We detect acetylene and water-vapor lines by using a difference-frequency generator in the spectral region around 3 µm. Both C2H2 and H2O lines belong to fundamental vibrational bands and exhibit a line strength of the order of 10-20 cm/mol. Acetylene molecules were detected either by pure absorption or by first-derivative wavelength-modulation spectroscopy. The minimum detection sensitivity achieved for C2H2 in nitrogen was 4 ppb (parts in 109). Moreover, we discuss the effects of C2H2 pressure reduction in the presence of nitrogen in order to estimate systematic errors in the concentration measurements. Finally, we tested the accuracy of our spectrometer by detecting water vapor present as an impurity in a nitrogen cylinder at a nominal concentration of ≃5 ppm.
© 2003 Optical Society of America
In recent years, detection and monitoring of trace gases has become an important topic in fields ranging from air-quality control , to green-house emissions monitoring , to control of industrial processes,  to biomedical applications . Tunable mid-infrared (mid-IR) sources are of particular interest, since numerous molecular species exhibit their fundamental vibrational bands in the region between 3 and 12 µm. As is well known, since the 1960s lead-salt diode lasers have been well-established sources for high-sensitivity detection of trace gases both in the laboratory and in the field . Nevertheless, these lasers present some practical drawbacks because of their liquid-nitrogen operation and their unpredictable frequency jumping, which make their use difficult for in situ detection. In addition, the highly astigmatic and divergent output of this source leads to beam instabilities, which can strongly limit measurement precision. Rapid progress has been reported recently in the development of quantum-cascade (QC) lasers ,; as a matter of fact, QC lasers are commercially available for wavelengths higher than 5 µm, whereas the region around 3 µm seems, at the moment, not easily accessible. In this region lie strong vibrational bands of several molecular species that are particularly important for environmental questions, such as H2O, N2O, C2H2, OH, and HCN. It is worth noting that the first three molecular species in fact can also be detected in other spectral bands, where available well-consolidated laser sources are available. However, a spectrometer operating in the region around 3 µm offers the possibility to detect all of them by use of the same system. Alternative approaches for producing cw tunable radiation in the mid-IR are represented by optical parametric oscillators (OPOs)  and difference-frequency generators. Whereas OPOs require enhancement optical cavities, difference-frequency generators achieve good conversion efficiencies even in single-pass geometries. For this reason, the development of coherent radiation sources based on difference-frequency generation (DFG) seems particularly attractive.
Compact DFG-based optical sensors have already been successfully developed for the detection of several molecular species –. In the present study we investigate a DFG spectrometer based on a periodically poled lithium niobate crystal, where the radiation of a Nd-YAG laser and of a diode laser (λ≃780 nm) are mixed. With this method, several microwatts of narrow-linewidth radiation tunable around 3 µm were obtained. Our spectrometer was applied for the detection of acetylene and water vapor. As is well known, acetylene plays an important role in many processes where this molecule is present at concentration ranging from 100 ppt (parts in 103) to 100 ppm (parts in 106). For instance, combustion chemistry it is produced as an intermediate species during hydrocarbon combustion at a level of several tens of ppm.
Different experimental techniques have been used for acetylene detection in flames. Coherent anti-Stokes Raman spectroscopy (CARS), applied to vibrational transitions, has been performed by several groups . However, this technique is effectively useful only for high C2H2 concentrations because of its relatively poor sensitivity. Pure rotational CARS was applied by Bood et al. , with mixtures of acetylene and nitrogen in the mole fraction range between 0.06 and 0.32. Laser-induced fluorescence (LIF) detection was tested by Raiche et al.  in flames where acetylene was the fuel. More recently, LIF detection of C2H2 was demonstrated in low-pressure propane and methane flames.  A mid-IR spectrometer, based on DFG, was developed by Chen et al.  that generated tens of nanowatts by mixing of two Ti:sapphire lasers in a birefringent GaSe crystal. In these conditions, a spectroscopic detection of 28 ppm per meter (ppmm)/Hz1/2 was demonstrated for lines in the fundamental ν 5 band.
In the present study we investigated the C2H2 ν 3 fundamental acetylene band around 3300 cm-1 by using wavelength-modulation spectroscopy (WMS) . The detection sensitivity reached by our spectrometer was 4 ppb (parts in 109), for an optical path of 30 m and at a total buffer gas pressure of 500 mbar.
