- Email: [email protected]
Laser-induced fluorescence instrument for measuring atmospheric SO2
ARTICLE IN PRESS
Atmospheric Environment 39 (2005) 3177–3185 www.elsevier.com/locate/atmosenv
Technical note
Laser-induced fluorescence instrument for measuring atmospheric SO2 Yutaka Matsumi, Hiroyuki Shigemori, Kenshi Takahashi Solar-Terrestrial Environment Laboratory and Graduate School of Science, Nagoya University, 3-13, Honohara, Toyokawa 442-8507, Japan Received 5 August 2004; received in revised form 28 January 2005; accepted 14 February 2005
Abstract We report on the development of a high-sensitive detection system for measuring atmospheric SO2 using a laserinduced fluorescence (LIF) technique at around 220 nm. Second harmonics of a tunable broad-band optical parametric oscillator (OPO) pumped by the third harmonic of a Nd:YAG laser is used as a fluorescence excitation source. The laser wavelength is alternatively tuned to the peak and the bottom wavelengths in the photoabsorption spectrum of SO2 at 220.6 and 220.2 nm, respectively, and the difference signal at the two wavelengths is used to extract the SO2 concentration. This procedure can give a good selectivity for SO2 and avoid interferences of fluorescent or particulate species other than SO2 in the sample air. The SO2 instrument developed has a sensitivity of 5 pptv in 60 s and S=N ¼ 2: The practical performance of the detection system is tested in the suburban area. The inter-comparisons between the LIF instrument and a commercial instrument using Xe flash lamp excitation for the fluorescence detection have been performed. The correlation between two instruments is measured up to 70 ppbv. A good linear relationship between the LIF measurements and commercial instrument measurements is obtained. r 2005 Published by Elsevier Ltd. Keywords: Sulfur dioxide; OPO; Detection limit; Instrument development
1. Introduction In the atmosphere, sulfur in both gaseous and aerosol forms impacts regional and global chemistry, climate change, as well as the health of various living organism. Anthropogenic emissions in the form of SO2 are about 75% of the total sulfur emission budget (Brasseur et al., 1999). The source of SO2 is primarily from coal and petroleum combustion, petroleum refining and metal smelting operations. The anthropogenic emissions of SO2 are converted to H2SO4, which is the dominant Corresponding author. Tel.: +81 533 89 5192; fax: +81 533 89 5593. E-mail address: [email protected] (Y. Matsumi).
1352-2310/$ - see front matter r 2005 Published by Elsevier Ltd. doi:10.1016/j.atmosenv.2005.02.023
precursor responsible for acid precipitation. The acidic precipitation in the form of rain, fogs, and mists has long been implicated in playing some role in forest decline and foliar damage. SO2 is also a common intermediate in the oxidation of all the reduced sulfur compounds such as dimethylsulfide (DMS), hydrogen sulfide (H2S), carbon disulfide (CS2), etc. The released sulfur compounds are converted to sulfuric acid (H2SO4) through the gaseous SO2 form. The mixing ratios of SO2 in polluted urban and rural areas, particularly near power plants, smelters or paper mills, may exceed several parts per billion by volume (109, ppbv). In the remote troposphere, the concentrations of SO2 are typically less than 100 parts per trillion by volume (1012, pptv). To provide useful information
ARTICLE IN PRESS 3178
Y. Matsumi et al. / Atmospheric Environment 39 (2005) 3177–3185
with which we may broaden our understanding of the sulfur cycles, measurement techniques must be sensitive, precise, and accurate enough to qualify the SO2 concentrations over the wide range from several hundred ppbv to several pptv. Measurements of SO2 in the remote marine atmosphere are extremely important for elucidating the long-range transport of sulfur compounds far from source regions, and understanding of atmospheric sulfur oxidation and deposition. For example, Yvon and Saltzman (1996) measured the concentrations of SO2 and its biogenic precursor DMS in the tropical Pacific marine boundary layer, and analyzed the atmospheric sulfur cycling. Tu et al. (2004) measured the SO2 concentrations on transit flights during the NASA Transport and Chemical Evolution over the Pacific mission (TRACE-P), and investigated the long-range transport of SO2 in the Pacific. Various techniques for the detection of atmospheric SO2 have been used. In the Gas-Phase Sulfur Intercomparison Experiment (GASIE), seven techniques for the field measurement of trace SO2 were compared simultaneously over 1 month in 1994 using samples produced in situ by dynamic dilution (Stecher et al., 1997). The seven techniques were aqueous chemiluminescence (estimated detection limit: 3 pptv in 10–35 min with 500 l sample gas) (Jaeschke et al., 1997), pulsed fluorescence detector (described later) (Luke, 1997), isotope dilution-gas chromatograph mass spectrometer (1 pptv in 3–8 min with 1.7 l gas) (Bandy et al., 1993), mist chamber ion chromatograph (5–47 pptv in 5–30 min with 400 l gas) (Talbot et al., 1997), diffusion denuder sulfur chemiluminescence detector (10 pptv in 3 min with 3 l gas) (Benner et al., 1997), high performance liquid chromatography fluorescence detector (4 pptv in 4 min with 6 l gas) (Gallagher et al., 1997), and carbonate filter ion chromatograph (3–12 pptv in 90 min with 6000 l gas) (Ferek et al., 1997). An atmospheric pressure ionization mass spectrometer (APIMS) has been developed to determine atmospheric SO2 by Thornton et al. (2002) using a chemical ionization mass spectrometry (CIMS) technique. They have used radioactive nickel-63 for an electron source to produce secondary ions. They continuously added 34 16 S O2 as an internal standard. Their detection limit was o1 pptv in 1-s integration time. They indicated that the sensitivity of the CIMS technique was strongly dependent on water vapor in the air and the necessity of a dryer in the air inlet was suggested especially for marine boundary layer measurements. The CIMS approach for the determination of SO2 has been also reported by several groups (Jost et al., 2003; Hunton et al., 2000; Reiner et al., 1998; Eisele and Berresheim, 1992). Differential optical absorption spectrometry (DOAS) measurements of atmospheric SO2 were performed by Platt and Perner. The detection limit of SO2
was estimated to be 10 pptv in 20-min measurement with 5-km optical path (Platt and Perner, 1980). Fluorescence detection methods with the optical excitation to electronically excited states of SO2 have been used for the determination of SO2 concentrations in air. In the 1970’s Okabe and co-workers (Okabe et al., 1973; Schwartz et al., 1974) used the atomic resonance line of Cd (228.8 nm) or Zn (213.8 nm) discharge lamp for the optical excitation sources. Excited SO2 molecules fluoresce in a broad band continuum from 240 to 420 nm, with an emission peak at approximately 320 nm. A bandpass filter was used to isolate the emission radiation and transmit it to an orthogonally oriented photomultiplier tube (PMT). They found that the detector response was linear from at least 8.6 ppbv to 1.8 ppmv. The detection limit was several ppbv levels in 1-min integration time. In the commercial pulsed fluorescence (PF) detector for SO2, Thermo Environmental Instruments (Franklin, Massachusetts) Model 43S, SO2 molecules are electronically excited with the output of a xenon flash lamp pulsed at 10 Hz with 130 ms time width (Thermo Environmental Instruments, Inc., 1992). This instrument is called a Thermo Electron 43S. In this instrument a series of refractive interference filters isolates and passes excitation radiation in the 190–230 nm wavelength range of the xenon flash lamp. The emission from SO2 is monitored with a gated PMT. The sampling window gate is delayed by 30 ms from the start of the Xe flash lamp trigger to avoid the electrical noise associated with the flash. Luke (1997) succeeded to detect as little as 30 pptv SO2 in a 25-min sampling interval using a modified Thermo Electron 43S instrument in the GASIE intercomparison experiment, although the specification of the original instrument is 200 pptv in 3-min detection time. The laser-induced fluorescence (LIF) detection of SO2 was tested in the laboratory by Bradshaw et al. (1982) to provide an assessment for the atmospheric detection. The excitation wavelength was 222.2 nm. The laser system consisted of a Nd:YAG laser pumped dye laser, the fundamental of which was frequency doubled and this output, in turn, was then frequency mixed with the Nd:YAG fundamental at 1064 nm. Based on the laboratory experiments, they estimated the detection limit of SO2 to be several pptv in 1-min integration time with the laser energy of 1 mJ at 222.2 nm and the repetition rate of 10 Hz for the excitation laser light and 6 PMTs for fluorescence detection. Here, we describe design and performance of an LIF instrument for SO2 detection that uses frequencydoubled light of a broad-band optical parametric oscillator (OPO). We have used a two-wavelength measurement technique to extract the concentration of SO2 to avoid the interferences by other fluorescent species. The detection limit of the instrument for SO2 is 5 pptv in 60-s integration time and S=N ¼ 2:
ARTICLE IN PRESS Y. Matsumi et al. / Atmospheric Environment 39 (2005) 3177–3185
We used the ultraviolet laser light at the wavelength around 220 nm for the excitation of SO2 molecule to detect the fluorescence. Fig. 1 shows a schematic diagram of the SO2 detection system which we have developed. The laser system consists of a broad-band OPO (optical parametric oscillator, Continuum, Surelite-OPO) that is pumped by the third harmonic (355 nm) of a Nd:YAG laser (Continuum, Surelite II), which was used in the detection of atmospheric NO2 (Matsumi et al., 2001). The 355 nm light from the YAG laser provides about 170 mJ pulse1 at the repetition rate of 10 Hz. The broad-band OPO contains a BBO crystal and the wavelength is tuned by the angle of the crystal. The angle of the BBO crystal in the OPO is driven by a linear actuator (Sigma Koki, DMY-25) with a sine bar. The output of the OPO around 440 nm is frequency doubled with another angle-tuned BBO crystal. The angle of the BBO crystal for the frequency doubling is driven by a high-resolution rotation stage with a stepper motor. The angle positions of the OPO and frequencydoubling crystals are computer controlled. The ultraviolet laser light generated around 220 nm is isolated from the fundamental laser light (440 nm) by two dielectric coating mirrors (CVI, Y5-1025-45) and fed into a LIF cell. The energy of the ultraviolet laser light around 220 nm is about 1 mJ pulse1 at 10 Hz (10 mW). The time width of the ultraviolet laser pulse is about 5 ns. The ultraviolet laser light of the frequency-doubled OPO output is always monitored at the exit position of the fluorescence cell with a power meter (SCIENTECH, AC2501). The transit time for changing the ultraviolet laser wavelength between the peak and bottom
~ 220 nm BBO crystal
Iris
~ 440 nm 15 mJ / pulse LIF Cell
OPO laser
Air PMT SO2
355 nm 10 Hz
SO2 þ hnðlo219 nmÞ ! SOð3 S Þ þ Oð3 PÞ.
