Sorption of Polycyclic Aromatic Hydrocarbons to Oil Contaminated

Oct 15, 2003 - Process Innovation (TNO-MEP), P.O. Box 342,. 7300 AH Apeldoorn, The Netherlands, and. The Netherlands Organization for Applied Scientif...
0 downloads 0 Views 97KB Size
Environ. Sci. Technol. 2003, 37, 5197-5203

Sorption of Polycyclic Aromatic Hydrocarbons to Oil Contaminated Sediment: Unresolved Complex? M I C H I E L T . O . J O N K E R , * ,† ANJA J. C. SINKE,‡ JOS M. BRILS,§ AND ALBERT A. KOELMANS† Aquatic Ecology and Water Quality Management Group, Department of Environmental Sciences, Wageningen University, P.O. Box 8080, 6700 DD Wageningen, The Netherlands, The Netherlands Organization for Applied Scientific ResearchsEnvironment, Energy, and Process Innovation (TNO-MEP), P.O. Box 342, 7300 AH Apeldoorn, The Netherlands, and The Netherlands Organization for Applied Scientific ResearchsEnvironment, Energy, and Process Innovation (TNO-MEP), P.O. Box 57, 1780 AB Den Helder, The Netherlands

Oil is ubiquitous in aquatic sediments and may affect partitioning and bioavailability of hydrophobic organic chemicals (HOCs). In contrast to other sedimentary hydrophobic carbon phases (natural organic matter, sootlike materials), oil residues have hardly received any attention as far as it concerns effects on HOC sorption. This paper describes experimental work dealing with such effects of oil on polycyclic aromatic hydrocarbon (PAH) sorption to sediments. Three different oils were spiked to a marine sediment in concentrations between 0 and 100 g/kg. Sediment-water distribution coefficients (Kd) for six deuterated PAHs were then determined either directly after spiking the oil or after a semi-natural weathering process in the lab (lasting for more than 2 yr). Resulting Kd values demonstrated sorption-reducing (competitive) effects at relatively low oil concentrations and sorptionenhancing effects at high oil concentrations. The latter effects only occurred above a certain threshold [i.e., ca. 15% (w/ w) of oil on a sedimentary organic carbon basis] marking the oil concentration at which the hydrocarbon mixture presumably starts forming separate phases. Assuming a twodomain (organic carbon + oil) distribution model, oilwater distribution coefficients (Koil) for PAHs were estimated. For fresh oils, log Koil values appeared to be very similar for different types of oils, proportional to log KOW values and indistinguishable from log KOC values. For weathered oils, Koil values were also rather independent of the type of oil, but the affinity of low molecular weight PAHs for weathered oil residues appeared to be extremely high, even higher than values reported for most types of soot. Because affinities of high molecular weight PAHs for oils had not changed upon weathering, sorption of all PAHs * Corresponding author present address: Institute for Risk Assessment Sciences, Utrecht University, P.O. Box 80176, 3508 TD Utrecht, The Netherlands; e-mail: [email protected]; phone: +31 30 2535338; fax: +31 30 2535077. † Wageningen University. ‡ TNO-MEP, Apeldoorn. § TNO-MEP, Den Helder. 10.1021/es0300564 CCC: $25.00 Published on Web 10/15/2003

 2003 American Chemical Society

studied (comprising a log KOW range of 4.6-6.9) to the weathered oil residues appeared to be more or less constant (averaged log Koil ) 7.0 ( 0.24). These results demonstrate that it is crucial to take the presence of oil and its weathering state into account when assessing the actual fate of PAHs in aquatic environments.

