Environ. Sci. Technol. 1984, 18, 580-587
Particulate Standard";U.S.EPA Washington, DC; Office of Air Quality Planning and Standards, memorandum dated Oct 30, 1981.
(10) Frank, N. H.; Curran, T. C. "Statistical Aspects of a 24-Hour National Ambient Air Quality Standard for Particulate Matter"; Air Pollution Control Association: Pittsburgh,PA,
1982; paper 82-23.8.
Received for review March 17,1983. Revised manuscript received January 23,1984. Accepted January 31,1984. This work has been supported in part by funding from the American Iron and Steel Institute.
Urban Runoff as a Source of Polycyclic Aromatic Hydrocarbons to Coastal Waters Eva J. Hoffman," Gary L. Mills,t James S. Latlmer, and James G. Quinn
Graduate School of Oceanography, University of Rhode Island, Narragansett, Rhode Island 02882
B Urban runoff samples collected from four storm drains, each serving a different land use, were analyzed for selected polycyclic aromatic hydrocarbons (PAHs) by gas chromatography. The PAH concentrations varied widely during the history of each storm and appeared to be most concentrated in first or second flushes. Higher molecular weight PAHs were mostly found associated with the particulates and were enriched on two different particle sizes, a small particle fraction and a larger particle fraction, perhaps from different sources. The annual input of PAHs to the upper Narragansett Bay watershed was calculated by using storm loads with local rainfall and land use data. The loads of PAHs [mass (drainage area)-' year1] in urban runoff were higher at the highway and industrial land uses in comparison to the commercial and residential areas. Urban runoff PAHs are similar in composition to that found in atmospheric fallout but unlike the PAHs in municipal effluents. Comparison of urban runoff PAH inputs to other sources entering Narragansett Bay showed that urban runoff accounted for 71% of the total inputs for higher molecular weight PAHs and 36% of the total PAHs.
Introduction The sources of polycyclic aromatic hydrocarbons (PAHs) to the marine environment have been the subject of a number of recent investigations (1-5). Several PAHs are naturally occurring (e.g., retene and perylene) (5), but recent evidence suggests the majority of PAHs found in estuarine and coastal sediments originate from atmospheric fallout of particulates from combustion sources, including forest fires (1)(which could account for PAHs in sediments with dates prior to the industrial revolution) and fossil fuel combustion (6),especially coal (5). Estuarine sediments also show evidence of PAH contributions from petroleum contamination. For example, Lake et al. (3) calculated that sediments from upper Narragansett Bay had PAH alkyl homologue distributions which could be approximated by 80% combustion sources and 20% used motor oil. There are a number of investigations which suggest that urban runoff is also an important pathway of PAHs into the aquatic environment by examination of patterns in sediments close to urban areas (7,8).Although this indirect evidence of urban runoff contributions to the aquatic environment has been observed in the sedimentary record, there have been very few direct studies of PAHs in urban runoff. A study of storm runoff in the Los Angeles River during one storm produced a mass emission rate of "total aromatics" of 4.1 tons/storm which was projected to an 'Present address: Savannah River Ecology Laboratory, Aiken, SC
29801. 580
Environ. Sci. Technol., Vol. 18, No. 8, 1984
annual discharge rate of 311 tons/year (9) or 100 g capita-l year-l. Another study, whose main emphasis was on determination of PAH extraction efficiencies, determined PAH concentrations in highway runoff for one sample (10). Another (11)examined dibenzothiophene in urban runoff from three storms. The most detailed study of PAHs in runoff is that of Herrmann (12)) who investigated the transport of four selected PAHs in a river flowing through the urbanized region of Bayreuth, West Germany. Although urban runoff was not examined directly, its chemical impact on the river during wet weather was studied. The objectives of this study were (1)to determine what factors control inputs of PAHs into urban runoff (i.e., suspended solids, land use, flush phenomena, antecedent dry periods, and flow rates) and (2) to estimate directly the input rate of PAHs to the coastal environment via urban runoff.
Methodology Description of Sampling Location. Urban runoff samples were collected from four different storm drains each serving a different land use: suburban residential, commercial (shopping center), heavy industrial, and multilane highway. A summary of the pertinent information concerning the sampling locations has been previously reported (13). The specific drains selected for study were chosen on the basis of several criteria. The most important of these are the following: (1)the drainage basin served by the drain is predominantly one land use type; (2) the drainage system is well charted so that the area served could be calculated; (3) the land use served by the drain is typical of the land use in Rhode Island; (4) the outfall is accessible. Description of Storm Events. Rain events for sampling were chosen to provide a variety of rainfull amounts at each land use. These events were selected with the aid of forecasts by local National Weather Service (NWS) personnel in Warwick, RI, and computer projections of expected rainfall amounts. Timing logistics were facilitated with cooperation of a local television station, WJAFL, which had a weather radar data receiver providing real time output from the NWS radars at Worcester and Chatham, MA. A total of 18 storms were monitored, with rain amounts varying from 0.13 to 6.76 cm. A complete list of the storm events has been given elsewhere (13). Sampling Procedures. Urban runoff samples at each location were collected manually with a metal bucket from the drain outfall. Sampling was started at the initiation of rainfall and ended when the drain flow subsided following the end of the storm. Samples were usually collected once every 30 min, more frequently when flow rates
0013-936X/84/0918-0580$01.50/0
0 1984 American Chemical Society
INCREASING
7iME
AND 1-EMPERATURE -*
Flgure 1. Gas chromatogram of the f, fraction of an urban runoff sample collected at the interstate highway location. The peaks labeled are those PAHs reported in this study (internal standard = anthracene).
