Environ. Sci. Technol. 2006, 40, 7868-7873
Cavity Ring-Down Spectroscopy of Ambient NO2 with Quantification and Elimination of Interferences JAMES HARGROVE, LIMING WANG,† KAREN MUYSKENS,‡ MARK MUYSKENS,‡ DAVID MEDINA, SUSAN ZAIDE, AND JINGSONG ZHANG* Department of Chemistry, University of California, Riverside, California 92521
Ambient detection of NO2 by cavity ring-down spectroscopy is examined in the wavelength region near 405.23 nm, and possible interferences by particulates, water vapor, and carbon dioxide are characterized. Particulates can be efficiently removed by the use of a 0.45 µm fluoropolymer filter. Water vapor has a response of 2.8 ppb (NO2 equivalent) for 1.0% water vapor (80% relative humidity at 10 °C) in air at 405.23 nm in a broad continuous absorption feature. Carbon dioxide has a response of 0.8 ppb (NO2 equivalent) for 1.0% CO2 attributable to Rayleigh scattering and would not contribute significant interference in ambient measurements due to the lower ambient CO2 levels. Water vapor interference and in general broad background in the absorption spectrum can be accounted for by removing NO2 selectively in the ambient air stream with an annular denuder coated with sodium hydroxide and methoxyphenol (guiacol). Subtraction of the resulting background signal provides NO2 measurements with a limit of detection of 150 ppt/10 s (S/N ) 3). Reliable NO2 measurements could be obtained by this method without the need for frequent calibration with calibration gas. Ambient NO2 measurements are carried out to demonstrate this method.
Introduction Nitrogen dioxide (NO2) is a major pollutant in the atmosphere of modern cities that is easily recognized by its reddishbrown color. NO2 is formed when nitric oxide (NO) is produced as a byproduct of combustion in internal combustion engines and power generators at temperatures greater than 800 °C and is oxidized by alkylperoxy radicals in the atmosphere. Currently, a principal source of NO in California is diesel engines, since auto emissions have been successfully reduced by use of catalytic converters. NO2 in the troposphere subsequently undergoes photolysis to ultimately form O3 in the presence of sunlight. In the stratosphere, however, NO2 is implicated in the destruction of O3. Mixing ratios for NO2 have been measured at sub-parts-per-billion levels in remote areas and up to hundreds of parts per billion (ppb) in urban areas (1). Many techniques have been developed to measure atmospheric NO2, but few have been demonstrated to be free from interferences from other atmospheric constituents. * Corresponding author fax: (951) 827-4713; e-mail:
[email protected]; additional address: Air Pollution Research Center, University of California, Riverside, CA 92521. † Permanent address: School of Chemistry, South China Univesity of Technology, Guangzhou 510640, P. R. China. ‡ Permanent address: Department of Chemistry and Biochemistry, Calvin College, Grand Rapids, MI 49546. 7868
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 24, 2006
Intercomparisons between different methods have been made (2, 3). The most commonly used method is chemiluminescence (4-7), in which conversion of NO2 to NO, either by catalytic thermal decomposition (which suffers from interferences from organic nitrates, HONO, HNO3, etc.) or photolysis (which is relatively immune from interferences), is followed by reaction of NO with O3 to produce electronically excited NO2*. The excited NO2* emits a broad continuum radiation in the region of 500-900 nm, with a signal strength that is proportional to the concentration of NO. Subtraction of the background NO concentration then yields the concentration of NO2. The chemiluminescence instruments have typical NO2 detection limits of ∼50 parts per trillion (ppt)/2 min (4, 7). The NO produced can also be detected using laser-induced fluorescence (LIF) with a sensitivity of 2 ppt/3 min (8). The Luminox method can be categorized as a chemiluminescence method as well, in which the emission at 420 nm is detected when NO2 reacts with luminol solution (9). A detection sensitivity of 300° C. As a result, it is actually a NOyanalyzer (33); species including HONO, HNO3, PAN, PPN (peroxypropionyl nitrate), particulate nitrate, and even HCN, CH3CN, and NH3 are all detected as NO2. This could result in a higher indicated NO2 concentration in a chemiluminescence analyzer when used for ambient analysis. Measurements by chemiluminescence analyzers are also sensitive to the humidity of ambient air. Water vapor and carbon dioxide can quench the fluorescence of NO2* by collision, resulting in lower NO2 readings. The comparisons of NO2 measurements by CRDS and NOx analyzer have been carried out on NO2 standard mixtures and ambient air. In this comparison, the gas samples passed through the CRDS cavity for the CRDS measurements and then entered into the NOx analyzer; the CRDS NO2 concentrations were obtained using the measured absorption coefficients by CRDS and the known NO2 absorption cross sections by Yoshino et al. (30). The CRDS and NOx analyzer measurements of the pure NO2 standards in clean air in the concentration range of 10-100 ppb were in excellent agreement, as shown in Figure 3. However, the CRDS and NOx analyzer measurements of NO2 in the ambient air samples showed some disagreement that seemed to vary with the ambient conditions. The chemiluminescence analyzer used in this study was found to have a 9% level of quenching for 1.0% water vapor [80% relative humidity (RH) at 10 °C]. This was measured by bubbling zero grade air through water held at 10 °C prior to blending with an NO2 calibration mixture. The level of quenching would be expected to vary between analyzers due to variations in operating pressure. This quenching effect could be difficult to deal with in chemiluminescence analysis.
