Multi-mechanism collaboration enhanced photoacoustic analyzer for trace H₂S detection

Abstract

To realize the real-time highly sensitive detection of SF₆ decomposition product H₂S, a multi-mechanism collaboration enhancement photoacoustic spectroscopy analyzer (MCEPA) based on acoustic resonance enhancement, cantilever enhancement and excitation light enhancement is proposed. An SF₆ background gas-induced photoacoustic cell (PAC) was used for acoustic resonance (AR) enhancement of the photoacoustic signals. A fiber-optic acoustic sensor based on a silicon cantilever is optimized and fabricated. The narrow-band acoustic signal enhancement based on cantilever mechanical resonance (MR) is realized in the optimal working frequency band of the PAC. A fiber-coupled DFB cascaded an Erbium-doped fiber amplifier (EDFA) realized the light power enhancement (LPE) of the photoacoustic signals excitation source. Experimental results show that the MR of the fiber-optic silicon cantilever acoustic sensor (FSCAS) is matched with the AR of the PAC and combined with the LPE, which realizes the multi-mechanism collaboration enhancement of weak photoacoustic signals. The Allan-Werle deviation evaluation showed that the minimum detection limit of H₂S in the SF₆ background is 10.96 ppb when the average time is 200 s. Benefiting from the all-optimization of photoacoustic excitation and detection, the MCEPA has near-field high-sensitivity gas detection capability immune to electromagnetic interference.

1. Introduction

Sulfur hexafluoride (SF₆) is a colorless, odorless industrial synthesis gas with high electrical insulating properties. Electrical equipment with SF₆ as insulation media, such as gas-insulated switches (GIS) and gas-insulated lines (GIL), have gradually replaced traditional insulation equipment in extra-high and ultra-high voltage power systems [1]. Compared with traditional electrical insulation equipment, SF₆-based GIS and GIL have outstanding advantages in insulation efficiency and reliability [2]. When the electrical insulation equipment is in a normal state, the chemical property of SF₆ is very stable. However, in case of overheating or partial discharge of electrical equipment, SF₆ at the insulation defect will be decomposed into low fluorine compounds (SF_x). The SF_x reacts with trace amounts of water and oxygen in the equipment. The reaction products H₂S, SO₂, CO and other gases will corrode electrical equipment and accelerate the attenuation of insulation performance. Through the real-time measurement and analysis of fault characteristic gas (H₂S) concentration in electrical equipment, the operation status of GIS and GIL can be effectively monitored and the loss caused by insulation failure can be avoided [3]. In GIS equipment with a normal operating state, the concentration of fault characteristic gas is usually in the order of ppm (H₂S＜2 ppm, SO₂＜2 ppm, CO＜10 ppm). For the detection of H₂S, 1 ppm is usually taken as the warning value. Therefore, the design of a highly sensitive H₂S analyzer is of great significance for the stable, safe and continuous operation of the power system.

Traditional H₂S concentration analysis is usually based on the gas chromatograph. The component analysis of the gas chromatograph takes a long time, which can not meet the real-time online monitoring of electrical equipment insulation. Online real-time analysis of H₂S concentration is usually based on semiconductor sensors and electrochemical sensors. Various metal oxide semiconductors, nanoparticles, and electrolytes are used as sensitive materials for H₂S [4]. The concentration information of H₂S can be obtained in real-time by monitoring the change of the electrical characteristics of the sensors. However, semiconductor H₂S sensors are usually susceptible to cross-interference and the performance of electrochemical gas sensors decays rapidly. Compared with electrical gas sensors, absorption spectroscopy gas analysis technology (ASGAT) has the advantages of online real-time analysis, minor cross-interference between gases, and high detection sensitivity [5], [6], [7]. The ASGAT mainly realizes the detection of gas composition and concentration based on the specificity of the molecules excited by the irradiated beam.

