, 2014
Fig. 5. Allan variance analysis of the system and CM-CLaDS time series (inset) with the laser wavelength tuned to the transition peak. Data were acquired every 200 ms with 5% duty cycle. Fig. 3. CM-CLaDS spectra recorded for (a) Ω 600 MHz, f d 1.2 GHz; (b) Ω 1.2 GHz, f d 1.2 GHz; (c) Ω 1.2 GHz, f d 2.4 GHz. Black points: measured data, red line: simulation based on HITRAN database. All plots are normalized with respect to the maximum value in plot (c).
Obtained experimental results are in perfect agreement with the HITRAN-based simulation (also shown in Fig. 3), including shape of recorded spectral features as well as their amplitudes. Noteworthy is the baselinefree nature of CM-CLaDS that significantly simplifies data analysis (no baseline correction is needed). Figure 4 shows three full CM-CLaDS spectra recorded after the gas cell was removed from the setup (thus the sample was laboratory air at atmospheric pressure and path length was 450 cm—conditions similar to simulation shown in Fig. 2). In this measurement f d was set to
Fig. 4. Three 2f CM-CLaDS spectra recorded for different RF beatnote powers (−10, −20, and −30 dBm) that correspond to different optical powers reaching the photodetector (averaging: 10 ms∕point; for comparison HITRAN simulation is shown in red).
2.4 GHz and Ω was set to 900 MHz in order to operate within −3 dB electrical bandwidth of the photodetector. (For higher values of Ω, responsivity of the photodetector drops causing increased noise level and no further improvement in SNR). Three spectra were recorded for different RF powers (−10, −20, and −30 dBm, varied by applying small changes to the optical alignment). For comparison all experimental spectra were normalized to the maximum signal amplitude obtained for the RF power of −10 dBm. From Fig. 5 it is clear that recorded signal amplitude does not depend on optical power that reaches the photodetector. This is extremely beneficial property of CLaDS when it comes to open-path sensing, especially when application of DFG-based source is considered. (In such a case there are many potential sources of power fluctuations that will not influence the CM-CLaDS measurement.) In terms of noise, a small reduction of SNR is visible for RF power −30 dBm. Moreover, a comparison between the simulated spectrum (red line) and measured data (black dots) reveals the presence of a fringe pattern in the background. Most likely the origin of this fringe signal are unwanted reflections from the facets of the PPLN crystal and the germanium filter. When the DFG wavelength was tuned to the center of CH4 line at ∼2948.1 cm−1 the detection limit and the stability of the system could be characterized. Results are shown in Fig. 5. With a relatively strong fringe pattern and both lasers with no active wavelength control (i.e., no line locking) system drift is observed after the integration time of approximately 5 s. From the measured time series, the minimum detection limit of 9 ppbv × Hz−1∕2 can be calculated (assuming continuous data acquisition and ambient methane concentration of 1.8 ppmv during the experiment). In this Letter a DFG-based CM-CLaDS system was described and the feasibility of methane sensing in openpath configuration was demonstrated. Thanks to the high speed modulators available at telecom wavelengths the
August 1, 2014 / Vol. 39, No. 15 / OPTICS LETTERS
presented approach allows for using CLaDS in the mid-IR range at close-to-optimal conditions for gases at atmospheric pressures. (Ω in the ∼1 GHz range is required.) This configuration provides much higher signal amplitude compared to previously described mid-IR setups that used an acousto-optical modulator [20]. Presented experimental results are in very good agreement with the CM-CLaDS model. Immunity of signal amplitude to power fluctuations as well as the baseline-free nature of 2f CM-CLaDS signals were also observed. Optical fringes were identified as the main limitation for the system performance. Because the presented setup requires multiple optical components (fiber coupled laser, amplifier with multiple fiber splices, modulator, collimator, nonlinear crystal, germanium plate) it would be difficult (or even impossible) to remove all existing etalons. Two solutions that can be applied in order to minimize their influence are: selecting wavelength separation Ω carefully, so that visibility of fringes in the dispersion spectrum is reduced [24] or washing-up fringes by varying physical parameters of the setup and averaging the output signal at the same time [25]. If the influence of optical fringes is reduced and active control of the DFG wavelength is applied the presented setup can be used for open-path sensing of atmospheric methane with a relatively short optical path (less than 1 m). Other hydrocarbons with absorption/dispersion features in the 3.2–3.5 μm region can also be detected when different telecom laser diodes with appropriate operating wavelength are chosen. Moreover, since near-IR modulators can operate using up to tens of GHz, further improvements in the minimum detection limit of the presented system can be obtained when Ω > 1 GHz and sufficiently fast detectors are used. Spacing up to 2–2.5 GHz could be still detected with MCT devices [26], whereas quantum well infrared photodetectors (QWIP) would be needed for larger Ω [27]. M. N. and R. K. would like to acknowledge financial support by the Homing Plus award 2012-6/8, funded by the Foundation of Polish Science and co-financed by the European Regional Development Fund. G. W. would like to acknowledge financial support by the NSF CAREER award CMMI-0954897. The research fellowship of K. K. is co-financed by the EU as part of the European Social Fund (Grant Plus program). References 1. R. Curl, F. Capasso, C. Gmachl, A. Kosterev, B. McManus, R. Lewicki, M. Pusharsky, G. Wysocki, and F. K. Tittel, Chem. Phys. Lett. 487, 1 (2010). 2. B. Tuzson, S. Henne, D. Brunner, M. Steinbacher, J. Mohn, B. Buchmann, and L. Emmenegger, Atmos. Chem. Phys. 11, 1685 (2011).
