Cite this Article
Chen Xiang, Yang Chen-Guang, Hu Mai, Shen Jian-Kang, Niu Er-Chao, Xu Zhen-Yu, Fan Xue-Li, Wei Min, Yao Lu, He Ya-Bai, Liu Jian-Guo, Kan Rui-Feng. Highly-sensitive NO, NO2, and NH3 measurements with an open-multipass cell based on mid-infrared wavelength modulation spectroscopy. Chinese Physics B, 2018, 27(4): 040701  
Permissions
Highly-sensitive NO, NO2, and NH3 measurements with an open-multipass cell based on mid-infrared wavelength modulation spectroscopy
Chen Xiang1, 2, Yang Chen-Guang1, Hu Mai1, Shen Jian-Kang3, Niu Er-Chao3, Xu Zhen-Yu1, Fan Xue-Li1, Wei Min1, 2, Yao Lu1, He Ya-Bai1, Liu Jian-Guo1, Kan Rui-Feng1, †
Key Laboratory of Environmental Optics and Technology, Anhui Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Hefei 2300314, China
University of Science and Technology of China, Hefei 230022, China
Jiangsu Environmental Monitoring Center, Nanjing 210036, China

 

† Corresponding author. E-mail: kanruifeng@aiofm.ac.cn

Abstract

A compact prototype based on mid-infrared wavelength modulation spectroscopy (WMS) is developed for the simultaneous monitoring of NO, NO2, and NH3 in the urban area. Three quantum cascade lasers (QCLs) with central frequencies around 1900.0 cm−1, 1600.0 cm−1, and 1103.4 cm−1 are used for NO, NO2, and NH3 detections, respectively, by time-division multiplex. An open-path multi-pass cell of 60-m optical path length is applied to the instrument for high sensitivity and reducing the response time to less than 1 s. The prototype achieves a sub-ppb detection limit for all the three target gases with an average time of about 100 s. The instrument is installed in the Jiangsu environmental monitoring center to conduct performance tests on ambient air. Continuous 24-hour measurements show good agreement with the results of a reference instrument based on the chemiluminescence technique.

PACS: ;07.07.Df;;33.20.-t;;42.55.Px;;42.62.-b;
Keyword:;wavelength modulation spectroscopy;;optical sensing;;atmospheric monitoring;;time division multiplexing;;quantum cascade laser;;mid-infrared laser spectroscopy;
1. Introduction

As the precursors of secondary fine particulate matters, NO, NO2, and NH3 in the air have been causing global environmental issues. NOx participates in a chain reaction that destroys the protective stratospheric ozone layer, resulting in an increase of ultraviolet radiation reaching the surface of the Earth.[1] Also, NOx promotes acid rain that causes soil acidification and affects the growth of plants.[2] Moreover, NOx plays an important role in forming photochemical smog.[3] Therefore, it is essential to reliably measure the NOx concentration in ambient air, as one of the most abundant nitrogen-containing molecules, ammonia, has important effects on atmospheric chemistry and the environment. The major ingredient is alkaline gas, and therefore it plays an important role in neutralizing the acidic gases and forming the particulate matters. However, NH3 adsorption onto the sampling manifolds can result in significant discrepancies between the measured and actual concentrations.[4] Open-path arrangements can help to make accurate measurements without being affected by adsorption.[5,6]