Photoacoustic spectroscopy (PAS) gas detection is an important branch of ASGAT [8], [9], [10]. When the gas to be measured is excited, part of the photons will be absorbed and transition to a high-energy state. The non-radiative transition of gas molecules releases part of the absorbed light energy in the form of heat [11]. When the excitation source is modulated, periodic thermal expansion of the gas produces photoacoustic signals [12], [13]. The concentration of the target gas can be obtained by measuring the photoacoustic signals through acoustic sensors [14], [15], [16]. Benefiting from the background-free detection mechanism, PAS as indirect absorption spectroscopy technology has a wider dynamic range and more sensitive gas detection performance than traditional direct ASGAT. Szabo et al. designed an H₂S analyzer based on PAS technology. A DFB (central wavelength = 1.57 µm) and an electret acoustic sensor were used for the excitation and acquisition of the H₂S concentration, respectively. A minimum detection limit (MDL) of 6 ppm was achieved and the detection dynamic range reached four orders of magnitude. However, for the field of power equipment insulation condition monitoring, the detection limit is difficult to meet the application requirements [17]. According to Lambert Beer's law, a higher absorption coefficient will enhance the photoacoustic signals. M. Siciliani et al. achieved highly sensitive detection of H₂S in the mid-infrared band with a stronger absorption coefficient [18]. When the integration time is 3 s, the detection limit of H₂S reaches 450 ppb. However, the high cost of quantum cascade laser (EC-QCL) makes it difficult for large-scale applications. S. Viciani. et al. reduced the cost of the mid-infrared H2S detection system by using a DFB laser with a central wavelength of 2.6 µm and the NNEA reached 2.4 × 10⁻⁹ W cm⁻¹ Hz^−1/2 [19]. However, due to the low power of the excitation light, the detection limit of H₂S is 4 ppm at the integration time of 1 s. It is difficult to realize real-time and highly sensitive detection of H₂S in GIS equipment which is usually less than 2 ppm. The excited photoacoustic signals can be enhanced by increasing the excitation power. To improve the detection sensitivity, Wu et al. used an erbium-doped fiber amplifier (EDFA) combined with a near-infrared DFB laser as the excitation light source for H₂S. A custom-made quartz tuning fork (QTF) with a resonance frequency of 7205 Hz was used as a photoacoustic signal detector [20]. Benefiting from the weak acoustic signal resonance enhancement ability of the QTF, this scheme not only reduces the cost but also maintains highly sensitive gas detection performance [21], [22]. With nitrogen as the background gas, the MDL of H₂S reaches 890 ppb at 1 s integration time. However, the physical properties such as density and viscosity of SF₆ are quite different from those of nitrogen. When the background gas is SF₆, the performance of the H₂S analyzer will be changed compared with the nitrogen atmosphere. Dong et al. designed an SF₆ background gas-induced high-Q photoacoustic cell (PAC) with a differential structure [23]. The excitation power of the DFB was amplified to 1.4 W by an EDFA. The highly sensitive detection of H₂S is achieved through excitation optical power enhancement combined with acoustic resonance amplification. The MDL of 109 ppb was obtained under 1 s integration time. However, the photoacoustic signal sensor based on a condenser microphone or QTF is vulnerable to strong electromagnetic interference around electrical equipment. Electromagnetic interference will affect the measured value during the transmission and conversion of weak electrical signals [24], [25].

In this paper, to realize the real-time and highly sensitive all-optical detection of trace H₂S in the SF₆ background, a multi-mechanism collaboration-enhanced photoacoustic analyzer (MCEPA) is proposed. A background gas-induced high-Q resonant PAC is used for acoustic resonance (AR) enhancement of weak photoacoustic signals. A silicon cantilever-based fiber-optic acoustic sensor is used for mechanical resonance (MR) enhancement of photoacoustic signals. Through the optimized design of the H₂S analyzer, the MR frequency of the fiber-optic silicon cantilever acoustic sensor (FSCAS) is matched with the AR frequency of the PAC. An EDFA was used to cascade the DFB to increase the excitation light power and enhance the photoacoustic signals. Combined with a variety of photoacoustic signal enhancement schemes, the highly sensitive detection of H₂S is realized. In addition, the MCEPA is immune to electromagnetic interference, which is especially suitable for monitoring the insulation state of electrical equipment.