4423
3. J. B. McManus, M. S. Zahniser, J. D. D. Nelson, J. H. Shorter, S. Herndon, E. Wood, and R. Wehr, Opt. Eng. 49, 111124 (2010). 4. P. Kluczynski, M. Jahjah, L. Nähle, O. Axner, S. Belahsene, M. Fischer, J. Koeth, Y. Rouillard, J. Westberg, A. Vicet, and S. Lundqvist, Appl. Phys. B 105, 427 (2011). 5. K. Krzempek, M. Jahjah, R. Lewicki, P. Stefański, S. So, D. Thomazy, and F. Tittel, Appl. Phys. B 112, 461 (2013). 6. S. Lundqvist, P. Kluczynski, R. Weih, M. von Edlinger, L. Nähle, M. Fischer, A. Bauer, S. Höfling, and J. Koeth, Appl. Opt. 51, 6009 (2012). 7. D. Richter, A. Fried, and P. Weibring, Laser Photon. Rev. 3, 343 (2009). 8. D. Richter, A. Fried, B. P. Wert, J. G. Walega, and F. K. Tittel, Appl. Phys. B 75, 281 (2002). 9. L. Naehle, S. Belahsene, M. von Edlinger, M. Fischer, G. Boissier, P. Grech, G. Narcy, A. Vicet, Y. Rouillard, J. Koeth, and L. Worschech, Electron. Lett. 47, 46 (2011). 10. I. Vurgaftman, W. W. Bewley, C. L. Canedy, K. Chul Soo, K. Mijin, J. R. Lindle, C. D. Merritt, J. Abell, and J. R. Meyer, J. Sel. Top. Quantum Electron. 17, 1435 (2011). 11. M. O. Fischer, M. von Edlinger, L. Nähle, J. Koeth, A. Bauer, M. Dallner, S. Höfling, L. Worschech, A. W. B. Forchel, S. Belahsene, and Y. Rouillard, Proc. SPIE 7945, 79450E (2011). 12. M. von Edlinger, J. Scheuermann, L. Nähle, C. Zimmermann, L. Hildebrandt, M. Fischer, J. Koeth, R. Weih, S. Höfling, and M. Kamp, Proc. SPIE 8993, 9 (2013). 13. C. S. Kim, M. Kim, J. Abell, W. W. Bewley, C. D. Merritt, C. L. Canedy, I. Vurgaftman, and J. R. Meyer, Proc. SPIE 8631, 86311O (2013). 14. K. Krzempek, G. Sobon, and K. M. Abramski, Opt. Express 21, 20023 (2013). 15. K. Whittaker, L. Ciaffoni, G. Hancock, M. Islam, R. Peverall, and G. Ritchie, Appl. Phys. B 109, 423 (2012). 16. J. J. Scherer, J. B. Paul, H. J. Jost, and M. Fischer, Appl. Phys. B 110, 271 (2013). 17. M. Nikodem and G. Wysocki, Ann. N.Y. Acad. Sci. 1260, 101 (2012). 18. M. Nikodem, G. Plant, Z. Wang, P. Prucnal, and G. Wysocki, Opt. Express 21, 14649 (2013). 19. P. Martín-Mateos, B. Jerez, and P. Acedo, Opt. Lett. 39, 2611 (2014). 20. M. Nikodem and G. Wysocki, Sensors 12, 16466 (2012). 21. G. Wysocki and D. Weidmann, Opt. Express 18, 26123 (2010). 22. M. Nikodem, D. Weidmann, and G. Wysocki, Appl. Phys. B 109, 477 (2012). 23. J. Z. Sotor, G. Dudzik, and K. M. Abramski, Opt. Commun. 291, 279 (2013). 24. M. Nikodem and G. Wysocki, Photon. Lett. Poland 5, 152 (2013). 25. M. Nikodem, D. Weidmann, C. Smith, and G. Wysocki, Opt. Express 20, 644 (2012). 26. I. Melngailis, W. E. Keicher, C. Freed, S. Marcus, B. E. Edwards, A. Sanchez, F. Tso Yee, and D. L. Spears, Proc. IEEE 84, 227 (1996). 27. A. Rogalski, J. Appl. Phys. 93, 4355 (2003).
