Quartz-enhanced photoacoustic spectroscopic methane sensor system using a quartz tuning fork-embedded, double-pass and off-beam configuration.

Development of a methane (CH4) sensor system was reported based on a novel quartz-tuning-fork (QTF)-embedded, double-pass, off-beam quartz-enhanced photoacoustic spectroscopy (DP-OB-QEPAS). A simplified and accurate numerical model was presented to optimize the DP-OB-QEPAS spectrophone and to enhance the detection sensitivity. A compact and fiber-coupled acoustic detection module (ADM) with a volume of 3 × 2×1 cm3 and a weight of 9.7 g was fabricated. A continuous-wave distributed feedback diode laser was used to target the CH4 absorption line at 6046.95 cm-1. With the combination of wavelength modulation spectroscopy (WMS) and second harmonic (2f) detection technique, the CH4 sensor system reveals a 1σ detection limit of 8.62 parts-per-million in volume (ppmv) for a 0.3 s averaging time with an optimized modulation depth of 0.26 cm-1. The proposed CH4 sensor shows a similar or even lower level in the normalized noise equivalent absorption coefficient (NNEA) (1.8 × 10-8 cm-1∙W/√Hz), compared to previously reported QEPAS-based CH4 sensors.

Introduction

Methane (CH4) detection plays an important role in many applications such as environmental monitoring in urban and rural areas, safety in coal mines, natural gas leak detection, and medical diagnostics [1]. Therefore, highly sensitive and stable detection of CH4 is urgently needed to meet requirements in these fields. CH4 detection usually employs chemical processes and optical spectroscopy. In comparison with chemical processes, optical spectroscopy is advantageous for CH4 sensing in terms of size, time resolution and cost and requires no pretreatment and/or accumulation of the concentration of the targeted gas samples [2]. Photoacoustic spectroscopy (PAS) is a widely used spectroscopic technique for trace gas detection due to high detection sensitivity and selectivity and compactness of the detection module [3]. Quartz-enhanced photoacoustic spectroscopy (QEPAS) technique, a significant innovation of photoacoustic spectroscopy, utilizes a quartz tuning fork (QTF) as the acoustic transducer, which has a small size, low cost and excellent immunity to ambient acoustic noise, is well suited for sensitive measurements and practical applications in trace gas sensing [4].

Acoustic micro-resonators (AmRs) are usually used in on-beam [[5], [6], [7]] or off-beam [[8], [9], [10]] configurations to enhance the sensitivity of the QEPAS-based sensors. The on-beam structure can realize a high acoustic coupling strength and detection sensitivity, but the excitation light beam should pass through a small gap (≈ 300 μm for a commercial QTF) between the QTF prongs, which constrained the inner diameter of the resonant tube and the beam size that passed through the tube, in order to avoid the thermal noise caused by the laser beam irradiating the QTF prongs [8]. To simplify the assembly and alignment process in QEPAS, off-beam configuration was proposed in 2009 [8]. The reported off-beam configuration without the need of the laser beam passing through the QTF prongs makes it easy to use excitation sources with a low spatial radiation quality for QEPAS. However, the acoustic coupling strength between the resonant tube and the QTF in off-beam configuration (such as common off-beam configuration [11] and T-tube off-beam configuration [9] proposed before), which can be evaluated by quality factor (Q-factor) and directly related to the response time τs of the QEPAS spectrophone (τs = Q/πf0) with f0 representing the resonant frequency, was much weaker than in common on-beam configuration. Moreover, the T-tube processing is more complicated than the use of commercial capillary.

In this work, a near-infrared highly-sensitive QEPAS CH4 sensor system was developed based on a novel QTF-embedded, double-pass, off-beam quartz-enhanced photoacoustic spectroscopy (DP-OB-QEPAS) technique, which realizes a high acoustic coupling strength and detection sensitivity in off-beam configuration (easier to assembly and alignment than the on-beam configuration), simultaneously [12]. At the same time, the processing is simpler than that used in the T-tube off-beam configuration. Off-beam QEPAS was realized by embedding the QTF partially inside the resonant tube, and the acoustic coupling was enhanced by a double-pass configuration by embedding the QTF in two resonant tubes and reflecting the light with a right-angle prism. A simplified numerical model was proposed to optimize the resonant tube length and inner diameter for enhancing the detection sensitivity. The simulation results are consistent with the previous experimental optimized results. A compact fiber-coupled acoustic detection module (ADM) with a volume of 3 × 2×1 cm3 and a total weight of 9.7 g was fabricated for photoacoustic detection. The performances of the developed sensor system were evaluated by the measurements on the prepared CH4 samples with a gas mixing system and a similar or higher detection sensitivity with previously reported QEPAS-based CH4 sensors was achieved.

