1. Introduction

A fundamental understanding of complex processes in such challenging fields as, e.g., gas-phase synthesis of nanoparticles, combustion, plasma, or medical breath analysis, requires suitable diagnostics techniques that offer high sensitivity and selectivity. Precise measurements of species concentrations and temperatures, coupled with a robustness to broadband losses (originating from, e.g., dust or window reflections) are prerequisites for the acquisition of valid data. Additionally, a capability of multi-species analysis is often required, e.g., to unravel complex chemical reaction kinetics. Such demanding requirements can currently be fulfilled satisfactorily only by few spectroscopic techniques.

In general, multi-species detection can be achieved either by utilizing broadband light sources, e.g. supercontinuum laser sources [1] or, alternatively, by multiplexing several narrow-band lasers [2]. However, to achieve detection limits below parts-per-million (ppm), the absorption path length L_eff needs to be substantially increased. Commonly, this is implemented by utilizing high-finesse passive cavities [3]. However, while such configurations enable substantial sensitivity enhancement compared to single-pass absorption, they impede in situ measurements in harsh environments, mainly due to a high susceptibility to broadband losses (e.g., from dust, soot or the presence of windows). As a consequence, such established spectroscopic technique as, e.g., cavity ring-down spectroscopy (CRDS) experience difficulties in harsh environments [4]. Although some emerging techniques based on cavity-enhanced frequency combs demonstrate promising results [5,6], they still suffer from aforementioned difficulties. Besides that, these frequency-comb techniques are complex, making the setup and operation demanding.

An alternative technique enabling broadband and sensitive spectroscopic measurements in harsh environments is intracavity absorption spectroscopy (ICAS) [7]. Its characteristic feature, and the significant difference compared to most other spectroscopic techniques, is the positioning of the absorber inside the resonator of a broadband laser source. The successive interaction of laser photons with the broadband laser gain and the narrowband absorber determines the emission spectrum of the laser. While the light travels through the resonator, the wavelength-independent losses inside the resonator, caused by, e.g., scattering or broadband absorption, are compensated by the gain of the laser medium. Depending on the laser medium used, these broadband losses can be several tens of percent per roundtrip, without significantly affecting the laser process. In contrast, with passive-cavity techniques, e.g. CRDS, broadband losses are not compensated and the light intensity decays fast below the noise level. These advantages make ICAS ideal for spectrally-resolved absorption measurements in “dirty” environments, such as particle-laden flows.

In contrast to the broadband losses, the narrow-band losses originating from narrow-band absorbers cannot be compensated by the laser gain in the same manner if the homogeneous linewidth of the gain is larger than the absorption linewidth. If some of the laser modes suffer from selective losses, caused by the narrowband absorber, they will possess smaller net gain than neighboring modes. With each round trip inside the resonator, this effect accumulates and leads to an imprint of the absorption lines directly onto the broadband laser emission spectrum. The depths of the imprinted absorption lines, however, do not grow infinitely. Depending on the gain medium and operation parameters of the laser, such processes as four-wave mixing, Rayleigh and Brillouin scattering can limit the mode competition, and thus, the sensitivity [8]. Consequently, these processes determine a certain saturation time t_s that marks the maximum sensitivity of the spectroscopic system, corresponding to the sensitivity in cw operation.

An appropriate measure of sensitivity of an ICAS system is the effective absorption path length L_eff that corresponds to the accumulated path through the absorption sample inside the resonator. In the experiment, L_eff can be determined as L_eff = K/α by measuring the absorption signal, K = ln(I₀/I), with I and I₀ being laser intensities with and without the intracavity absorber. The absorber is characterized by the absorption coefficient α(ν), which is a product of the concentration n and the absorption cross-section σ(ν). This determination of L_eff enables a subsequent determination of species concentrations from the observed absorption.

The highest sensitivity is achieved in cw operation of the laser; however, in this case the sensitivity can depend on various laser parameters, such as pump rate, gain profile and cavity loss. In pulsed operation mode shorter than t_s, the spectral sensitivity can be precisely determined by recording the spectrum at a time t after the onset of laser oscillations

Since the absorption signal K is measured as the relative change of the laser intensity at the center of an absorption line, fluctuations of the total laser power do not influence the sensitivity and accuracy of ICAS measurements. This is one of the most important advantages of ICAS compared to other spectroscopic techniques, as it enables in situ measurements of multiline absorption spectra of various molecules in hostile environments, as demonstrated, e.g., in human breath [9], flames [10,11], plasma discharges [12] and shock-tubes [13].

