Injecting radio frequency (RF) white noise to the current driving of the laser can broaden the laser emission linewidth and efficiently suppress cavity-mode noise in off-axis integrated cavity output spectroscopy (OA-ICOS). The effect of the injected RF noise level on the cavity-mode noise and the deformation of the absorption line shape in off-axis integrated cavity output spectroscopy (OA-ICOS) with a distributed feedback laser (DFB) at 1.65 µm were investigated. We measured methane at different concentrations between 0.1 ppmv and 2 ppmv associated with a −20 dBm RF noise injection. A linear spectral response of the intensity of the cavity output spectra with the CH4 concentration was observed. A threefold improvement in the detection limit was achieved compared to unperturbed OA-ICOS. The response time of the improved OA-ICOS system is about 30 s and the minimum detectable concentration (MDC) of CH4 is 7.6 ppbv, which corresponds to a minimum detectable fractional absorption scaled to the path length of 7.3 × 10−10 cm−1. The noise equivalent absorption sensitivity of the system is 5.51 × 10−9 cm−1Hz-1/2.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
The ocean is an important source and sink of various gases in the atmospheric environment, such as CO2 and CH4. It is of considerable importance to detect the concentration of dissolved gases in seawater. In particular, the detection of dissolved methane (CH4) in seawater is used to study marine ecology and climate but can also help to detect submarine combustible ice resources. Fukasawa et al. [1–3] developed a methane sensor for the measurement of dissolved methane in water (METS, CAPSUM Technologie GmbH). The detection range of CH4 is from 50 nmol/L to 10 μmol/L. Using Raman spectroscopy and caged remotely operated vehicles (ROV) systems, Zhang et al. developed a Deep Ocean Raman In Situ Spectrometer (DORISS) with a detection accuracy of 1 mmol/L . Given the latest developments in membrane degassing technology, direct absorption spectroscopy is now widely used in the detection of trace gases dissolve in water [5–10]. Tsunogai et al.  developed an in situ methane analyzer based on infrared (IR) spectroscopy around 3.3 μm and using a gas-permeable membrane tube with a CH4 concentration range of 40 ppm (25 nmol/L) to 320 ppm (200 nmol/L) in the gas phase. The Deep-Sea Gas Analyzer from Los Gatos Research, Inc. uses a gas-permeable membrane and off-axis integrated cavity output spectroscopy (OA-ICOS) to achieve a detection range between 1 and 500 ppm and a response time of 5 min [6–8]. Currently, off-axis integrated cavity spectroscopy technology has great potential for high-precision in situ measurement of marine methane. Further research on this technology is of great significance in improving the accuracy and response time of dissolved gases in water.
OA-ICOS, or off-axis cavity-enhanced absorption spectroscopy (OA-CEAS) , usually consists of two high-reflectivity mirrors (>99.9%) where the light can bounce up to thousands of times to provide a long optical absorption path. It is expected to have a low detection limit for methane dissolved in water because of its high sensitivity and fast response time, which are limited by the semi-permeable membrane [6–10]. OA-ICOS relies on the excitation of a dense spectrum of transverse-longitudinal cavity modes in the cavity combined with the time averaging of the total signal transmitted through the optical cavity. Unlike in cavity ring-down spectroscopy (CRDS),the OA-ICOS technique does not require sophisticated optical and electronic components for optical mode matching and coupling of single frequency laser to the optical cavity modes. It makes OA-ICOS very suitable for real-time and continuous in situ monitoring of trace gases [12,13]. In the OA-ICOS approach, a narrow-band laser is used and is coupled “off-axis” into the cavity at an angle with respect to the optical axis of the cavity. Such off-axis visually non-resonant cavity results in a significant reduction of the cavity free spectral range (FSR). The laser beam is not only coupled into the cavity through the fundamental longitudinal cavity modes but also through high-order longitudinal cavity modes and a large number of transverse cavity modes. This makes the energy concentrated on the fundamental modes (in on-axis cavity) evenly distributed to the higher-order modes (in off-axis cavity), which will lead to the gradual disappearance of the lower order mode, and the FSR tends to zero (as shown in Fig. 3(a) and Fig. 4(a)). Light of any frequency can pass through the cavity when the FSR approaches zero, which means that the cavity mode disappears, ie the output signal fluctuations (“cavity mode noise”) are suppressed.