2. Theoretical analysis and optimal design

2.1. Design of SF₆ background gas-induced high-Q PAC

The density, viscosity and sound velocity of SF₆ are significantly different from those of air or nitrogen. To realize the highly sensitive all-optical photoacoustic detection of trace H₂S in SF₆ (background gas), it is necessary to optimize the design of PAC and FSCAS. Photoacoustic gas detection technology is based on the selective interaction between photons and molecules. The specific absorption wavelength and intensity of each substance molecule is the basis for quantitative sensing of gas concentration. The photoacoustic spectroscopy gas detection process mainly includes three stages. First, the target gas molecules absorb excitation light of a specific wavelength and transition to a high-energy state. Second, the absorbed light energy is released through transitions. In the infrared band, at atmospheric pressure, the rate of collisional deactivation is much greater than the rate of radiative decay. Therefore, the light energy is mainly converted into heat energy by molecular collisions and causes the temperature of the gas to increase. Finally, the sound waves generated by the periodic thermal expansion of the gas are sensed by the acoustic sensors. For a target gas with a concentration of C and an absorption coefficient of α, when the excitation source is I(r,t), the thermal power density source H(r, t) generated in the PAC can be expressed as [26]:

$$H\left(r,t\right)=C\alpha I\left(r,t\right)$$ (1)

When the excitation light source is modulated with the angular frequency ω, the gas pressure in the PAC periodically fluctuates and photoacoustic signals are generated. The wave equation of sound pressure p can be expressed as [27]:

$$\left({\nabla }^2+\frac{{\omega }^2}{{{\upsilon }_s}^2}\right)p\left(r,\omega \right)=\frac{\gamma -1}{{{\upsilon }_s}^2}i\omega H\left(r,\omega \right)$$ (2)

Where v_s and γ and are the sound speed and heat capacity ratio of the gas in the PAC, respectively. The p(r, ω) is the superposition of multiple modes of sound waves, which can be expanded as:

$$p\left(r,\omega \right)=\sum _j{A}_j\left(\omega \right)p_j\left(r\right)$$ (3)

Where A_j(ω) is the amplitude of the sound field. The p_j(r) is the acoustic vibration mode, which is determined by the shape of the PAC. When the angular frequency of the photoacoustic signal is equal to the j_th order resonance angular frequency (ω = ω_j) of the PAC, the sound field amplitude can be expressed as [28]:

$$A_j\left(\omega \right)=-\frac{Q_{j}C\alpha }{{\omega }_j}\left[\frac{\gamma -1}{V_c}\right]\int p_j^{*}\left(r\right)I(r,\omega )dV$$ (4)

Where p*_j(r) is the complex conjugate of p_j(r), V_c is the volume of the resonance tube, ω_j is the j_th-order normal frequency, Q_j is the acoustic resonance quality factor of the acoustic vibration mode p_j(r).

For a cylindrical PAC, when it operates in the first-order longitudinal resonance mode, the sound pressure A(ω₁) can be expressed as:

$$A\left({\omega }_1\right)=P_0{F}_{cell}C\alpha$$ (5)

Where P₀ is the excitation light power. The F_cell is the cell constant, which is used to describe the acoustic cumulative amplification characteristics of the PAC [29]：

$$f_1=\frac{v_s}{2L_{eff}}$$ (6)

$$L_{eff}=L_{PAC}+\frac{16}{3\pi }{R}_{PAC}$$ (7)

$$F_{cell}=-\left(\gamma -1\right)Q_1\frac{4L_{eff}^2}{V_c{v}_s{\pi }^2}$$ (8)

Where f₁ and L_eff are the first-order AR frequency and effective length of the PAC, respectively. The L_PAC and R_PAC are the length and radius of the PAC, respectively. When the excitation source is modulated at an eigenfrequency of the resonance tube, the energy from the photoacoustic signal of multiple cycles is accumulated in the standing wave, and the PAC plays the role of acoustic amplification [30]. After the initial transient (energy accumulation in the standing wave), the sound field inside the PAC reaches a steady state. The energy of steady-state sound loss is equal to that of photon absorption. The signal amplification capacity of PAC is determined by the total loss of sound waves. When the resonance tube operates in the first-order longitudinal resonance mode, the Q₁ is used to represent the acoustic amplification performance of the PAC. At atmospheric pressure, the main factors affecting the Q₁ are the viscous gas in the cell and the surface losses caused by heat conduction. The Q₁ can be expressed as [31]：

$$Q_1=\frac{R_{PAC}}{{\delta }_{visc}+\left(\gamma -1\right){\delta }_{therm}\left(1+2R_{PAC}/L_{PAC}\right)}$$ (9)