In this paper we present a new ICAS system based on a homemade broadband Cr⁴⁺:forsterite laser that is continuously tunable in the spectral range of 1196–1380 nm. We demonstrate its high sensitivity to intracavity absorption, corresponding to absorption lengths of L_eff = 2500 km. The Cr⁴⁺:forsterite laser is applied to highly sensitive measurements of HCl and HF in the intracavity absorption cell, and to atmospheric absorption measurements of H₂O, O₂, CO₂, and CH₄. The demonstrated detection limit for HF of p_min = 2 ppt emphasizes the possibilities of our spectroscopic system.

From a spectroscopic point of view, the spectral range of 1.2–1.4 µm is highly interesting. Besides the well-documented absorption bands of such species as H₂O, O₂, OH, CO, CO₂, CH₄, C₂H₂, NO, N₂O, NH₃, HF, HCl, HBr, HI, OCS, HCN, H₂S, and H₂ [14], it also contains absorption lines of other important molecules, which are either investigated only sparsely, e.g., HO₂ [15], or have so far not been listed in this spectral range at all, e.g., CH₃OH.

Despite the excellent suitability of this spectral range for spectroscopic detection, no suitable laser source operating in the range 1.2–1.4 µm has been used for ICAS up to now. Although few ICAS measurements with color-center lasers operating in this spectral range have been reported earlier [16,17], the operation of these lasers at room temperature is possible only in a pulsed mode due to the requirement of high pump power to reach the laser threshold. As a result, the sensitivity to intracavity absorption was limited.

The Cr⁴⁺:forsterite laser presented here fills the spectral gap between 1.2 and 1.4 µm enabling various challenging ICAS applications. Although the first Cr⁴⁺:forsterite laser has been realized already in 1988 [18], it has never been applied to spectroscopic measurements so far. Instead, the main research activities are directed towards improvements of high-power efficiencies [19,20] and to the development of ultrafast lasers and frequency combs [21,22]. Although the main goal of our work is directed towards the development of a spectroscopic tool based on the Cr⁴⁺:forsterite laser, it should be noted that the achieved laser threshold of P_thr = 0.8 W and the laser efficiency of η_cw = 6% (at an output-mirror transmission of T = 0.4%) are comparable with results achieved in works that focus on the efficiency of laser performance [19,20,23,24].

2. Experimental setup

The experimental setup of our ICAS system is shown in Fig. 1.

 

Fig. 1 Experimental setup.

Download Full Size | PDF

The Cr⁴⁺:forsterite crystal (Solix Ltd., Minsk, Belarus) has the dimensions of 4 × 4 × 12 mm³. It is cut along the crystallographic a-axis and at Brewster’s angle of 58.55° to the b-axis. The polarization-dependent absorption coefficients at the pump wavelength of λ ≈1070 nm are estimated to be α_b  =  1.4 cm^–1 for E ǁ b, and α_c  =  0.8 cm^–1 for E ǁ c. This corresponds to an absorption in the crystal of about 80% for pump light with E ǁ b, and of 60% when E ǁ c. The figure of merit (FOM) for E ǁ b is specified as α_1.07/α_1.25 ≈200, with α_1.07 and α_1.25 being the absorption coefficients at wavelengths of 1.07 and 1.25 µm, respectively. The high FOM reflects a high optical quality of the crystal, indicating low losses for laser light (α_1.25 = 0.007 cm^–1) and facilitating efficient laser operation. It should be noted that so far, most of the published works on Cr⁴⁺:forsterite lasers were performed with crystals having a factor 5–10 smaller FOM [25,26]. This may be partly attributed to the trend of using high doping concentrations (to reach high output powers) that comes with additional losses [27]. Nevertheless, the high FOM of the crystal used in our work obviously implies significant progress in crystal growth of Cr⁴⁺:forsterite by the manufacturer, presumably by prolonged (~800 h) high-temperature oxidizing annealing. This process was recently shown to significantly increase the Cr⁴⁺ content and, more importantly, drastically reduce the parasitic Cr²⁺ (and Cr³⁺) content in the crystal, thus substantially enhancing the FOM [27].

For thermal management, the crystal is wrapped in an indium foil (Haines & Maassen) and incorporated into a copper block, which is held at room temperature (~20°C) by water cooling. The copper block is mounted onto an alignment unit consisting of two linear translation stages, one rotation stage and one goniometer. As a result, this unit enables precise crystal alignment in 4 coordinates: linear shifts of the crystal along the x- and y-axes, rotation around the z-axis to adjust the Brewster’s angle, and rotation around the axis perpendicular to the entrance surface of the crystal (at Brewster’s angle to the b-axis) to align the laser polarization with the crystallographic axis b, thus avoiding birefringence effects (see section 3).

