PAHs in Dated Sediments of Ashtabula River, Ohio, USA - American

sediment cores from Ashtabula River in Ohio, USA. The Ashtabula River drains a watershed of 350 km2 in northeast Ohio and northwest Pennsylvania and ...
0 downloads 0 Views 91KB Size
Environ. Sci. Technol. 2001, 35, 2896-2902

PAHs in Dated Sediments of Ashtabula River, Ohio, USA KAI LI,† ERIK R. CHRISTENSEN,* RYAN P. VAN CAMP,‡ AND IPEK IMAMOGLU Department of Civil Engineering and Mechanics and Center for Great Lakes Studies, University of WisconsinsMilwaukee, Milwaukee, Wisconsin 53201

Polycyclic aromatic hydrocarbons (PAHs) in dated sediments from Ashtabula River, Ohio, were determined, and a chemical mass balance (CMB) model was used to apportion sources. Three cores (AR-1, AR-2, AR-3) were dated by correlating uranium-supported 210Pb peaks with 1964, 1972, 1977, and 1979 maxima in the discharge record for Ashtabula River. These cores had sedimentation rates between 7.1 and 4.4 cm/year, while a fourth (AR-4) exhibited a much higher rate of 27.8 ( 18 cm/year. The highest PAH concentration was 11500 ng/g found in layer 6 of AR-1 (1986), and the lowest was 621 ng/g found in layer 8 of AR-2 (1982). The source contributions to the total PAH concentrations estimated by the CMB model are 0.12.2%, 16.8-22.8%, and 78.1-83.8% for wood burning (WB), coke oven (CO), and highway dust (HWY), respectively. Petroleum generated PAHs have maximal contribution during 1977-79, and wood burning PAHs show minimal emissions during 1975-77 in accordance with U.S. consumption records and other studies. Among six PAH markers, only phenanthrene may be subjected to aerobic biodegradation or photolysis with an apparent half-life of 0.005-0.025 year. No anaerobic degradation was observed based on the CMB model. The model works well for the nonmarker compounds, fluoranthene, and benzo[b]fluoranthene/benzo[k]fluoranthene.

Introduction Polycyclic aromatic hydrocarbons (PAHs) are a class of compounds that consist of two or more fused aromatic rings, and some of them are carcinogens in animals and are suspected carcinogens in humans (1, 2). PAHs are produced mainly by incomplete combustion of organic matter. Major sources include emissions from wood and coal burning, coke ovens, automobile exhaust, heat and power generation, and refuse burning. After entering the environment, PAHs are widely dispersed by atmospheric transport or through stream pathways, and eventually accumulate in soils and aquatic sediments. U.S. Environmental Protection Agency (EPA) has listed 16 PAHs as priority pollutants in wastewater and 24 PAHs in soils, sediments, hazardous solid wastes, and groundwater (3). Several studies indicate that atmospheric input is the main source of PAHs to sediments and soils, and * Corresponding author phone: (414) 229-4968; fax: (414) 2296958; e-mail: [email protected]. † Present address: Pharmaceutical Products Development, Inc. 8500 Research Way, Middleton, WI 53562. ‡ Present address: Ramaker and Associates, Inc. 1120 Dallas Street, Sauk City, WI 53583. 2896