2. Experimental setup
A scheme of the experimental setup is shown in Fig. 1. Two solid-state near-IR lasers were used for the DFG: a Nd-YAG laser and a semiconductor diode laser. The diode-pumped Nd-YAG laser (Innolight, Model Mephisto 500) consists of a monolithic ring cavity and emits a maximum power of 500 mW in a spectral bandwidth of a few kilohertz. The diode laser (Sacher Lasertechnik, Model TEC 100) was mounted in an external cavity (Littrow mounting); proper alignment of the external grating allows a tuning range between 770 and 785 nm.
The two laser beams were combined on a dichroic mirror and focused into a 19-mm-long periodically poled lithium niobate crystal (PPLN) . The PPLN crystal used in this experiment (Crystal Technology) consists of an array of eight crystals whose periods vary from 20.4 to 21.8 µm. When the proper one is selected, it becomes possible to produce mid-IR radiation (idler) from 2.8 to 3.2 µm. Typical idler power was of the order of 5 µW, starting from 60 mW and 300 mW of the pump and signal power, respectively. This corresponds to a conversion efficiency of 0.0146%/W cm.
The mid-IR beam emerging from the crystal was collimated with a 5-cm–focal-length lens and sent into a 50-cm-long, Herriott-type multipass cell (SIT, Model CMP-30), presenting a total optical path of 30 m. A power transmission as high as 35% was obtained when the collimated idler beam was focused at the center of the multipass cell, by use of a 50-cm–focal-length lens. The radiation emerging from the cell was finally detected by an InSb liquid-nitrogen–cooled photodiode (Hamamatsu, Model PN5968).
The cell was evacuated by means of a turbo-molecular pump at a pressure of 10-6 mbar. The gas pressure in the cell was measured with an absolute Baratron pressure gauge (MKS, Model 122A) working in the range of 10-3 to 10 mbar. In addition, a cold-cathode gauge (Varian, Model senTorr CC2C) was used to monitor lower pressure values.
Frequency scans up to 8 GHz of the DFG radiation were easily obtained by tuning of the diode-laser frequency. Nonlinearity observed in the frequency scan, resulting from the diode-laser PZT response, was corrected by a computer program. We calibrated the mid-IR radiation by measuring the diode-laser wavelength with a traveling Michelson interferometer (accuracy one part in 107) and by using the calibration curve of the Nd-YAG laser provided by the manufacturer. Since InSb detectors are sensitive to environmental mid-IR radiation, pure absorption spectra were revealed by a standard phase-sensitive detection scheme: the Nd-YAG beam was mechanically chopped at a frequency of 1 kHz, and the output photodiode signal was sent into a lock-in amplifier. To detect very low absorption signals, WMS was used. In this case, the diode-laser frequency was modulated at ~1 kHz at a modulation index (the ratio between the modulation amplitude and the width of the absorption line) of ~2. The relatively low frequency modulation was limited by the time response of the InSb detector (≃1 ms). Thanks to the WMS scheme, we could observe the absorption signals on an almost flat background, which allowed us to detect absorbance of the order of 6×10-5.
3. C2H2 detection
To evaluate the performance of our DFG spectrometer, we detected the R(25) C2H2 line belonging to the fundamental ν 3 band. This line is free from interference by other molecular species and, according to the HITRAN database, presents a line strength of 2.339×10-20 cm/mol. This means that, in our experimental conditions (30-m optical path length), a relative absorption of ≃20% can be obtained with a gas pressure of only 1.5 µbar. We proceeded with the determination of the absorption cross section and the nitrogen pressure-broadening coefficient of the selected line.
We performed the cross-section measurement by filling the absorption cell with 99.97%- purity acetylene at a variable pressure and estimating the absorbance given by
where is the acetylene pressure; L is the optical path length; σ is the absorption cross section (cm-1 mbar-1); and I 0 and I(L) are the intensity of the radiation incident on the cell and emerging from it, respectively. The gas pressure was varied from 0.3 to 1.5 µbar. A fit of the absorbance as a function of the gas pressure furnished σ=(4.79±0.09)×10-2 cm-1 mbar-1, which is in good agreement with the value 4.71×10-2 cm-1 mbar-1 reported in the HITRAN database.
