Abstract
A compact and highly sensitive photoacoustic spectroscopy (PAS) sensor for dissolved acetylene (C2H2) detection was developed for transformer fault diagnosis applications. A 1.53 μm distributed-feedback (DFB) diode laser targeting the ν1 + ν3 absorption line of acetylene at ∼1530.37 nm was employed, benefiting from strong line intensity and high transparency of transformer oil in the near-infrared region. Wavelength modulation spectroscopy combined with second-harmonic (2f) demodulation was implemented, and the modulation depth was optimized to 120 mVpp to achieve the maximum normalized 2f signal. A high-Q resonant photoacoustic cell was designed to enhance the acoustic response and improve the signal-to-noise ratio. Experimental results demonstrated excellent linearity (R2 = 0.999) over the concentration range of 200–1,000 ppm. For a standard 100 ppm C2H2/N2 mixture, a peak 2f signal amplitude of 5.4 × 10−5 V was obtained, with a noise level of 4.05 × 10−7 V (1σ). Allan deviation analysis further confirmed the outstanding long-term stability of the system, achieving a detection sensitivity of 0.75 ppm at 1 s and a minimum detection limit of 50 ppb at an integration time of 3,200 s. These results indicate that the proposed PAS sensor enables accurate, real-time, and trace-level acetylene detection, demonstrating strong potential for dissolved gas analysis in transformer oil and early warning of incipient arcing faults.
1 Introduction
Power transformers are critical and expensive assets in the electric power grid. Their reliable operation is paramount, and even incipient faults can lead to serious failures if left undetected []. Dissolved Gas Analysis (DGA) of transformer oil has long been established as one of the most informative techniques for early fault diagnosis []. When thermal or electrical faults (e.g., overheating, partial discharge, arcing) occur in a transformer, insulating oil decomposes and generates various gases (H2, CH4, C2H6, C2H4, C2H2, CO, CO2) whose identity and concentrations correlate with fault type []. Acetylene (C2H2) in particular is widely recognized as a key indicator of high-energy electrical discharges (arcing) and severe overheating faults [–]. The presence of even trace amounts of C2H2 in oil is considered a serious warning sign of arcing faults. Industry standards (IEC and IEEE) have codified DGA interpretation, assigning acetylene a central role in fault diagnostics []. Accurate and timely detection of dissolved acetylene is thus critically important for condition monitoring and preventive maintenance of power transformers.
Common methods for detecting dissolved gases (including acetylene) in transformer oil include laboratory gas chromatography (GC), on-line electrochemical or semiconductor gas sensors, and emerging optical techniques. Gas chromatography is the traditional reference method for DGA and provides accurate quantitative analysis of multiple gases [–]. However, GC systems are large, slow (requiring batch sampling), and not well suited for real-time on-site monitoring. On-line gas sensor devices (e.g., catalytic hot-wire, metal-oxide semiconductor, or electrochemical cells) offer faster in-situ measurements, but they often suffer from cross-sensitivity to other gases, drift in calibration, and interference from electrical noise in the substation environment []. For example, solid-state electrochemical sensors based on yttria-stabilized zirconia have achieved acetylene detection limits around 0.5–1 ppm, but require high operating temperatures and careful calibration []. Moreover, all these conventional solutions depend on first extracting the dissolved gas from oil (via techniques like headspace equilibrium, membrane permeation, or vacuum degassing) before measurement. The oil–gas separation step can be slow (equilibrium times of hours or days in membrane methods) and can introduce uncertainties in gas quantification.
