Open Access
Add to BibSonomyAbstract
We demonstrate a second-harmonic-generation (SHG) based method for the detection of gaseous elemental mercury by using a newly available green diode laser. Multimode ultraviolet radiation at 253.7 nm is generated through a process of SHG. Correlation spectroscopy is introduced into the scheme to guarantee the measurement accuracy. The limit of detection achieved is 0.6 μg/m3 (0.07 ppb) for 1-m pathlength and 10-s integration time. The measurement accuracy is estimated to be 1.2%. The linear response range is estimated to be 0~60 μg/m2 (6.7 ppb·m), within which the linearity error is less than 1%. Real-time monitoring of mercury volatilization is demonstrated with a time resolution of 1 s. The results of performance characterization show that the proposed method has great potentials for mercury sensing in environmental and industrial fields.
© 2016 Optical Society of America
1. Introduction
Mercury is recognized as a toxic pollutant having adverse effects on human health, especially on the sensitive nervous system and kidneys [1]. Anthropogenic sources contribute to about 30% annual mercury emissions to the atmosphere [2]. In order to promote reduction of anthropogenic mercury emissions, Minamata Convention on Mercury, a multilateral environmental treaty was signed by more than 120 countries in 2013 and will enter into force after 50 countries have ratified [3]. Effective control of mercury emissions requires high-performance mercury sensors. Mercury in the air is commonly grouped into three categories: gaseous elemental mercury (GEM), reactive gaseous mercury (RGM) and particle-bound mercury (PBM) [4]. Detection of GEM is of particular importance as it is the most common form (>95%) in the atmosphere [5]. GEM is fairly stable in the atmosphere with a residence time of several months and can readily transport long distances leading to global pollution [6]. In addition, coal-combustion has been proven to be a major anthropogenic mercury source [2]; however, the effective removal of GEM in flue gas is rather difficult due to its low reactivity and solubility. Hence, it calls for effective methods for measuring GEM to efficiently monitor the mercury present in the air and evaluate the removal efficiency of mercury in industrial emissions.
To meet the demands of GEM detection various methods have been developed most of which are based on optical principles, mainly the spectroscopic methodology [5]. Cold vapor atomic absorption spectroscopy (CVAAS) and cold vapor atomic absorption spectroscopy (CVAFS) are the two most commercially available and widely used methods [7–10]. When employed for measuring mercury in the atmosphere where the background GEM levels are typically ~2 ng/m3 [11], CVAFS is often preferred as it can provide a detection limit at sub-ng/m3 levels. Both CVAAS and CVAFS require sample preconcentration through gold amalgamation to avoid interference from other relevant gases, which, however, is not desirable for long-term monitoring with timely data update. Zeeman-modulated atomic absorption spectroscopy (ZAAS) and differential optical absorption spectroscopy (DOAS) are another two commercially available methods [7, 12]. Although their sensitivities are not as good as CVAFS, the ZAAS and DOAS dispense with the preconcentration unit and thus have the advantages of simplicity and quick response. To meet the different requirements on the measurement of GEM other methods have also been developed such as cavity ring-down spectroscopy (CRDS) [13–15], laser induced fluorescence (LIF) [16, 17], laser-induced breakdown spectroscopy (LIBS) [18], light detection and ranging (LIDAR) [19, 20] and gas chromatography/mass spectrometry (GC/MS) [21, 22]. In order to reduce the cost of measurement, numerous researches have been focused on the development of micro-fabricated GEM sensors using gold thin film or gold nanostructures. These GEM micro-sensors are typically based on surface plasmon resonance [23, 24], conductivity [25–27] and piezoelectric devices such as quartz crystal microbalance and surface acoustic wave sensors [28–30].
In the last few decades, diode-laser-based gas sensing techniques have been greatly developed and widely used because of their high sensitivity, high selectivity and high speed [31, 32]. When diode lasers (DLs) below 600 nm were not commercially available, mercury measurement using DLs was initially demonstrated at 365 nm by laser radiation produced by sum frequency generation (SFG) [33], although the strongest absorption transition of mercury at 253.7 nm is mostly preferred. In 2000, Alnis et al for the first time demonstrated mercury detection at 253.7 nm based on SFG using a red DL and a newly available blue DL [34]. In their work, the ultraviolet (UV) radiation power generated was only 1 nW and the mode-hop-free tuning range was only 35 GHz which was not adequate to cover the whole absorption feature of mercury (80 GHz is required). Subsequently, Carruthers et al developed SFG-DL-based UV lasers at 253.7 nm extending the UV power to 50 nW [35]. In 2007, Anderson et al developed a SFG-DL-based mercury sensor and achieved for the first time a wide enough mode-hop-free scan range (up to 100 GHz) allowing for acquisition of sufficient off-resonant baseline and thus effective discrimination against broadband attenuation coming from absorption or scattering [36]. They also demonstrated for the first time the capability of DL-based mercury sensors for in site monitoring the atomic mercury in the exhaust stream of a coal-fired combustor [36]. Although 253.7 nm laser radiations can also be produced through two successive frequency-doubling (second-harmonic generation, SHG) steps [37, 38], they are not suitable for mercury detection due to their narrow tuning range of only a few GHz. In addition, compared with SFG based methods [34–36], the two-stage SHG would involve a more complex and expensive system.