To achieve a high signal-to-noise ratio (SNR) in OA-ICOS absorption spectra, or equivalently a low detection limit, the main issue is to minimize the fluctuations in the spectral baseline resulting from the residual cavity-mode noise that is the ultimate sensitivity-limiting factor. Very large cavity mirrors (diameter ≥ 50 mm) [12–14] can be used to increase the off-axis angle leading to the excitation of more transverse-longitudinal modes of different orders and improve the suppression of cavity-mode noise. However, this results in a large optical cavity, which requires a larger gas sample, more space, and a detector with larger photo surface. In particular, an optimum trade-off must be evaluated between the cavity mirror size and the detector performance due to the degradation of the sensitivity for the very large detector. Besides, the degassing efficiency of a membrane-based water-gas separation system is limited mainly by the properties of the membrane materials [15,16]. In our study, the gas flow rate of the water-gas separation is typically 4 ml/min, and therefore the cavity volume cannot be too large, which makes it unrealistic to use large-size cavity mirrors to reduce the cavity mode noise. In our case, it is obvious that the cavity mode noise cannot be suppressed by increasing the mirrors size and cavity volume. We need to find some other methods to suppress cavity mode noise .
Alternative methods for the removal of cavity mode noise include adjusting the cavity length using a piezoelectric device [14,18] or modulating the laser current [20,21] However, both methods are accompanied by adverse effects, such as fringing from the piezoelectric modulation of the cavity length additional noise, and a broadened absorption spectrum from the varying laser current [18–20].
In principle, the basic idea is to achieve a cavity FSR lower than the injected laser linewidth so that the laser beam can be non-resonantly coupled into the cavity. Then, the transmission spectrum from an empty cavity becomes a frequency-independent baseline as in a conventional multi-pass cell. When the laser linewidth is widened, its coherence will be weakened, so that there will be no interference effect even if the beams coincide. In this case, the FSR is also tends to zero to make the optical cavity equivalent to a multi-pass cell, and the output signal is considered continuous. Engel et al.  suggested broadening the laser linewidth wider than the cavity FSR by injecting filtered white noise to the laser current to reduce the spurious coupling of the longitudinal cavity modes. Ciaffoni et al.  first implemented this method to improve the SNR of an OA-ICOS setup for oxygen measurement using a 764-nm vertical cavity surface emitting laser (VCSEL). When RF noise (1-1500 MHz, - 174 dBm) filtered with a 30 MHz low-pass (LP) filter was added to the laser current, a four-fold gain in the SNR was obtained leading to a minimum detectable absorption (MDA) of 4.3 × 10−5 Hz-1/2 in 1 s on average. Katherine et al.  demonstrated the performance improvement of an OA-ICOS with a quantum cascade laser for CO2 absorption at 1890 cm−1 through the injection of RF perturbations (50-1500 MHz) to achieve a tenfold improvement from the unperturbed regime.
In our recent work , RF white noise was added to a near-IR (1653 nm) DFB laser to improve the performance of a wavelength modulated OA-ICOS system (WM-OA-ICOS), and more than six-fold improvement in detection limit is achieved. In the present work, we improved the performance of an ICOS instrument using an RF white noise perturbation. The impact of the RF noise power on the cavity modes and the absorption spectra were investigated to obtain the optimal RF noise level that minimizes the deformation of the absorption line and improves the SNR. The experiments were carried out on both on-axis ICOS and off-axis ICOS. Additionally, an RF white noise of −20 dBm filtered by a 70 MHz LP filter was selected to evaluate the stability and the precision of the OA-ICOS instrument developed.
2. Materials and methods
2.1. Concentration retrieval in ICOS
In an OA-ICOS approach based on a double-mirror cavity, the absorption coefficient of the target molecular trace gas can be expressed as [25,26]:
Where I0 is the incident laser intensity and I is the output intensity of the cavity. d is the distance between two cavity mirrors in cm, R is the mirror reflectivity (assuming the same reflectivity for both mirrors: R1 = R2 = R), and α(ν) = N × σ(ν) absorption coefficient with N the number of absorbing molecules in molecules/cm3 and σ(ν) the frequency-dependent absorption cross-section in cm2/ molecule.
The molecular number concentration N can then be expressed as:
The molecular concentration N is therefore proportional to . The dimensionless mixing ratio of the trace gas χ can be obtained as:
Where NL = 2.6868 × 1019 molecules/cm3 is the Loschmidt number at Tref = 273.15 K and P0 = 760 Torr, T is the working temperature in K, and P is the working pressure in Torr.