In recent years, optical methods have shown great promise for more sensitive and direct detection of fault gases. For instance, Fourier-transform infrared (FTIR) spectroscopy can simultaneously analyze multiple gases and has been applied to DGA (detecting C2H2 down to a few tenths of a ppm) []. Raman spectroscopy has also been explored, though Raman signals are inherently weak and require enhancement techniques to reach useful sensitivity []. Among optical techniques, photoacoustic spectroscopy (PAS) has emerged as especially promising due to its high sensitivity and potential for on-line implementation [–]. Photoacoustic spectroscopy is a laser-based gas sensing technique that measures sound waves generated by intermittent light absorption in the target gas []. When a modulated laser beam at a wavelength resonant with a gas absorption line irradiates the sample, the gas molecules absorb energy and subsequently release it via non-radiative relaxation, producing periodic local heating. This causes pressure oscillations—i.e., acoustic waves—that can be detected by a sensitive microphone or acoustic transducer []. PAS has a zero-background advantage (the signal is generated only by absorbed light, not by the transmitted intensity), enabling extremely low detection limits []. Very recent progress in PAS gas sensing has focused on compact resonant-cell miniaturization and geometry/structure optimization, multi-gas detection architectures, and quartz-enhanced variants (QEPAS/LITES), which further push detection limits toward the ppb–ppt regime while enabling more integrated and cost-effective implementations [–25].
In this work, we present a resonant PAS sensor for dissolved acetylene in transformer oil, leveraging a 1.53 μm distributed-feedback (DFB) laser and optimized acoustic resonance to achieve high sensitivity. 1.53 μm is chosen as the operating wavelength because transformer oil is relatively transparent in the near-infrared around this wavelength and acetylene has a strong absorption line (the ν1 + ν3 band P9 line) at ∼1530.37 nm. The use of a telecom-band DFB laser (with wavelength modulation and second-harmonic detection) offers an economically favorable and stable light source compared to mid-IR lasers, while still accessing a sensitive acetylene absorption. We designed a resonant PA cell and carefully tuned the laser modulation frequency to coincide with the cell’s acoustic resonance, maximizing the signal-to-noise ratio. An Allan deviation analysis was employed to characterize the sensor’s long-term stability and to identify the optimal integration time. Different from many recent ppb-level PAS/QEPAS C2H2 sensors that depend on optical amplification (e.g., EDFA), multipass/cavity enhancement, or specialized acoustic transducers (e.g., fiber-optic/cantilever sensors), the present work targets a simpler and more deployable architecture based on cost-effective telecom components and a conventional microphone, while still achieving ppb-level detection capability.
2 Experiments
The selection of a strong, interference-free absorption line is essential for high-sensitivity detection of acetylene (C2H2) using photoacoustic spectroscopy. Based on the HITRAN database, the C2H2 absorption spectrum near 1.53 μm was simulated to identify a line with high intensity and minimal spectral overlap. In this study, the P (9) line at 1530.37 nm (≈6535.57 cm-1), with an absorption line strength of 1.19 × 10−20 cm/molecule at 296 K, was selected as the target transition due to its strong signal response and practical availability of DFB laser sources in this region.
In practical transformer applications, gas samples may contain trace moisture. In PAS measurements, humidity can affect the sensor response mainly through (i) spectral overlap and (ii) humidity-dependent collisional relaxation. Regarding spectral interference, the selected C2H2 P (9) transition at 1530.37 nm (6534.37 cm-1) has been reported to be free from spectral interference of other molecules based on HITRAN-assisted line selection, and thus the contribution from H2O absorption at this wavelength is expected to be negligible. Regarding non-spectral effects, water vapor can modify the radiation-to-sound conversion efficiency by influencing molecular relaxation processes; therefore, for field deployment, an inline moisture trap/dryer can be applied upstream of the photoacoustic cell if the moisture level is non-negligible, which stabilizes the gas matrix and minimizes any humidity-related influence.
Figure 1a presents the simulated absorption spectrum of C2H2 in the 1,529–1,550 nm range under standard conditions. The selected target line at 1530.37 nm exhibits a relatively isolated profile with minimal neighboring interference, making it ideal for selective detection. To ensure precise alignment of the laser emission with the selected absorption feature, wavelength tunability of the DFB diode laser was characterized.
