Methane (CH4), a primary component of natural gas, is a greenhouse gas that has a significant effect on the Earth’s climate^[1]. Anthropogenic CH4 sources include landfills, the oil and natural gas industry, agricultural activities, coal mining, combustion, wastewater treatment, and industrial processes. CH4 emission estimates are fraught with uncertainty^[2–4]. The oil and gas industry produces CH4 leakages, resulting in significant energy waste and unclear climate assessments. To ensure the security and integrity of the natural gas industry, a continual process must be established for searching, locating, and repairing CH4 leakage. On the other hand, meeting the criteria for large-scale deployment of CH4 detectors in industry driven by cost reduction and practicality is a considerable challenge. Thus, finding economical and feasible monitoring methods for CH4 leakage detection and repair operations has become urgent across the oil and gas industry^[5].

Passive optical gas imaging is affected by the differences in the background radiation of the surroundings, which can alter the signal-to-noise ratios of the received signals, and thus necessitates careful analysis by expert operators^[6,7]. The CH4 leakages monitoring equipment in factories should have high spatial resolutions and be capable of continuous and quantitative detection of the leakage, which ensures quick identification of the CH4 emitting points and quick repair of the leakage^[8–10]. Satellites have low spatial resolutions and are not capable of timely CH4 detection, and the algorithm presently used to quantify CH4 concentrations based on satellite observations is not sufficiently mature^[11,12]. Compared with passive optical methods, active laser-based methods are more effective in quantitative detection of CH4. The narrow linewidth tunable diode laser absorption spectroscopy (TDLAS) method has been used for quantitative CH4 leakage monitoring^[13,14]. The method, employed in laboratories for many years, is now employed in the industry with an open laser beam path and the environment replacing the measuring cell^[15–17]. However, TDLAS devices cannot simultaneously detect background target distances and CH4 concentrations and have restricted detection ranges because of low signal strength. Simultaneous measurement of the physical distribution of the leak and the gas concentration not only provides information on the location of the leak, but is essential in the actual calculation of the CH4 leak rate.

Integrated path differential absorption (IPDA) light detection and ranging (lidar) and differential absorption lidar (DIAL) are also two remote-sensing techniques that can be used to detect atmospheric gases^[18–20]. Although these devices are highly effective when used in large-scale surveys, their high costs and large sizes prevent them from being used for the continuous monitoring of individual CH4 emissions^[21]. The CH4 remote sensors that can autonomously and accurately locate CH4 leakage points and quantify the leakages are currently not available. The single-photon avalanche diode detector (SPAD) is a device that can respond at single-photon signal level. Currently, SPADs are employed for weak-signal imaging^[22–24]. Studies on the use of lidar monitoring systems employing SPAD for the detection of atmospheric aerosols and gases have facilitated the detection of low received signals^[25–29]. In order to adapt to the characteristics of CH4 leak monitoring equipment, SPAD can be applied to gas remote-sensing equipment so that remote CH4 spectroscopy and ranging can be performed using low power semiconductor diode lasers. The use of SPAD for gas detection has good prospects and advantages. However, only a few remote-sensing techniques using SPAD are available for gas concentration detection. The methods adopted to determine the signal characteristics of SPAD for gas concentration inversion are still immature and need further study. Gas remote monitoring using SPAD requires further study to improve industrialized CH4 leakage monitoring ability.

In this study, for current high CH4 concentration monitoring scenarios of leakage, a CH4 remote methodology was proposed based on a near-infrared SPAD, a continuous wave laser, and an electro-optical modulator (EOM). First, the signal modulation scheme, CH4 concentration inversion algorithm, and the structure of sensor are presented. Using an amplitude-modulated pseudo-random binary sequence (PRBS) and time-correlated single-photon counting (TSCPC), the sensor accomplishes the detection of background distance and CH4 concentration. Next, a series of experiments and field tests validate the proposed detection method and system performance. The developed CH4 remote method and verification sensor offer a new solution used for the observation of CH4 leakage.