(1)
With the photoexcitation at wavelengths shorter than 219 nm, SO2 gives weak fluorescence. Around 220 nm, there is a structure which is suitable for the peak and bottom measurements. The linewidth of the ultraviolet laser light at 220 nm was measured to be 6 cm1 FWHM (full-width at half-maximum), using a monochromator (f ¼ 500 mm). The structure of the absorption spectrum shown in Fig. 2a is well reproduced in the excitation spectrum with the broad-band OPO system. This
σabs (10-18 cm2)
2.1. Excitation system
wavelengths is less than 0.2 s. Since the OPO is in solid state and does not need a circulation system of liquid dye solution, it is easy to operate. Fig. 2a shows the absorption spectrum of SO2 in the wavelength region of 215–225 nm (Manatt and Lane, 1993). Fig. 2b shows the fluorescence excitation spectrum of SO2 with the frequency-doubled OPO-YAG laser system under the condition of the SO2 concentration of 10 ppbv, where the OPO wavelength is scanned while monitoring the fluorescence intensity with a photomultiplier. The fluorescence excitation spectrum is not proportional to the absorption spectrum at wavelengths shorter than 219 nm. This is due to the predissociation of the upper electronic state of SO2:
Fluorescence Intensity (Arb.)
2. Description of the instrument
3179
60
(a)
40
20
0 5 (b)
Peak Wavelength
4 Bottom Wavelength
3 2 1 0
216
218
220
222
224
Wavelength (nm) YAG laser
Pump PM
Fig. 1. Schematic diagram of the SO2 instrument developed in this work. (PMT) photomultiplier tube; (PM) power meter.
Fig. 2. (a) Absorption cross-section (sabs ) of SO2 as a function of wavelength, which was taken by Manatt and Lane (1993). (b) Fluorescence excitation spectrum of SO2 measured with the instrument shown in Fig. 1 at SO2 concentration of 10 ppbv and total pressure of 10 Torr. The arrows indicate the bottom and peak wavelengths used in this study.
ARTICLE IN PRESS Y. Matsumi et al. / Atmospheric Environment 39 (2005) 3177–3185
3180
indicates that the spectral resolution of the frequencydoubled broad-band OPO system (6 cm1 equivalent to about 0.03 nm) is sufficient to separate the spectral peak and bottom positions in the SO2 spectrum around 220 nm. We have chosen 220.6 nm for the peak wavelength and 220.2 nm for the bottom wavelength. The ratio of the fluorescence intensities at the peak and bottom wavelengths is more than 10. For the peak absorption cross section of 3 1018 cm2 at 220.6 nm and the laser power of 1 mJ pulse1 with the beam diameter of 5 mm, it is estimated that about 0.25% of SO2 molecules in the laser irradiation zone is photoexcited. Therefore, the energy of our laser excitation system is far from the saturation level of the SO2 absorption.
2.2. Fluorescence monitoring system Fig. 3 shows a detailed schematic diagram of the fluorescence cell and the signal detection systems. Optical baffles with cone-shaped apertures are placed within both arms of the fluorescence cell to minimize detection of light scattered by the entrance and exit windows. The diameters of the aperture holes are 12 mm in the entrance baffle arm and 14 mm in the exit baffle arm. The fluorescence from SO2 lies in the 220–450 nm (Mettee, 1968). For the detection of the fluorescence from SO2, a side-on photomultiplier (PMT, Hamamatsu 1P28A) is used, which has sensitivity in the ultraviolet and visible wavelength region. The emitting image of the SO2 volume along the excitation laser beam in the cell is focused onto the photocathode of the PMT by a set of four quartz lenses (50 mm dia., f ¼ 100 mm). A band-pass optical filter (Corning, 7-54) is equipped in front of the PMT to isolate the fluorescence band from 240 to 420 nm. Since the fluorescence lifetime of SO2 is as short as about 40 ns with the excitation wavelength of 220 nm, a photon counting method cannot be used for the processing of the PMT signal output due to the pile-up phenomena. The output of the PMT is amplified and fed into an analog
Laser Beam
Laser Baffles PM Lens+Window
Pre-amp. Baratron Gauge
PMT
Optical Filter
Analog Gated Integrator
A/D
Computer
Fig. 3. Cross section of the fluorescence cell and fluorescence detection system.
gated integrator (Stanford Research, SR250) with the integration over 10 successive laser pulses. The gate of the integrator is opened at the rise of the laser pulse and its duration is 100 ns. The integrated signal is sampled with an A/D board and a microcomputer. 2.3. Gas handling system The ambient air is pulled into the fluorescence cell with a booster pump (ULVAC PMB001C) and a rotary pump (Alcatel 2021). The pumping speed at the cell pressure of 10 Torr is about 1000 l min1. The pressure in the cell is monitored with a capacitance manometer (MKS Baratron 122A, full scale 20 Torr). The sample flow rate in the fluorescence cell is 1.3 slm (standard liter per minute). For the calibration of the detection system, SO2 from a standard mixing ratio gas cylinder (Nihon Sanso, 1.03 ppmv) is diluted with the synthetic air flow. The flow rates of all the gases are controlled by mass flow meters. All the mass flow meters are calibrated using a soap film flow meter (STEC, SF1). The weight and size of the whole system of our prototype instrument are about 250 kg and 2 m width 1 m depth 1 m height, respectively.
3. Instrument performance 3.1. Instrument sensitivity Fig. 4 shows a typical example of the fluorescence intensity measurements with our SO2 instrument, when the SO2 gas concentration was 10 ppbv. The vertical scale is the fluorescence intensity in arbitrary units, while the horizontal axis is time. The OPO wavelength was fixed at the peak position of the SO2 absorption (220.6 nm) for 30 s and then the wavelength was changed and fixed at the bottom position (220.2 nm) for 30 s. The measurements at the peak and bottom positions were alternately repeated several times. Then, the SO2 gas concentration was reduced to zero to measure the background signal intensity. The differences between the fluorescence intensities at the peak and bottom wavelengths versus SO2 concentrations with the standard gas sample in the range of 0–140 ppbv are plotted in Fig. 5. The LIF signal intensities are linearly proportional to the SO2 concentrations up to 140 ppbv. The differences between the fluorescence intensities at the peak and bottom wavelengths versus SO2 concentrations in the range of 0–1500 pptv are plotted in Fig. 6. The detection sensitivity of LIF instrument, S SO2 ; is proportional to the product of the fluorescence quantum yield, Ff ; and the ratio of the pressures in the fluorescence cell and in the ambient air, Pcell and Pambient : SSO2 / Ff Pcell =Pambient .
(2)
ARTICLE IN PRESS Y. Matsumi et al. / Atmospheric Environment 39 (2005) 3177–3185
1.5
0.15 SO2 OFF
Fluorescence Intensity (Arb.)
Fluorescence Intensity (Arb.)
SO2 10 ppbv
1.0
0.5
0.10
0.05
0 0
200
400
600
Time (s) Fig. 4. Typical example of the fluorescence intensity measurements with our SO2 instrument. The sample SO2 gas concentration is 10 ppbv, which is generated by the dilution of the standard SO2 gas with the synthetic air. The fluorescence intensity is measured while alternatively tuning the laser wavelength at the peak (220.6 nm) and bottom (220.2 nm) positions every 30 s. Then the SO2 gas is reduced to zero to measure the back ground signal intensity.