Figure 2 shows the pressure-broadened lineshapes of the R(25) line (~ν=3352.2871 cm-1), obtained by varying the N2 pressure from 5.33 to 446.6 mbar and keeping the C2H2 pressure to 350 nbar. To cancel residual interference fringes resulting from spurious etalon effects and to eliminate the diode-laser amplitude modulation, the empty cell absorption baseline was subtracted from each lineshape of Fig. 2. The lineshapes of Fig. 2 were fitted with a Voigt profile, by use of the Levenberg–Marquardt least-squares procedure. In the fitting procedure, the Doppler width was fixed at the theoretical value (γD =333 MHz), whereas the line center, the amplitude, and the homogeneous width were considered to be free parameters. Figure 3 shows the linear behavior of the homogeneous width (FWHM) versus the nitrogen pressure, from which we estimated a collisional broadening coefficient of (3.30±0.05) MHz/mbar. This value is similar to the air-broadening coefficient value 3.37 MHz/mbar reported by HITRAN.
To obtain high detection sensitivity, we found it advantageous to use the WMS technique. For this purpose, the diode-laser frequency was sinusoidally modulated at a frequency of ≃1 kHz. The modulation amplitude Δνm was adjusted to reach the optimum value of the modulation index m(m=Δνm /γ, with γ as the line half-width) .
Figure 4 shows the first-derivative absorption signals of the selected R(25) line for pure acetylene samples at pressure values ranging from 20 to 100 nbar. The signals were acquired with a 1-ms lock-in time constant and by averaging of 20 line profiles; in this condition, an effective bandwidth of ≃50 Hz was achieved. In the inset of Fig. 4 the first-derivative absorption signal corresponding to a C2H2 pressure of ≃3.3 nbar is shown. This pressure value was extrapolated from the linear relationship between the first-derivative absorption amplitude versus the C2H2 pressure which was measured within the working range of our pressure gauge. For each of the signals of Fig. 4 we estimated the signal-to-noise ratio (SNR) by measuring the peak-to-peak amplitude signal out of the resonance line. The SNR is reported in Fig. 5 as a function of the absorbing gas pressure. From a linear fit a minimum detectable C2H2 pressure (at SNR=1) of 0.4 nbar was obtained, which corresponds to a minimum detectable absorbance of ≃6×10-5.
In most applications, such as in combustion processes, acetylene detection occurs in presence of buffer gases at high pressure. At this purpose we choose nitrogen as the buffer gas, since, as shown above, it is very similar to air in the line-broadening mechanism.
An accurate estimation of the detection sensitivity in the presence of buffer gases requires particular care during sample preparation. In our investigation, we prepared C2H2-N2 mixtures by filling the multipass cell at low C2H2 pressure and then adding nitrogen gas. To test the accuracy of C2H2 concentration obtained in such a way, we recorded the pure absorption profile of the R(25) C2H2 line in the presence of nitrogen and estimated the acetylene concentration C using the relation 
where P is the total gas pressure (mbar), T is the temperature (K), NL is the Loschmidt number, γ is the pressure-broadened linewidth (Hz), and S(T) is the line intensity (cm/mol). In Fig. 6 we report the nominal concentration as experimentally prepared, i.e., , and the measured concentration, C meas, found by use of Eq. (2). We obtained the data by introducing in the absorption cell an initial C2H2 pressure of 2 µbar and adding a nitrogen gas pressure ranging between 200 and 450 mbar. In particular, the value assumed for S (cm/mol) in Eq. (2) was calculated from the relation 
whereas γ was estimated by fitting of the absorption signals with a Lorentzian profile, with the lineshape essentially pressure broadened. As can be seen in Fig. 6(b), a difference Δ between C nom and C meas exists, which is essentially constant over the investigated nitrogen pressure range [see Fig. 6(b)]. This behavior cannot be ascribed to systematic errors introduced in the gas pressure measurement. Indeed, in this case, the measured concentration should be
where a and b are the possible systematic errors, and as a consequence, the difference Δ should result dependent on the N2 pressure. In contrast, we assumed that the lower values of the measured concentration are related to some process that leads to a reduction of the partial C2H2 pressure (adsorption, dimerization, reaction with impurities contained in the nitrogen cylinder, and so on). In other words, the concentration in the cell can be described by
where α is the acetylene loss coefficient. It is easy to verify that in this case that Δ is constant and equal to α. We performed a further check to verify our hypothesis by studying the behavior of the area under the absorption line profile versus the nitrogen pressure. As seen in Fig. 7, the measured area decreases linearly as a function of
where A 0 is the area corresponding to and and β is the loss coefficient referred to the area. Since the area is proportional to the partial C2H2 pressure, we have
which, if compared with Eq. (5), leads to the relation
From the data of Fig. 6 we found an average value of 〈α〉=(2.41±0.06)×10-6. Instead, fitting the data of Fig. 7, we found A 0=(2.01±0.02) and β=(2.40±0.06)×10-3 mbar-1. These values allow us to give a further estimation of α [see Eq. (8)], which results in , in good agreement with the loss coefficient previously estimated. The knowledge of the acetylene loss coefficient makes it possible to provide an estimation of the minimum detectable concentration not affected by systematic errors. In Fig. 8 we report the first-derivative absorption signal of the investigated R(25) C2H2 line (effective bandwidth ≃50 Hz), at a concentration of (32.0±0.7) ppb ( equal to 500 mbar). For this signal we estimated SNR ≃8; this means that a minimum detectable concentration (at SNR=1) corresponds to 4 ppb. To our knowledge, this result corresponds to the highest detection sensitivity obtained for this molecular species with a potentially transportable system, which therefore promises to be successfully applied for highly sensitive detection of acetylene in flames. In fact, in a combustion system the acetylene concentration ranges between 1 and 103 ppm, depending on the combustion conditions. Therefore, the obtained detection sensitivity of 4 ppb should allow us to detect higher concentration in a shorter optical path (single-pass geometry).