The V-shaped laser resonator is defined by the mirrors M1, M2, and M3 with their radii of curvature being r₁ = 75 mm, r₂ = 150 mm, and r₃ = ∞. As a result, the resonator has a focusing point between M1 and M2 at which the crystal is positioned, and a collimated arm between M2 and M3, where a sample cell can be introduced. The V-shape is necessary to compensate the astigmatism of the crystal by an opposite-sign astigmatism introduced by the tilted mirror M2 [28]. For this purpose, the optimum angle between the two arms of the resonator is estimated to be θ = 27.8°. The mirrors M1 and M2 provide high transmission for the pump light at 1070 nm and are simultaneously highly reflective in the range of the expected laser generation (between 1.2 and 1.4 µm), while the end mirror M3 is made to transmit 0.4% of the laser light. Furthermore, to suppress interference fringes from back-reflected laser light, the backsides of all mirrors are anti-reflection coated. The total length of the resonator (M1–M2–M3) is L = 115 cm, the distance between M1 and M2 is 15 cm. An open sample cell with a length of about l = 29 cm can be placed inside the resonator. For this cell, the filling factor of the resonator with an absorber is β = l/L = 25%.

As a pump source we use an ytterbium fiber laser (IPG Laser, YLM-20-SC) that provides an optical power of up to 20 W at λ ≈1070 nm with diffraction-limited beam quality. The pump light is focused into the crystal by an anti-reflection coated plano-convex lens (f = 80 mm) positioned in front of the mirror M1. For the purpose of spectral tuning of the Cr⁴⁺:forsterite laser, a pellicle (etalon) with a thickness of about 2 µm (Thorlabs, BP108) is placed between M2 and M3. By rotating the pellicle perpendicularly to the optical axis of the resonator, its transmission maximum can be spectrally shifted, providing an effective way of broadband spectral tuning of the laser. The total tuning range achieved with this laser is 7246–8361 cm^–1 (1196–1380 nm). The laser threshold in the gain maximum is P_thr = 0.8 W.

The laser light is coupled to a spectrometer (Jarell Ash 78-467, 1 m, 295 grooves/mm) and recorded with a CCD line-scan camera (Goodrich Sensors Unlimited, SU-LDV, 1 × 1024 pixels) with a controllable detection window. The maximum spectral resolution of the recording system is Δν = 0.11 cm^–1, corresponding to 3.3 GHz or Δλ ≈0.02 nm at λ = 1.3 µm.

Depending on the spectroscopic task, the laser can be operated either in cw or in pulsed mode. If the highest achievable sensitivity is not required, measurements are performed in a pulsed-repetition mode with a selected effective absorption path length L_eff. Duty cycles of 10% are usually used to reduce the thermal load on the crystal in order to keep a constant threshold. For example, if the laser is switched on for 1 ms every 10 ms, precise measurements of intracavity absorption with effective absorption paths between 3 and 300 km can be performed over a broad spectral range by controlling the position of the detection window. If higher sensitivity is required, we have also used other “On/Off” patterns of laser pulses, such as 3/30 ms or 10/50 ms. For the purpose of noise-reduction typical for multi-mode lasers [8], each spectrum is averaged over 2000 laser pulses. The residual spectral noise is usually less than 1% (RMS). All experimental spectra presented in this work are recorded at ambient conditions, i.e. at room temperature of T ≈300 K and at normal atmospheric pressure of p_tot ≈1 bar.

The synchronization of the pump-laser excitation with the recording window of the CCD camera is achieved using a two-channel pulse generator (Tektronix, AFG3022B). Figure 2 demonstrates the typical behavior of the pump and Cr⁴⁺:forsterite laser power, with and without a DC offset in the pump power.

 

Fig. 2 Diagrams of the control signal for the pump laser (green), the pump laser emission (blue) and the Cr⁴⁺:forsterite laser emission (red). Left: modulation of the pump laser at full modulation depth. Right: modulation with a DC offset of the control signal by 0.4 V.

