Open Access
Add to BibSonomy
Abstract
A novel method for laser frequency locking and intensity normalization in wavelength modulation spectroscopy (WMS)–based gas sensor system is reported. The center spacing between two second harmonic peaks demodulated from the rising and falling edges of a scanning triangular wave (for wavelength scan) is employed as a frequency locking reference. Amplitude of the directly acquired sine signal (for wavelength modulation) in the spectral region far away from the absorption feature is employed as an intensity normalization reference. A 50 ppm CH4:N2 sample sealed in a multi-pass cell at 1 atm was employed as the target analyte for demonstration. The frequency locking significantly improves measurement accuracy, and the introduced intensity normalization method realized a ~3 times SNR improvement as compared to the commonly used 1f normalization method under frequency locking conditions. A minimum measurement precision of ~2.5 ppbv was achieved with a normalized noise equivalent absorption coefficient of 1.8 × 10−9 cm−1Hz-1/2.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Over the last 40 years, tunable diode laser absorption spectroscopy (TDLAS) techniques have been widely used for a variety of applications in environmental sensing, medical diagnostic and industrial process control, owning to its ability of providing non-intrusive, selective and time-resolved in situ measurements of trace gases [1–5]. Combining wavelength modulation spectroscopy (WMS) [6–8] with long optical path length designs (such as White cell, Herriot cell or high finesse optical cavity [9–13]), detection of trace gas species at ppbv (parts per billion by volume) and pptv (parts per trillion by volume) levels becomes feasible. A calibration-free WMS method based on the use of the ratio of second harmonic signal (2f) to first harmonic signal (1f) was recently developed and applied to the measurements of gas concentration and temperature [15–17], which not only effectively broadens the measurable scope of TDLAS, but also simplifies the measurement setups for applications in harsh environment at high temperature and high pressure.
Tunable lasers employed in WMS are usually operated in free-running mode, whereas the laser center frequency may drift with time from dozens of megahertz (MHz) to hundreds of MHz, which makes the system unstable leading to additional noises. To stabilize laser emission frequency, a feedback servo loop is often used. 1f or 3f signal of a target gas absorption line from a reference cell [22–25], is used as an error signal for stabilization of the laser frequency. In general, the laser frequency is firstly scanned to map out the spectral region of the target absorption line, the feedback loop is then activated to lock the laser frequency to the absorption line center (no longer scanning the whole spectrum for trace gas quantification). However, the whole absorption spectrum contains additional information on baseline (the non-absorption part of demodulated 2f signal) and spectrum broadening that are sometimes very interesting. Consequently, it's desired to lock laser frequency while keeping its frequency scanned across the absorption line feature.
Another key factor affecting the sensitivity of the WMS-based sensor is the fluctuations in laser intensity. In order to eliminate the impact of this fluctuation on the 2f signal, 1f signal is usually employed as an amplitude reference for normalization of the 2f signal. Up to now, the 1f based 2f normalization methods can be sorted as peak value normalization and line shape normalization [14–17,19–21].
In the present work, the peak value normalization method is considered for comparison. Once the working pressure, temperature, optical path length, absorption lines and laser parameters (including modulation index) are fixed, the lineshape function becomes insensitive to the species concentration under lower absorbance condition (typically <5%) [7,8,14,18], and then the relationship between the target species concentration and 2f peak height can be described as:
where X donates the species concentration, I0 is the incident laser intensity, v0 is the center frequency of the absorption line, and P2f is the 2f peak height. When using 1f peak height P1f to replace the incident laser intensity I0, an additional demodulation channel is needed to obtain the 1f signal.In this paper, a simple but effective method is introduced for suppressing long term frequency drift without reference cell. In addition, a new internal intensity reference using “direct signal” is proposed to replace 1f in the 2f normalization.