2.2. ICOS setup
The experimental ICOS setup used is shown in Fig. 1. A pair of 1”-diameter spherical mirrors (Advanced Thin Films) with a reflectivity of 99.999% (for 0°-5°) and a curvature radius of 1000 mm was used to form a linear optical cavity. The mirrors were separated by a stainless steel tube with a distance of 28 cm between both mirrors to yield and an FSR of 535 MHz, as calculated by a previously published method . The inner cavity diameter was 10 mm and the volume of the cavity was approximately 22 cm3.
An optical cage system (30 mm, Thorlabs) using four rigid steel rods to mount all optical components, including the collimator, the focusing lens, and the photodetector, along the optical cavity axis was integrate into the optical cavity, as shown in Fig. 1(a). An adjustable aspheric collimator (Thorlabs, CFC-8X-C) was mounted before the cavity using a threaded kinematic mount with a slip plate (Thorlabs, KC1-S) that provides ± 4° tip/tilt along the optical axis and ± 1 mm of X and Y adjustment perpendicular to the optical axis for the laser injection. The adjustable collimator was used to adjust the match between the laser mode and the cavity mode. The length from optic collimator to the photodetector (including the collimator and photodetector) is approximately 45 cm with an external diameter of about 6.6 cm.
A distributed feedback (DFB) laser (NEL, NLK1U5EAAA) operating near 1653.74 nm (at 20 °C) with a linewidth of about 2 MHz and a power of 12 mW was used to measure the methane (CH4) concentration. The laser was powered by a laser driver (Stanford Research Systems, LDC 501) combined with a function generator (RIGOL, DG4162) for the laser wavelength scan. A power-adjustable RF noise source (Shenzhen Dakofeng Technology Co., Ltd. NF-1000) with an output frequency range of 5 MHz-1.5 GHz was first filtered with a 70 MHz bandwidth LP filter (Crystek Corporation, CLPFL-0070-BNC), then injected into the DFB laser through a bias tee (Thorlabs, LM14S2-BT) to broaden the laser linewidth [21–24]. The adjustable range of the RF noise power was from −40 dBm to −10 dBm. The spectral lines near 6046.96 cm−1 (mainly 002 0F2←000 01A band) are commonly used as target lines for measuring CH4 . An InGaAs photodiode (Thorlabs, FGA 10) was used to measure the output signal from the cavity. The detected signal was then connected to a data acquisition card (National Instruments, NI 6210).
The relationship between the laser current and the emission wave number was calibrated at 20 °C using a wavelength meter (Bristol Instruments Inc. 621B). The calibration curve was then used to convert the experimental “Row number” to a “Wavenumber” for the spectral calibration.
In the marine environment, higher depths have a lower water temperature, from 30 to −2 °C, and higher pressure. Such broad temperature and pressure changes have a significant impact on the accuracy of absorption spectroscopy. Therefore, a temperature control box was used to provide a stable working environment for optical detection. The pressure and the temperature in the cavity were controlled and stabilized using a home-made electronic card (PT Card) via two proportional valves (Clippard, EV-PM-10-1325), a pressure sensor (MEAS, US266-000006-002PA) and a temperature sensor (Heraeus, M222 PT100), as shown in Fig. 1. The pressure and the temperature in the cavity were controlled and stabilized using a home-made electronic card (PT Card) via two proportional valves (Clippard, EV-PM-10-1325), a pressure sensor (MEAS, US266-000006-002PA) and a temperature control box, as shown in Fig. 1. The temperature control box consists of a temperature sensor (Heraeus, M222 PT100), a heating plate, a radiator, and a fan. The heating plate heats the surrounding air through the radiator. A fan is used to uniformly transfer the heat to the optical setup. The temperature sensor feeds back the temperature of the cavity in real-time, and the PT-Card controls the working efficiency of the heating plate according to the set working temperature and the actual temperature of the cavity.
A gas mixing system (Environics. Inc, Series 4000) and a 2.0 ppmv CH4 reference cylinder were used to prepare various CH4 concentrations required for the experiment.
3. Results and discussion
3.1. Power-adjustable RF white noise source
We experimentally optimized the bandwidth of the used LP filter. Seven LP filters with bandwidths in the range of 15 - 400 MHz have been tested. 70 MHz was selected because of its significant capacity of suppressing cavity-mode noise while without introducing any distortions. The performance of the RF white noise source filtered by the LP filter was investigated first. Figure 2 shows that the RF noise power after the filter was about −55 dBm, −45 dBm, −35 dBm, and −25 dBm within 1-70 MHz, respectively. The laser linewidth is widened by RF white noise to 81.8 MHz (−40 dBm), 124.4 MHz (−30 dBm), 271.9 MHz (−20 dBm), and 855.4 MHz (−10 dBm), respectively.