FIGURE 1
As shown in Figure 1b, the emission wavelength of the DFB laser was measured as a function of driving current and chip temperature. The results indicate a clear monotonic dependence, allowing accurate adjustment of the laser wavelength to match the selected C2H2 line. The dashed red line indicates the trajectory of the target absorption line in the wavelength–temperature–current domain, which provides practical guidance for selecting the optimal operating parameters. This characterization ensures reliable spectral alignment during PAS-based sensing and supports robust field deployment.
The schematic of the resonant photoacoustic spectroscopy (PAS) system for C2H2 detection is illustrated in Figure 2. A dual-channel function generator (RIGOL, Beijing, China) was used to simultaneously produce a low-frequency triangular waveform and a high-frequency sinusoidal waveform. The triangular waveform was employed to linearly scan the laser wavelength across the target absorption line, while the sinusoidal modulation (at the cell’s acoustic resonance frequency) enabled 2f wavelength modulation spectroscopy. These two signals were superimposed and fed into the current input of the laser driver to realize composite modulation of the distributed-feedback (DFB) laser. The DFB laser, operating near 1530.37 nm, was current- and temperature-tunable. Coarse wavelength tuning was achieved by adjusting the TEC setpoint, while fine modulation was realized through modulation of the injection current. The laser output was delivered through a fiber collimator and focused into the photoacoustic cell (PAC) using a fiber focuser and lens system. Inside the resonant PAC, modulated light was absorbed by acetylene molecules, generating periodic local heating and resulting in pressure oscillations. These acoustic waves were detected by a central microphone mounted at the acoustic antinode. The microphone signal was first passed through a custom-built pre-amplifier for initial signal conditioning, and then routed into a lock-in amplifier (SR830, Stanford Research Systems, Sunnyvale, CA, USA), which performed phase-sensitive detection at the second harmonic of the modulation frequency (2f). The lock-in amplifier received a reference signal from the function generator to ensure phase coherence, and the demodulated photoacoustic signal was transmitted via RS232 interface to a personal computer for data acquisition and analysis through LabVIEW software (National Instruments, Austin, TX, USA). The demodulated 2f signal was obtained using the SR830 internal low-pass filter (time constant τ = 1 s, filter slope 12 dB), which defines the effective noise-equivalent bandwidth; unless otherwise stated, no additional digital filtering was applied beyond the lock-in output, and the recorded values correspond to the steady-state lock-in readings. The optical module is modular and can be readily extended to multi-gas detection by switching/adding additional diode lasers at other wavelengths targeting interference-free absorption lines, while using the same photoacoustic cell and lock-in based demodulation (e.g., time-division or frequency-division multiplexing).
FIGURE 2
The PAS system employs a conventional cylindrical resonator–buffer PAC architecture widely used in resonant PAS to enhance acoustic response and suppress flow/window noise. Typical longitudinally resonant PACs use a centimeter-scale resonator tube (inner diameter on the order of ∼5–10 mm and length ∼90–120 mm) connected to buffer volumes. For this class of resonant cells, the fundamental longitudinal resonance typically lies in the ∼1–2 kHz range (e.g., 1,310 Hz and 1730 Hz reported in representative resonant PAC implementations), with an acoustic quality factor typically on the order of ∼10–35 depending on viscous/thermal losses and inlet/outlet configuration. The acoustic signal is commonly detected by a commercial condenser/electret microphone with sensitivity in the tens of mV/Pa (e.g., 50 mV/Pa reported in a representative resonant PAC system). Buffer volumes and 2f lock-in demodulation help reject ambient acoustic disturbances, while stabilizing pressure/flow mitigates slow resonance drift since the resonance frequency depends on gas composition, temperature and pressure.