In recent years, the detecting of NO, NO2, and NH3 based on infrared WMS[7–10] has been widely studied for industrial processes, environmental monitoring and medical diagnoses. For example, a quantum cascade laser (QCL)[11,12] with a center wavelength ∼5263 nm was used to detect NO emission.[2] With a resolution of line strength, a resolution about 2 ppm was realized and the relative error was less than 3%. In order to estimate the NO and NO2 background level in the free troposphere, a QCL-based absorption spectrometer was employed at the high-altitude site Jungfraujoch (3850 m a.s.l) in Switzerland.[13] The precisions of 10 ppt and 3 ppt for NO and NO2 were achieved under field conditions with 180-s averaging time, resepectively. The linear dynamic range of the instrument for both species increased up to 45 ppb. Also, it is essential to measure NO and its metabolites in various body compartments to enhance our understanding of biology. Wang et al. presented a mid-IR Faraday rotation spectrometer for detecting the NO isotopes with detection limits of 3.72 ppbv/Hz1 / 2 for 14NO and 0.53 ppbv/Hz1 / 2 for 15NO, respectively.[14] Generally, a single-laser system is less complex and consumes less power than systems using multiple laser modules. Jagersk et al. presented the practical realization of such a pulsed QCL dual-wavelength setup, which emitted single-mode radiation at 1600 cm−1 and 1900 cm−1 and was suited for simultaneously detecting NO and NO2.[15] In industrial processes, the NOx emission in exhaust gasses was strictly restricted, whereas ammonia was widely used in flue gas treatment for deNOx applications. Therefore, a highly sensitive setup based on near-infrared lasers for the detection of excess NH3 has been investigated.[16] The extractive gas cell was heated to a temperature of 180 °C to avoid condensing the moisture and sticking the ammonia molecules. For an ammonia breath sensor, using absorption lines located at the near-infrared absorption band may not satisfy the demand for detection limit. A calibration-free ammonia sensor using a QCL with the WMS technique was demonstrated for monitoring the ammonia in exhaled breath.[17] The minimum detectable ammonia mole fraction was found to be 7 ppb with a 76.45-m long extractive multi-pass gas cell. Compared with the extractive type, the open-path sensor configuration minimized sampling artifacts associated with NH3 surface adsorption onto the inlet tubing.[18] This open-path sensor has a sensitivity and noise-equivalent limit of 0.15-ppbv NH3 at 10 Hz.

To be able to do further research on the formation of secondary fine particulate matters, we report the development of a QCL-based instrument for monitoring NO, NO2, and NH3 in the open air. Three absorption lines located at 1900.0 cm−1, 1600.0 cm−1, and 1103.4 cm−1 were selected for detecting NO, NO2, and NH3 respectively, where the line strengths were two orders stronger than those in the near-infrared absorption region. In contrast to closed-path systems, our non-destructive open-path design minimized the challenges of sampling artifacts and provided a fast time response. The time division multiplexing technology and second harmonic analysis with WMS can be used to simultaneously detect NO, NO2, and NH3 at a sub-ppb level. To test its reliability and long-term operation, the prototype was installed in the Jiangsu environmental monitoring center and compared with the measuring instrument based on chemiluminescence in the center.

2. Theory and absorption line selection
2.1. WMS principles

When a QCL injection current is modulated by a sinusoidal modulation signal, the laser frequency ν(t) and output intensity I(t) can be expressed as where is the center laser frequency, is the average laser intensity, a is the amplitude of frequency modulation, i1 and i2 are the linear amplitude and the nonlinear amplitude of the intensity modulation, respectively, ω is the angular modulation frequency, while and are the phase shifts of the linear term and the nonlinear term, respectively. In most of the situations, i2 is much smaller than i1 and their effects of nonlinearities can be ignored.[19] When the modulated laser radiation passed through the gas sample, the transmittance was given by the Beer–Lambert relation where S(T) is the absorption line strength at temperature T, P is the pressure, X is the mole fraction of absorbing species, L is the optical path length, and is the lineshape function. For an optically thin sample, the transmittance can be described approximately as

Expand the transmittance in Fourier series of laser angular modulation frequency and its harmonics kω and combine with simplified Beer–Lambert relation for weak absorption, then H terms will be written as

In the experiment, a digital lock-in amplifier with appropriate parameters was used to attain the first and second (i.e., 1f and 2f) harmonic signal of the transmitted light intensity through the gas sample. By normalizing the 2f signal against the magnitude of the 1f signal, the dependence on laser intensity was eliminated. To a good approximation, the amplitude of the normalized 2f signal was found to be

After the line strength, the air pressure and optical path length have been predetermined, and the gas concentration can be acquired based on Eq. (8). For long-term monitoring measurements, the ambient temperature T and pressure P are recorded for calculating the line strength and the mole fraction X value.