3.2. Reduction of the cavity-mode noise by RF white noise in on-axis ICOS
In an on-axis ICOS approach, the laser beam is coupled into the cavity along the optical axis of the cavity. The effect of the RF noise power (from −10 dBm to −40 dBm) on the suppression of cavity-mode noise was studied. The on-axis cavity-mode spectra were measured by scanning the laser using a triangular signal with an amplitude of 100 mV at 1 Hz, which provides a wavenumber range from 6046.83 cm−1 to 6046.9 cm−1. Under the same conditions, the on-axis cavity-mode spectra were measured without and with noise injection at different noise powers of −40 dBm, −30 dBm, −20 dBm, and −10 dBm.
Figure 3 shows the effect of the RF noise on the reduction of the cavity-mode noise amplitude in the on-axis ICOS approach developed. Figure 3(a) plots a time-series measurement of the cavity modes without (red line) and with RF noise injected at different powers of −40 dBm, −30 dBm, −20 dBm, and −10 dBm. The spacing between the fundamental modes is 517.9 MHz, which is close to the theoretical FSR value of 535 MHz . When −40 dBm RF noise was injected in the DFB laser, the second-order cavity mode disappeared, and the fundamental mode spacing remained unchanged. After −30 dBm RF noise was injected, the fundamental mode and first-order mode amplitude were significantly reduced, and many visible higher-order modes appeared around them. When −20 dBm RF noise was injected, the first-order mode disappeared as well, and the fundamental mode almost disappeared whereas higher-order modes appeared. When the RF noise was increased to −10 dBm, no distinct cavity mode was visible. It illustrates that the effect of −40 dBm RF noise on the on-axis cavity modes is negligible and the suppression of the cavity-mode noise becomes detectable for a −30 dBm noise injection. A significant reduction of about 33.5 dB of the cavity-mode noise amplitude was obtained with a −10 dBm RF noise perturbation (Fig. 3(b)). In the present work, increasing the RF noise power higher than −10 dBm did not further reduce the cavity-mode noise amplitude, whereas the shape of the absorption line was affected by the RF noise injection, particularly at higher RF noise power, in agreement with previously published reports [22,23].
The amplitude spectra of the cavity-mode noise were obtained through a Fourier transformation, as shown in Fig. 3(b). The −40 dBm RF noise reduces only slightly the high-frequency noise (> 1000 Hz) and has almost no effect on the low-frequency noise (< 800 Hz). A −30 dBm RF noise significantly eliminates cavity-mode noise at frequencies higher than 300 Hz from −42.5 dB (red line) down to −59.6 dB (yellow line), or a 17.1 dB reduction, and has less effect on the cavity-mode noise at frequencies lower than 300 Hz, which means the lower-order modes or the fundamental longitudinal modes are not entirely suppressed. Similarly, the −20 dBm RF noise drastically reduces the noise at frequencies higher than 150 Hz by about 19.8 dB, and the −10 dBm RF noise provides a very efficient elimination of the cavity-mode noise below 12000 Hz by about 25.3 dB, which leads to a full suppression of the cavity-mode noise (Fig. 3(a)).
2.0 ppmv of CH4 was filled into the on-axis ICOS cavity at 60 Torr to study the effect of the RF noise effects on the CH4 absorption spectra. The on-axis ICOS output spectra without and with RF noise perturbation were measured under the same conditions (scanning the laser using a triangular signal with an amplitude of 1.6 V at 10 Hz). The results are shown in Fig. 4. The plotted curves are averaged spectra (n = 1000) of the original signal from the detector.
The red trace in Fig. 4(a) is the output signal without noise perturbation. The following curves are the output signals obtained after the injection of −40 dBm, −30 dBm, −20 dBm, and −10 dBm RF noise into the laser current. We obtained the baseline intensity I0 by fitting the cavity output signal that does not contain the portion of the absorption feature, and the absorption spectra (Fig. 4(b)) was calculated using (I0/I-1). As can be seen in Fig. 4(a), without any RF noise injection, the CH4 absorption peak is almost indiscernible from the cavity-mode structured noise even when it is averaged over 1000 times. The injection of −40 dBm RF noise does not improve the absorption signal. The absorption signal becomes visible when the RF noise power is higher than −30 dBm. The cavity-mode structured noise disappears with −10 dBm RF noise injection and the absorption peak is obtained with a high SNR of about 81. The SNR is obtained by dividing the peak of the absorption spectra by the standard deviation of a baseline (from 6074.10 cm−1 to 6047.2 cm−1) of the absorption spectra.