On the gas handling side, a calibrated mass flow controller (MFC; Alicat Scientific, Tucson, AZ, USA) was used to regulate the flow rate of the target gas mixture into the PAC. The internal pressure of the cell was maintained by a pressure controller to ensure stable acoustic resonance conditions. After flowing through the cell, the gas was exhausted through a dedicated outlet. All components of the system were monitored and coordinated via a LabVIEW-based control program, enabling fully automated scanning, modulation, and signal acquisition. All tests reported in this work were performed in the gas phase using certified C2H2/N2 mixtures; no oil–gas extraction module was integrated in the present laboratory setup. The total flow rate was fixed at 200 sccm for the 2f characterization, calibration, and N2 purge noise measurement. The PAC pressure was stabilized by the pressure controller at ∼1 atm (101 kPa) (exhausted to ambient) to keep the acoustic resonance condition constant. For practical transformer-oil monitoring, the PAS analyzer can be interfaced with standard dissolved-gas extraction approaches such as vacuum extraction, stripping, or headspace partitioning, as recommended in IEC guidance, or with a membrane-degassing unit commonly used in online DGA instruments.
3 Results and discussion
In wavelength modulation spectroscopy (WMS) combined with second-harmonic (2f) demodulation, the modulation depth is a key operating parameter that directly affects the magnitude of the detected signal and, consequently, the sensitivity of the PAS-based sensing system. In this work, the modulation depth is defined as the peak-to-peak voltage (Vpp) of the sinusoidal signal generated by the function generator and applied to the laser driver, which determines the amplitude of the laser wavelength modulation.
Figure 3 illustrates the optimization of the modulation depth for the developed PAS acetylene sensing system. The horizontal axis represents the modulation depth in terms of the peak-to-peak driving voltage applied to the laser current controller, while the vertical axis corresponds to the normalized 2f signal amplitude, defined as the peak amplitude of the demodulated 2f photoacoustic signal normalized to the maximum value obtained in this modulation-depth sweep. This optimization was carried out using a 100 ppm C2H2/N2 mixture under the same gas-handling conditions as the subsequent 2f measurement. As the modulation depth increases from 40 mVpp to 150 mVpp, the normalized second-harmonic signal first increases and reaches its maximum at 120 mVpp (marked by the star). Beyond this point, the signal slightly saturates or decreases, which can be attributed to an over-modulation effect: excessively large wavelength excursion leads to a non-optimal modulation index relative to the absorption linewidth and reduces the efficiency of 2f signal generation, potentially accompanied by increased non-ideal modulation effects. In addition, the sinusoidal modulation frequency was selected to maximize the resonant PA response (i.e., such that the detected 2f component coincides with the PAC acoustic resonance), while the triangular scan rate was chosen to be sufficiently slow compared with the lock-in time constant to avoid dynamic distortion of the 2f lineshape.
FIGURE 3
Based on this quantitative optimization result, a modulation depth of 120 mVpp was selected as the optimal operating parameter for all subsequent measurements. This choice ensures that the PAS sensor operates at its maximum signal response while maintaining stable and reproducible performance. The optimized modulation depth plays a crucial role in achieving high signal-to-noise ratio and, ultimately, in enabling sensitive detection of trace-level acetylene in transformer oil.
To characterize the second-harmonic (2f) response of the developed PAS system to acetylene, a standard gas mixture containing 100 ppm C2H2 balanced with N2 was introduced into the photoacoustic cell. The total gas flow rate was precisely controlled at 200 sccm using calibrated mass flow controllers to ensure stable and reproducible measurement conditions. During the experiment, the injection current of the DFB laser was scanned from 80 mA to 120 mA to cover the target absorption line of acetylene, and the corresponding PAS 2f signal was recorded in real time.
Figure 4 presents the measured 2f signal waveform obtained under these conditions. The horizontal axis represents the laser injection current, while the vertical axis denotes the voltage amplitude of the 2f signal generated by the PAS system. As the laser current is scanned across the absorption feature, a clear second-harmonic lineshape is observed, exhibiting a pronounced central peak accompanied by two symmetric side lobes. This characteristic profile confirms that the laser wavelength modulation and 2f demodulation are properly synchronized with the selected acetylene absorption line.