Very recently, Fabry-Perot (FP) type green DLs around 507 nm have been commercially available and used by Almog et al to produce 253.7 nm laser radiation with a single frequency-doubling process [39]. Since all the available green DLs operated with multimode emissions, in Almog’s work an external cavity configuration was employed to generate single-mode laser light with a mode-hop free tuning range of a few GHz, which, however, was not adequate for mercury sensing. In fact, some methods have been developed to directly use the low-cost multimode DLs for gas sensing. One of them is termed multimode absorption spectroscopy (MUMAS) in which all the longitudinal modes of a DL are simultaneously tuned over a mode interval and the total absorption of intensity is detected [40, 41]. However, this approach requires a well-behaved laser source and thus its application for mercury detection would be hampered by the frequent and random mode hops of FP type green DLs. Another gas sensing technique using multimode DLs is multimode diode laser correlation spectroscopy (MDL-COSPEC) which has been successfully employed for single and multiple gas detections [42, 43]. We have recently demonstrated the capability of MDL-COSPEC for mercury detection [44, 45]. In our previous works, UV laser radiations were generated by SFG using blue and red MDLs. The frequent mode-hop behavior of FP MDLs normally considered as a poor quality was turned into advantage, as it could provide an off-resonant baseline which helped to eliminate interferences from other gas species. Furthermore, mercury absorption signals were readily acquired by interacting a set of laser modes to the sharp absorption line of mercury, yielding high system stability.
In this paper, we demonstrate measurements of GEM by a SHG-based 253.7 nm laser radiation using a newly available 507 nm green DL. Compared with the mercury sensing systems based on SFG, the scheme using SHG can provide advantages of high generation of UV laser power and simplicity for laser alignment. MDL-COSPEC technique is employed to guarantee the measurement accuracy. The experimental setup is described in detail and the system performance for mercury detection is evaluated in terms of detection limit, accuracy, linearity and response time.
2. Principle of MDL-COSPEC
The principle of MDL-COSPEC has been described previously [42–45] and will only be outlined here. When using a MDL with longitudinal modes and in the condition of low absorbance, the effective absorption cross section can be expressed as
where is the gas absorption cross section corresponding to the nth mode, denotes the intensity partition coefficient of the nth mode and obeys . It indicates that the total effective absorption is simply a sum of power-weighted absorptions by all individual modes. Even if the of each individual mode varies largely, the total output power of the laser almost remains unchanged because of the highly organized antiphase dynamics of the mode oscillation [46].To perform MDL-COSPEC, the MDL radiation is split into two beams passing through the sample gas to be analyzed and the reference gas containing well-calibrated target gas, respectively. The sample and the reference laser beams are simultaneously detected. Thus, is always identical for both sample and reference paths, although it varies with time due to the change of and even mode hops. Consequently, the target gases in the two optical paths would generate absorption signals well correlated in line shape. The correlation can be well maintained even when the absorption line shape is distorted due to mode hops or overlaps of absorptions from different laser modes. Also, line shape distortions due to the broad linewidth and wavelength-tuning nonlinearity of the laser would not affect the degree of correlation and thus have no impact on the gas measurement. By contrast, signals originating from interfering gases or light intensity variations having totally different correlations can be readily discriminated. By comparing the recorded absorption signals of the two paths, we can identify the target gas in the sample and determine its concentration.
Under the condition of weak absorption, the ratio between the path-integrated concentrations can be obtained by division:
where is the gas number density, is the optical pathlength through the gas, is the absorption amplitude, and the subscripts S and R denote the physical parameters belonging to the sample and the reference paths, respectively. Equation (2) is the key expression in MDL-COSPEC, by using which the concentration of the target gas in the sample can be retrieved through a linear-regression procedure [42].3. Experimental
3.1 Setup
A schematic diagram of the experimental setup is shown in Fig. 1. The setup is similar to those employed in our previous works [44, 45] except for the part of light source. Thus, the laser source is here described in detail and the other parts of the system are briefly outlined. Tunable multimode UV radiation covering 253.65 nm was generated by frequency-doubling a green MDL (Toptica, LD-0505, 80 mW) in a Beta-Barium-Borate (BBO) crystal (Caston Inc.). The laser was driven by a low-noise current driver (Thorlabs, LDC205C). The original center wavelength of the laser was about 505 nm. In order to achieve the required wavelength around 507.3 nm, a Littrow-type external cavity was constructed with a reflective holographic grating (2400/mm). The grating mount (Thorlabs, KC1-PZ) had screw and piezoelectric adjusters. By adjusting the angle of the grating, the center wavelength of the laser was set to 507.3 nm. Another function of the external cavity was to reduce the number of laser modes and thus increase the effective absorption cross section defined by Eq. (1). In order to achieve a more stable laser emission, the center wavelength of the laser gain curve was adjusted to about 506 nm (a wavelength closer to 507.3 nm) by setting the operating temperature of the laser to 45 °C using a temperature controller (Thorlabs, TED200C). The laser emission spectra were analyzed by a grating spectrometer (Andor, SR 750) with a resolution of 0.02 nm and accuracy 0.1 nm. Figure 2 shows the measured emission spectra of the MDL with and without the equipment of external cavity. It can be clearly seen that the Littrow cavity successfully shifted the center wavelength of the fundamental green laser radiation to the required 507.3 nm and concentrated most of the laser power in one laser mode. Due to the imperfect optical feedback there remained a few laser light around 505 nm (with a relative intensity <20%), which, however, would not affect the measurement accuracy by using COSPEC technique. The laser beam (40 mW) was focused by a BK7 glass lens (f = 25 mm) into the BBO crystal (5 × 5 × 7 mm3) which was cut at θ = 51.2° to achieve type I phase matching (o + o→e). The maximum UV light power obtained through the SHG process was measured to be 16.6 nW according to the radiant sensitivity of the photomultiplier (PMT, Hamamatsu, H10493). However, the BBO crystal was slightly titled to avoid severe optical fringe noise. Consequently, the conversion efficiency of SHG was reduced and only 4.9 nW UV light was generated and employed.
