Environ. Sci. Technol. 2008, 42, 5428–5434
Radiocarbon-Based Assessment of Fossil Fuel-Derived Contaminant Associations in Sediments H E L E N K . W H I T E , * ,† CHRISTOPHER M. REDDY, AND TIMOTHY I. EGLINTON Department of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543
Received February 15, 2008. Revised manuscript received May 15, 2008. Accepted May 19, 2008.
Hydrophobic organic contaminants (HOCs) are associated with natural organic matter (OM) in the environment via mechanisms such as sorption or chemical binding. The latter interactions are difficult to quantitatively constrain, as HOCs can reside in different OM pools outside of conventional analytical windows. Here, we exploited natural abundance variations in radiocarbon (14C) to trace various fossil fuel-derived HOCs (14Cfree) within chemically defined fractions of contemporary OM (modern 14C content) in 13 samples including marine and freshwater sediments and one dust and one soil sample. Samples were sequentially treated by solvent extraction followed by saponification. Radiocarbon analysis of the bulk sample and resulting residues was then performed. Fossil fuel-derived HOCs released by these treatments were quantified from an isotope mass balance approach as well as by gas chromatography-mass spectrometry. For the majority of samples (n ) 13), 98-100% of the total HOC pool was solvent extractable. Nonextracted HOCs are only significant (29% of total HOC pool) in one sample containing p,p-2,2-bis(chlorophenyl)1,1,1-trichloroethane and its metabolites. The infrequency of significant incorporation of HOCs into nonextracted OM residues suggests that most HOCs are mobile and bioavailable in the environment and, as such, have a greater potential to exert adverse effects.
Introduction Elucidating the extent and type of associations between hydrophobic organic contaminants (HOCs) and natural organic matter (OM) in soils and sediments is important in determining the fate and bioavailability of xenobiotics in the environment (1). To assess these interactions, samples are first solvent-extracted to isolate noncovalently bound OM associated by hydrophobic sorption, charge transfer complexes, and hydrogen bonding. The HOCs and residues released from this process are referred to as free or unbound (e.g., ref 2). Those that remain associated are defined as bound, are nonextracted, and are released only by harsh chemical treatments such as acid or base hydrolysis or via thermal decomposition (pyrolysis: see ref 1 and references therein). As to whether a HOC is free or bound can affect its * Corresponding author phone: (617)495-9156; fax: (617)495-8848; e-mail:
[email protected]. † Present address: Harvard University Biological Laboratories, Room 3092, 16 Divinity Ave., Cambridge, MA 02138. 5428
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 15, 2008
bioavailability, toxicity, and mobility, and bound residue formation in soils can be a major sink for contaminants (ref 3 and references therein). The significance of bound residue formation in the aquatic environment, however, is less wellstudied and may vary due to different chemical and microbial conditions in aquatic sediments as compared to soil (ref 4 and references therein). Some studies of aquatic environments, however, have indicated the presence of nonextracted HOCs (2, 5–10), but the proportions that are free versus bound are not always well-constrained. In this study, we utilized natural abundance variations in radiocarbon content (expressed here in terms of ∆14C notation) as a tool to assess the significance of binding of HOCs in aquatic sediments. Previously, ∆14C measurements have been used to trace fossil fuel or petrochemical-derived HOCs into various pools of organic matter (4, 11) and to provide quantitative assessments of fossil carbon contributions through the principle of isotope mass balance (e.g., refs 12 and 13). In these studies, the absence of 14C in fossil fuel-derived contaminants serves as an inverse tracer intrinsic to the contaminant, negating the need to use a labeled derivative in experimental simulations that may not accurately reflect in situ processes (14). Considering a conservative estimate of 10‰ ∆14C analytical error, at least 1% of the carbon in a particular OM pool of interest must be fossil fuel-derived for it to significantly influence the overall ∆14C values (4). Examination of petroleum residues in saltmarsh sediments with this approach (4) revealed no significant binding of petroleum residues to OM. This was attributed to the chemical inertness of petroleum hydrocarbons and a lack of more reactive petroleum metabolites due to the apparent absence of active biodegradation at the site (13). Understanding as to whether more reactive HOCs (e.g., those with functional groups and therefore potential reactive sites for binding to OM), or HOCs in different sedimentary environments (e.g., aerobic as opposed to anaerobic, with varying types and amounts of natural OM), would give rise to more binding is the main focus of this study. To address this, surface sediments, as well as one soil and one dust sample, were taken from various locations impacted with a range of contaminants. The latter included petroleum, creosote, coal tar, polychlorinated biphenyls (PCBs), benzotriazoles (BZTs), p,p-2,2-bis(chlorophenyl)1,1,1-trichloroethane (DDT), and other chlorinated pesticides (Table 1). Many of these sites including the Hudson River, New Bedford Harbor, and Palos Verdes have been the focus of historic studies (e.g., refs 15–17, respectively), and this work also aims to build upon our understanding of the fate of HOCs in these well-characterized environments.