DP-OB-QEPAS spectrophone and ADM

DP-OB-QEPAS spectrophone

In off-beam QEPAS, the acoustic oscillations of gas are usually exited in the acoustic micro-resonator (AmR) by a modulated laser source and the photoacoustic signal amplitude in the AmR can be written aswhere C is the cell constant that describes the characteristic of the AmR at a specified frequency f, and ω = 2πf [11]; the absorption coefficient α characterizes the gas absorption as the laser beam passes through a gas sample, which is related to gas concentration and absorption cross section; W is the optical power of the excitation source. Generally, the laser source is modulated at half of the resonant frequency f0 of the QTF. With wavelength modulation spectroscopy (WMS), a photoacoustic signal with a frequency of f0 will be generated via photoacoustic effect and enhanced in the AmR. The acoustic oscillations in the AmR give rise to a sound wave radiated via a slit placed in its center and detected by the QTF. The QEPAS signal (S) generated by the QTF is proportional to the amplitude of the photoacoustic signal A and the Q-factor of the QTF. Therefore, S can be described aswhere, k is a parameter. The system noise N should also be considered for an accurate assessment of the QEPAS-based sensor performance. It is theoretically predicted and verified through experiment that the QEPAS system noise is primarily determined by the thermal noise of the QTF [13]. The QTF thermal noise can be expressed in terms of root mean square (RMS), aswhere, VN is the voltage noise at the transimpedance amplifier output, Rg is the feedback resistor (≈ 10 MΩ), kB is the Boltzmann constant, T is the QTF temperature, R is the equivalent resistance of the QEPAS spectrophone, and is the detection bandwidth [14]. R represents QTF oscillator loss and is related to other QTF parameters as , where L is an equivalent inductance representing the QTF oscillator equivalent mass, and C represents the inverse force constant [13]. Both L and C do not change more than 0.1 % due to the interaction with gas molecule [13]. Therefore, Eq. (3) can be re-written as

According to Eqs. (2) and (4), the signal to noise ratio (SNR) of the QEPAS sensor system can be written as

SNR can be used to evaluate the detection sensitivity of the QEPAS sensor system at a specific environment temperature, gas concentration, laser power and detection bandwidth. Multiple parameters, including the length and inner diameter of the resonant tube, the gap between the tube and QTF, gas pressure, the dimension of the QTF, could play important roles on the acoustic coupling and consequently the QEPAS signal and noise. The relationship among the parameters cannot be easily obtained [15].

However, a simplified theoretical model, which only considers the resonance condition of the AmR at a specific frequency (the resonant frequency f0 of the QTF), has been successfully applied to various QEPAS configurations (on beam [16], off-beam [17], and T-tube off-beam [9,18]), and the theoretical predictions are in agreement with the experimental results. The models provide a simplified way to optimize the AmR used in QEPAS, which is based on the facts that: (1) the displacement of the QTF prongs is very small (pm or nm-level) [[19], [20], [21]], and thus the sound field produced by the QTF has little effect on the sound field inside the resonant tube; (2) the resonant frequency of the QTF has little change with or without the resonant tube, so the spectrophone parameters can be optimized at the resonant frequency of the bare QTF; (3) the piezoelectric current generated by the QTF is proportional to the pressure acting on its surface at a specified frequency [[21], [22], [23]], so the relative magnitude of the piezoelectric signal can be expressed by the acoustic pressure acting on the surface of the QTF prongs.

Compared with theoretical model, a numerical model is simpler and more flexible to optimize AmRs’ parameters [15,24,25]. However, the calculation efficiency of the developed numerical model is limited by complex interactions among optical, sound, and electrical fields when multiple physical field modules were employed. In this work, we propose a simplified numerical model to optimize the resonant tube length and inner diameter of the DP-OB-QEPAS spectrophone based on the simplified theoretical model proposed in Refs. [9,[16], [17], [18]]. With this simplification, only the pressure acoustic module was employed in the numerical model, which is helpful to improve the calculation efficiency.

The structure of the DP-OB-QEPAS spectrophone is shown in Fig. 1. The design parameters are in agreement with those in Ref. [12]. The outer diameter of the resonant tube is 1.2 mm. A side slit was made in the middle of the resonant tube. The slit depth d and width w are 0.6 mm and 0.4 mm, respectively. The vertical distance between the opening of the QTF and the axis of the resonant tube is h (≈ 1 mm) [26].

