Open Access
Add to BibSonomy
Abstract
In this manuscript, a highly sensitive methane (CH4) sensor based on light-induced thermoelastic spectroscopy (LITES) using a 2.33 µm diode laser with high power is demonstrated for the first time. A quartz tuning fork (QTF) with an intrinsic resonance frequency of 32.768 kHz was used to detect the light-induced thermoelastic signal. A Herriot multi-pass cell with an effective optical path of 10 m was adopted to increase the laser absorption. The laser wavelength modulation depth and concentration response of this CH4-LITES sensor were investigated. The sensor showed excellent long term stability when Allan deviation analysis was performed. An adaptive Savitzky-Golay (S-G) filtering algorithm with χ2 statistical criterion was firstly introduced to the LITES technique. The SNR of this CH4-LITES sensor was improved by a factor of 2.35 and the minimum detection limit (MDL) with an integration time of 0.1 s was optimized to 0.5 ppm. This reported CH4-LITES sensor with sub ppm-level detection ability is of great value in applications such as environmental monitoring and industrial safety.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Methane (CH4) is one of the main gases causing the greenhouse effect, and its absorption of surface radiation energy per unit time is about 20 times higher than that of carbon dioxide [1,2]. At the same time, the widespread use of CH4 as civil and industrial fuel is accompanied by great safety hazards. Furthermore, the explosions of coal mine gas [3,4] and sewage pipe [5,6] occur endlessly. Therefore, the sensitive detection of CH4 is important for alleviating the greenhouse effect and ensuring the safe production and life.
Gas sensing technology based on absorption spectroscopy has the advantages of fast response, high sensitivity, multi-component measurement and continuous monitoring [7–13]. Therefore, it is emerging as an ideal method for trace gas concentration detection. However, the direct absorption technique of tunable diode laser absorption spectroscopy (TDLAS) is prone to optical interference during the test [14,15]. The indirect absorption technique of quartz-enhanced photoacoustic spectroscopy (QEPAS) is a contact measurement method. When corrosive or oxidizing gases are measured [16,17], the performance of quality factor and resonance characteristic of acoustic wave transducer of quartz tuning fork (QTF) in QEPAS sensor may be degraded due to the exposure to the measured environment.
QEPAS technique has limitations on its scope of analyzed gas, and it also does not allow for remote trace gas detection and combustion diagnostics. To effectively address these issues, light-induced thermoelastic spectroscopy (LITES) technique was proposed by researchers in 2018 [18]. When a laser is irradiated on the surface of QTF, the laser power is absorbed by the quartz crystal and converted into thermal energy. The fork fingers of QTF undergoes a thermoelastic deformation after being heated and form a periodic mechanical vibration due to the laser modulation [19–22]. When the laser modulation frequency coincides with the resonance frequency of QTF, the mechanical vibration amplitude of QTF will be enhanced. According to the piezoelectric effect, the mechanical vibration of QTF is converted into an electrical signal, which is collected by the silver layer on the surface of QTF [23]. The concentration of the gas can be obtained by demodulation of the electrical signal. The LITES technique has been widely applied in the detection of trace gases [24–27]. A miniature fiber-coupled CH4-LITES sensor with an absorption line of 6046.95 cm−1 was reported in 2020, and the minimum detection limit (MDL) of the sensor was ∼48.8 ppm at an integration time of 0.3 s [28]. A wavelength-locked LITES (WL-LITES) gas sensor system with the same absorption line of 6046.95 cm−1 was reported in 2021 for CH4 detection, and the MDL of ∼11 ppm at an integration time of 0.3 s was achieved [28].
For the most gas molecules, the distribution of absorption lines can be distinguished into the fundamental vibrational-rotational band and the overtone band. The intensity of the absorption lines in the fundamental band tends to be much stronger than the overtone band. Quantum cascade lasers (QCLs) and interband cascade lasers (ICLs) emitting the mid-infrared light are usually used to cover this band. But the expensive price and requiring water cooling restrict their applications. In comparison, diode lasers with the merits of compact size and low cost are widely adopted to target the overtone region in the gas sensing field. For CH4 detection, the spectral line intensity of CH4 located in the overtone band of 2.3 µm is approximately one order larger than it in the overtone band of 1.6 µm. However, a laser of 1.65 µm was chosen to detect CH4 in the literature [28,29], which limits the sensor performance.