Where δ_visc and δ_therm are used to describe the thickness of the thermal viscosity boundary layer and thermal boundary layer, respectively:

$${\delta }_{visc}=\sqrt{\frac{2\mu }{\omega {\rho }_0}}$$ (10)

$${\delta }_{therm}=\sqrt{\frac{2kM}{\omega {\rho }_0{C}_p}}$$ (11)

Where ρ_0, μ, k and M are the density, dynamic viscosity, thermal conductivity and molar mass of the gas to be measured, respectively.

The distribution of the sound field in axial resonant PAC is simulated by the finite element method, as shown in Fig. 1(a). The SF₆ was used as the background gas in the simulation. The maximum value of the photoacoustic signals is in the middle of the resonance tube. To enhance the sound amplification performance, the structure of the PAC was optimized based on the theoretical analysis results. According to Eq. (4), the photoacoustic signals can be increased by reducing the AR frequency of the PAC. From Eq. (6), the AR frequency can be reduced by increasing the PAC length. Eq. (8) shows that an increase in length will enhance the cell constant of the PAC. However, increasing the length of the PAC will increase the volume of the sensing module, and the consumption of insulating gas inside the electrical equipment will also be increased. According to Eq. (9), increasing the radius of the PAC increases the Q-value. However, as the radius increases, the cell constant decreases. In comprehensive consideration, to obtain a high Q-value and cell constant while maintaining a smaller volume, the designed PAC has a length of 100 mm and a radius of 4 mm. In order to analyze the sound cumulative amplification performance of the PAC, the acoustic field distribution at different frequencies was simulated by COMSOL. As shown in Fig. 1(b), the energy accumulation efficiency of sound waves in PAC will vary with the change of frequency. At the frequency of 630 Hz (AR frequency of the PAC), the enhancement of the sound wave is the largest. Therefore, the maximum sensitivity can be obtained by detecting the photoacoustic signals in the middle of the resonance tube at the laser modulation frequency of 630 Hz.

Fig. 1.

(a) Axial distribution of photoacoustic signals in PAC, (b) the energy accumulation efficiency of sound waves with different frequencies in PAC.

Nitrogen or air is usually used as background gas in traditional photoacoustic spectroscopy for trace H₂S detection. In the field of electrical equipment operating status monitoring, direct measurement of trace fault characteristic components in insulating gas SF₆ without separation is the basis for real-time online measurement of H₂S concentration. The acoustic resonance amplification performance of the PAC will be affected by the physical parameters of the background gas. The physical properties of SF₆ and N₂ as shown in Table 1. According to (6), (7) and Fig. 1(b), when the background gas is N₂ and SF₆, the resonant frequencies of the PAC are 1617 Hz and 630 Hz, respectively. From the Eqs. (9–11) and the Table 1, the quality factor of the PAC change with the resonance frequency is shown in Fig. 2.

Table 1.

Physical properties of SF₆ and N₂.

Gas	vs (m/s)	ρ0 (kg/m3)	γ	M (kg/mol)	μ (Pa/s)	k W/ (mk)	Cp (J/molk)
SF6	133	6.52	1.1	0.146	1.53 × 10-5	0.013	97.5
N2	340	1.16	1.4	0.028	1.75 × 10-5	0.026	29.1

Fig. 2.

The Q-value of the PAC under different background gases.

Analysis results show that the photoacoustic signal generated by the weakly absorbing gas H₂S can be amplified by the acoustic cumulative effect of the resonant PAC. Compared with N₂ background, the quality factor of the SF₆ gas-induced PAC is significantly improved. SF₆ is used as the background gas to detect the concentration of decomposition product H₂S, which can avoid the gas separation process and help improve the acoustic amplification performance of the PAC.