were high or rapidly changing, and less frequently during quiescent periods. At the time of sample collection, flow rates, cumulative rainfall, and local time were recorded. After collection, the samples were stored in precleaned 4-L amber glass bottles with Teflon-lined caps. Further sampling details have been reported elsewhere (13). Analytical Procedures. Shortly upon return to the laboratory, the urban runoff samples were filtered through preweighed, precombusted glass fiber filters (Whatman GF/C, particle size retention of 1.2 pm). After drying at room temperature, usually for less than 12 h, the filters were reweighed to determine the amount of suspended solids in each sample and then frozen until further analysis. Experiments on particle size distributions were also performed soon after return to the laboratory (14). This procedure involved passing 12-16 L of urban runoff through metal sieves of increasingly smaller mesh openings and then finally through a Whatman GF/C glass-fiber filter. The trapped particulates on the various sieves were back-rinsed onto filters, and each size class was analyzed separately. After addition of internal standards, a 3-L aliquot of each filtrate containing “soluble” components (operationally defined as that which passes through the filter) was extracted once with 300 mL of dichloromethane in the original sample container. The organic extract was then isolated and stored for further analysis. The sample filters and filtrate extracts were analyzed for particulate and soluble hydrocarbons, respectively. Briefly, the filters were shredded and transferred to a round bottom flask, internal standards were added, and then the filters were extracted with 0.5 N KOH in 9O:lO methanol-water under reflux conditions for 2 h. After additional water was added, the nonpolar components were partitioned from the basic aqueous-methanol phase into petroleum ether. The petroleum ether extracts (containing extractable components from the particulate phase) and the dichloromethane extracts (containing the extractable components from the soluble phase) were reduced in volume and solvent exchanged into hexane. The hydrocarbons in the hexane extracts were isolated from the other nonpolar components by silica gel chromatography. Two fractions were collected: fl, containing primarily the saturated hydrocarbons (including n-alkanes, branched alkanes, and cycloalkanes), and f2, containing polyunsaturated hydrocarbons and PAHs. The results of the total hydrocarbon analyses (the sum of the fl and f2 fractions) are reported elsewhere (13, 14).
The fi fractions containing the PAHs were analyzed by glass capillary gas chromatography on a 15-m Durabond (DB-5) fused silica column (J & W Scientific) using temperature programming from 35 to 270 “C at 4 OC/min employing a Hewlett-Packard 5840 FID gas chromatograph. The PAHs were identified on the basis of their chromatographic retention times and were quantitated by comparison of each peak area to the peak area of a known quantity of internal standard added prior to extraction. The identity of the PAHs studied was confirmed for one sample by GC-MS (Finnigan 4000 equipped with a 30-m DB-5 fused silica column). An example of a gas chromatogram of an urban runoff sample (f2 fraction) is given in Figure 1. During early stages of this work, anthracene was used as an internal standard since it was present only in very low amounts in urban runoff. Later o-terphenyl was used because it was completely absent in runoff samples and has a similar response factor. However, no response factor corrections were used in this study. Blanks contained some unidentified components, but these peaks did not interfere with the determination of the individual PAHs reported in this study. We are reporting results of the analyses of 14 PAHs, which have been divided into two classes for convenience. The lower molecular weight PAHs (LMW) include naphthalene, 2-methylnaphthalene, 1methylnaphthalene, biphenyl, 2-ethylnaphthalene, fluorene, and dibenzothiophene. The higher molecular weight PAHs (HMW) include phenanthrene, fluoranthene, pyrene, benz[a]anthracene, chrysene + triphenylene, benzo[elpyrene, and benzo[a]pyrene. Other PAHs were present also, and not all of the listed PAHs were detected in every sample. The individual PAH concentrations are reported in nanograms per liter or micrograms per liter. When the sum of PAHs is discussed, it refers to the sum of the 14 listed PAHs. The precision of the analyses for these PAHs was between 11and 2570, except for benzo[e]pyrene which was 417O (based on replicate analyses of simulated urban runoff samples made with National Bureau of Standards Urban Dust, SRM 1649) (15). Total f2 is quantified by summing the areas of all the resolved peaks in the fi chromatogram and the area of the unresolved complex mixture. These areas are then compared to the area and mass of the internal standard added prior to extraction. It should be noted that the determinations reported in this study are often operationally defined. For example, the particulate fraction is operationally defined as that which is retained by a Whatman GF/C filter. Experiments Envlron. Scl. Technol., Vol. 18, No. 8, 1984 581
0.4
1st flush:) \
-
J
2000
0
-J
1000
t-
I
0
LL
L particulate
A . . - a 4
0
‘.--__-.soluble
r
*
(e)
0 1200
2100
L O C A L TIME
Flgure 2. Variatlon of urban runoff parameters as a functlon of local tlme for the storm of Oct 25, 1980, at the interstate highway locatlon. (a) Rain intensity and drain flow; (b) suspended solids: (c) particulate and soluble hydrocarbons; (d) partlculate and soluble PAHs; (e) total fluoranthene and phenanthrene.
with a river sediment have revealed that a number of lower molecular weight compounds can be washed off the sediment particles into the soluble phase by the process of filtration. In the more severe cases, this transferral could be up to 50%. Filtration did not significantly affect the higher molecular weight PAH compounds. While this process does not affect the total result, since both the “soluble” and “particulate” phases were analyzed, there is a problem in determining soluble-particulate relationships which are environmentally meaningful for LMW PAHs. These experiments have also shown that some lower molecular weight PAHs can be lost as the filter dries prior to extraction. The losses are about 50% for alkylated naphthalenes. No losses due to drying were observed for naphthalene, biphenyl, or higher molecular weight PAHs, however.