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 14, 2001

FIGURE 1. Sampling sites on Ashtabula River, Ohio, USA the PAHs in aquatic sediments may serve as tracers for the identification of the sources of these organic pollutants (49). Knowledge of sources and pathways of pollutants in the environment is important for effective pollution abatement. An effective approach in quantifying the actual contribution of various sources is to use receptor models, such as a chemical mass balance (CMB) model (10). Source apportionment of PAHs in sediments has been studied by various groups using CMB models (6, 11). Recently, we have apportioned PAH sources for the sediments from several rivers, such as Fox River and Kinnickinnic River in Wisconsin and the Black River in Ohio (9, 12-14). In this paper, attention was focused on the source apportionment for PAHs in dated sediment cores from Ashtabula River in Ohio, USA. The Ashtabula River drains a watershed of 350 km2 in northeast Ohio and northwest Pennsylvania and empties into the central basin of Lake Erie at the city of Ashtabula. There is concentrated industrial development in Ashtabula along the river and a tributary stream, Fields Brook. In 1983, the 14 km2 Fields Brook watershed was designated by U.S. EPA as a CERCLA (“Superfund”) National Priorities List site based on elevated sediment concentrations of several contaminants including uranium and daughters, PAHs, polychlorinated biphenyls (PCBs), hexachlorobenzene, hexachlorocyclopentadiene, and heavy metals such as Cd, Cr, Hg, Pb, Zn, and As (15). Because of the heavily polluted sediments, the area south of the 5th Street bridge including the portion of Ashtabula River shown in Figure 1 has not been dredged since 1962 (16). The present study is a further examination of PAH source apportionment based on a CMB model, for dated (19651998) sediments of a Great Lakes tributary. Attention is focused on the evaluation of possible degradation of PAHs by modification of a CMB model. We expect that lack of fit in a CMB model in which degradation is not considered may indicate possible degradation. Apparent half-life will be estimated based on the optimal fits in a CMB model when source profiles are modified to include a degradation factor. The CMB model will also be applied to some nonmarker compounds.

Materials and Methods Sampling and Sediment Dating. Collection of vibra cores from the Fields Brook area of Ashtabula River (Figure 1) was carried out in May, 1998 from the U.S. EPA research vessel, R/V Mudpuppy. The 10.2-cm diameter cores were contained in polycarbonate tubes. The cores were between 131 and 237 cm long. After collection, they were frozen at -18 °C for several days and then sliced into 15 sections. The porosity 10.1021/es001790f CCC: $20.00

 2001 American Chemical Society Published on Web 06/07/2001

and organic content (loss on ignition) were measured, and the samples were also dated by measuring 210Pb and 137Cs activities (17). The dating was established from a plot of log 210Pb excess activity vs depth (AR-4) or by correlating uranium-supported 210Pb peaks with episodic high discharges of the Ashtabula River (Figure A, Supporting Information). Storm events occurred on 3/5/64, 3/2/72, 12/15/77, and 12/ 25/79. Material from each section was transferred into amber glass bottles and stored at -18 °C prior to PAH analysis. Materials. All PAH standards, surrogate standards (2fluorobiphenyl), and internal standard (triphenylmethane) were purchased from AccuStandard Inc. (New Haven, CT). Solvents used in analysis were obtained from Fisher Scientific (Chicago, IL) and are ACS grade or higher. PAH Analysis. Measurement of PAHs in sediments is performed based on the procedures established in our laboratory (18) and U.S. EPA Method 8270 (19) and briefly summarized here. Sediment samples were thawed, freeze-dried, and then extracted in Soxhlet using hexane/acetone (1:1, v/v). The extract was cleaned up by a silica gel chromatographic column and then analyzed by GC-MS using a DB-5 column (30 m × 0.25 mm × 0.25 µm, J & W). The following 17 parent PAHs and 3 methylated PAHs were determined: naphthalene (NaP), acenaphthylene (AcNP), acenaphthene (AcN), fluorene (Fl), phenanthrene (PhA), anthracene (AN), 1-methylphenanthrene (1-mPhA), 2-methylphenanthrene (2-mPhA), 3,6-dimethylphenanthrene (3,6-dmPhA), fluoranthene (FlA), pyrene (Py), benz[a]anthracene (BaA), chrysene (Chy), benzo[b]fluoranthene (BbFlA), benzo[k]fluoranthene (BkFlA), benzo[a]pyrene (BaP), benzo[e]pyrene (BeP), indeno[1,2,3-cd]pyrene (IP), dibenz[a,h]anthracene (dBahA), and benzo[ghi]perylene (BghiP). Chemical Mass Balance (CMB) Model. The CMB model is used for solving contributions from different sources to a sample at a receptor site (9, 20, 21). The basic idea is that the measured chemical pollutants in a sample are the sums of the contributions from several sources. The model relies on two basic assumptions. The first is the amount of a chemical pollutant in the sediment at a receptor site is the sum of the amount of this chemical from each independent source. The second is that the ratio of the concentration in the sample of a chemical from a given source, and the concentration in the source material, is the same for all chemicals in the study (i.e., no change in source profile between source and receptor). Therefore, n

Fj )