2.2. Absorption line selection

Since the detection was performed on ambient air with an open-path setup, water vapor, carbon dioxide and other species in the air may cause interferences due to their overlapping absorption features or pressure broadening effects. Therefore, spectral features in the ranges of 1900.0 cm−1, 1600.0 cm−1, and 1103.4 cm−1 were selected for NO, NO2, and NH3, where their absorption features overlap with the minimum of the adjacent interfering gas absorption feature, and the line strength was relatively strong. To demonstrate the feasibility further, the detailed simulations of absorption spectra of NO, NO2, and NH3, together with typical ambient mixing ratios of atmospheric absorbers were performed. Figure 1 shows the spectra simulated based on HITRAN databases for 10-ppb NO, 40-ppb NO2, 15-ppb NH3, 80-ppb O3, 400-ppm CO2, 1800-ppb CH4, 5-ppb SO2, and 2.0% H2O at 296 K and 1 atm (1 atm = 1.01325×105 Pa) for a 60-m optical path length. In the region of 1900 cm−1, the NO absorption feature is a superposition of five absorption transitions, and the major influence comes from CO2 and H2O. For NO2 and NH3 absorption lines near 1600.0 cm−1 and 1103.4 cm−1, respectively, the H2O lines still have a remarkable influence. To analyze the influences of H2O, CO2 and other atmospheric gases quantitatively, the second harmonic signals were simulated with the same conditions and are presented in Fig. 2. From the results, 2.0% H2O along with 400 ppm CO2 will result in 1.5% deviation of the amplitude of NO (0.15 ppb). The amplitudes of NO2 and NH3 would vary about 0.7% (0.28 ppb) and 0.18% (0.03 ppb), respectively, while H2O changes from 0.0% to 2.0%. Other atmospheric species such as O3, CH4, and SO2 can be neglected for the retrieval of second harmonic signals. In the ambient monitoring experiment, the average ambient concentrations of H2O and CO2 for a given time period were used to calculate the background 1f-normalized 2f signals based on the QCLs output. These H2O and CO2 background 1f-normalized 2f signals were then subtracted from the demodulated measurement signals to remove their contribution. A typical daily concentration fluctuation of CO2 in the air is about 50 ppm. The concentration of H2O can be detected by a hygrometer and the short-term fluctuation was typically a few hundred ppm. Based on the simulation, the concentration fluctuations of CO2 and H2O in air have much smaller influences on measurements of NO, NO2, and NH3 than their system detection limits ( ). The detailed simulation analysis confirms that absorption lines located at 1900.0 cm−1, 1600.0 cm−1 and 1103.4 cm−1 are appropriate for detecting the NO, NO2, and NH3 with open-path cells.

Fig. 1. (color online) Direct absorption HITRAN simulations of (a) NO, (b) NO2, and (c) NH3, with typical ambient mixing ratios of atmospheric absorbers, based on 60-m path length, 296 K, 1 atm.
Fig. 2. (color online) Second harmonic spectra of (a) NO, (b) NO2, and (c) NH3, with typical ambient mixing ratios of atmospheric absorbers calculated based on 60-m path length, 296 K, 1 atm.
3. Instrument setup

The schematic diagram of the open-path sensor is shown in Fig. 3. The open-path design was optimized for field portability with a 131 cm×41 cm×35 cm sensor head and a 49 cm ×50 cm×62 cm control unit. The total weight of the sensor is 68 kg. The quantum cascade lasers (Hamamatsu Lasers), were installed on the surface of Peltier devices with excellent heat dissipation controlled by low-noise current drivers (Wavelength electronics, PLD1250) and precision temperature controllers (Wavelength electronics, PTC10K-CH), respectively. Furthermore, the temperature of the enclosure of QCLs and drivers were controlled by fans, along with Peltier devices with high-efficiency temperature controllers (Analog Technologies, TECA1-XV) to achieve high stability in field conditions.

Fig. 3. (color online) Schematic diagram of an open-path instrument for atmospheric gas sensing, based on the wavelength modulation spectroscopy technique and mid-infrared quantum cascade lasers.

In the experiment three 50-Hz-frequency time division multiplexing sawtooth ramp signals superimposed with a 30-kHz-frequency sinusoidal signal were used to modulate laser injection currents. The ramp and modulation signals were generated by a data acquisition device (DAQ, NI-USB 6363, 16-bit resolution). Three QCLs output beams were combined by wedged dielectric beam splitters ( and on both sides), and passed through a 25-cm-long reference gas cell equipped with wedged CaF2 windows before being injected into an open-air multi-pass cell. By using wedged beam splitters, no optical interferences would be formed. The reflectors were coated with gold film ( ). The multi-pass cell consisted of two 10.2-cm-diameter gold-coated spherical mirrors ( ), with a base length of 0.87 m, which could provide an optical path length of 60 m. A green indication light beam was used to guide optical beam path adjustments.