Download Full Size | PDF

Figure 2(a) shows a full-depth modulation of the control voltage of pump-laser excitation from zero to 0.7 V, corresponding to a pump-power modulation in the range from zero to 2 W (Fig. 2(b)), which is 150% above the threshold of the Cr⁴⁺:forsterite laser. In this case, the pump-laser emission appears 250 µs after the beginning of excitation (marked by a dashed vertical line) and exhibits a strong spike of relaxation oscillations at the beginning of pump-laser emission. About 20 µs later, the emission of the Cr⁴⁺:forsterite laser follows the spiking pattern of the pump laser (Fig. 2(c)). The delay between the beginning of pump-power excitation and the start of laser emission depends on the time required to build up the required inversion in the pump and Cr⁴⁺:forsterite lasers and, therefore, varies with pump power, cavity loss and spectral tuning. In contrast, the position of the detection window of the CCD camera is fixed in respect to the beginning of excitation. Consequently, induced relative shifts between the start of laser emission and the detection time can lead to a variation in the effective absorption path length at which the measurements are performed, thus reducing the accuracy of absorption measurements.

To solve this problem, we add a DC offset of 0.4 V to the modulation amplitude of the control voltage of pump-laser excitation (Fig. 2(d)), such that the low pump-power level is held just below the threshold of the Cr⁴⁺:forsterite laser, as shown in Fig. 2(e). In this case, the delay between the excitation of the pump laser and the beginning of Cr⁴⁺:forsterite laser emission is very small and almost constant (Fig. 2(f)), enabling time-resolved absorption measurements with fixed L_eff. Additionally, relaxation oscillations in both pump- and Cr⁴⁺:forsterite-laser emission almost disappear (Figs. 2(e) and 2(f)).

3. Spectral characteristics of the Cr⁴⁺:forsterite laser emission

Spectral quality

Since forsterite (Mg₂SiO₄) is a biaxially birefringent material, a proper alignment of the crystal in the cavity is very important to achieve high spectral quality. As for all orthorhombic crystals, the crystallographic axes (a, b, and c) coincide with the axes of the refractive-index ellipsoid. If the incident light is not polarized along one of these axes, the individual polarization components experience different phase-velocities, since the refractive indices differ for all polarization directions, i.e., n_a ≠ n_b ≠ n_c. Consequently, to avoid birefringence effects, our crystal needs to be oriented such that the laser polarization coincides with the b-axis. If this is not the case, then there will be periodic reflection losses at the Brewster cut surfaces. This is because the Brewster’s angle determines the direction of laser polarization, which is independent upon rotation around the axis normal to the Brewster-cut surface. However, a rotation of the crystal around this axis changes the relative orientation of the laser polarization to the crystallographic b-axis. Since different polarization components experience different phase velocities, the net polarization changes after a passage through the crystal, and this transforms into higher reflection losses at the Brewster-cut surfaces. Only wavelengths for which the phase difference between the polarization components is equal to a multiple integer of 2π experience no additional losses.

In some cases, this birefringence effect can be very useful, e.g., it is the basis for the Lyot filter [29], which is often used as a wavelength-selective element for laser tuning. For ICAS, however, wavelength-selective losses can be very detrimental as they usually lead to fringes in the broadband spectra. In our experiment, the elimination of such fringes is performed by precise rotation of the crystal around the axis perpendicular to the Brewster-cut surface. For this purpose, the crystal is incorporated into a goniometer. Figure 3 demonstrates broadband emission spectra of the Cr⁴⁺:forsterite laser with atmospheric absorption in the cavity, recorded with the effective absorption path length of L_eff = 10 km at different angles between the crystallographic b-axis and the polarization of optical field.

 

Fig. 3 Influence of crystal rotation on the emission spectrum of the Cr⁴⁺:forsterite laser, shown for different angles between the crystallographic b-axis and the polarization of optical field in the cavity.

Download Full Size | PDF

As can be seen from Fig. 3, birefringence effects lead to strong channeling in the emission spectrum already for a misalignment angle of 1°. Therefore, a proper crystal adjustment is crucial for highly sensitive measurements of intracavity absorption spectra.

Homogeneous broadening of the gain

Sensitive measurements of intracavity absorption are only possible if the homogeneous broadening of the laser gain is larger than the absorption linewidth. In the following experiment the estimation of the homogeneous gain broadening of the Cr⁴⁺:forsterite laser is performed by recording the so-called spectral condensation typical for lasers with a homogeneously broadened gain profile [8]. The emission spectrum of such laser narrows down with the duration of laser oscillations t, according to

Figure 4 shows selected emission spectra of the Cr⁴⁺:forsterite laser recorded at different times t after the onset of laser oscillation, with a spectral resolution of Δν_res = 1 cm^–1.