Materials and Methods Study Areas and Samples. Geographic locations, dates, depths, and ancillary details of the sites examined are provided in Table 1. To test the inverse tracer approach on samples known to contain bound HOCs, soil contaminated with mineral oil from a site near Hamburg, Germany (SS (2)) was obtained as well as freshwater sediments contaminated with BZTs from the PR (ref 6 and references therein) and freshwater sediments contaminated with DDT from TC (7, 9). The PV shelf is a coastal marine site with a history of DDT contamination (17, 18), and sediments from this site were analyzed for comparison. Creosote-contaminated sediments from both a marine site, EH, and a freshwater site, CC, as well as PCB-contaminated sediments from NBH and the HR were examined. Sites contaminated with petroleum deposited in oxic sediments were sampled from PC, BH, and WC. 10.1021/es800478x CCC: $40.75
2008 American Chemical Society
Published on Web 06/20/2008
TABLE 1. Location, Date, and Depths for Samples water depth (m)
sediment depth (cm)
core type used
2003
0.5
0-5
GSa
2003 1999 2001 1995 2001 1999 2003 2005 2003
∼0.5 nd nd ∼1 60 1.5 36 0 5 nae
ndb 0-15 0-5 0-12 0-2 0-2 0-1 0-2 4-6 na
PCc GS GS GS BCd PC GS PC BC na
39°,12.3′ N, 76°,31.4′ W
1998
nd
surface
GS
Washington, DC area
1976-1977
na
na
na
location Pawtuxet River (PR), RI Station (St.) 1 St. 2 St. 3 Chattanooga Creek (CC), TN Teltow Canal (TC), Berlin, Germany Paleta Creek (PC), San Diego Bay, CA New Bedford Harbor (NBH), MA Palos Verdes (PV), CA Hudson River (HR), Glen Falls, NY Boston Harbor (BH), MA Winsor Cove (WC), MA Eagle Harbor (EH), WA soil sample (SS), Hamburg, Germany NIST SRM 1941b “organic marine sediment” (1941) NIST SRM 1649a “urban dust” (1649) a
GS: grab sampler.
b
date sampled
details 41°,45.9′ N, 71°,23.7′ W 41°,46.0′ N, 71°,23.8′ W 41°,43.4′ N, 71°,28.9′ W 35°,00.4′ N, 85°,18.2′ W sample T2(7) 32°,40.4′ N, 117°,06.9′ W 41°,40.5′ N, 70°,54.9′ W 33°,42.5′ N, 118°,21.2′ W 43°,20′ N, 73°,40′ W 42°,23.2′ N, 70°,49.4′ W 41°,39.3′ N, 70°,37.1′ W 47°,37.2′ N, 122°,30.6′ W nd
nd: not determined. c PC: push-core.
d
BC: box-core.
e
na: not applicable.
TABLE 2. OC (%), Stable Carbon, and Radiocarbon Abundancesa for Bulk Samples and Respective Residues sample
% OC TOCb
% OC EXRESc
% OC SARESd
% TOC extractablee
δ13C (‰) TOC
δ13C (‰) EX-RES
δ13C (‰) SA-RES
∆14C (‰) TOC
∆14C (‰) EX-RES
∆14C (‰) SA-RES
PR St. 1 PR St. 2 PR St. 3 CC TC PC NBH PV HR BH WC EH SS NIST SRM 1941b (1941) NIST SRM 1649a (1649) West Falmouth 12-14 cm (WF)f
1.8 5.4 1.1 4.3 7.9 1.6 11.2 1.5 2.2 1.6 4.0 1.1 2.9
1.4 3.6 2.2 4.0 6.1 1.2 5.2 1.0 1.7 1.7 2.8 0.3 1.4
1.3 2.6 0.9 2.8 3.2 1.0 5.0 0.8 2.5 1.3 2.4 0.3 0.1
22 26 17 11 12 45 3 22 67 9 9 93 58
-27.5 -27.2 -27.7 -25.2 -28.4 -24.9 -24.5 -23.4 -24.8 -22.5 -17.7 -23.6 -26.6
-27.5 -27.2 -27.7 -24.8 -28.4 -24.5 -23.0 -23.1 -25.3 -22.1 -17.1 -22.4 -27.7
-27.2 -26.6 -26.7 -24.6 -27.7 -24.5 -22.9 -23.2 -25.8 -22.8 -17.1 -22.9 -27.7
-292 -364 -241 -604 -356 -545 -581 -314 -764 -316 75.0 -869 -659
-211 -187 -192 -582 -376 -298 -367 -151 -511 -342 92.7 -425 -311
-323 -302 -388 -701 -531 -337 -512 -315 -351 -537 89.5 -392 -306
3.1
2.7
2.3
14
-24.8
-24.5
-24.6
-534
-487
-599
17.8
13.4
10.8
47
-25.7
-25.1
-25.2
-589
-549
-631
11.6
11.1
7.9
15
-17.9
-16.8
-16.7
164
102
68.3
a All ∆14C values displayed in Table 2 are corrected for 14C decay for time of collection to time of measurement (see text). b TOC is untreated sample prior to solvent extraction. c EX-RES is the extracted residue remaining after solvent extraction. d SA-RES is the saponified residue remaining after saponification. e % TOC extractable was calculated by weighing a portion of the total lipid extract (TLE) and converting this to carbon equivalents as described in the text. This value was then divided by the amount of organic carbon in the bulk unextracted sediment (TOC, Table 3). f Data for this sample taken from White et al. (4).