The output beams from the multi-pass cell were focused onto a room-temperature MCT photodiode (Vigo, PC-2TE) by an off-axis parabolic mirror (f = 50.4 mm) and fine adjustment was made to optimize the detector signal. Data acquisition of the detector output signal was performed by the same DAQ card with a sampling rate of 2 MHz and a resolution of 16 bits. The raw spectral signal was averaged over 50 scans to enhance the signal-to-noise ratio. In operation, digital lock-in processing of the acquired data can be performed in real time, which led to a response time of 1 s. Meanwhile, ambient temperature (T) and pressure (P) were recorded with measuring equipment (OMEGA, iBTHX). The measuring accuracies of temperature and pressure were 0.5 °C and 2 mbar (1 bar = 105 Pa), respectively. The concentration detection responsivities of the instrument at T and P (see Eq. (8)) were obtained by multiplying its calibration values, which were measured at T0 and P0 (see Fig. 5), with the ratios of and .

4. Experimental results and discussion
4.1. Modulation signal optimization and sensitivity calibration

The sweep cycle rate of time division multiplexing signals is interleaved at 50 Hz and the fast modulation frequency is at 30 kHz. When QCLs are not in the emitting state, their bias currents are set slightly below their thresholds without emitting any light or causing interference. This method of operating the laser current just below its threshold rather than turning off the current completely avoids rapidly declining the injection current and helps to achieve a longer operational lifespan for the lasers. In addition, the time division multiplexing technique reduces the average light-emitting time of QCLs as well as the produced heat, and improved the thermal stability of the system. Another reported approach is to modulate each of the lasers at different frequencies. This required complex signal demodulation also increases the photodetectorʼs total amount of received light emitted by multiple lasers, which will contribute to higher noise.

Since the modulation depth varies with modulation voltage directly, an appropriate modulation voltage is required to maximize the 1f-normalized 2f amplitudes of NO, NO2, and NH3. We investigate the modulation characteristics by using the fixed concentrations of NO, NO2, and NH3 in the 25-cm reference cell alternatively and by filling the enclosed open-path cell with pure N2. The amplitudes of NO, NO2, and NH3 1f-normalized 2f signals at different modulation voltages are recorded in Fig. 4. Based on these results, the modulation amplitudes of 140 mV for NO, 120 mV for NO2, and 240 mV for NH3 are selected and used in the subsequent experiments. These correspond to modulation depths of 0.048 cm−1, 0.062 cm−1, and 0.10 cm−1 for NO, NO2, and NH3, respectively.

Fig. 4. (color online) 1f-normalized 2f signal amplitudes of (a) NO, (b) NO2, and (c) NH3 at different modulation depths.
Fig. 5. (color online) Calibrations of the detection responsivity of the instrument for (a) NO, (b) NO2, and (c) NH3.

Due to the large volume of the 60-m-long multi-pass cell and the sticky characteristics of NOx and NH3, serious deviations may be introduced when we calibrate this system directly with gas samples at a ppb level. According to the Specifications and Test Procedures for the Ambient Air Quality Continuous Automated Monitoring System promulgated by the Chinese Ministry of Environmental Protection,[20] the 2.5-cm-diameter short reference-gas cell can be used for accurate in-line calibration. While examining the responses of the system at different concentrations of gas samples, the multi-pass cell is enclosed and purged with pure N2 gas. After calibration is finished, the enclosure is removed and the reference cell is filled with pure N2. For an open-path setup, this approach minimizes the calibration deviations and prevents the open-path cell from being contaminated.

The responses of the sensor system at different concentrations of gas samples are studied by using reference gas mixtures (Guangming Research & Design Institute of Chemical Industry, with 2% relative uncertainty). Pure N2 and reference gas samples of NO (10 ppm, 20 ppm, 30 ppm, 40 ppm, 50 ppm), NO2 (5 ppm, 15 ppm, 25 ppm, 35 ppm, 45 ppm), and NH3 (10 ppm, 20 ppm, 30 ppm, 40 ppm, 50 ppm) are injected into the reference cell successively. Due to the characteristics of NOx and NH3 strongly adsorbing the cell surface, an empirical relatively fast gas flow rate of 5 L/min is used to stabilize the cell surface in a short time of ∼5 min. The gas flow of each concentration is maintained for fifteen mins before acquiring stable 1f-normalized 2f signals. The experimental data are recorded for five mins in order to obtain an accurate instrumental response. The peak values of 1f-normalized 2f signals are calculated and the good linear relationships between gas concentration and peak value for different gases are observed as shown in Fig. 5.