In this paper, a highly sensitive CH4-LITES sensor using a 2.33 µm diode laser with high optical power was reported for the first time. A QTF with a resonant frequency f0 of 32.768 kHz was used as the thermoelastic detector. Multi-pass gas absorption technique was adopted to enhance the laser absorption. The wavelength modulation depth of the CH4-LITES sensor was optimized to achieve the maximum 2f signal amplitude at the selected absorption line. In addition, the adaptive Savitzky-Golay (S-G) algorithm was firstly introduced by using the χ2 statistical criterion to improve the signal-to-noise ratio (SNR) of the LITES sensor. The long term stability of the reported sensor was investigated through the Allan deviation analysis.
2. Experimental setup
2.1 Absorption line selection
According to the HITRAN 2016 database, the absorption lines of CH4 in 2.33 µm region is simulated and shown in Fig. 1. It can be seen that the absorption coefficient of CH4 is much larger than that of other gases of water vapor (H2O) and carbon dioxide (CO2) when the optical path reaches 1000 cm. Based on the simulation shown in Fig. 1 and the availability of laser sources in the laboratory, a continuous wave (CW), distributed feedback (DFB) diode laser with a wavelength of 2.33 µm was selected as the laser source and an CH4 absorption line located at 4294.55 cm−1 was selected in this experiment. The temperature and driving current tuning coefficients of diode laser wavelength were 0.322 cm−1/°C and 0.017 cm−1/mA, respectively.
Fig. 1. Absorption lines of CH4, H2O and CO2 near 4295 cm−1 spectral region at atmosphere pressure, room temperature and optical absorption length of 1000 cm according to the HITRAN 2016 database.
Diode lasers typically use both temperature-tuned and current-tuned methods to change output wavelength to target specific wavelength bands. The current-tuned approach allows for modulating output wavelength of laser up to GHz, a feature that guarantees the implementation of wavelength modulation spectroscopy (WMS). The output characteristics of this 2.33 µm CW-DFB diode laser are shown in Fig. 2. The laser can operate in the temperature range of −5 to 70 °C with the maximum injection current of 300 mA. The laser output wavelength as a function of TEC operating temperature and injection current is shown in Fig. 2(a). It can be found that the setting of TEC operating temperature at 11 °C completely covers the absorption line of 4294.55 cm−1. Also, as can be seen in Fig. 2(b), the laser output power is as high as 7.62 mW when the injection current reached 300 mA.
Fig. 2. The output characteristics of 2.33 µm laser. (a) The trend of laser output wavelength versus TEC temperature and injection current. (b) The output power of laser versus TEC temperature and injection current.
2.2 Sensor system configuration
The experimental setup for the reported CH4-LITES sensor is shown in Fig. 3. A CW-DFB laser (Model #: KELD1G5BAAH, NEL Corp., Japan) at 2.33 µm was selected as the excitation source. The laser controller modulated the output wavelength of the laser according to the input signal from a signal generator. A triangular wave signal with frequency of 1 Hz and duty cycle of 50% was generated by the signal generator to scan the selected absorption line. A sine wave with a frequency at half of the resonant frequency f0 of QTF was superimposed on a triangular wave and used as the input signal to the laser controller. The diode laser beam from the pigtail was firstly collimated by a fiber collimator (FC), and then sent into a Herriot multi-pass cell (Model #: PCI-HC10 m, Port City Instruments, LLC, USA) with optical length of 10 m to enhance the laser absorption. The exiting laser beam from multi-pass cell was focused by a planar-convex lens with a focal length of 30 mm onto the QTF surface. A QTF (Model #: DT38-C12Q-32.768 kHz, King Yu Xing Technology Co., Ltd.) with a resonance frequency f0 of 32.759 kHz and a quality factor (Q-factor) of 14250 at 1 atm was utilized as the thermoelastic detector. According to the relationship Q = f0/Δf between f0 and Q-factor of the QTF, the detection bandwidth Δf of the QTF was determined as 2.42 Hz. Due to the light-thermo-elastic effect, the electrical signal was generated by the QTF and it was sent to a lock-in amplifier (Model #: MFLI DC-500 kHz, Zurich Instruments, Switzerland) for demodulation and analysis. The bandwidth of the lock-in amplifier was set to 1.16 Hz, and the filter roll off was 18 dB/oct. The integration time of the sensor was set to 100 ms. A 2% CH4:N2 mixture was used as the target gas. Two flow controllers (Model #:SC117 D07-19B, Bei Jing Sevenstar Flow Co., Ltd., ±1% F.S.) were selected to mix CH4 with N2 in different ratios in order to verify the detection performance of the sensor, and the flow rate was controlled at 120 mL/min.
