Dry Deposition Fluxes and Deposition Velocities of ... - ACS Publications

Observations were conducted at a study site in the Tokyo metropolitan area. The dry deposition fluxes were compared to the wet deposition fluxes measu...
0 downloads 0 Views 178KB Size
Download PDF
Environ. Sci. Technol. 2004, 38, 2190-2197

Dry Deposition Fluxes and Deposition Velocities of Trace Metals in the Tokyo Metropolitan Area Measured with a Water Surface Sampler MASAHIRO SAKATA* AND KOHJI MARUMOTO Komae Research Laboratory, Central Research Institute of Electric Power Industry (CRIEPI), 2-11-1 Iwato-kita, Komae, Tokyo 201-8511, Japan

Dry deposition fluxes and deposition velocities ()deposition flux/atmospheric concentration) for trace metals including Hg, Cd, Cu, Mn, Pb, and Zn in the Tokyo metropolitan area were measured using an improved water surface sampler. Mercury is deposited on the water surface in both gaseous (reactive gaseous mercury, RGM) and particulate (particulate mercury, Hg(p)) forms. The results based on 1 yr observations found that dry deposition plays a significant if not dominant role in trace metal deposition in this urban area, contributing fluxes ranging from 0.46 (Cd) to 3.0 (Zn) times those of concurrent wet deposition fluxes. The deposition velocities were found to be dependent on the deposition of coarse particles larger than approximately 5 µm in diameter on the basis of model calculations. Our analysis suggests that the 84.13% diameter is a more appropriate index for each deposited metal than the 50% diameter in the assumed undersize log-normal distribution, because larger particles are responsible for the flux. The deposition velocities for trace metals other than mercury increased exponentially with an increase in their 84.13% diameters. Using this regression equation, the deposition velocities for Hg(p) were estimated from its 84.13% diameter. The deposition fluxes for Hg(p) calculated from the estimated velocities tended to be close to the mercury fluxes measured with the water surface sampler during the study periods except during summer.

Introduction It is known that the atmospheric deposition is an important source of various substances toxic to surface waters and terrestrial environments (1-3). These substances include metals such as As, Cd, Hg, and Pb and organic compounds such as polychlorinated dibenzo-p-dioxins and furans (PCDD/ Fs) and polycyclic aromatic hydrocarbons (PAHs). In recent years, there has been an increase in human health and environmental concerns related to the toxicity of methylmercury bioaccumulated in fish through the food chain. Significant amounts of methylmercury are produced by in situ methylation of inorganic mercury deposited into aquatic environments (3). Wet deposition fluxes of atmospheric constituents are conventionally calculated using their concentrations mea* Corresponding author phone: +81-3-3480-2111; fax: +81-33480-1942; e-mail: [email protected]. 2190

