Highly sensitive optical gas sensors have garnered significant attention for their outstanding sensitivity, exceptional selectivity, and rapid response. These sensors are now extensively used in atmospheric monitoring, medical diagnostics, interstellar detection, and so on [1–7]. One of the various optical sensors available is based on quartz-enhanced photoacoustic spectroscopy (QEPAS), which uses a tiny quartz tuning fork (QTF) as an acoustic transducer. This enables the detection of gas concentrations at levels as low as parts per million (ppm) or even parts per billion (ppb) [8–14]. Additionally, the QTF can serve as a photodetector, as demonstrated in light-induced thermoelastic spectroscopy (LITES) [15,16]. Since LITES does not require measurement of the acoustic signal generated by the photoacoustic effect of gases, the QTF can be removed from the measurement environment. This capability facilitates non-contact measurement, significantly advancing the field of optical gas sensing, particularly for detecting acidic and corrosive gases [17–23].

Exploring and developing a QTF with a high quality factor (Q-factor) has always been one of the important subjects in LITES techniques for improving the detection performance [16,24,25]. Commercial QTF is extracted from a quartz crystal oscillator shelled in a metal cylindrical case with a Q-factor of 104–105. However, the Q-factor would be reduced to ∼104 when the metal cylindrical case is peeled due to the damping effect of ambient gas around the QTF [26–30]. Reducing the ambient gas pressure around the QTF can improve the Q-factor 3.4-fold, but the process is complex [16]. Besides, the stability of the QTF’s resonance frequency depends heavily on the stability of the pressure controller. Furthermore, it would also introduce additional noise during implementation. Another way to obtain high Q-factor QTF is to redesign and fabricate a custom one [31]. The Q-factor can be improved slightly below 20,000 by carving the rectangular grooves on two prong surfaces. However, this method is very expensive. Meanwhile, with a Q-factor of less than 20,000, there are limitations to the performance improvements of QTF-based LITES sensors.

In this paper, a novel hydrogen (H2)-enhanced LITES sensor is reported for the first time, where H2 is used to surround the QTF, reducing viscous damping and thereby increasing the Q-factor. This method not only eliminates the need for a pressure controller and gas pump, which are necessary in traditional pressure reduction approaches, but also boosts the Q-factor to over 30,000. Other gases, such as helium (He), methane (CH4), and carbon dioxide (CO2), were also explored to demonstrate the Q-factor enhancement mechanism. The H2-enhanced LITES sensor’s performance was evaluated using acetylene (C2H2) as the analyte, owing to its significance in detecting fault gases in transformers and in ethylene streams for polyethylene production. Finally, a self-designed round-head QTF with a low resonance frequency (f0) and a fiber coupled multipass cell (MPC) with an optical length of 40 m were utilized to increase the energy accumulation time of QTF and the optical absorption of the target gas, respectively, in this new technique to achieve ultra-high sensitivity in C2H2 detection.

Exposing a QTF to gas or air results in a shift (Δf) and damping of its f0. As the gas with different thickness adheres to the surface of vibrating prongs, the effective mass of the fork increases. In the case of constant pressure, the relative variation of its resonance frequency Δf/f0 due to the presence of gas can be calculated according to Ref. [32]: Δff0=aρg+bρgη,(1)where a, b are coefficients that depend on the QTF’s density, size, and the resonance frequency; ρg is the density of gas and η is the viscosity of the gas. Meanwhile, 1/Q reflects the QTF intrinsic energy loss in surrounding gas, summarized based on Ref. [33]: 1Q=Aη+Bρgη+C,(2)where A, B are coefficients that depend on the QTF’s density, size, and the resonance frequency; C is a coefficient related to the QTF’s support losses and the intrinsic energy loss in absolute vacuum conditions.

