Environ. Sci. Technol. 2009, 43, 494–502
Removal and Fate of Polycyclic Aromatic Hydrocarbon Pollutants in an Urban Stormwater Bioretention Facility CATHERINE J. DIBLASI,† HOUNG LI,‡ A L L E N P . D A V I S , ‡ A N D U P A L G H O S H * ,† Department of Civil and Environmental Engineering, University of Maryland, Baltimore County, Baltimore, Maryland 21250, and Department of Civil and Environmental Engineering, University of Maryland, College Park, Maryland 20742
Received July 26, 2008. Revised manuscript received October 22, 2008. Accepted October 23, 2008.
This research investigated the removal and fate of 16 USEPA priority pollutant polycyclic aromatic hydrocarbons (PAHs) from urban stormwater runoff through a bioretention cell. Bioretention is an infiltration/filtration practice containing a mixed layer of about 90 cm of soil, sand, and organic matter, planted with appropriate vegetation. Field water quality monitoring and bioretention media core analyses were performed. The results indicate that bioretention is a promising management practice to control runoff PAH pollutants. The PAH event mean concentration (EMC) reduction ranged from 31 to 99%, with a mean discharge EMC of 0.22 µg/L. The mass load decreased from a mean value of 0.0180 kg/ha yr to 0.0025 kg/ha yr, suggesting an average PAH mass load reduction of 87% to the discharging watershed. The most dominant PAH species monitored were fluoranthene and pyrene. Influent PAHs indicated strong affiliation with runoff total suspended solids (TSS). As such, PAH removal positively correlated with TSS removal. Low rainfall depth was associated with high influent PAH concentration and resulted in favorable PAH removal. Source investigation suggested that the PAHs measured in the monitored cell were from pyrogenic sources, likely resulting from vehicle combustion processes. Sealers used in parking lots and driveway coatings were also a possible source of PAHs. Media core analyses indicated that the intercepted PAH compounds transported only a few centimeters vertically in the soil media near the runoff entrance location, suggesting that a shallow cell design may be adequate for systems focusing on PAH removal.
Introduction Polycyclic aromatic hydrocarbons (PAHs) represent the largest class of suspected carcinogens (1) and they are widely distributed in the air, water, and sediments of urban environments (2-5). More than 100 different PAH compounds exist, and although many are naturally occurring, the majority of PAHs in the environment are from anthropogenic sources (6). Anthropogenic sources of PAHs include the release of petroleum products (petrogenic) or combustion * Corresponding author phone: 410-455-8665; e-mail: ughosh@ umbc.edu. † University of Maryland, Baltimore County. ‡ University of Maryland. 494
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 2, 2009
of organic matter such as petroleum, coal, wood, and oil (pyrogenic) (5). Pyrogenic sources of PAHs have been noted to dominate urban environments (5, 7, 8). Specifically, urban sources may include vehicle exhaust, home heating through coal and wood burning stoves, trash burning, power plants and other industrial processes, and the leaching of PAHs in sealants used to coat parking lots and driveways (9). Because PAHs are ubiquitous in urban environments, they are also a very accessible contaminant to be mobilized and transported by stormwater runoff. Urban stormwater runoff is considered an important source of many pollutants, including PAHs, to aquatic environments. Studies have found that patterns of PAHs observed in near-shore sediments closely reflect the PAH patterns found in urban stormwater runoff. It is estimated that urban runoff contributes about 14-36% of the total PAH load to aquatic ecosystems (8, 10) However, the management of PAHs in urban stormwater has received little attention. In the Chesapeake Bay and in many other coastal and estuarine environments, eutrophication resulting from excess nutrient loading is the primary cause of poor water quality and aquatic habitat loss. Therefore, reducing nutrient and sediment loads in stormwater is the primary concern for many municipalities. Reduction of toxics is often a lower priority, partially because of the lack of information on nonpoint toxics loads. Although past research on the effectiveness of stormwater best management practices (BMP) has not focused specifically on PAHs, practices that remove sediment from stormwater may also be capable of removing PAHs, since PAH compounds generally have low solubilities (see Table S1, Supporting Information) and are expected to be associated with the suspended solids in stormwater runoff. Bioretention is an increasingly popular stormwater BMP. Bioretention is typically designed as individual cells (sized at approximately 2-5% of the drainage area), which include a mixed layer of about 90 cm of soil, sand, and organic matter, planted with appropriate vegetation, engineered to store, infiltrate, and treat stormwater runoff (11-14). Bioretention studies both in the laboratory and in the field have shown high-efficiency removal of suspended solids (15-18). Dissolved PAHs are expected to be strongly retained in the organic matter stipulated in bioretention treatment media design (19). The primary objective of this study was to determine the effectiveness of a field bioretention facility in removing PAHs from urban stormwater runoff. Flow-weighted influent and effluent samples to the bioretention cell were examined for the 16 EPA priority pollutant PAHs. A second objective of this study was to evaluate possible PAH sources and affiliations with particles. A final objective was to observe the distribution of PAHs in the vertical profile of the bioretention cell media. This information is important in examining long-term performance and operational characteristics of bioretention media.