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 7, 2004

sured in precipitation samples and the amount of precipitation recorded for each collection period. Compared to the case for wet deposition, however, many uncertainties exist in the methods used to quantify dry deposition (4-7). These methods include both direct measurements and modeled estimates. To date, there is no accepted technique that can be used to evaluate the accuracy of these methods. This is due to the fact that quantification of dry deposition is difficult because of large spatial and temporal variations of meteorological conditions and surface characteristics. Modeled estimates are typically generated by multiplying the measured atmospheric concentration by the modeled deposition velocity for each constituent. The electrical resistance analogue is often used for the parametrization of the dry deposition velocity (8). On the other hand, direct measurements using various surrogate surfaces, mainly solid surfaces such as Teflon plates, filters, and buckets, have been conducted in an attempt to quantify dry deposition (5, 9-11). It has been shown that both the collector geometry and the surface characteristics have a large impact on the amount of collected material. Recently, Yi et al. (12) developed a circular water surface sampler to measure the dry deposition fluxes of atmospheric gases and particles such as SO2 and SO42-. In contrast to solid surfaces, a water surface can act as an infinite sink for both constituents. The sampler has a water surface plate with an airfoil leading edge to minimize airflow disruptions caused by the collector geometry. The SO2 fluxes measured directly with a water surface sampler were found to agree well with those from model calculations (12). In this study, a water surface sampler was used to measure the dry deposition of trace metals including Hg, Cd, Cu, Mn, Pb, and Zn. Mercury is deposited on the water surface in both gaseous (reactive gaseous mercury, RGM) and particulate (particulate mercury, Hg(p)) forms. Deposition of gaseous elemental mercury, Hg0, which primarily exists in the atmosphere, is ignored because of its extremely low solubility in water. The sampler of Yi et al. (12) was improved to prevent the contamination and adsorption of trace metals and also vaporization of mercury, and to perform long-term dry deposition measurements. Observations were conducted at a study site in the Tokyo metropolitan area. The dry deposition fluxes were compared to the wet deposition fluxes measured at the same site. The dry deposition fluxes for all metals measured with the water surface sampler are those obtained on the same deposition surface (water surface) under the same meteorological conditions. Thus, the difference in deposition velocity ()deposition flux/atmospheric concentration) between trace metals is expected to be mainly due to the difference in their size distribution in atmospheric particles. In general, the trace metals associated with larger particles will have larger deposition velocities. On the basis of the size distribution data measured during dry deposition sampling, we investigated the relationship between deposition velocity and particle size distribution. For mercury, both RGM and Hg(p) are deposited on the water surface as described earlier. If the relationship between deposition velocity and particle size distribution is quantified for trace metals other than Hg(p), which exist entirely in particulate form, then the deposition velocity of Hg(p) can be estimated from its size distribution. This allows us to calculate the Hg(p) deposition flux using this deposition velocity and the measured atmospheric concentration. In addition, this makes it possible to estimate the dry deposition flux and deposition velocity of RGM, because the RGM flux 10.1021/es030467k CCC: $27.50

 2004 American Chemical Society Published on Web 03/05/2004

FIGURE 1. Schematic diagram of the improved water surface sampler. corresponds to the difference between the Hg flux measured with the water surface sampler and the estimated Hg(p) flux. On the basis of this concept, we evaluated the dry deposition fluxes and deposition velocities of Hg(p) and RGM individually.

Methods Improvement of the Water Surface Sampler. The water surface sampler used is similar to that of Yi et al. (12). A schematic diagram of the sampler is presented in Figure 1. It has four major parts: a water surface holder, a water surface plate, a pump, and a reservoir system (5 L Teflon bottle). The water surface plate is automatically covered when a moisture sensor detects rain or snow and is exposed to the atmosphere during dry periods. The sampler has a sensitive moisture sensor. This minimizes the amount of precipitation collected at the onset of a precipitation event. The sampler also records cumulative time when dry deposition sampling is conducted. Water is continuously supplied to the water surface plate with a pump and discharges over a weir on the outside edge of the plate. Several improvements of the sampler of Yi et al. (12) were conducted to prevent the contamination and adsorption of trace metals and also vaporization of mercury, and to reduce water evaporation during sampling. The water surface holder and plate were entirely made of Teflon. A tubing pump was used to simplify tubing and fittings, which often cause contamination. The circulated water on the water surface plate was kept at a temperature which is roughly 5-10 °C lower than air temperature by refrigerating the reservoir system to reduce water evaporation. This makes it possible to conduct a continuous deposition sampling without adding water for about 2 weeks in the case of using 5 L of water. The amount of residual water at the end of the sampling periods described in the next section varied from 1.4 to 5.8 L. In July, atmospheric water vapor conversely condensed on the water surface due to refrigeration. In this case, it appears that dew mixed rapidly with the water. Thus, dew may not be an important deposition pathway for atmospheric Hg, which is different from the case on the solid surface (13). In general, water evaporation is enhanced with increasing atmospheric water vapor saturation deficit that is related to air temperature and relative humidity, along with wind speed. If the extent of refrigeration for water is controlled on the basis of the monitored air temperature and relative humidity, evaporation or condensation may be suppressed effectively. The refrigeration of the circulated water may also be favorable for preventing mercury volatilization. In addition, to prevent the splash of water on the surface plate due to strong wind, the water surface plate was kept covered when the wind speed