Introduction Sorption of hydrophobic organic chemicals (HOCs) to sediments is usually considered and modeled as a simple partitioning process between water and sedimentary natural organic matter (NOM). NOM is a hydrophobic phase, which results from the decomposition of biomass (e.g., algae, plants, animal remains) and practically is present in all sediments. However, as advanced by Luthy et al. (1), sediments are often contaminated with “anthropogenic carbon phases” such as combustion residues (soot or black carbon) and so-called nonaqueous phase liquids (NAPLs) formed by oil or tar residues. These carbon matrixes can serve as additional sorption pools for HOCs, causing overall sorption behavior that may deviate from behavior predicted on the basis of partitioning into NOM only. Recently, it was demonstrated that, in particular, HOCs with planar molecular structures can have extremely high affinities for combustion residues and related materials such as coal and charcoal (2). Therefore, sorption of HOCs to sediments containing these additional sorbents is much stronger than anticipated on the basis of organic matter partitioning estimates (3, 4). Because of the importance of “hard carbon materials” in controlling the actual fate of HOCs in the aquatic environment, several researchers have recently focused on the effects of these materials on both sorption and bioaccumulation processes. In contrast, possible effects of sedimentary oil phases on HOC sorption have hardly received any attention, even though these phases are widespread in the aquatic environment as a result of the enormous worldwide use of oil (products). Research in the past has focused on soils using either fresh oil spikes or field-contaminated samples. Partitioning of HOCs into fresh oils was found to be approximately equal to partitioning into octanol (5-7), which suggests that fresh oils could act as additional sorption phases with sorption properties similar to natural organic matter. However, fresh oils were shown to cause so-called “cosolvent effects” (i.e., enhancement of aqueous solubilities of HOCs by water accommodated oil fractions), lowering apparent soil-water distribution coefficients (8, 9). In contrast, in soil samples from the field, weathered oil residues did appear to behave as an additional sorption phase (10-15), reducing aqueous concentrations (13, 14) and thus lowering bioavailability and biodegradation (11). This is in line with expectations because the light, water-soluble oil fractions responsible for solubility enhancement of HOCs usually get washed away or degrade during natural weathering of oil (16), leaving only the nonsoluble so-called “unresolved complex mixture” (UCM) (17). This oil residue can be characterized as a highly persistent and complex mixture composed of thousands of predominantly high molecular weight hydrocarbons such as branched alkanes, (alkylated) cycloalkanes, and polycyclic aromatic hydrocarbons (PAHs), naphtenoaromatics, heteroatomic aromatics, etc. (17) having an extremely hydrophobic nature. Affinities of HOCs for weathered oil phases have been reported to be somewhat higher (up to ca. 35 times for a hydraulic PCB oil) than for natural soil organic carbon (12-15). Although not experimentally demonstrated yet, oil has also been suggested to act as (a) “plasticizer”, VOL. 37, NO. 22, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5197

serving as a swelling agent for natural organic matter, causing an increase in accessibility of sorption sites for HOCs, and thus increasing sorption (10, 18) and (b) “competitor”, occupying potential sorption sites for HOCs, thus on the contrary lowering overall sorption of these compounds (19). From the above, it will be clear that, analogous to soils, the presence of residual oil phases in sediments may affect sorption of HOCs in different ways; the overall effect probably depending on the type (different types of oils have different physicochemical characteristics), concentration, and age or weathering state of the oil. So far, however, no study has systematically researched this complex behavior. Therefore, in the present study, we investigated sorption of PAHs to sediment contaminated with different types and concentrations of (weathered) oils. The main goals were to quantify (a) concentration-dependent effects of oil on PAH sorption to sediment and (b) affinities of PAHs for oil. Resulting data were interpreted in terms of different types of oils and PAHs as well as the weathering state of the oils.