Our results show that an appropriate RF noise perturbation efficiently suppresses the cavity-mode noise and increases the SNR of the absorption peak in on-axis ICOS.
3.3. Reduction of the cavity-mode noise by RF noise in off-Axis ICOS (OA-ICOS)
The laser beam is off-axis coupled to the ICOS cavity. We present the improvement of the performance of an OA-ICOS setup using RF noise perturbation. The off-axis cavity-modes without and with RF noise perturbations are compared in Fig. 5(a). Figure 5(a) shows that the cavity modes with the −40 dBm RF noise perturbation (blue line) are still very similar to the original cavity modes (red trace) without RF noise injection. The amplitude was only reduced by a factor 2. The FSR of the cavity modes without and with −40dBm RF noise is 6.4 MHz and 6.8 MHz, respectively. When the RF noise power increases, distributed cavity modes are no longer observed. A mode-noise-free spectral baseline is obtained for an RF noise power higher than −30 dBm, similarly than with conventional multi-pass cells. The Fourier analysis of the time-series cavity modes of Fig. 5(a) was performed. The spectral amplitudes expressed in dB were plotted in Fig. 5(b). The noise caused in the electrical signal by the cavity modes is mainly located at 1983 Hz, 4300 Hz, and 6300 Hz. The peak amplitudes of the cavity-mode noise at these frequencies were reduced from −17.8 dB to −24.7 dB, −19 dB to −31.9 dB, and −20.4 dB to −35.1 dB, respectively, by the injection of RF noise at −40 dBm. The cavity-mode noise can be efficiently suppressed with an RF noise power higher than −30 dBm. The injection of −30 dBm, −20 dBm, and −10 dBm RF noise results in a reduction of the cavity-mode noise by about 23.4 dB, 28.8 dB, and 33.6 dB, respectively, compared to the situation without RF noise injection.
The OA-ICOS absorption spectra of 2.0 ppmv CH4, “perturbed” by different RF noise powers, are shown in Fig. 6. Figure 6(a) represents the CH4 absorption spectra from1000 averaged original signals from the detector. For a higher RF noise power, the baseline of the absorption signal is smoother and the SNR is higher. The corresponding spectral term (I0/I-1) is shown in Fig. 6(b). The impact of the RF noise level on the shape of the CH4 absorption line is evaluated through the SNR of the spectra and the broadening of the absorption line that is related to the line shape deformation. The absorption spectra are fitted to Vogit profiles to determine the peak height and the linewidth of the absorption spectra. These parameters are summarized in Table 1.
The standard deviation of the baseline was obtained by statistical analysis of the baselines in the range of 6047.10 cm−1 to 6047.20 cm−1. The shapes of the absorption spectra under a perturbation by −40 dBm and −30 dBm RF noise coincide well with the absorption spectra without any RF noise perturbation where no deformation of the molecular absorption line shape occurs. The −20 dBm RF noise perturbation results in the highest SNR and a weak deformation of the line shape, which is characterized by a smaller peak value and a wider full-width at half-maximum (FWHM). The absorption spectrum with a −10 dBm RF noise perturbation is severely deformed compared to the absorption spectrum without any RF noise injection. The absorbance peak height is significantly reduced from 0.644 to 0.480, and the spectral FWHM increases from 0.0414 to 0.0541 cm−1. However, the integrated area of all absorption spectra and the effective optical path lengths are notably similar. It suggests that the deformation caused by the RF noise does not influence the integrated absorbance, as discussed by Yalin .
We showed that a −20 dBm RF noise perturbation efficiently suppresses the off-axis cavity-mode noise and significantly improves the SNR in the absorption spectrum while not significantly deforming the absorption spectral profile. It is considered that −20 dBm is the best RF white noise power for improving the signal of the OA-ICOS system we used in this paper.