2. Experimental setup
The experimental setup is shown in Fig. 1. A fiber-coupled distributed feedback (DFB) diode laser (NEL, NLK1E5GAAA) operating at 1653 nm was employed as TDLAS light source for this WMS experience. The laser current and temperature were set at 70 mA and 22°C (ILX Lightwave, LDC-3724) to probe the CH4 absorption lines located around 6046.95 cm−1.
Wavelength scanning and modulation were implemented by applying a triangle voltage ramp (20 Hz, 2 V), from a function generator summed with a sinusoidal dither (10 kHz, 0.3 V) from a lock-in amplifier, to the laser controller LDC-3724.
After passing through a compact multi-pass cell (MPC) with an effective optical path length of 26.4 m (based on a basal length of 12 cm), the laser beam was detected with a photodetector (New Focus, 2011-FC). Subsequently, signals from three channels (for DS, 1f and 2f respectively) were acquired at a sampling rate of 400 kHz via a laptop equipped with a DAQ card (National Instruments, USB-6366): (1) the direct signal (DS) from the detector is used for monitoring optical intensity fluctuations and for 2f signal normalization; (2) the 1f signal from a lock-in amplifier (Stanford Research Systems, SR830) used for performance comparison between 1f-based and DS-based normalization; (3) the 2f signal from another lock-in amplifier used for gas concentrations retrieval and for monitoring laser frequency drift. Synchronously, the acquired signals were processed in Labview environment.
A feedback server loop is designed to lock the laser frequency through a USB cable involving a Labview based proportional integral derivative (PID) algorithm.
3. Results and discussion
3.1 Frequency locking
The principle of the proposed method for laser frequency locking is schematically presented in Fig. 2. When scanning the laser current with a triangle wave overlapped with a sinuous wave, two absorption peaks can be obtained from the rising and falling edges of the triangular wave in one scan cycle. The spacing between two peaks might change with the shift of the laser wavelength. This center spacing can be employed as a reference for frequency locking by keeping the spacing constant. In order to improve the locking accuracy, a 5th order wavelet denoising method is employed to reduce the noise of 2f signal (so as to the center spacing). Comparison between the denoised signal (red curve) and raw signal (black line) is shown in the inset of Fig. 2, where a significant improvement of the signal-to-noise ratio is obtained by which the locking accuracy is finally improved.
Fig. 2 Schematic diagram of the proposed method for frequency locking. Inset figure: denoised 2f signal (in red) and raw 2f signal (in black).
The relationship between laser current, wavenumber and center spacing is given in Fig. 3, obtained by changing the laser center current from 60 to 80 mA with a step of 0.005 mA. This current range corresponds to a frequency (in wavenumber) range from 6046.69 to 6047.23 cm−1. According to Fig. 3, the center spacing is monotonically changing with the laser current and wavenumber, which can be used therefore as the frequency locking reference. When the feedback server loop is activated, an error signal will be continuously generated from the difference between a set point and the current center spacing. Therefore, the center spacing can be locked at a certain value by adjusting laser center current. It should be noted that the PID parameters are determined by the relationship between laser current and center spacing (Fig. 3, blue curve). In addition, the relationship between wavenumber and center spacing (Fig. 3, red line) can be used to estimate the frequency shift in cm−1 (or MHz) during the measurement.
In the present work, a 50 ppm CH4:N2 sample sealed in the MPC at 1 atm was employed as the target analyte. Continuous measurements of the analyte over 20 hours with a 0.5 s interval were performed to study the stability of the WMS system. Long-term frequency fluctuations observed through the variation in the center spacing under free-running (Lock OFF) and frequency locking (Lock ON) conditions are shown in Fig. 4(a). A frequency drift of about 400 MHz (peak to peak value) under free-running condition was reduced to less than 40 MHz by frequency locking, where the frequency drift is determined through the relationship between the center spacing and the corresponding wavenumber. The 2f signals under free-running and frequency locking conditions are shown in Fig. 4(b), where an obvious improvement of the 2f drift can be observed under frequency locking conditions.
Fig. 4 Comparisons of the results under laser frequency locking OFF and ON. (a) Results of the center spacing; (b) Results of the 2f values.