A schematic of the QTF frequency shift in H2-enhanced LITES is shown in Fig. 1. The QTF is surrounded by pure H2. At room temperature, the density of air is 1.225kg/m3, while that of H2 is 0.082kg/m3. Additionally, the viscosities of air and H2 are 1.82×10−5Pa·s and 0.88×10−5Pa·s, respectively. As a consequence, the presence of H2 will increase the f0 of QTF. With lower gas density and smaller viscosity, f0 and the Q-factor will increase, leading to a narrower response frequency bandwidth. It is worth mentioning that the frequency shift is usually less than several hertz, and such frequency shift will not make any impact on the performance of the LITES sensor [34,35].

The configuration of the H2-enhanced LITES sensor for proof-of-principle is detailed in Fig. 2. A ramp signal was superimposed on a sine waveform to control the laser driver to perform the wavelength modulation spectroscopy (WMS). A 1.53 μm distributed feedback (DFB) fiber coupled diode laser with an output power of ∼16mW was used as the excitation source. The output laser was collimated by a fiber collimator (FC) with a focal length of 8 mm, and then propagated into a gas cell with an optical length of 200 mm. The gas cell was filled with 2% C2H2:N2 gas mixture. The output laser beam from the gas cell was focused by a plano-convex lens with a focal length of 40 mm and transmitted into the QTF’s gas cell. A gas mixer was used to inject the H2 into the QTF’s gas cell in order to investigate the gas damping effect. The piezoelectric signal of the QTF was amplified by a transimpedance amplifier (TA) with equivalent impedance of 10 MΩ and then sent to a lock-in amplifier (Zurich Instruments, MFLI) that was used to demodulate the piezoelectric signals with the reference of the sine signal. All the data stream and control data were processed by a PC.

To investigate the frequency response characteristics of the QTF in the H2 surrounding and evaluate the Q-factor enhancement performance, H2 with different concentrations diluted from pure nitrogen (N2) was injected into the QTF’s gas cell. First, the f0 response of QTF was measured by utilizing the optical excitation method and is shown in Fig. 3(a). It can be observed that, at the same optical excitation intensity, both the central f0 and amplitude increased with rising H2 concentration, indicating that the intrinsic energy loss of the QTF decreased as H2 concentration increased. The f0 and the Q-factor were extracted by fitting the frequency response curves with the Lorentz function. As shown in Fig. 3(b), the f0 of the QTF shift presents a linear relationship with the H2 concentration. Meanwhile, we can also notice that the frequency shift was about 6 Hz before and after injecting H2. Compared with the f0 of 32.768 kHz, the relative variation of its f0 accounts for 10−4-level, and such frequency shift has almost no effect on the LITES sensor. However, the Q-factor improves significantly (from 12,600 to 32,020) with increased H2 concentration, though not linearly, which is notable and advantageous for the LITES sensor.

To preliminarily evaluate the performance of the sensor, the second harmonic (2f) signal was measured when the QTF was surrounded by pure N2 and H2. The QTF’s gas cell was injected by N2 or H2 at a constant flow rate of 10 mL/min. As shown in Fig. 5, the 2f signal amplitude has been improved significantly when the QTF was surrounded by H2, and 2.5 times improvement was achieved compared with that of N2. The noise level of such two conditions was evaluated when the gas cell was filled in pure N2. As shown in Fig. 5, the noise levels are comparable. Therefore, the signal to noise ratio (SNR) reached a 2.9 times improvement in the H2-enhanced LITES sensor.

In order to characterize the effect of introducing H2 on the long-term performance of the LITES sensor, the Allan deviation analysis was implemented by measuring and averaging the LITES signal at zero C2H2 concentration for more than 1 h. For comparison, the QTF surrounded by pure H2 and N2 was applied during the measurement, and the results are shown in Fig. 6. It can be seen that the stability of such two conditions is quite similar; the Allan deviation shows a white noise behavior before 70 s as indicated by the red line with a slope of ∼τ−1/2. Therefore, the H2-enhancement introduced almost no extra noise in the LITES sensor system. In the H2-enhanced LITES sensor, the minimum detection limit (MDL) can reach 78 ppb for an integration time of 70 s.

