Oxygen sensors play a key role in various applications, including industrial process control, medical devices, and biomanufacturing. In these contexts, high sensitivity detection of low concentration oxygen is essential for tasks such as monitoring residual oxygen in syringe vials, controlling combustion processes, tracking metabolic processes, and ensuring air quality in food packaging. Current trace oxygen sensors primarily utilize electrochemical, zirconia, and paramagnetic technologies. Electrochemical sensors are the most widely used due to their low cost and high precision, but they suffer from long response time and limited lifetime. While zirconia and paramagnetic sensors offer longer operating lives, their high operating temperatures and costs restrict their broader application. Therefore, developing trace oxygen sensors that provide long lifetimes, low costs, fast responses, and high detection limits at room temperature is of significant practical importance.

We develop a compact, high-sensitivity oxygen (O₂) detection system using the quartz-enhanced photoacoustic spectroscopy (QEPAS) technique and an on-beam resonator tube configuration. A high-power tunable diode laser with a center wavelength of 763 nm is used in combination with wavelength modulation spectroscopy to detect O₂, and we investigate the influence of water vapor (H₂O) on O₂ molecular relaxation. We also explore the linearity, stability, and minimum detection limit of the system.

To optimize the second harmonic photoacoustic signal, we adjust the wavelength modulation depth of the QEPAS system. Figure 4(a) shows the relationship between photoacoustic signal intensity and the modulation amplitude. From Fig. 4(b), the oxygen photoacoustic amplitude reaches a maximum of 2.24 V at a modulation amplitude of 55 mV. We determine this modulation depth as optimal and conduct subsequent experiments under these conditions. Oxygen has a low relaxation rate, while water vapor exhibits a high relaxation rate, making H₂O an effective catalyst for enhancing relaxation. The blue curve in Fig. 5 illustrates the measured O₂ photoacoustic signal amplitude as a function of volume fraction of water vapor, while the red curve represents the fit of experimental results using Eq. (2), yielding a fitting coefficient of 0.97. We observe a significant positive influence of H₂O on the O₂ photoacoustic signal, particularly at 0 to 0.8% volume fraction of water vapor. Beyond this range, the O₂ signal stabilizes (mean value: 1.75 V, standard deviation: 0.01 V). Figure 5 also indicates that the photoacoustic signal of O₂ at 8% volume fraction of water vapor is enhanced by a factor of 7 compared to that at dry conditions. In the QEPAS technique, a good linear relationship exists between photoacoustic signal amplitude and gas volume fraction. Figure 6(b) depicts the relationship between the QEPAS signal for O₂ and volume fraction of O₂, with a linear fitting coefficient of R²=0.999, confirming theoretical expectations. To evaluate the minimum detection sensitivity of this QEPAS system, we set the integration time of the lock-in amplifier to 100 ms and the fading frequency to 24 dB. We measure the photoacoustic signal of O₂ with a volume fraction of 0.21, as shown in Fig. 7, resulting in a peak-to-peak value of 2.27 V, a noise level of 0.01 V, and a signal-to-noise ratio of 227, which corresponds to a minimum detection limit of 9.25×10⁻⁴. The normalized noise equivalent absorption coefficient of the system is 4.70×10^-9 W·cm^-1·Hz^-1/2. Employing lasers with higher output power and stronger absorption line strength can further improve the minimum detection limit. To assess the long-term stability of this sensing system and determine its detection limit, we analyze the oxygen photoacoustic signal amplitude over an 8 h period, as shown in Fig. 8(a). A systematic Allan variance analysis reveals that the optimal integration time is 925 s, resulting in a detection limit for the oxygen QEPAS system of 5.5×10^-⁴.

We develop an O₂ sensor based on the quartz-enhanced photoacoustic spectroscopy (QEPAS) technique and an on-beam resonator tube configuration, utilizing a TO-packaged DM 763 nm light source. The absorption spectrum of O₂ at 13114.10 cm^-1 (762.54 nm) with a line intensity of 3.884×10^-24 molecule^-1·cm is selected. We optimize the modulation depth of the sensor, identifying 55 mV as optimal. The influence of H₂O molecules on the O₂ relaxation rate is investigated, and we establish the relationship between O₂ photoacoustic signals and H₂O volume fraction. At a water vapor volume fraction of 8%, the O₂ photoacoustic signals are enhanced by a factor of 7 compared to dry conditions. We calibrate the detection system with varying O₂ volume fraction, obtaining a linear fitting coefficient of 0.999. The system achieves a detection limit of 9.25×10^-4 at an integration time of 12.5 s, with a normalized noise equivalent absorption coefficient of 4.70×10^-9 W·cm^-1·Hz^-1/2; at an optimal integration time of 925 s, the detection limit is 5.5×10^-4. The QEPAS oxygen sensor developed in this study features a compact structure and high sensitivity, meeting the demands for O₂ gas detection in applications such as combustion process control, metabolic process monitoring, and food atmosphere packaging.