FIGURE 4
As shown in Figure 4, the peak amplitude of the 2f signal reaches approximately 5.4 × 10−5 V at the center of the absorption line when the acetylene concentration is 100 ppm. To quantify the system noise, the photoacoustic cell was subsequently purged with pure nitrogen under the same experimental conditions, and the background signal was recorded. The standard deviation of the noise was determined to be 4.05 × 10−7 V (1σ). This baseline noise at the lock-in output is mainly attributed to the electronic noise of the microphone pre-amplifier and the lock-in front-end, with minor contributions from residual acoustic/flow-induced fluctuations and cell/window background; laser-related noise is largely suppressed by resonance operation and narrow-band 2f lock-in detection. Based on these values, the signal-to-noise ratio (SNR) for 100 ppm C2H2 can be calculated to be 133.3. This determines a minimum detection limit of the sensor of ∼75 ppb for C2H2 detection.
To evaluate the quantitative detection capability of the developed PAS acetylene sensor, calibration experiments were performed at multiple C2H2 concentrations. Standard C2H2/N2 mixtures with concentrations of 200, 400, 600, 800, and 1,000 ppm were generated using calibrated mass flow controllers under the same operating conditions as those used in the signal characterization experiments. For each concentration point, the corresponding PAS 2f signal amplitude was recorded as the sensor response.
Figure 5 summarizes the measured responses at different acetylene concentrations. The bar chart shows that the sensor signal increases monotonically with increasing C2H2 concentration, indicating a clear concentration-dependent response. To assess the linearity, the measured signal amplitudes were extracted and fitted using a linear regression model, as indicated by the dashed red line in Figure 5. The fitted result yields an excellent determination coefficient of R2 = 0.999, demonstrating that the sensor output is highly proportional to the acetylene concentration over the investigated range. This outstanding linearity confirms the reliable quantitative detection capability of the proposed PAS sensor. It also indicates that the system exhibits stable modulation/demodulation performance and minimal concentration-dependent distortion, which are essential for accurate online monitoring applications. The high linear response across 200–1,000 ppm provides a solid basis for subsequent measurements, including sensitivity analysis and detection limit evaluation.
FIGURE 5
To further evaluate the long-term stability and ultimate detection capability of the developed PAS acetylene sensor, an Allan deviation analysis was performed. During this test, the photoacoustic system was operated continuously while the photoacoustic cell was filled with pure N2, so that the recorded output represented the intrinsic system noise and drift rather than a true gas absorption signal. The sensor signal was acquired over an extended period, and the Allan deviation (σ) was calculated as a function of the averaging (integration) time to determine the optimal integration time and the corresponding minimum detection limit.
As shown in Figure 6, the Allan deviation decreases as the averaging time increases, indicating that random noise dominates the system output at short integration times and can be effectively suppressed by time averaging. At an averaging time of 1 s, the Allan deviation corresponds to a detection sensitivity of 0.75 ppm. With increasing integration time, σ continues to decline, demonstrating that the sensor maintains stable operation without significant short-term fluctuation.
FIGURE 6
When the averaging time reaches 3,200 s, the Allan deviation achieves its minimum value, corresponding to an estimated detection limit of 50 ppb. Here, the Allan deviation σ(τ) is treated as a 1σ stability metric (two-sample standard deviation), and thus the reported detection sensitivity of 0.75 ppm at 1 s and the minimum detection limit of 50 ppb at 3,200 s are based on 1σ. For reference, the corresponding 3σ-equivalent detection limits are ∼2.25 ppm at 1 s and ∼150 ppb at 3,200 s. Beyond the optimum averaging time, σ(τ) tends to level off and may slightly increase, indicating that slow drift starts to dominate. Possible drift sources include laser power/wavelength fluctuations, temperature variations, pressure/flow changes that shift the acoustic resonance condition, and electronic offset drift in the preamplifier/lock-in detection chain. To contextualize the proposed system, recent PAS-based acetylene sensors have achieved lower MDLs by introducing optical power/path enhancement and/or specialized acoustic transducers. For instance, a differential Helmholtz PAS sensor combined an EDFA and dense-spot multi-pass excitation and reported an MDL of 5 ppb at 200 s [26, ]. An all-fiber DDR-PAS sensor employing fiber-optic acoustic sensors achieved 3.8 ppb at 299 s []. A gourd-type PA cell leveraging a 3D optical-path extension reported 0.59 ppb at 100 s []. For dissolved-gas monitoring in transformer oil, an EDFA-boosted QEPAS system reported 469 ppb sensitivity after gas extraction. Compared with these approaches, our system is intentionally designed to avoid optical amplification/cavity locking and exotic microphones, and thus prioritizes cost-effectiveness and practical deployability while still reaching 50 ppb at 3,200 s (Allan deviation) using a telecom-band DFB laser and a conventional microphone.