15 Fluorescence Intensity (Arb.)
3181
10
5
0
50
100
1500
Fig. 6. Calibration plots of the fluorescence intensities at the peak wavelength (triangle), that at the bottom wavelength (square), and the difference of the two intensities (circle). The concentration range of SO2 is 0–1500 pptv. Error bars are one standard deviations.
where kr and knr are radiative and non-radiative rate constants of the photoexcited SO2, respectively, and kM q is a quenching rate constant for M such as O2 and N2. The radiative and non-radiative rate constants are 1.5 107 and 7.4 106 s1, respectively, with the excitation around 220 nm and the quenching rate constants are in the order of 1010 cm3 molecule1 s1 (Hui and Rice, 1972). According to expression (2), the optimum pressure for the LIF detection of SO2 around 220 nm is in the range of 1–100 Torr. The total pressure of 10 Torr at the fluorescence cell was used in most experiments in this study. The flow rate in the fluorescence cell is high enough to avoid the irradiation of an air sample by two successive laser pulses before it leaves the detection volume, since fragments from the photodissociation of some molecules such as sulfur compounds and aromatics can produce extra LIF.
150
SO2 Concentration (ppbv)
3.2. Minimum detectable limit
Fig. 5. Calibration plot of the fluorescence intensity which is the subtraction of the intensity at the bottom wavelength (220.2 nm) from that at the peak wavelength (220.6 nm). The concentration range of SO2 is 0–150 ppbv. The scale of the fluorescence intensity is normalized with the level when the SO2 concentration is 10 ppbv. Error bars are one standard deviations.
The fluorescence quantum yield of SO2 in the cell is determined by the equation: Ff ¼ kr =ðkr þ knr þ SkM q ½M Þ,
500 1000 Concentration (pptv)
(3)
Fig. 7 shows the fluorescence intensities recorded at the peak and bottom wavelengths at 60 pptv of SO2 concentration and the background fluorescence intensity. In the first 0–30 s period, the laser wavelength was fixed at the peak wavelength and in the 30–60 s period the laser wavelength was fixed at the bottom wavelength and then in the 60–160 s period the laser wavelength was fixed at the peak wavelength again. In the 0–100 s period the concentration of SO2 was kept to be 60 pptv, while at 100 s the supply of the SO2 gas was turned off and the background signal intensity was recorded. When the signal intensities are very weak, the dominant source for
ARTICLE IN PRESS Y. Matsumi et al. / Atmospheric Environment 39 (2005) 3177–3185
3182
0.51 counts s1 pptv1 from the estimation of the equivalent photon numbers for the vertical scale in Fig. 6. The background signal level was estimated to be 50 photons s1. The background signal came mainly from the wall and Rayleigh scattering of the laser light. The background level was stable on the time scale of the on/off measurements. With S=N ¼ 2; the minimum detectable limit of our SO2 instrument is calculated to be 5 pptv with the integration time of t ¼ 60 s and 1.6 pptv with t ¼ 600 s:
120 Fluorescence Intensity (counts s-1)
SO2 60 pptv
SO2 OFF
Bottom wavelength 80
40
3.3. Possible interferences Peak wavelength 0
50
100
150
Time (s) Fig. 7. Fluorescence intensities recorded at the peak and bottom wavelengths at 60 pptv of SO2 concentration and the background fluorescence intensity. In the 0–100 s period the concentration of SO2 was kept to be 60 pptv, while at 100 s the supply of the SO2 gas was turned off and the background signal intensity is recorded. The vertical scale is equivalent photon number detected by the PMT per second.
the fluctuation of the signal intensities is photon number statistics described by a Poisson distribution. Under this condition, the standard deviation pffiffiffiffisffi of the photon population can be taken as s ¼ N ; where N is the number of photons detected by the PMT for a given integration time. We calculate the standard deviation s of the signal intensity and then determine the relationship between the number of photons per second and the signal intensity detected using the gated integrator assuming that the standard of the signal pffiffiffiffideviation ffi intensities follow the s ¼ N expression. Thus the equivalent photon numbers are estimated from the signal intensity from the gated integrator as shown in Fig. 7. When the predominant source of the noise is fluctuations of the detected photon numbers with Poisson distributions, the minimum detectable SO2 mixing ratio of the instrument, [SO2]min, is expressed by pffiffiffiffiffiffiffi S=N Sbg ½SO2 min ¼ pffiffi , t ðC peak C bottom Þ
(4)
where S/N is the required signal-to-noise ratio, C peak and C bottom are the sensitivity factors of the LIF signal of SO2 for the peak and bottom wavelengths, respectively, in counts s1 pptv1, S bg is the background signal level of the instrument in photons s1, and t is the averaging time of the instrument in s. The value of (C peak 2C bottom ) was determined to be
The possible fluorescent species with the excitation around 220 nm are compounds such as aromatics, and molecules with unsaturated chemical bonds. No testing for the interference of aromatics has been done with the present instrument. In principle the interference from species other than SO2 can be eliminated by the measurements with tuning the laser wavelength at the top and bottom wavelengths in the SO2 photoabsorption spectrum. We have checked the interference by nitrogen dioxide (NO2) molecules. We could detect no fluorescence at both peak and bottom wavelengths at the NO2 concentration of 9.7 ppmv. NO2 molecules dissociate at the excitation wavelengths shorter than 400 nm and emits no fluorescence (Okabe, 1978). We also checked the interference by nitric oxide (NO) molecules. No emission was detected from NO molecules at both peak and bottom wavelengths at the NO concentration of 1.0 ppmv. The interference measurements indicate that over 1000 times more NO2 or NO is needed to produce an equivalent signal to that from one part SO2. The fluorescence of SO2 is quenched by water molecules as well as N2 and O2 molecules. The change of H2O concentration in the ambient air is large under various conditions. We have checked the dependence of the fluorescence intensity of SO2 on the water vapor concentration. Fig. 8 shows the fluorescence intensity of SO2 as a function of water vapor concentration, when the concentration of SO2 is 10 ppbv. The fluorescence intensity of SO2 is reduced by about 10% by the presence of 16 g m3 H2O which corresponds to the relative humidity of 90% at 298 K. Thus, the fluorescence quenching of SO2 by water vapor is not so fast when the excitation laser wavelength is 220.6 nm. Since this negative interference by water vapor is quantitative, it can be corrected by the simultaneous measurements of SO2 and H2O in sample air, when very accurate values of SO2 concentration are necessary. Luke (1997) also suggested a decrease of 7% and 15% in the response of the pulsed fluorescence detector for SO2 (Thermo Electron 43S) at the relative humidities of 35% and 50%, respectively, at 20–25 1C.
ARTICLE IN PRESS Y. Matsumi et al. / Atmospheric Environment 39 (2005) 3177–3185
Relative Humidity (%) at 298 K 20 40 60 80
15 100 SO2 concentration (ppbv)
0
Fluorescence Intensity (arb.)
1
0.5
0
3183
5 10 15 Absolute Humidity (g m-3)
Fig. 8. Fluorescence intensity of SO2 as a function of water vapor concentration, when the concentration of SO2 is 10 ppbv.
10
5
0 15:00
16:00
17:00
Local time Fig. 9. Ambient measurement data obtained on December 18, 2002. The results of the SO2 concentration measurements with our LIF instrument (open square) and with the commercial pulsed fluorescence (PF) detector for SO2 (closed circles).
4. Ambient measurement and inter-comparison Ambient measurements were performed in December of 2002 in the south yard in Toyokawa campus of Nagoya University. The sampling point is located at a height of 2.0 m from the ground and 10 m south from the laboratory building which is about 9 m tall. The yard is covered with lawn and has an area of about 4000 m2. Toyokawa campus is located in a suburban factory area and Tomei expressway runs 1 km north of the campus. The length of the PTFE tube (inner dia. 4 mm) from the sampling point to the fluorescence cell is about 15 m. A PTFE filter with a pore size of 1.2 mm (Millipore, 80 mm dia.) is installed at the top of the sampling tube to remove particles which can cause large laser light scattering. The loss of SO2 in the PTFE filter was checked using the standard gas of 10 ppbv concentration, and the LIF signal intensity was unchanged with and without the filter. The ambient air is aspirated through the tube with a diaphragm pump (ULVAC DA50) at the rate of 6 l per min. The residence time in the 15-m tube was 0.6 s from the outside sampling point to the fluorescence cell. In the measurements the laser wavelength was tuned at the top and bottom absorption wavelengths, 220.6 and 220.2 nm, alternatively, every 30 s. During the ambient air measurements, the concentration of SO2 was also measured by a commercial pulsed fluorescence (PF) detector for SO2 (Thermo Electron 43S). The zero level and signal span of both instruments were calibrated using the 10 ppbv SO2 gas which was produced by mixing of synthetic air and standard SO2 gas (1.03 ppm). Fig. 9 shows an example of the ambient air measurements. The concentration of SO2 was in the range of 4–12 ppbv in this measurement.