monitored by an anemometer exceeded a predetermined limit (11 m s-1 in this study). However, those periods were negligibly small relative to the sampling periods. Dry Deposition Measurements Using the Water Surface Sampler. Dry deposition sampling using the water surface sampler was conducted during the following periods on the rooftop of a building (about 12 m above ground) on the property of Komae Research Laboratory, CRIEPI: May 1531, 2002 (dry period 320 h); July 16-25, 2002 (dry period 169 h); Sept 24 to Oct 4, 2002 (dry period 166 h); Nov 18 to Dec 2, 2002 (dry period 294 h). Meteorological data including wind speed and air temperature were also obtained concurrently on the rooftop (about 16 m above ground). The study site is located at Komae City in the western Tokyo metropolitan area. Nine municipal solid waste (MSW) incinerators (total incineration capacity 4350 tons d-1) are located within 10 km of the site, and no coal combustion facilities exist within the radius. Heavily industrialized areas including steel mills and petrochemical plants in addition to many MSW incinerators are present within a 10-20 km radius. Previously, Sakata et al. (14, 15) reported that the MSW incinerators are the predominant sources of mercury and other trace metals such as As, Cd, Pb, and Zn in the atmosphere. Prior to sampling, the water surface plate and 5 L Teflon bottle were cleaned with 6 mol L-1 HCl while being heated. Five liters of 0.5 mol L-1 HCl solution, hereafter referred to as the sampling solution, was used for collection of deposited materials. The bulk of the trace metals associated with atmospheric particles are expected to be soluble in this solution. This was supported by the fact that >90% of Cu, Pb, and Zn and >80% of Mn in rain samples (% for the total content with the HF-HNO3-HClO4 treatment) collected at the study site were dissolved in 0.25 mol L-1 HNO3, the acidity of which is lower than that of 0.5 mol L-1 HCl, during the storage of samples (>1 week) (16). Mercury in the HCl solution exists predominantly in forms of HgCl42- and HgCl3-; thus, mercury is probably stabilized without adsorption and volatilization after collection (17). This was confirmed by measuring the change in the concentration ratio of Hg (54.8 ng L-1) and Sr (4.37 mg L-1) previously added into the solution under atmospheric conditions as actual deposition sampling. The Hg/Sr ratio was used to cancel the increase in the Hg concentration due to water evaporation. The result indicated that initial and final Hg/Sr ratios during the period from Oct 4 to Oct 18, 2002 (average air temperature 19.5 ( 3.9 °C) were 1.26 × 10-5 and 1.18 × 10-5, respectively, when the ratios were corrected for the dry deposition of mercury. Thus, there was minimal difference between both ratios. On the other hand, there is the possibility that ambient Hg0 is being oxidized by the sampling solution due to its relatively high HCl concentration. Stratton et al. (17) tested the artifact formation of Hg(II) within the mist chamber using 0.25 mol L-1 HCl by oxidation of Hg0. The result showed that the formation of Hg(II) was enhanced with increasing ozone, but the rate of conversion was very small, reaching only 0.3% conversion of Hg0 in 50 min at 35 ng m-3 Hg0 and 100 ppb ozone. We believe that the artifact formation of Hg(II) in the sampling solution by oxidation of Hg0 was considerably smaller than that in the mist chamber because of the reaction under static conditions. Sampling began by pumping the sampling solution to the water surface plate at a rate of about 400 mL min-1. The water surface sampler was allowed to run for about 10 min. Then, 100 and 50 mL samples (blanks) were taken from the water surface using a 50 mL syringe to determine the background concentrations of mercury and other trace metals, respectively. At the end of the sampling period, two samples were taken in the same manner as blanks. The pump was shut off, and the residual sampling solution was weighed. BrCl (0.002 mol L-1) was added into the samples for mercury VOL. 38, NO. 7, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2191