Fig. 1 Experimental setup for mercury detection using a frequency-doubled green diode laser combined with the MDL-COSPEC technique.
Fig. 2 Measured emission spectra of (a) the free-running MDL operating at 110 mA and 25°C and (b) the Littrow-cavity equipped MDL operating at 110 mA and 45°C.
The overlapped green and UV laser radiations exiting the BBO crystal were collimated with an f = 20 mm fused-silica lens and then separated by a triangular prism. The separated UV laser beam was split into two beams passing through the sample and the reference cells, respectively. According to Eq. (2), to vary absorption amplitudes, changing gas cell length is equivalent to changing gas concentration. Thus, both gas cells were filled with a consistent saturated concentration of mercury vapor at room temperature and atmospheric pressure. The concentration of mercury vapor was determined by the temperature of the liquid mercury in the cell [47]. The reference cell length was 2.144 mm, whilst sample cells with various thicknesses were employed to get different absorptions for proof-of-principle evaluation of the performance of the mercury sensing system. The mercury cell lengths were measured using a micrometer with an uncertainty of 0.005 mm. Each transmitted UV laser beam was detected by a PMT module in front of which a bandpass interference filter (Edmund Optics, 10 nm bandwidth, >15% transmission at 254 nm) was placed to block the scattered green laser radiation and ambient light. Signals from the two PMTs were simultaneously acquired by a data acquisition (DAQ) card (Adlinktech, DAQ-2010) mounted on a personal computer (PC).
3.2 Mercury absorption measurements
When using laser light at nW level, shot noise often becomes the dominant noise source. In this case, a straightforward and effective strategy to improve signal-to-noise ratio (SNR) is to increase the scan rate of laser wavelength and thus promote the averaging of noise. In this work, the laser wavelength was rapidly scanned at 800 Hz by a function generator which applied a ramp sweeping voltage to the laser driver changing the laser operating current from 80 mA to 120 mA. The resonance between laser modes and cavity modes would induce intense fluctuations of the laser output power. Figure 3(a) shows an example of the raw signals collected from the PMTs with an integration time of 4 s. The length of the sample mercury cell used was 0.520 mm. As can be seen from Fig. 3(a), absorption structures are overwhelmed by the mode coupling noise. In order to suppress the mode resonance, the length of the Littrow cavity was dithered by modulating the grating at 2.1 Hz through PZTs. In fact, suppressing the cavity coupling noise by dithering the cavity length is a common strategy in integrated cavity output spectroscopy and is also used here [48, 49]. When the mode coupling noise was effectively suppressed, the absorption structure stood out, as shown in Fig. 3(b).
Fig. 3 Example of recorded signals (a) with a constant external laser cavity length and (b) with the external laser cavity length dithered. The lengths of the sample and the reference cells were 0.520 and 2.144 mm, respectively. The integration time was 4 s.
In order to further increase the SNR, the signals shown in Fig. 3(b) were processed using a software-based low-pass Butterworth filter with the cutoff frequency at 6 kHz. The baseline was evaluated by a cubic fit to the absorption-free region of the filtered signal. Using the fitted baseline the absorption signal was then obtained by normalization. Figure 4 shows a pair of absorption signals belonging to the sample and the reference channels, respectively. According to Eq. (2), the sample-to-reference absorption amplitude ratio equals the corresponding path-integrated concentration ratio. The absorption amplitude ratio was determined by fitting the sample signal to the reference signal employing a linear-least-squares method combined with a multiple-linear-regression fitting approach [42, 45].
Fig. 4 Example of typical mercury absorption signal pairs obtained from the signals shown in Fig. 3(b).