2.2. Design of fiber-optic cantilever acoustic sensor

According to the sound wave enhancement characteristics of the resonance tube, it is of great significance to design an acoustic sensor with high detection sensitivity at the AR frequency of the PAC. A cantilever diaphragm is a rectangular narrow-band acoustic MR enhancement element with one end clamped and the other end free to vibrate [32], [33]. Compared with the traditional circular acoustic-sensitive diaphragm, the oscillatory of the cantilever is almost not inhibited by the surface tensile stress [34]. Therefore, the sound pressure sensitivity can be improved while maintaining a compact structure [35]. The MR of the cantilever is beneficial to enhance the acoustic sensing ability of the FSCAS. When the photoacoustic signals match the natural frequency of the cantilever, the optimal acoustic sensing performance of the FSCAS can be obtained. The MR frequency can be adjusted by optimizing the three-dimensional dimension. When the cantilever MR matches the AR of the PAC, weak photoacoustic signals will be multi-enhanced.

The FSCAS based on the Fabry-Perot (F-P) interferometer is composed of a cantilever and a fiber tip as two optical reflecting surfaces combined with a shell. The fiber is fixed by a ceramic ferrule. The structure of FSCAS is shown in Fig. 3. The FSCAS is immune to electromagnetic interference due to components that are all passive. The broad-spectrum light used for cavity-length demodulation is transmitted to the F-P interferometer through single-mode fiber. The cavity length of the F-P interferometer is proportional to the photoacoustic signals. The H₂S can be measured by demodulating the dynamic variation of the F-P cavity length through a white light interference demodulation algorithm [36], [37].

Fig. 3.

Schematic diagram of FSCAS structure.

The performance of the cantilever is crucial for enhancing the gas detection sensitivity. The nanocrystalline silicon-made cantilever has stable properties and excellent sound pressure response characteristics. For a rectangular silicon cantilever, the MR frequency can be expressed as [15], [37], [38]:

$$f_0=\frac{t_c}{2\pi {L_c}^2}\sqrt{\frac{2E_{si}}{3\times 0.647{\rho }_{si}}}$$ (12)

Where E_si and ρ_si are Young's modulus and density of monocrystalline silicon, respectively. According to Eq. (12), the first-order MR frequency of a cantilever is mainly determined by the length L_c and the thickness t_c.

The relationship between the first-order resonant frequency and the size of the cantilever is shown in Fig. 4. To obtain high sound pressure sensitivity while maintaining mechanical strength, the thickness of the cantilever is selected as 6 µm. The effect of air damping of the cantilever with thickness in the micrometer scale cannot be ignored. The amplitude-frequency response of the cantilever under the action of thermal viscous damping was simulated by COMSOL. The result of the vibration mode finite element analysis of the acoustic element is shown in Fig. 5(a). The sound pressure amplitude-frequency response curve of the free end of the cantilever was shown as the solid line in Fig. 5(b). The dotted line is the amplitude-frequency response of the simulated photoacoustic signals in the PAC. When the length, width, and thickness of the cantilever are 4.4 mm, 1 mm, and 6 µm, respectively, the first-order MR frequency of the FSCAS is consistent with the AR enhancement frequency of the PAC. The most sensitive photoacoustic signal sensing performance can be obtained by placing the FSCAS in the middle of the resonance tube.

Fig. 4.

Relationship between MR frequency and size of a silicon cantilever.

Fig. 5.

(a) Simulation of the vibration mode, (b) simulation of frequency response curves.

2.3. Excitation light power enhancement

The photoacoustic signals can be enhanced by a stronger gas absorption line and higher optical power absorbed by the H₂S molecules. The H₂S has strong absorption bands around 1.6 µm, 2.6 µm and 7.8 µm [39], [40], [41]. The absorption coefficients at 2.6 µm and 7.8 µm are one to two orders of magnitude higher than those at 1.6 µm. However, at 7.8 µm, the detection of H₂S will be disturbed by the light absorption of background gas SF₆ molecules. Lasers with a center wavelength of 2.6 µm have low output power and are expensive. In the absorption band of 1.6 µm, the background gas SF₆ and trace impurity gases (such as H₂O and CO₂) that exist in electrical equipment have almost no absorption (about two orders of magnitude lower than H₂S) at 1574.56 nm and will not interfere with the measurement of H₂S. In addition, the near-infrared DFB laser has the advantages of high stability, long service life and low price. Therefore, a DFB (central wavelength=1574.56 nm) was used as the excitation laser of H₂S. The enhancement of the excitation power is realized by cascading the DFB with an EDFA. The multiple enhancement of the photoacoustic signals can be realized by the EDFA-based excitation light power enhancement combined with the background gas-induced high-Q PAC and the narrow-band resonant FSCAS. Fig. 6.