Materials and Methods Site Description. The monitored bioretention cell is located on the University of Maryland campus (College Park, MD). The drainage area consists of approximately 90% impervious surface, including asphalt parking lots and roads,and concrete sidewalks for commuter students and athletic event visitors. This facility was retrofitted into the existing storm drain network in spring 2004. The bioretention cell is nearly linear, but trapezoidal in shape (length ) 50.3 m, width ) 2.4-4.8 m, and cell area ) 181 m2, Figure 1). It serves a design drainage 10.1021/es802090g CCC: $40.75
2009 American Chemical Society
Published on Web 12/12/2008
FIGURE 1. The monitored bioretention cell, adjacent to a parking lot on University of Maryland, College Park campus, and sampling locations. area of approximately 0.28 ha, producing a cell surface area: drainage area ratio ) 6%. The cell has a sloped surface away from the inlet, with an average ponding storage depth of 15 cm. Two 15 cm perforated PVC pipes run the length of the bioretention cell 0.5-0.8 m below the media, collecting and conveying infiltrated water to nearby Campus Creek. The cell was not lined. Stormwater Sample Collection and Preparation. The monitored cell intercepts incoming runoff through a 20 cm Tracom Cutthroat flume for influent rate measurement and water sampling. The cell underdrain directs infiltrated water to a discharge manhole with a 20 cm PVC pipe, which is outfitted with a 20 cm Thel-Mar plug-in weir for effluent rate measurement and water quality sampling. Two ISCO 6712FR refrigerated autosamplers were assigned to the influent and effluent. Each autosampler was equipped with a bubble flow meter (ISCO 730) positioned at the flume/weir. One factorycalibrated ISCO 674 tipping bucket rain gauge with 0.0254
cm sensitivity was connected with the influent autosampler to record rainfall. Five rainfall events were monitored from April 2006 to July 2007. The rainfall depth for these five events ranged from 0.28 to 0.83 cm. Flow-weighted composite sampling was employed for both input and output measurement. The stormwater samples were stored in large glass jars with aluminum foil-lined caps on ice in a cooler and transported to the laboratory within 24 h of collection. Sodium azide (100 mg/L) was added to the stormwater samples to prevent biodegradation of PAHs before they were analyzed. The flowweighted composite samples were directly analyzed for PAH levels to obtain the pollutant event mean concentration (EMC) of that event. All samples were analyzed for PAHs within five days of collection. Flow rates were monitored and PAH masses were determined as the product of total runoff volume and the respective EMC. Total suspended VOL. 43, NO. 2, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
495
FIGURE 2. Dissolved (top) and particulate (bottom) PAH concentration in stormwater influent and effluent samples of the monitored bioretention cell. Reference line represents laboratory detection blank (0.08 µg/L). Concentrations shown are the averages of the duplicate samples, and error bars represent (1 standard deviation where available. solids (TSS) concentrations were measured in an USEPAcertified laboratory. Duplicate samples (900 mL) of each influent and effluent stormwater were used for PAH analysis. Particle-associated PAHs were separated using alum flocculation using 9.0 mL of 1.0 M alum and 4.5 mL of 1.0 N NaOH based on the method of Ghosh et al. (20). After adding the alum and NaOH, the samples were gently mixed on a shaker for about 3 min and then placed in a refrigerator to settle for approximately 16 h. PAH Extraction, Cleanup, and Analysis. After settling, the overlying liquid was transferred to a 1 L separatory funnel for liquid-liquid extraction with hexane based on USEPA method SW-846 3510C (21). The settled solids were filtered through a folded Whatman GF-C 125 mm diameter filter paper (1.2 µm particle retention). The filter paper with filtered solids was extracted by sonication using 1:1 hexane:acetone based on USEPA method SW-846 3550B (21). Solvent extracts were concentrated using a Rotovap (Buchi model R-200) 496
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 2, 2009