4. Performance analysis
4.1 Detection limit and stability
To evaluate the detection limit of the proposed GEM detection system, Allan-Werle variance analysis was performed on continuous measurements of a certain low concentration of mercury vapor [50]. During the measurements the room temperature was measured to be 23.0 ± 0.3 °C with the corresponding saturated mercury vapor concentration being 16.9 ± 0.4 mg/m3. The length of the sample mercury cell used was 0.215 mm, yielding a path-integrated mercury concentration of 3.63 μg/m2. Figure 5(a) shows the plots of 1-h continuous measurement results with each measurement taking 1 s. Figure 5(b) shows the corresponding Allan-Werle deviation (square-root of the Allan-Werle variance) plots indicating a system stability time of 300 s. This stability time reflects the optimum time for signal averaging, after which the system drift starts to dominate and the average result becomes worse. Using the 300-s optimum integration time a best precision of 40 ng/m2 (4 ppt·m) can be achieved. The detection limit (3-times the Allan-Werle deviation) is 0.6 μg/m3 (0.07 ppb) for 1-m pathlength and 10-s integration time.
Fig. 5 (a) Plots of continuous 1-s measurement results of 3.63 μg/m2 mercury vapor during 1 h and (b) the corresponding Allan-Werle deviation plots.
4.2 Accuracy and linearity
To evaluate the measurement accuracy and linearity, measurements were performed by using mercury cells with various cell lengths yielding different path-integrated concentrations from 3.63 to 204.8 μg/m2. Figure 6(a) shows the plots of measured path-integrated concentrations versus calculated values. Each measured concentration was an average of 20 successive measurements with each taking 4 s. The calculated mercury concentration was obtained by directly multiplying the cell length and the saturated mercury vapor concentration at room temperature (16.9 mg/m3). The error bars are included but not clearly shown due to their small sizes. The error of the measured concentration was evaluated from the standard deviation of the 20 measurements. The uncertainty of the calculated concentration mainly originated from the measurement error of the cell length (0.005 mm). Because the sample and the reference cells placed in the same environment had a consistent saturated mercury vapor concentration inside, the measurement error of the room temperature was not considered for linearity analysis. It can be seen from Fig. 6(a), the mercury sensing system is linear under the concentration of 60 μg/m2 and present obvious nonlinear behavior with higher concentrations. The data between 0 and 60 μg/m2 were consequently selected and quantitatively analyzed through linear regression, as shown in Fig. 6(b). The regression analysis yields a slope of 0.988 ± 0.008, indicating a measurement accuracy of 1.2%. The intercept of 0.45 ± 0.25 μg/m2 is insignificant and indicates that the measurement results were not much biased. The coefficient of determination (R2) of the linear fit is nearly equivalent to 1, which proves that under the concentration of 60 μg/m2 (6.7 ppb·m) the system has a nearly perfect linear response. The fit residuals illustrate that the linearity error is within the range of 1%.
Fig. 6 (a) Plots of the measured mercury concentrations versus the calculated ones; (b) Plots selected from the low concentration region in (a) and the corresponding linear fit results.
4.3 Capability of real-time monitoring
Finally, to evaluate the capability of the system for real-time monitoring, a process of mercury volatilization was measured. A 20-mm sample cell with a cold finger holding a drop of mercury was used. The temperature of the cold finger was stabilized at 4 ± 0.3 °C corresponding to a saturated mercury vapor concentration of 3.21 ± 0.08 μg/m3. At first, the liquid mercury was shut off using the valve of the cold finger and the cell was purged with clean air. Then the valve of the cold finger was opened and the volatilized mercury vapor diffused into the cell. Figure 7 shows 2-h real-time measurement results of the GEM concentration in the cell after the cold finger valve was opened. Each measurement took 4 seconds, which yielded an efficient time resolution for monitoring the process of mercury volatilization. As shown in Fig. 7, there was a sharp rising at the beginning of the process, which was due to the quick diffuse of the initial mercury vapor inside the cold finger into the cell. As expected, the rising rate of the GEM concentration decreased with time. After 2 h, the mercury vapor in the cell got nearly saturated. This experiment demonstrated the capability of the proposed GEM sensing system for mercury monitoring applications requiring a time resolution at second level.
5. Discussion and conclusions
The detection limit achieved in this study (0.6 μg/m3 or 0.07 ppb for 1-m pathlength and 10-s integration time, determined by 3-times the Allan-Werle deviation) is 4 times better than that obtained by Anderson et al (0.1 ppb for 10-s integration time, determined by 1-σ noise level) [6]. The SFG-based UV laser radiation used in Anderson’s work is single mode and widely tuneable, which earns the system an advantage of self-calibration by completely capturing the mercury absorption spectrum [6]. However, for SFG the two fundamental laser beams have to be finely adjusted and shaped to achieve a high beam overlap and thus to guarantee the SFG efficiency. By contrast, for SHG the laser beam overlap is inherently not a problem. Accordingly, under the same condition SHG would always achieve a higher nonlinear conversion efficiency than SFG. In the current work, a nonlinear conversion efficiency of 4.2 × 10−7 is achieved even without any beam shaping, which is slightly better than that obtained in Anderson’s work (3.9 × 10−7). The UV laser power of 16.6 nW achieved by us is 3 times better than that obtained by Anderson et al (3.7 nW) [36], which is mainly due to the higher fundamental laser power we employed.