Results and Discussion PAH Concentrations as a Function of Storm History. An example of the type of data generated by the 582
Envlron. Sci. Technol., Vol. 18, No. 8, 1984
storm event sampling is illustrated in Figure 2. This storm, occurring in the afternoon of Oct 25, 1980, was monitored at the Interstate Highway location. Figure 2a shows that the storm had two distinct rain pulses, each followed by rapid increases in flow rate. We will refer to the first peak in flow as the first flush, the shoulder of this peak as the second flush, and the final peak in flow as the third flush. Figure 2b shows the behavior of suspended solids in the runoff. There is a peak in concentration associated with each flush. Rapidly increasing flow rates over urban surfaces suspend the particles and transport them most effectively during these periods of high flow. Figure 2c shows the behavior of total hydrocarbons for each phase as a function of time in the storm. Again there is a peak in concentration of hydrocarbons, especially associated with the particulate phase, during each flush. As illustrated by this figure, most of the hydrocarbons in urban runoff are predominately associated with particles. Figure 2d illustrates how the sum of the 14 quantified PAHs varies with time during the storm. There are two peaks in concentration, one associated with the first flush and one with the second flush. There is no response at all to the third flush. The concentration of particulate PAHs predominates especially during the first part of this storm. The significance of the soluble fraction varies from storm to storm, however. The absence of a PAH response to the third flush, even when the suspended solids and total hydrocarbons did respond, suggests that the PAHs available for incorporation into urban runoff are mobilized easily during a storm; during longer storms, the supply available on the urban surfaces may have been depleted. The possible influence of particle size will be discussed later. Figure 2e illustrates the behavior of two specific PAHs, phenanthrene and fluoranthene, which were found to have the highest concentrations of the PAHs in this series of samples. There is a peak in concentration of phenanthrene associated with the first flush, only a minor increase associated with the second flush, and no response at all to the third flush. For fluoranthene, there was a small peak in response to the first flush, a larger peak in concentration associated with the second flush, and, like phenanthrene, no response to the third flush. While there are similarities in the response of these two PAHs, they do not track each other exactly. Differences among the PAHs as a function of particle size distribution, solubility, volatility, susceptibility to degradation, and other factors influencing the supply may have precluded strictly analogous storm event discharge behavior. A summary of PAH concentrations in urban runoff is given in Appendix I in the supplementary material (see paragraph at end of paper regarding supplementary material). Because both the concentrations and flow rates vary by several orders of magnitude within a storm, only a few random urban runoff samples are of little use in estimating the magnitude of PAH loads via urban runoff. The total amount of any PAH 0’) discharged by a storm event with any number of samples (i) CM be calculated by use of the following equation: storm discharge of PAHj = i
(concentrationj
X
drain flow rate X time interva1)i
is1
An example of this calculation for phenanthrene in the storm event illustrated in Figure 2 (11-26-79, interstate highway, Appendix I) is the following:
phenanthrene storm discharge = (0 pg/L X 98.4 L/s X 3600 s)1 + (0 pg/L X 89.3 L/s X 3600 s)2 (2.21 pg/L X 1423 L/s X 2700 s)S (0.70 pg/L X 2114 L/s X 1800 ~ ) 1+ (0.69 pg/L X 2591 L/s X 1800 s)6 (0.69 pg/LX 2076 L/S x 1800 S)6 (0.25 pg/L x 1806 L/S x 1800 s), (0.49 pg/L X 1837 L/s X 1800 ~ ) + g (0.30 pg/L X 901 L/s X 1800 s)g + (0.22 pg/L X 1003 L/s X 2250 ~ ) 1 0 (0.076 pg/L X 2346 L/s X 3150 ~ ) 1 1 + (0.080 pg/L X 1170 L/s X 3150 ~ ) 1 2+ (0.11 pg/L X 5520 L/s X 2700 s)13= 21.4 g of phenanthrene
+
+
+
+
+
+
Similarly, the amount of water discharged by the drain during a storm event is calculated by using the following equation: volume of water discharged = 5
C (drain flow rate X time interval); i= 1
Because individual concentrations of PAHs in individual samples have limited value by themselves in calculation of discharge rates, we have presented PAH concentration data in terms of flow-weighted means (Appendix I in supplementary material), using the following equation: flow weighted mean j = (total mass of PAHj discharged by drain) /(total water volume discharged by drain) In the case of illustrated storm phenanthrene flow weighted mean = 21.4 g of phenanthrene 39.7
x
IO6 L of water
= 0.54 pg/L
