1. Introduction

It is both scientifically and technologically challenging to devise instrumental techniques with adequate absolute quantitative accuracy and reliability for trace-level detection of greenhouse-gas components at ambient concentrations in our atmospheric environment. A capability to make measurements of this type is relevant to environmental issues that are of great public concern [1]. In this regard, near-infrared (IR) continuous-wave (cw) cavity-ringdown (CRD) absorption spectroscopy is a particularly promising technique for such measurements.

CRD spectroscopy [2–4] is well established as a highly sensitive technique for absolute measurement of optical absorption. It depends on determining the decay rate of coherent light in a high-finesse optical cavity containing a sample (usually gas-phase) of interest. The intensity of this intra-cavity light typically decays as a simple exponential function, with the decay rate (or inverse time constant τ^–1) depending on cavity mirror reflectivity and additional intra-cavity optical losses (due to processes such as absorption and scattering). Optical absorption causes the decay rate τ^–1 to increase relative to that of an absorber-free cavity or to that at non-absorbing wavelengths, thereby providing a direct measure of optical absorption by components of the sample medium inside the cavity. The inherent sensitivity of the CRD technique stems both from the extremely long effective path lengths of optical interaction (up to tens of kilometers) that are attained by using highly reflective, low-loss cavity reflectors and from the insensitivity of the decay rate to fluctuations in laser intensity.

Here we focus on our special variant of cw-CRD spectroscopy [5–8], in which ringdown decay is established and subsequently measured simply either by rapidly sweeping the cavity length (and hence its optical resonance frequencies) while fixing or slowly scanning the laser wavelength [5,8] or by rapidly sweeping the laser wavelength (with fixed or progressively stepped cavity length) [6,7]. Coherent interaction between the laser light and the optical cavity then facilitates build-up and subsequent decay of laser radiation in the cavity at its optical resonance frequencies, without needing to incorporate a fast electro-optic or acousto-optic switch to interrupt or modulate the cw input radiation. Moreover, the laser frequency does not need to be locked to resonate with the ringdown cavity and is required to be stable only during the optical build-up time of the cavity. A further advantage of these rapidly swept approaches to cw-CRD absorption spectroscopy is that the appropriate instrumentation can be significantly simpler, more compact, and more rugged than other typical cw-CRD spectrometers. In addition, they facilitate a single-ended, fiber-coupled transmitter-receiver configuration in a retro-reflected, optical-heterodyne-detected (OHD) mode [5,8].

1.1 Background to our research on multi-species spectroscopic sensing

We are concerned in this paper with quantitative detection of greenhouse-gas molecules such as carbon dioxide (CO₂), methane (CH₄), and water vapor (H₂O) in natural isotopic abundance at ambient concentrations in air. Our rapidly swept cw-CRD approach is readily amenable to such multi-species detection schemes. Moreover, this method enables all three molecular species to be detected simultaneously, within the period (a few ms) of the cavity sweep cycle.

A conventional (but time-consuming) cw-CRD spectroscopic approach is to scan the wavelength of the monochromatic laser radiation through adjacent spectroscopic features of different chemical species. As a laborious prelude to implementing our simultaneous multi-laser cw-CRD spectroscopic detection scheme, Fig. 1 (top trace, with τ^–1 plotted on the CRDS ordinate) presents a superimposed set of three survey absorption spectra (plotted in green, red, and blue) of outdoor air with pressure P ≈750 Torr and T ≈300 K, sampled from the environment outside our laboratory building and flowed continuously at a rate of ~50 L h^–1 through the ringdown cell. Each spectrum was measured by means of a swept-cavity cw-CRD spectrometer using a single widely tunable 3-mW external-cavity diode laser (ECDL; Photonetics, model Tunics Plus), scanned very slowly with high spectroscopic resolution through the wide wavelength range of 1540–1640 nm (spanning 100 nm in steps of 0.005 nm at 1 data point per second). Note that the higher baseline level and noise around 1620–1640 nm are attributable to reduced mirror reflectivity and laser stability in that region.

 

Fig. 1 Top trace: three superimposed survey CRD absorption spectra of outdoor air measured by a swept-cavity near-IR cw-CRD spectrometer with a single tunable diode laser. Lower traces: stick spectra for the three greenhouse-gas components (CO₂, CH₄, and H₂O vapor), based on the HITRAN spectroscopic database [9].