3. Results and discussions
Wavelength modulation spectroscopy (WMS) with 2nd harmonic (2f) detection technique was utilized to simply the data processing and reduce the background noise. The wavelength modulation depth was optimized firstly to improve the 2f signal amplitude of this CH4-LITES sensor. The relationship between the modulation depth and 2f signal amplitude was investigated and shown in Fig. 4.
As shown in Fig. 4, the optimum modulation depth of the sensor was experimentally determined to be 0.81 cm−1 for the selected CH4 absorption line of 4294.55 cm−1. Because the used 2.33 µm CW-DFB diode laser can be tuned to cover 4292–4296 cm−1 wavelength range, the 2f signal of the CH4-LITES sensor at different modulation depth for this range was investigated and is shown in Fig. 5. Figure 5(a) depicts the absorption coefficients between 4292–4296 cm−1 obtained through the HITRAN 2016 database. It can be found that multiple absorption peaks appeared between 4292–4296 cm−1. As shown in Fig. 5(b)-(e), when the wavelength modulation depth increased from 0.18 to 0.81 cm−1, the 2f signal amplitude for the absorption line located at 4294.55 cm−1 gradually increased. But for the absorption lines located at 4293.09 cm−1 and 4293.58 cm−1 they had the different optimum modulation depth, which mainly result from the fact that different absorption lines have different line widths.
Fig. 5. (a) Simulated absorption spectra based on the HITRAN 2016 database; (b)-(e) Measured 2f CH4-LITES signal at different modulation depths for the wavenumber range of 4292–4296 cm−1.
The concentration response of this CH4-LITES sensor was verified. The 2% CH4:N2 was diluted with pure N2 using two mass flow controllers to obtain different concentrations of the gas mixture. The 2f signal for different CH4 concentrations is shown in Fig. 6(a). The 2f signal amplitude at the absorption line of 4294.55 cm−1 extracted from Fig. 6(a) for different concentrations is displayed in Fig. 6(b). A linear fit of the amplitude showed that the value of the coefficient of determination R2 is about 0.99 (close to the constant 1), which indicated that this CH4-LITES sensor showed good linear response for different concentrations of CH4.
Fig. 6. The concentration response of CH4-LITES sensor: (a) 2f signal with different CH4 concentrations; (b) Linear fit of the 2f signal amplitude as a function of CH4 concentrations.
The long term stability of this CH4-LITES sensor was investigated. The 2f signal amplitude was recorded by continuous demodulation for 90 minutes after replacing CH4:N2 with pure N2 in the gas chamber. The obtained data was used for Allan deviation analysis, and the result is shown in Fig. 7. The Allan deviation analysis showed that the minimum detection limit (MDL) of this CH4-LITES sensor could reach 0.4 ppm when the optimum integration time was 100 s, which indicated that this CH4-LITES sensor has good detection capability and system stability.
In order to improve the sensor performance, the filtering algorithm was further used for noise reduction. The Savitzky-Golay (S-G) filtering algorithm has been widely used in the data analysis of direct absorption spectroscopy [30] due to its good denoising effect and simple parameter settings [31]. The S-G filtering algorithm essentially uses local modeling, which means a filtering method based on a local polynomial least squares fit in the time domain. Based on pre-set parameters, the filtering algorithm is able to remove the noise from the original signal at a time. The algorithm contains only two parameters, polynomial fitting order and window length, which usually need to be set empirically. However, when the sensor system is perturbed, it is difficult for the S-G filter with fixed parameters to achieve the desired filtering effect. Therefore, a calibration-free adaptive S-G filtering algorithm was proposed in this manuscript for the effective noise reduction. Gabriel et al. considered the S-G filter to be optimal when the filtered noise had the highest autocorrelation coefficient with the instrument noise [31]. Lu et al. considered that when the similarity between the signal of test optical path and the signal of reference optical path was maximum, the parameters of the filtering algorithm should be optimal [32]. All these above methods inevitably require the provision of reference signals, which are often difficult to obtain during the actual test. The similarity index χ2 [31] is introduced in the adaptive S-G as a judgment criterion to avoid the use of reference signal in this paper.