measurement to oxidize all Hg compounds to Hg(II). All samples were stored in a refrigerator until analysis. Mercury concentration in the samples was measured by cold-vapor atomic fluorescence spectrometry (CVAFS; Tekran, model 2600), following Hg0 generation with SnCl2 as the reducing agent (18). The Cu, Mn, and Zn concentrations and the Cd and Pb concentrations were determined by inductively coupled plasma atomic emission spectrometry (ICP-AES; Seiko Instruments, SPS5000) and inductively coupled plasma mass spectrometry (ICP-MS; Micromass, platform ICP), respectively. The deposition flux (ng m-2 h-1) for trace metals was calculated by dividing the increase in their amount in the sampling solution during the sampling period by the hours of the dry period. Those values are independent of the occurrence of precipitation during the sampling period. Wet Deposition Measurements. Precipitation samples were collected at the study site using an automatic wet-only sampler, which is similar to the sampler of Landis and Keeler (19). The sampler was equipped with two sampling trains, which consisted of a borosilicate glass funnel (177 cm2 collection area), a Teflon tubing (10 mm inner diameter), and a 5 L Teflon sample bottle. A vapor lock to minimize vapor-phase Hg exchange was not incorporated. Sampling was conducted every half-month from Apr 2, 2002 to Apr 1, 2003 (total precipitation 1525 mm). The samples were stored in a sample bottle previously containing about 50 mL of 5 mol L-1 HCl solution to stabilize mercury and to dissolve particulate trace metals. BrCl (0.002 mol L-1) was added into the samples for mercury measurement. The concentrations of mercury and other trace metals in the samples were determined by CVAFS and ICP-MS, respectively. The wet deposition fluxes (µg m-2) for each month were calculated from these concentrations in the precipitation samples and the amount of precipitation. Measurements of Atmospheric Concentrations and Particle Size Distribution. For the determination of Hg(p), Cd, Cu, Mn, Pb, and Zn concentrations in air, airborne particles were collected on a quartz fiber filter (8 in. × 10 in.) using a high-volume air sampler (Shibata HV-1000F). Sampling was performed at a flow rate of 300 L min-1 during dry deposition sampling. Similarly, airborne particles in 13 size fractions between 12.1 µm were collected on a quartz fiber filter (80 mm diameter) using a multistage, multiorifice cascade impactor (Andersen air sampler, Tokyo Dylec, LP-20) to determine the size distribution of trace metals. The 50% cutoff diameters were 0.06, 0.13, 0.22, 0.33, 0.52, 0.76, 1.25, 2.5, 3.9, 5.7, 8.5, and 12.1 µm at a flow rate of 23.3 L min-1. Prior to sampling of particles, the filters were heated at 500 °C overnight to remove any Hg present on the filters. The filters were not greased because a quartz fiber filter minimizes particle bounce. Saido (20) found that there is little difference in size distribution between silicon greasecoated stainless steel and glass fiber filters. A glass fiber filter has a surface roughness similar to that of a quartz fiber filter. After collection of the samples, the filters were immediately placed in glass containers and stored in a refrigerator until analysis. Hg(p) concentration in the sample filter was measured by atomic absorption spectrometry (AAS) following thermal desorption and gold trap amalgamation (Sugiyamagen, 2537/2538). For the determination of Cd, Cu, Mn, Pb, and Zn concentrations, a portion of the sample filter was digested using HF-HNO3-HClO4. After evaporation to dryness, the residue was dissolved in 50 mL of 0.3 mol L-1 HNO3. The metals in this solution were measured by ICP-AES or ICP-MS. The RGM concentration in air was also measured by an automated mercury vapor analyzer (Tekran, 2537A) with a KCl-coated annular denuder sampling unit (Tekran, 1130) during dry deposition sampling. A sampling time of 2 h at a flow rate of 10 L min-1 was used for each measurement. 2192