∑ Φ R + e (1 e j e m) ji i

j

(1)

i)1

where Fj is the measured concentration of the jth PAH compound in the sample, Φji is the concentration of jth PAH in the ith source, Ri is the source contribution factor of the ith source, ej is the error associated with the jth PAH, n is the number of sources, and m is the number of PAH marker compounds used in the model. The contribution factors (Ri’s) are determined by multiple linear regression using a leastsquares method in which the weighted error (equivalent to χ2) is minimized. This is an effective variance method that includes errors in both source and sample profiles (9). Source contributions were calculated as shown in Christensen et al. (9). The number of marker compounds (m) must be greater than the number of sources (n) (9). Both χ2 and the multiple linear correlation coefficient R2 (21) were used to assess the fit between the calculated and the measured PAH concentrations. During the calculation of χ2, the relative errors of the sample and source profiles were used to account for the uncertainty of both profiles. The relative source profile errors were determined to be 0.35, 0.37, and 0.40 (CV) for

TABLE 1. PAH Source Profiles Used in CMB Modeling (ppm) WBa

COb

marker compounds naphthalene (NaP) 12.5 17.0g fluorene (Fl) 0.4 17.2 phenanthrene (PhA) 2.54 541.0 anthracene (AN) 0.44 391.0 pyrene (PY) 0.91 826.0 benz[a]anthracene (BaA) 0.14 1658.0 benzo[a]pyrene (BaP) 0.018 2148.0 CVe 0.35 0.37 MWf 138.9 222.7 nonmarker compounds fluoranthene (FlA) 1.13 704.00 benzo[b]fluoranthene 0.11 1356.10 (BbFlA)/benzo[k]fluoranthene (BkFlA)

HWY-1c

HWY-2d

0.25 0.82 4.62 0.96 3.86 0.80 0.77 0.40 189.0

0.02 0.20 2.20 0.83 3.34 1.01 0.63 0.38 196.6

5.48 1.84

4.70 2.58

a Wood burning (open fire). b Coke oven emission. c Highway dust collected in Milwaukee, WI, 1990. d Highway dust collected in Milwaukee, WI, 2000. e Coefficient of variation (relative error). f Average molecular weight (9). g Upper limit.

WB, CO, and HWY, respectively (Table 1). The levels of relative errors for samples were set at 0.1, 0.2, 0.4, 0.5, and 0.6 in the CMB model. PAH Source Profiles. Although PAHs in the environment may be contributed from many sources, only three source profiles were chosen as the major contributors for PAHs in the sediments based on our previous studies (9, 12-14). These are emissions from coke ovens (CO), particularly in steel manufacturing, wood burning (WB), and highway dust (HWY). The compounds of interest, i.e., PAH marker compounds, are given in Table 1. The PAH data for WB are from Jenkins et al. (22). The PAH data for CO were originally generated by Lao et al. (23). Since the data from samples collected from a silver membrane filter provided a better fit in CMB modeling (12), these PAH data are selected and used in this study. The HWY profile (HWY-1) was obtained from Singh et al. (6). An alternative HWY profile (HWY-2) was determined based on six individual samples collected and analyzed during April and May, 2000 (Table A, Supporting Information). This profile is listed for comparison only since it did not provide acceptable fits, when it was used in CMB modeling, to measured PAH concentrations either because of several negative source contributions (< -3%), or poor statistics (χ2 = 6-12 for df ) 4). Average values were calculated as geometric averages (13). Note the following special conditions regarding the source profiles. The NaP concentrations of 12.5 and 17.0 ppm (estimated upper limit) were found to be satisfactory in the WB and CO profiles, respectively, despite potential problems (12). Also, the BaA concentration in the CO profiles is based in part on the value for the coarse fraction of samples collected on glass fiber filters (12, 23). Two nonmarker compounds, FlA and BbFlA/BkFlA, were used for checking the model. Photolysis and Biodegradation. The second assumption in CMB modeling does not hold if differential loss of PAHs occurs from source to receptor. This is true typically for low molecular-weight PAHs because of photooxidation in the atmosphere or photolysis or biodegradation in sediments. The lack of fit in a CMB model in which degradation is not considered may indicate possible degradation. In fact, lowerthan-expected measured concentrations will be taken as an indication that one or several of these processes may have occurred. It is well-known that biological or chemical processes can degrade organic contaminants in soil, sediments, and water. A detailed overview of photolysis and biodegradation VOL. 35, NO. 14, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2897