 

Fig. 4 Emission spectra of the Cr⁴⁺:forsterite laser recorded at different times t after the onset of laser oscillation.

Download Full Size | PDF

To avoid additional spectral narrowing, these spectra are recorded without the pellicle in the cavity. The laser spectra in Fig. 4 clearly show spectral condensation and additionally a spectral shift. The shift occurs since the laser emission starts at the maximum of the gain profile and gradually shifts to the effective gain maximum given by the difference between gain and loss.

Figure 5 shows the development of the emission bandwidth of the laser with the laser generation time.

 

Fig. 5 Emission widths of laser spectra (circles) fitted by Eq. (2) (line).

Download Full Size | PDF

As can be seen, Eq. (2) is well suitable for the description of this development for the initial stage of laser oscillations. From a fit to the experimental data for t ≤ 100 µs using Eq. (2), and taking into account the cavity loss rate of γ = 4.3 × 10⁶ s^–1, we obtain the homogeneous linewidth of the gain of Q = 770 cm^–1 (FWHM).

For t > 100 µs, the experimental bandwidth is larger than predicted by Eq. (2). This deviation occurs due to various processes in the laser, such as residual inhomogeneous broadening, mode coupling and spontaneous emission, retarding the spectral condensation. As a consequence, the spectral width of the laser in cw operation is still about Δν_cw = 3 cm^–1, despite the fact that it should condense to a single mode according to Eq. (2). With the obtained value for the homogeneous linewidth of the gain of Q = 770 cm^–1, we can expect to achieve a laser tuning range of more than 1000 cm^–1 by applying sufficient pump power. The largest tuning range for a 1.06-µm pumped Cr⁴⁺:forsterite laser reported in literature was 1130–1367 nm (Δν = 1535 cm^–1) [30].

Tuning range

Figure 6 (top) demonstrates individual emission spectra of our Cr⁴⁺:forsterite laser recorded by stepwise rotation of the intracavity pellicle with a spectral resolution of Δν_res = 0.6 cm^–1. All intermediate spectral peak positions can be reached as well.

 

Fig. 6 Top: Emission spectra of the Cr⁴⁺:forsterite laser recorded by stepwise rotation of the pellicle in the cavity filled with ambient air. A spectral shift (color change) corresponds to an increment of the pellicle’s rotation angle. Bottom: Calculated transmission spectra (HITRAN database) of air containing H₂O (L_eff = 1 km, p_H2O = 0.8%) and O₂ (L_eff = 10 km, p_O2 = 21%).

Download Full Size | PDF

The total tuning range achieved with this laser is 7246–8361 cm^–1 (1196–1380 nm). To adapt the spectral sensitivity to the magnitude of atmospheric absorption, the effective absorption length has been adjusted (by shifting the recording-window position) to L_eff = 1 km between 7200 and 7700 cm^–1, and to L_eff = 10 km between 7700 and 8400 cm^–1. Note that laser emission is almost completely suppressed between 7280 and 7380 cm^–1 due to strong water-vapor absorption, while only distinct peaks appear between 7240 and 7280 cm^–1. For comparison, Fig. 6 (bottom) shows calculated transmission spectra (HITRAN database [14]) of air containing typical 0.8% of H₂O (L_eff = 1 km) and 21% of O₂ (L_eff = 10 km). An operation of the laser in a water-vapor-free environment should enable a tunability beyond 1.38 µm. Other factors that limit the tuning range include the spectral reflectivity profiles of the cavity mirrors and the thickness of the pellicle.

4. Sensitivity to intracavity absorption

The sensitivity of our ICAS-system is estimated by recording laser emission spectra containing the weak absorption line of atmospheric oxygen at ν = 8030.17 cm^–1. Figure 7 shows selected experimental spectra, together with calculated spectra of atmospheric absorption of O₂ and H₂O from the HITRAN database. Blue spectra are recorded at a pump rate of η = P/P_thr = 1.5, while the red spectrum is recorded near the laser threshold at η = 1.1.

 

Fig. 7 Emission spectra of the Cr⁴⁺:forsterite laser with atmospheric absorption of H₂O and O₂ recorded at different durations of the laser oscillation t. Blue spectra are recorded at a pump rate of η = P/P_thr = 1.5, while the red spectrum is recorded near the laser threshold at η = 1.1. Calculated spectra of H₂O and O₂ (HITRAN) are shown on top.