The signal averaging times for the two instruments used in the ambient air comparisons were both 2 min. As shown in Fig. 9, the results of the ambient measurements with the LIF instrument developed in this study are in good agreement with those with the commercial PF detector. The correlation plot between the LIF and PF measurement results shown in Fig. 9 gives the slope of 0.976 and the correlation factor of R2 ¼ 0:984: The inter-comparison between the LIF and the commercial PF instruments was also performed in low (0–1500 pptv) and high (0–70 ppbv) concentration ranges, using the standard SO2 gas diluted with the synthetic air in the laboratory. Fig. 10 shows the correlation plots of the SO2 concentrations measured with the LIF instrument versus those measured with the PF instrument in the 0–1500 pptv range. The slope of the least-squares regression line in Fig. 10 is 0.987 and the correlation factor is R2 ¼ 0:982: The correlation plots for the concentration range of 0–70 ppbv gave the slope of 0.992 and correlation factor of R2 ¼ 0:999: Thus, the good linear correlations between the LIF measurements and the PF measurements were obtained in the wide range of SO2 concentration.
5. Conclusion The instrument developed in this study demonstrates that SO2 can be measured directly with high sensitivity and good selectivity by laser induced fluorescence. The laser system consists of a broad-band OPO that is
ARTICLE IN PRESS Y. Matsumi et al. / Atmospheric Environment 39 (2005) 3177–3185
3184
the Advancement of Environmental Protection Technology (Y.M).
LIF SO2 (ppbv)
1.5
References
1
0.5
0
0.5
1
1.5
PF SO2 (ppbv) Fig. 10. Correlation plot of the SO2 concentrations measured with the LIF instrument and the commercial PF detector for the 0–1500 pptv range using the standard SO2 gas diluted with the synthetic air in the laboratory. The slope of the leastsquares regression line indicated is 0.987 and the correlation factor is R2 ¼ 0:982:
pumped by a Nd:YAG laser. Since the OPO is solidstate and does not need a circulation system of liquid dye solution, it is easy to operate. Singling out SO2 by its unique spectral features practically eliminates any possibility of interference from the other species in the ambient air. The SO2 instrument developed has the sensitivity of 5 pptv in 60-s integration time and S/N ¼ 2. Moreover, the instrument proved the feasibility and potential of the laser induced fluorescence technique for measuring SO2. The inter-comparisons between our LIF instrument and the commercial Thermo 43S instrument have been performed using standard gas and ambient air under laboratory conditions. Good correlations between the two instruments have been measured up to 70 ppbv.
Acknowledgements The support of this work by Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, Japan, is acknowledged. We also thank Drs. Hiroshi Tanimoto and Shiro Hatakeyama of National Institute for Environmental Studies for their advices. The research grant for Dynamics of the Sun–Earth–Life Interactive System, No. G-4, the 21st Century COE Program from the Ministry is also acknowledged. This work was also supported in part by the Mitsubishi Chemical Corporation Fund (K.T) and the Steel Industrial Foundation for
Bandy, A.R., Thornton, D.C., Driedger III, A.R., 1993. Airborne measurements of sulfur dioxide, dimethyl sulfide, carbon disulfide, and carbonyl sulfide by isotope dilution gas chromatography/mass spectrometry. Journal of Geophysical Research 98, 23,423–23,433. Benner, R.L., Wu, J., Nicks Jr., D.K., 1997. Development and evaluation of the diffusion denuder-sulfur chemiluminescence detector for atmospheric SO2 measurements. Journal of Geophysical Research 102, 16,287–16,291. Bradshaw, J.D., Rodgers, M.O., Davis, D.D., 1982. Single photon laser-induced fluorescence detection of NO and SO2 for atmospheric conditions of composition and pressure. Applied Optics 21, 2493–2500. Brasseur, G.P., Orlando, J.J., Tyndall, G.S. (Eds.), 1999. Atmospheric Chemistry and Global Change. Oxford University Press, New York. Eisele, F.L., Berresheim, H., 1992. High-pressure chemical ionization flow reactor for real-time mass spectrometric detection of sulfur gases and unsaturated hydrocarbons in air. Analytical Chemistry 64, 283–288. Ferek, R.J., Covert, P.A., Luke, W., 1997. Intercomparison of measurements of sulfur dioxide in ambient air by carbonateimpregnated filters and Teco pulsed-fluorescence analyzers. Journal of Geophysical Research 102, 16,267–16,272. Gallagher, M.S., King, D.B., Whung, P.-Y., Saltzman, E.S., 1997. Performance of the HPLC/fluorescence SO2 detector during the GASIE instrument intercomparison experiment. Journal of Geophysical Research 102, 16,247–16,254. Hui, M.-H., Rice, S.A., 1972. Decay of fluorescence from single vibronic states of SO2. Chemical Physics Letters 17 (4), 474–478. Hunton, D.E., Ballenthin, J.O., Borghetti, J.F., Federico, G.S., Miller, T.M., Thorn, W.F., Viggiano, A.A., Anderson, B.E., Cofer III, W.R., McDougal, D.S., Wey, C.C., 2000. Chemical ionization mass spectrometric measurements of SO2 emissions from jet engines in flight and test chamber operations. Journal of Geophysical Research 105, 26,841–26,855. Jaeschke, W., Beltz, N., Haunold, W., Krischke, U., 1997. Improvement of the tetrachloromercurate absorption technique for measuring low atmospheric CO2 mixing ratios. Journal of Geophysical Research 102, 16,279–16,286. Jost, C., Sprung, D., Kenntner, T., Reiner, T., 2003. Atmospheric pressure chemical ionization mass spectrometry for the detection of tropospheric trace gases: the influence of clustering on sensitivity and precision. International Journal of Mass Spectrometry 223/224, 771–782. Luke, W.T., 1997. Evaluation of a commercial pulsed fluorescence detector for the measurement of low-level SO2 concentrations during the Gas-Phase Sulfur Intercomparison Experiment. Journal of Geophysical Research 102, 16,255–16,265. Manatt, S.L., Lane, A.L., 1993. A compilation of the absorption cross-sections of SO2 from 106 to 403 nm.
ARTICLE IN PRESS Y. Matsumi et al. / Atmospheric Environment 39 (2005) 3177–3185 Journal of Quantitative Spectroscopy & Radiative Transfer 50, 267–276. Matsumi, Y., Murakami, S., Kono, M., Takahashi, K., Koike, M., Kondo, Y., 2001. High-sensitivity instrument for measuring atmospheric NO2. Analytical Chemistry 73, 5485–5493. Mettee, H.D., 1968. Fluorescence and phosphorescence of SO2 vapor. Journal of Chemical Physics 49, 1784–1793. Okabe, H., 1978. Photochmeistry of small molecules. Willy, New York. Okabe, H., Splitstone, P.L., Ball, J.J., 1973. Ambient and source SO2 detector based on a fluorescence method. Journal of the Air Pollution Control Association 23, 514–516. Platt, U., Perner, D., 1980. Direct measurements of atmospheric CH2O, HNO2, O3, NO2, and SO2 by differential optical absorption in the near UV. Journal of Geophysical Research 85, 7453–7458. Reiner, T., Mo¨hler, O., Arnold, F., 1998. Improved atmospheric trace gas measurements with an aircraft-based tandem mass spectrometer: ion identification by massselected fragmentation studies. Journal of Geophysical Research 103, 31,309–31,320. Schwartz, F.P., Okabe, H., Whittaker, J.K., 1974. Fluorescence detection of sulfur dioxide in air at the parts per billion level. Analytical Chemistry 46, 1024–1028. Stecher III, H.A., Luther III, G.W., MacTaggart, D.L., Farwell, S.O., Crosley, D.R., Dorko, W.D., Goldan, P.D.,
3185
Beltz, N., Krischke, U., Luke, W.T., Thornton, D.C., Talbot, R.W., Lefer, B.L., Scheuer, E.M., Benner, R.L., Wu, J., Saltzman, E.S., Gallagher, M.S., Ferek, R.J., 1997. Results of the Gas-Phase Sulfur Intercomparison Experiment (GASIE): Overview of experimental setup, results and general conclusions. Journal of Geophysical Research 102, 16,219–16,236. Talbot, R.W., Scheuer, E.M., Lefer, B.L., Luke, W.T., 1997. Measurements of sulfur dioxide during GASIE with the mist chamber technique. Journal of Geophysical Research 102, 16,273–16,278. Thermo Environmental Instruments, Inc., 1992. Instruction Manual for Model 43S High Sensitive Pulsed Fluorescence SO2 Analyzer. Thermo Environ. Instrum., Inc., Franklin, MA. Thornton, D.C., Bandy, A.R., Tu, F.H., Blomquist, B.W., Mitchell, G.M., Nadler, W., 2002. Fast airborne sulfur dioxide measurements by Atmospheric Pressure Ionization Mass Spectrometry (APIMS). Journal of Geophysical Research 107 (D22), 4632. Tu, F.H., Thornton, D.C., Bandy, A.R., Carmichael, G.R., Tang, Y., Thornhill, K.L., Sachse, G.W., Blake, D.R., 2004. Long-range transport of sulfur dioxide in the central Pacific. Journal of Geophysical Research 109, D15S08. Yvon, S.A., Saltzman, E.S., 1996. Atmospheric sulfur cycling in the tropical Pacific marine boundary layer (121S, 1351W): A comparison of field data and model results 2. Sulfur dioxide. Journal of Geophysical Research 101 (D3), 6911–6918.
