Environ. Sci. Technol. 2004, 38, 3689-3697
Compound-Specific Carbon and Hydrogen Isotope Analysis of Sub-Parts per Billion Level Waterborne Petroleum Hydrocarbons Y I W A N G , † Y O N G S O N G H U A N G , * ,† JAMES N. HUCKINS,‡ AND JIMMIE D. PETTY‡ Department of Geological Sciences, 324 Brook Street, Brown University, Providence, Rhode Island 02912, and Columbia Environmental Research Center, United States Geological Survey, Columbia, Missouri 65201
Compound-specific carbon and hydrogen isotope analysis (CSCIA and CSHIA) has been increasingly used to study the source, transport, and bioremediation of organic contaminants such as petroleum hydrocarbons. In natural aquatic systems, dissolved contaminants represent the bioavailable fraction that generally is of the greatest toxicological significance. However, determining the isotopic ratios of waterborne hydrophobic contaminants in natural waters is very challenging because of their extremely low concentrations (often at sub-parts ber billion, or even lower). To acquire sufficient quantities of polycyclic aromatic hydrocarbons with 10 ng/L concentration for CSHIA, more than 1000 L of water must be extracted. Conventional liquid/liquid or solid-phase extraction is not suitable for such large volume extractions. We have developed a new approach that is capable of efficiently sampling sub-parts per billion level waterborne petroleum hydrocarbons for CSIA. We use semipermeable membrane devices (SPMDs) to accumulate hydrophobic contaminants from polluted waters and then recover the compounds in the laboratory for CSIA. In this study, we demonstrate, under a variety of experimental conditions (different concentrations, temperatures, and turbulence levels), that SPMD-associated processes do not induce C and H isotopic fractionations. The applicability of SPMD-CSIA technology to natural systems is further demonstrated by determining the δ13C and δD values of petroleum hydrocarbons present in the Pawtuxet River, RI. Our results show that the combined SPMD-CSIA is an effective tool to investigate the source and fate of hydrophobic contaminants in the aquatic environments.
Introduction Compound-specific carbon and hydrogen isotope analysis (CSCIA and CSHIA) has become a powerful tool for studying the sources and environmental fate of organic contaminants. The isotopic ratios of the contaminant compounds are useful tracers for their sources (e.g., 1, 2), while the isotopic fractionation can be used to quantitatively assess the progress of an environmental process such as biodegradation (e.g., 3, * Corresponding author phone: (401)863-3822; fax: (401)863-2058; e-mail:
[email protected]. † Department of Geological Sciences. ‡ Columbia Environmental Research Center. 10.1021/es035470i CCC: $27.50 Published on Web 05/22/2004
2004 American Chemical Society
4). A wide range of compounds have been studied, including alkanes in crude oil (1, 5), BTEX compounds (benzene, toluene, ethylbenzene, o-xylene, and p-xylene), tert-butyl methyl ether (MTBE) in gasoline (2, 4), the chlorinated solvents trichloroethene (TriCE) and tetrachloroethene (TCE) (6), polycyclic aromatic hydrocarbons (PAHs) in soil (7), and polychlorinated biphenyl (PCB) congeners in animal tissues (8). In natural aquatic systems, the freely dissolved fractions of hydrophobic organic contaminants generally have the greatest impacts on aquatic organisms and, thus, represent the most ecotoxicologically relevant environmental residues. Unfortunately, environmental concentrations of this fraction of residues are typically low (often at the sub-parts per billion level). Sampling sufficient quantities of these compounds for compound-specific isotope analysis (CSIA; particularly CSHIA) using conventional liquid/liquid extraction (LLE) and solid-phase extraction (SPE) approaches is difficult and timeconsuming. For example, ∼300 ng is needed for an alkane and 600∼800 ng is needed for each two to three ring PAH for a single run of CSIA of D/H ratios. The lower detection limits of alkanes than of PAHs is due to the smaller number of H atoms in PAH molecules. For