Fig. 1

Schematic of the QTF-embedded double-pass photoacoustic spectrophone, (a) 3D side view and (b) top view.

As shown in Fig. 2, the numerical model was established according to the schematic of the QTF-embedded DP-OB-QEPAS spectrophone shown in Fig. 1. Considering that the parameters of the two resonant tubes are identical, only one resonant tube is fabricated in the simplified numerical model. The QTF was treated as two slabs with a dimension of 0.47 mm × 0.25 mm × 3.3 mm and a bottom support. The dimension was obtained by measuring the bare QTF used in our experiment using a high-definition scientific digital microscope (Supereyes, Model A005). The acoustic pressure wave was regarded to be originated from a line source along the axis of the tube and propagate all around, as shown in Fig. 2. Considering that only relative pressure is needed to evaluate the optimal length of the resonant tube, the source term is set to a unit strength. The boundary of gas involved is assumed to be of a spherical shape and an outer spherical shell is introduced as the perfectly matched layer (PML) to absorb the reflected acoustic wave from the boundary [15]. The sound velocity in the air is set to 343.5 m/s based on an assumed environment temperature of 293 K [16]. The resonant tube and the QTF are considered to be rigid and not active in pressure acoustic module to simplify the numerical model.

Fig. 2

The simplified numerical model for QTF-embedded, DP-OB-QEPAS spectrophone optimization (unit: m). PML: perfectly matched layer.

The normalized sound pressure at point A (shown in Fig. 2) as a function of the tube length L was recorded by using frequency domain analysis in COMSOL 5.3 for different inner diameters. The simulation frequency is 32.75 kHz, which is the resonant frequency f0,b of the bare QTF. As shown in Fig. 3(a), a smaller inner diameter is preferred for the accumulation of acoustic signal and enlarging acoustic pressure signal when resonance occurs. The larger the sound pressure, the larger piezoelectric signal is generated. Therefore, smaller inner diameter can improve sensitivity. However, too small inner diameter will increase the difficulty in laser beam collimation. The optimal tube length as a function of inner is shown in Fig. 3(b). For comparison, the optimal result obtained by experiment is also given. And the experimental and theoretical results are highly consistent. In addition, the linear fit shows that the optimal tube length decreases linearly with the increase of inner diameter. This phenomenon is due to the end correction, which is also consistent with the theoretical model predictions [9,[16], [17], [18]].

Fig. 3

Numerical simulation results of the QTF-embedded, DP-OB-QEPAS spectrophone. (a) The normalized pressure at point A as a function of the tube length L for different inner diameters. (b) The optimal tube lengths obtained by experiment and numerical calculation for different inner diameters.

We also investigated the acoustic pressure distribution inside the resonant tube with an optimized inner diameter of 0.6 mm and an optimal length of 8.6 mm. The simulation results are shown in Fig. 4. The sound pressure decreases at the center of the resonant tube, which results from the sound leakage through the side slit in the middle of the resonant tube. The normalized acoustic pressure distribution inside the resonant tube is similar to that of the common on-beam QEPAS configuration [13]. The optimized resonant tube parameters (inner diameter = 0.6 mm, length = 8.6 mm) was employed in the developed DP-OB-QEPAS sensor for CH4 detection.

Fig. 4

The normalized acoustic pressure distribution inside the resonant tube. (a) 3D side view and (b) top view. (c) The normalized acoustic pressure along the axis of the resonant tube (unit: mm).

ADM

The ADM, which consists of a spectrophone enclosed by a gas enclosure, is considered to be the core part in a QEPAS gas sensor system [27]. The 3D laser printing technique was employed to realize a compact, cost-effective DP-OB-QEPAS ADM. A QEPAS spectrophone holder was used to fix the two resonant tubes and a commercial QTF in DP-OB-QEPAS. The two resonant tubes are machined by wire cutting with a processing accuracy of 1 μm. A schematic 3D view of the DP-OB-QEPAS spectrophone with the 3D-printed spectrophone holder is given in Fig. 5(a). The design parameters of the DP-OB-QEPAS spectrophone was identical with the optimal results in Section 2.1 (tube length = 8.6 mm, inner diameter = 0.6 mm). The QEPAS spectrophone holder, a right-angle prism and a fiber-coupled focuser (OZ Optics Ltd., Model LPF-05) were mounted on a 3D printed bottom cover containing a gas inlet and a gas outlet with high-strength vacuum epoxy. The focusing diameter of the output laser beam can be decreased to 150−312 μm within a propagation length of 30 mm. A 3D-printed top cover was fabricated to seal the gas enclosure with UV-curing glue. A compact, cost-effective and fiber-coupled DP-OB-QEPAS ADM with a size of 3 × 2×1 cm3 in volume and a total weight of 9.7 g was finally obtained for photoacoustic detection. A schematic 3D view and a photograph of the DP-OB-QEPAS ADM are shown in Fig. 5(b) and (c), respectively.