Download Full Size | PDF

A steady growth of the absorption signals with time t can be observed in Fig. 7 until 9 ms. However, it is obvious, that the absorption signal at η = 1.1 is larger in comparison to η = 1.5. Figure 8 shows the dependence of the spectral sensitivity, i.e., of L_eff, on the generation time for three different pump rates.

 

Fig. 8 Effective absorption path length L_eff vs. generation time t for different pump rates η. The straight line corresponds to L_eff = ct.

Download Full Size | PDF

The evaluation has been performed by measuring the absorption signal K = ln(I₀/I) of the oxygen absorption line at ν = 8030.17 cm^–1 for various laser generation times t and calculating the corresponding effective absorption path length L_eff = K/α from the HITRAN database [14]. Figure 8 shows that the sensitivity grows linearly, according to L_eff = ct, in the beginning of the laser process for all presented pump rates. However, depending on the pump rate, the sensitivity saturates at different generation times. The reason for the saturation behavior is the influence of various laser processes that limit the maximum sensitivity. Typical examples are spontaneous emission, Rayleigh scattering, or non-linear mode coupling, e.g., by four-wave mixing or stimulated Brillouin scattering [8]. While some of these parasitic effects are intrinsic for a specific laser type, others can be controlled by appropriate operation parameters of the laser. The latter being especially true for non-linear effects, which can be reduced by operating the laser at lower pump power as it is demonstrated in Fig. 8. As a result, the maximum sensitivity demonstrated in the experiment for η = 4.5 corresponds to L_eff = 400 km. In contrast, for η = 1.1 the sensitivity continues further growing to L_eff = 2500 km and is not yet saturated at t = 9 ms.

An exact evaluation of the sensitivity for generation times t > 9 ms is hampered here by the overlap of the reference absorption line with strong absorption lines of H₂O. Nevertheless, the maximum sensitivity for η = 1.1 can be roughly estimated to be at least L_eff = 10,000 km. An exact estimation of the maximum sensitivity achievable with η ≤ 1.1 requires a suitable sample with a well-defined, isolated and weak absorption line. In the following, estimations of the noise-equivalent detection limits of various species absorbing in the spectral range of the Cr⁴⁺:forsterite laser will be performed assuming the conservative value of the maximum effective absorption path length of L_eff = 2500 km and a noise level of 1%.

5. ICAS of atmospheric H₂O, O₂, CO₂, and CH₄

As introduced above, the emission range of the Cr⁴⁺:forsterite laser is well suited for the monitoring various atmospheric gases and pollutants, such as H₂O, O₂, CO₂, CH₄, N₂O, and NH₃. In this section, we present demonstrations of sensitive detection of the important atmospheric gases H₂O, O₂, CO₂, and CH₄.

Figure 9 shows an emission spectrum of the Cr⁴⁺:forsterite laser (red), recorded in ambient air around 7915 cm^–1 and at L_eff = 10 km, containing absorption lines atmospheric O₂ and H₂O. The absorption band of O₂ around 7880 cm^–1 shows only a small interference with water-vapor absorption and enables precise multiline measurements of O₂, as well as simultaneous measurements of H₂O and O₂.

 

Fig. 9 Bottom: Emission spectrum of the Cr⁴⁺:forsterite laser with intracavity atmospheric absorption (red). Top: Normalized experimental spectrum (blue) and calculated spectra (HITRAN, L_eff = 10 km) of O₂ (brown) and H₂O (black).

Download Full Size | PDF

After a normalization of the laser emission spectrum by the spectral envelope (obtained by polynomial fitting of laser intensities without absorption), the resulting spectrum in Fig. 9 (blue) can be compared with theoretical spectra of atmospheric O₂ (brown) and H₂O (black) absorption calculated from the HITRAN database using L_eff = 10 km and by taking into account Doppler- and collisional broadening and the spectral resolution of the recording system. The good agreement between experimental and calculated absorption features proves an accurate control of the specified sensitivity and illustrates the capability of our system to perform precise and sensitive spectroscopic measurements. With L_eff = 2500 km the detection limit for the strongest O₂ line located at ν = 7880.64 cm^–1 is estimated to be 3 ppm.

A demonstration of sensitive monitoring of atmospheric CO₂ with our spectroscopic system is presented in Fig. 10. The laser is operated around 8206 cm^–1 and the spectral sensitivity is set to L_eff = 100 km.

 

Fig. 10 Bottom: Emission spectrum of the Cr⁴⁺:forsterite laser with intracavity atmospheric absorption (red). Top: Calculated spectra (HITRAN, L_eff = 100 km) of CO₂ (green) and H₂O (grey).

