A Fiber-Coupled Quartz-Enhanced Photoacoustic Sensor for Dissolved Gas Detection

Abstract

Detection of dissolved gases in oceans is critically needed for global carbon cycle investigation. However, most in situ optical detection techniques, as far as we know, have restricted measurement efficiency due to large gas consumption. Herein, we develop a sub-mL photoacoustic gas sensor with a simple configuration. A single-mode fiber directly guides the incident laser into the photoacoustic cell without any other free-space optics. Thus, a reduced inner size of 12 mm × 6 mm × 4 mm enables the effective detection of limited dissolved gas. We employ methane (CH₄) as an example to demonstrate its sensing performance. The sensor achieves a good linear response with an R-square value of 0.9989 and a minimum detection limit of 1.1 ppmv, corresponding to a normalized noise equivalent absorption coefficient of 7.75 × 10⁻⁸ W·cm⁻¹·Hz^−1/2.

1. Introduction

As an important part of the global carbon cycle, oceans store 38,000 Gt of carbon [1], which is dozens of times greater than that achieved by atmospheric and terrestrial carbon storage. Greenhouse gases such as methane (CH₄) and carbon dioxide (CO₂) are crucial dissolved gases, especially in the extreme window of cold springs and hydrothermal solutions [2,3]. Precise measurement of the components, concentrations and spatiotemporal distribution of dissolved gases is essential to investigate the global carbon cycle, the deep sea ecological environment, and global climate change [4,5,6]. The main current techniques for dissolved gas detection include sampling analysis and in situ optical detection. Sampling analysis with precise laboratory instruments, such as gas chromatography [7,8,9], is a common method to analyze dissolved gases in surface seawater in the early stage of marine detection. However, potential sample change and discontinuous sampling may result in systematic errors, which must be considered in data analysis. Spectroscopic technologies, such as evanescent wave and Raman spectroscopy, have been implemented for high-concentration dissolved gas measurement [10,11], demonstrating the advantages of high precision and non-contact measurement, as well as the challenges of limited sensitivity. Further, the emergence of polymer materials with hydrophobic permeability [12] enables in situ deep-sea dissolved gas detection by encapsulating instruments in a cavity with high pressure-bearing capacity and separating the dissolved gases from seawater [13,14,15].

Infrared absorption-based spectroscopic sensors with high sensitivity and high selectivity have been developed for dissolved gas measurement. These techniques include non-dispersive infrared spectroscopy (NDIR), tunable diode laser absorption spectroscopy (TDLAS) and cavity-enhanced absorption spectroscopy (CEAS). Using a broadband laser as light source, the NDIR selects strong absorption lines of target gases for detection with filters [16]. TDLAS uses a narrow linewidth laser to obtain high spectral resolution and leverages wavelength modulation spectroscopy to further improve its sensitivity and immunity to interference [17]. CEAS enhances the detection sensitivity by significantly increasing the light–gas interaction length for a limited chamber size. Using the CEAS with an off-axis integrated cavity, Wankel et al. measured the dissolved CH₄ and its isotopes in oceans [18]. A long optical path contributes to the improvement of detection sensitivity. However, an increased gas consumption of as large as several hundred milliliters is needed. Limited by the low permeation rate of the water–gas separation membrane such as polydimethylsiloxane (PDMS), typically 20–40 μL/min [19], the equilibration of the detection system could take a very long time. Therefore, real-time dissolved gas detection remains unsolved for absorption-based spectroscopic techniques.

Photoacoustic spectroscopy (PAS) enables target gas measurement by the detection of the gas-absorption-induced acoustic wave using an acoustic transducer rather than the attenuated laser after long-distance absorption [20,21,22,23,24,25,26]. Therefore, PAS could offer a choice with low gas consumption. Since 2002, the invention of quartz-enhanced photoacoustic spectroscopy (QEPAS) used a tiny quartz tuning fork (QTF), known as the wrist watch crystal, as the acoustic transducer [27]. Its unique properties enable QEPAS gas sensors with excitation wavelength independence, low gas consumption and high noise immunity [28]. Continuous efforts have been contributed to this field towards practical measurement of numerous atmospheric trace gases, such as CO₂ with a minimum detection limit of 300 ppt and 90 ppb for CH₄ [29,30], and even state-of-the-art performance with an ultra-low detection limit and an ultra-wide dynamic range [31,32]. Usually, QEPAS systems force the laser to pass through the gap between the QTF prongs to avoid touching any surfaces by the employment of collimators, matching lenses and other spatial filtering mechanics if necessary [33]. This limits the further optimization of the QEPAS sensor footprint, which still requires gas samples of a few to a dozen mL for effective analysis.

In this paper, we develop a QEPAS sensor with sub-mL gas consumption for dissolved gas measurement. The spectrophone has a simple configuration with a micro-resonator for acoustic wave enhancement and a standard QTF for signal detection. Only a piece of single-mode fiber guides the excitation laser into the spectrophone without any other free-space optics. All the above components are assembled inside a stainless chamber with a total gas consumption of ~300 μL. We selected CH₄, as an example, to evaluate the sensor performance. An in situ dissolved CH₄ measurement has been carried out at the Haima Cold Spring Area of the South China Sea.