Standard Reference Materials 1941b “organics in marine sediment” (Baltimore Harbor, MD) and 1649a “urban dust” (Washington, DC area) from the National Institute of Standards and Technology (NIST) were selected because of the extensive information regarding their compositions and varied contamination from polyaromatic hydrocarbons (PAHs), PCBs, and pesticides, as well as the opportunity to examine the significance of binding in a different environmental matrix (atmospheric particulates) with the latter sample. Data from West Falmouth sediments impacted by No. 2 fuel oil as previously described (4) are included in this study for comparison. Sequential Chemical Treatment. Samples of 1-10 g of air-dried sediment were sieved (1.4 mm) and homogenized with a mortar and pestle. Sediments were sequentially treated as in Figure S1 (Supporting Information) similar to White et al. (4).
Bulk Sediment Analyses. Aliquots of dried unextracted sediment (TOC), solvent-extracted residues (EX-RES), and saponified sediment residues (SA-RES) were analyzed for OC content, stable carbon isotope ratio (δ13C), and radiocarbon abundance (∆14C) as in White et al. (4). Stable carbon isotopic compositions were determined by isotope ratio mass spectrometry (irMS) and 14C content by accelerator mass spectrometry (AMS) at the National Ocean Sciences Accelerator Mass Spectrometry (NOSAMS) facility at Woods Hole Oceanographic Institution (WHOI), Woods Hole, MA. All 14C measurements are expressed as ∆14C values (Table 2), which is the per mille (‰) deviation from the international standard for 14C dating, SRM 4990B “oxalic acid”. Precision for δ13C and ∆14C measurements was ∼0.1 and 1-5‰, respectively. For each sample, reported ∆14C ) [fme(1950-x)λ 1]1000 (19), where λ is 1/8267 (year-1), fm is the fraction of VOL. 42, NO. 15, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5429
TABLE 3. Quantities of Fossil Fuel-Derived Contaminants Measured by GC
sample PR St. 1 PR St. 2 PR St. 3 CC TC PC NBH PV HR BH WC EH SS NIST SRM 1941b (1941) NIST SRM 1649a (1649) West Falmouth 12-14 cm (WF)e
extractable contaminanta (mg g-1)
extractable UCMb (mg g-1)
UCM carbon range
bound contaminantc (mg g-1)
0.01 0.2 ndd 0.3 0.7 nd 4 0.002 0.2 0.004 nd 8 2 0.01 0.06 nd
2 8 0.8 0.7 nd 3 10 1 4 0 2 nd 9 3 15 6
C14-C24 C14-C24 C14-C24 C14-C38 nd C13-C38 C22-C38 C14-C34 C22-C38 C15-C38 C14-C24 nd C14-C34 C15-C38 C18-C38 C14-C24
nd 0.001 nd 0.007 0.3 nd nd nd nd nd nd 0.002 0.2 nd nd nd
a Nature of contaminant in each sediment is detailed in the text. b UCM refers to the unresolved complex mixture. Nature of contaminant in each sediment is detailed in the text. d nd: not detectable. e Data for this sample taken from White et al. (4).