Fig. 5

(a) Schematic 3D view of the DP-OB-QEPAS spectrophone with a 3D-printed spectrophone holder. (b) Schematic 3D view of the DP-OB-QEPAS ADM. (c) Photograph of the fabricated ADM for gas sensing. 1, pins to be connected with the trans-impedance amplifier; 2, gas inlet; 3, gas outlet; 4, top cover; 5, right-angle prism; 6, resonant tube; 7, spectrophone holder; 8, fiber-coupled focuser; 9, bottom cover.

We evaluated three important parameters (resonant frequency f0, quality factor Q and equivalent resistance R) of the ADM by an electric excitation method [28]. In QEPAS, an ADM are usually used under resonance condition. The results are listed in Table 1, where the response time of the QEPAS spectrophone τs = Q/πf0, and the equivalent resistance R was used to calculate the QTF theoretical thermal noise in Section 4. A strong acoustic coupling leads to a decrease in the Q-factor because the high-Q QTF transfers energy primarily via the coupling to the low-Q AmR oscillator (the resonant tube) [26]. As can be seen in Table 1, the designed DP-OB-QEPAS spectrophone realizes a similar acoustic coupling with the common on-beam QEPAS spectrophone and a stronger acoustic coupling than any other off-beam QEPAS spectrophone proposed before. Although the response time of the QEPAS spectrophone cannot be used to characterize the response time of the sensor system since many factors will affect the response time of the whole sensor system, such as the integration time of the lock-in amplifier, the gas exchange time of the chamber. The improved acoustic coupling shortens the response time of the QEPAS spectrophone and is helpful to realize a rapid spectral measurement in off-beam QEPAS with the effects of other factors are eliminated experimentally.

Table 1

Comparison among the DP-OB-QEPAS spectrophone and the previously reported spectrophone configurations.

Spectrophone configuration	Resonant frequency f0 (Hz)	Q	R (kΩ)	τs (ms)
DP-OB-QEPAS [this work]	32,755	3127	482.80	30
T-tube off-beam QEPAS [9]	32,740	5457	–	53
Common off-beam QEPAS [11]	32,753	8000	–	78
Common on-beam QEPAS [13]	32,735	3630	–	35
Common on-beam QEPAS [28]	32,761	1958	–	19

Structure and design of the sensor system

CH4 absorption line selection

Based on the high-resolution transmission (HITRAN) absorption database 2012, the CH4 absorption spectra is shown in Fig. 6(a). There are four fundamental absorption bands in the mid-infrared spectral range at v1 = 2913 cm−1, v2 = 1533.3 cm−1, v3 = 3018.9 cm−1, v4 = 1305.9 cm−1 and two overtone-rotational vibrational combination bands in the near-infrared range at 1.65 μm (2v3) and 1.33 μm (v2 + 2v3). The absorption line intensity for CH4 in the mid-infrared region is much stronger than those in the near-infrared region. However, near-infrared continuous-wave distributed feedback (CW-DFB) diode laser has a lower cost than the mid-infrared laser [29]. Meanwhile, the 2v3 harmonic band at 1.65 μm is the most suitable band for CH4 detection free from the interference of other common gas species in the near-infrared region, such as water vapor and carbon dioxide [30]. Therefore, a near-infrared CW DFB diode laser with an emission wavelength around 1.65 μm was employed as the excitation source. The absorption spectra for 1000 ppmv CH4 at an environmental temperature of 293 K and 1 atm pressure with a path length of 1 cm is shown in Fig. 6(b). The R-branch R(3) of the 2v3-band of CH4 located at 6046.95 cm−1 was used to perform the experiments.

Fig. 6

(a) Absorption line of CH4 in the near-infrared and mid-infrared region. (b) The absorbance for 1000 ppmv CH4 at an environmental temperature of 293 K, 1 atm pressure and a path length of 1 cm.

