Environ. Sci. Technol. 2002, 36, 3763-3771
Precipitation Scavenging of Polychlorinated Biphenyls and Polycyclic Aromatic Hydrocarbons along an Urban to Over-water Transect J O H N H . O F F E N B E R G * ,† A N D JOEL E. BAKER Chesapeake Biological Laboratory, The University of Maryland Center for Environmental Science, Solomons, Maryland 20688
Concentrations of semi-volatile organic compounds in dissolved and particulate forms in precipitation are related to those measured concurrently in air measured at multiple locations along the tracks of several storms. In 14 paired (air and precipitation) samples collected at urban, over-water, and downwind stations around southern Lake Michigan, compound-specific total (gas + particle) precipitation scavenging ratios range from 180 to 8.2 × 107 (fluorene and benzo[a]pyrene, respectively) for PAHs and from 0.5 to 1.1 × 107 for PCBs (PCB26 and PCB194, respectively). Particle scavenging, rather than gas scavenging, is the dominant removal mechanism for both PAHs and PCBs in all samples collected along the urban to overwater transect. Variations in measured total scavenging ratios within the 14 samples are large (from 69 to 1.00 × 106 for ∑-PAHs and from 290 to 88 000 for total ∑-PCBs), with larger variation between samples of differing storms collected at a single location than among samples of the same storm collected along the urban to rural transect. This minimal variance of scavenging ratio during the transport of a storm across multiple locations holds in two of three storms measured at multiple locations along the urban to over-water to rural storm track. This suggests that scavenging mechanisms are largely variable between storms and that the relative importance of each mechanism can occasionally also vary greatly during the progression of the storm from the urban to downwind locations. Furthermore, precipitation scavenging appears to be more variable between storms and generally less variable during progression of a storm down transect, indicating that meteorologic and precipitation characteristics, generally referred to here as storm type, control precipitation scavenging.
Introduction Semi-volatile organic compounds are removed from the atmosphere via gas exchange with the surface, by dry deposition to the surface, and because of precipitation scavenging, also know as washout. The residence time and * Corresponding author phone: (732)932-3097; fax: (732)932-8644; e-mail:
[email protected]. † Present address: Department of Environmental Sciences, Rutgers, The State University of New Jersey, 14 College Farm Rd., New Brunswick, NJ 08901. 10.1021/es025608h CCC: $22.00 Published on Web 08/02/2002
2002 American Chemical Society
distribution of semi-volatile organic compounds in the atmosphere depends on their distributions between vapor, gaseous, and aqueous phases (1, 2). The distribution between these phases is controlled by the compound’s vapor pressure, Henry’s law constant, and aqueous solubility along with the concentration, size distribution, and physical characteristics of aerosols in the atmosphere. Gaseous compounds dissolve into water both within clouds and into falling rain at rates determined by the equilibrium condition (as described by Henry’s law) and by mass transfer and mixing characteristics (i.e., molecular diffusion and in-drop mixing). Simultaneously, particles are scavenged from the atmosphere via physical processes controlled by size, number density, and solubility of aerosols along with meteorological conditions and microphysics. Particles less than 0.05 µm collide with the falling raindrop by Brownian motion whereas particles greater than 1 µm are primarily removed by impaction with the front of the falling raindrop. Alternatively, they may be swept into the water droplet by eddies formed in the lee of the falling water droplets (3). Aerosol particles between 0.05 and 1 µm are strongly influenced by phoretic and Coulombic forces (4, 5). Slinn (6) demonstrated that particle scavenging is a sensitive function of particle size for particles with radii between 0.1 and 10 µm, with an observed minimum in collision efficiency between particles and raindrops near 0.1 µm with maxima at radii much less than 0.1 and greater than 10 µm. Particle scavenging is an extremely efficient mechanism of removal of PAHs from the atmosphere (7-9) based upon the thorough investigataion of seven and five storms, respectively. Similarly, Duinker and Bouchertall (10) found particle scavenging to be the dominant mechanism for PCBs in three samples. Poster and Baker (11) found seemingly supporting evidence in five rain samples, although particlephase air concentrations of PCBs were not measured a part of their study. There have been no confirmatory experiments to date largely because of the complexities involved with simultaneous collection of event-based precipitation samples coupled with short duration air sampling. Field campaigns (12, 13) have provided the occasion for such simultaneous sampling efforts and thus allow for more detailed investigations into the mechanisms of precipitation scavenging of semi-volatile organic contaminants from the atmosphere. Furthermore, these campaigns have allowed for the unique opportunity of investigating the changes in