The accuracy of our spectrometer was also tested for water-vapor detection. The test was accomplished by detection of water vapor present as impurity in a nitrogen cylinder at a nominal concentration of 5 ppm.
We filled the multipass cell at a total N2 pressure between 5 and 20 mbar and monitored the absorption profile of the 33,0←44,1 H2O line in the fundamental ν 1 band. This line has a line intensity of 5.049×10-21 cm/mol and, according to HITRAN database, is one of the strongest lines in the ν 1 vibrational band. To estimate the H2O concentration, we used a relation similar to Eq. (2), which holds for Voigt profiles:
In Eq. (9) γD and γL are the Doppler and Lorentzian width, respectively, estimated by fitting the acquired absorption signal, β=γD /[γD +(ln2)1/2 γL ], while S was assumed equal to the value reported by HITRAN database. In Fig. 9 we report, for instance, the water-vapor absorption line at a total pressure of 10 mbar; a fit of the curve with a Voigt profile is also reported. Particular care was taken in subtracting the background obtained with the evacuated cell to eliminate the broad lineshape contribution resulting from the absorption in the air of the laboratory. Concentration measurements were repeated at 20 different total pressure values. From these data, an average value of water-vapor concentration equal to ppm was estimated, which is in good agreement with the value of 5 ppm certified as the impurity level by the manufacturer. The standard deviation of the concentration measurement (70 ppb) furnishes the accuracy of our spectrometer for the detection of water vapor.
In this study we have investigated the performances of a DFG-based spectrometer for acetylene and water-vapor detection. We demonstrated that our system is useful when sensitive and accurate detection of the investigated molecular species is required. In particular, acetylene molecules were detected in a multipass cell at a sensitivity limit of 4 ppb. Further improvement of the apparatus can easily allow acetylene detection in a combustion system, where these molecules are present at concentration levels quite higher than the sensitivity here reached. In addition, the possibility here demonstrated of an accurate H2O detection allows the simultaneous monitoring of another combustion relevant molecule.
References and links
1. D. Richter, A. Fried, B. P. Wert, J. G. Walega, and F. K. Tittel, “Development of a tunable mid-IR difference frequency laser source for highly sensitive airborne trace gas detection,” Appl. Phys. B 75, 281–288 (2002). [CrossRef]
2. F. K. Tittel, D. Richter, and A. Fried, Mid-Infrared Laser Sources, I. T. Sorokina and K.L. Vodopyanov, eds. (Springer-Verlag, Berlin, 2002).
3. R. E. Neuhauser, U. Panne, R. Niessner, and P. Wilbring, “On line monitoring of chromium aerosols in industrial exhaust stream by laser-induced plasma spectroscopy (LIPS),” J. Anal. Chem. 364, 720–726 (1999).