Experimental Design Sorption of PAHs to (oil-contaminated) sediment was quantified by measuring sediment-water distribution coefficients. These coefficients were determined according to an equilibrium solid-phase extraction approach using polyoxymethylene (POM-SPE) (2, 20). A big advantage of polyoxymethylene (POM) is that it is fully resistant to oil. This makes the POM-SPE method highly suitable for the current setup. Distribution coefficients were measured for six added deuterated PAHs (phenanthrene-d10, anthracene-d10, fluoranthene-d10, benz[a]anthracene-d12, benzo[k]fluoranthened12, and benzo[g,h,i]perylene-d12) spiked into systems containing synthetic seawater and samples of a marine sediment that had been contaminated artificially with three different oils, all at nine concentration levels (between 0 and 100 g/kg). These oil concentrations were chosen to represent a fieldrealistic contamination range. The implication of this range, however, is that cosolvent effects could not be detected using the current setup because aqueous oil concentrations [always much less than 0.02% (v/v)] were too low to produce such effects. Sediment-water distribution of PAHs was determined directly either after the addition of fresh oils or after a seminatural weathering period. During this period, which lasted for more than 2 yr, natural weathering was mimicked in the laboratory by incubating oil-contaminated sediments under aerobic, flooded conditions at 20 °C and frequent replacement of the water column (washing away dissolved oil fractions). The three oils selected for this study were (a) a gasoils Distillate Marine grade A (DMA) [used as fuel in midsized to larger ships such as coasters, large cutters, and large inland vessels (21)]; (b) a light crude oilsArabian Crude Light (ACL); and (c) a waste lubricating oil (BIL). The last-mentioned oil was collected as the oil layer from so-called “bilge water” from yachts and may in fact be a mixture of several types of lubricating oils and some grease. The three oils above were chosen because they have different physicochemical properties (22) and because they are all widely used. Accidental and deliberate spills during use, transport, transfer, and cleaning of tanks and holds may cause the presence of these oils in particular in sediments from harbors and shipping routes. Note that gasoil (as represented by DMA) has been identified as the major source of oil contamination in Dutch harbor sediments (21).

Material and Methods Chemicals. Deuterated PAHs (purity g98%) were obtained from Cambridge Isotope Laboratories (CIL), Andover, MA. 2-Methylchrysene (purity 99.2%) was supplied by The Community Bureau of Reference (BCR), Geel, Belgium. Other 5198

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 22, 2003

chemicals used were as follows: hexane and acetone (picograde; Promochem, Wesel, Germany); methanol (HPLC gradient grade; Mallinckrodt Baker, Deventer, The Netherlands); acetonitrile (HPLC grade; Lab-Scan, Dublin, Ireland); heptane (pestanal grade; Riedel-deHae¨n, Seelze, Germany); tetracontane (98%; Sigma-Aldrich); sodium azide (99%; Sigma-Aldrich); sodium chloride, calcium chloride (dihydrate), magnesium chloride (hexahydrate), sodium sulfate, and potassium sulfate (all analytical grade; Merck, Darmstadt, Germany); water (Nanopure; Barnstead, Dubuque, IA); and aluminum oxide-Super I (ICN Biomedicals, Eschwege, Germany). All chemicals were used as received except aluminum oxide, which was deactivated with 10% (w/w) of Nanopure water prior to use. Oils. DMA gasoil was supplied by Gulf Oil, Nigtevecht, The Netherlands. ACL oil from Saudi Arabia was donated by the Dutch Shell refinery at Pernis, The Netherlands. BIL was collected at the yacht basin of Wageningen by scooping the floating oil layer from a bilge water collection depot. The oils were homogenized by shaking and used without further modifications. Sediment and Spiking Procedure. Sediment was sampled from a tidal flat (Oesterput) in the Eastern Scheldt Estuary, The Netherlands. This sediment was selected because of its unpolluted nature (oil concentration below detection limit) and its fine composition, which resembles that of Dutch harbor sediments (21). The upper 5 cm was collected during the summer of 1999 (sediment temperature on site: 18 °C). It was passed through a 500-µm sieve, mechanically homogenized, and stored at 20 °C until use (10 d). Organic carbon content was determined (see Chemical Analyses section) to be 0.82 ( 0.007% (n ) 9). Subsamples of this sediment were spiked with oils as follows: For each type of oil, nine brown 500-mL jars were filled with 175 g of wet sediment (100 g dw) and 300 mL of sand bed-filtered Eastern Scheldt water (salinity ca. 32 ‰). While being stirred intensively (700 rpm) using metal magnetic stirrers, oil was added drop by drop in such amounts that oil-sediment ratios of 0, 0.03, 0.1, 0.3, 1, 3, 10, 30, and 100 g/kg were created. Then, the jars were closed with aluminum-coated tops and stirred at 300 rpm for 3 d. Following this period, the magnetic stirrers were removed, and the resulting 27 systems were subjected to a weathering procedure (see below). In addition to this set of sedimentoil systems, a duplicate set was produced for which, however, the highest oil concentration was omitted. The resulting duplicate set of 24 systems was not allowed to weather, but subsamples were used directly for the determination of distribution coefficients. Weathering Procedure. Weathering of the first set of oilspiked sediments was allowed to take place during 27 months in a dark air-conditioned room (20 ( 0.2 °C). Jar tops were replaced by tops containing a hole (L 12 mm) covered with a glass fiber filter (0.5-1 µm pore size). This way, air exchange was possible, but dirt entry was prevented. All jars were then shaken on a rotation shaker for 2.5 months. Following this period, sediments were allowed to settle during 4 d, after which time the supernatant was replaced by fresh sand bedfiltered Eastern Scheldt water of 20 °C, and the filters were renewed. From then on, this washing procedure was repeated every 2 months. Between times, jars were constantly shaken on a rotation shaker to facilitate gas-phase exchange. Determination of Distribution Coefficients. After a 2-d settling period (following the 3-d homogenization period of fresh oil-spiked sediments and the final 2-month washing period of weathered sediments), supernatants of all 500-mL jars were discarded and the sediments were homogenized. Then, sediment-water distribution coefficients were determined using the POM-SPE method (20). Sediment samples (ca. 2 g dw) were weighted into 300-mL all-glass covered