Atmospheric Environment 39 (2005) 3177–3185 www.elsevier.com/locate/atmosenv
Technical note
Laser-induced fluorescence instrument for measuring atmospheric SO2 Yutaka Matsumi, Hiroyuki Shigemori, Kenshi Takahashi Solar-Terrestrial Environment Laboratory and Graduate School of Science, Nagoya University, 3-13, Honohara, Toyokawa 442-8507, Japan Received 5 August 2004; received in revised form 28 January 2005; accepted 14 February 2005
Abstract We report on the development of a high-sensitive detection system for measuring atmospheric SO2 using a laserinduced fluorescence (LIF) technique at around 220 nm. Second harmonics of a tunable broad-band optical parametric oscillator (OPO) pumped by the third harmonic of a Nd:YAG laser is used as a fluorescence excitation source. The laser wavelength is alternatively tuned to the peak and the bottom wavelengths in the photoabsorption spectrum of SO2 at 220.6 and 220.2 nm, respectively, and the difference signal at the two wavelengths is used to extract the SO2 concentration. This procedure can give a good selectivity for SO2 and avoid interferences of fluorescent or particulate species other than SO2 in the sample air. The SO2 instrument developed has a sensitivity of 5 pptv in 60 s and S=N ¼ 2: The practical performance of the detection system is tested in the suburban area. The inter-comparisons between the LIF instrument and a commercial instrument using Xe flash lamp excitation for the fluorescence detection have been performed. The correlation between two instruments is measured up to 70 ppbv. A good linear relationship between the LIF measurements and commercial instrument measurements is obtained. r 2005 Published by Elsevier Ltd. Keywords: Sulfur dioxide; OPO; Detection limit; Instrument development
1. Introduction In the atmosphere, sulfur in both gaseous and aerosol forms impacts regional and global chemistry, climate change, as well as the health of various living organism. Anthropogenic emissions in the form of SO2 are about 75% of the total sulfur emission budget (Brasseur et al., 1999). The source of SO2 is primarily from coal and petroleum combustion, petroleum refining and metal smelting operations. The anthropogenic emissions of SO2 are converted to H2SO4, which is the dominant Corresponding author. Tel.: +81 533 89 5192; fax: +81 533 89 5593. E-mail address: [email protected] (Y. Matsumi).
1352-2310/$ - see front matter r 2005 Published by Elsevier Ltd. doi:10.1016/j.atmosenv.2005.02.023
precursor responsible for acid precipitation. The acidic precipitation in the form of rain, fogs, and mists has long been implicated in playing some role in forest decline and foliar damage. SO2 is also a common intermediate in the oxidation of all the reduced sulfur compounds such as dimethylsulfide (DMS), hydrogen sulfide (H2S), carbon disulfide (CS2), etc. The released sulfur compounds are converted to sulfuric acid (H2SO4) through the gaseous SO2 form. The mixing ratios of SO2 in polluted urban and rural areas, particularly near power plants, smelters or paper mills, may exceed several parts per billion by volume (109, ppbv). In the remote troposphere, the concentrations of SO2 are typically less than 100 parts per trillion by volume (1012, pptv). To provide useful information
ARTICLE IN PRESS 3178
Y. Matsumi et al. / Atmospheric Environment 39 (2005) 3177–3185
with which we may broaden our understanding of the sulfur cycles, measurement techniques must be sensitive, precise, and accurate enough to qualify the SO2 concentrations over the wide range from several hundred ppbv to several pptv. Measurements of SO2 in the remote marine atmosphere are extremely important for elucidating the long-range transport of sulfur compounds far from source regions, and understanding of atmospheric sulfur oxidation and deposition. For example, Yvon and Saltzman (1996) measured the concentrations of SO2 and its biogenic precursor DMS in the tropical Pacific marine boundary layer, and analyzed the atmospheric sulfur cycling. Tu et al. (2004) measured the SO2 concentrations on transit flights during the NASA Transport and Chemical Evolution over the Pacific mission (TRACE-P), and investigated the long-range transport of SO2 in the Pacific. Various techniques for the detection of atmospheric SO2 have been used. In the Gas-Phase Sulfur Intercomparison Experiment (GASIE), seven techniques for the field measurement of trace SO2 were compared simultaneously over 1 month in 1994 using samples produced in situ by dynamic dilution (Stecher et al., 1997). The seven techniques were aqueous chemiluminescence (estimated detection limit: 3 pptv in 10–35 min with 500 l sample gas) (Jaeschke et al., 1997), pulsed fluorescence detector (described later) (Luke, 1997), isotope dilution-gas chromatograph mass spectrometer (1 pptv in 3–8 min with 1.7 l gas) (Bandy et al., 1993), mist chamber ion chromatograph (5–47 pptv in 5–30 min with 400 l gas) (Talbot et al., 1997), diffusion denuder sulfur chemiluminescence detector (10 pptv in 3 min with 3 l gas) (Benner et al., 1997), high performance liquid chromatography fluorescence detector (4 pptv in 4 min with 6 l gas) (Gallagher et al., 1997), and carbonate filter ion chromatograph (3–12 pptv in 90 min with 6000 l gas) (Ferek et al., 1997). An atmospheric pressure ionization mass spectrometer (APIMS) has been developed to determine atmospheric SO2 by Thornton et al. (2002) using a chemical ionization mass spectrometry (CIMS) technique. They have used radioactive nickel-63 for an electron source to produce secondary ions. They continuously added 34 16 S O2 as an internal standard. Their detection limit was o1 pptv in 1-s integration time. They indicated that the sensitivity of the CIMS technique was strongly dependent on water vapor in the air and the necessity of a dryer in the air inlet was suggested especially for marine boundary layer measurements. The CIMS approach for the determination of SO2 has been also reported by several groups (Jost et al., 2003; Hunton et al., 2000; Reiner et al., 1998; Eisele and Berresheim, 1992). Differential optical absorption spectrometry (DOAS) measurements of atmospheric SO2 were performed by Platt and Perner. The detection limit of SO2
was estimated to be 10 pptv in 20-min measurement with 5-km optical path (Platt and Perner, 1980). Fluorescence detection methods with the optical excitation to electronically excited states of SO2 have been used for the determination of SO2 concentrations in air. In the 1970’s Okabe and co-workers (Okabe et al., 1973; Schwartz et al., 1974) used the atomic resonance line of Cd (228.8 nm) or Zn (213.8 nm) discharge lamp for the optical excitation sources. Excited SO2 molecules fluoresce in a broad band continuum from 240 to 420 nm, with an emission peak at approximately 320 nm. A bandpass filter was used to isolate the emission radiation and transmit it to an orthogonally oriented photomultiplier tube (PMT). They found that the detector response was linear from at least 8.6 ppbv to 1.8 ppmv. The detection limit was several ppbv levels in 1-min integration time. In the commercial pulsed fluorescence (PF) detector for SO2, Thermo Environmental Instruments (Franklin, Massachusetts) Model 43S, SO2 molecules are electronically excited with the output of a xenon flash lamp pulsed at 10 Hz with 130 ms time width (Thermo Environmental Instruments, Inc., 1992). This instrument is called a Thermo Electron 43S. In this instrument a series of refractive interference filters isolates and passes excitation radiation in the 190–230 nm wavelength range of the xenon flash lamp. The emission from SO2 is monitored with a gated PMT. The sampling window gate is delayed by 30 ms from the start of the Xe flash lamp trigger to avoid the electrical noise associated with the flash. Luke (1997) succeeded to detect as little as 30 pptv SO2 in a 25-min sampling interval using a modified Thermo Electron 43S instrument in the GASIE intercomparison experiment, although the specification of the original instrument is 200 pptv in 3-min detection time. The laser-induced fluorescence (LIF) detection of SO2 was tested in the laboratory by Bradshaw et al. (1982) to provide an assessment for the atmospheric detection. The excitation wavelength was 222.2 nm. The laser system consisted of a Nd:YAG laser pumped dye laser, the fundamental of which was frequency doubled and this output, in turn, was then frequency mixed with the Nd:YAG fundamental at 1064 nm. Based on the laboratory experiments, they estimated the detection limit of SO2 to be several pptv in 1-min integration time with the laser energy of 1 mJ at 222.2 nm and the repetition rate of 10 Hz for the excitation laser light and 6 PMTs for fluorescence detection. Here, we describe design and performance of an LIF instrument for SO2 detection that uses frequencydoubled light of a broad-band optical parametric oscillator (OPO). We have used a two-wavelength measurement technique to extract the concentration of SO2 to avoid the interferences by other fluorescent species. The detection limit of the instrument for SO2 is 5 pptv in 60-s integration time and S=N ¼ 2:
ARTICLE IN PRESS Y. Matsumi et al. / Atmospheric Environment 39 (2005) 3177–3185
We used the ultraviolet laser light at the wavelength around 220 nm for the excitation of SO2 molecule to detect the fluorescence. Fig. 1 shows a schematic diagram of the SO2 detection system which we have developed. The laser system consists of a broad-band OPO (optical parametric oscillator, Continuum, Surelite-OPO) that is pumped by the third harmonic (355 nm) of a Nd:YAG laser (Continuum, Surelite II), which was used in the detection of atmospheric NO2 (Matsumi et al., 2001). The 355 nm light from the YAG laser provides about 170 mJ pulse1 at the repetition rate of 10 Hz. The broad-band OPO contains a BBO crystal and the wavelength is tuned by the angle of the crystal. The angle of the BBO crystal in the OPO is driven by a linear actuator (Sigma Koki, DMY-25) with a sine bar. The output of the OPO around 440 nm is frequency doubled with another angle-tuned BBO crystal. The angle of the BBO crystal for the frequency doubling is driven by a high-resolution rotation stage with a stepper motor. The angle positions of the OPO and frequencydoubling crystals are computer controlled. The ultraviolet laser light generated around 220 nm is isolated from the fundamental laser light (440 nm) by two dielectric coating mirrors (CVI, Y5-1025-45) and fed into a LIF cell. The energy of the ultraviolet laser light around 220 nm is about 1 mJ pulse1 at 10 Hz (10 mW). The time width of the ultraviolet laser pulse is about 5 ns. The ultraviolet laser light of the frequency-doubled OPO output is always monitored at the exit position of the fluorescence cell with a power meter (SCIENTECH, AC2501). The transit time for changing the ultraviolet laser wavelength between the peak and bottom
~ 220 nm BBO crystal
Iris
~ 440 nm 15 mJ / pulse LIF Cell
OPO laser
Air PMT SO2
355 nm 10 Hz
SO2 þ hnðlo219 nmÞ ! SOð3 S Þ þ Oð3 PÞ.