By comparison with other methods for GEM detection, the detection limit achieved in this work is at middle level. The most sensitive mercury sensing techniques are based on fluorescence spectroscopy (CVAFS [9, 10] and LIF [16, 17]) or GC/MS [21, 22] which can achieve detection limits at sub ng/m3 levels. Current commercial instruments for mercury monitoring at ambient levels are mostly based on CVAFS, which, however requires sample preconcentration on gold taking several minutes or more. The response time of a LIF-based system can be only a few seconds but it is bulk and expensive due to the utilization of pulsed dye lasers. The approaches based on GC/MS suffer from the complex measurement procedure and slow response, and therefore infrequently used for GEM detection. The detection limits achieved by GEM sensing methods based on absorption spectroscopy including CVAAS [8], ZAAS [7], DOAS [7], CRDS [13–15], LIDAR [19, 20] as well as the DL-based techniques [36, 44, 45] are at levels from ng/m3 to sub μg/m3. All these AS-based methods can provide a response time at second level (if optional sample preconcentration by gold amalgamation is not adopted). The differences of detection limits are mainly due to the different absorption pathlengths employed. Among these AS-based methods CRDS and LIDAR can achieve the best sensitivities at ng/m3 levels by using absorption pathlenghs from hundreds to thousands of meters. But the systems based on CRDS and LIDAR require bulk pulsed lasers and consequently are complex and expensive. The typical detection limits obtained by CVAAS, ZAAS, DOAS and DL-based mercury sensors are at sub μg/m3 levels. By comparison, the typical life time of a DL (>10000 hours) is much longer than the mercury lamp (typically 2000 hours) used in CVAAS, ZAAS and DOAS, which brings the DL-based mercury sensors an advantage of low cost in operation maintenance. The detection limits of LIBS [18] and micro-fabricated mercury sensors [23–30] are at μg/m3 levels. The LIBS requires a bulk and expensive pulsed laser and thus is seldom used for GEM detection. Although the sensitivities of micro-sensors are at present not good enough, they have still gained increasing attentions due to their advantages of low cost and small volume.
Compared with our latest work based on SFG [45], the detection limit obtained in the current study is only slightly improved, although many components of the systems are same. The reason is that in the current work the BBO crystal is slightly tilted off the optimum phase-matching angle to mitigate optical feedbacks, and consequently only 4.9 nW UV laser power was employed which is not much better than that used in our previous work. However, by using a beveled BBO crystal a greatly improved detection limit can reasonably be expected. As stated above, the SHG approach makes the mercury sensing system in this work much simpler than those in our previous works especially in terms of laser beam alignment [44, 45].
However, what’s more worth noting proposed in this paper is the new DL-based diagnostic method for GEM. For DL-based GEM sensors, the ideal laser source should be single mode around 253.7 nm, widely and rapidly tunable (>80 GHz tuning range with a >1 kHz scan rate) and high power (at mW level). However, the commercially available DLs at present only down to 370 nm and thus the commercial availability of 253.7 nm DLs cannot be expected in the near future. Thus, frequency conversion through a nonlinear process is currently the only approach to acquire a 253.7 nm laser source, like those did in Anderson’s and our previous works, although the generated UV laser powers were only a few nW. Up to now, the reported DL-based GEM sensors are all based on SFG. In fact, compared with SFG the strategy of SHG using 507 nm DLs would be more preferable because of its high efficiency of nonlinear conversion and simplicity for laser alignment. Very recently, 507 nm GaN-based green DLs have been commercially available. However, they are all FP type and operated with tens of laser modes (as shown in Fig. 2) and thus not suitable for sensitive mercury analysis. Distributed feedback techniques to achieve single-mode emission for GaN-based lasers are still under development. It is worth noting that GaN-based blue diode lasers have been commercially available for over two decades, but DFB-type ones are still not available. Although equipment with an external cavity can realize single mode operation, but the typical mode-hop-free tuning range of an ECDL is less than 50 GHz and not adequate to completely capture the mercury absorption spectrum. It is the reason that in Anderson’s work the mercury spectrum is obtained by scanning the near-infrared DFB DL rather than the violet ECDL [36]. Based on this context, in this paper we propose a method that can realize the combination of the newly available green DLs with the SHG approach for GEM detection. This new strategy includes four aspects: (a) Most of the emitted power of the DL is concentrated into a single mode by using a Littrow-type external cavity. (b) The DL is wavelength-tuned across the mercury transition by rapidly sweeping its injection current rather than mechanically changing the grating angle of the external cavity. The resulted high wavelength tuning rate can greatly promote the averaging of the shot noise which is often dominated when the laser power is at nW level. To our best knowledge, current scanning is for the first time applied to an ECDL for gas detection. (c) To eliminate broadband interferences, the off-resonant baseline is obtained by taking advantage of the mode hops of the DL. (d) Correlation spectroscopy is introduced into the scheme to deal with the imperfect single-mode emission and mode hops of the laser source, guaranteeing the measurement accuracy. By using this new strategy, we realize for the first the demonstration of a SHG-based GEM sensing method.