followed by nitrogen evaporation (Organomation Inc., Model N-EVAP 111). The concentrated extracts were cleaned using activated silica gel based on USEPA method SW-846 3630C (21) and further concentrated to 1 mL. All samples were analyzed using an Agilent 6890 gas chromatograph with a fused silica capillary column (HP-5, 60 m × 0.25 mm I.D.) and a 5973N mass spectrometer detector operated in the single ion monitoring mode, based on EPA method SW-846 8270 for PAHs (21). Calibration standards for the 16 USEPA priority PAH pollutants (Table S1, Supporting Information) and internal standards (1-fluoronaphthalene and p-terphenyl-d14) were obtained from Ultra Scientific. Phenanthrened10 (Ultra Scientific) was added to all samples as a surrogate standard. Samples with surrogate recoveries less than 70% were not included in the final data analysis. Laboratory blanks were analyzed along with each set of stormwater samples by performing liquid-liquid extraction of approximately 900 mL of deionized water. The instrument detection limits (IDL)
TABLE 1. Total PAH Removal Efficiency (in terms of concentration reduction and mass removal), TSS Concentrations, Rainfall Depth, and Runoff Volume for the Five Monitored Storm Events
influent total PAH (µg/L) effluent total PAH (µg/L) % reduction in PAH concn influent TSS (mg/L) effluent TSS (mg/L) % reduction in TSS rainfall depth (cm) influent volume (L) effluent volume (L) total PAH mass in (mg) total PAH mass out (mg) annual PAH mass load in (kg/ha yr) annual PAH mass load out (kg/ha yr) % reduction in annual PAH mass load a
4/21/2006
6/19/2006
7/4/2006
9/14/2006
7/10/2007
mean
5.16 0.057 99 30 1 97 0.28 4440 3090 22.9 0.2 0.0446 0.0004 99
0.48 0.33 31 NAa NA NA 0.62 10500 16500 5.0 5.4 0.0041 0.0045 -8
0.29 0.19 34 16 10 38 0.83 11500 30000 3.3 5.7 0.0025 0.0043 -71
0.98 0.37 62 16 3 81 0.43 17300 14600 17.0 5.4 0.0085 0.0027 68
3.5 0.14 96 68 37 46 0.57 13300 8120 46.6 1.1 0.0302 0.0007 98
2.08 0.22 90 33 13 61 0.55 11408 14462 23.8 3.2 0.0180 0.0025 87
NA: not available.
based on signal-to-noise ratio were typically one-third of the lowest calibration level reported in Table S1 (Supporting Information). PAH concentrations falling between the IDL and the lowest calibration level are reported and used for data interpretation. The laboratory DI blank (0.08 µg/L) was determined by taking the mean of six deionized water blanks. Most PAH compounds in the DI water blanks were below the IDL, except for naphthalene, phenanthrene, anthracene, pyrene, and chrysene, which often showed up in the blank samples. Media Core Collection and Analyses. A JMC environmentalist’s subsoil probe (2 cm inner diameter, 71 cm length) was used in sampling the bioretention media. Two media cores were taken from the monitored bioretention cell on June 12, 2007. One core was taken specifically close to the inlet discharge to capture any surface deposit (influent core), as shown in Figure 1. It was also expected that this location would have the highest PAH accumulation, since it receives the greatest influent runoff load. For comparison purposes, the second core was chosen at approximately the midpoint of the cell (midpoint core). Both core samples were taken from the surface to 40-50 cm deep. The JMC corer was lined with precleaned PETG (glycol-modified polyethylene terephthalate) liners, which encased the core sample. Vinyl caps were used to close the ends of the liners after sampling. The influent soil core was approximately 51 cm long and split into five 10.2 cm segments. The midpoint core was 41 cm long and split into four 10.2 cm segments. Each segment was separated with a precleaned saw and homogenized in the field. From an obvious accumulation of deposited particulate matter near the runoff entrance, two additional samples were collected from the surface of the bioretention cell. Several grams of the top crust material (upper 1 to 2 mm) were scraped off the surface and collected along with a sample of nearby loose gravel material. This material is representative of captured runoff particles, a mixture of naturally occurring sediments and anthropogenic materials such as asphalt attrition particles, soot, worn tire treads/brake pads, and atmospheric deposition solids. All media samples were placed in clean glass jars, transported to the laboratory within 1 h, and stored at 4 °C. All media samples were analyzed for the 16 USEPA priority pollutant PAHs using methods described earlier. Moisture content was measured by placing 5 g of each sample in an oven at 110 °C for 16 h. Duplicate samples were used and the moisture content values were calculated as mass percentages of water in the media (wet basis).