Download Full Size | PDF

The dominant features of these survey spectra are attributable to weak absorption by CO₂, CH₄, and H₂O in outdoor air, as indicated by their stick spectra based on the HITRAN spectroscopic database [9]. Since all three of these greenhouse-gas molecules have absorption features in the wavelength region used for fiber-optical telecommunications, our CRD spectrometer is able to take advantage of low-cost, monolithic, miniature tunable diode lasers, as well as fiber-optical components and networks for remote, distributed-cavity CRD-based sensing systems [5,8].

Recording extensive survey absorption spectra as in Fig. 1 is not a practical way to pursue active environmental sensing. However, data thus collected help us either to identify more localized spectral regions in which two or more molecular species of interest have adjacent characteristic features or to identify interference-free isolated features in absorption spectra, using less elaborate diode lasers with limited tuning range. For instance, we have adopted such a method at near-IR laser wavelengths around ~1.5765 μm to examine dilute gas-phase mixtures of carbon monoxide (CO) and CO₂ via their overlapping absorption spectra [10]. We note that gas-phase CO:CO₂ concentration ratios are a key control measure of combustion efficiency in industrial processes such as smelting. Likewise, we have used a similar approach to monitor acetylene (C₂H₂) diluted in moist N₂ gas at ~1.5201 μm by scanning part of the (ν₁ + ν₃) band of C₂H₂ and the overlapping spectrum of H₂O vapor [8].

However, a more efficient method for instantaneous multi-species cw-CRD spectroscopy is to use several fixed-wavelength tunable diode lasers, each with its output pre-tuned to probe a characteristic absorption feature of a particular gas-phase molecular species of interest. We have previously demonstrated this multi-laser, multi-species cw-CRD spectroscopic concept with separate diode lasers in the case of on- and off-resonance wavelengths for CO₂ gas at ~1.54 μm [10]. In later (unpublished) work, we achieved simultaneous gas-phase detection of CO and CO₂ at ~1.56 and ~1.57 μm, respectively. We have now extended this approach using tailored multi-wavelength coherent radiation from a set of pre-tuned near-IR diode lasers and have reported preliminary results [11] on greenhouse-gas molecules (CO₂, CH₄, H₂O) in outdoor air. This paper reports details and outcomes of that research.

Prior to the present work, near-IR wavelength-modulation absorption spectra of ambient CO₂ and CH₄ in air have been measured [12,13], providing useful reference points for the present multi-species cw-CRD spectroscopy. Using an open-path multiple-reflection cell, minimum detectable mixing ratios (in parts per million by volume, ppmv) were measured at 1.578 μm and 1.653 μm to be 10 ppmv and <0.1 ppmv for CO₂ and CH₄, respectively [12], in air at a total pressure of 1 atm; by comparison, our previous best-effort swept-cavity 1.578-μm cw-CRD spectroscopic measurements of CO₂ in air yield a minimum detectable mixing ratio of ~2.7 ppmv [8]. More recently, a Fresnel lens and remotely positioned retro-reflector have been used in a time-sharing mode of long-path multi-species wavelength-modulation spectroscopic measurements of CH₄ and hydrogen sulfide (H₂S), to monitor leakage of natural gas [13]; this yields an estimated minimum detectable mixing ratio of ~0.01 ppmv for CH₄ in air.

1.2 Background to other cw-CRD approaches to multi-species spectroscopic sensing

There are various other examples of multi-species cw-CRD spectroscopy, including an early multiplexed near-IR study [14], in which acousto-optic modulators (AOMs) were used to toggle between two separate 1.4-μm distributed feedback (DFB) diode lasers, each pre-tuned to absorption features of vapor-phase species such as methanol (CH₃OH) and isopropanol ((CH₃)₂CHOH) in the presence of residual H₂O vapor. A similar approach has been applied for rapid fire detection via the gas-phase combustion products CO, CO₂, C₂H₂, and hydrogen cyanide (HCN) [15], with an AOM used to toggle between two DFB diode lasers operating at ~1.57 μm (to monitor CO and CO₂) and ~1.53 μm (to monitor C₂H₂ and HCN). More recently, a multi-species near-IR cw-CRD spectrometer, using two multiplexed DFB diode lasers, has been used to monitor ¹²CO₂ and ¹³CO₂ (at ~1.57 μm) and CH₄ (at ~1.65 μm) [16]. Likewise, another advanced near-IR cw-CRD spectrometer employed two multiplexed DFB diode lasers for multi-species detection of CH₄ and H₂O (at ~1.651 μm) and of CO₂ (at 1.603 μm) [17]. Addition of a chromatographic separation stage to a continuous-flow near-IR cw-CRD spectrometer has facilitated isotopic ¹³C/¹²C analysis of gas-phase mixtures of the hydrocarbons CH₄, ethane (C₂H₆), and propane (C₃H₈), with ¹³CO₂ and ¹²CO₂ detected at ~1.603 μm after catalytic conversion of the hydrocarbons [18].