(1)
With the photoexcitation at wavelengths shorter than 219 nm, SO2 gives weak fluorescence. Around 220 nm, there is a structure which is suitable for the peak and bottom measurements. The linewidth of the ultraviolet laser light at 220 nm was measured to be 6 cm1 FWHM (full-width at half-maximum), using a monochromator (f ¼ 500 mm). The structure of the absorption spectrum shown in Fig. 2a is well reproduced in the excitation spectrum with the broad-band OPO system. This
σabs (10-18 cm2)
2.1. Excitation system
wavelengths is less than 0.2 s. Since the OPO is in solid state and does not need a circulation system of liquid dye solution, it is easy to operate. Fig. 2a shows the absorption spectrum of SO2 in the wavelength region of 215–225 nm (Manatt and Lane, 1993). Fig. 2b shows the fluorescence excitation spectrum of SO2 with the frequency-doubled OPO-YAG laser system under the condition of the SO2 concentration of 10 ppbv, where the OPO wavelength is scanned while monitoring the fluorescence intensity with a photomultiplier. The fluorescence excitation spectrum is not proportional to the absorption spectrum at wavelengths shorter than 219 nm. This is due to the predissociation of the upper electronic state of SO2:
Fluorescence Intensity (Arb.)
2. Description of the instrument
3179
60
(a)
40
20
0 5 (b)
Peak Wavelength
4 Bottom Wavelength
3 2 1 0
216
218
220
222
224
Wavelength (nm) YAG laser
Pump PM
Fig. 1. Schematic diagram of the SO2 instrument developed in this work. (PMT) photomultiplier tube; (PM) power meter.
Fig. 2. (a) Absorption cross-section (sabs ) of SO2 as a function of wavelength, which was taken by Manatt and Lane (1993). (b) Fluorescence excitation spectrum of SO2 measured with the instrument shown in Fig. 1 at SO2 concentration of 10 ppbv and total pressure of 10 Torr. The arrows indicate the bottom and peak wavelengths used in this study.
ARTICLE IN PRESS Y. Matsumi et al. / Atmospheric Environment 39 (2005) 3177–3185
3180
indicates that the spectral resolution of the frequencydoubled broad-band OPO system (6 cm1 equivalent to about 0.03 nm) is sufficient to separate the spectral peak and bottom positions in the SO2 spectrum around 220 nm. We have chosen 220.6 nm for the peak wavelength and 220.2 nm for the bottom wavelength. The ratio of the fluorescence intensities at the peak and bottom wavelengths is more than 10. For the peak absorption cross section of 3 1018 cm2 at 220.6 nm and the laser power of 1 mJ pulse1 with the beam diameter of 5 mm, it is estimated that about 0.25% of SO2 molecules in the laser irradiation zone is photoexcited. Therefore, the energy of our laser excitation system is far from the saturation level of the SO2 absorption.
2.2. Fluorescence monitoring system Fig. 3 shows a detailed schematic diagram of the fluorescence cell and the signal detection systems. Optical baffles with cone-shaped apertures are placed within both arms of the fluorescence cell to minimize detection of light scattered by the entrance and exit windows. The diameters of the aperture holes are 12 mm in the entrance baffle arm and 14 mm in the exit baffle arm. The fluorescence from SO2 lies in the 220–450 nm (Mettee, 1968). For the detection of the fluorescence from SO2, a side-on photomultiplier (PMT, Hamamatsu 1P28A) is used, which has sensitivity in the ultraviolet and visible wavelength region. The emitting image of the SO2 volume along the excitation laser beam in the cell is focused onto the photocathode of the PMT by a set of four quartz lenses (50 mm dia., f ¼ 100 mm). A band-pass optical filter (Corning, 7-54) is equipped in front of the PMT to isolate the fluorescence band from 240 to 420 nm. Since the fluorescence lifetime of SO2 is as short as about 40 ns with the excitation wavelength of 220 nm, a photon counting method cannot be used for the processing of the PMT signal output due to the pile-up phenomena. The output of the PMT is amplified and fed into an analog
Laser Beam
Laser Baffles PM Lens+Window
Pre-amp. Baratron Gauge
PMT
Optical Filter
Analog Gated Integrator
A/D
Computer
Fig. 3. Cross section of the fluorescence cell and fluorescence detection system.
gated integrator (Stanford Research, SR250) with the integration over 10 successive laser pulses. The gate of the integrator is opened at the rise of the laser pulse and its duration is 100 ns. The integrated signal is sampled with an A/D board and a microcomputer. 2.3. Gas handling system The ambient air is pulled into the fluorescence cell with a booster pump (ULVAC PMB001C) and a rotary pump (Alcatel 2021). The pumping speed at the cell pressure of 10 Torr is about 1000 l min1. The pressure in the cell is monitored with a capacitance manometer (MKS Baratron 122A, full scale 20 Torr). The sample flow rate in the fluorescence cell is 1.3 slm (standard liter per minute). For the calibration of the detection system, SO2 from a standard mixing ratio gas cylinder (Nihon Sanso, 1.03 ppmv) is diluted with the synthetic air flow. The flow rates of all the gases are controlled by mass flow meters. All the mass flow meters are calibrated using a soap film flow meter (STEC, SF1). The weight and size of the whole system of our prototype instrument are about 250 kg and 2 m width 1 m depth 1 m height, respectively.
3. Instrument performance 3.1. Instrument sensitivity Fig. 4 shows a typical example of the fluorescence intensity measurements with our SO2 instrument, when the SO2 gas concentration was 10 ppbv. The vertical scale is the fluorescence intensity in arbitrary units, while the horizontal axis is time. The OPO wavelength was fixed at the peak position of the SO2 absorption (220.6 nm) for 30 s and then the wavelength was changed and fixed at the bottom position (220.2 nm) for 30 s. The measurements at the peak and bottom positions were alternately repeated several times. Then, the SO2 gas concentration was reduced to zero to measure the background signal intensity. The differences between the fluorescence intensities at the peak and bottom wavelengths versus SO2 concentrations with the standard gas sample in the range of 0–140 ppbv are plotted in Fig. 5. The LIF signal intensities are linearly proportional to the SO2 concentrations up to 140 ppbv. The differences between the fluorescence intensities at the peak and bottom wavelengths versus SO2 concentrations in the range of 0–1500 pptv are plotted in Fig. 6. The detection sensitivity of LIF instrument, S SO2 ; is proportional to the product of the fluorescence quantum yield, Ff ; and the ratio of the pressures in the fluorescence cell and in the ambient air, Pcell and Pambient : SSO2 / Ff Pcell =Pambient .
(2)
ARTICLE IN PRESS Y. Matsumi et al. / Atmospheric Environment 39 (2005) 3177–3185
1.5
0.15 SO2 OFF
Fluorescence Intensity (Arb.)
Fluorescence Intensity (Arb.)
SO2 10 ppbv
1.0
0.5
0.10
0.05
0 0
200
400
600
Time (s) Fig. 4. Typical example of the fluorescence intensity measurements with our SO2 instrument. The sample SO2 gas concentration is 10 ppbv, which is generated by the dilution of the standard SO2 gas with the synthetic air. The fluorescence intensity is measured while alternatively tuning the laser wavelength at the peak (220.6 nm) and bottom (220.2 nm) positions every 30 s. Then the SO2 gas is reduced to zero to measure the back ground signal intensity.