3.4. Measurement of CH4 trace concentrations
Based on our results, the off-axis ICOS setup and a −20 dBm RF noise perturbation were used for the quantitative measurement of CH4 concentrations. 14 groups of different CH4 concentrations (0.1 ppmv, 0.2 ppmv, 0.3 ppmv, 0.4 ppmv, 0.5 ppmv, 0.6 ppmv, 0.7 ppmv, 0.8 ppmv, 0.9 ppmv, 1.0 ppmv, 1.2 ppmv, 1.5 ppmv, 1.7 ppmv, and 2.0 ppmv) were prepared by diluting 2.0 ppmv CH4 from a reference cylinder with nitrogen using a commercial gas mixing system (Environics. Inc, Series 4000). The pressure in the cavity was controlled at 38 Torr and the temperature was controlled at 30 °C. The x-axis is the peak height of the absorption spectrum from a fit and the Y-axis is the expected concentration.
Figure 7 shows the evolution of the CH4 concentration with the absorption term (I0/I-1). A linear relationship was obtained with a correlation coefficient R2 = 0.99985:
This result indicates that the injection of −20 dBm RF white noise does not affect the proportional relationship between the peak height and the concentration. This equation is only valid for the low CH4 concentration regime.
In addition, we measured the response time of the system for an inlet speed of 4 ml/min by repeatedly switching the injection gas concentration between 1.0 ppmv and 2.0 ppmv. Figure 8 shows that the response time of the OA-ICOS system to the changes of the gas concentration is about 30 s. Since the gas concentrations used in this experiment was generated from a gas mixing system that needed about 3-5 s to prepare a new concentration, the actual response time of the OA-ICOS system can be estimated at 27 s.
Time-series measurements of 2.0 ppmv CH4 were performed to evaluate the stability and the precision of the OA-ICOS instrument developed without and with a −20 dBm RF noise perturbation. The signals from the detector were averaged 10 times and used to determine the CH4 concentration. The spectral data sampling rate was 0.5 Hz. The test time for each method was approximately 2.2 hours, and the experimental results are summarized in Fig. 9.
The black line corresponds to the OA-ICOS system without any RF noise and the red line represents the system with a −20 dBm RF noise perturbation. The injection of a −20 dBm RF white noise allowed significantly reducing the fluctuations (the difference value between the maximum and the minimum, D-value) in the measured concentration was reduced from 623.5 to 195.6 ppbv. The standard deviation (SD) of the measured concentration was reduced from 102.6 ppbv to 37.0 ppbv, and the corresponding relative measurement accuracy was reduced from 4.9% to 1.8%, leading to an improvement of almost a factor 3.
The Allan variance analysis (Fig. 9(b)) confirms the improvement from the injection of RF noise. At the optimum integration time about 1500 s, the OA-ICOS set up provided an Allan deviation of 2.4 ppbv without any RF noise injection that was reduced to 1 ppbv with the injection of the −20 dBm RF noise. As shown in Table 2, the minimum detectable concentration (MDC) was 16.5 ppbv for a 10-s integration time and 7.6 ppbv for 1-min integration time. The corresponding minimum detectable absorption coefficient were 1.7 × 10−9 cm−1 and 7.3 × 10−10 cm−1, respectively. Using the method described by Moyer et al. , the two noise equivalent absorption sensitivities (NEAS) were calculated according to the integration time and the corresponding minimum detectable absorption coefficients, which were 5.38 × 10−9 cm−1Hz-1/2 and 5.65 × 10−9 cm−1Hz-1/2, respectively. And the average of the two value 5.51 × 10−9 cm−1Hz-1/2 was taken as the NEAS of the system.
In conclusion, the introduction of an RF white noise perturbation into the laser drive current effectively suppresses the cavity mode noise and improves the measurement sensitivity of the ICOS instrument. Higher powers of the injected RF noise yield a lower cavity-mode noise. However, the injection of RF noise also reduces the absorption peak and broadens the linewidth of the absorption line. A trade-off must be made between an efficient suppression of the cavity-mode noise and the lowest distortion of the spectral line shape. In the present work, a −20 dBm RF white noise was used for both on-axis and off-axis ICOS experiments. It contributed to a threefold improvement of the detection limit. Methane concentrations ranging from 0.1 ppmv to 2 ppmv were measured using the −20 dBm RF noise injection and a linear response between the cavity-enhanced absorption intensity and the gas concentration was obtained. The response time of the improved OA-ICOS system is about 30 s. The MDC for CH4 is 7.6 ppbv when the integration time is 1 min, and the corresponding minimum detectable absorption coefficient is 7.3 × 10−10 cm−1. The NEAS of the system is 5.51 × 10−9 cm−1Hz-1/2.