scavenging mechanisms along the track of several storms, from the urban to over-water to downwind atmospheres surrounding southern Lake Michigan. The size-dependent scavenging model of Poster and Baker (9) is applied to concentrations of PAHs and PCBs measured in event-based precipitation and simultaneously collected samples in and around Chicago, IL. Using temperaturecorrected Henry’s law constants (14-16) and measured phase distributions of each compound in both air and precipitation (12, 13, 17), we present scavenging ratios for gas phase as well as large and small particle-bound individual PAHs and PCB congeners calculated from concentrations measured in precipitation events and coincidental air samples in the atmosphere over Chicago, over southern Lake Michigan, and over southwestern Michigan. We find strong support of very efficient particle washout in all atmospheres tested for both PAHs and PCBs. This study provides a unique perspective on the development of scavenging mechanisms during the course of a storm as it is the first study of semi-volatile organic contaminant scavenging ratios conducted on storms measured at multiple locations along the storm track. Spatial VOL. 36, NO. 17, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3763
and temporal variations in washout ratios are quantified and discussed relative to previous measurements at single locations.
Background Previous models of total removal of semi-volatile organic contaminants from the atmosphere by precipitation relate the total contaminant concentration in the surrounding air to that observed in precipitation (7, 8):
Wt ) Crain/Cair ) Wpφ + Wg(1 - φ)
(1)
where Wt, Wp, and Wg are total, particle, and gas scavenging ratios, respectively, and φ is the fraction of contaminant in the air that is bound to particles. The particle scavenging ratio is defined as Cp,rain/Cp,air where Cp,rain is the particleassociated concentration in rainwater and Cp,air is the concentration of the compound of interest bound to airborne particles. Likewise, the gas scavenging ratio is defined as Cd/Cg where Cd is the dissolved organic contaminant concentration in rainwater and Cg is the gaseous contaminant concentration in the atmosphere. The fraction of contaminant bound to particles (φ) is defined as Cp,air(Cp,air + Cg). This expression of the compound’s gas/particle distribution is directly related to compound vapor pressure, temperature, and aerosol surface characteristics (1). The portion of the contaminant in the gas phase will behave in accordance with Henry’s law when distributing into the water droplets of precipitation. Thus, at equilibrium the gas scavenging ratio equals the dimensionless Henry’s law constant (RT/H) where R is the universal gas constant (8.21 × 105 m3 atm-1 mol-1 K-1), T is temperature (K), and H is the Henry’s law constant (atm m-3 mol-1). Particle scavenging, however, is highly complex and is determined by both meteorology and size of both particles and rain droplets (18-20). The relative importance of gas and particle scavenging on the total removal of contaminants from the atmosphere depends on the relative magnitudes of the terms Wpφ and Wg(1 - φ). Poster and Baker (9) modified eq 1 to include the scavenging of submicron particles that are present in the filtrate of precipitation samples:
Wt ) Crain/Cair ) Wp,fφp,f + Wp,nfφnf + Wg(1 - φt) (2) where φp,f, φnf, and φt are the fractions of contaminant associated with filter-retained particles, nonfilter-retained particles, and all particles. This model relates the respective scavenging of operationally defined large and small particles from the atmosphere with the equilibrium partitioning of contaminants between the gas phase and the dissolved phase within falling raindrops. This relationship assumes no changes in particle size upon entrainment in the water droplet such as reduction in particle size due to partial dissolution of the aerosol or growth of the particle due to coagulation of small particles upon entrainment within the falling precipitation. Furthermore, this model assumes minimal importance of re-distribution once the contaminant has entered the drop while falling and before it is filtered in the rain sampling apparatus. A model of this type could be expanded to include a broader range of particle sizes, each exhibiting its own washout ratio, with the total of all sizes summing to the aforementioned Wp as in eq 1:
Wt ) Crain/Cair ) Wp,1φp,1 + Wp,2φp,2 + ... + Wp,xφp,x + Wg(1 - φt) (3) The difficulty in the application of such a model revolves around the complexities in attributing particles of various sizes to their original atmospheric diameter and inefficiencies 3764
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 17, 2002
FIGURE 1. Location of precipitation sampling sites around southern Lake Michigan. in separating small particles in rainwater. As such, the model of Poster and Baker (9) can be applied to examine the relative influences of operationally defined small and large particles on the total contaminant removal from the atmosphere by precipitation. This model is applied to samples collected in and downwind of Chicago, IL, in order to examine the influences of the urban atmosphere on precipitation scavenging of semi-volatile organic contaminants.