χ2 is used to characterize the degree of dispersion between the data before and after denoising, which can be expressed as [33]
Before and after applying the adaptive S-G filtering algorithm to the 2f signal of this CH4-LITES sensor, the denoised results are shown in Fig. 9(a). A local signal comparison between the original signal and the denoised signal are depicted in Fig. 9(b). It can be found from Fig. 9 that after processing with the adaptive S-G filtering algorithm, the noise level in the original signal was significantly suppressed. The SNR increased by 2.35 times from 953.18 to 2239.99. At the same time, the MDL of CH4-LITES sensor was reduced to 0.5 ppm with an integration time of 0.1 s. Table 1 shows the detection ability of sensors based on different LITES methods for CH4 sensing. It can be found that the sensor in this work has a better detection performance than the existing ones. The adaptive S-G algorithm was proved to be effective for LITES technique.
Table 1. Comparison of different CH4-LITES sensors.
4. Conclusions
In this paper, a highly sensitive CH4-LITES sensor using a 2.33 µm diode laser was reported for the first time. A QTF with an intrinsic resonance frequency of 32.768 kHz and a Q factor of 14250 was used as a detector for the light-induced thermoelastic effect. A Herriot multi-pass cell with an effective optical path of 10 m was adopted to increase the laser absorption. The laser wavelength modulation depth was optimized to be 0.81 cm−1 for the selected CH4 absorption line located at 4294.55 cm−1. CH4 gas mixtures with different concentrations were detected, indicating that the sensor has an excellent linear concentration response. To evaluate the system stability for this CH4-LITES sensor, an Allan deviation analysis was carried out. The measured results showed that the MDL of the sensor reached 0.4 ppm with an integration time of 100 s. We introduced an adaptive S-G filtering algorithm with χ2 statistical criterion to LITES technique for the first time. The SNR of the CH4-LITES sensor was improved and the MDL with an integration time of 0.1 s was reduced to 0.5 ppm. This reported CH4-LITES sensor with sub ppm-level detection ability is of great importance in applications such as environmental monitoring and industrial safety. The MDL of the CH4-LITES sensor can be further improved by increasing the output power of laser based on power amplification or choosing a multi-pass cell with long optical length.
Funding
National Natural Science Foundation of China (61505041, 61875047, 62022032); Natural Science Foundation of Heilongjiang Province (YQ2019F006); Fundamental Research Funds for the Central Universities; Heilongjiang Provincial Postdoctoral Science Foundation (LBH-Q18052).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. J. R. Evans, “Resolving methane fluxes,” New Phytol. 175(1), 1–4 (2007). [CrossRef]
2. M. Borowski, P. yczkowski, J. Cheng, R. uczak, and K. Zwolinska, “The combustion of methane from hard coal seams in gas engines as a technology leading to reducing greenhouse gas emissions-electricity prediction using ANN,” Energies 13(17), 4429 (2020). [CrossRef]
3. A. E. Dursun, “Statistical analysis of methane explosions in Turkey's underground coal mines and some recommendations for the prevention of these accidents: 2010–2017,” Nat. Hazards 104(1), 329–351 (2020). [CrossRef]