3.2 Intensity normalization
Besides the laser frequency drift, fluctuation in optical intensity is another key factor that affects the stability of a WMS-based sensor. As shown in Fig. 4(b), though the frequency locking improves the 2f drift obviously, the 2f drift still exists. For intensity normalization, amplitude of the directly acquired sine signal (termed as DS-sine, see Fig. 5(b)), in the spectral region where there is no disturbance from the interference of other molecules or from far wing absorbance, far away from absorption peak was employed as an alternative intensity normalization reference to 1f value. As a comparison to the traditional intensity normalization, 1f signal is also acquired and the offset of the absorption center to the zero level (see Fig. 5(c)) is used as 1f value for the 2f normalization. Compared with the 1f normalization method, the DS-sine normalization method adds only one acquisition channel while 1f normalization method needs an additional demodulation module (lock-in amplifier).
Fig. 5 Schematic diagram of the proposed method for 2f signal normalization. (a) 2f signal and 2f peak value; (b) direct signal (DS) and DS-sine value (inset figure); (c) 1f signal and 1f value.
In order to test the DS-sine normalization method, an iris is employed to vary the optical intensity impacting the detector. By this way, direct intensity signals (without scan and modulation, termed as DA) and DS-sine signals at different optical intensities were acquired.
A linear fit of the DA signal to the DS-sine signal, shown in Fig. 6, yields an R-square value of > 0.9999, which verifies the feasibility of using DS-sine value for 2f normalization. Specially, it should be noted that the DS-sine value was calculated by using a Labview subroutine named as “extract single tone information”.
When the frequency locking is ON, the fluctuation in laser intensity becomes the major factor affecting the measurement precision in a WMS-based sensor. The measured results of 2f values, DS-sine values and 1f values over 10 hours are shown in Figs. 7(a)-7(c), respectively. As shown in Fig. 7(a), there is a slow deviation of 2f values from 10 to 15 hours, at the same period, a similar deviation of DS-sine values is observed in Fig. 7(b). This indicates that the deviation is caused by the laser intensity fluctuation, and this fluctuation can be accurately normalized by using the DS-sine values. Though the 1f signal (in Fig. 7(c)) can also track the intensity fluctuation, the SNR of the DS-sine signal is much better for accurate normalization.
3.3 Comparison of different methods
For performing a uniform comparison, the results of 2f/DS-sine and 2f/1f under different conditions are transformed to concentrations (in ppm). The measured 2f values normalized by DS-sine and by 1f values, respectively, are shown in Fig. 8.
Fig. 8 Time-series concentration measurements using different normalization methods under different conditions. (a) Normalized 2f signal by DS-sine value (2f/DS-sine); (b) Normalized 2f signal by 1f value (2f/1f).
Fluctuations in the measured concentration were about 2 ppm for 2f/DS-sine normalization and 3.1 ppm for 2f/1f normalization under free-running condition. These fluctuations were reduced to less than 0.5 ppm 2f/DS-sine and 1.6 ppm for 2f/1f under frequency locking condition.
These results illustrate that the measurement accuracy of the WMS sensor is mainly affected by frequency drift under free-running conditions, and it can be efficiently improved by locking laser frequency. In addition, under frequency locking conditions the standard deviation of about 0.228 ppm for 2f/1f, shown in Fig. 8(b), is roughly three times greater than that from 2f/DS-sine (Fig. 8(a)).
Allan variance analysis [26] under different conditions are plotted in Fig. 9. The results indicate that the optimal integration time has been improved from ~100 s for free-running mode to ~400 s for frequency locking mode. Minimum measurement precisions (MMP) of 12.5 ppbv, 7.4 ppbv, 5.3 ppbv, and 2.5 ppbv are obtained for 2f/1f–free, 2f/1f–lock, 2f/DS-free, and 2f/DS–lock, respectively. The MMP was significantly improved by employing frequency locking, and further improved by using DS-sine to normalize 2f signal. Based on the experimental results, normalized noise equivalent absorption coefficients (NNEA) of 5.5 × 10−9 cm−1Hz-1/2, and 1.8 × 10−9 cm−1Hz-1/2 were achieved for 2f/1f –lock and 2f/DS-lock, respectively.