Methods Precipitation events and ambient air concentrations were measured concurrently at multiple locations during a series of field experiments to determine the influence of Chicago’s urban atmosphere on the concentrations of atmospheric pollutants in Lake Michigan (see Figure 1; 12, 13, 21-30). Event-based precipitation samples were collected at the onland sites by automated, wet-only collectors equipped with a 1-m2 stainless steel funnel connected to an in situ filtration system consisting of a glass fiber filter (90 mm diameter, Schleicher & Schuell No. 25) followed by a column of adsorbant (Amberlite XAD-2; 12, 31). To include particles that remain on the funnel rather than being washed into the filtration system, the funnel was manually scrubbed with clean glass wool wetted with deionized water after each precipitation event. This funnel wipe sample was combined with the filter sample for analysis. Prior study has shown that the funnel wash contained an important portion of the total particulate PCBs in rural areas and largely consists of particles. Details describing operation of these automated samplers have been presented elsewhere (9, 11, 12, 31, 32). Over-water precipitation samples were manually collected using similar techniques (12). At all locations, 12-h, groundlevel air samples were collected simultaneously by collocated high-volume air samplers, thus giving particle-bound and gas-phase concentrations of the compounds of interest (13).
The use of time-integrated, ground-based air measurements for the calculation of air column-integrated scavenging mechanisms does leave open some uncertainties. Because of operational constraints, air column and in-cloud sampling were not performed as part of this field experiment. Detailed sampling of the precipitation at various heights in the air column would allow for a better understanding of the processing of these chemicals by precipitation and of the relative importance of in-cloud versus below-cloud scavenging mechanisms and is further discussed later in the text. The results presented herein focus on the differences in precipitation washout of semi-volatile organic contaminants as measured at the ground level in the urban atmosphere from those measured along the downwind gradient. Model Application. The SOC scavenging model of Poster and Baker was applied to concurrently measured precipitation and air samples collected in the Chicago/Lake Michigan area during July 1994 and January 1995. Samples were collected simultaneously at three locations, thereby allowing observation of the development of each precipitation event across the urban/coastal atmosphere. Scavenging ratios are calculated following the method of Poster and Baker (9) for precipitation events using both the ambient atmospheric concentrations and the measured concentrations in the precipitation collected simultaneously at urban, over-water, and rural locations in the southern basin of Lake Michigan. Phase distributions for both air and precipitation samples were operationally defined as the filter-retained and the adsorbant-retained measured concentrations. Temperature corrected Henry’s law values are calculated for the ambient atmospheric temperature at the time of the precipitation event (12). The values are based upon the measured relationships of Bamford et al. (14) for 13 PAHs and of Bamford et al. (15) for 26 PCBs. The remaining PCBs were estimated based upon the work of Bamford et al. (16), while the remaining PAHs were interpolated based upon the method of Gigliotti et al. (33). The fraction of particle-bound PAH associated with large and small particles was apportioned according to measured distributions made at the same locations during the sampling intensives (27). PAH size distributions were measured over 12-h periods while PCB distributions were measured over longer periods, typically 72 h, to collect enough material. Not all such size distributions were concurrent to precipitation sampling, although systematic differences according to wind regime were observed for PAHs making prediction of distribution possible for offshore samples (27). Unfortunately, concentrations of PCBs on particulate matter are too low to allow for short duration measurements of size distributions, which could allow for development of meteorologically based predictive models (23). Thus, for PAHs the total particlebound concentrations measured simultaneously at all locations with all precipitation samples are distributed between operationally defined large and small fractions according to the size distributions measured during the same sampling intensives, although not always at the same times. PCB distributions were estimated using a more generalized method due to the lengthy integrated periods of sampling. Nevertheless, neither PAHs nor PCB size distributions varied greatly enough to play a significant role in the importance of large versus small particle washout by precipitation. To determine if measured scavenging ratios that differ by more than 1 order of magnitude are truly different, the total error of each scavenging ratio was estimated. Errors were based upon reported standard errors of the weighted mean of the volume-weighted mean concentration at each location (12) in conjunction with the reported uncertainties of the measured air concentrations (13). Figures 1 and 2 represent these estimations of total error with error bars representing, on average, ∼(100% total error in the measured scavenging
ratios. These estimates may be off by a factor of roughly 2-3 due mainly to the small number of precipitation samples at each location. Even if the error associated with the calculated scavenging ratios is approximately (500%, then differences of 2 orders of magnitude (i.e., ∼10000% increase in Chicago over that measured over Lake Michigan on July 20, 1994) are different.