Laser characterization

The tuning performances of the CW-DFB diode laser (wavenumber and output optical power) were characterized using a Fourier transform spectrometer (Thermo Fisher Scientific, Model Nicolet iS50) and an optical power meter (Ophir Photonics, Model Nova II). Fig. 7(a) and (b) depict the measured laser wavelength and output optical power as a function of injection current at different operating temperatures ranging from 16 °C to 18 °C. The calculated temperature and current tuning coefficients of the diode laser were −0.407 cm−1/°C and −0.073 cm−1/mA, respectively. The diode laser was tuning to target the CH4 absorption line at 1.65 μm (6046.95 cm−1), which is shown in Figs. 6(b) and 7 (a), with an operating temperature of 17 °C and an injection current of 68 mA. Under this case, the output optical power is 3.74 mW as depicted in Fig. 7(b).

Fig. 7

Wavelength tuning performance of the 1.65 μm CW-DFB diode laser. (a) Emission wavenumber as a function of the injection current. (b) Output optical power at different operating temperature and injection current.

Sensor configuration

The structure of the CH4 sensor system is shown in Fig. 8. A 1.65 μm near-infrared CW-DFB diode laser was used as the excitation source. The output fiber of the diode laser was connected to the fiber-coupled focuser by a pair of FC/APC connectors with a 0.25 dB loss. The CW-DFB diode laser was controlled by a commercial temperature controller (Thorlabs, Model TED200C) and a laser current driver (Thorlabs, Model LDC202C). Modulation of the laser current was realized by applying a sinusoidal at half of the ADM resonant frequency dither to a triangular wave generated by an analog adder and a function generator (Good Will Instrument, Model AFG-2225). The piezoelectric signal of the QTF was converted to voltage by a trans-impedance amplifier with a feedback resistor of 10 MΩ. Subsequently, the voltage signal was demodulated by a lock-in amplifier (Stanford Research System, Model SR830) for the measurement of the 2f component. The time constant of the lock-in amplifier and the slope filter were set to 0.3 s and 18 dB/octave, respectively, leading to a detection bandwidth of Δf =0.44 Hz. A data acquirement card (DAQ, National Instrument, Model USB-6211) was used to acquire the orthogonal components (X and Y) from the lock-in amplifier. The 2f-signal S2 was calculated by a LabVIEW program to eliminate the influence of signal phase on the sensor performance [(S2 = ((X + offset)2 + (Y + offset)2)1/2 - baseline)], where offset is a constant for obtaining a normal 2f signal. CH4 samples were prepared by a commercial gas mixing system (Environics, Model S4000). The gas flow rate of the gas mixing system was limited to 500 standard-state cubic centimeters per minute (sccm) to reduce airflow noise.

Fig. 8

Schematic of the DP-OB-QEPAS CH4 sensor system by utilizing a compact fiber-coupled ADM and a near-infrared CW-DFB diode laser. CD, laser current driver; TC, laser temperature controller; DAQ, data acquisition card.

Modulation depth optimization

The modulation depth was optimized to maximize the 2f-signal amplitude of the DP-OB-QEPAS sensor system. The normalized 2f-signal amplitude as a function of the modulation depth is shown in Fig. 9. Initially, the DP-OB-QEPAS normalized 2f-signal amplitude increases with the modulation depth. However, no further significant change was observed when the modulation depth is > 0.28 cm−1. The maximum normalized 2f-signal amplitude was obtained at a modulation depth of 0.26 cm−1, leading to a modulation index (defined as the ratio between the modulation depth and the half width at the half maximum (HWHM, ≈ 0.076 cm−1)) of 3.42. Therefore, the optimized modulation depth (0.26 cm−1) was employed in the developed DP-OB-QEPAS CH4 sensor.

Fig. 9

Normalized 2f-signal amplitude and modulation index as a function of modulation depth for a 10,000 ppmv CH4:N2 sample, where the modulation frequency is f0/2 = 16.38 kHz.