Theoretical equations for the proposed method are shown in Supplementary Note 1 (see Supplementary Material). The proposed method is aimed at detecting the path-integrated column concentration (PICC) of CH4 released to the atmosphere by measuring the backscattered echoes from the hard target. First, CH4 absorption spectroscopy was obtained by using the approximate megahertz sweeping modulation frequency of the continuous laser. Next, the CH4 absorption spectroscopy was modulated by the PRBS, and the TSCPC method was adopted to ranging the background target. Near-infrared SPAD is sensitive to the single photons of light between 800 and 1700 nm, and can be used for 1653.7 nm wavelength CH4 detection. As shown in Fig. 1(b), the distributed feedback laser (DFB) is driven by the temperature and current unit, the output power is about 10 mW, and the wavelength modulation ranges from 1653.5 to 1653.85 nm. The rapid ramp sweep and M-sequence PRBS modulation signal are supplied by a signal generator (Tektronix, AWG 31252). The M-sequence modulation signal is applied to a commercial EOM (iXblue, MXER-LN-10) to encode the emitted laser. The modulated laser is converted into a parallel beam using a fiber port collimator. A half-wave plate is used to adjust the polarization direction of the emitted laser to match the polarizing beam splitter (PBS). A 10 cm gas cell (Ganwei Technology, GW-1000b-10 cm) is installed in the optical path to calibrate the sensor. The optical design uses individual coaxial optical transmitting and receiving channels integrated using the PBS, which maintains the required transmitting and receiving path alignments over long distances. The PBS, which has polarization selection characteristics, protects the SPAD from the laser reflections within the device. The received signal is obtained through the optical lens and converged to the SPAD and photon-counting unit. The M-sequence and continuously ramp sweeping laser were employed to determine the CH4 concentration and background target distance from the sensor. A hybrid signal comprising a high-frequency ramp wave (∼1MHz) and an M-sequence was generated and separately applied to the laser controller and the EOM to enable the modulation of the emitted laser. The M-sequence chip time Tc=10ns, corresponding to a bin rate of f=100MHz, with a distance resolution of Δd=c/2f=1.5m. The length of the M-sequence N=27−1=127 and the unambiguous distance d=Nc/2f=190.5m. The laser sweeping covered the CH4 absorption line, simultaneously encoded by the M-sequence. The theoretical schematic of transmitting and receiving signals sequence is shown in Fig. 1(a). The time cycle of the M-sequence was 127×10ns=1.27μs, corresponding to a frequency of 787.4 kHz. Because the M-sequence and ramp sweep signal were to be released simultaneously, the ramp sweeping period was set to 787.4 kHz. The inversion algorithm flow for ranging and CH4 concentration is shown in Supplementary Note 2 (see Supplementary Material).

The ramp sweep and the M-sequence modulation signals were simultaneously added. The time counting unit was used to collect the encoded signal to measure CH4 absorption. As shown in Fig. 2(a), the calibration system uses two gas cells with fiber-optic interfaces to detect CH4 absorption. The signal distributions of the modulation modes for various filled gas concentrations are displayed in Figs. 2(b) and 2(c). In Fig. 2(b), with only the ramp sweep signals added (NoPrbs mode), the pure N2 and 1000 (1 k), 2000 (2 k), and 10,000 (10 k) ppm (parts per million) of CH4 were flushed into a 3 m gas chamber for absorption testing. In the absorption band, the number of received photons decreases as the concentration of filled CH4 gas increases. In Fig. 2(c), the M-sequence and ramp modulation signals are simultaneously applied to measure the CH4 absorption (Prbs mode). As the gas concentration increases, the signal intensity of the peak M-sequence signal decreases and attains its minimum value at 1653.7 nm. The magnified subplot in the red box shows the CH4 absorption characteristics of different CH4 concentrations. The tests enabled the initial verification of CH4 gas-absorption characteristics of the SPAD. The absorption signal obtained by the SPAD differs from that produced by a linear mode detector, which is prone to jitter and deviation owing to factors such as its dead time (DT), which will be discussed in the next section.

As is shown in Fig. 3, the CH4 concentrations and SPAD’s input powers were changed for absorption signals testing and validation. CH4 concentrations of 2000 and 10,000 ppm were flushed into a 3 m gas chamber for absorption testing. The input laser power was adjusted using the attenuator. P₁, P₂, and P₃ represent different input laser powers, and the energy magnitude relationship is P₁<P₂<P₃. The solid lines represent the transmittance curves for different DTs at 2000 ppm CH4 concentration, and the dashed lines represent the 10,000 ppm concentration. Currently used SPAD can adjust the setting of the DT parameter according to the actual applied process. The effect of DT on the CH4 absorption was also analyzed for both NoPrbs and Prbs modes under the same input laser power. The results show that the values of the CH4 absorption intensity of the two modulation modes are closely related to the DT for different input laser powers. However, as the DT increased from 10 to 200 ns with the same CH4 concentration, the CH4 absorption intensity tended to increase, and the peak signal amplitude tended to decrease. The DT setting affects the actual CH4 absorption signal. The reason for this may be that a shorter DT results in increased noise interference such as SPAD’s after-pulses, leading to distortion of the signal, whereas as the DT setting grows, the distortion of the signal corresponding to the absorption spectra decreases, and thus improves the signal to a certain extent. Thus, when using the SPAD for accurate CH4 concentration detection, the DT effect should be fully considered. The gas absorption signal should be calibrated under different DTs, and the appropriate parameter should be selected for CH4 detection.