Results and Discussion Concentrations and speciations (operationally defined filterretained and filtrate phases) of PAHs and PCBs in 14 precipitation events and their corresponding air samples at three locations around southern Lake Michigan have been reported previously (12, 13). The distribution of particlebound contaminants between small and large particles is based upon the measured size distributions of Offenberg and Baker (27) for PAHs and of Franz et al. (23) for PCBs. The results of the following calculations show a limited effect of reasonably small changes in the size distributions on the attribution of importance of removal mechanisms for both PCBs and PAHs. Scavenging ratios calculated from speciated air and precipitation samples are available from the authors and are in the Supporting Information, along with detailed lists of all compounds included in the analyses described by Offenberg and Baker (12) and Simcik et al. (13). Total scavenging ratios (gas + particle) for individual polycyclic aromatic hydrocarbons ranged from 180 to 8.2 × 107 (fluorene, at site LM1 on July 21, 1994, and benzo[a]pyrene at South Haven, MI, on July 17, 1994, respectively). Similarly, total scavenging ratios for individual polychlorinated biphenyl congeners range from 2.7 to 1.1 × 107 (PCB118, 2,3′,4,4′,5-pentachlorobiphenyl, at site LM5 on July 20, 1994, and PCB194, 2,2′,3,3′,4,4′,5,5′-octachlorobiphenyl, at site LM5 on July 25, 1994, respectively) while total ∑-PCB scavenging ratios ranged from 290 to 88 000 (site LM 5 on July 20, 1994, and LM5 on July 25, 1994, respectively). Generally, total scavenging ratios vary among the individual PAHs by more than 3 orders of magnitude at each of the three locations (Figure 2). Likewise, total scavenging ratios for PCBs are consistently widespread across the three locations (Figure 3), spanning over 4 orders of magnitude for a given congener across the limited number of samples described herein (n ) 14). Total scavenging ratios tended to be greater for less volatile compounds, ones with relatively more in the particle-bound phase in air for both PCBs and PAHs, although no clear trend existed among all samples collected. For two of the three events that were sampled at multiple locations, total compound-specific scavenging ratios for both PAHs and PCBs did not change significantly during the course of the progression of the storm from the urban to over-water to downwind locations. There is some scatter in the relative enrichment for both the July 19 and July 21 storms for both PCBs and PAHs but little more than a 10-fold difference between the measured washout ratios at the urban location and the downwind stations. Atmospheric concentrations of both PAHs and PCBs decreased with distance from the urban source region (13), and aerosol concentrations and contaminant size distributions changed along the urban to downwind transect (27). The changes in the physiochemical character of the aerosols being scavenged by the precipitation would suggest that the efficiency at which the particles are scavenged should be expected to change along the urban to rural transect. Thus, the general constancy of scavenging ratios within the two single events on July 19 and 21, 1994, as measured sequentially at multiple locations, is likely related to the characteristics of the storms themselves and not to the aerosols to which the compounds of interest are bound. Thus, while contaminant concentrations and aerosol characteristics have been VOL. 36, NO. 17, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3765
FIGURE 2. Total washout ratios (Crain/Cair) measured at urban, over-water, and downwind rural locations plotted vs total washout ratio measured at urban location for individual PAHs in three storms measured at multiple locations along each storm track: July 19, July 20, and July 21, 1994. measured to change along the transect, the changes or apparent lack of changes in the storm characteristics (e.g., precipitation rate, drop size, etc.) are the dominant factors controlling the observed constancy of the measured scavenging ratios of these compound classes. This is further complicated by the use of time-integrated (i.e., 12 h), groundlevel measurements of the concentrations and phase distributions of the contaminants being scavenged by the 3766