Progress in multi-species near-IR cw-CRD spectroscopy has also been made possible by the availability of broadly tunable telecommunication lasers that can cover a much wider range of characteristic wavelengths than is possible with lasers that operate at a single discrete wavelength (which may be tuned relatively slowly). Using DFB diode-laser arrays [19], molecular species have been detected at ambient concentrations in air with high sensitivity. Alternatively, miniature swept-frequency lasers based on microelectromechanical structure (MEMS) technology have facilitated rapid acquisition of wide-ranging near-IR spectra in the telecommunications C-band and have been used to detect CO₂ and H₂O vapor in N₂ [6,7].

Other recent developments in the area of multi-species near-IR cw-CRD spectroscopy include advanced instrumentation based on a broadband optical frequency comb emitted by a mode-locked laser source [20,21]. Such measurements of oxygen (O₂), ammonia (NH₃) and H₂O at 0.76–0.85 μm [20] and of NH₃, C₂H₂ and CO at 1.46–1.60 μm [21] have demonstrated a huge multi-species tuning range relative to conventional CRD spectroscopy with discrete-wavelength near-IR cw laser sources, traded off against limited sensitivity, spectral resolution and increased instrumental complexity.

1.3 Specific objectives of this paper

In this paper, we report details of highly sensitive trace-level environmental monitoring of three greenhouse-gas molecules (CO₂, CH₄, and H₂O), which have been the subject of many previous spectroscopic investigations including those mentioned above [5–21]. Our technique (which has not previously been reported, apart from a preliminary conference presentation [11]) uses a multi-laser swept-cavity cw-CRD spectrometer operating in the near-IR region around 1.5 μm. This innovative form of simultaneous multi-species spectroscopic sensing has evolved from our own previous work on rapidly swept cw-CRD spectroscopy (see Sec. 1.1 [5–8,10]) and its variants for multi-species cw-CRD spectroscopy (see Sec. 1.2 [6,7,14–21]).

Our approach also depends on evaluating various forms of compact tunable diode laser, including those with DFB or distributed Bragg reflector (DBR) designs, that are commercially available in integrated packages. DFB and DBR diode lasers are typically more robust, more compact and less expensive than ECDL sources (as used in our previous cw-CRD spectroscopic studies [5–8,10]). The survey spectra of Fig. 1 show that ECDLs offer the advantage of wide-range wavelength tuning. However, targeted monitoring of specific gas-phase molecules has more relaxed requirements for active wavelength-tuning performance of laser sources, so that DFB and/or DBR lasers may be conveniently employed.

2. Instrumentation

The layout of our rapidly swept IR cw-CRD spectrometer [5,8,10,11] is illustrated in Fig. 2 . It includes several fixed-wavelength, near-IR cw diode lasers (DL₁, DL₂, …) and two photodetectors (PD₁, PD₂). The high-finesse ringdown cavity and its accompanying gas cell have been described in detail elsewhere [8]; its two concave mirrors (with ~99.993% reflectivity at ~1.55 μm and a 1-m radius of curvature) have a separation of ~0.53 m which can be rapidly varied by a piezoelectric translator (PZT). With appropriate control electronics, this instrument can operate in either a conventional, forward-propagating mode (using photodetector PD₂) or in a backward-propagating OHD mode (using photodetector PD₁). In the latter case, the backward-propagating light beam is coupled into the same optical fiber that delivered it from the laser, with a fiber-optical three-port circulator (AOC Technologies model OCIR-3-155, isolation, directivity and return loss greater than 50 dB [8]) used to divert the backward-propagating reflected and ringdown radiation to photodetector PD₁ for OHD operation [5,8,10]. The laser source(s) and ringdown light detection can therefore be integrated into a single transmitter-receiver module (left-hand side of Fig. 2). One or more ringdown cavity modules (right-hand side of Fig. 2) can then be located far away from their transmitter-receiver module, for remote-sensing applications by means of fiber-optical networks. Analog circuits allow real-time processing of resulting ringdown signals [5,8], generating a voltage output that is proportional to the ringdown time τ. This yields a direct measure of absorption and concentration of molecules in the optical cavity, without any digital computing process. The instrument can operate in three different configurations: (i) with a single tunable laser [5,7,10], (ii) with a swept-frequency laser [6,7], and (iii) with multiple lasers set at selected on- and off-resonance wavelengths for molecules of specific interest [6,10,11]. Configuration (iii), as portrayed in Fig. 2, is the multi-laser variant of rapidly swept cw-CRD spectroscopy that is of central interest in this paper.