Distribution of PAHs between Soluble and Particulate Phases in Urban Runoff. Although the percentages of HMW PAH concentrations associated with particulate matter are variable from storm to storm, in general, the HMW PAHs and total fi (aromatics) are largely associated with particulate matter, averaging from 79 to 93% (see Appendix I1 in supplementary material). It should be noted, however, that the concentration of the PAHs in urban runoff is much less than the water solubility of these compounds, listed by Mackay et al. (16). Therefore, simple solubility alone does not entirely explain the PAH-particulate associations. For example, the solubility for fluoroanthrene is 260 pg/L (16))but the maximum flow weighted mean for fluoranthene in urban runoff was 8.1 pg/L (see Appendix I in supplementary material). Association of PAHs with particulate matter in urban runoff is undoubtedly a complex function of the nature of the source material (petroleum product and/or combustion material), physical and chemical weathering on the road surface prior to the rain, and interaction of the PAHs with particulates and surfactants in the storm water. Even further interactions are possible within the receiving water body. PAHs as a Function of Particle Size. At least one sample during first flush conditions at each land use was examined to determine the particle size distributions of PAHs in urban runoff. An example of this particle size distribution for the most abundant PAHs at the commercial site is given in Table I. During flush conditions, the urban runoff contains both large and small particles, but in more quiescent conditions, the larger particle population becomes much less important (14). When expressed in pg/gm (Table IB), the three most abundant
PAHs in this sample all had maxima in the 250 pm > d > 125 pm range and, except for phenanthrene, had another maxima in the d < 45 pm size. This pattern was also present in the total f2 concentrations. It has been previously postulated that PAHs in lacustrine and oceanic sediments may have two sources. Prahl and Carpenter (17) found PAH chemical differences among different particle size categories which they attributed to two sources of PAHs to the oceanic sediments off the continental shelf of Washington. On the basis of the methylphenanthrene/ phenanthrene ratio, they postulated that a series of PAHs dominated by phenanthrenes associated with a high-density sand size (>64 pm) sediment fraction were derived from a fossil petroleum source, presumably weathered rock. Combustion-derived PAHs were found enriched in the lower density organic-rich sand size fraction. Wakeham et al. (8) also suggested that PAHs in lacustrine sediment could come from two sources. They found that street dust and sediment samples both contained many tarlike particles resembling asphalt with particle sizes ranging up to 500 pm in diameter. Atmospheric particles, are found in sizes much less than 45 pm (18). Our data show that PAHs are found in two different particle size categories in urban runoff and that the larger particle size (250 pm > d > 125 pm) is enriched in phenanthrene compared to the smaller particle sizes. We therefore suggest that the two sources of PAHs in urban runoff are (1)from asphalt abrasion particles (8)which is the source of PAHs in the larger, heavier particle sizes (and could also be the fossil source of the phenanthrene-rich particles observed in Frahl and Carpenter (17))and (2) from atmospheric fallout which is the source of the smaller particles. More particle size studies would be needed to confirm this possibility. Although the larger particle population becomes much less important during more quiescent conditions, storm water during flushes (periods of high flow) contains a significant population of these larger particles (14). The size distribution of the PAHs also has fortunate implications for treatment of urban runoff by simple methods such as the use of settling basins. Even though a fraction of the PAHs are present on the smaller particles, the PAHs on the larger particles could be removed simply by settling. Removal of these larger particles in urban runoff would result in the large reductions of suspended solids, total hydrocarbons, and PAHs observed in the retention basin study of Latimer (19). PAHs and Aromatics as a Function of Storm Rainfall. The storm loads of the f2 fraction of the hydrocarbons (aromatics) increased with increasing rainfall amounts in a manner similar to that observed for the total hydrocarbons (13). This suggesh that the flow of the storm water over the pavement surfaces continues to incorporate f2 hydrocarbons in runoff throughout the storm and implies an adequate supply of these hydrocarbons available for incorporation in the runoff. However, there was no similar relationship of increasing storm mass loads with increasing rainfall for any of the individual PAHs. The lack of a physically meaningful relationship of PAH loads with rainfall (or runoff since runoff is well correlated to rainfall) suggests that flow dynamic considerations are not as important a factor for PAHs as is the case for total and f2 hydrocarbons. The PAHs, especially the lower molecular weight ones, are also more soluble than the corresponding saturated hydrocarbons containing the same number of carbon atoms. This may also affect their transport as the rainwater passes over the urban surfaces. It is also possible that the storm event PAH loads may be more a function of the available supply than was the case Environ. Scl. Technol., Vol. 18, No. 8, 1984 583
Table I. Size Distribution of PAHs on Urban Runoff Particles (Commercial Site, J u l y 6, 1980, First Flush Sample) d>lmm phenanthrene fluoranthene pyrene f2
total HC
lmm>d> 250pm
250pm>d> 125 pm
125pm>d> 63 pm
10.7 25.3 17.6 12 300 30 800
8.9 17.7 17.1 7 600 10 200
(A) Concentrations, ng/L 8.5 9.1 14.6 17.3 19.7 13.1 8 600 7 600 44 200 171000
0.62 1.47 1.02 715 17 900
2.12 4.21 4.07 1810 24 300
2.58 4.42 5.96 2 606 13400
63pm>d> 45 pm
d
< 45 ,um
total particulate
soluble 26.3 3.4 5.6 2 000 95 400
19.8 32.3 18.4 11300 211 000