In conclusion, detection of gaseous elemental mercury has been for the first time demonstrated by using a second-harmonic-generation based method. The UV radiation at 253.7 nm was generated through SHG using a newly available green diode laser. The absorption signal was obtained by applying a fast current ramp to the diode laser and scanning its wavelength across the absorption line of mercury. The free-running Fabry-Perot type laser source used in this work operated with multimode emission and mode hops. Hence, correlation spectroscopy was employed to guarantee the measurement accuracy and stability. The sensitivity and stability of the system were evaluated by Allan-Werle variance analysis, showing that a detection limit of 0.6 μg/m3 or 0.07 ppb for 1-m pathlength and 10-s integration time could be achieved. The accuracy and linearity of the system were evaluated by measuring 15 different path-integrated concentrations of mercury. By a regression analysis, the measurement accuracy was evaluated to be 1.2% and the linearity error was less than 1% within the path-integrated concentration range of 60 μg/m2 (6.7 ppb·m). The developed GEM sensing system was used to measure a process of mercury volatilization validating its capability of real-time monitoring with a response time at second level. Compared with other diode-laser-based mercury sensing methods using SFG, the scheme based on SHG demonstrated in this work possesses obvious advantages of high efficiency of nonlinear conversion and simplicity for laser alignment.
Funding
Natural Science Foundation of China (NSFC) (61008027); Jiangsu Provincial Key Research and Development Program (BE2015653); Science and Technology Department of Zhejiang Province (2010R50007); Postdoctoral Scientific Research Developmental Fund of Heilongjiang Province (LBH-Q14069).
References and links
1. U.S. Agency for Toxic Substances and Disease Registry, “Toxicological profile for mercury,” (1999).
2. United Nations Environment Programme, “Global Mercury Assessment 2013: Sources, Emissions, Releases and Environmental Transport,” (2013).
3. United Nations Environment Programme, “Minamata convention on mercury,” (2013).
4. R. J. Valente, C. Shea, K. Lynn Humes, and R. L. Tanner, “Atmospheric mercury in the Great Smoky Mountains compared to regional and global levels,” Atmos. Environ. 41(9), 1861–1873 (2007). [CrossRef]
5. S. K. Pandey, K. H. Kim, and R. J. C. Brown, “Measurement techniques for mercury species in ambient air,” Trac-Trend Anal. Chem. 30, 899–917 (2011).
6. C. D. Holmes, D. J. Jacob, and X. Yang, “Global lifetime of elemental mercury against oxidation by atomic bromine in the free troposphere,” Geophys. Res. Lett. 33(20), L20808 (2006). [CrossRef]
7. D. L. Laudal, J. S. Thompson, J. H. Pavlish, L. A. Brickett, and P. Chu, “Use of continuous mercury monitors at coal-fired utilities,” Fuel Process. Technol. 85(6-7), 501–511 (2004). [CrossRef]
8. J. Wang, W. Wang, W. Xu, X. Wang, and S. Zhao, “Mercury removals by existing pollutants control devices of four coal-fired power plants in China,” J. Environ. Sci. (China) 23(11), 1839–1844 (2011). [CrossRef] [PubMed]
9. P. C. Swartzendruber, D. A. Jaffe, and B. Finley, “Improved fluorescence peak integration in the Tekran 2537 for applications with sub-optimal sample loadings,” Atmos. Environ. 43(22-23), 3648–3651 (2009). [CrossRef]
10. J. Zhu, T. Wang, R. Talbot, H. Mao, C. B. Hall, X. Yang, C. Fu, B. Zhuang, S. Li, Y. Han, and X. Huang, “Characteristics of atmospheric Total Gaseous Mercury (TGM) observed in urban Nanjing, China,” Atmos. Chem. Phys. 12(24), 12103–12118 (2012). [CrossRef]
11. K. H. Kim, R. Ebinghaus, W. H. Schroeder, P. Blanchard, H. H. Kock, A. Steffen, F. A. Froude, M. Y. Kim, S. M. Hong, and J. H. Kim, “Atmospheric mercury concentrations from several observatory sites in the northern hemisphere,” J. Atmos. Chem. 50(1), 1–24 (2005). [CrossRef]
12. K. H. Kim, V. K. Mishra, and S. Hong, “The rapid and continuous monitoring of gaseous elemental mercury (GEM) behavior in ambient air,” Atmos. Environ. 40(18), 3281–3293 (2006). [CrossRef]