 

Fig. 2 Layout of an optical fiber-coupled, multi-laser near-IR cw-CRD spectrometer. The length of the ringdown cavity is rapidly varied by a piezoelectric translator (PZT). Radiation from several fixed-wavelength diode lasers (DL₁, DL₂, …) is coupled to the cavity to allow simultaneous measurement of multiple gas species.

Download Full Size | PDF

Several fixed-wavelength near-IR cw diode lasers (DL₁, DL₂, …) are each pre-tuned to a characteristic on- or off-resonance feature for the gas-phase molecules of interest, with their outputs combined into a single-mode optical fiber. Each wavelength-component of the diode-laser light is propagated to the ringdown cavity, where it attains CRD resonance at a distinct portion of the cavity-sweep cycle. The N × 1 fiber-optical combiner or optical switch provides flexible access to multiple laser outputs, as indicated schematically in Fig. 2.

Figure 3 shows how swept-cavity CRD spectroscopy with several fixed-wavelength near-IR cw diode lasers enables rapid specific multi-wavelength detection, with one, two and three lasers and with the ringdown cell statically filled with transparent pure nitrogen (N₂) gas. With a single monochromatic laser, CRD signal events occur repetitively each time the fixed-wavelength laser radiation passes through optical resonance with the rapidly swept ringdown cavity, its length controlled by the PZT voltage ramp depicted in Fig. 3(a). Figure 3(b) shows that such CRD signal events occur repetitively (period ~1.1 ms) whenever the PZT-controlled cavity length is changed by a half-wavelength, so that the cavity round-trip optical pathlength is an integer multiple of the laser wavelength. Figures 3(c) and 3(d) display corresponding CRD signal events with two- and three-wavelength irradiation. For each laser wavelength (λ₁, λ₂, λ₃), the CRD signal events can be cleanly separated in time by finely adjusting the laser wavelengths and, therefore, their resonance points. Such adjustments correspond to only a fraction of the cavity-mode separation, and are small (several percent) in comparison to the spectral widths of particular absorption features at atmospheric pressure. Incidentally, it may be noted that the spacings between CRD features in Figs. 3(c) and 3(d) appear to be slightly irregular; this is attributable to the fact that, although the PZT voltage ramp in Fig. 3(a) is linear, the linearity of the PZT-driven mirror motion is imperfect. Actual laser wavelengths used to record Figs. 3(b)–3(d) were: λ₁ ≈1573 nm and λ₂ ≈1586 nm (each generated by Santec TSL-210 ECDLs); λ₃ ≈1635 nm (generated by our 3-mW Photonetics Tunics Plus ECDL). As will be established in Sec. 3.2 below, these three specific laser wavelengths λ₁, λ₂, λ₃ are suitable for interference-free trace-level detection of CO₂, H₂O vapor, and CH₄, respectively. Each laser runs continuously (which helps us to operate it at a fixed wavelength and on a single longitudinal mode), with the CRD timing sequence spontaneously determined by the setting of the ringdown cavity length and its resonance with each laser wavelength (λ₁, λ₂ or λ₃). We have developed LabVIEW software to control each laser, to log ringdown curves via an analog-to-digital board (National Instruments, model PCI-5122, 100 MHz, 14 bit, 32 MByte in each of two channels), and to analyze ringdown decays. We usually select one laser (the most stable one) as master, and actively control frequencies of the other lasers (via their operating temperature and current) by stabilizing the temporal separation of their ringdown events to that of the master laser within a cavity sweep cycle.

 

Fig. 3 Signal voltages for a set of 0–3.5-ms rapidly swept scans of the optical cavity of a near-IR cw-CRD spectrometer. Trace (a): piezoelectric control voltage, spanning more than two free spectral ranges of the cavity. Trace (b): CRD signal events recorded with a single laser wavelength (λ₁); these occur repetitively whenever the PZT-controlled cavity length is changed by a half-wavelength, so that the cavity round-trip optical pathlength is an integer multiple of the laser wavelength. Traces (c and d): corresponding CRD signal events with two and three laser wavelengths (λ₁, λ₂) and (λ₁, λ₂, λ₃), respectively; the wavelengths have been finely adjusted to ensure that each signal event is cleanly separated in time within the cavity-sweep cycle of period ~1.1 ms.