2. Materials and Methods

When the incident laser frequency is consistent with the molecular absorption line of the detection gas, the gas molecules absorb the optical radiation and release heat, which causes local expansion of the gas [34]. For QEPAS, a QTF serves as the acoustic transducer instead of a microphone in traditional PAS. Commercially standard QTFs have a resonant frequency of approximately 32.768 kHz. When the acoustic wave has the same frequency, QTF prongs symmetrically vibrate to generate piezoelectric signal, which can be governed by [35]

$$S=C\frac{\alpha PQ}{f}$$

(1)

where α is the absorption coefficient per unit concentration of the target gas, C is a constant describing the transfer function, P is the laser power, Q is the quality factor of the folk, and f is the resonant frequency. The QTF has a typical Q-factor of >100,000 in vacuum, while approximately 10,000 under atmospheric pressure due to the environmental gas-induced damping effect. In the QEPAS system, the resonant frequency and absorption coefficient can be precisely calibrated, thus the target gas absorption coefficient per unit concentration can be obtained by detecting the PAS signal.

In the implementation, we use a short microtube to further amplify the weak acoustic signal. It works as a one-dimensional resonator to form a standing wave along its length [36,37], and the acoustic wave can thus be enhanced at the antinode of the standing wave. The resonator operates as an orifice ended tube with a short gap of 50 μm between the QTF and the resonator.

3. Sensor Configuration and Experimental Setup

3.1. Fiber-Coupled Spectrophone

We design a spectrophone with an external size of 30 mm × 20 mm × 10 mm as depicted in Figure 1. Different from the on-beam resonators widely used in QEPAS [38,39], a half on-beam configuration [40] is used to amplify the acoustic wave and to minimize the internal size to 12 mm × 6 mm × 4 mm. Without other free-space optics for collimation and convergence, the fiber pigtail directly guides laser to the spectrophone. A groove with a diameter of 1 mm is etched for coaxial installation of the fiber and the micro-resonator tube. Vertical distance from the top surface of QTF prongs to resonator’s axis is adjusted to maximize the QEPAS signal by referring to the experimentally optimized result [41]. A through hole with a 2 mm diameter holds the QTF. To suppress potential airflow disturbance on the acoustic detection, the gas inlet is designed at the corner with a buffer (2 mm × 6 mm × 4 mm) etched. The total gas consumption is evaluated to be approximately 300 μL, which is at least 100-fold less than most direct absorption-based gas cells.

3.2. Experimental Setup

The QEPAS sensor with the fiber-coupled spectrophone is illustrated in Figure 2. A distributed feedback (DFB) diode laser with an emission at 1653.73 nm, as the excitation source, targets the absorption of CH₄. The laser wavelength is controlled by a commercial laser driver (LDC 501, Stanford Research System, Montana, USA) with its temperature fixed at 20℃. A LabVIEW program simultaneously scans and modulates the laser wavelength by superposing a sinusoidal modulation on a ramp current from 62.5 to 87.5 mA. To perform the second harmonic detection, modulation frequency is chosen at half of the QTF resonant frequency, which is calibrated to be f₀ = 32.716 kHz. The 0.2 Hz ramp scanning signal leads to a response time of 5 s. Acoustic wave-induced piezoelectric current is converted into voltage signal by a transimpedance amplifier, which is then digitized for demodulation by a software-based lock-in amplifier.

4. Experimental Results and Discussion

4.1. System Parameter Optimization

For the half on-beam QEPAS configuration, the micro-resonator generates enhanced acoustic signal by constructive interference. To demonstrate the performance of the half on-beam configuration, we compared the QEPAS signal of 5000-ppmv CH₄ with a base QTF to that with a micro-resonator. As shown in Figure 3a, the magnification contributed by the micro-resonator (length: 4.7 mm) is evaluated to be approximately 8 fold. When the modulation frequency is fixed at f₀/2, the enhancement performance would depend on the micro-resonator size, especially its length, which will affect how much the signal can be amplified. Figure 3b plots the normalized QEPAS signals with micro-resonators at varying lengths for a selected diameter of 0.6 mm.

Different from the laser beam in common acoustic detection modules, the laser emitted from the fiber has a non-negligible divergence angle. In the case of a long distance between the fiber end and the resonator, the divergent laser could hit on the tube wall or be partially blocked by the QTF. While in the case of a too short or negative distance, less gas molecules would be illuminated by the laser at the entrance of the resonator. The fiber end location from the acoustic resonator can hence affect laser-gas interaction efficiency, then the QEPAS signal.

When the laser passes through the resonator tube, the sound waves generated by the interaction between the laser and the gas oscillate inside the resonator tube. The higher the intensity of the sound wave amplified by the resonator tube is, the stronger the photoacoustic signal will be. An experiment to optimize the fiber location was carried out. The fiber was moved by a 1D precision stage to obtain accurate position information. As the results shown in Figure 4, the signal amplitude presents the strongest value at a distance of 75 μm.