Download Full Size | PDF

The experimental spectrum shown in Fig. 10 is recorded with a tenfold increase in sensitivity compared to the spectrum in Fig. 9, and thus it is narrower due to spectral condensation. Nevertheless, it enables broadband spectroscopic measurements of atmospheric CO₂ and H₂O. It should be particularly noted that in the spectral region of 8205–8208 cm^–1 CO₂ can be measured almost without interferences with water-vapor lines. In the operation mode with L_eff set to 100 km, the detection limit for the CO₂ line located at 8206.40 cm^–1 is estimated to be 5 ppm, while the detection limit for the H₂O line located at 8210.55 cm^–1 is about 3 ppm. With L_eff = 2500 km the estimated detection limit for the strongest CO₂ line located at 8304.34 cm^–1 results in 150 ppb.

An extremely high sensitivity for the detection of H₂O can be achieved in the spectral range around 7350 cm^–1, cf. Figure 6. However, due to strong water absorption in ambient air in this spectral range, an experimental validation would require a hermetic encapsulation of the experimental setup from the atmosphere. With L_eff = 2500 km the detection limit estimated for the strongest H₂O line located at 7327.68 cm^–1 is 25 ppt.

Another atmospheric species that can be monitored with our spectroscopic system is methane. Together with CO₂ and H₂O, CH₄ plays an important role in the greenhouse effect. Figure 11 shows an experimental spectrum (green) recorded around ν = 7596 cm^–1 with a sensitivity set to L_eff = 660 km.

 

Fig. 11 Laser emission spectrum (green) containing absorption lines of atmospheric CO₂, CH₄, and H₂O recorded with L_eff = 660 km, superimposed with a corresponding calculated spectrum (red, HITRAN: p_H2O = 1.5%, p_CO2 = 400 ppm and p_CH4 = 2.2 ppm).

Download Full Size | PDF

The experimental spectrum contains the absorption line of atmospheric CH₄ at ν = 7596.21 cm^–1, as well as absorption lines of CO₂ and H₂O on the wings. By fitting the experimental spectrum with a calculated spectrum (red, HITRAN), the atmospheric concentration of the species H₂O, CO₂ and CH₄ have been determined to be p_H2O = 1.5%, p_CO2 = 400 ppm and p_CH4 = 2.2 ppm. The determined methane concentration is slightly higher than the recent (July 2018) globally averaged mean atmospheric value of 1.85 ppm, determined from marine surface sites [31]. However, our measured value is in good agreement with typical CH₄ concentrations of 1.8–3.9 ppm in urban environments [32].

The detection limit for the recorded absorption line of CH₄ with L_eff = 660 km is determined to be 35 ppb. However, with L_eff = 2500 km, the detection limit for the strongest CH₄ absorption line located at 7510.15 cm^–1 is estimated to be 2 ppb. The presented measurements demonstrate a good applicability of our spectroscopic system for sensitive multicomponent atmospheric diagnostics.

6. Measurements of HCl and HF

Measurements of HCl and HF are carried out in an open gas cell located in the laser cavity. To generate a small concentration of HCl in the laser cavity, a droplet of hydrochloric acid is released into the cell. During the evaporation of the acid, and thus, time-variable HCl concentrations in the gas phase, a series of laser emission spectra with different HCl absorption signals is recorded. For normalization, laser spectra without absorption by HCl are recorded in advance. Figure 12 shows the corresponding recorded and evaluated spectra. The absorption spectra with HCl in the resonator (red and blue in the bottom diagrams) are normalized by spectra recorded without HCl (grey). The resulting transmission spectra (red and blue in the top diagrams) are subsequently fitted by calculated HITRAN spectra (black) by varying the partial concentration of HCl.

 

Fig. 12 Emission spectra of the Cr⁴⁺:forsterite laser recorded with L_eff = 10 km (left) and with L_eff = 100 km (right) with intracavity atmospheric absorption (bottom, grey) and with different concentrations of HCl in the gas cell (bottom, red and blue). Normalized emission laser spectra with HCl absorption (top, red and blue) superimposed with calculated HCl spectra from the HITRAN database (black) are shown in the top diagrams. Calculated spectra of H₂O (top, green) are shown for comparison.