Fig. 6.

Absorption coefficient of H₂S, H₂O and CO₂ near 1574 nm.

3. Experimental results and discussion

To achieve multi-mechanism collaboration enhancement of photoacoustic signals, a cantilever with a length of 4.4 mm, a width of 1 mm and a thickness of 6 µm was designed. The dry etching technology process is used to batch fabricate silicon cantilever acoustic wave-sensitive diaphragms on a silicon-on-insulator (SOI) wafer, as shown in Fig. 7(a). The fabricated silicon cantilever diaphragm was fixed to the stainless steel shell with epoxy. The single-mode fiber end was cut flat and held in place by a ceramic ferrule. The completed FSCAS is shown in Fig. 7(b). To verify the acoustic response performance of the sensor, the FSCAS was placed in the full anechoic room (296.15 K, 58%RH, 99.15 kPa), as shown in Fig. 7(c). Fig. 7(d) is the schematic diagram of the FSCAS test system. A SLED with a center wavelength of around 1550 nm was used as a broad-spectrum demodulation light source. The broad-spectrum demodulated beam enters the FSCAS after passing through the fiber circulator. The F-P interference spectrum with cavity length information is formed by two beams of light reflected on the end plane of the fiber and the surface of the cantilever respectively. The high-speed spectrometer receives the interference spectrum and transmits it to the computer. The dynamic cavity length variation Δd is demodulated from the F-P interference spectrum by a white light interference demodulation algorithm [42], [43], [44].

Fig. 7.

(a) The SOI wafer for fabrication of silicon cantilever diaphragm, (b) the fabricated FSCAS, (c)the test environment of the FSCAS, (d) the diagram of the FSCAS test system.

To test the sound pressure response characteristics of the FSCAS, the synchronous modulation signal generated by the spectrometer drives the speaker after passing through the power amplifier to emit acoustic waves of different frequencies. The reference microphone unit (B&K 4190) and FSCAS were placed symmetrically on both sides of the acoustic axis. The electrical signal output of the reference microphone unit is collected by the data acquisition (DAQ) module. The sound field information is calculated according to the nominal sensitivity and responsivity of the B&K4190. The sensitivity of FSCAS at different frequencies can be obtained by referring to the sound pressure amplitude monitored by the reference microphone unit combined with the length variation of the F-P cavity. The frequency response curve obtained according to the sensitivity of FSCAS at different frequencies is shown in Fig. 8(a). The experimental results show that the resonance frequency of FSCAS is 620 Hz, which is basically consistent with the simulation results. The FSCAS obtains the best weak acoustic signals sensing ability at 620 Hz, due to the enhanced MR of the cantilever. The vibration amplitude of the cantilever under different sound pressure at the resonance frequency (620 Hz) is collected by changing the loudness of the speaker. The linear fitting result is shown in Fig. 8(b). The FSCAS has good linearity of sound pressure response at 620 Hz, with a sensitivity of 80.17 nm/mPa.

Fig. 8.

(a) The frequency response curve of the FSCAS, (b) linear fitting of sound pressure response.

The multi-mechanism collaboration enhanced photoacoustic spectroscopy H₂S analyzer is shown in Fig. 9. According to the acoustic characteristics of PAC, the AR enhancement efficiency is the highest in the middle of the resonance tube. Therefore, the FSCAS was installed in the middle of the PAC to collect weak photoacoustic signals. The DFB cascaded with an EDFA is used as the excitation light source of H₂S.

Fig. 9.

The main components of the MCEPA.