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 7, 2004

FIGURE 2. Change in dry deposition fluxes with time for trace metals (December 2001 to November 2002). Landis et al. (21) found significant RGM artifact on quartz fiber filter samples collected for Hg(p) determination when KCl-coated annular denuders were not utilized. However, they observed no significant artifact when the RGM concentration was 12.1 µm, deposition velocity was obtained assuming a uniform diameter of 20 µm. The ratio of modeled fluxes to measured fluxes (a ratio of 1 means perfect agreement) varied by 0.1-2.6 for Cd, 0.40.9 for Cu, 0.3-1.3 for Mn, 0.2-1.2 for Pb, and 0.2-0.9 for Zn. The modeled fluxes generally did not contradict but tended to underestimate the measured fluxes. This may be primarily due to the uncertainty of the size distribution for particles larger than 12.1 µm, which highly affects the calculated deposition fluxes. In addition, the above findings support that the 84.13% diameter is a more appropriate index for each deposited metal than the 50% diameter (or mass median diameter, MMD) in the undersize distribution (Figure 5), because larger particles are responsible for the flux. The 84.13% diameter corresponds to a diameter of 50% diameter plus the geometric standard deviation as follows: ln dp84.13 ) ln dp50 + ln σg. Then, the deposition velocities for trace metals other than mercury, which exist entirely in particulate form, were plotted against their 84.13% diameters during each study period (Figure 7). There was a linear relationship between logarithms of the parameters. This regression equation was used for the estimation of the deposition velocity for Hg(p), as described in the next section. Figure 7 also indicates that there is no great difference in the relationship of vd - dp84.13 between the study periods. Among meteorological parameters, wind speed strongly affects deposition velocity. The average wind speed based on 10 min mean values during each period is given in Table 2. The lack of difference in the relationship of deposition velocity and particle size between the study periods is likely due to the similar wind speeds obtained in all periods. In the cases of long-term dry deposition measurements, the average wind speeds during the study period were similar; thus, the effect of wind speed on overall deposition velocity was very small. In addition, a large error associated with the measurements makes it difficult to distinguish the actual difference in the relationship of deposition velocity and particle size between the study periods. Estimation of Dry Deposition Fluxes and Deposition Velocities for Hg(p) and RGM. The dry deposition flux of mercury, FHg, is given as the sum of the deposition fluxes of Hg(p), FHg(p), and RGM, FRGM:

FHg ) FHg(p) + FRGM ) vd,Hg(p)CHg(p) + vd,RGMCRGM where vd,Hg(p) and vd,RGM are dry deposition velocities for Hg(p) and RGM and CHg(p) and CRGM are atmospheric concentrations for Hg(p) and RGM, respectively. The deposition velocity for Hg(p) was estimated on the basis of its 84.13% diameter using the regression equation provided in Figure 7. The appropriateness of using the 84.13% diameter was confirmed by the underestimates for deposition velocity based on the 50% diameter of up to 65% (July 2002). The result gave vd,Hg(p) values from 0.31 to 0.94 cm s-1, which were much higher than the values used for the previous model calculations. For example, Pai et al. (8) used 0.1 cm s-1 for water regions and 0.2 cm s-1 for land regions as the average Hg(p) annual dry deposition velocities. This may be attributed to Hg(p) being assumed to be primarily associated with fine particles and the velocity for Hg(p) associated with coarse particles being underestimated due to the model used. In urban areas and sites near the sources where the levels of coarse particles are much higher, the degree of underestimation for Hg(p) deposition velocity is supposed to be larger.