c
modern 14C (corrected for isotopic fractionation using δ13C), and x is the year of collection (see Table 1). Gas Chromatography/Mass Spectrometry (GC/MS) Analysis. PR sediment extracts were acetylated and prepared for GC/MS analysis as previously described (6). All other extracts were derivatized with bis-(trimethylsilyl)-trifluoroacetamide (BSTFA) and pyridine. Aliquots of extracts from all samples were spiked with an internal standard of n-C36 alkane (4 µg) and analyzed by electron ionization (EI) using GC as described in White et al. (4). Compounds were identified from both mass spectra and GC retention characteristics. Contributions of an unresolved complex mixture (UCM) were quantified by integrating the total FID area of the UCM and using response factors determined from No. 2 fuel oil standards from the Marine Ecosystem Research Laboratory (MERL) as in Peacock et al. (20). Saponification extracts from HR were also analyzed by electron capture negative ionization (ECNI) with methane as a reagent gas and using an Agilent 6890 series gas chromatograph interfaced to an Agilent 5973 mass selective detector (MSD), with a J&W Scientific DB-XLB column (60 m × 0.25 mm i.d., 0.25 µm film thickness). The gas chromatograph oven had an initial temperature of 50 °C (1 min hold) and was ramped at 30 °C min-1 until 120 °C and then at 8 °C min-1 to 300 °C (30 min hold). Spectra were acquired between m/z 50 and 750 at a scan rate of 1 cycle s-1.
Results and Discussion Solvent-extractable Contaminants. Samples examined in this study had a wide range of TOC values (1.1-17.8%; Table 2), and thus, the quantities of HOCs extracted from the samples will be discussed in terms of their contribution to the sediment as a whole as well as relative to TOC. Several BZT compounds including C1-BZT, C4-BZT, C4-Cl-BZT, C5Cl-BZT, C8-BZT, and C10-BZT were isolated from PR sediments from St. 1 and St. 2 and were dominated by C1BZT with maximum concentrations of 6 µg g-1 dry weight (dw) sediment (0.33 mg g-1 OC) and 100 µg g-1 dw (1.9 mg g-1 OC), respectively. BZT compounds were not identified at Station 3 due to its location upstream of the chemical plant that released BZTs into the river. Sediment extracts contaminated with PCBs from New Bedford Harbor were dominated by di-, tri-, and tetrachlorobiphenyls and also contained penta- and hexachlorobiphenyls with a summed concentration of 4200 µg g-1dw (37.5 mg g-1 OC). Sediments 5430
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 15, 2008
from the HR were less contaminated with summed mono-, di-, tri-, and tetrachlorobiphenyl concentrations of 170 µg g-1 dw (7.7 mg g-1 OC). Sediments from TC contained various extractable DDT metabolites including 2,2-bis(4-chlorophenyl)-1,1-dichloroethane (DDD), 2,2-bis(4-chlorophenyl)-1,1dichloroethene (DDE), and bis(4-chlorophenyl)acetic acid (DDA) with a total concentration of 740 µg g-1 dw (9.4 mg g-1 OC), whereas PV sediments were much less contaminated (0.002 µg g-1 dw and 0.1 µg g-1 OC of DDE). Creosotecontaminated extracts from CC and EH were dominated by PAHs including fluoranthene, pyrene, phenanthrene, benzo[b]fluoranthene, benz[a]anthracene, benzo[a]pyrene, chrysene, fluorene, and anthracene with individual PAH concentrations between 10 and 50 µg g-1 dw (0.23-1.16 mg g-1 OC) and 100-1400 µg g-1 dw (9.1-127.3 mg g-1 OC), respectively. PAHs were also the dominant HOCs in sediment extracts from BH with fluoranthene, pyrene, and benz[a]anthracene at concentrations of 1-2 µg g-1 dw (0.06-0.12 mg g-1 OC) as well as the soil sample, which contained phenanthrene, anthracene, pyrene, fluoranthene, and chrysene at concentrations of 60-570 µg g-1dw (2.1-19.7 mg g-1 OC). Solvent-extractable contaminant identifications and abundances in NIST SRM 1941b and NIST SRM 1649a are discussed in their corresponding certificates of analysis but contribute total contaminant concentrations of ∼10 µg g-1 dw (0.32 mg g-1 OC) and ∼60 µg g-1 dw (0.34 mg g-1 OC), respectively. Varying quantities of UCM with different carbon ranges also were detected in most samples except TC and EH (Table 3). In all samples, no additional HOCs or HOC derivatives were observed in the saponified TLE. Assuming that all contaminants are solvent extractable, the relative contribution of solvent-extractable fossil fuelderived HOCs can be calculated based on an isotope mass balance in terms of radiocarbon abundances (∆) f ) (∆t - ∆n)/(∆x - ∆n)