The QEPAS signal also relates to the modulation depth. With a small modulation depth, it is difficult for the sinusoidal modulation to cover the gas absorption line width, generating a small second harmonic signal from a Lorentz absorption profile. Similarly, a degraded signal amplitude can also be inferred from an excessive modulation [42]. We experimentally obtained the response of the photoacoustic signal to the modulation depth, which was adjusted by varying the modulation current applied on the laser source. A calibrated 5000 ppmv CH₄ was sealed inside the acoustic detection module. As shown in Figure 5, the amplitude of QEPAS signal increases at the onset of the curve and then decreases slowly. According to the experimental results, the optimized modulation current of 12.5 mA was chosen in the following experiments.

4.2. Sensor Performance

We measured the photoacoustic signal under different CH₄ concentrations to acquire the linear relationship between the signal amplitude and gas concentration. The CH₄ samples ranging from 1000 ppmv to 10% were generated by diluting the calibrated pure CH₄ with pure N₂. A continuous measurement of each sample was performed over 10 min and the result is shown in Figure 6. The R-square of the fitted curve can reach 0.9989, which proves the good linearity of this sensor.

As the laser is not collimated or shaped by the free-space optics, its influence on the system noise should be investigated. The laser diameter from the fiber end is approximately 10 μm, while the gap between QTF prongs is approximately 300 μm [43]. Considering the fiber divergence angle of approximately 8° and the resonator length of 4.7 mm, the spot diameter is approximately 400 μm when the uncollimated laser passes through the QTF. Thus, the laser will inevitably touch the micro-resonator during its propagation. Further, part of laser also hits the QTF prongs. Periodic laser intensity variations during the turning could lead to thermal deformation, which therefore generates the thermoelastic noise [44,45]. Thanks to the large irradiated area, corresponding to a low power intensity, the noise is much smaller than that of imperfectly shaped free-space laser. To numerically evaluate the effect of fiber coupling on system noise, we conducted a comparative experiment while the cell was filled with pure N₂. The noise with modulated incident laser and without laser was successively measured. As the experimental data shown in Figure 7, within 1500 s, the averaged signal amplitude with incident laser is Offset₁ = 3.6 μV, while the averaged amplitude without laser is Offset₂ = 2.5 μV. The standard deviations are σ₁ = 0.6 and σ₂ = 0.7 μV, respectively. Experimental results show that fiber-coupled operation has a little impact on the amplitude of the system noise.

To evaluate the stability and minimum detection limit of the system in the long-term detection, an Allan–Werle deviation analysis was performed with the results shown in Figure 8. The measurement was conducted for 5 h at 760 Torr with pure N₂. The detection limit is determined to be 16.7 ppmv at an averaging time of 5 s. The ultimate detection limit can be improved to 1.1 ppmv at a much longer averaging time of 1245 s, within which the white noise dominates, demonstrating a very stable performance for the fiber-coupled spectrophone. Taking the laser power into consideration, we calculate the normalized noise equivalent absorption (NNEA) coefficient as 7.75 × 10⁻⁸ W·cm⁻¹·Hz^−1/2, which is comparable to the traditional QEPAS technology.

4.3. In situ Detection of Deep-Sea Dissolved CH₄

With the fiber-coupled sensor calibrated, we performed an in situ measurement of dissolved CH₄ at the Haima Cold Spring Area in the South China sea. In this measurement, our sensor was integrated into a metal cavity that can protect the inside instruments from a seawater pressure as high as 35 MPa. The acoustic detection module and the dissolved gas separation unit with a PDMS membrane were connected by a short 1/8 stainless steel pipe. The gas separated from the PDMS membrane diffused into the photoacoustic cell for measurement with the result shown in Figure 9. The instrument was placed by the submersible near the cold spring vent and then began to operate after 1 h. With a low separating rate of 30 μL/min, the measured dissolved CH₄ concentration reached a plateau level of approximately 6% after 90 min. Compared with a recently reported demonstration of near-coast CO₂ using cavity ring-down spectroscopy (CRDS) technique, the balancing time has been shortened by more than 7 fold [46]. This measurement illustrates that the dissolved CH₄ concentration near the vent of the cold spring is much higher than that of the background seawater, which is usually approximately 10–20 ppmv [47].

5. Conclusions

In conclusion, we have reported a sub-mL QEPAS sensor with a compact spectrophone for the detection of deep-sea dissolved gases. It has a gas consumption of approximately 300 μL, at least two orders of magnitude smaller than traditional optical absorption-based gas chambers, which is important to reduce the balance time for in situ dissolved gas detection. Instead of free-space laser alignment, the use of fiber coupling without any other optics offers a simple configuration for assembly. The detection limit and linearity have been experimentally evaluated, achieving a minimum detection limit of 1.1 ppmv for CH₄ measurement and a good linear response with an R-square value of 0.9989. The normalized noise equivalent absorption (NNEA) coefficient is calculated to be 7.75 × 10⁻⁸ W·cm⁻¹·Hz^−1/2. An in situ dissolved gas measurement was carried out at the Haima Cold Spring Area to verify the practicability of this sensor.