Results and Discussion Stormwater Runoff PAH and Source Investigation. Total (dissolved + particulate) influent PAH EMCs ranged from
0.29 to 5.16 µg/L (mean of 2.08 µg/L), as shown in Table 1. Significant storm event variability in pollutant EMCs is expected (22) due to variable pollutant deposition and hydrologic parameters. Total runoff PAH concentrations measured in this study are similar to those of previous urban runoff studies (23). Menzie et al. (8) analyzed runoff samples in eastern Massachusetts for 39 PAH compounds and reported total PAH values of 1.33 µg/L for nonurban sites and 12.1 µg/L for urban sites. Stein et al. (5) measured 26 PAH compounds in urban stormwater samples in the greater Los Angeles (CA) region and the detected EMC ranged from 0.14 to 5.82 µg/L. Dissolved and particulate PAH distribution patterns observed in this study are also similar to previous studies. Hoffman (7) and Menzie et al. (8) both found that fluoranthene, phenanthrene, pyrene, and chrysene were the dominant PAH compounds in their stormwater samples. These four compounds contribute 60-79% (mean 69%) of the dissolved PAH concentrations in the runoff influent. These four compounds also contribute 50-70% (mean 58%) of the particulate PAH concentration in the influent stormwater samples. The contribution of different PAH sources can be identified by examining the ratios of the low molecular weight (LMW, two or three rings) and high molecular weight (HMW, four to six rings) compounds in the influent stormwater samples. HMW compounds contributed an average 72% of the total PAH (particulate + dissolved) concentration in influent runoff samples, indicating that the detected PAHs were the result of combustion processes (8). The most likely source of combustion-derived PAHs in roadway and parking areas is vehicle exhaust. It is expected that a higher percentage of the LMW compounds exist in the dissolved form because of higher solubility compared to the HMW. Among the dissolved PAH samples of all five storms, on average 47% of the PAHs were LMW, while the LMW makeup was only 23% for the particulate phase samples. Another method for apportioning the general (pyrogenic vis-a`-vis petrogenic) sources of PAHs is characterization of the ratios of individual compounds, specifically fluoranthene/ pyrene (F/P) and phenanthrene/anthracene (P/A). Literature criteria for these ratios vary since previous studies were subject to different source conditions; however, pyrogenic sources generally dominate when the F/P value is close to 1 and the P/A value is generally between 3 and 26 (5, 8). It should be noted that these ratio methods were initially established for sediment samples but have been applied to stormwater runoff studies as well. Among the five influent runoff samples for total PAHs (liquid + solid phase), the average F/P value was 1.3 (range ) 1.04-1.61) and the average VOL. 43, NO. 2, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
497
FIGURE 3. Dissolved (top) and particulate (bottom) PAH concentrations in the influent and effluent runoff samples collected on April 21, 2006 (left) and July 10, 2007 (right). All values are averages of duplicate samples. Note that the values for effluent concentrations are typically below the laboratory DI blank measurement. PAH abbreviations are listed in Table S1 (Supporting Information). P/A value was 6.8 (range ) 1.02-10.33), which also implies pyrogenic sources. Another relevant PAH source in urban environments is leaching from sealers used in parking lots and driveway coatings. These sealers, used to improve the appearance of asphalt pavement, are most frequently made with a coal-tar base. However asphalt-emulsion-based sealers are also used (9). A study in Texas (9) reported that the mean PAH concentration affiliated with runoff suspended solids from parking lots with coal-tar emulsion seal coat was 3500 mg/ kg, 65 times greater than those from unsealed asphalt and concrete parking lots. Bioretention PAH Performance. The monitored bioretention cell exhibited beneficial PAH removal capability (Table 1). Effluent PAH EMCs were less than input, ranging from 0.057 to 0.37 µg/L (mean ) 0.22 µg/L) as shown in Table 1. Percent reduction of EMC for total (dissolved + particulate) PAHs ranged from 31 to 99% (mean EMC reduction ) 90%). Overall percent reductions are higher when influent PAH