As one of the important raw materials of organic synthesis, C2H2 is an important industrial gas and is widely used in synthetic rubber, fiber, and plastic production. Besides, it also has important applications in the detection of fault gases in transformers. But C2H2 is flammable and explosive, which poses significant risks to human health and safety. Recently, it has been demonstrated that C2H2 can be treated as a marker of human health [36], with several hundred ppb measured in breath directly after smoking with a fast decay down to ambient levels within 3 h. Therefore, accurate measurement of C2H2 is crucial in a wide range of applications.

The schematic diagram of the ultra-highly sensitive C2H2-LITES sensor is shown in Fig. 7. To further improve the detection sensitivity, a self-designed round-head QTF depicted in the insert of Fig. 7 with low f0 of 9527 Hz was used, which leads to a long light-induced thermoelastic energy accumulation time in the QTF [37]. The length, width, and thickness of the round-head QTF are 9.1 mm, 2.2 mm, and 0.25 mm, respectively. Besides, a fiber coupled MPC with an optical length of 40 m, in which the transmission loss of the fiber coupled MPC is 3.98 dB, was adopted to improve the optical absorption. The DFB laser output was directly connected to the fiber coupled MPC, and the MPC was fed into the 20 ppm C2H2:N2 gas mixture diluted with pure N2. Additionally, there was optical feedback in the fiber input. Such optical feedback of the fiber coupled MPC was directly sent into a spectrum analyzer by a fiber circulator. Thus, the laser wavelength obtained from the spectrum analyzer was input into the data acquisition card (Zurich Instruments, MFLI) and served as the locking feedback signal. A digital proportion integration differentiation (PID) program based on LabVIEW was developed to control the laser modulation current to realize the wavelength locking. The output laser from the MPC was focused by a plano-convex lens with a focal length of 20 mm and then incident into the QTF’s gas cell. The laser power incident on the QTF is 5 mW. The QTF’s gas cell was sealed with pure H2 in advance to maintain a high Q-factor and non-disturbance caused by gas flow. Similarly, the QTF’s piezoelectric signal was amplified by the TA and then demodulated by a lock-in amplifier with the WMS technique.

For the self-designed round-head QTF, the f0 and Q-factor were measured first by the light excitation method. The laser focused on QTF was modulated by a square wave with frequency scanning from 9520 to 9536 Hz. Therefore, the frequency response can be obtained by means of the demodulation of the lock-in amplifier referenced with the square wave. As shown in Fig. 8, the f0 of the self-designed round-head QTF in H2 shifts slightly compared with that in pure N2. It is obvious that the Q-factor increased a lot, which presents a 2.55 times improvement from 12,610 to 32,140.

The location of the laser focal point on the QTF’s surface was experimentally optimized; it was found to be located at the joint position of the QTF’s prongs as shown in Fig. 7. The laser wavelength modulation depth was optimized to obtain the highest 2f signal. The LITES signal amplitude as a function of the laser modulation current is shown in Fig. 9. The maximum signal was achieved with a laser modulation current of 15 mA, so this optimal value was used in the subsequent experiments.

The LITES 2f signal was measured when the QTF was surrounded by pure H2 and pure N2. The fiber coupled MPC was filled with 20 ppm C2H2:N2 gas mixture in a constant flow rate of 120 mL/min. Triangular waves superimposed sine waves were applied to perform laser wavelength modulation. The 2f signals of three triangular wave periods were measured constantly, as shown in Fig. 10. It can be seen that the 2f signal shows a very good symmetry in the ascending and descending regions of the triangular wave. Meanwhile, the 2f signal amplitudes of 358 μV and 141 μV were obtained in the conditions of pure H2 and pure N2, respectively. As a result, 2.54 times improvement for 2f signal amplitude of H2-enhancement was achieved.