Measurements and performances

NNEA measurements

The measured DP-OB-QEPAS 2f-signal with and without the right-angle prism are shown in Fig. 10(a) at a CH4 concentration level of 10,000 ppmv. A 1.8 times enhancement of QEPAS signal was realized by use of the right-angle prism to realize a double-pass structure. The background noise was measured by flushing the gas cell with ultra-highly pure N2 to ensure no residual CH4 in the ADM. As can be shown in Fig. 10(b), the standard deviation of the background noise was 1.716 μV, which is primarily determined by the fundamental thermal noise of the QTF. With an environmental temperature of 293 K, the theoretical noise is determined to be 1.215 μV according to Eq. (3), which is in basically agreement with the experimental value of 1.716 μV. Therefore, the thermal noise caused by the laser beam irradiating the QTF prongs was avoided due to an off-beam configuration was employed and precise optical path design. The 2f-signal amplitude was 2.05 mV with a background noise of 1.716 μV, leading to a signal-to-noise ratio (SNR) of 1195. Therefore, the 1σ minimum detection limit (MDL) of the CH4 sensor system is 8.36 ppmv for a 0.3 s integration time, leading to a normalized noise equivalent absorption coefficient (NNEA) of 1.804 × 10−8 cm−1∙W/√Hz.

Fig. 10

(a) The measured DP-OB-QEPAS 2f-signal waveform with and without the right-angle prism. (b) The measured 2f-signal waveform by flushing the gas cell with pure N2.

Calibration

To calibrate the DP-OB-QEPAS CH4 sensor system, CH4 samples with different concentration levels diluted in N2 were prepared using the commercial gas dilution system. For accuracy, CH4 samples with low concentration levels (< = 1000 ppmv) were prepared using a certified 1000 ppmv CH4:N2 mixture, and the samples with high concentration levels were generated using a 10,000 ppmv CH4:N2 mixture. The measured 2f-signal waveforms as a function of CH4 concentration ranging from 0 ppmv to 10,000 ppmv are plotted in Fig. 11(a). The relation curve of the 2f signal amplitude versus the CH4 concentration and the corresponding fitting curve are shown in Fig. 11(b). A R-square value of 0.999 implies that the sensor system exhibits a good linear response between the 2f signal amplitude and the monitored CH4 concentration levels.

Fig. 11

(a) The measured 2f-signal waveforms for nine CH4 concentration levels of 0, 250, 500, 750, 1000, 2500, 5000, 7500, and 10,000 ppmv, where the balance gas was N2. (b) Experimental data dots and fitting curve of 2f-signal amplitude as a function of CH4 concentration.

Allan deviation

The DP-OB-QEPAS sensor was further investigated by an Allan deviation analysis, which was performed by the measurement of the QEPAS signal in pure N2 (0 ppmv CH4) at a 1 atm pressure and a temperature of 293 K. The QEPAS signal was converted into CH4 concentration by use of the calibration curve shown in Fig. 11(b). The measured CH4 concentrations as a function of observation time are shown in Fig. 12(a). A 1σ detection limit of 8.62 ppmv with a 0.3 s averaging time was achieved, which agrees well with the result of 8.36 ppmv determined by SNR in Section 4.1. The Allan deviation shown in Fig. 12(b) indicates that the thermal noise of the QTF is the dominant noise [36] and the DP-OB-QEPAS-based sensor allows data averaging without a base line or sensitivity drift for a time scale of 50 s.

Fig. 12

(a) The measured CH4 concentration by injecting pure N2 into the ADM. (b) Allan deviation plot as a function of averaging time for the sensor operated at a standard atmospheric pressure and environment temperature.

Comparison and discussion

The detection performance and the difficulty of assembly and alignment process are two key criterions for the designed QEPAS spectrophone. The common off-beam QEPAS configuration-based CH4 sensors were rarely reported due to a much weak acoustic coupling and low detection performance. As a comparison, the performances of several reported QEPAS CH4 sensors based on the on-beam and T-tube off-beam QEPAS spectrophones are listed in Table 2. To be fair, the 1σ limits of detection (LoD) of different CH4 sensors are given with an averaging time of 1 s. As can be seen in Table 2, the normalized noise equivalent absorbance (NNEA), which is independent of absorption line strength, optical power and the detection bandwidth, is a more important parameter for performance evaluation [34]. The proposed CH4 sensor reveals a similar NNEA level with the T-tube QEPAS CH4 sensor and a lower NNEA level with the common on-beam QEPAS CH4 sensors proposed before. Therefore, by using a novel DP-OB-QEPAS spectrophone, this sensor configuration possesses a similar or even better detection performance with the reported common on-beam QEPAS [[31], [32], [33],35] and the T-tube off-beam QEPAS configurations.

Table 2

Several reported QEPAS CH4 gas sensors based on different spectrophone configurations. * not mentioned. # Calculation results according to the data in the reference paper.