Download Full Size | PDF

3. Results

The significant experimental results obtained in this project (apart from those depicted already in Figs. 1 and 3 to explain the methodology) are as follows. In Sec. 3.1, we have investigated the key performance characteristics of assorted diode lasers that are available for use. In Sec. 3.2, we have identified well-isolated molecular absorption lines for possible co-existing gas species that are of interest, to optimize detection sensitivity and to minimize interferences that tend to impair unambiguous quantitative analysis. Finally, in Sec. 3.3, we report the crucial outcomes of simultaneous multi-laser, multi-species measurements by rapidly swept cw-CRD spectroscopy on the greenhouse-gas molecules CO₂, CH₄, and H₂O in air.

3.1 Performance of assorted near-IR diode lasers

Operation of cw-CRD spectroscopic systems in this near-IR wavelength region is greatly facilitated by the wide availability of compact, low-cost optical fiber components and cw tunable diode lasers operating around 1.55 μm. However, selection of suitable components requires considerable care, particularly because data sheets and other specifications are often aimed at the telecommunications market, providing characteristics of typical performance that may be too general or lacking in detail for our particular purposes. We have therefore tested several diode lasers thoroughly, with our specific CRD-spectroscopic applications in mind.

Figure 4 displays a set of fixed-wavelength baseline sensitivity traces recorded as noise-equivalent absorbance (NEA) by using our swept-cavity cw-CRD spectrometer (as described in Sec. 2 above) to test six different diode-laser light sources and an absorber-free ringdown cell. Three of the six (left-hand block of Fig. 4, with their performance traced in red) are ECDL-type light sources: New Focus, model 2630 with 6200 controller (tunable over ~1.51–1.59 μm with ~5 mW output power); Photonetics, model Tunics Plus (tunable over ~1.50–1.64 μm at ~3 mW, also used to record Figs. 1, 3, 5 , and 6 ); Santec, model TSL-210 (tunable over ~1.52–1.62 μm at ~10 mW, also used to record Fig. 3). The other three of the six light sources (right-hand block of Fig. 4, with their performance traced in blue) are DFB-type diode lasers, each: Eblana Photonics, model EP1550-NLW-B01-400-FA (~1.57 μm); Fitel, model FOL15CDCWD (~1.54 μm); NTT Electronics, model NEL NLK1U5FAAA (~1.64 μm). Each measurement was conducted under similar conditions, with ~2 mW of optical power delivered at the input to the ringdown cell. Each data point is derived from a numerical fit of the exponential decay rate for an average of 256 cavity-ringdown curves. We note that these measurements apply only to the particular laser units used in our tests. Nevertheless, they provide useful information that may be more widely applicable to this representative range of models, despite the fact that manufacturers are often improving and releasing new models.

 

Fig. 4 Comparison of cw-CRD spectroscopic sensitivity (NEA) using assorted diode-laser sources. The performance of three ECDL-type light sources is depicted in the left-hand block (red traces) whereas that of three DFB-type diode lasers is depicted in the right-hand block (blue).

Download Full Size | PDF

 

Fig. 5 Observed cw-CRD (a) and HITRAN-simulated (b) absorption spectra of gas mixtures of CO₂, CO, CH₄, and H₂O diluted in N₂ at 1 atm in a region where CO₂ (red) predominates over other components CO (green) and H₂O (blue). The interference-free CO₂ features marked <1> and <2> are preferred for CO₂ detection by multi-laser cw-CRD spectroscopy.

Download Full Size | PDF

 

Fig. 6 Observed cw-CRD (a) and HITRAN-simulated (b) absorption spectra of gas mixtures of CO₂, CO, CH₄, and H₂O diluted in N₂ at 1 atm in a region where H₂O (blue) predominates over other components CO (green) and CO₂ (red). The interference-free H₂O features marked <1> and <2> are preferred for H₂O detection by multi-laser cw-CRD spectroscopy.

Download Full Size | PDF

It should be noted that the ordinate (NEA) scale for the red (ECDL) traces in Fig. 4 is finer by a factor of 10 than that for the blue (DFB laser) traces in Fig. 4. These measurements therefore confirm that, as a general rule, ECDL-type lasers are significantly more stable than DFB-type lasers in terms of their optical frequency (at least on sub-ms time scales) and that the resulting ringdown decay curves are closer to single-exponential with less-pronounced deviations/fluctuations. However, DFB-type lasers are much more compact, robust and cheap. The frequency stability of the light source is a key factor that limits the detection sensitivity of a CRD-spectroscopic system. In these measurements, the standard deviations σ for NEA sensitivities recorded by the ECDL lasers approach ~2 × 10⁻⁹ cm⁻¹; the corresponding NEA, normalized to the data rate of ~1 Hz, is ~2 × 10⁻⁹ cm⁻¹ Hz^-1/2. By contrast, the DFB lasers yield sensitivities and NEAs that are generally inferior by an order of magnitude.