13. S. Q. Tao, F. J. Mazzotti, C. B. Winstead, and G. P. Miller, “Determination of elemental mercury by cavity ringdown spectrometry,” Analyst (Lond.) 125(6), 1021–1023 (2000). [CrossRef]
14. S. Spuler, M. Linne, A. Sappey, and S. Snyder, “Development of a cavity ringdown laser absorption spectrometer for detection of trace levels of mercury,” Appl. Opt. 39(15), 2480–2486 (2000). [CrossRef] [PubMed]
15. A. M. Pierce, C. W. Moore, G. Wohlfahrt, L. Hörtnagl, N. Kljun, and D. Obrist, “Eddy Covariance Flux Measurements of Gaseous Elemental Mercury Using Cavity Ring-Down Spectroscopy,” Environ. Sci. Technol. 49(3), 1559–1568 (2015). [CrossRef] [PubMed]
16. D. Bauer, P. Campuzano-Jost, and A. J. Hynes, “Rapid, ultra-sensitive detection of gas phase elemental mercury under atmospheric conditions using sequential two-photon laser induced fluorescence,” J. Environ. Monit. 4(3), 339–343 (2002). [CrossRef] [PubMed]
17. D. Bauer, S. Everhart, J. Remeika, C. T. Ernest, and A. J. Hynes, “Deployment of a sequential two-photon laser-induced fluorescence sensor for the detection of gaseous elemental mercury at ambient levels: fast, specific, ultrasensitive detection with parts-per-quadrillion sensitivity,” Atmos. Meas. Tech. 7(12), 4251–4265 (2014). [CrossRef]
18. C. Lazzari, M. De Rosa, S. Rastelli, A. Ciucci, V. Palleschi, and A. Salvetti, “Detection of mercury in air by time-resolved laser-induced breakdown spectroscopy technique,” Laser Part. Beams 12(03), 525–530 (1994). [CrossRef]
19. T. Nayuki, K. Marumoto, T. Fujii, T. Fukuchi, K. Nemoto, A. Shirakawa, and K. Ueda, “Development of a differential-absorption lidar system for measurement of atmospheric atomic mercury by use of the third harmonic of an LDS-dye laser,” Appl. Opt. 43(35), 6487–6491 (2004). [CrossRef] [PubMed]
20. L. Mei, G. Y. Zhao, and S. Svanberg, “Differential absorption lidar system employed for background atomic mercury vertical profiling in South China,” Opt. Lasers Eng. 55, 128–135 (2014). [CrossRef]
21. J. Cai, G. Ouyang, Y. Gong, and J. Pawliszyn, “Simultaneous sampling and analysis for vapor mercury in ambient air using needle trap coupled with gas chromatography-mass spectrometry,” J. Chromatogr. A 1213(1), 19–24 (2008). [CrossRef] [PubMed]
22. R. Gras, J. Luong, and R. A. Shellie, “Direct measurement of trace elemental mercury in hydrocarbon matrices by gas chromatography with ultraviolet photometric detection,” Anal. Chem. 87(22), 11429–11432 (2015). [CrossRef] [PubMed]
23. J. Z. James, D. Lucas, and C. P. Koshland, “Gold nanoparticle films as sensitive and reusable elemental mercury sensors,” Environ. Sci. Technol. 46(17), 9557–9562 (2012). [CrossRef] [PubMed]
24. J. S. Crosby, D. Lucas, and C. P. Koshland, “Fiber optic based evanescent wave sensor for the detection of elemental mercury utilizing gold nanorods,” Sensors Actuat. B 181, 938–942 (2013). [CrossRef]
25. B. Mazzolai, V. Mattoli, V. Raffa, G. Tripoli, D. Accoto, A. Menciassi, and P. Dario, “A microfabricated physical sensor for atmospheric mercury monitoring,” Sensors Actuat. A 113(3), 282–287 (2004). [CrossRef]
26. M. Sarajlić, Z. G. Duric, V. B. Jovic, S. P. Petrovic, and D. S. Dordevic, “An adsorption-based mercury sensor with continuous readout,” Microsyst. Technol. 19(5), 749–755 (2013). [CrossRef]
27. M. J. Griffin, K. M. M. Kabir, V. E. Coyle, A. E. Kandjani, Y. M. Sabri, S. J. Ippolito, and S. K. Bhargava, “A nanoengineered conductometric device for accurate analysis of elemental mercury vapor,” Environ. Sci. Technol. 50(3), 1384–1392 (2016). [CrossRef] [PubMed]
28. D. P. Ruys, J. F. Andrade, and O. M. Guimaraes, “Mercury detection in air using a coated piezoelectric sensor,” Anal. Chim. Acta 404(1), 95–100 (2000). [CrossRef]
29. Y. M. Sabri, S. J. Ippolito, J. Tardio, V. Bansal, A. P. O’Mullane, and S. K. Bhargava, “Gold nanospikes based microsensor as a highly accurate mercury emission monitoring system,” Sci. Rep. 4, 6741 (2014). [CrossRef] [PubMed]
30. K. M. M. Kabir, Y. M. Sabri, G. I. Matthews, L. A. Jones, S. J. Ippolito, and S. K. Bhargava, “Selective detection of elemental mercury vapor using a surface acoustic wave (SAW) sensor,” Analyst (Lond.) 140(16), 5508–5517 (2015). [CrossRef] [PubMed]