example, naphthalene (C10H8) has 14 fewer H atoms than decane (C10H22). The ratios of H/C decrease with increasing ring numbers and fusion of PAHs. For a three to four ring PAH at a 10 ng/L concentration, 100 L of water must be extracted to obtain 1000 ng of mass for a single injection. In practice, however, to deliver the 1000 ng mass onto the gas chromatography high-temperature conversion-isotope ratio mass spectrometer (GC-TC-IRMS system) using a standard syringe injection, the compounds need to be dissolved in >20 µL of a solvent, and 800 Da) of triolein greatly reduces its permeability or losses through the LDPE membrane during organic solvent dialysis to recover accumulated contaminant molecules. The random thermal motions of the polymer chains on LDPE form transient cavities with maximum diameters of approximately 10 Å (9). Because the crosssectional diameters of many environmental contaminant molecules are smaller than or approach the size of LDPE cavities, only dissolved and readily bioavailable organic contaminants can diffuse into the membrane and be concentrated in the SPMD. Concentration factors, i.e., VOL. 38, NO. 13, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3689
concentration in the SPMD divided by concentration in the exposure medium, for various hydrophobic contaminants by SPMD can be greater than 4 × 105 (9), allowing high efficiency accumulation of trace-level waterborne contaminants. The upper limit is generally controlled by the differential in the thermodynamic solubility of the particular chemical in triolein plus LDPE and water (10). The objectives of this study are as follows: (1) to conduct laboratory tests to determine if the SPMD sampling process results in any fractionation of C and H isotopes under different conditions; (2) to test the effectiveness of the SPMD-CSIA method for petroleum hydrocarbon contaminants in river water (the Pawtuxet River, RI); and (3) to discuss the implications of study results relative to the future use of SPMD-CSIA technology for assessing sources and fate of aquatic petroleum contaminants.
Experimental Section Materials. Mixed PAH standards (g95% purity, Supelco) used in these tests include naphthalene, acenaphthylene, acenaphthene, fluorene, phenanthrene, anthracene, benzo[c]phenanthrene, benz[a]anthracene, chrysene, benzo[a]pyrene, and dibenzo[a,l]pyrene. Mixed standards of n-alkanes (99.9%, Alltech) used in these tests have a carbon number range from C10 to C36. All solvents used in this work were gas chromatography (GC) grade. Aluminum foil was baked (500 °C) overnight before use. Glassware was baked at 500 °C overnight and rinsed with solvent prior to use. Thin layer chromatograph plates (TLC, LK6F silica gel 60 Å, 20 × 20 cm, layer thickness 250 µm, Waterman) were used for PAH purification. Triolein {1,2,3-tri[(cis)-9-octadecenoyl] glycerol; 95% purity, Sigma Chemical Co.} was further purified by the method of Lebo et al. (11) and contained no detectable PAH and n-alkane residues. LDPE tubing (no. 940, untreated) from Environmental Sampling Technologies, St. Joseph, MO, was used to prepare the SPMDs. The wall thickness of the lot of layflat tubing ranged from 84 to 89 µm, and the width was 2.54 cm. SPMDs. The triolein SPMDs used in this study were made in a clean room at Columbia Environmental Research Center, USGS, Columbia, MO. Layflat LDPE tubing was soaked in hexane beforehand to remove contaminants and was cut into segments 91-cm long. Triolein (1 mL) was pipetted into each segment of LDPE tubing and spread into thin films. The tubing was heat-sealed three times at each end. The triolein SPMDs of standard 450-cm2 design (12) were used for this study. Prior to sampling, the SPMDs were stored at -18 °C, in sealed solvent-cleaned metal cans. Laboratory Simulations. SPMDs were exposed to 5 and 10 µg/L concentrations of PAHs in water, respectively. Oneliter conical flasks with glass stoppers were used for the individual SPMD exposures. All flasks were completely wrapped by double-folded aluminum foil to prevent the photolysis of PAHs. These tests were conducted for 28 days under the