3.2 Selection of characteristic transitions to minimize interference for unique identification

Many gas-phase molecules of interest (e.g., CH₄, CO₂, H₂O, CO, C₂H₂, NH₃) have absorption spectra in the 1.5–1.65-μm wavelength region. Their individual rovibrational absorption features span a broad wavelength range, with lineshapes (and widths) affected by Doppler and pressure broadening. For such molecules at ~300 K, full-width-at-half-maximum (FWHM) Doppler widths are ≤0.6 GHz (≤0.02 cm^–1), while pressure-broadened linewidths are typically ~5 GHz (~0.17 cm^–1) FWHM in air at atmospheric pressure. For a multi-component gas mixture, an absorption measurement at one particular wavelength can often be complicated by contributions from adjacent absorption lines of several molecules. Careful examination of line positions for possible co-existing gas species is therefore necessary, to identify well-isolated absorption lines for each molecule and thereby minimize any interferences that might otherwise obscure their unambiguous identification and quantitative analysis. Our investigation of interferences between various atmospheric species of interest has identified several relatively strong, well-isolated lines for the purpose of spectroscopic detection.

Figures 5(a) and 6(a) show cw-CRD absorption spectra measured for a gas mixture with partial pressures of CO, CO₂, and CH₄ in the ratio 700:500:1, diluted 500-fold in N₂ at a total pressure P = 1 atm, and with residual H₂O vapor. The ringdown cell contained this synthetic gas mixture and spectra were recorded by tuning our 3-mW Photonetics Tunics Plus ECDL.

Absorption features suitable for detection of CO₂ and H₂O vapor are highlighted in the measured cw-CRD spectra in Figs. 5(a) and 6(a), respectively. Figures 5(b) and 6(b) show corresponding HITRAN-based simulations for a mixture of CO₂, CO, and H₂O.

In the case of CO₂ in Fig. 5, the preferred transitions marked <1> (at 1572.6598 nm) and <2> (at 1573.3317 nm) are well isolated from interference by other gas species. These are quite distinct from assorted characteristic wavelengths in the range 1560–1603 nm that have been used [8,11,19–21,23,27] in other CRD-spectroscopic studies of CO₂.

Although near-IR absorption by H₂O vapor is much stronger outside the wavelength region of the fiber-optical telecommunication C- and L-bands, it is convenient here to use less prominent lines in regions where CO₂ and CO also absorb, so that a single set of optical coatings and other components can be used to detect all three of these molecules. The H₂O absorption lines designated <1> (at 1586.288 nm) and <2> (at 1591.686 nm) in Fig. 6 are relatively strong in this wavelength range which overlaps spectra of CO₂ and CO; as far as we know, they have not previously been used for near-IR CRD-spectroscopic sensing of H₂O. We note that, although H₂O vapor is not of direct environmental concern as an anthropogenic additive, it is nevertheless one of the most important greenhouse gases, in view of its widespread IR absorption spectrum, its significant continuum absorption, and the effect of global warming on its natural vapor pressure. Likewise, CO is included in our synthetic gas mixture because it often accompanies greenhouse gases of primary concern and is implicated (together with CO₂) in combustion processes [10,15,21].

Ambient CH₄ is typically very much less abundant than CO₂, but it can still contribute significantly to global warming and is also implicated in tropospheric ozone production. Several recent near-IR spectroscopic investigations have focused on CH₄ [6–8,14,18,22–27]; these include our own studies of CRD-detected stimulated Raman-gain spectra [22,26] and monitoring of CH₄ in ambient and leakage situations [23,24].

To guide our experiments here, frequencies and line strengths from the HITRAN spectroscopic database [9] have been used to simulate absorption spectra for CH₄ and H₂O in the 1635–1655-nm region, slightly outside the C- and L-band optical-fiber telecommunication wavelength ranges (1530–1565 nm and 1565–1625 nm, respectively), but still accessible to some diode lasers (in particular, DFB-type diode lasers). Rovibrational structure in spectra of CH₄ occurs in closely-spaced clusters of lines, which are merged by Doppler and pressure broadening; such spectra of trace CH₄ in air at 1 atm and 300 K have a FWHM Doppler width of 0.565 GHz (0.0188 cm^–1) and a pressure-broadened profile with a FWHM of ~3.9 GHz (~0.13 cm^–1), so that its net Voigt convolution is effectively Lorentzian. The measurements of CH₄ that we report in Sec. 3.3 were performed at 1635.414 nm, because other more prominent features revealed by our HITRAN simulations (e.g., at 1650.956 nm and 1653.725 nm) are beyond the 1640-nm operating limit of our available tunable diode lasers and cavity reflectors.