4 Conclusion
In this work, a compact and highly sensitive photoacoustic spectroscopy (PAS) sensor for dissolved acetylene (C2H2) detection was developed and systematically evaluated. A 1.53 μm distributed-feedback (DFB) diode laser targeting the ν1 + ν3 absorption line of acetylene at ∼1530.37 nm was employed, together with a high-Q resonant photoacoustic cell and wavelength modulation spectroscopy combined with second-harmonic (2f) demodulation. The modulation depth was optimized to 120 mVpp to maximize the normalized 2f signal. Under this condition, a well-defined 2f response was obtained for a 100 ppm C2H2/N2 mixture, yielding a peak signal amplitude of 5.4 × 10−5 V and a noise level of 4.05 × 10−7 V (1σ). Calibration experiments over 200–1,000 ppm demonstrated an excellent linear response with a correlation coefficient of R2 = 0.999. Allan deviation analysis further confirmed the outstanding long-term stability of the system, achieving a detection sensitivity of 0.75 ppm at 1 s and a minimum detection limit of 50 ppb at an integration time of 3,200 s. These results demonstrate that the developed PAS sensor combines high sensitivity, excellent linearity, and robust stability, enabling reliable trace-level acetylene detection.
From an application perspective, the proposed PAS sensor shows strong potential for dissolved gas analysis (DGA) in transformer oil. By coupling the sensor with a compact oil–gas extraction interface, such as a headspace or membrane degassing unit, the system can be readily adapted for online monitoring of incipient electrical faults, particularly arcing-related failures indicated by trace-level C2H2. The achieved ppb-level detection capability enables early fault warning well before critical alarm thresholds are reached, thereby improving transformer reliability and operational safety. Future work will focus on further system miniaturization, multi-gas detection capability (e.g., CH4, C2H4, CO, and H2), and long-term field deployment, paving the way toward an integrated, intelligent DGA platform for next-generation smart grid and condition-based maintenance applications. Finally, although trace moisture may be present in practical transformer gas samples, the selected 1530.37 nm C2H2 transition is spectrally isolated according to HITRAN-based line selection, and an upstream moisture trap/dryer can be employed when necessary to suppress humidity-related matrix effects. In practical transformer monitoring, multiple diagnostic gases are relevant (e.g., H2, CO, CO2, CH4, C2H6, C2H4 and C2H2). The proposed PAS architecture is compatible with multi-gas sensing by incorporating additional laser wavelengths for other target species and applying multiplexing strategies such as time-division multiplexing (sequential measurement) or frequency-division multiplexing (simultaneous measurement). Each added channel would require selecting spectrally isolated absorption features and performing an independent calibration. In future work, the proposed PAS module will be coupled with a standard oil–gas extraction front-end (e.g., headspace/vacuum/stripping or membrane degassing) to enable direct dissolved-gas measurement in transformer oil for field deployment.
Statements
Data availability statement
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.
Author contributions
LP: Writing – original draft, Writing – review and editing. YQ: Formal Analysis, Software, Validation, Methodology, Writing – original draft. YaZ: Data curation, Investigation, Writing – original draft. QW: Writing – original draft, Formal Analysis, Visualization. YiZ: Resources, Supervision, Writing – review and editing. QF: Supervision, Funding acquisition, Writing – original draft, Project administration.