FIGURE 5. Undersize distribution for trace metals in atmospheric particles. Lines represent regression lines, which are obtained when the log-normal distribution is applied. dp84.13 denotes the 84.13% diameter, corresponding to a diameter of 50% diameter (dp50) plus the geometric standard deviation (σg) as follows: ln dp84.13 ) ln dp50 + ln σg. Next, the deposition fluxes for Hg(p) during the study periods were calculated from the estimated velocities and measured atmospheric concentrations. This result is given in Figure 8 as the ratio of the calculated flux for Hg(p) to the flux for Hg measured with the water surface sampler. The ratio varied from 0.24 to 2.8, and significantly decreased in July. Essentially this ratio could not be higher than 1. The calculated fluxes are expected to have a large error on the basis of the 95% confidence interval for the estimated vd

(Figure 7). Particularly for vd in December, a larger error might have been induced because estimation was carried out in the border region (dp84.13 ) 2.4 µm), as shown in Figure 7. Thus, it is regarded that the calculated fluxes for Hg(p) are almost equal to the measured flux for Hg during the study periods except in July if the large error is taken into consideration. On the other hand, the ratio of the calculated Hg(p) flux to the measured Hg flux was very low in July. This suggests VOL. 38, NO. 7, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2195

be questionable to evaluate their deposition on the basis of their average concentrations. The uncertainty associated with a large error makes it difficult to estimate the deposition velocity for RGM, although the RGM flux corresponds to the difference between the Hg flux measured with the water surface sampler and the estimated Hg(p) flux. In summary, the limited results presented here tend to support the hypothesis that deposited mercury is mainly in particulate form except during summer, and that significant amounts of RGM are deposited during summer. However, further research is required to elucidate the relative importance of Hg(p) and RGM deposition more accurately and to consequently validate this hypothesis.

Acknowledgments FIGURE 6. Calculated deposition velocities as a function of particle size, based on a model of Noll and Fang (29). The deposition velocities measured in this study are also given in the figure.

We thank T. Okabe (Techno Service, Inc.) for assistance in the sampling and analyses.

Literature Cited

FIGURE 7. Relationship between dry deposition velocity and 84.13% diameter for trace metals. The deposition velocities were calculated from the water surface measurements. Dotted lines in the figure represent the 95% confidence interval for the vd estimated using the regression equation.

FIGURE 8. Ratio of calculated deposition flux for particulate mercury to mercury deposition flux measured with the water surface sampler. that the deposition flux for RGM increased abruptly in July. The reason seems to be that summer generally favors the formation of RGM due to higher air temperature, solar radiation intensity, and atmospheric ozone concentration (34). The average concentrations of RGM observed during four study periods (May, July, September, and November 2002) were 12 ( 19 pg m-3 (n ) 195), 11 ( 14 pg m-3 (n ) 176), 6 ( 7 pg m-3 (n ) 113), and 3 ( 5 pg m-3 (n ) 145), respectively. This shows that their concentrations were extremely varied during each period and that the projected high mean value was not observed in July. For substances whose atmospheric concentrations are highly variable, it may 2196