3.3 Simultaneous multi-laser multi-species measurements

Our conclusive simultaneous multi-species experimental measurements were performed with up to three diode lasers, as proposed in Fig. 3. Guided by Sec. 3.2, three distinct diode-laser wavelengths (1572.660 nm, 1635.414 nm, and 1586.288 nm) were set to monitor the greenhouse-gas molecules CO₂, CH₄, and H₂O in air, respectively. As in Figs. 3(b) and 3(d), the two shorter wavelengths (detecting CO₂ and H₂O vapor) were generated by a pair of 10-mW Santec TSL-210 lasers, while our 3-mW Photonetics Tunics Plus ECDL was used to generate the longest wavelength (detecting CH₄).

Representative results are shown in Fig. 7 . During the first 100 s of the measurements depicted, an absorber-free baseline CRD decay rate was recorded for each simultaneously monitored wavelength (with red, orange, and blue traces indicating CO₂, CH₄, and H₂O vapor in air, respectively), after flushing the ringdown cell with pure dry N₂ gas at atmospheric pressure. A two-way valve was then used to switch the gas flow in the ringdown cell (where P ≈750 Torr and T ≈300 K) from pure N₂ to air sampled from outside the laboratory building via a 5-m length of 6.35-mm-diameter nylon pressure tubing at a flow rate of ~50 L h^–1.

 

Fig. 7 Use of swept-cavity cw-CRD spectroscopy to monitor greenhouse gases CO₂ (red), CH₄ (orange) and H₂O (blue) in outdoor air at fixed diode-laser wavelengths of 1572.660 nm, 1635.414 nm, and 1586.288 nm, respectively. The decay rate of the absorber-free ringdown cavity at ~1635 nm is relatively high because of reduced mirror reflectivity, as indicated by measurements during the first 100 s of each trace when the cavity is flushed by pure N₂ gas.

Download Full Size | PDF

It should be noted that the baseline of the CRD decay rate for the absorber-free cell at ~1635 nm (orange trace, for CH₄) is ~20% higher than for the other two wavelengths (~1573 nm and ~1586 nm, for CO₂ and H₂O vapor, respectively). This is attributed to inferior reflectivity of the ringdown cavity mirrors beyond ~1620 nm (as is already evident in the survey spectrum of Fig. 1). Note also that the blue trace for H₂O vapor takes at least 1 minute to settle, owing presumably to equilibration by adsorption of H₂O on the ringdown cell walls, etc. after the gas flow is switched from pure N₂ to outdoor air.

From the results in Fig. 7, we deduce that the partial pressures of CO₂, CH₄, and H₂O vapor in outdoor air as sampled are 285 mTorr, 1.7 mTorr, and 10.5 mTorr, respectively. The corresponding minimum detectable mixing ratios from these measurements are ~2 ppmv CO₂, ~0.065 ppmv CH₄, and ~100 ppmv H₂O vapor in air. In future measurements of CH₄, using a laser that is more stable around ~1650 nm, we project that the ~3-times stronger CH₄ absorption feature at 1650.956 nm will improve the minimum detectable mixing ratio by an order of magnitude to <0.005 ppmv. These simultaneously recorded CRD-spectroscopic results compare favorably with our previous results obtained by non-CRD methods such as temporally multiplexed wavelength-modulation spectroscopy [12,13] (see Sec. 1.1).

The wavelength range that has been used, as described above for simultaneous CRD-spectroscopic trace-level monitoring of CO₂, CH₄, and H₂O vapor in air, is amenable to multi-species detection of various other gases (e.g., NH₃, CO, and C₂H₂) which have absorption features in the same wavelength range. These could be detected by incorporating additional laser sources or by operating existing lasers at different wavelengths.

We find that our scheme of simultaneous multi-laser, multi-species cw-CRD spectroscopy tends to become imprecise and/or impractical with more than three lasers. The problem is that the dominant longitudinal ringdown-cavity mode associated with the wavelength of one monochromatic laser (from which CRD decay rates are measured) may be corrupted by weak overlapping transverse ringdown-cavity modes associated with the wavelengths of other monochromatic lasers. Our previous research (as in the context of Fig. 2 in [8]) has shown that such coupling to residual transverse modes can be reduced to less than 0.5% of the power of the overlapping longitudinal ringdown-cavity mode, by careful matching of the single-mode optical fiber to the longitudinal-mode geometry of the ringdown cavity. In the present work, the residual transverse modes are similarly minimized (as they are not visible in Fig. 3), but they can nevertheless still limit the CRD measurement accuracy.