Funding
The author(s) declared that financial support was received for this work and/or its publication. China Southern Power Grid Project (GDKJXM20231566).
Conflict of interest
Authors LP, YQ, YaZ, QW, YiZ, and QF were employed by Electric Power Research Institute of Guangdong Power Grid Co., Ltd.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
1.
IbrahimAHMSadanandanSKGhaoudTRajkumarVSSharmaM.Incipient fault detection in power distribution networks: review, analysis, challenges and future directions. IEEE Access (2024). 12:112822–112838. 10.1109/ACCESS.2024.3443252
2.
HuangY-CSunH-C. Dissolved gas analysis of mineral oil for power transformer fault diagnosis using fuzzy logic. IEEE Trans Dielectrics Electr Insul (2013) 20(3):974–81. 10.1109/tdei.2013.6518967
3.
NouriBOjaghiM. Power transformer fault diagnosis via gas chromatography: a review of techniques and standards. Electromechanical Energ Convers Syst (2022) 2(4):17–29. 10.30503/EECS.2024.479067.1061
4.
ShutenkoO. Analysis of gas composition in oil-filled faulty equipment with acetylene as the key gas. Energetika (2019) 65.1. 10.6001/energetika.v65i1.3973
5.
ShutenkoO. Faults diagnostics of high-voltage equipment based on the analysis of the dynamics of changing of the content of gases. Energetika (2018) 64.1. 10.6001/energetika.v64i1.3724
6.
AbinayaMAswiniASriSDAnupriyaPKumarKVigneshwaranB.Power transformer fault diagnosis using optimized fuzzy C-Means clustering. In: 2025 7th International Conference on Inventive Material Science and Applications (ICIMA). IEEE (2025) 527–532.
7.
ShutenkoOVKulykOSPonomarenkoSG. Comparative analysis of existing standards and methodologies for interpreting DGA results. (2021).
8.
WanjareSBSwamiPS. DGA interpretation for increasing the percent of accuracy by different methods: a review. In: 2018 International Conference on Computation of Power, Energy, Information and Communication (ICCPEIC). IEEE (2018).
9.
GolarzJ. Understanding dissolved gas analysis (DGA) techniques and interpretations. In: 2016 IEEE/PES Transmission and Distribution Conference and Exposition (T&D). IEEE (2016).
10.
ZhaoAXTangXJLiuJHZhangZH.The on-site DGA detecting and analysis system based on the fourier transform infrared instrument. In: 2014 IEEE International Instrumentation and Measurement Technology Conference (I2MTC) Proceedings. IEEE (2014) 1036–1040.
11.
KorotcenkovGChoBK. Chemical sensors selection and operation guide. Chemical sensors: comprehensive sensor technologies. In: KorotcenkovD, editor New York: Momentum Press (2011) 6: 281–348.
12.
WangJZhaoLLiJLiuFYouRLvSet alStabilized zirconia-based solid state electrochemical gas sensor coupled with CdTiO3 for acetylene detection. Sensors Actuators B: Chem (2020) 316:128199. 10.1016/j.snb.2020.128199
13.
TangXWangWZhangXWangELiX. On-line analysis of oil-dissolved gas in power transformers using fourier transform infrared spectrometry. Energies (2018) 11(11):3192. 10.3390/en11113192
14.
DasRSAgrawalYK. Raman spectroscopy: recent advancements, techniques and applications. Vibrational Spectroscopy (2011) 57(2):163–76. 10.1016/j.vibspec.2011.08.003
15.
LiJChenWYuB. Recent progress on infrared photoacoustic spectroscopy techniques. Appl Spectrosc Rev (2011) 46(6):440–71. 10.1080/05704928.2011.570835
16.
YangTChenWWangP. A review of all-optical photoacoustic spectroscopy as a gas sensing method. Appl Spectroscopy Reviews (2021) 56(2):143–70. 10.1080/05704928.2020.1760875
17.