7.3 50.4 45.2 21 400 150000
64.3 157.5 131.1 68 800 1040 000
1.92 3.13 1.78 1097 20 500
1.23 8.54 7.66 3 627 25 400
1.35 3.32 2.76 1448 21 900
(B)Concentrations, pg/g phenanthrene fluoranthene pyrene fi
total HC
1.38 2.62 1.98 1151 25 900
Table 11. Loading Factors for PAHs as a Function of Land Use [log
naphthalene 2-methylnaphthalene 1-methylnaphthalene biphenyl 2-ethylnaphthalene fluorene dibenzothiophene phenanthrene fluoranthene pyrene benz[ a]anthracene chrysene benzo[e]pyrene benzo[a]pyrene f2
cm-' km-2)]
residential
commercial
industrial
highway
3.846 f 0.607 4.471 f 1.574 4.234 & 1.640 4.057 f 0.578 2.924 3.234 f 0.608 3.866 f 0.383 4.841 f 0.204 5.921 f 0.300 5.681 f 0.260 5.059 f 0.225 5.322 f 0.498 5.357 & 0.252 5.287 f 0.633 7.945 & 0.449
5.401 f 1.202 5.199 f 0.937 5.076 f 1.115 4.920 f 1.129 4.997 f 1.016 4.825 f 1.025 4.653 f 1.525 5.500 f 1.411 6.159 f 0.305 6.006 f 0.381 5.403 f 0.489 6.068 f 0.159 5.704 f 0.300 5.215 f 0.795 8.667 f 0.268
6.097 6.492 f 0.514 6.563 f 0.504 6.005 f 0.164 6.511 f 0.647 6.397 f 1.229 6.720 f 0.229 6.920 f 0.737 7.019 f 0.437 6.420 f 0.426 6.622 f 0.752 6.354 f 0.712 6.088 f 0.793 6.569 f 0.490 9.699 f 0.359
6.036 f 0.927 6.314 f 0.457 5.909 f 0.149 5.500 f 0.348 5.958 f 0.039 6.518 f 0.041 6.206 f 0.026 7.128 f 0.645 7.625 f 0.457 7.166 f 1.084 7.389 f 0.105 7.554 6.626 f 0.221 6.714 f 0.827 9.739 f 0.158
with total hydrocarbons or fi hydrocarbons. PAHs as a Function of Antecedent Dry Period. If PAHs in runoff were strictly a function of supply and the supply were accumulated between rainfalls, then the PAH storm loads could be more a function of the length of the dry period preceding the monitored storm than the total rainfall. However, when total PAH storm mass loads and the most important individual PAH loads were examined as a function or antecedent dry periods, no consistent meaningful relationships were found. The influence of antecedent dry periods on pollutant mass loads in urban runoff is controversial. For suspended solids, some studies have found increasing loadings with increasing antecedent dry periods (20,21),but other studies find no relationship with antecedent conditions (22, 23). In the case of hydrocarbons and PAHs, all of the previous investigators could find no effect of antecedent dry periods (9,14,24). The lack of a simple relationship may indicate that the PAH supply available a t any time may be a complex function of meteorological variables. For example, windy conditions prior to a storm could remobilize smaller particles containing PAHs from the surfaces into the atmosphere, particles which might otherwise become a part of urban runoff. Or insolation prior to rainfall may cause losses by volatilization, degradation, and/or photooxidation of the PAHs. PAHs in Urban Runoff as a Function of Land Use. To compare mass loadings at the different land uses, each mass loading was normalized to the storm rainfall and the drainage area to yield PAH, urban runoff loading factors in unita of fig of PAH (cm of rain)-l km-2 of land drainage area. Since these factors could vary for some PAHs by an order of magnitude from storm to storm within the same land use, and because of the trace quantities involved, log normal statistics were used as suggested by Belli et al. (25). 584
(pg
Environ. Scl. Technoi., Vol. 18, No. 8, 1984
These loading factors are given in Table 11. We used Student's t test to determine if these factors are different at the various land uses. Briefly, the PAH loading factors at the industrial and highway sites were usually higher (P I 0.05) than the residential loading factors (12 out of 14 PAHs). About half of the PAHs at the industrial and highway sites were higher than at commercial sites. However, most PAH loading factors at the residential location were not significantly different than commercial loading factors, and most industrial and highway PAH loadings were also not different from each other. Therefore, for the individual PAHs studied, generally the urban runoff loading factors are residential = commercial < industrial = highway. This assessment is also true of the total f2 hydrocarbon loading factors as well. The observation that industrial and highway locations contribute more PAHs to urban runoff than other land uses is consistent with the proximity of these two locations to points of both automotive and industrial combustion emissions of PAHs to the atmosphere (26,27). These two locations also contribute the highest loadings of total hydrocarbons in urban runoff as well (13),largely from crankcase oil (13, 28) which has also been shown to contain PAHs (3,10,28). Although the quantitative inputs of PAHs show differences among the land use types, the chemical distributions of the PAHs are not statistically different at the different land uses when normalized to the total (see Table 111). This would suggest that the conditions which produce PAHs or the sources of PAHs to urban runoff in all land uses are the same or that the PAH distributions of the various sources are similar. Therefore, for PAHs, land use differences can influence the magnitude of the PAH discharge rate but do not appear, on the average, to statistically affect the chemical distributions relative to each other.
Table 111. Chemical Distribution of PAHs in Urban Runoff from Four Different Land Uses (Mean Percent of Sum of 14 Quantified PAHs) residential naphthalene 2-methylnaphthalene 1-methylnaphthalene biphenyl 2-ethylnaphthalene fluorene dibenzothiophene phenanthrene fluoranthene pyrene benz[a]anthracene chrysene benzo [e]pyrene benzolal pyrene