Spectrophone configuration	Target line (cm−1)	Carrier gas	Laser Power (mW)	LoD (1σ)	NNEA (cm−1∙W/√Hz)
T-tube off-beam [18]	2958.02	dry N2	0.5	500 ppbv	1.1 × 10−8
on-beam [31,32]	1275.04	dry N2	150	13 ppbv	3.0 × 10−8 #
on-beam [33]	2988.8	dry N2	11	90 ppbv	5.0 × 10−8 #
on-beam [35]	6057.1	dry N2	15.8	-*	2.9 × 10−8
QTF-embedded double-pass off-beam [this work]	6046.95	dry N2	3.74	4.63 ppmv	1.8 × 10−8

In addition, the difficulty of assembly and alignment process is also an important criterion to evaluate the proposed DP-OB-QEPAS spectrophone. Generally, the spectrophone realizes an easier assembly and alignment process than the common on-beam configuration. (1) Using the same resonant tube with an inner diameter of 600 μm, in the common on-beam and DP-OB-QEPAS spectrophones, the laser beam should pass a ≈ 300 μm gap (the gap between the QTF prongs) and a ≈ 450 μm gap (the distance between the walls of the resonant tube and the QTF prong), respectively. Therefore, it is a bit more difficult in aligning the laser beam to pass a smaller gap in common on-beam configuration. (2) The two tubes and the QTF should be seriously on-beam in common on-beam approach, which requires more steps in the assembly process. However, the two tubes used in the DP-OB-QEPAS spectrophone are independent and thus can be adjusted independently. (3) The background noise of the QEPAS spectrophone is primarily determined by the fundamental thermal noise of the QTF. A high noise level will be generated by the laser beam irradiating the QTF prongs, but the laser beam irradiating the wall of the resonant tube will not obviously degrade the detection performance. The DP-OB-QEPAS spectrophone reduces the possibility of the beam radiation to the QTF surface and further simplify the alignment process. (4) In the common off-beam configuration, the laser beam should pass a 600 μm gap (the inner diameter of the employed resonant tube), which is larger than that in DP-OB-QEPAS. Therefore, the alignment and assembly process of the DP-OB-QEPAS spectrophone seems to be a bit more complex. However, the laser beam diameter can be decreased to 150−312 μm within a propagation length of 30 mm using a fiber focuser, which makes it easy for the alignment of the dual-tube with a total light propagation length of ≈ 20 mm (2×tube length + light forward/backward length). (5) Compared to T-tube off-beam QEPAS spectrophone, the processing of commercial capillary is simpler than that of a T-tube, which should be customized to realize a high detection sensitivity.

To summarize, the proposed DP-OB-QEPAS spectrophone realizes a high detection performance and an easy assembly and alignment process in QEPAS, simultaneously.

Consideration for field application

For field application of the developed DP-OB-QEPAS CH4 sensor system, some possible interfering factors must be considered. The change of pressure, temperature and background gases will affect the accuracy of the sensor. When the pressure and temperature change, the line strength and width of the selected CH4 absorption line will change. The change of absorption line intensity leads to the change of gas absorption coefficient α, which will directly affect the QEPAS signal S as shown in Eq. (2). The change of absorption linewidth will lead to the failure of optimized modulation depth in Section 3.4, which will also affect the detection performance of the system [2,37]. Therefore, it is suggested to control the temperature and pressure of the DP-OB-QEPAS ADM in the detection environment with obvious change of ambient temperature and pressure. Alternatively, both pressure and temperature compensation can also be employed in the sensor. But when both of them act at the same time, the effects on the sensor are complex and a large number of experimental investigations must be carried out to obtain the actual concentration in time.

N2 is usually used as carrier gas (or background gas) to evaluate the performance of the developed sensor in the laboratory. The interference introduced by optical absorption of the background gas in field application can be ignored, since the emission wavelength of the DFB laser was tuned precisely to a specific absorption line of CH4 that is free of interfering absorption from other gases (typically e.g. H2O, CO2, see Ref. [37]). This assures the selectivity of the sensor. However, the change of the background gases may change the sound velocity and violate the resonance condition of the optimal resonant tube parameters at ≈ 32 kHz. In a majority of application environments, the sound velocity does not change significantly, so the sensor can still achieve accurate concentration measurements. However, the resonant tube parameters should be re-optimized to realize a high detection sensitivity and the sensor should be recalibrated in special background gas, such as hydrogen (H2) [35]. Moreover, when exploiting a standard ≈ 32 kHz QTF, the modulation frequency of ≈ 16 kHz is comparable to or exceeds the vibrational to translational (V–T) energy transfer rate in slow-relaxing gas species such as CH4, thereby violating the condition for photoacoustic spectroscopy that the molecular relaxation rate should be much higher than the modulation frequency [38,39]. As a result, the observed QEPAS signal exhibits strong reduced amplitude compared to a conventional photoacoustic spectroscopy sensor. The H2O molecules act as V-T relaxation process promoter for CH4. Therefore, the V-T energy transfer rate will be reduced and the photoacoustic signal will be improved with the present of H2O. In order to eliminate the influence of ambient humidity on the performance of the sensor, the sample gas should be pre-dried. Another possible method is to obtain the H2O concentration in real time to calibrate the sensor, which has been realized in Ref. [37].