4.2. Allan deviation

The minimum detection limits (MDL) of the system for the three gases are investigated as a function of averaging measurement time. To remove the fluctuation influence derived from concentration fluctuations in the ambient air, a pure N2 purge gas flow is used to fill the laser beam path including the reference cell and the enclosed open-path cell. Continuous measurement results with the three QCLs (for NO, NO2, and NH3, respectively) are recorded for 25 min with time-division multiplex at the 1-Hz cycle rate. Allan variance analysis[21–23] is performed and the results are plotted in Fig. 6, showing the precision of gas measurements as a function of the averaging time. As indicated in the figure, MDLs of 0.12 ppb for NO, 0.11 ppb for NO2, and 0.11 ppb for NH3 with optimum averaging times of 80 s, 150 s, and 100 s can be achieved with this setup. As the averaging time increases further, the detection limits start to degrade as shown in Fig. 6. The plots demonstrate 1-s detection limits of 0.96 ppb, 0.94 ppb, and 0.86 ppb for NO, NO2, and NH3 respectively. These similar detection limits for all three gases are just a coincidence despite differences in absorption line strength among the three gases and in operation characteristic among three corresponding QCLs. While the concentrations of NO, NO2, and NH3 molecules in the air fluctuate at a ppb level, this instrument is suitable for their environmental monitoring with a response time of 1 s.

Fig. 6. (color online) Allan deviation plots of (a) NO, (b) NO2, and (c) NH3 detection sensitivity as a function of averaging time.
4.3. Wavelength stability

The wavelength stability of QCL is essential as severe wavelength drift may tune the target absorption lines outside of the scanning range. An active wavelength stabilization is an option, but would increase the system complexity and cost. We find that maintaining a high temperature stability is sufficient to minimize the wavelength drift. Since the wavelength drift correlates well with the temperature of the laser case and drivers, the precise temperature control is performed to suppress the drift. An example of 1f-normalized 2f signals of ambient air recorded at 1 Hz is depicted in Fig. 7. The temperatures of laser shells and drivers are stabilized at 20 ±0.15 °C and 25 ±1 °C respectively with Peltier devices. The laser wavelength drifts as determined by the peak position of wavelength scan signals, which are recorded in Fig. 8, for 24 hours in the steps of 1 min. Based on the result, daily wavelength drift is minimized to less than 0.01 cm−1. Therefore, the prototype is able to make measurements continuously without any adjustment, as the laser wavelength drift changes within a small range of the absorption spectra profile.

Fig. 7. (color online) Measurement examples of 1f-normalized 2f signals of (a) NO, (b) NO2, and (c) NH3 in ambient air.
Fig. 8. (color online) Wavelength drifts of three QCLs for (a) NO, (b) NO2, and (c) NH3 over 24 hours.
4.4. Ambient air monitoring

The long period monitoring of NO, NO2, and NH3 is performed at the urban site of the Jiangsu environmental monitoring center to examine the reliability and accuracy of the prototype. Restricted by the available laboratory space at the experimental site, our prototype is installed in a laboratory on the third floor of the building in a good ventilation condition, whereas a commercial instrument (Teledyne, model T201) is used for comparison and is placed on the sixth floor, which is about 15 meters above the ground. The observation site is beside a street with relatively heavy traffic. To account for the influence of temperature and pressure variations during a long measurement period, the real-time temperature and pressure are measured and used for correcting the absorption line strengths and 2f amplitudes of the three target gases. The model T201 used for making comparative measurements is based on chemiluminescence (CL) technology[24] with an external thermal converter to give stable and repeatable NO, NO2, and NH3 measurements. The instrument uses an innovative sample path with specially conditioned materials, to preserve gas samples and minimize the interference. A gas mixing system, together with pure N2 and standard gas samples of NO (10 ppm), NO2 (5 ppm), and NH3 (10 ppm) is used to make daily calibration of the instrument at a sub-ppm level. The detection limit of this instrument is 1 ppb in a 10-min sampling time.