Bonilla-ManriqueOEPérez Gonzalez-BanfiAViñuela PérezJDessenaG. Photoacoustic spectroscopy combined with a multipass circular cell to detect low concentrations of ammonia. Appl Sci (2025) 15(12):6727. 10.3390/app15126727
18.
WijesingheDRZobairMAEsmaeelpourM. A review on photoacoustic spectroscopy techniques for gas sensing. Sensors (2024) 24(20):6577. 10.3390/s24206577
19.
FathyASabryYMHunterIWKhalilDBourouinaT. Direct absorption and photoacoustic spectroscopy for gas sensing and analysis: a critical review. Laser and Photon Rev (2022) 16(8):2100556. 10.1002/lpor.202100556
20.
WangSHoffmannMYetisenAKWangKBrändleFKurzWet alOptical interferometer-based methods for photoacoustic gas sensing: a review. Appl Spectrosc Rev (2024) 59(3):382–421. 10.1080/05704928.2023.2196729
21.
QiaoSLiuXLangZHeYChenWMaY. Quartz-enhanced laser spectroscopy sensing. Light: Sci and Appl (2026) 15(1):5. 10.1038/s41377-025-02075-7
22.
ChangEQWangGLiYFYuYWangYLuZ. Study on research progress of quartz-enhanced photoacoustic spectroscopy for trace gas detection. Front Phys (2025) 13:1709349. 10.3389/fphy.2025.1709349
23.
WangWFengZJiangYZhangJYanSCaoXet alTopological optimization method “MMA-BP” for photoacoustic resonator and implementations in ppt-level gas sensor using miniaturized vase-type photoacoustic cells. Photoacoustics (2025) 46:100767. 10.1016/j.pacs.2025.100767
24.
LiMZhuTWuGGuanYGongZPengWet alA double dual-resonance enhanced photoacoustic spectroscopy gas sensor for simultaneous C2H2 and CH4 detection. Sensors Actuators B: Chem (2025) 445:138617. 10.1016/j.snb.2025.138617
25.
LuoBDaiJWangGZhangHXiaoWZhongZet alQuartz‐enhanced photoacoustic spectroscopy sensor for detection of dissolved carbon monoxide in transformer oil. Microwave Opt Technology Lett (2026) 68(1): e70516. 10.1002/mop.70516
26.
ZhangCHeYQiaoSLiuYMaY. High-sensitivity trace gas detection based on differential helmholtz photoacoustic cell with dense spot pattern. Photoacoustics (2024) 38:100634. 10.1016/j.pacs.2024.100634
27.
QianCNiWYangCZhaoZZhangLRanSet alThree-dimensional optical path extended gourd-type photoacoustic cell for highly sensitive trace acetylene sensing. Photoacoustics (2025) 45:100762. 10.1016/j.pacs.2025.100762
28.
WuYMaTLiuCFanYShiSGuoSet alHigh-sensitivity dynamic detection of dissolved acetylene in transformer oil based on high-power quartz-enhanced photoacoustic spectroscopy sensing system photonics. MDPI (2025) 12(7):713. 10.3390/photonics12070713
Summary
Keywords
acetylene, gas sensor, laser spectroscopy, photoacoustic spectroscopy, photoacustic
Citation
Peng L, Qian Y, Zhao Y, Wang Q, Zhao Y and Fu Q (2026) A resonant photoacoustic spectroscopy sensor for ppb-level detection of acetylene employing a 1.53 μm laser diode. Front. Phys. 14:1791434. doi: 10.3389/fphy.2026.1791434
Received
19 January 2026
Revised
07 February 2026
Accepted
10 February 2026
Published
05 March 2026
Volume
14 - 2026
Edited by
Xukun Yin, Xidian University, China
Reviewed by
Wang Jiapeng, Shanxi University, China
Li Biao, Chongqing University of Posts and Telecommunications, China
Updates
Copyright
© 2026 Peng, Qian, Zhao, Wang, Zhao and Fu.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Lei Peng, 13657209328@163.com
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.