4. S. K. Kundu, J. Zanganeh, D. Eschebach, and B. Moghtaderi, “Explosion severity of methane–coal dust hybrid mixtures in a ducted spherical vessel,” Powder Technol. 323, 95–102 (2018). [CrossRef]
5. B. Su, Z. M. Luo, T. Wang, and L. Liu, “Experimental and numerical evaluations on characteristics of vented methane explosion,” J. Cent. South Univ. 27(8), 2382–2393 (2020). [CrossRef]
6. W. Li, T. Zheng, Y. Ma, and J. Liu, “Current status and future prospects of sewer biofilms: Their structure, influencing factors, and substance transformations,” Sci. Total Environ. 695, 133815 (2019). [CrossRef]
7. S. D. Qiao, Y. F. Ma, P. Patimisco, A. Sampaolo, Y. He, Z. T. Lang, F. K. Tittel, and V. Spagnolo, “Multi-pass quartz-enhanced photoacoustic spectroscopy based trace gas sensing,” Opt. Lett. 46(5), 977–980 (2021). [CrossRef]
8. R. Rousseau, Z. Loghmari, M. Bahriz, K. Chamassi, R. Teissier, A. N. Baranov, and A. Vicet, “Off-beam QEPAS sensor using an 11-µm DFB-QCL with an optimized acoustic resonator,” Opt. Express 27(5), 7435–7446 (2019). [CrossRef]
9. Y. F. Ma, “Research progress of quartz-enhanced photoacoustic spectroscopy based gas sensing,” Acta Phys. Sin. 70(16), 160702 (2021). [CrossRef]
10. H. D. Zheng, Y. Liu, H. Lin, B. Liu, and F. K. Tittel, “Quartz-enhanced photoacoustic spectroscopy employing pilot line manufactured custom tuning forks,” Photoacoustics 17, 100158 (2020). [CrossRef]
11. Y. F. Ma, R. F. Lewicki, M. Razeghi, and F. K. Tittel, “QEPAS based ppb-level detection of CO and N2O using a high power CW DFB-QCL,” Opt. Express 21(1), 1008–1019 (2013). [CrossRef]
12. J. P. Waclawek, H. Moser, and B. Lendl, “Compact quantum cascade laser based quartz-enhanced photoacoustic spectroscopy sensor system for detection of carbon disulfide,” Opt. Express 24(6), 6559–6571 (2016). [CrossRef]
13. S. D. Qiao, Y. He, and Y. F. Ma, “Trace gas sensing based on single-quartz-enhanced photoacoustic-photothermal dual spectroscopy,” Opt. Lett. 46(10), 2449–2452 (2021). [CrossRef]
14. T. Y. Zhang, J. W. Kang, D. Z. Meng, H. Wang, Z. M. Mu, M. Zhou, and C. Chen, “Mathematical methods and algorithms for improving near-infrared tunable diode-laser absorption spectroscopy,” Sensors 18(12), 4295–4310 (2018). [CrossRef]
15. R. Yang and Y. Zhang, “A method of low concentration methane measurement in tunable diode laser absorption spectroscopy and Levenberg-Marquardt algorithm,” Optik 224, 165657 (2020). [CrossRef]
16. Y. F. Ma, Y. He, X. Yu, C. Chen, R. Sun, and F. K. Tittel, “HCl ppb-level detection based on QEPAS sensor using a low resonance frequency quartz tuning fork,” Sens. Actuators B 233, 388–393 (2016). [CrossRef]
17. H. P. Wu, A. Sampaolo, L. Dong, P. Patimisco, X. L. Liu, H. D. Zheng, X. K. Yin, W. G. Ma, L. Zhang, W. B. Yin, V. Spagnolo, S. T. Jia, and F. K. Tittel, “Quartz enhanced photoacoustic H2S gas sensor based on a fiber-amplifier source and a custom tuning fork with large prong spacing,” Appl. Phys. Lett. 107(11), 111104 (2015). [CrossRef]
18. Y. F. Ma, Y. He, Y. Tong, X. Yu, and F. K. Tittel, “Quartz-tuning-fork enhanced photothermal spectroscopy for ultra-high sensitive trace gas detection,” Opt. Express 26(24), 32103–32110 (2018). [CrossRef]