To test the frequency response characteristics of the analyzer, the mixed gas of H₂S (certified concentration of cylinder: 100.29 ppm) and SF₆ (certified concentration of cylinder: 99.999%) were filled into the PAC. The volume concentration of H₂S in the gas to be measured is 100 ppm. The excitation light power output by the EDFA is 200 mW. When the modulation frequency of the DFB is changed, the response of the photoacoustic signals is shown in Fig. 10. According to the unimodal characteristics of amplitude-frequency response, the matching of AR and MR is realized in the working frequency band of the analyzer. The MR of the cantilever and the AR of the PAC were matched at 644 Hz. The power of the excitation light was enhanced by an EDFA. Fig. 11 shows the relationship between the excitation light power and the photoacoustic signal amplitude. With the increase of the excitation light power, the amplitude of the photoacoustic signal is also enhanced. The photoacoustic signals have a good linear relationship with the excitation light power.

Fig. 10.

The amplitude-frequency response of the MCEPA.

Fig. 11.

The relationship between the excitation light power and the photoacoustic signal amplitude.

To reduce the interference of cell wall absorption on H₂S detection, the second harmonic (2 f) signal detection technology based on a lock-in amplifier is adopted. The photoacoustic signal reached its peak when the laser modulation frequency was 322 Hz [45]. The modulation frequency of the DFB was fixed at 322 Hz, and the excitation optical power was enhanced to 1000 mW by the EDFA. Different concentrations of H₂S were mixed by a mass flow controller (MFC) and filled into the PAC. The second harmonic of H₂S is obtained by modulating the wavelength of the excitation light by superimposing a sawtooth wave with a sine wave. The second harmonic of H₂S with different concentrations is shown in Fig. 12(a). The integration and acquisition time of each data point is 1 s. The amplitude of the 2f-signal was linearly fitted. As shown in Fig. 12(b), the amplitude of the 2f-signals has a good linear relationship with the H₂S concentration. When the excitation light power is 1000 mW, the sensitivity (cantilever vibration amplitude/concentration) of the MCEPA is 12.1 pm/ppm.

Fig. 12.

(a) The 2f-signals of different concentrations, (b) the linear fitting of the amplitude of 2f-signals.

In order to clarify the detection limit of the analyzer, the background noise was collected and analyzed. High-purity SF₆ was filled into the PAC and 1000 data were collected continuously. The trace water and CO₂ molecules in the gas to be measured cannot be completely removed. The absorption of trace impurity molecules and PAC wall results in a non-zero background level. As shown in Fig. 13(a), the standard deviation (1σ) of the noise was calculated to be 1.85 pm with an integration time of 1 s, and the noise equivalent MDL of H₂S under the SF₆ background is 0.15 ppm. The excitation light power and integration time were normalized, and the NNEA [46], [47] was calculated to be 2.49 × 10^-9 cm^-1 W/Hz^1/2. Allan-Werle deviation [48] was used to evaluate the performance of the H₂S analyzer. As shown in Fig. 13(b), when the average time is 200 s, the detection limit of H₂S reaches 10.96 ppb.

Fig. 13.

(a) Background noise of the MCEPA, (b) Allan-Werle deviation estimation of the MCEPA.

The experimental results show that the MCEPA achieves highly sensitive detection of trace H₂S in SF₆ due to the collaboration enhancement effect of multiple mechanisms. The background gas-induced high-Q PAC was used for AR enhancement of the photoacoustic signals. The designed and fabricated FSCAS has ultrasensitive acoustic perception performance at the resonant frequency of the PAC. Compared to traditional H₂S analyzers based on QTFs and condenser microphones, the MCEPA has the advantage of being immune to electromagnetic interference [23], [49], [50], [51], [52]. In addition, the MCEPA achieves or even exceeds the detection performance of traditional H₂S analyzers in N₂ background under SF₆ background gas [49], [50], [51]. The tedious and time-consuming gas separation process is avoided, and real-time high-sensitivity H₂S detection is realized. Table 2.

Table 2.

Comparison with other photoacoustic spectroscopy trace H₂S detectors.