In addition, the QEPAS signal is directly proportional to the laser power as shown in Eq. (2). Therefore, the optical power fluctuation of the diode laser must be monitored and the sensor can be calibrated by the power linearly. The resonant frequency of the ADM must be calibrated periodically to avoid the shift of resonant frequency caused by the change of QTF characteristics [40]. This is because the QTF based ADM generally has a higher Q-factor, and the modulation frequency deviation will directly affect the detection performance. Moreover, the optical collimation structure requires the laser to pass through the resonant tube smoothly without irradiating the QTF surface. Therefore, in the long-term detection for a field application, it is necessary to ensure the stability of the mechanical structure to avoid the introduction of unnecessary laser radiation noise [27,41]. The mentioned interfering factors and corresponding suggestions for the DP-OB-QEPAS sensor system have been summarized in Table 3, which should be considered carefully in the field application.

Table 3

The possible interfering factors and corresponding suggestions for the proposed DP-OB-QEPAS CH4 sensor system in the field application.

Interfering factor	Suggestion
Pressure & temperature change	(1) Control the temperature and pressure in the ADM; (2) Pressure and temperature compensation
Background gas (include humidity)	(1) pre-dried; (2) V-T relaxation self-calibration; (3) re-optimize the spectrophone parameters and re- recalibrate the sensor (if necessary)
Resonant frequency of the ADM	Calibrate the resonant frequency periodically (once a month)
Power fluctuation	Monitor output power and corrected the result
Optical alignment structure	Strengthen mechanical structure

Consideration for reproducibility of the ADM

There are some considerations for assuring the reproducibility of the sensor, whose core module is the ADM. The detection performance of the proposed DP-OB-QEPAS ADM was determined by the resonant tube parameters and the bare QTF parameters. Before assembly into the ADM, these QTFs should be strictly selected to ensure that the resonance frequency is close to 32.75 kHz and the Q-factor is > 10,000 at room temperature and atmospheric pressure. The parameters of the resonant tube can be optimized by the proposed COMSOL finite element model and verified by the experimental results. The optimized results d = 0.6 mm, w = 0.4 mm, outer diameter = 1.2 mm, inner diameter = 0.6 mm, length = 8.6 mm) are suggested to be used to process the resonant tubes. The resonant tubes can be machined by wire cutting with an accuracy of 1 μm, which ensures the consistency of the parameters of the processed resonant tube. Through the fine assembly of the resonant tube, QTF and right-angle prism under the microscope, the achieved ADM will obtain a similar detection sensitivity and acoustic coupling strength.

Conclusions

A near-infrared CH4 sensor system was developed using a novel QTF-embedded, DP-OB-QEPAS technique, which realizes a high acoustic coupling strength and detection sensitivity in off-beam configuration, simultaneously. A simplified numerical model was proposed to optimize the resonant tube length and inner diameter. The model was proved to be accurate through experiment. A compact, cost-effective and fiber-coupled ADM was fabricated using 3D laser printing technique. The sensor performance was evaluated by CH4 detection at a temperature of 293 K and a pressure of 1 atm. By using a novel DP-OB-QEPAS spectrophone, this sensor configuration realizes a similar or even better detection performance with the reported common on-beam QEPAS and the T-tube off-beam QEPAS CH4 sensor. In addition, the assembly and alignment process are simpler than common on-beam configuration and the resonant tube is easier to process than machining T-tube in T-tube off-beam configuration. With high detection sensitivity and compact, low-cost ADM design, this sensor is more suitable for application in civil, industrial and environmental monitoring.

Funding

This work was supported by the National Key R&D Program of China (No. 2017YFB0405300), the National Natural Science Foundation of China (Nos. 61960206004, 61775079, 61627823), the Key Science and Technology R&D program of Jilin Province, China (Nos. 20180201046GX, 20190101016JH).

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.