19. Z. T. Lang, S. D. Qiao, Y. He, and Y. F. Ma, “Quartz tuning fork-based demodulation of an acoustic signal induced by photo-thermo-elastic energy conversion,” Photoacoustics 22, 100272 (2021). [CrossRef]
20. S. D. Russo, A. Zifarelli, P. Patimisco, A. Sampaolo, T. T. Wei, H. P. Wu, L. Dong, and V. Spagnolo, “Light-induced thermo-elastic effect in quartz tuning forks exploited as a photodetector in gas absorption spectroscopy,” Opt. Express 28(13), 19074–19084 (2020). [CrossRef]
21. Y. F. Ma, Y. He, P. Patimisco, A. Sampaolo, S. D. Qiao, X. Yu, F. K. Tittel, and V. Spagnolo, “Ultra-high sensitive trace gas detection based on light-induced thermoelastic spectroscopy and a custom quartz tuning fork,” Appl. Phys. Lett. 116(1), 011103 (2020). [CrossRef]
22. S. Zhou, K. Chen, L. G. Xu, B. L. Yu, T. T. Jiang, and J. S. Li, “Ultrathin two-dimensional Fe-doped cobaltous oxide as a piezoelectric enhancement mechanism in quartz crystal tuning fork (QCTF) photodetectors,” Opt. Lett. 46(3), 496–499 (2021). [CrossRef]
23. Y. He, Y. F. Ma, Y. Tong, X. Yu, and F. K. Tittel, “Ultra-high sensitive light-induced thermoelastic spectroscopy sensor with a high Q-factor quartz tuning fork and a multi-pass cell,” Opt. Lett. 44(8), 1904–1907 (2019). [CrossRef]
24. S. D. Qiao, Y. F. Ma, Y. He, P. Patimisco, A. Sampaolo, and V. Spagnolo, “Ppt level carbon monoxide detection based on light-induced thermoelastic spectroscopy exploring custom quartz tuning forks and a mid-infrared QCL,” Opt. Express 29(16), 25100–25108 (2021). [CrossRef]
25. Y. Hu, S Qiao, Y. He, Z. Lang, and Y. Ma, “Quartz-enhanced photoacoustic-photothermal spectroscopy for trace gas sensing,” Opt. Express 29(4), 5121–5127 (2021). [CrossRef]
26. C. G. Lou, X. T. Li, H. J. Chen, X. Yang, Y. Zhang, J. Q. Yao, and X. L. Liu, “Polymer-coated quartz tuning fork for enhancing the sensitivity of laser-induced thermoelastic spectroscopy,” Opt. Express 29(8), 12195–12205 (2021). [CrossRef]
27. Y. F. Ma, Y. Q. Hu, S. D. Qiao, Y. He, and F. K. Tittel, “Trace gas sensing based on multi-quartz-enhanced photothermal spectroscopy,” Photoacoustics 20, 100206 (2020). [CrossRef]
28. L. Hu, C. Zheng, Y. Zhang, J. Zheng, Y. Wang, and F. K. Tittel, “Compact all-fiber light-induced thermoelastic spectroscopy for gas sensing,” Opt. Lett. 45(7), 1894–1987 (2020). [CrossRef]
29. L. Hu, C. Zheng, M. Zhang, K. Zheng, and F. K. Tittel, “Long-distance in-situ methane detection using near-infrared light-induced thermo-elastic spectroscopy,” Photoacoustics 21, 100230 (2021). [CrossRef]
30. J. Li, B. Yu, and H. Fischer, “Wavelet transform based on the optimal wavelet pairs for tunable diode laser absorption spectroscopy signal processing,” Appl. Spectrosc 69(4), 496–506 (2015). [CrossRef]
31. G. V. Truyols and J. Peter Schoenmakers, “Automatic selection of optimal Savitzky-Golay smoothing,” Anal. Chem. 78(13), 4598–4608 (2006). [CrossRef]
32. Y. B. Lu, W. Q. Liu, Y. J. Zhang, K. Zhang, H. E. Ying, and K. You, “An adaptive hierarchical Savitzky-Golay spectral filtering algorithm and its application,” Spectrosc. Spectr. Anal. 39(09), 2657–2663 (2019). [CrossRef]
33. H. G. Schulze, R. B. Foist, A. Ivanov, and R. F. B. Turner, “Fully automated high-performance signal-to-noise ratio enhancement based on an iterative three-point zero-order Savitzky-Golay Filter,” Appl. Spectrosc. 62(10), 1160–1166 (2008). [CrossRef]