15 Fluorescence Intensity (Arb.)
3181
10
5
0
50
100
1500
Fig. 6. Calibration plots of the fluorescence intensities at the peak wavelength (triangle), that at the bottom wavelength (square), and the difference of the two intensities (circle). The concentration range of SO2 is 0–1500 pptv. Error bars are one standard deviations.
where kr and knr are radiative and non-radiative rate constants of the photoexcited SO2, respectively, and kM q is a quenching rate constant for M such as O2 and N2. The radiative and non-radiative rate constants are 1.5 107 and 7.4 106 s1, respectively, with the excitation around 220 nm and the quenching rate constants are in the order of 1010 cm3 molecule1 s1 (Hui and Rice, 1972). According to expression (2), the optimum pressure for the LIF detection of SO2 around 220 nm is in the range of 1–100 Torr. The total pressure of 10 Torr at the fluorescence cell was used in most experiments in this study. The flow rate in the fluorescence cell is high enough to avoid the irradiation of an air sample by two successive laser pulses before it leaves the detection volume, since fragments from the photodissociation of some molecules such as sulfur compounds and aromatics can produce extra LIF.
150
SO2 Concentration (ppbv)
3.2. Minimum detectable limit
Fig. 5. Calibration plot of the fluorescence intensity which is the subtraction of the intensity at the bottom wavelength (220.2 nm) from that at the peak wavelength (220.6 nm). The concentration range of SO2 is 0–150 ppbv. The scale of the fluorescence intensity is normalized with the level when the SO2 concentration is 10 ppbv. Error bars are one standard deviations.
The fluorescence quantum yield of SO2 in the cell is determined by the equation: Ff ¼ kr =ðkr þ knr þ SkM q ½M Þ,
500 1000 Concentration (pptv)
(3)
Fig. 7 shows the fluorescence intensities recorded at the peak and bottom wavelengths at 60 pptv of SO2 concentration and the background fluorescence intensity. In the first 0–30 s period, the laser wavelength was fixed at the peak wavelength and in the 30–60 s period the laser wavelength was fixed at the bottom wavelength and then in the 60–160 s period the laser wavelength was fixed at the peak wavelength again. In the 0–100 s period the concentration of SO2 was kept to be 60 pptv, while at 100 s the supply of the SO2 gas was turned off and the background signal intensity was recorded. When the signal intensities are very weak, the dominant source for
ARTICLE IN PRESS Y. Matsumi et al. / Atmospheric Environment 39 (2005) 3177–3185
3182
0.51 counts s1 pptv1 from the estimation of the equivalent photon numbers for the vertical scale in Fig. 6. The background signal level was estimated to be 50 photons s1. The background signal came mainly from the wall and Rayleigh scattering of the laser light. The background level was stable on the time scale of the on/off measurements. With S=N ¼ 2; the minimum detectable limit of our SO2 instrument is calculated to be 5 pptv with the integration time of t ¼ 60 s and 1.6 pptv with t ¼ 600 s:
120 Fluorescence Intensity (counts s-1)
SO2 60 pptv
SO2 OFF
Bottom wavelength 80
40
3.3. Possible interferences Peak wavelength 0
50
100
150
Time (s) Fig. 7. Fluorescence intensities recorded at the peak and bottom wavelengths at 60 pptv of SO2 concentration and the background fluorescence intensity. In the 0–100 s period the concentration of SO2 was kept to be 60 pptv, while at 100 s the supply of the SO2 gas was turned off and the background signal intensity is recorded. The vertical scale is equivalent photon number detected by the PMT per second.
the fluctuation of the signal intensities is photon number statistics described by a Poisson distribution. Under this condition, the standard deviation pffiffiffiffisffi of the photon population can be taken as s ¼ N ; where N is the number of photons detected by the PMT for a given integration time. We calculate the standard deviation s of the signal intensity and then determine the relationship between the number of photons per second and the signal intensity detected using the gated integrator assuming that the standard of the signal pffiffiffiffideviation ffi intensities follow the s ¼ N expression. Thus the equivalent photon numbers are estimated from the signal intensity from the gated integrator as shown in Fig. 7. When the predominant source of the noise is fluctuations of the detected photon numbers with Poisson distributions, the minimum detectable SO2 mixing ratio of the instrument, [SO2]min, is expressed by pffiffiffiffiffiffiffi S=N Sbg ½SO2 min ¼ pffiffi , t ðC peak C bottom Þ
(4)
where S/N is the required signal-to-noise ratio, C peak and C bottom are the sensitivity factors of the LIF signal of SO2 for the peak and bottom wavelengths, respectively, in counts s1 pptv1, S bg is the background signal level of the instrument in photons s1, and t is the averaging time of the instrument in s. The value of (C peak 2C bottom ) was determined to be
The possible fluorescent species with the excitation around 220 nm are compounds such as aromatics, and molecules with unsaturated chemical bonds. No testing for the interference of aromatics has been done with the present instrument. In principle the interference from species other than SO2 can be eliminated by the measurements with tuning the laser wavelength at the top and bottom wavelengths in the SO2 photoabsorption spectrum. We have checked the interference by nitrogen dioxide (NO2) molecules. We could detect no fluorescence at both peak and bottom wavelengths at the NO2 concentration of 9.7 ppmv. NO2 molecules dissociate at the excitation wavelengths shorter than 400 nm and emits no fluorescence (Okabe, 1978). We also checked the interference by nitric oxide (NO) molecules. No emission was detected from NO molecules at both peak and bottom wavelengths at the NO concentration of 1.0 ppmv. The interference measurements indicate that over 1000 times more NO2 or NO is needed to produce an equivalent signal to that from one part SO2. The fluorescence of SO2 is quenched by water molecules as well as N2 and O2 molecules. The change of H2O concentration in the ambient air is large under various conditions. We have checked the dependence of the fluorescence intensity of SO2 on the water vapor concentration. Fig. 8 shows the fluorescence intensity of SO2 as a function of water vapor concentration, when the concentration of SO2 is 10 ppbv. The fluorescence intensity of SO2 is reduced by about 10% by the presence of 16 g m3 H2O which corresponds to the relative humidity of 90% at 298 K. Thus, the fluorescence quenching of SO2 by water vapor is not so fast when the excitation laser wavelength is 220.6 nm. Since this negative interference by water vapor is quantitative, it can be corrected by the simultaneous measurements of SO2 and H2O in sample air, when very accurate values of SO2 concentration are necessary. Luke (1997) also suggested a decrease of 7% and 15% in the response of the pulsed fluorescence detector for SO2 (Thermo Electron 43S) at the relative humidities of 35% and 50%, respectively, at 20–25 1C.
ARTICLE IN PRESS Y. Matsumi et al. / Atmospheric Environment 39 (2005) 3177–3185
Relative Humidity (%) at 298 K 20 40 60 80
15 100 SO2 concentration (ppbv)
0
Fluorescence Intensity (arb.)
1
0.5
0
3183
5 10 15 Absolute Humidity (g m-3)
Fig. 8. Fluorescence intensity of SO2 as a function of water vapor concentration, when the concentration of SO2 is 10 ppbv.
10
5
0 15:00
16:00
17:00
Local time Fig. 9. Ambient measurement data obtained on December 18, 2002. The results of the SO2 concentration measurements with our LIF instrument (open square) and with the commercial pulsed fluorescence (PF) detector for SO2 (closed circles).
4. Ambient measurement and inter-comparison Ambient measurements were performed in December of 2002 in the south yard in Toyokawa campus of Nagoya University. The sampling point is located at a height of 2.0 m from the ground and 10 m south from the laboratory building which is about 9 m tall. The yard is covered with lawn and has an area of about 4000 m2. Toyokawa campus is located in a suburban factory area and Tomei expressway runs 1 km north of the campus. The length of the PTFE tube (inner dia. 4 mm) from the sampling point to the fluorescence cell is about 15 m. A PTFE filter with a pore size of 1.2 mm (Millipore, 80 mm dia.) is installed at the top of the sampling tube to remove particles which can cause large laser light scattering. The loss of SO2 in the PTFE filter was checked using the standard gas of 10 ppbv concentration, and the LIF signal intensity was unchanged with and without the filter. The ambient air is aspirated through the tube with a diaphragm pump (ULVAC DA50) at the rate of 6 l per min. The residence time in the 15-m tube was 0.6 s from the outside sampling point to the fluorescence cell. In the measurements the laser wavelength was tuned at the top and bottom absorption wavelengths, 220.6 and 220.2 nm, alternatively, every 30 s. During the ambient air measurements, the concentration of SO2 was also measured by a commercial pulsed fluorescence (PF) detector for SO2 (Thermo Electron 43S). The zero level and signal span of both instruments were calibrated using the 10 ppbv SO2 gas which was produced by mixing of synthetic air and standard SO2 gas (1.03 ppm). Fig. 9 shows an example of the ambient air measurements. The concentration of SO2 was in the range of 4–12 ppbv in this measurement.