The 24-hour monitoring results of our WMS open-path sensor and the CL reference instrument on November 24, 2016 are shown in Figs. 9–11 for NO, NO2, and NH3 respectively. Data from the reference instrument are averaged hourly by the instrument. The measurement is made by using our WMS instrument every second. The hourly-averaged WMS data are also presented in these figures to illustrate the comparison results. As shown in these figures, the measured NO, NO2, and NH3 concentrations by both instruments reveal that they have the same features and are in good agreement with each other. Compared with the CL instrument, our open-path sensor is able to capture subtle fluctuations of gas concentration with a fast response time of 1 s. Since the location of the open-path sensor is closer to the ground, the rapid fluctuations of gas concentration due to human activities on the street are observed and lead to their rapid changes. The NO concentration shows two high concentration peaks during a day starting at 9 o’clock in the morning and ending at 10 o’clock at night. The NO2 concentration presents an upward trend during the test. Because the NOx is mainly generated by automobile exhaust and the measuring site is beside the street, a high concentration level in the daytime as well as large concentration changes of NO and NO2 is reasonable. Agricultural activities are the major source of NH3 discharges. The measured NH3 concentration increases from a low value of 4 ppb in the early morning to a value of about 9 ppb in the evening, which decreases back to a low value in the morning.

Fig. 9. (color online) NO diurnal profiles measured with WMS and CL technologies.
Fig. 10. (color online) NO2 diurnal profiles measured with WMS and CL technologies.
Fig. 11. (color online) NH3 diurnal profiles measured with WMS and CL technologies.

For the regression analysis, data from the WMS instrument are block-averaged hourly values in order to match the time resolution of the CL instrument. Figure 12–14 show that the NO, NO2, and NH3 concentrations from the WMS instrument are compared with those from the CL instrument, together with the results of linear regression analysis. Overall, similar trends of concentration changes for target gases between two instruments are observed, where the correlation factor is larger than 0.85. However, the ratios between the concentrations measured by the two instruments deviate from the ideal ratio of 1:1. The regression line of NO has a slope slightly greater than 1, but has a negative intercept of −1.33 ppb. Relatively high intercept and small slope are found for NO2 as shown in Fig. 13. The NH3 measurements have the lowest correlation factor in the measurements of the three gases. There are several possible reasons for the discrepancies between WMS sensor and CL instrument for three target gases. Firstly, the two instruments sample the ambient air from two nearby but different locations. Secondly, standard gas cylinders of NO, NO2, and NH3 used for calibrating both the instruments are different. Thirdly, measurement error is introduced by the instrument itself. For example, the ambient reactive nitrogen species and the conversion efficiency can lead to the discrepancy between the NOx concentrations measured by standard CL monitors.[25] Furthermore, due to the strong absorption characteristics of NH3 and the local contamination source on sampling tubing and inlet filters, errors are inevitable for CL ambient data,[26] whereas our open-path WMS instrument is able to directly measure the samples without treating any of them.

Fig. 12. (color online) Scatter NO concentration plots of WMS against CL at 1-hour time resolution, including regression line (red) and 1:1 reference line (black).
Fig. 13. (color online) Scatter NO2 concentration plots of WMS against CL at 1-hour time resolution, including regression line (red) and 1:1 reference line (black).
Fig. 14. (color online) Scatter NH3 concentration plots of WMS against CL at 1-hour time resolution, including regression line (red) and 1:1 reference line (black).
5. Conclusions

By applying the time division multiplexing technique to the WMS, an optical instrument for simultaneously monitoring NO, NO2, and NH3 in the air is developed and demonstrated. To realize the gas detection at a sub-ppb level, the QCLs operating in the mid-infrared wavelength region are used, where the target gases have their stronger absorption lines. The time division multiplexing technology reduces the average light-emitting time of QCLs and the produced heat and improves the thermal stability in continuous operation. The instrument shows an excellent linear response to gas sample concentration over a large range. With an optimum integration time of about 100 s, the detection limit for each of all three gases is less than 0.2 ppb. The measurements are well consistent with the measurements by a commercial CL instrument. It should also be noted that our nondestructive open-path design minimizes the challenges of gas sampling artifacts in contrast to extractive sensors. The stability and reliability of the instrument are tested by field measurements and demonstrate that this detecting system is suitable for real-time monitoring NO, NO2, and NH3 in the air with the advantages of quick response, excellent stability and low minimal detection limit. Our future work will focus on continuous monitoring applications in several cities with different meteorological and geographic conditions for several months. Investigations on the reaction and relation among NO, NO2, and NH3 will also be performed to study their influence on the atmosphere.