Download Full Size | PDF

The concentrations of HCl given in Fig. 12 are determined from HITRAN fits (black) performed with the assumption that the whole cavity is filled with HCl. The real concentration in the open cell is about a factor of four higher. Owing to the large emission bandwidth of the spectra recorded with L_eff = 10 km, two absorption lines of HCl, located at 8218.86 and 8224.52 cm^–1, can be monitored simultaneously (Fig. 12, left). Although the weaker absorption line at 8218.86 cm^–1 interferes with a H₂O line at 8219.21 cm^–1, the normalization procedure enables a complete elimination of H₂O absorption lines and a clean detection of HCl absorption. The noise-equivalent detection limit for HCl with L_eff = 100 km corresponds to 300 ppb. With L_eff = 2500 km, the expected detection limit for the strongest HCl line located at ν = 8278.74 cm^–1 is estimated to be 6 ppb.

An outstanding sensitivity is demonstrated with our Cr⁴⁺:forsterite laser to intracavity absorption of HF. We were able to detect the absorption of gaseous HF released when giving a droplet of HCl solution (imitating stomach acid) in the open intracavity cell on a pinch of table salt containing 0.2 ppm of potassium fluoride. Figure 13 shows laser emission spectra recorded in the atmosphere at L_eff = 100 km with (bottom, red) and without (black) additional HF absorption around 7790 cm^–1.

 

Fig. 13 Bottom: Laser emission spectra recorded at t = 330 µs in air (black) and with additional 40 ppb of HF (red). Top: The normalized experimental spectrum (top, red) agrees well with the calculated spectrum of HF form the HITRAN database (blue). Calculated spectra of H₂O and O₂ are shown for comparison.

Download Full Size | PDF

The normalized experimental spectrum (top, red), consisting of only one absorption line at 7788.86 cm^–1, agrees well with the calculated HITRAN spectrum. The concentration of HF, determined under the assumption that the whole cavity is filled with HF, is equal to 40 ppb. It should be noted that all atmospheric absorption lines are completely eliminated in the normalized spectrum and the residual noise is as low as 0.3% (RMS), resulting in a detection limit for HF of 150 ppt. With L_eff = 2500 km and a noise level of 0.3%, the detection limit for the strongest HF line located at 7823.82 cm^–1 results in 2 ppt, being one of the lowest detection limits ever obtained for HF using a spectroscopic technique. In fact, only one, recently published experiment based on the CEPAS technique has demonstrated a lower noise-equivalent HF-detection of 0.65 ppt [33]. However, CEPAS is limited to single species at a time and clean environments, whereas our ICAS system enables in situ multicomponent measurements in various environments. Moreover, we believe that with an optimization of our setup, the current ICAS system is capable of achieving even lower detection limit for HF. As mentioned above, the maximum sensitivity of our system could be increased by at least a factor of four, while an increased spectral resolution combined with a reduced total pressure in the sample cell could easily give another enhancement factor of five. With these technical improvements, a detection limit for HF of about 0.1 ppt should be feasible.

Summary

This work describes the development of a broadband Cr⁴⁺:forsterite laser and its first applications for intracavity absorption spectroscopy of high sensitivity. The laser threshold at room temperature at the gain maximum around ν = 7900 cm^–1 is about P_thr = 0.8 W (λ_pump = 1.07 µm). The slope efficiency with an output-mirror transmission of T = 0.4% is η_cw = 6%. We estimate the homogeneous line width of the gain to be Q = 770 cm^–1 (FWHM). In the experiment, the laser was continuously tunable over the spectral range of 7246–8361 cm^–1 (1196–1380 nm). The spectral widths of individual spectra vary from 3 to 150 cm^–1, depending on the duration of laser oscillations. The broadband emission enables multiline and multicomponent spectroscopic measurements.

The spectral sensitivity of absorption measurements is precisely controlled by recording laser emission spectra at different times after the onset of laser oscillations providing the required effective absorption path length, L_eff = ct, up to the saturated value ct_s. The maximum sensitivity demonstrated in the experiment corresponds to L_eff = 2500 km. However, it has been shown that the sensitivity limit of our spectroscopic system amounts to at least L_eff = 10,000 km. The spectral noise in 2000-fold averaged spectra is less than 1%.

The ICAS-system has been applied for sensitive measurements of atmospheric compounds: H₂O, O₂, CO₂ and CH₄, as well as for measurements of HCl and HF molecules in the laser cavity. The detection limits estimated for the strongest lines of these species within the emission range of our laser are summarized in Table 1.

Table 1. Detection limits in ppb for species studied in this work.

View Table

The obtained results demonstrate the versatile possibilities of our spectroscopic system to perform highly sensitive multicomponent measurements in a broad spectral range, qualifying the Cr⁴⁺:forsterite laser for many challenging spectroscopic applications.