The signal averaging times for the two instruments used in the ambient air comparisons were both 2 min. As shown in Fig. 9, the results of the ambient measurements with the LIF instrument developed in this study are in good agreement with those with the commercial PF detector. The correlation plot between the LIF and PF measurement results shown in Fig. 9 gives the slope of 0.976 and the correlation factor of R2 ¼ 0:984: The inter-comparison between the LIF and the commercial PF instruments was also performed in low (0–1500 pptv) and high (0–70 ppbv) concentration ranges, using the standard SO2 gas diluted with the synthetic air in the laboratory. Fig. 10 shows the correlation plots of the SO2 concentrations measured with the LIF instrument versus those measured with the PF instrument in the 0–1500 pptv range. The slope of the least-squares regression line in Fig. 10 is 0.987 and the correlation factor is R2 ¼ 0:982: The correlation plots for the concentration range of 0–70 ppbv gave the slope of 0.992 and correlation factor of R2 ¼ 0:999: Thus, the good linear correlations between the LIF measurements and the PF measurements were obtained in the wide range of SO2 concentration.
5. Conclusion The instrument developed in this study demonstrates that SO2 can be measured directly with high sensitivity and good selectivity by laser induced fluorescence. The laser system consists of a broad-band OPO that is
ARTICLE IN PRESS Y. Matsumi et al. / Atmospheric Environment 39 (2005) 3177–3185
3184
the Advancement of Environmental Protection Technology (Y.M).
LIF SO2 (ppbv)
1.5
References
1
0.5
0
0.5
1
1.5
PF SO2 (ppbv) Fig. 10. Correlation plot of the SO2 concentrations measured with the LIF instrument and the commercial PF detector for the 0–1500 pptv range using the standard SO2 gas diluted with the synthetic air in the laboratory. The slope of the leastsquares regression line indicated is 0.987 and the correlation factor is R2 ¼ 0:982:
pumped by a Nd:YAG laser. Since the OPO is solidstate and does not need a circulation system of liquid dye solution, it is easy to operate. Singling out SO2 by its unique spectral features practically eliminates any possibility of interference from the other species in the ambient air. The SO2 instrument developed has the sensitivity of 5 pptv in 60-s integration time and S/N ¼ 2. Moreover, the instrument proved the feasibility and potential of the laser induced fluorescence technique for measuring SO2. The inter-comparisons between our LIF instrument and the commercial Thermo 43S instrument have been performed using standard gas and ambient air under laboratory conditions. Good correlations between the two instruments have been measured up to 70 ppbv.
Acknowledgements The support of this work by Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, Japan, is acknowledged. We also thank Drs. Hiroshi Tanimoto and Shiro Hatakeyama of National Institute for Environmental Studies for their advices. The research grant for Dynamics of the Sun–Earth–Life Interactive System, No. G-4, the 21st Century COE Program from the Ministry is also acknowledged. This work was also supported in part by the Mitsubishi Chemical Corporation Fund (K.T) and the Steel Industrial Foundation for
Bandy, A.R., Thornton, D.C., Driedger III, A.R., 1993. Airborne measurements of sulfur dioxide, dimethyl sulfide, carbon disulfide, and carbonyl sulfide by isotope dilution gas chromatography/mass spectrometry. Journal of Geophysical Research 98, 23,423–23,433. Benner, R.L., Wu, J., Nicks Jr., D.K., 1997. Development and evaluation of the diffusion denuder-sulfur chemiluminescence detector for atmospheric SO2 measurements. Journal of Geophysical Research 102, 16,287–16,291. Bradshaw, J.D., Rodgers, M.O., Davis, D.D., 1982. Single photon laser-induced fluorescence detection of NO and SO2 for atmospheric conditions of composition and pressure. Applied Optics 21, 2493–2500. Brasseur, G.P., Orlando, J.J., Tyndall, G.S. (Eds.), 1999. Atmospheric Chemistry and Global Change. Oxford University Press, New York. Eisele, F.L., Berresheim, H., 1992. High-pressure chemical ionization flow reactor for real-time mass spectrometric detection of sulfur gases and unsaturated hydrocarbons in air. Analytical Chemistry 64, 283–288. Ferek, R.J., Covert, P.A., Luke, W., 1997. Intercomparison of measurements of sulfur dioxide in ambient air by carbonateimpregnated filters and Teco pulsed-fluorescence analyzers. Journal of Geophysical Research 102, 16,267–16,272. Gallagher, M.S., King, D.B., Whung, P.-Y., Saltzman, E.S., 1997. Performance of the HPLC/fluorescence SO2 detector during the GASIE instrument intercomparison experiment. Journal of Geophysical Research 102, 16,247–16,254. Hui, M.-H., Rice, S.A., 1972. Decay of fluorescence from single vibronic states of SO2. Chemical Physics Letters 17 (4), 474–478. Hunton, D.E., Ballenthin, J.O., Borghetti, J.F., Federico, G.S., Miller, T.M., Thorn, W.F., Viggiano, A.A., Anderson, B.E., Cofer III, W.R., McDougal, D.S., Wey, C.C., 2000. Chemical ionization mass spectrometric measurements of SO2 emissions from jet engines in flight and test chamber operations. Journal of Geophysical Research 105, 26,841–26,855. Jaeschke, W., Beltz, N., Haunold, W., Krischke, U., 1997. Improvement of the tetrachloromercurate absorption technique for measuring low atmospheric CO2 mixing ratios. Journal of Geophysical Research 102, 16,279–16,286. Jost, C., Sprung, D., Kenntner, T., Reiner, T., 2003. Atmospheric pressure chemical ionization mass spectrometry for the detection of tropospheric trace gases: the influence of clustering on sensitivity and precision. International Journal of Mass Spectrometry 223/224, 771–782. Luke, W.T., 1997. Evaluation of a commercial pulsed fluorescence detector for the measurement of low-level SO2 concentrations during the Gas-Phase Sulfur Intercomparison Experiment. Journal of Geophysical Research 102, 16,255–16,265. Manatt, S.L., Lane, A.L., 1993. A compilation of the absorption cross-sections of SO2 from 106 to 403 nm.
ARTICLE IN PRESS Y. Matsumi et al. / Atmospheric Environment 39 (2005) 3177–3185 Journal of Quantitative Spectroscopy & Radiative Transfer 50, 267–276. Matsumi, Y., Murakami, S., Kono, M., Takahashi, K., Koike, M., Kondo, Y., 2001. High-sensitivity instrument for measuring atmospheric NO2. Analytical Chemistry 73, 5485–5493. Mettee, H.D., 1968. Fluorescence and phosphorescence of SO2 vapor. Journal of Chemical Physics 49, 1784–1793. Okabe, H., 1978. Photochmeistry of small molecules. Willy, New York. Okabe, H., Splitstone, P.L., Ball, J.J., 1973. Ambient and source SO2 detector based on a fluorescence method. Journal of the Air Pollution Control Association 23, 514–516. Platt, U., Perner, D., 1980. Direct measurements of atmospheric CH2O, HNO2, O3, NO2, and SO2 by differential optical absorption in the near UV. Journal of Geophysical Research 85, 7453–7458. Reiner, T., Mo¨hler, O., Arnold, F., 1998. Improved atmospheric trace gas measurements with an aircraft-based tandem mass spectrometer: ion identification by massselected fragmentation studies. Journal of Geophysical Research 103, 31,309–31,320. Schwartz, F.P., Okabe, H., Whittaker, J.K., 1974. Fluorescence detection of sulfur dioxide in air at the parts per billion level. Analytical Chemistry 46, 1024–1028. Stecher III, H.A., Luther III, G.W., MacTaggart, D.L., Farwell, S.O., Crosley, D.R., Dorko, W.D., Goldan, P.D.,
3185
Beltz, N., Krischke, U., Luke, W.T., Thornton, D.C., Talbot, R.W., Lefer, B.L., Scheuer, E.M., Benner, R.L., Wu, J., Saltzman, E.S., Gallagher, M.S., Ferek, R.J., 1997. Results of the Gas-Phase Sulfur Intercomparison Experiment (GASIE): Overview of experimental setup, results and general conclusions. Journal of Geophysical Research 102, 16,219–16,236. Talbot, R.W., Scheuer, E.M., Lefer, B.L., Luke, W.T., 1997. Measurements of sulfur dioxide during GASIE with the mist chamber technique. Journal of Geophysical Research 102, 16,273–16,278. Thermo Environmental Instruments, Inc., 1992. Instruction Manual for Model 43S High Sensitive Pulsed Fluorescence SO2 Analyzer. Thermo Environ. Instrum., Inc., Franklin, MA. Thornton, D.C., Bandy, A.R., Tu, F.H., Blomquist, B.W., Mitchell, G.M., Nadler, W., 2002. Fast airborne sulfur dioxide measurements by Atmospheric Pressure Ionization Mass Spectrometry (APIMS). Journal of Geophysical Research 107 (D22), 4632. Tu, F.H., Thornton, D.C., Bandy, A.R., Carmichael, G.R., Tang, Y., Thornhill, K.L., Sachse, G.W., Blake, D.R., 2004. Long-range transport of sulfur dioxide in the central Pacific. Journal of Geophysical Research 109, D15S08. Yvon, S.A., Saltzman, E.S., 1996. Atmospheric sulfur cycling in the tropical Pacific marine boundary layer (121S, 1351W): A comparison of field data and model results 2. Sulfur dioxide. Journal of Geophysical Research 101 (D3), 6911–6918.