To demonstrate the concentration response of such a H2-enhanced C2H2-LITES sensor, the 20 ppm C2H2:N2 gas mixture was diluted 200 times to 0.1 ppm concentration levels with pure N2 by a gas mixer, and the laser wavelength was locked at the absorption peak of the C2H2 gas molecule during the continuous measurement. The 2f signal amplitude was recorded for more than 5000 s, and the signal of the concentration step scanned from 20 ppm to 0.1 ppm and back to 20 ppm was extracted. As shown in Fig. 11, it can be seen that the measured signal amplitude holds a good consistency during the concentration variation. A constant several μV level background signal with 1σ of 52 nV was observed in pure N2 gas, which resulted in an MDL of 2.8 ppb with an 800 ms integration time, corresponding to a normalized noise equivalent absorption (NNEA) coefficient of 5.8×10−11cm−1WHz−1/2. It is important to note that the noise level of the H2-sealed QTF has decreased significantly. The insert in Fig. 11 shows the averaged signal amplitude as a function of C2H2 concentrations. The measured signal with linear fitting R-square of 0.999 indicates an excellent linearity concentration response in the investigated two orders dynamic range.

The Allan deviation analysis was performed to investigate the long-term stability of the H2-enhanced C2H2-LITES sensor system. The signal amplitude at zero C2H2 concentration (pure N2) was measured and averaged for more than 1 h. As shown in Fig. 12, the measured signal satisfies the normal distribution and indicates a typical Gaussian white noise. According to the Allan deviation analysis, the optimum integration time was 140 s, which resulted in an MDL of 290 ppt (0.29 ppb). The optimum integration time is much longer than the one of commercial QTF with f0 of 32.768 kHz, which is mainly due to the higher laser power transition loss in the fiber coupled MPC holding a lower laser power focused on the QTF with less thermal noise.

The main parameters (e.g., QTF’s f0, optical length, and MDL) of the reported C2H2-LITES sensors are summarized in Table 2. The H2-enhanced LITES sensor developed in this work shows an MDL superior to the reported C2H2-LITES sensors using QTFs with different f0 and optical lengths. Most remarkably, with almost the same QTF’s f0 and optical length but higher optical transmittance of MPC, the MDL in this work is 241 times lower than that in a previous study [42]. To sum up, it has been proven to be the most sensitive C2H2-LITES sensor.Table 2.

Main Parameters of the Reported C₂H₂-LITES Sensors

In conclusion, this paper presents a H2-enhanced LITES sensor for the first time. The mechanism for Q-factor enhancement was both theoretically and experimentally demonstrated by examining various gases with different densities and viscosities. The research revealed that gases with lower viscous damping and viscosity lead to reduced energy loss in the QTF. The Q-factor of a QTF can be improved by 2.5 times to 30,000 with H2-enhancement without introducing additional noise into the LITES sensor system. Based on the H2-enhancement effect, a self-designed round-head QTF with low f0 of 9527 Hz and a fiber coupled MPC with an optical length of 40 m were utilized to increase the energy accumulation time and optical absorption, respectively, to demonstrate an ultra-highly sensitive C2H2-LITES sensor. As a result, an MDL of 2.8 ppb for C2H2 detection was obtained with 800 ms integration time, corresponding to a normalized noise equivalent absorption (NNEA) coefficient of 5.8×10−11cm−1WHz−1/2. The Allan deviation analysis was developed to evaluate the stability of the H2-enhanced LITES sensor system. With an optimal integration time of 140 s, the MDL can be further improved to 290 ppt. Compared to other reported state-of-the-art C2H2-LITES techniques with similar parameters, this sensor shows a 241-fold improvement in the MDL. This H2-enhancement technique is illustrated as an excellent method to realize a QTF with a high Q-factor, which holds the merits of simplicity and effectiveness and has a great prospect in QTF-based gas sensing.