FIGURE 3. Comparison of NO2 measurements by CRDS and NOx analyzer on pure NO2 standards in clean air. The 8-bit data sets are from two measurements separated by several weeks and are corrected for the -8.8% error in CRDS. The CRDS and NOx analyzer measurements of the pure NO2 standards have an excellent linear correlation in the range of 10-100 ppb; the least-square fit line has a slope of 1.01 ( 0.01, an intercept of 0.24 ( 0.21 ppb, and a linear correlation coefficient R of 0.999. The humidity could be measured simultaneously to determine a quenching correction factor, but the water response is a variable that depends on the pressure and temperature within the fluorescence chamber. Baseline subtraction with removal of water vapor response could not be used on chemiluminescence analyzers, because as the NO2 in the sample stream is removed the quenching effect is removed as well. Water vapor was found to give a response in our CRDS measurements equivalent to 2.8 ppb of NO2 for 1.0% water vapor in clean air (80% RH at 10 °C) near 405 nm. The NO2 denuders tested in this study passed ambient water vapor completely, though there may be some time needed for equilibration if there are large fluctuations in humidity. The CRDS signal remaining after denuding ambient air samples was completely removed by the addition of silica gel downstream of the denuder, suggesting that the remaining response was from water vapor. After both the NO2 denuder and the silica gel trap, NO2 and water vapor were removed from the ambient air, and the denuded dry air was shown to have the same CRDS response as the clean zero air. Separate experiments with high-purity water also supported the origin and magnitude of the water absorption response. In the experiments with high-purity water, the CRDS spectrum of the response to water was shown to be relatively flat with a slightly higher absorption at longer wavelengths near 405 nm. The water vapor response showed a quadratic dependence on water vapor concentration when measured with different concentrations of water vapor; this water response is suspected to be due to water dimer and will be the focus of subsequent research. Measurements of background water levels in ambient air found a range between 2 and 7 ppb NO2 equivalent during testing as the humidity varied. Subtraction of the background level due to the water vapor and the Rayleigh scattering of air (i.e, the CRDS signal level with the NO2 denuder) from the total CRDS signal level (without the denuder) was found to give reliable measurements of NO2. The observed water response in this work is similar to a previously measured response of water vapor in this spectral region (34). It was suggested that the water vapor response was consistent with the effect of scattering by water droplets with a 1 µm diameter, but water droplets would not be expected to form to such an extent at well below the dew point of water. In our experiment, there was no reduction
in water response when the CRDS mirrors were protected with dry cylinder air, suggesting that the effect was not due to absorption on the mirrors; heating the mirrors also had no effect on the water response. Recently, the water vapor response has also been detected by the CAPS technique in the 420-nm region, with a somewhat smaller cross section (35). CO2 had a response of 0.8 ppb (NO2 equivalent) for 1.0% CO2. This is consistent with our estimated effect of Rayleigh scattering. It is expected that CO2 levels will usually be much smaller under ambient conditions so that the CO2 response is negligible. Other interferences to the CRDS measurement in the 400410 nm range are small for ambient NO2 measurements. NO2 has its peak absorption near 400-410 nm, while interferences due to absorption from other species are minimized in this region. The nitrogen oxide compounds such as HONO, CH3ONO, N2O4, CH3NO2, CH3ONO2, and PAN, which would be detected as NO2 by the chemiluminescence analyzer, do not have absorptions near 400 nm. Other species such as ClNO, ClNO2, ClONO2, and BrONO2 have small absorption cross sections around 400 nm and small ambient concentrations, compared to NO2 (1). Non-nitrogen oxide compounds, such as glyoxal and its methyl substitutes have absorptions peaking around 440-450 nm, and their absorption cross sections are an order of magnitude lower than NO2 in the region of 405 nm (