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 17, 2002
precipitation. The true concentrations and distributions at the point of interaction between the air and the falling rain are likely not well characterized by integrated surface measurements. The July 20, 1994, storm, which had the highest measured scavenging ratios of any of the storms at any of the locations, provides an interesting exception to the otherwise limited variability of washout ratios during the progression described
FIGURE 3. Total washout ratios (Crain/Cair) measured at urban, over-water, and downwind rural locations plotted vs total washout ratio measured at urban location for PCB congeners in three storms measured at multiple locations along each storm track: July 19, July 20, and July 21, 1994. above. The total scavenging ratios of the four PAHs with highest values of Wt (benzo[b+k]fluoranthene, benzo[ghi]perylene, retene, and 1-methylphenanthrene) and all PCB congeners decreased 20-100 times at the two subsequent, downwind locations relative to the those at the urban site during this storm. Interestingly, lower molecular weight PAHs showed little change in scavenging ratios as the storm progressed over Lake Michigan to South Haven. This event
deposited rainwater with the highest concentration of ∑-PCBs and second highest concentration of ∑-PAH of any of the storms measured (12). This event deposited a much smaller proportion of the ∑-PCBs as nonfilter-associated material (i.e., gas + small particle scavenging) at nonurban locations than all of the others (12). This large particle dominated scavenging occurred in a strongly convective thunderstorm, while the other storms described herein were either localized VOL. 36, NO. 17, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3767
FIGURE 4. Relative contributions of submicron (small particle), filter-retained (large particle), and gaseous (Henry’s law) individual PAHs in urban, over-water, and rural downwind locations for three storms progressing across Chicago/southern Lake Michigan: July 19, July 20, and July 21, 1994. thunderstorms or frontal systems. These differences may be temperature-corrected Henry’s law constant. Thus, dimeninterrelated and account for the greatly increased scavenging sionless gas scavenging ratios range from 130 (acenaphthene ratios at the urban location during the event on July 20, 1994. in Chicago on July 19, 1994) to 12700 (indeno[1,2,3-cd]pyrene At equilibrium, gas scavenging of PAHs and PCBs, as in Chicago on January 19, 1995) for PAHs and from 25 (PCB described in the above equations, is equal to the compound’s 207, 2,2′,3,3′,4,4′,5,6,6′-nonachlorobiphenyl in Chicago on 3768
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 17, 2002
FIGURE 5. Relative contributions of submicron (small particle), filter-retained (large particle), and gaseous (Henry’s law) PCB homologues in urban, over-water, and rural downwind locations for three storms progressing across Chicago/southern Lake Michigan: July 19, July 20, and July 21, 1994. July 20, 1994) to 230 (PCB 194, 2,2′,3,3′,4,4′,5,5′-octachlorobiphenyl in Chicago on January 19, 1995) for PCB congeners. In all cases, gas scavenging has the least relative importance as a scavenging mechanism for individual PAHs (Figure
4) and PCB congeners (Figure 5). The temperature dependence of Henry’s law constants does not affect major changes in the relative importance of gas scavenging over the temperature range of the samples explored in this study (3VOL. 36, NO. 17, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3769
24 °C). Gas scavenging decreases in relative importance with decreasing compound volatility for both PCBs and PAHs. Low vapor pressure compounds are not present in the gas phase in sufficient quantities to be scavenged by the falling precipitation in appreciable quantities relative to the particle scavenging mechanisms, which are enhanced due to higher proportions of low volatility compounds bound to the particle phase. Scavenging ratios for nonfilter-retained (i.e., submicron) particles range from 350 to 3.1 × 108 for PAHs (indeno[1,2,3cd]pyrene, at LM1 on July 21, 1994, and 1-methylfluorene, in South Haven on July 17 1994, respectively) and from 830 to 7.6 × 107 for PCBs (PCBs 31+28, 2,4′,5-trichlorobiphenyl + 2,4,4′-trichlorobiphenyl, in Chicago on January, 19, 1995, and PCB 91, 