0.2 f 0.2 3.3 f 5.4 2.1 f 3.5 0.4 f 0.4 0.02 f 0.03 0.2 f 0.3 0.2 f 0.2 3.7 f 3.0 33.6 f 4.0 19.8 f 5.6 4.9 f 1.9 7.7 f 8.1 13.3 f 12.8 10.9 f 11.2
commercial
industrial
highway
5.6 f 9.9 3.0 f 3.7 3.4 f 4.7 1.7 f 2.9 1.4 f 2.4 1.3 f 1.7 3.0 f 5.2 12.4 f 7.0 20.4 f 7.9 16.5 f 10.3 4.8 f 3.6 11.2 f 14.9 5.8 f 4.6 8.9 f 11.4
1.8 f 2.3 1.3 f 1.2 1.1 f 1.5 0.2 f 0.3 1.3 f 1.6 7.8 f 11.5 3.3 f 2.1 10.6 f 6.3 21.5 f 6.3 19.2 f 12.3 22.9 f 22.2 1.6 f 2.9 3.2 f 4.6 5.6 f 5.0
0.7 f 0.7 1.2 f 1.2 0.6 f 0.8 0.5 f 0.8 0.8 f 1.1 2.9 f 4.2 1.5 f 2.3 15.8 f 15.6 33.6 f 2.2 19.3 f 15.5 8.7 f 10.3 3.5 f 6.0 3.8 f 2.4 3.0 f 2.9
Table IV. Annual Inputs (kg/year) of PAHs via Urban Runoff to the Upper Narragansett Bay Watershed
napkthalene 2-methylnaphthalene 1-methylnaphthalene biphenyl 2-ethylnaphthalene fluorene dibenzothiophene phenanthrene fluoranthene pyrene benz[a]anthracene chrysene benzo[e]pyrene benzo[a]pyrene sum PAHs f2
total hydrocarbons sum PAHs/ total hydrocarbons, % fi hydrocarbons/ total hydrocarbons, %
residential
commercial
industrial
highway
total
0.18 7.5 0.35 0.28 0.02 0.04 0.18 1.7 20.8 12.0 2.9 5.2 5.7 4.8 55.1 2 200 36 250
1.7 1.0 0.79 0.55 0.66 0.44 0.30 2.1 9.4 6.7 1.7 7.7 3.3 1.1 37.4 3 069 31 880
4.8 12.1 14.3 3.9 12.6 9.7 20.4 32.4 40.7 32.4 16.3 8.8 4.8 14.4 22.8 19500 444 700
2.6 4.9 1.9 0.76 2.2 7.9 3.9 32.2 101.2 35.1 58.7 86.0 10.1 12.4 360.0 13200 152 700
9.3 25.5 17.3 5.5 15.4 18.1 24.8 68.4 172.1 86.2 79.6 107.7 23.9 32.7 681.0 38 000 665 300
0.15
0.12
0.05
0.23
0.10
6.1
9.6
4.3
8.6
5.7
Annual Inputs of PAHs into the Upper Narragansett Bay Watershed. The annual input rate of PAHs into a typical urban watershed via urban runoff was calculated by multiplying the loading factors (logs are given in Table 11) by the average annual precipitation (121 cm) and the respective land use areas (13) in the watershed. The annual input rate for the PAHs in the upper Narragansett Bay watershed (population of 686 000) is given in Table IV. For comparative purposes, the mass inputs of f~ hydrocarbons and total hydrocarbons (13) are also reported in this table. The total PAH discharge was approximately 680 kg/year, 2.2 kg/km2 of land drainage arealyear, or 1.0 g capita-l year-l. fi (aromatic) hydrocarbons in urban runoff to this watershed were 38 metric tons/year, 112 kg/km2 of land drainage area/year, or 51 g capita-' year-'. Eganhouse et al. (11) studied the discharge of the Los Angeles River basin during one rainstorm. They projected their results to estimate an annual discharge rate of fi (aromatic) hydrocarbons of 100 g capita-l yeail, approximately twice the urban runoff discharge rate determined in this study. The discharge during this storm event in Los Angeles had a f2/total hydrocarbon ratio of 0.12, whereas the ratio in our study averaged 0.057 (see Table IV). This ratio is, however, variable from storm to storm with a relative standard deviation of 44% (six storms, commercial site). Therefore, we cannot determine at this time if geographical differences are involved, either in comparison between land uses or in comparison of different metropolitan areas.
Comparison of PAH Inputs into Narragansett Bay via Urban Runoff with Other PAH Sources. A comparison of the PAH urban runoff inputs with those of other PAH sources to the upper Narragansett Bay drainage basin is given in Table V. The PAH distribution in urban runoff is similar to that of atmospheric fallout. The similarity of chemical distributions alone would suggest that atmospheric fallout on land surfaces is a source of PAHs in urban runoff. If the PAH atmospheric fallout rates as a function of land use reported by Latimer (28) are used for the upper Bay drainage basin, atmospheric fallout could account for half of the PAHs in urban runoff. Urban runoff PAH distributions do not resemble the PAHs in sewage treatment effluents which were analyzed by using exactly the same procedures (29). PAHs in both urban runoff and atmospheric particulates are subjected to weathering resulting in losses of low molecular weight PAHs via degradation and evaporation. Sewage effluents containing petroleum hydrocarbons largely from industrial discharges (29) are not as weathered and still contain many naphthalenes when discharged. The rates of entry of PAHs into upper Narragansett Bay for a number of sources are estimated in Table VI. In terms of the 14 quantified PAHs, urban runoff represents 36% of the total. The contributions of urban runoff to the total lower molecular weight PAH inputs to the estuary are minor, representing only 16% of the total. However, urban runoff is the major source of higher molecular weight PAHs to the estuary. Seventy-one percent of the HMW PAHs come Environ. Sci. Technol., Vol. 18, No. 8, 1984
585
__ ___ -_____ - _ _ -___ Table V. Comparison of PAHs in Urban Runoff with Other Sources of PAHs to t h e Marine Environment (Mean Percent of Sum of 14 Quantified PAHs) ___ _______I_ _I
urban runoff (this study)
-.
naphthalene 2-methylnaphthalene I-methylnaphthalene biphenyl 2-ethglnaphthalene fluorene dibenzothiophene phenanthrene fluoranthene pyrene benz[a]anthrarene chrysene benzo[e]pyrene benzo[a]pyrene - . -_I_I__. __-I
__._ .
atmospheric fallout (28)
sewage treatment (29) dry conditions
0.7 1.2 0.6 0.5 0.8 2.9 1.5 15.7 33.5 19.3 8.7 3.4 3.8 3.0
32.0 21.8 17.9 1.7 11.2 2.4 0.6 4.7 3.0 2.9 0.2 0.7 0.4 - -0.2 .