following conditions: (1) temperature, 20 °C (airconditioned room temperature) and 4 °C (in refrigerator); (2) turbulence, magnetic stirring rates of 0, 100, and 200 rpm (Figure 1); and (3) blanks (quality control samples), two replicate SPMDs, each in separate flasks, were exposed to the extremes of the treatment regimes (i.e., 4 °C, 0 rpm and 5 µg/L; 20 °C, 200 rpm and 10 µg/L). Two additional SPMDs were used; one was kept frozen sealed in the can as a laboratory control, and the other was exposed to 1 L of unamended deionized water at ≈20 °C as an SPMD accumulation blank. SPMD Sampling in a Petroleum-Contaminated River. The Pawtuxet River in the state of Rhode Island is contaminated by effluent from three sewage treatment plants in Warwick, Cranston, and West Warwick (13). It also receives stormwater runoff that carries high concentrations of pol3690
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 13, 2004
lutants such as oil, silt, and grease from nearby interstate 95 and route 10 and from city streets and residential lawns. Previous studies indicated high levels of petroleum hydrocarbons, including PAHs, present in the sediments of the Pawtuxet Cove, where the Pawtuxet River joins the sea (14). Thus, the Pawtuxet River was selected as the site for a pilotfield study of the SPMD-CSIA approach. At the site of Pawtuxet River mouth (41°45′53′′ N latitude, 71°23′27′′ W longitude), two replicate SPMDs were placed on separate steel racks inside a stainless steel mesh cage (10), and the sampling apparatus was submersed in river water to a depth of 50 cm. A field blank SPMD was exposed to the air adjacent to the deployment site, to account for any airborne contamination during field sampling. The metal cage was designed to protect the SPMDs from sunlight, exclude large-sized debris, and reduce particle deposition on membranes. Although flow velocity/turbulence was not measured inside the deployment apparatus, the structure did not appear to greatly impede flow around the SPMDs. The average linear velocity of water at the deployment site was 50 cm/s when the samples were collected. The SPMDs were exposed to water for 7 days in June, 2003. The water temperature was approximately 18-20 °C at the times of SPMD deployment and retrieval. Following retrieval, the SPMDs were stored at -18 °C in solvent-cleaned sealed cans. A sample of 8 L of river water was also collected in an amber glass container, fitted with a Teflon lined cap, and shipped to the laboratory for LLE and analysis. The results of these analyses were compared with the results obtained from the SPMD samples. Sample Processing and Residue Enrichment. Processing of exposed SPMDs involved the complete removal of any exterior periphytic growth. No biofouling was visible on SPMDs exposed in laboratory studies. A light brown coloration was observed on the field-deployed SPMDs. Fieldexposed SPMDs were mechanically brushed in water, shaken in n-hexane, rinsed with organic free hydrochloride acid (1 M), and dried with acetone and isopropyl alcohol (10). Recovery of target compounds from each SPMD was accomplished by dialysis in hexane (200 mL) for 18 h at 18 °C in the darkness followed by dialysis for another 6 h in fresh hexane (200 mL). Dialysates from each field-deployed SPMD were concentrated by rotary evaporation (30 °C). These extracts were further cleaned on a TLC plate to remove polyethylene waxes and triolein impurities and to separate the hydrocarbons into aliphatic and aromatic fractions. The mobile phase used for these separations was hexane and dichloromethane (DCM; 95/5, v/v). Both TLC fractions were recovered by scraping and were extracted by ethyl acetate. Sample extracts were concentrated by rotary evaporation (30 °C) to 1 mL prior to analysis. The recoveries for spiked alkanes and PAHs through the whole sample preparation process were determined to be >85%. Water samples from the Pawtuxet River were filtered through a 0.45-µm glass-fiber filter. The 8-L filtrate was extracted by partitioning with DCM. Each liter of water was extracted by three 300-mL portions of DCM, and the total 2.4 