loads are higher, as seen for the April 2006 and July 2007 storms. When influent PAH loads are low, reductions also appear low, which may be partly due to laboratory detection limit issues. Although particle size fractions in water samples were not measured in this study, it is important to note that at low influent PAH loads, the effluent concentrations of PAHs are also low, indicating that a breakthrough of PAHs associated with excess fines is likely not causing the low reductions. Effluent total PAH concentrations for both dissolved phase and particulate phase samples were close to laboratory deionized water (DI) water blank measurements for three of the five storms measured. In general, the effluent total PAH concentrations for all storms are relatively similar and removal efficiency is largely dependent on the influent PAH load entering the cell. Consequently, percent reductions should be used judiciously when evaluating BMP performance, and greater emphasis should be placed on effluent water quality. 498
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 2, 2009
Data of annual PAH mass load per unit drainage area (Lin, in kg/ha yr) are important in watershed management for BMP deployment. Lin can be estimated using the simple method (25): Lin )
PCFRVC 100
(1)
where P is the average annual precipitation [1067 mm/yr for the State of Maryland (24)], CF is a factor that corrects for events that do not produce runoff; a typical value ) 0.9 for impervious area (25) was used in this study; RV ) the runoff coefficient for the drainage area (0.9), and C is the influent PAH EMC in mg/L. Similarly, data of annual PAH mass release per unit drainage area after BMP treatment (Lout, in kg/ha yr) are capable of evaluating the watershed impact of the BMP and can be used to compare to predevelopment loads. Lout values were estimated as the product of Lin and the fraction PAH mass discharged (Table 1). The estimated Lin ranged from 0.0025 to 0.0446 kg/ha yr (mean value ) 0.0180 kg/ha yr), and Lout ranged from 0.0004 to 0.0045 kg/ha yr (mean value ) 0.0025 kg/ha yr). These data provide a PAH mass inventory basis for total maximum daily loads (TMDLs) development. Reductions in annual PAH mass loadings ranged from -71 to 99% (mean ) 87%). The negative values were caused by two events in which the effluent volume was larger than influent volume. This is assumed to have been caused by subsurface surge from the nearby (a few meters away) Campus Creek during the two more intense rainfall events. Although PAH measurements from the surge and the creek are not available, Campus Creek flows through a parking lot intensive area and may have elevated PAH concentrations. Because the goal of this study was to monitor events that produced outlet flow, overall performance may be biased because many smaller, common rainfall events will not result in flow discharge.
FIGURE 4. A comparison between PAH/TSS percent removal and rainfall depth for monitored events.
TABLE 2. log Kd (L/kg) and log Koc (L/kg) Values for Phenanthrene (phen) and Pyrene (pyr) in Stormwater Influent Samples, and a Comparison of Literature Values for Sediment, Soot, Bitumen, and Parking Lot Runoff log Kd (L/kg) storm date 4/21/2006 7/4/2006 9/14/2006 7/10/2007 stormwater influent mean bitumen, Kbitumen-water (29) runoff, coaltar-sealed lots (9) runoff, asphaltsealed lots (9) runoff, unsealed lots (9)
phen
pyr
5.55 5.06 5.29 5.69 5.46
5.58 5.20 5.70 5.82 5.63
5.7
6.4
4.88 (4.31-5.73) 5.62 (5.10-6.34) 4.69
4.63
4.16
4.60 log Koc (L/kg) phen
stormwater influent mean in sediment, Koc (28) soot, Ksoot (30)
pyr
6.58
6.75
3.7-7.0 6.24
4.0-7.5 6.79
Figure 2 shows the dissolved total PAH concentrations for the influent (0.09-0.39 µg/L) and effluent (0.03-0.09 µg/ L) stormwater flows. In comparison, particulate total PAH concentrations ranged from 0.09 to 4.76 µg/L (influent) and from 0.03 to 0.25 µg/L (effluent), as indicated in Figure 2. Influent dissolved PAH concentration measurements showed large variability, especially for the April 2006 storm event. The last three storm events shown in Figure 2 were associated with high influent flows and lower dissolved phase PAH concentrations. The influent and effluent PAH concentrations in the dissolved phase are not significantly different in these three samples. Nonetheless, these dissolved PAH concentrations are extremely low,