4. H. Dahnke, D. Kleine, P. Hering, and M. Mürtz “Real-time monitoring of ethane in human breath using mid-infrared cavity leak-out spectroscopy,” Appl. Phys. B 72, 971–975 (2001). [CrossRef]
5. P. Werle, “A review of recent advances in semiconductor laser based gas monitors,” Spectrochim. Acta Part A 54, 197–236 (1998). [CrossRef]
6. A. A. Kosterev, R. F. Curl, F. K. Tittel, C. Gmachl, F. Capasso, D. L. Sivco, J. N. Baillargeon, A. L. Hutchinson, and A. Y. Cho, “Effective utilization of quantum-cascade distributed-feedback lasers in absorption spectroscopy,” Appl. Opt. 39, 4425–4430 (2000). [CrossRef]
7. D. D. Nelson, J. H. Shorter, J. B. Mcmanus, and M. S. Zahniser, “Sub-part-per-billion detection of nitric oxide in air using a thermoelectrically cooled mid-infrared quantum cascade laser spectrometer,” Appl. Phys. B 75, 343–350 (2002). [CrossRef]
8. M. M. J. W. Van Herpen, S. Li, S. E. Bisson, S. Te Lintel Hekkert, and F. J. M. Harren, “Tuning and stability of a continuous-wave mid-infrared high-power single resonant optical parametric oscillator,” Appl. Phys. B 75, 329–333 (2002). [CrossRef]
9. D. Richter, D. G. Lancaster, R. F. Curl, W. Neu, and F. K. Tittel, “Compact mid-infrared trace gas sensor based on difference-frequency generation of two diode lasers in periodically poled LiNbO3,” Appl. Phys. B 67, 347–350 (1998). [CrossRef]
10. Y. Mine, N. Melander, D. Richter, D. G. Lancaster, K. P. Petrov, R. F. Curl, and F. K. Tittel, “Detection of formaldehyde using mid-infrared difference-frequency generation,” Appl. Phys. B 65, 771–774 (1997). [CrossRef]
11. W. Chen, G. Mouret, D. Boucher, and F. K. Tittel, “Mid-infrared trace gas detection using continuous-wave difference frequency generation in periodically poled RbTiOAsO4,” Appl. Phys. B 72, 873–876 (2001). [CrossRef]
12. G. Pesce, G. Rusciano, and A. Sasso, “Detection and spectroscopy of OH fundamental vibrational band based on a difference frequency generator at 3 µm,” Chem. Phys. Lett. 374, 425–431 (2003). [CrossRef]
13. G. Pesce, G. Rusciano, and A. Sasso “High sensitivity spectrometer at 3 µm based on difference frequency generation for N2O detection,” IEEE Sensor J. 3(2), 206–211 (2003). [CrossRef]
14. A. Bruno, G. Pesce, G. Rusciano, and A. Sasso, “Detection and spectroscopy of the ν1+ν3 band of N2O by difference-frequency spectrometer at 3 µm,” Spectrochim. Acta Part A 58, 2481–2488 (2002). [CrossRef]
15. A. Khorsandi, U. Willer, P. Geiser, and W. Schade, “MIR-difference frequency laser spectrometer for CO detection in combustions,” Iranian J. Phys. Res. 3, 4 (2003).
16. R. P. Lucht, R. L. Farrow, and R. E. Palmer, “Acetylene measurements in flames by coherent anti-stokes Raman spectroscopy,” Combust. Sci. Technol. 45, 261 (1986). [CrossRef]
17. J. Bood, P. E. Bengtsson, and M. Aldén, “Temperature and concentration measurements in acetylene-nitrogen mixtures in the range 300–600 K using dual-broadband rotational CARS,” Appl. Phys. B 70, 607–620 (2000). [CrossRef]
18. G. A. Raiche, D. R. Crosley, and R. A. Copeland, Am. Inst. Phys. Proc.191758 (1989).
19. B. A. Williams and J. W. Fleming, “Laser-induced fluorescence detection of acetylene in low-pressure propane and methane flames,” Appl. Phys. B 75, 883–890 (2002). [CrossRef]
20. W. Chen, G. Mouret, and B. Boucher, “Difference-frequency laser spectroscopy detection of acetylene trace constituent,” Appl. Phys. B 67, 375–378 (1998). [CrossRef]
21. F. S. Pavone and M. Inguscio “Frequency- and wavelength-modulation spectroscopies: comparison of experimental methods using an AlGaAs diode laser,” Appl. Phys. B 56, 118–122 (1993). [CrossRef]
22. M. M. Fejer, G. A. Magel, D. H. Jundt, and R. L. Byer, “Quasi-phase-matched second harmonic generation: tuning and tolerances,” IEEE J. Quantum Electron. 28, 2631–2654 (1992). [CrossRef]
23. USF HITRAN-PC Version 2.0 (University of South Florida, Tampa, Fla., 1992).
24. M. Erdélyn, D. Richter, and F. K. Tittel “13CCO2 isotopic ratio measurements using a difference frequency-based sensor operating at 4.35 µm,” Appl. Phys. B 75, 289–295 (2002). [CrossRef]
25. B. J. Finlayson-Pitts and J. N. Pitts, Atmospheric Chemistry: Fundamentals and Experimental Techniques (Wiley-Interscience, New-York, 1987), p. 65.