This work demonstrates a significant improvement in the detection limit of an OA-ICOS system using an RF white noise perturbation. It will further promote the development of OA-ICOS gas sensors for various field applications.
National Key Research and Development Program of China (2016YFC0303900 and 2017YFC0209700); National Natural Science Foundation of China (NSFC) (41730103).
The authors declare no conflicts of interest.
1. T. Fukasawa, S. Hozumi, M. Morita, T. Oketani, and M. Masson, “Dissolved methane sensor for methane leakage Monitoring in methane hydrate production,” in OCEANS 2006 (IEEE, 2006), pp. 1–6.
2. T. Fukasawa, T. Oketani, M. Masson, J. Groneman, Y. Hara, M. Hayashi, “Optimized METS Sensor for Methane Leakage Monitoring,” in OCEANS 2008 - MTS/IEEE Kobe Techno-Ocean (IEEE, 2008), pp, 1–8.
3. C. Boulart, M. C. Mowlem, D. P. Connelly, J. P. Dutasta, and C. R. German, “A novel, low-cost, high performance dissolved methane sensor for aqueous environments,” Opt. Express 16(17), 12607–12617 (2008). [CrossRef] [PubMed]
4. X. Zhang, K. C. Hester, W. Ussler, P. M. Walz, E. T. Peltzer, and P. G. Brewer, “In situ Raman - based measurements of high dissolved methane concentrations in hydrate - rich ocean sediments,” Geophys. Res. Lett. 38, L08605 (2011). [CrossRef]
5. U. Tsunogai, S. Kou, H. Kimoto, and T. Kimoto, “Development of in-situ methane sensor in seawater,” in 51th Annual Meeting of the Geochemical Society of Japan, 23 February 2007.
6. S. D. Wankel, Y. W. Huang, M. Gupta, R. Provencal, J. B. Leen, A. Fahrland, C. Vidoudez, and P. R. Girguis, “Characterizing the Distribution of Methane Sources and Cycling in the Deep Sea via in Situ Stable Isotope Analysis,” Environ. Sci. Technol. 47(3), 1478–1486 (2013). [CrossRef] [PubMed]
7. S. D. Wankel, M. Gupta, J. Leen, R. A. Provencal, V. Parsotam, and P. R. Girguis, “In Situ Stable Isotopic Detection of Anaerobic Oxidation of Methane in Monterey Bay Cold Seeps Via Off-Axis Integrated Cavity Output Spectroscopy,” American Geophysical Union, Fall Meeting 2010, December2010.
8. P. M. Michel, S. D. Wankel, J. Kapit, P. R. Girguis, M. Gupta, “Advancing a deep sea near-infrared laser spectrometer for dual isotope measurements,” in CLEO: 2015, OSA Technical Digest (online) (Optical Society of America, 2015), paper ATu4J.6.
9. R. Grilli, J. Triest, J. Chappellaz, M. Calzas, T. Desbois, P. Jansson, C. Guillerm, B. Ferré, L. Lechevallier, V. Ledoux, and D. Romanini, “Sub-ocean: subsea dissolved methane measurements using an embedded laser spectrometer technology,” Environ. Sci. Technol. 52(18), 10543–10551 (2018). [CrossRef] [PubMed]
10. R. Gonzalez-Valencia, F. Magana-Rodriguez, O. Gerardo-Nieto, A. Sepulveda-Jauregui, K. Martinez-Cruz, K. W. Anthony, D. Baer, and F. Thalasso, “In Situ Measurement of Dissolved Methane and Carbon Dioxide in Freshwater Ecosystems by Off-Axis Integrated Cavity Output Spectroscopy,” Environ. Sci. Technol. 48(19), 11421–11428 (2014). [CrossRef] [PubMed]
11. G. Gagliardi, and H. P. Loock. Cavity-Enhanced Spectroscopy and Sensing (Springer 2014).
13. D. S. Baer, J. B. Paul, M. Gupta, and A. O’keefe, “Sensitive absorption measurements in the near infrared region using off-axis integrated-cavity output spectroscopy,” Appl. Phys. B 75(2-3), 261–265 (2002). [CrossRef]
14. Y. A. Bakhirkin, A. A. Kosterev, C. Roller, R. F. Curl, and F. K. Tittel, “Mid-infrared quantum cascade laser based off-axis integrated cavity output spectroscopy for biogenic nitric oxide detection,” Appl. Opt. 43(11), 2257–2266 (2004). [CrossRef] [PubMed]
15. P. W. Heo and I. S. Park, “Separation Characteristics of Dissolved Gases from Water Using a Polypropylene Hollow Fiber Membrane Module with High Surface Area,” Internat. Scholarly Sci. Res. Innov. 8(7), 1295–1298 (2014).