rates for PAHs in water and sediments is available in Mackay et al. (24). Because of the limitation of light input, photodegradation of PAHs is only considered to be likely in the shallowest areas. Since several factors, such as temperature, light, and bacterial communities are involved, the half-lives of PAHs can vary widely. Degradation of two- and three-ring PAHs, such as naphthalene, anthracene, and phenanthrene, commonly occurs via aerobic bacteria (25, 26), and therefore, aerobic biodegradation of PAHs may also occur in the sediment. However, much of the degradation data in the literature are from petrogenic PAHs while combustiongenerated PAHs are more resistant to degradation (27). When degradation is considered, the previously used CMB model should be modified. We made three assumptions for the degradation model in sediments: that PAH degradation is a first-order reaction, that the PAH flux into sediments is similar to plug flow (i.e., minimal mixing occurs), and that biodegradation or photolysis occurs in the upper sediment layers, close to the sediment-water interface, where aerobic conditions prevail. Therefore, the fundamental equation can be modified with an exponential factor exp (-kj’t) to n

Fj )

∑Φ

ji

Ri e-kj′t + ej (1 e j e m)

(2)

i)1

kj′ ) ln 2/t′1/2

(3)

where kj′ (years-1) is an apparent first-order rate constant of degradation of compound j, t′1/2 is an apparent half-life of compound j, and t (years) is the reaction time for aerobic biodegradation or photolysis. The following equation applies for the aerobic reaction time (t):

t ) ∆z/ν

(4)

where ∆z (mm) is the thickness of the top aerobic layer and ν (mm/year) is the sedimentation velocity in the top sediment layer. The aerobic layer is defined as the top section of the sediment where the dissolved oxygen level of the interstitial water is g2 mg/L. The thickness of the aerobic layer of Ashtabula River sediments is 1.5 mm as determined by Van Camp (17). The sedimentation rates were 7.1, 4.7, 4.4, and 27.8 cm/year for AR-1, 2, 3, and 4, respectively (17).

Results and Discussion Sediment Dating. On the basis of the 210Pb and 137Cs dating results (Figure B, Supporting Information), the sediment cores span the time period of 33 years from 1965 to 1998 (17). The dating for AR-1, 2, and 3 was established by assigning the top of the cores to the sampling date (5/13/98) and the deepest full 210Pb peak to the flooding event of 3/2/72, while other layers were dated by linear inter- or extrapolation. The resulting dating is consistent with the 137Cs data. In particular, the highest 137Cs activity in AR-1 is, as expected, at the bottom layer (1966) where the dates are close to the time of max 137Cs activity (1963). Note also that dates assigned to other 210 Pb peaks, that is 1977 and 1981 for AR-1, 1980 for AR-2, and 1976 and 1979 for AR-3 are in good agreement with the previously mentioned dates of high discharge for the Ashtabula River. Even though the discharge of March 5, 1964 predates the 210Pb peaks, there is a suggestion from the relatively high 210Pb activity in layer 15 that there may be a corresponding 210Pb peak just below this bottom layer of core AR-1. The 210Pb peak corresponding to 12/15/77 is absent, or hidden as a shoulder, in core AR-2 possibly because of lack of settling due to the location of this site at the confluence of the Fields Brook and the Ashtabula River. 210Pb levels are 2898