1.3 3.7 2.5 0.8 2.3 2.7 3.6 10.0 25.2 12.6 11.6 15.8 3.5 4.8._.._I
--_ - _ _ _ - . -______ Table VI. Estimated Annual Input Rate of PAHs to the upper Narragansett Bay Drainage Basin (kg/year)
-_
direct atmos sewage urban runoff fallout to treatment (this study) water surfaceo effluent8 (29) total total PAHs LMW PAHs HMW PAHs
681 184 502
81 7 74
1120 990 130
1880 1180 706
“Using Latimer (28) rates. These values do not include atmospheric fallout on land surfaces because it is incorporated in urban runoff. - __ - - __ - - ---_---
from urban runoff, 10% from direct atmospheric fallout, and 18% from sewage treatment effluents. We estimate the total input rate of higher molecular weight PAH compounds to upper Narragansett Bay to be approximately 700 kg/year. The PAH sediment accumulation rate for Narragansett Bay has not been determined. However, if the accumulation rate for outer Boston Harbor (31)is used for an area the size of Narragansett Bay, the removal rate for higher molecular weight PAHs via sedimentation would approximate 400 kg/year. The similarity in chemical distribution of the higher molecular weight PAHs between urban runoff and Narragansett Bay sediments as reported by Pruell (30) and the similarity in estimated inputs and sedimentary accumulation rates give circumstantial evidence that urban runoff PAHs strongly influence the composition of coastal marine sediments especially close to citiefi. Comparison of urban runoff inputs to other proposed sources of PAHs to the marine environment confirms that urban runoff is the predominant source of higher molecular weight PAHs in Narragansett Ray but is a less important source of lower molecular weight PAHs.
Conclusions Urban runoff is a source of PAHs in rivers and estuarine waters. The highest PAH conrentrations occur in the first or second storm flushes; the highest PAH loadings were found at industrial and highway land uses. There was no direct relationship of PAH urban runoff discharges with either rainfall amounts or the length of the dry periods preceding the storms. The higher molecular weight PAHs are associated with the suspended solids in the runoff and some are enriched on smaller particle sizes. The annual rate of total PAH and total aromatic hydrocarbon urban runoff inputs to waters in the upper Narragansett Ray watershed is approximately 1.0 and 51 g capita-’ year-’, 588
Environ. Sei. Technot., Vot. 18, No 8 , 1984
---
-
Providence, RI, sediment (30) 0.5 0.6 0.3 0.4 1.2 0.2
-_
I
7.1 17.9 18.3 7.4 13.2 14.8 17.3 -. - _____ .
- .-
-
respectively. Urban runoff PAH chemical distributions are similar to that of atmospheric fallout both of which contain few lower molecular weight PAH compounds. Urban runoff contributes approximately 71 % of the higher molecular weight PAHs to the estuary and 36% of the total PAHs.
Acknowledgments In addition to those previously acknowledged (13) for their aid in sampling and land use data collection, we thank Richard Pruell, Graduate School of Oceanography, University of Rhode Island, for his analytical advice and review of the manuscript and Curt Norwood of the US.Environmental Protection Agency’s Environmental Research Laboratory in Narragansett, RI, for GC/MS analyses.
Supplementary Material Available Storm flow weighted means of PAH concentrations in urban runoff (Appendix I) and percentage of PAHs associated with particulate matter (Appendix 11) (3 pages) will appear following these pages in the microfilm edition of this volume of the journal. Photocopies of the supplementary material from this paper or microfiche (105 X 148 mm, 24X reduction, negatives) may be obtained from Microforms Office, American Chemical Society, 1155 16th St., N.W., Washington, DC 20036. Full bibliographic citation (journal, title of article, author, page number) and prepayment, check or money order of $6.00 for photocopy ($8.00 foreign) or $6.00 for microfiche ($7.00 foreign), are required. Registry No. Naphthalene, 91-20-3; %methylnaphthalene, 91-57-6; 1-methylnaphthalene, 90-12-0; biphenyl, 92-52-4; 2ethylnaphthalene, 939-27-5;fluorene, 86-73-7;dibenzothiophene, 132.65-0; phenanthrene, 85-01-8; fluoranthene, 206-44-0;pyrene, 129-00-0; benz[a]anthracene, 56-55-3; chrysene, 218-01-9; benzo[e]pyrene, 192-97-2; benzo[a]pyrene, 50-32-8.
Literature Cited (1) Yonngblood, W. W.; Blumer, M. Geochim. Cosmochim. Acta 1975, 39, 1303-1314. (2) Laflamme, R. E.; Hites, R. A. Geochim. Cosmochim. Acta 1978, 42, 289-303. (3) L,ake, J. L.; Norwood, C.; Dimock, C.; Bowen, R. Geochim. Cosmochim. Acta 1979, 43, 1847-1854. (4) Hites, R. A.; Laflamme, R. E.; Windsor, J. G., Jr. Adu. Chem. Ser. 1980, No. 185, 289-311. ( 5 ) Hites, R. A.; Laflamme, R. E.; Windsor, J. G., dr.; Farrington, J. W.; Dueser, W. G. Geochim. Cosmochim. Acta 1980,44,873-878. (6) Lee, M. L.; Prado, G. P.; Howard, J. B.; Hites, R. A. Riomed. Mass Spectrom. 1977,4, 182-186. (7) Windsor, J. G., Jr.; Hites, R. A. Geochim. Cosmochim. Acto 1979, 43, 27-33.
Environ. Sci. Technol. 1984, 18,587-591
(8) Wakeham, S. G.; Schaffner, C.; Giger, W. Geochim. Cosmochim. Acta 1980, 44, 403-413. (9) Epanhouse, R. P.; Kaplan, 1.R. Environ. Sci. Technol. 1981, 1.5, 310-315. (10) Acheson, M. A,; et al. Water Res. 1976, 10, 207-212. (11) MacKenzie, M. J.; Hunter, J. V. Environ. Sci. Technol. 1979, 13, 179-183. (12) Herrmann, R. Water, Air, Soil Pollut. 1981, 16, 445-467. (13) Hoffman, E. J.; Mills, G. L.; Latimer, J. S.; Quinn, J . G. Can. J. Fish. Aquat. Sci. 1983, 40 (Suppl. 2), 41-53. (14) Hoffman, E. J.; Latimer, J. S.; Mills, G. L.; Quinn, J . G. J. Water Pollut. Control Fed. 1982,54, 1517-1525. (15) Hoffman, E. J.; T,atimer, J. S.; Mills, G. L.;Quinn, J. G. Oct 1982, NOAA Quality Assurance Report on Grant NA80RAD00047, pp 1-18. (16) Mackay, D.;Babra, A.; Shiu, W. Y. C'hemosphere 1980,9, 701-711. (17) Prahl, F. G.; Carpenter, R. Geochim. Cosmochim.Acta 1983, 47,1013-1023. (18) Pierce, R. C.; Katz, M. Environ. Sci. Technol. 1975, 9, 347--353. (19) Latimer, J. S., Graduate School of Oceanography, University of Rhode Island, personal communication. (20) Marsalek, J. Proc. Am. SOC.Civ. Eng. 1976, 564.