16. M. Aleksanyan, “Methane sensor based on SnO2 /In2O3 TiO2 nanostrucure,” J. Contemp. Phys. 45(2), 77–80 (2010). [CrossRef]
17. X. Chao, G. Shen, K. Sun, Z. Wang, Q. Meng, S. Wang, and R. K. Hanson, “Cavity-enhanced absorption spectroscopy for shocktubes: Design and optimization,” Proc. Combust. Inst. 37(2), 1345–1353 (2019). [CrossRef]
18. J. J. S. O’Keefe and J. B. Paul, “CW integrated cavity output spectroscopy,” Chem. Phys. Lett. 307(5-6), 343–349 (1999). [CrossRef]
19. J. H. van Helden, N. Lang, U. Macherius, H. Zimmermann, and J. Röpcke, “Sensitive trace gas detection with cavity enhanced absorption spectroscopy using a continuous wave external-cavity quantum cascade laser,” Appl. Phys. Lett. 103(13), 131114 (2013). [CrossRef]
20. L. Ciaffoni, G. Hancock, J. J. Harrison, J. P. van Helden, C. E. Langley, R. Peverall, G. A. D. Ritchie, and S. Wood, “Demonstration of a mid-infrared cavity enhanced absorption spectrometer for breath acetone detection,” Anal. Chem. 85(2), 846–850 (2013). [CrossRef] [PubMed]
21. G. S. Engel, E. J. Moyer, F. N. Keutsch, and J. G. Anderson, “Innovations in cavity enhanced laser absorption spectroscopy: using in situ measurements to probe the mechanisms driving climate change,” in NASA Earth Science Technology Conference (ESTC) Proceedings, College Park, Maryland (2001).
22. L. Ciaffoni, J. Couper, G. Hancock, R. Peverall, P. A. Robbins, and G. A. D. Ritchie, “RF noise induced laser perturbation for improving the performance of non-resonant cavity enhanced absorption spectroscopy,” Opt. Express 22(14), 17030–17038 (2014). [CrossRef] [PubMed]
23. K. M. Manfred, J. M. R. Kirkbride, L. Ciaffoni, R. Peverall, and G. A. D. Ritchie, “Enhancing the sensitivity of mid-IR quantum cascade laser-based cavity-enhanced absorption spectroscopy using RF current perturbation,” Opt. Lett. 39(24), 6811–6814 (2014). [CrossRef] [PubMed]
24. J. Wang, X. Tian, Y. Dong, J. Chen, T. Tan, G. Zhu, W. Chen, and X. Gao, “High-sensitivity off-axis integrated cavity output spectroscopy implementing wavelength modulation and white noise perturbation,” Opt. Lett. 44(13), 3298–3301 (2019). [CrossRef] [PubMed]
25. H. Zhao, G. S. Wang, T. D. Cai, and X. M. Gao, “Off-axis cavity enhanced absorption spectroscopy detection techniques for the measurement of carbon dioxide,” Guangpuxue Yu Guangpu Fenxi 32(1), 41–45 (2012). [PubMed]
26. V. L. Kasyutich, P. A. Martin, and R. J. Holdsworth, “An off-axis cavity-enhanced absorption spectrometer at 1605 nm for the 12CO2 / 13CO2 measurement,” Appl. Phys. B 85(2-3), 413–420 (2006). [CrossRef]
27. C. K. Laue, R. Knappe, K.-J. Boller, and R. Wallenstein, “Wavelength tuning and spectral properties of distributed feedback diode lasers with a short external optical cavity,” Appl. Opt. 40(18), 3051–3059 (2001). [CrossRef] [PubMed]
28. P. Yalin, “Laser lineshape effects on cavity-enhanced absorption spectroscopy signals,” Appl. Phys. B 78(3-4), 477–483 (2004). [CrossRef]
29. E. J. Moyer, D. S. Sayres, G. S. Engel, J. M. S. T. Clair, F. N. Keutsch, N. T. Allen, J. H. Kroll, and J. G. Anderson, “Design considerations in high-sensitivity off-axis integrated cavity output spectroscopy,” Appl. Phys. B 92, 467–474 (2008). [CrossRef]