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 14, 2001

higher in AR-3 because this core is downstream of the Fields Brook source. Cessation of significant uranium supported 210 Pb activity occurs at about the same time (1979-1981) in all three cores and appears to be coincident with the flooding event of 12/25/79. The timing is consistent with the fact that the main uranium source, the Millennium Chemical plant, discarded concentrated uranium containing wastes on site until 1976 (16). The hypothetical correlation between 210Pb peaks and episodic discharges is supported by the uranium peaks in cores 17, 24, 45, and 47 from the Ashtabula River, shown in the study by Ketterer et al. (16). In fact, several uranium peaks are especially clear in core 45, close to our core AR-3. Assuming that the largest and deepest peak is from 1964, and using linear interpolation, other peaks occur in 1971, 1975, 1978, and 1979 in good agreement with times of maximum discharges (Figure A, Supporting Information). The dating of the deepest peak is supported by a similar peak in core 47 from the same area, since this peak can be quite accurately dated to 1964, based on 137Cs data. To provide further support for the dating established by 210Pb and 137Cs, we have plotted total PCB concentration vs depth in the lower panel of Figure B, Supporting Information. Analytical procedures are described in Li and Christensen (28). Maximum PCB concentration occurs between 1970 and 1976 for cores AR-1, 2, and 3. Since this was the time PCBs were effectively banned, the above dating is in fact supported. Core AR-4 that comes from the side of the river in a quiescent area has a high sedimentation rate (27.8 ( 18 cm/ year). On the basis of this rate, the cycles in the 210Pb and 137Cs data (Figure B, Supporting Information) appear to be annual, supporting 6 years of deposition from 1992 to 1998. PAH Concentrations in the Sediment Cores. A total of 20 PAHs and methylated PAHs were quantified by GC-MSD. To ensure the reliability of the analysis, method blanks, standard reference sample (NIST 1941a, Organics in Marine Sediment), sample replicates and sample spike replicates were performed prior to analysis of real sediment samples. Analysis of NIST 1941a in four replicates showed good agreement between measured and certified values (Table B, Supporting Information). The average recovery of the surrogate standard (2-fluorobiphenyl) was 89%, ranging from 49 to 125%. The measured PAH concentrations in the sediment cores were corrected by surrogate recovery from individual samples. Results from the 7 marker compounds in core AR-1 are shown in Table 2. In addition, results for all PAHs for all cores are included in the Supporting Information (Tables C-F). It is seen that PhA, AN, and PY are dominant in most layers indicating HWY dust as a likely source (Table 1). However, several top layers, e.g., #1, 3, 4, and 6 have also high values of BaA and BaP reflecting significant contributions from coke ovens. Wood burning is seen as a source when the naphthalene concentration is high, as in layers 1 and 2. Results for total PAHs indicate large variations in PAH concentrations between layers (Figure 2). For example, total PAH concentrations ranged from 1710 to 11500 ng/g in AR-1 and from 621 to 6980 ng/g in AR-2. This may have resulted from changing patterns of industrial activities and vehicle traffic during that period of time. The average PAH concentrations were 6190, 3880, 4970, and 3650 ng/g for sediment cores, AR-1, AR-2, AR-3, and AR-4, respectively. The highest PAH concentration was 11500 ng/g found in layer 6 of AR-1 (1986) and the lowest was 621 ng/g found in layer 8 of AR-2 (1982). There is a minimum in the total PAH concentrations in AR-1, 2, and 3 around 1982-1983. This is consistent with a 1981 minimum found in the sediments from the Milwaukee Basin of central Lake Michigan (12). This may be a result of the reduced gasoline supplies during that time. Our results regarding the 1982-83 minimum are similar to the findings of a recent study of PAHs in sediments of lakes and reservoirs in United States (29).

TABLE 2. Concentrations (ng/g) of 7 PAH Marker Compounds Found in Sediment Core AR-1 layer

depth (cm)

dates

NaP

Fl

PhA

AN

PY

BaA

BaP

AR-1-1 AR-1-2 AR-1-3 AR-1-4 AR-1-5 AR-1-6 AR-1-7 AR-1-8 AR-1-9 AR-1-10 AR-1-11 AR-1-12 AR-1-13 AR-1-14 AR-1-15

0-15.8 15.8-31.6 31.6-47.4 47.4-63.2 63.2-79.0 79.0-94.8 94.8-110.6 110.6-126.4 126.4-142.2 142.2-158.0 158.0-173.8 173.8-189.6 189.6-205.4 205.4-221.2 221.2-237.0

1998-1996 1996-1994 1994-1992 1992-1989 1989-1987 1987-1985 1985-1983 1983-1981 1981-1978 1978-1976 1976-1974 1974-1972 1972-1969 1969-1967 1967-1965