(21) Litwin, Y. J.;Anthony, S. D. J. Water Pollut. Control Fed. 1978, 50, 2348. (22) Bedient, P. B.; Lambert, J. L.; Springer, N. K. J . Water Pollut. Control Fed. 1980,52, 2396. (23) Whipple, W.; Hunter, J. V.; Yu, S . L. J . Water Pollut. Control Fed. 1977, 49, 2243. (24) Hunter, J. V.; Sabatino, T.; Gamperts, R.; MacKenzie, M. J. J. Water Pollut. Control Fed. 1979, 5 1 , 2129. (25) Belli, G.; et al. Chemosphere 1983, 12, 517-521. (26) Gordon, R. J. Enuiron. Sci. Technol. 1976, 10, 370-373. (27) Katz, M.; Sakuma, T.; Ho, A. Environ. Sci. Technol. 1978, 12, 909-915. (28) Latimer, J. S. MS Thesis, University of Rhode Island, 1984. (29) Hoffman, E. J., unpublished data, 1983. (30) Pruell, R. J., Graduate School of Oceanography, University of Rhode Island, personal communication, 1983. (31) Gschwend, P. M.; Hites, R. A. Geochim. Cosmochim. Acta 1981, 45, 2359-2367.
Received for review May 16, 1983. Revised manuscript received January 12, 1984. Accepted February 23,1984. This work was funded by the U.S. National Oceanic and Atmospheric Administration through the Office of Marine Pollution Assessment (Grant NA80RAD00047).
Solubility of Organic Mixtures in Water Sujlt Banerjee* Life and Environmental Sciences Division, Syracuse Research Corporation, Syracuse, New York 13210
-
.
- --
--
-
--
The solubilities of several chlorobenzenes and other mixtures in water have been determined. The results varied with the phase of the solute mixture and the hydrophilicity of the components and were interpreted through activity coefficients calculated by the UNIFAC equation. It was found that mixtures of structurally related hydrophobic liquids were near ideal in the organic phase; in the aqueous phase the activity coefficient of a component was unaffected by the presence of cosolutes. Increasing hydrophilicity of the solutes led to deviations from ideality in the organic phase, but these could be largely accounted for by the UNIFAC equation. For mixtures of solids which did not interact, the components tended to behave independently of one another, and their solubilities were approximately additive. The behavior of mixtures of liquids and solids was intermediate between that of liquid mixtures and that of mixtures of solids. The application of these results to the toxicity of organic mixtures in water is discussed. -- - - .-- -- --_ I
-_I_----
The growing number of organic compounds under present or potential regulation has prompted extensive work on structure-activity relationships and on environmental models (1, 2). Present techniques for assessing environmental effects from laboratory studies typically rely on data such as solubility and the octanol-water partition coefficient for the calculation of bioconcentration factors ( 3 , 4 ) ,sediment adsorption coefficients (ij),toxicity (6),and biodegradation rates (7,8). Environmental contaminants, however, are frequently encountered as mixtures, and the behavior of a compound in a mixture may not correspond to that predicted from pure component data. The work of Sutton (9)and others (10, 11) on the solubility of hydrocarbons has shown that the interaction of components __
--__
--
-
_.-_
_"I___
*Address correspondence to this author at the Safety and Environmental Protection Division, Rrookhaven National Laboratory, IJpton, NY 11973. 0013-936X/84/0918-0587$01.50/0
in a mixture can cause complex and substantial changes in the solubilities of its constituents. We have measured solubilities in water of mixtures of liquid components, of solid components, and of both liquid and solid components, and in this paper we present our results and identify behavior typical of each category. In addition we discuss the potential application of our conclusions to the toxicity of organic mixtures dissolved in water. Experimental Section
For the solubilitystudies the substrates (50 mg-2 g) were separately weighed into 25-mL glass tubes to each of which 5-15 mL of distilled water was then added. The tubes were shaken in a water bath maintained at 25 f 0.05 OC for at least 48 h, after which the shaking was stopped and the phases were allowed to separate over 24 h. In some of the mixtures containing solid components, partial or complete liquefaction of the solids occurred at equilibrium. For example, in the mixture containing 1,2,3-trichlorobenzene, 1,2,3,4-tetrachlorobenzene, and water, the solids liquefied to various degrees depending upon the composition of the mixture, as shown in Table VIII. Following equilibration, the water in each tube was sampled at least in duplicate, and the solution was diluted with an equal volume of acetonitrile to prevent any deposition of material and then analyzed by high-pressure liquid chromatography (HPLC). A Waters Associate M6000A pump fitted with either a Lichrosorb RP-2 or an Altex Ultrasphere ODS column and a LDC Spectromonitor I11 detector was used, and the mobile phase consisted of various mixtures of acetonitrile and water. Five replicate determinations of the solubility of chlorobenzene yielded a standard error of 5.470,and a similar degree of uncertainty was associated with the other solubility measurements. Pure component solubilities of all the compounds measured in this study are listed in Table I. Activity coefficients were calculated from composition by the UNIFAC method introduced by Prausnitz and
0 1984 American Chemical Society
Environ. Sci. Technol., Vol. 18, No. 8, 1984 587