4. Conclusions
In conclusion, we demonstrated improved performance in WMS approach for trace gas sensing by implementation of laser frequency locking while scanning laser frequency across absorption line and DS-sine intensity normalization. Compared to conventional WMS system without frequency locking, the optimal integration time and the MMP were both significantly improved by a factor of ~4. In addition, the DS-sine based 2f normalization method shows a more accurate concentration determination than commonly used 1f normalization method under frequency locking conditions. The method described in this article has a high potential for long-term trace gas detection.
Funding
National Key Research and Development Program of China (2017YFC0209700 and 2016YFC0303900); National Natural Science Foundation of China (41730103, 61775221, and 41405022).
References
1. H. I. Schiff, G. I. Mackay, and J. Bechara, “The use of tunable diode laser absorption spectroscopy for atmospheric measurements,” Res. Chem. Intermed. 20(3–5), 525–556 (1994). [CrossRef]
2. M. Lackner, “Tunable diode laser absorption spectroscopy (TDLAS) in the process industries–a review,” Rev. Chem. Eng. 23(2), 65–147 (2007). [CrossRef]
3. J. Shemshad, S. M. Aminossadati, and M. S. Kizil, “A review of developments in near infrared methane detection based on tunable diode laser,” Sens. Actuators B Chem. 171, 77–92 (2012). [CrossRef]
4. W. Ren, L. Luo, and F. K. Tittel, “Sensitive detection of formaldehyde using an interband cascade laser near 3.6 μm,” Sensor. Actuat. B-Chem. 221, 1062–1068 (2015).
5. J. Shao, J. Xiang, O. Axner, and C. Ying, “Wavelength-modulated tunable diode-laser absorption spectrometry for real-time monitoring of microbial growth,” Appl. Opt. 55(9), 2339–2345 (2016). [CrossRef] [PubMed]
6. S. Schilt, L. Thévenaz, and P. Robert, “Wavelength modulation spectroscopy: combined frequency and intensity laser modulation,” Appl. Opt. 42(33), 6728–6738 (2003). [CrossRef] [PubMed]
7. H. Li, G. B. Rieker, X. Liu, J. B. Jeffries, and R. K. Hanson, “Extension of wavelength-modulation spectroscopy to large modulation depth for diode laser absorption measurements in high-pressure gases,” Appl. Opt. 45(5), 1052–1061 (2006). [CrossRef] [PubMed]
8. S. Schilt and L. Thévenaz, “Wavelength modulation photoacoustic spectroscopy: Theoretical description and experimental results,” Infrared Phys. Technol. 48(2), 154–162 (2006). [CrossRef]
9. D. R. Herriott and H. J. Schulte, “Folded optical delay lines,” Appl. Opt. 4(8), 883–889 (1965). [CrossRef]
10. J. U. White, “Long optical paths of large aperture,” J. Opt. Soc. Am. 32(5), 285–288 (1942). [CrossRef]
11. L. Dong, C. Li, N. P. Sanchez, A. K. Gluszek, R. J. Griffin, and F. K. Tittel, “Compact CH4 sensor system based on a continuous-wave, low power consumption, room temperature interband cascade laser,” Appl. Phys. Lett. 108(1), 011106 (2016). [CrossRef]
12. K. Liu, L. Wang, T. Tan, G. Wang, W. Zhang, W. Chen, and X. Gao, “Highly sensitive detection of methane by near-infrared laser absorption spectroscopy using a compact dense-pattern multipass cell,” Sens. Actuators B Chem. 220, 1000–1005 (2015). [CrossRef]