(1)
where f is the fraction of the mixture that is fossil fuel-derived, ∆t is the decay-corrected ∆14C value for the untreated sediment (TOC; Table 2), ∆n is the decay-corrected ∆14C value for the extracted sediment residue (EX-RES; Table 2), and ∆x is the ∆14C of HOCs derived from fossil fuel (i.e., ∆14C ) -1000‰). Solutions of f obtained are expressed as the percentage of TOC that is fossil fuel-derived (Table 4). These values are compared to those calculated from chromatographically resolvable contaminant and UCM concentrations
TABLE 4. Comparison of Quantities of Fossil Fuel-Derived Contaminants Measured by Isotope Mass Balance and GC % OC of TLE that is fossil fuelderived by
% OC of SAE that is fossil fuelderived by
sample
mass balance
GCa
mass balance
GCa
PR St. 1 PR St. 2 PR St. 3 CC TC PC NBH PV HR BH WC EH SS NIST SRM 1941b (1941) NIST SRM 1649a (1649) West Falmouth 12-14 cm (WF)b
10 22 6 5 -3 35 34 19 52 -4 2 77 51
8 13 6 2 0.5 18 10 6 16 0 3 72 32
0 0 0 0 0 0 0 0 25 0 0 5 1
0 0.002 0 0.02 0.4 0 0 0 0 0 0 0.1 1
9
9
0
0
9
7
0
0
8
6
0
0
a To calculate % OC of TLE and SAE that is fossil fuel-derived, the mass of contaminant and UCM was converted to carbon equivalents, which varied depending on the contaminant in question. See text for details. b Data for this sample taken from White et al. (4).
(measured by GC-FID; Table 3) relative to TOC content (determined by elemental analysis; Table 2). For this, the mass of UCM was converted to carbon equivalents considering the mass contribution of hydrogen to the hydrocarbon, which in an alkane typical of the UCM (n-C10 to n-C38 as before) is ∼15% of the total. It is assumed that the UCM is entirely fossil fuel-derived (21). Similar conversions of contaminant quantities to carbon equivalents were performed for the other contaminants examined considering noncarbon elements such as oxygen, nitrogen, chlorine, and hydrogen. On average, 95% PAHs, 70% BZTs, 60% PCBs, and 50% DDT and its metabolites are derived from carbon. To evaluate the quantity of bulk OC that these compounds represent, the percent of TOC that is extractable was calculated by weighing a portion of the TLE and converting this weight to carbon equivalents. The percent of TLE that is GC-amenable was then calculated (Table S1). For the majority of sites examined (n ) 10), there is good agreement (within 0-5%) between the percent TOC that is fossil fuel-derived determined by GC-FID as compared to that calculated by isotope mass balance (Table 4). For PR St. 2, PC, NBH, PV, HR, and SS, significant differences were observed, and in each case, the extractable fossil fuel-derived OC calculated by isotope mass balance was greater (9-36%) than that quantified by GC-FID. These samples each contain g19% fossil fuel-derived OC as calculated by isotope mass balance, whereas the samples with good agreement between the two methods contain e10% (Table 4). The previous study of West Falmouth sediments (4) (a representative contaminated sediment horizon is shown in Table 4) showed good agreement between the % fossil fuel-derived OC calculated by GC-FID as compared to isotope mass balance, and the contribution of petroleum was 10% TOC. One exception is the sample from EH, which contains 77 and 72% fossil fuel-derived OC calculated by isotope mass balance and GC-FID, respectively. EH sediment differs from the other
FIGURE 1. Difference between the % TLE that is non-GC-amenable and the difference in % fossil fuel-derived OC calculated by GC-FID as compared to isotope mass balance. samples in that it is largely composed of coarse, sandy material and that the majority of OC is removed by solvent extraction (93%; Table S1) with the latter mainly comprised of HOCs (72-77%; Table 4). After solvent extraction, only a small quantity (0.3%) of OC remains in EX-RES (Table 2). This anomaly suggests that the offset observed in the other sites reflects the influence of biogenic OC that is coextracted with fossil fuel-derived OC. The TLE in these samples is a mixture of fossil fuel-derived OC (19-52%; Table 4) combined with biogenic compounds (primarily lipids). Biogenic OC extracted with the HOCs can dilute the 14C-dead fossil carbon signature and/or mask the contribution of HOCs observable by GC. Extractable, GC-amenable biogenic lipid-derived components contribute significantly (10-50%) to sediment extracts from HR, PV, and all PR sites but comprise