2,2′,3,4′,6-pentachlorobiphenyl, in Chicago on July 20, 1994, respectively). Generally, nonfilter-retained particle scavenging is of the same relative importance as gas scavenging. The relative importance of small particle scavenging increases for low volatility compounds, which are not found in the gas phase and are not strongly influenced by gas scavenging. For these exceptions, the small particle scavenging becomes relatively more important and occasionally outweighs scavenging of large particles. Scavenging ratios for large, filter-retained particles range from 8700 to 4.7 × 108 for PAHs (dibenz[ah,ac]anthracene, at LM5 on July 20, 1994, and acenaphthene, in South Haven on July 17, 1994, respectively) and from 72 to 4.1 × 109 for individual PCBs (PCB26, 2,3′,5-trichlorobiphenyl, at LM5 on July 20, 1994, and PCBs 24+27, 2,3′,6-trichlorobiphenyl + 2,3′,6′-trichlorobiphenyl at LM5 on July 25, 1994, respectively). In almost all cases, large particle scavenging is the most important mechanism for PAHs and PCBs. Most of the cases in which large particle scavenging is not the most important mechanism measured are due to measured concentrations that were below the operationally defined limits of detection in the precipitation filter-retained, particulate-phase measurements. The concentrations on large particles, which were below limits of detection, thereby eliminated the large particle scavenging from these results and thus allowed small particle and/or gas scavenging to dominate. The relative importance of large particle scavenging increases with decreasing compound volatility (i.e., increasing molecular weight and/ or chlorine substitution) for both compound classes. Low vapor pressure compounds are not present in the gas phase in sufficient quantities to exhibit significant gas scavenging. These compounds are also more prevalent in the particlebound phase and thus are predominantly scavenged from that phase. There is little change in the relative importance of the mechanisms of scavenging along the urban to downwind transect. Variation in measured scavenging ratios within the 14 samples are extremely large, with greater variation among the samples from different storms collected at a single location than between samples of the same storm collected along the urban to rural transect. This suggests that while scavenging mechanisms are largely variable between storms, the relative importance of each mechanism generally does not vary greatly as a storm progresses along the urban to downwind transect. Large particles dominate the scavenging of both PAHs and PCBs in all samples collected. The few instances where large particle scavenging appears not to be the clearly dominant process most often result from missing (unmeasurable) concentrations of the large particle fraction in either rain or air samples used for these calculations. While both PAHs and PCBs span differing ranges in physical properties, such as Henry’s law and vapor pressure, these compound classes undergo similar scavenging mechanisms. There is generally good agreement between the two compound classes, suggesting that the scavenging mechanisms are controlled by the same governing principles and 3770
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 17, 2002
mechanisms. Both PAHs and PCBs are predominantly in the gas phase (>90%; 13), but particle scavenging mechanisms dominate their removal during precipitation events for both compound classes. Furthermore, the wide range in Henry’s law values across these compound classes (0.20-98.3 Pa m3 mol-1 at 298 K) does not appear to play a large role in determining the relative importance of gas scavenging from any of the atmospheres measured. Previous measurements of washout ratios for PAHs have been calculated for a limited number of samples at locations both adjacent to the Lake Superior and Chesapeake Bay and in Portland, OR, and Kiel, Germany (34, 35, 7-11, 32). The earliest of these studies showed a widely varying importance of gas and particle scavenging among a broad range of compounds (7, 8). Another effort focusing only on the washout of PCBs near the Baltic Sea (10) found that scavenging of particle-bound PCBs was the major source of PCBs measured in the three precipitation samples collected. Despite the small fraction of PCBs found bound to the airborne particulate matter (