For high-precision cw-CRD sensing applications, dual-wavelength operation is straightforward and effective, because it enables one monochromatic laser wavelength to be set at a characteristic absorption feature for a particular species of interest, with a second monochromatic laser wavelength set off resonance to record a simultaneous baseline reference. This approach [5] helps us to correct any drift of the baseline decay rate in real-world applications, where operating conditions might change substantially and frequently.

In order to monitor more than three distinct species with high sensitivity and precision, it would be best to combine our novel simultaneous multi-laser approach to cw-CRD spectroscopy with well-proven temporally multiplexed techniques (as surveyed in Sec. 1.2). The number of distinct species that can then be monitored is limited primarily by the number of suitable single-wavelength tunable lasers that are available.

4. Conclusions

Our multi-laser, multi-species approach to near-IR cw-CRD spectroscopic sensing employs communications-band fiber-optical and photonics components to devise a compact, robust, field-deployable, economical instrument that is well suited to highly sensitive detection of trace gas mixtures in various industrial, medical, agricultural, or environmental settings. In the context of this paper, this technique has been adapted to achieve accurate, quantitative monitoring of greenhouse-gas molecules at trace levels in the atmosphere.

We have shown in this paper that active scanning of laser wavelength(s) is not necessarily required to perform reliable trace-level monitoring of particular gas-phase molecules for which interference-free spectral characteristics have been established. This reduces the required wavelength-tuning performance of the laser sources that can be employed. The multi-laser variant of our distinctive swept-cavity near-IR cw-CRD technique uses several fixed-wavelength tunable diode lasers, each pre-tuned to a characteristic on- or off-resonance feature of a gas molecule of interest, with their outputs combined into a single-mode optical fiber and directed simultaneously to the ringdown cavity. Each component of the combined laser radiation attains CRD resonance at a separate part of the cavity-sweep cycle. This provides an attractive alternative to traditional multi-species spectroscopic approaches, in which different characteristic wavelengths are spatially separated via dispersive optics or wavelength-selective filters and monitored by multiple photodetectors or a CCD array.

As in most forms of spectroscopic sensing (including those in which a continuously scanned monochromatic source or a dispersed broadband source is used to record an extensive portion of a spectrum), “false positive” signals due to spectral interferences occasionally arise. However, as demonstrated in Secs 3.1 and 3.2, we have carefully characterized the frequency stability of the diode lasers to be used and critically selected spectral features for the molecular species of interest, thereby ensuring that multi-laser, multi-species cw-CRD signals as in Sec. 3.3 (notably Fig. 7) are virtually free of interferences from any unidentified species that is likely to accompany the targeted greenhouse-gas molecules in ambient air. Nevertheless, in applying our multi-laser, cw-CRD method to real-world environmental sensing, we need to be alert to avoid potential interferences (e.g., from unidentified atmospheric pollutants). Moreover, our approach is likely to be less viable when applied to sensing of trace gas mixtures for which the chemical composition is unknown or difficult to anticipate.

Finally, to summarise this paper, our novel simultaneous multi-species detection technique has been described in detail for the first time, after a survey (in Sec. 1.1) of our own earlier work on rapidly swept cw-CRD spectroscopy [5–8,10] and a review (in Sec. 1.2) of other previous near-IR variants of multi-species cw-CRD spectroscopy [14–21], including an extensive (1540–1640 nm) slowly scanned survey spectrum of the trace-level components CO₂, CH₄, and H₂O vapor in outdoor air, as in Fig. 1. In Sec. 2, appropriate operating conditions have been established for a rapidly swept cw-CRD spectrometer, capable of simultaneously detecting three distinct species with characteristic wavelengths λ₁, λ₂, λ₃ and avoiding potentially troublesome interferences via residual coupling into transverse modes of the ringdown cavity. After careful evaluating and selecting suitable tunable diode lasers (in Sec. 3.1), we have identified (in Sec. 3.2) appropriate characteristic wavelengths for simultaneous detection of CO₂, CH₄, and H₂O vapor in air by rapidly swept cw-CRD spectroscopy; some of these characteristic wavelengths have not previously been used in other published work. Sec. 3.3 reports definitive rapidly swept cw-CRD spectroscopic measurements that have been made simultaneously for CO₂, CH₄, and H₂O vapor in outdoor air. This first detailed report (after a preliminary conference presentation [11]) of our novel multi-species, multi-laser approach to simultaneous cw-CRD detection allows us to envisage further applications to real-time spectroscopic sensing of multi-component gas mixtures.