L of DCM extract was concentrated to 1 mL by rotary evaporation (30 °C) prior to analysis. GC. Quantification of PAHs and n-alkanes was conducted using a Hewlett-Packard 6890+ gas chromatograph, fitted with a 60-m fused silica column (HP-1MS, 0.32-mm i.d., 0.25µm film thickness), a flame ionization detector (FID), a split/ splitless injector, and an HP7683 autosampler. Helium (UHP 5.0 grade) was used as the carrier gas, with a flow rate of 1.7 mL/min. The temperature program for both PAHs and n-alkanes was isothermal at 40 °C for 1 min, followed by heating to 315 °C at 6 °C/min and holding for 30 min. The injector temperature was 300 °C, and it was operated in the splitless mode. The FID temperature was isothermal at 320
FIGURE 1. Accumulated amounts of 11 PAH standards in SPMDs under different concentrations, temperatures and turbulence levels in the laboratory simulation experiments. °C. DCM was used as the solvent to dilute samples for GC analyses. Quantification of PAHs and n-alkanes was accomplished using a five-point external standard curve (10, 20, 50, 100, and 200 ng/µL). Gas Chromatography-Mass Spectrometry (GC-MS). The presence of target PAH analytes and n-alkanes in fielddeployed SPMDs and water samples was confirmed by GCMS (HP 6890, HP 5973A) using the full-scan electron impact mode. Reagent blanks and the SPMD control did not contain measurable levels of any target compounds in this study. Carbon Isotope Analyses of PAHs and n-Alkanes. Carbon isotope analyses of PAHs and n-alkanes were performed using a gas chromatograph combustion-isotope ratio mass spectrometer (Thermo Finnigan). An HP 6890 GC, equipped with an AS 200 autosampler, was connected to a Finnigan MAT Delta+-XL MS via the combustion interface. The temperature program and capillary column used were identical to those used for GC analysis. Helium (UHP 5.5 grade) was used as the carrier gas operating in constant flow mode with a rate of 1.1 mL/min. Injection was performed in the splitless mode to deliver typical sample amounts (∼80 ng per alkane or
∼160 ng per PAH) on the column. Compounds separated by the GC column were converted to CO2 and H2O through the combustion furnace (0.5 mm i.d. × 1.5 mm o.d. × 34 cm) operated at 940 °C and loaded with CuO and Pt wires as the oxidant and catalyst, respectively. Six pulses of CO2 reference gas with known δ13C values (-46.96‰) were injected via the interface to the IRMS, for the determination of δ13C values of sample compounds. Typically, the standard deviation of triplicate analyses is smaller than (0.3‰. Values of δ13C are reported in the usual δ notation relative to the Vienna Pee Dee Belemnite standard. Hydrogen Isotope Analyses of PAHs and n-Alkanes. Hydrogen isotope analyses of PAHs and n-alkanes were performed using a GC-TC-IRMS system (Thermo Finnigan). The GC described previously was connected via a high TC interface to a Finnigan MAT Delta+-XL MS. The temperature program and capillary column were identical to those used for C isotope analysis. Typical sample amounts (∼300 ng per alkane or 600-800 ng per PAH, because less hydrogen is contained in larger PAHs, greater mass is needed to get the same peak size) were delivered on the column. Compounds VOL. 38, NO. 13, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3691
separated by GC were converted to H2 through a pyrolysis furnace (0.5 mm i.d. × 1.5 mm o.d. × 34 cm) operated at 1445 °C. Six pulses of H2 reference gas with known δD values (-273.20‰) were injected via the interface to the IRMS, for the computation of δD values of sample compounds. Typically, the standard deviation of triplicate analyses was smaller than (2‰. Values of δD are reported in the usual δ notation relative to the Vienna Standard Mean Ocean Water standard. Quality Control. The SPMD control, SPMD laboratory accumulation blank, and SPMD field blank were run with the samples on the GC-MS to determine if contaminants were present in the materials or introduced during sampling, extraction, and cleanup. All of the blanks were clean, with only small amounts of phenanthrene being detected in the field blank (