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 7, 2004

(1) Strachan, W. M. J.; Eisenreich, S. J. Mass Balance of Toxic Chemicals in the Great Lakes: The Role of Atmospheric Deposition; International Joint Commission: Windsor, ON, Canada, 1988. (2) Landis, M. S.; Keeler, G. J. Environ. Sci. Technol. 2002, 36, 45184524. (3) Rolfhus, K. R.; Sakamoto, H. E.; Cleckner, L. B.; Stoor, R. W.; Babiarz, C. L.; Back, R. C.; Manolopoulos, H.; Hurley, J. P. Environ. Sci. Technol. 2003, 37, 865-872. (4) Sehmel, G. A. Atmos. Environ. 1984, 14, 983-1011. (5) Davidson, C. I.; Lindberg, S. E.; Schmidt, J. A.; Cartwright, L. G.; Landis, I. R. J. Geophys. Res. 1985, 90, 2123-2130. (6) Nicholson, K. W. Atmos. Environ. 1988, 22, 2653-2666. (7) Holsen, T. M.; Noll, K. E. Environ. Sci. Technol. 1992, 26, 18071815. (8) Pai, P.; Karamchandani, P.; Seigneur, C. Atmos. Environ. 1997, 31, 2717-2732. (9) Dolske, D. A.; Gatz, D. F. J. Geophys. Res. 1985, 90, 2076-2084. (10) Davidson, C. I.; Wu, Y. Deposition of Particles and Vapors. Acidic Precipitation: Sources, Deposition, and Canopy Interactions; Springer-Verlag: New York, 1990; Vol. 3, pp 152-170. (11) Noll, K. E.; Fang, K. Y.; Watkins, L. A. Atmos. Environ. 1988, 22, 1461-1468. (12) Yi, S.-M.; Holsen, T. M.; Noll, K. E. Environ. Sci. Technol. 2002, 36, 2815-2821. (13) Malcolm, E. G.; Keeler, G. J. Environ. Sci. Technol. 1997, 31, 272-278. (14) Sakata, M.; Kurata, M.; Tanaka, N. Geochem. J. 2000, 34, 23-32. (15) Sakata, M.; Marumoto, K. Atmos. Environ. 2002, 36, 239-246. (16) Marumoto, K. Unpublished results. (17) Stratton, W. J.; Lindberg, S. E.; Perry, C. J. Environ. Sci. Technol. 2001, 35, 170-177. (18) Fitzgerald, W. F.; Gill, G. A. Anal. Chem. 1979, 51, 1714-1720. (19) Landis, M. S.; Keeler, G. J. Environ. Sci. Technol. 1997, 31, 26102615. (20) Saido, J. Study on Prevention of Particle Bounce in Size Distribution Measurements Using a Multinozzle Cascade Impactor. Master’s Thesis, Kitasato University, 1998 (in Japanese). (21) Landis, M. S.; Stevens, R. K.; Schaedlich, F.; Prestbo, E. M. Environ. Sci. Technol. 2002, 36, 3000-3009. (22) Sakata, M.; Marumoto, K. J. Phys. IV 2003, 107, 1177-1180. (23) Yi, S.-M.; Shahin, U.; Sivadechathep, J.; Sofuoglu, S. C.; Holsen, T. M. Atmos. Environ. 2001, 35, 1133-1140. (24) Shahin, U.; Yi, S.-M.; Paode, R. D.; Holsen, T. M. Environ. Sci. Technol. 2000, 34, 1887-1892. (25) Odabasi, M.; Muezzinoglu, A.; Bozlaker, A. Atmos. Environ. 2002, 36, 5841-5851. (26) Herut, B.; Nimmo, M.; Medway, A.; Chester, R.; Krom, M. D. Atmos. Environ. 2001, 35, 803-813. (27) Sehmel, G. A. Aerosol Sci. 1973, 4, 125-138. (28) Slinn, S. A.; Slinn, W. G. N. Atmos. Environ. 1980, 14, 10131016. (29) Noll, K. E.; Fang, K. Y. P. Atmos. Environ. 1989, 23, 585-594. (30) Kondo, J. Meteorology in Water Environment; Asakura-Shoten: Tokyo, 1994 (in Japanese). (31) Holsen, T. M.; Noll, K. E.; Fang, G. C.; Lee, W. J.; Lin, J. M. Environ. Sci. Technol. 1993, 27, 1327-1333.

(32) Zufall, M.; Davidson, C.; Caffrey, P. F.; Ondov, J. M. Environ. Sci. Technol. 1998, 32, 1623-1628. (33) Paode, R. D.; Sofuoglu, S. C.; Sivadechathep, J.; Noll, K. E.; Holsen, T. M. Environ. Sci. Technol. 1998, 32, 1629-1635. (34) Lindberg, S. E.; Stratton. Environ. Sci. Technol. 1998, 32, 49-57.

Received for review May 9, 2003. Revised manuscript received January 21, 2004. Accepted January 27, 2004. ES030467K

VOL. 38, NO. 7, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2197