31. P. Werle, “A review of recent advances in semiconductor laser based gas monitors,” Spectrochim. Acta A 54(2), 197–236 (1998). [CrossRef]
32. J. Hodgkinson and R. P. Tatam, “Optical gas sensing: a review,” Meas. Sci. Technol. 24(1), 012004 (2013). [CrossRef]
33. J. Franzke, “Sum frequency generation at 365 nm by two diode lasers applied to the detection of mercury,” Spectrochim. Acta B At. Spectrosc. 53(11), 1595–1599 (1998). [CrossRef]
34. J. Alnis, U. Gustafsson, G. Somesfalean, and S. Svanberg, “Sum-frequency generation with a blue diode laser for mercury spectroscopy at 254 nm,” Appl. Phys. Lett. 76(10), 1234–1236 (2000). [CrossRef]
35. A. E. Carruthers, T. K. Lake, A. Shah, J. W. Allen, W. Sibbett, and K. Dholakia, “Single-scan spectroscopy of mercury at 253.7 nm by sum frequency mixing of violet and red microlensed diode lasers,” Opt. Commun. 255(4-6), 261–266 (2005). [CrossRef]
36. T. N. Anderson, J. K. Magnuson, and R. P. Lucht, “Diode-laser-based sensor for ultraviolet absorption measurements of atomic mercury,” Appl. Phys. B 87(2), 341–353 (2007). [CrossRef]
37. J. Paul, Y. Kaneda, T. L. Wang, C. Lytle, J. V. Moloney, and R. J. Jones, “Doppler-free spectroscopy of mercury at 253.7 nm using a high-power, frequency-quadrupled, optically pumped external-cavity semiconductor laser,” Opt. Lett. 36(1), 61–63 (2011). [CrossRef] [PubMed]
38. M. V. Romalis, W. C. Griffith, J. P. Jacobs, and E. N. Fortson, “New limit on the permanent electric dipole moment of 199Hg,” Phys. Rev. Lett. 86(12), 2505–2508 (2001). [CrossRef] [PubMed]
39. G. Almog, M. Scholz, W. Weber, P. Leisching, W. Kaenders, and T. Udem, “A simplified scheme for generating narrow-band mid-ultraviolet laser radiation,” Rev. Sci. Instrum. 86(3), 033110 (2015). [CrossRef] [PubMed]
40. Y. Arita and P. Ewart, “Infra-red multi-mode absorption spectroscopy of acetylene using an Er/Yb:glass micro-laser,” Opt. Express 16(7), 4437–4442 (2008). [CrossRef] [PubMed]
41. J. H. Northern, S. O’Hagan, B. Fletcher, B. Gras, P. Ewart, C. S. Kim, M. Kim, C. D. Merritt, W. W. Bewley, C. L. Canedy, J. Abell, I. Vurgaftman, and J. R. Meyer, “Mid-infrared multi-mode absorption spectroscopy using interband cascade lasers for multi-species sensing,” Opt. Lett. 40(17), 4186–4189 (2015). [CrossRef] [PubMed]
42. X. Lou, G. Somesfalean, and Z. Zhang, “Gas detection by correlation spectroscopy employing a multimode diode laser,” Appl. Opt. 47(13), 2392–2398 (2008). [CrossRef] [PubMed]
43. X. T. Lou, G. Somesfalean, B. Chen, Y. G. Zhang, H. S. Wang, Z. G. Zhang, S. H. Wu, and Y. K. Qin, “Simultaneous detection of multiple-gas species by correlation spectroscopy using a multimode diode laser,” Opt. Lett. 35(11), 1749–1751 (2010). [CrossRef] [PubMed]
44. X. Lou, G. Somesfalean, S. Svanberg, Z. Zhang, and S. Wu, “Detection of elemental mercury by multimode diode laser correlation spectroscopy,” Opt. Express 20(5), 4927–4938 (2012). [CrossRef] [PubMed]
45. H. Z. Lin, X. T. Lou, W. J. Zhong, and S. L. He, “Continuous monitoring of elemental mercury employing low-cost multimode diode lasers,” Meas. Sci. Technol. 26(8), 085501 (2015). [CrossRef]
46. A. M. Yacomotti, L. Furfaro, X. Hachair, F. Pedaci, M. Giudici, J. Tredicce, J. Javaloyes, S. Balle, E. A. Viktorov, and P. Mandel, “Dynamics of multimode semiconductor lasers,” Phys. Rev. A 69(5), 053816 (2004). [CrossRef]
47. M. L. Huber, A. Laesecke, and D. G. Friend, “Correlation for the vapor pressure of mercury,” Ind. Eng. Chem. Res. 45(21), 7351–7361 (2006). [CrossRef]
48. A. O’Keefe, J. J. Scherer, and J. B. Paul, “cw Integrated cavity output spectroscopy,” Chem. Phys. Lett. 307(5-6), 343–349 (1999). [CrossRef]
49. X. T. Lou, Y. K. Dong, D. C. Wu, J. T. Wei, and Z. W. Lu, “Integrated cavity output spectroscopy by using a sub-centimeter short optical cavity combined with a free-running Fabry-Perot diode laser,” Appl. Phys. B 121(2), 171–178 (2015). [CrossRef]
50. P. Werle, “Accuracy and precision of laser spectrometers for trace gas sensing in the presence of optical fringes and atmospheric turbulence,” Appl. Phys. B 102(2), 313–329 (2011). [CrossRef]