After the laboratory calibration and signal testing, the developed single-photon CH4 remote system was used for outfield signal verification experiments. As shown in Fig. 4(a), the experiment was performed in a laboratory hallway by placing the sensor on a movable cart to detect a long-range reflective target in the hallway. The target could be moved to vary the distance between it and the sensor. The 15 and 29 m were selected as the two hard target distances for the signal verification test. The theoretical distance of the hard target was calibrated using a laser tachymeter. Figure 4(b) shows the physical composition diagram of the sensor. A small 10 cm pass-through-type gas cell was mounted on the transmitter optical path for the calibration of CH4 absorption spectroscopy. The components used in the remote sensor are shown in Fig. 1(b). The SPAD and other supporting instruments signal was controlled by a PC. Raw echo signals received at the hard target distances 15 and 29 m are shown in Figs. 5(a) and 5(b). The blue solid line represents the initial bipolar signal, and the purple solid line represents the received original signal. The number of shifted bins can be obtained by calculating the cross-correlation between the received signal and the reference bipolar sequence. The red dashed line corresponds to the initial position of the emitted signal of two distances, which corresponds to 15- and 25-bin shifts, respectively, in agreement with the position calculated in Fig. 5(c). In Fig. 5(c), each horizontal coordinate bin represented a theoretical distance resolution of 1.5 m; 15 m corresponds to a moving bin number of 15, which represents a detection distance of 22.5 m with a bias. In addition to the distance detected by the sensor, a distance bias will be caused by the fiber optic and the signal data line, which introduces a fixed distance bias. Thus, the distance calibration is required in practical applications. The 29 m hard target signal corresponded to a rightward shifting bin count of 25, 10 bins short of the 15 m. The calculated distance difference was 15 m, and the actual distance was 14 m. The distance error was within 1 m, which is within the allowable error range. Thus, the distance performance of the system was verified by the cross-correlation results.

After the CH4 absorption signal is recovered by shift operation, the recovered signal can be utilized for CH4 absorption spectroscopy calculations and Lorentz fitting. In Fig. 5(d), the original CH4 signal is recovered using the shift calculations. The signal was calibrated by flushing 10,000 ppm of CH4 into a 10 cm optical range gas cell, and the corresponding PICC is 1000 ppm · m. The gas absorption signals recovered after the shift operation at 15 m target distances were calculated, including the pure N2 without absorption and the 1000 ppm · m CH4 concentration. By extracting the peak N2 signals and the peak signals of CH4, the CH4 absorption curves were obtained. As shown by the solid blue line, the absorption spectra of CH4 can be obtained using the Lorentz fitting. Experimental results validate the proposed CH4 absorption spectroscopy modulation methods. Then, in Fig. 5(e), the measured Lorentz-fitted CH4 absorption spectrum (1k15m) were compared with the theoretical CH4 absorbance spectrum (1kT). The two spectral lines have small differences and good consistency, which initially validate the proposed CH4 measurement method.

In this study, a remote method to detect CH4 spectra and background target distance was established using a near-infrared SPAD and a continuous DFB laser. A superimposed M-sequence and ramp laser modulation was utilized for emitted laser encoding, which in turn enabled simultaneous target ranging and CH4 PICC measurement by the TSCPC method. The influence of the SPAD’s DT on CH4 absorption spectra was discussed. Laboratory calibration of the proposed method was performed for different CH4 concentrations. The intensity of CH4 absorption spectra increases with increasing DT. Thus, when using the SPAD for accurate CH4 concentration detection, the DT effect should be fully considered. Then, through field tests, the hard target distance was obtained. The acquired CH4 absorption spectra were compared and analyzed with the theoretical absorption spectra, which demonstrated consistent results.