13. J. B. Paul, L. Lapson, and J. G. Anderson, “Ultrasensitive absorption spectroscopy with a high-finesse optical cavity and off-axis alignment,” Appl. Opt. 40(27), 4904–4910 (2001). [CrossRef] [PubMed]
14. J. T. C. Liu, J. B. Jeffries, and R. K. Hanson, “Wavelength modulation absorption spectroscopy with 2f detection using multiplexed diode lasers for rapid temperature measurements in gaseous flows,” Appl. Phys. B 78(3–4), 503–511 (2004). [CrossRef]
15. G. B. Rieker, J. B. Jeffries, and R. K. Hanson, “Calibration-free wavelength-modulation spectroscopy for measurements of gas temperature and concentration in harsh environments,” Appl. Opt. 48(29), 5546–5560 (2009). [CrossRef] [PubMed]
16. C. S. Goldenstein, C. L. Strand, I. A. Schultz, K. Sun, J. B. Jeffries, and R. K. Hanson, “Fitting of calibration-free scanned-wavelength-modulation spectroscopy spectra for determination of gas properties and absorption lineshapes,” Appl. Opt. 53(3), 356–367 (2014). [CrossRef] [PubMed]
17. K. Sun, X. Chao, R. Sur, C. Goldenstein, J. Jeffries, and R. Hanson, “Analysis of calibration-free wavelength-scanned wavelength modulation spectroscopy for practical gas sensing using tunable diode lasers,” Meas. Sci. Technol. 24(12), 125203 (2013). [CrossRef]
18. J. Shao, J. Guo, L. Wang, C. Ying, and Z. Zhou, “Self-calibration methodology by normalized intensity for wavelength modulation spectroscopy measurement,” Opt. Commun. 336, 67–72 (2015). [CrossRef]
19. G. Zhao, W. Tan, J. Hou, X. Qiu, W. Ma, Z. Li, L. Dong, L. Zhang, W. Yin, L. Xiao, O. Axner, and S. Jia, “Calibration-free wavelength-modulation spectroscopy based on a swiftly determined wavelength-modulation frequency response function of a DFB laser,” Opt. Express 24(2), 1723–1733 (2016). [CrossRef] [PubMed]
20. T. R. S. Hayden and G. B. Rieker, “Large amplitude wavelength modulation spectroscopy for sensitive measurements of broad absorbers,” Opt. Express 24(24), 27910–27921 (2016). [CrossRef] [PubMed]
21. T. R. S. Hayden, D. J. Petrykowskia, A. Sancheza, S. P. Nigama, C. Lapointea, J. D. Christophera, N. T. Wimera, A. Upadhyeb, M. Strobelb, P. E. Hamlingtona, and G. B. Rieker, “Characterization of OH, H2O, and temperature profiles in industrial flame treatment systems interacting with polymer films,” Proc. Combust. Inst. 37(2), 1571–1578 (2018).
22. T. Yanagawa, S. Saito, and Y. Yamamoto, “Frequency stabilization of 1.5-μm InGaAsP distributed feedback laser to NH3 absorption lines,” Appl. Phys. Lett. 45(8), 826–828 (1984). [CrossRef]
23. L. Dong, W. Yin, W. Ma, and S. Jia, “A novel control system for automatically locking a diode laser frequency to a selected gas absorption line,” Meas. Sci. Technol. 18(5), 1447–1452 (2007). [CrossRef]
24. P. Gong, L. Xie, X. Q. Qi, and R. Wang, “QEPAS-based central wavelength stabilized diode laser for gas sensing,” IEEE Photonics Technol. Lett. 27(5), 545–548 (2015). [CrossRef]
25. Q. Wang, Z. Wang, and W. Ren, “Wavelength-stabilization-based photoacoustic spectroscopy for methane detection,” Meas. Sci. Technol. 28(6), 065102 (2017). [CrossRef] [PubMed]
26. P. O. Werle, R. Mücke, and F. Slemr, “The limits of signal averaging in atmospheric trace-gas monitoring by tunable diode-laser absorption spectroscopy (TDLAS),” Appl. Phys. B 57(2), 131–139 (1993). [CrossRef]