Reference
[1] Cabot A Marsal A Arbiol J Morante J R 2004 Sensors and Actuators 99 74
[2] Yang X T Xie W Q Yuan Z G 2016 Optik 127 3788
[3] Dong L Spagnolo V Lewicki R Tittel F K 2011 Opt. Express 19 24037
[4] Peng W Y Sur R Strand C L Spearrin R M Jeffries J B Hanson R K 2016 Appl. Phys. 122 188
[5] He Y Jin C J Kan R F Liu J G Liu W Q Hill J Jamie I M Orr B J 2014 Opt. Express 22 13170
[6] Shadman S Rose C Yalin A P 2016 Appl. Phys. 122 194
[7] Xia H Dong F Z Wu B Zhang Z R Pang T Sun P S Cui X J Han L Wang Y 2015 Chin. Phys. 24 034204
[8] Andersson M Persson L Svensson T Svanberg S 2007 Rev. Sci. Instrum. 78 113107
[9] Mei L Svanberg S 2015 Appl. Opt. 54 2234
[10] Qu D S Hong Y J Wang G Yand Pan H 2017 Chin. Phys. 26 064204
[11] Liu C W Zhang J C Yan F L Jia Z W Zhao Z B Zhuo N Liu F Q Wang Z G 2017 Chin. Phys. Lett. 34 034209
[12] Zhao Y Zhang J C Jia Z W Liu Y H Zhuo N Zhai S Q Liu F Q Wang Z G 2016 Chin. Phys. Lett. 33 124201
[13] Tuzson B Zeyer K Steinbacher M McManus J B Nelson D D Zahniser M S Emmenegger L 2013 Atmos. Meas. Tech. 6 927
[14] Wang Y Nikodem M Zhang E Zhang E Cikach F Bames J Comhair S Kao G Wysocki G 2015 Sci. Rep. 5 9096
[15] Jágerská J Jouy P Tuzson B Looser H Mangold M Soltic P Hugi A Brönnimann R Faist J Emmenegger L 2015 Opt. Express 23 1512
[16] Panella B Maas D Brändle H Galletti B 2013 Optical Instrumentation for Energy and Environmental Applications November 3–7, 2013 Tucson, USA in Renewable Energy and the Environment, OSA Technical Digest (OSA, 2013), paper EW1A.3
[17] Owen K Farooq A 2014 Appl. Phys. 116 371
[18] Miller D J Sun K Tao L Khan M A Zondlo M A 2014 Atmos. Meas. Tech. 7 81
[19] Wei M Ye Q H Kan R F Chen B Yang C G Xu Z Y Chen X Ruan J Fan X L Wang W Hu M Liu J G 2016 Chin. Phys. 25 094210
[20] Chinese Ministry of Environmental Protection 2013 National Environmental Standard HJ654-2013
[21] Werle P O Mücke R Slemr F 1993 Appl. Phys. 57 131
[22] Sanchez N P Yu Y Dong L Griffin R Tittel F K 2016 Proc. SPIE 9755 975508
[23] Pakmanesh N Cristescu S M Ghorbanzadeh A Harren F J M Mandon J 2016 Appl. Phys. 122 10
[24] Villena G Bejan I Kurtenbach R Wiesen P Kleffmann J 2012 Atmos. Meas. Tech. 5 149
[25] Dunlea E J Herndon S C Nelson D D Volkamer R M San Martini F Sheehy P M Allwine E J 2007 Atmos Chem. Phys. 7 2691
[26] Williams E J Sandholm S T Bradshaw J D Schendel J S Langford A O Quinn P K LeBel P J Vay S A Roberts P D Norton R B Watkins B A Buhr M P Parrish D D Calvert J G Fehsenfeld F C 1992 J. Geophys. Res. 97 11591