Authors	Detection method/ Background gas	λ/Power	α (cm-1)	NNEA (cm-1 W/Hz1/2)
Sampaolo et al.[53]	QEPAS/N2	104.6 µm/150 mW	3.12	3.1 × 10−8
S. Viciani et al.[19]	QEPAS/N2	2.6 µm/3 mW	9.25 × 10−2	2.4 × 10−9
Helman et al.[50]	QEPAS/N2	8.1 µm/160 mW	6.70 × 10−2	3.05 × 10−9
Yin et al.[54]	Condenser microphone/SF6	1.58 µm/1.35 W	8.18 × 10−3	2.9 × 10−8
Dong et al.[55]	Condenser microphone/SF6	1.58 µm/1.4 W	8.18 × 10−3	2.9 × 10−9
This work	F-P interference/SF6	1.57 µm/1 W	8.25 × 10−3	2.49 × 10−9

4. Conclusion

In conclusion, a multi-mechanism collaboration enhancement photoacoustic spectroscopy analyzer is proposed for the weakly absorbing gas H₂S. Theoretical analysis and finite element simulation are used for the fusion and matching of multiple sensitivity enhancement mechanisms. According to the optimized design results, an ultra-sensitive FSCAS and a resonance PAC are fabricated. The performance of FSCAS was tested in an anechoic room, and the acoustic detection sensitivity reaches 80.17 µm/pa. The DFB laser in the near-infrared band cascaded EDFA realizes the enhancement of effective light power absorption. The experimental results of photoacoustic spectroscopy gas analysis show that the MDL of H₂S is 10.96 ppb when the integration time is 200 s. The MCEPA integrates multiple photoacoustic signal enhancement mechanisms such as the MR of the silicon cantilever, the AR of the resonant PAC, and the LPE of the excitation source. The MCEPA is immune to electromagnetic interference, and all-optical high-sensitivity sensing and determination of trace H₂S in SF₆ background gas is realized.

Funding

This work was supported by the National Nature Science Foundation of China [61905034,62275040]; The Science and Technology Project of State Grid [521205190014]; The Fundamental Research Funds for the Central Universities [DUT21JC03].

Declaration of Competing Interest

The authors declare no conflicts of interest.

Biographies

Min Guo is currently pursuing a Ph.D. degree in optics engineering at the School of Optoelectronic Engineering and Instrumentation Science, Dalian University of Technology (DUT). His current research interests include fiber-optic sensors and photoacoustic spectroscopy.

Xinyu Zhao is currently pursuing a Ph.D. degree in optics engineering at the School of Optoelectronic Engineering and Instrumentation Science, Dalian University of Technology (DUT). Her current research interests include fiber-optic sensors and photoacoustic spectroscopy.

Ke Chen is currently an associate professor at the School of Optoelectronic Engineering and Instrumentation Science, Dalian University of Technology (DUT). He received a Ph.D. degree in 2015 from the School of Physics and Optoelectronic Technology, DUT, Dalian, China. His research interests are fiber-optic sensors and photoacoustic spectroscopy.

Dongyu Cui is currently working towards his master's degree at the School of Optoelectronic Engineering and Instrumentation Science, Dalian University of Technology (DUT). His research interest is fiber-optic sensors and photoacoustic spectroscopy.

Guangyin Zhang is currently working towards his master's degree at the School of Optoelectronic Engineering and Instrumentation Science, Dalian University of Technology (DUT). His research interest is fiber-optic sensors and photoacoustic spectroscopy.

Chenxi Li received her B.S. degree in 2020 from the School of Physics, University of Jinan (UJN), Jinan, China. She is currently pursuing a Ph.D. degree in optics engineering at the School of Optoelectronic Engineering and Instrumentation Science, Dalian University of Technology (DUT). Her current research interests include fiber-optic sensors and photoacoustic spectroscopy.

Zhenfeng Gong received his Ph.D. degree in optical engineering from DUT, Dalian, China, in 2018. He is currently an associate professor at the School of Optoelectronic Engineering and Instrumentation Science, DUT. His current research interests include fiber-optic sensors and photoacoustic spectroscopy.

Qingxu Yu is currently a professor at the School of Optoelectronic Engineering and Instrumentation Science, at the Dalian University of Technology. His current research interests are fiber-optic sensors and laser spectroscopy.

Data Availability

The authors are unable or have chosen not to specify which data has been used.