32.4 10.3 9.9 1.9 12.0 1.9 1.9 1.9 2.4 1.9 1.9 3.5 34.0 1.9 3.2

59.8 31.0 84.0 84.7 140 191 57.7 181 145 201 191 487 234 117 311

259 121 309 281 882 814 161 538 395 526 419 1023 666 294 827

299 147 341 315 157 900 201 202 429 571 451 1107 712 319 881

1352 434 789 584 1103 1624 356 393 803 560 419 1107 646 346 1048

779 15.5 213 185 366 619 67.7 765 288 101 109 317 95.9 104 324

1121 33.6 294 241 347 667 120 244 145 189 70.5 306 122 92.3 231

FIGURE 2. Fitted curves for total PAH (ng/g) found in AR sediments versus sediment year Source Apportionment for PAHs. Although 20 PAHs in AR sediments were quantified by GC-MSD, only seven PAHs were selected as marker compounds due to the limitation of compounds in the available source profiles. These seven marker compounds were used in source apportionment. They are naphthalene (NaP), fluorene (Fl), phenanthrene (PhA), anthracene (AN), pyrene (PY), benz[a]anthracene (BaA), and benzo[a]pyrene (BaP) (Table 1). The results demonstrate that WB, CO, and HWY made different contributions to PAH content in the sediment cores (Table 3). Some layers were excluded in the modeling process because of the nature of samples, such as high sand content and low value in loss-on-ignition, and nondetection of some marker compounds. Three layers in AR-1 and two layers in AR-3 showed negative contributions (< -3%) and were not included in the calculation of averages. Sample AR-1-1 shows an unusually high CO content. However, because the other three cores show more moderate CO levels in the upper layers (30-38%), we believe that AR-1-1 represents a random contaminated sample rather than a trend. In all four cores, HWY was the predominant source and contributed 78-84% of the total PAHs. This result is similar to that of Lim et al. (30) who found that the contribution of traffic to concentrations of 4-7 ring PAHs was 80-82% and 61-67% for city center and campus sites, respectively. The implication of automobile traffic as a major source of PAHs

to urban areas is also consistent with the findings of Van Metre et al. (29). CO contributed around 17-23%. The average contribution from WB is less than 3% and the individual contribution is never beyond 5%. HWY contributions reached a maximum during 1977 to 1979, and this is in agreement with the high petroleum consumption during that period (12). Contributions from WB were negligible in AR-1 and AR-2, and increased to 2.2 and 1.1% in AR-3 and AR-4, respectively. There is minimal contribution from WB around 1977 in AR-3. This is in good agreement with similar minima during 1970-76 determined for Lake Michigan (12). Although the CMB model is quite robust to changes in CO source, WB source, and sample profiles (12), it is sufficiently sensitive to the HWY profile to not accept the alternative HWY profile HWY-2 (Table 1). This is despite the fact that the average molecular weight for the latter profile (196.6) is fairly close to the one for HWY-1 (189.0), while still being significantly lower than the CO molecular weight of 222.7. There are several possible reasons for this including local industrial sources, changing automobile emissions, and different gasoline vs diesel mix. In our experience, there are few high quality PAH profiles available in the literature from these and related sources. Nevertheless, the three sources WB, CO, and HWY-1 do provide a satisfactory statistical fit to measured PAH data, with average χ2 values ranging from 4.48 to 5.93 (df ) 7 - 3 ) 4). The ranges of average correlation coefficient R2 and relative error for χ2 ) df ) 4 are 0.78-0.82 and 44-54%, respectively. Note that sample HWY-1 was collected at an exit ramp (72C, south) from I-43. Also, two of the six samples for HWY-2 were obtained near interstates I-43 and I-94. For these, the average molecular weights were 193.0 and 193.2, respectively, that is very close to the value of 189.0 for HWY-1. The other four samples were taken at local city streets (Table A, Supporting Information). The failure of HWY-2 to provide an adequate source profile may, therefore, be because the main highway source to Ashtabula River is the nearby interstate I-90 rather than city streets. The stop-and-go traffic on city streets is associated with accelerated engines and apparently higher molecular weight PAHs. As for the CO emissions, they are probably coming from coke ovens for steel making and other industrial purposes in Cleveland, Youngstown, Pittsburgh, and Buffalo, depending on the prevailing wind direction. We demonstrated previously that coal-fired power plants have very low PAH emissions, typically