bottles together with precleaned POM strips (Vink Kunststoffen BV, Didam, The Netherlands) of 1.0 g. The bottles were filled with synthetic seawater (containing 24.5 g of sodium chloride, 9.8 g of magnesium chloride, 0.53 g of calcium chloride, 3.22 g of sodium sulfate, 0.85 g of potassium sulfate (23), and 25 mg of sodium azide/L of Nanopure water), spiked with 15-50 µL of a cocktail solution containing six deuterated PAHs in acetone (each at 30 mg/L) and equilibrated during 4 weeks by shaking them horizontally at 100 rpm and 20 (( 0.2) °C. Handling protocol for equilibrated POM strips and other details on POM-SPE are presented in previous papers (2, 20). In addition, POM-seawater partition coefficients for PAHs (KPOM; n ) 10) were determined in synthetic seawater according to methods presented in ref 20. Chemical Analyses. Oil concentrations in weathered sediments were determined as follows: Subsamples were homogenized with Na2SO4 (kiln-fired for 3 h at 550 °C) in a mortar. The dried samples were transferred to pre-extracted glass fiber extraction timbles and Soxhlet-extracted with hexane/acetone (3:1) for 16 h. The resulting extracts were concentrated and cleaned up using Al2O3 columns as described previously (20, 24). Finally, extracts were solventexchanged to heptane, and tetracontane (C40) was added as the internal standard. Extraction of POM strips and cleanup of the resulting extracts was performed according to procedures reported elsewhere (2, 20). PAHs were analyzed on a HP 1100 HPLC (for details see refs 20 and 24), and total oil concentrations (operationally defined as C10-C40) were quantified by oncolumn injection of 0.5 µL of the extracts in a Chrompack CP-9000 GC equipped with a CP-SimDist fused silica capillary column (10 m, df 0.1 µm, i.d. 0.32 mm) and a flame ionization detector (FID). Total organic carbon (TOC) contents of all sediments were determined on an EA 1110 CHN elemental analyzer after descaling freeze-dried subsamples with HCl (see ref 2 for more details). Quality assurance precautions for PAH, oil, and TOC analyses (including blanks, recoveries, and light reduction) were identical to those presented in previous papers (2, 20).

Results and Discussion Effects of Oil on PAH Sorption. POM-SPE experiments resulted in almost 300 sediment-water distribution coefficients (Kd) that were plotted against corresponding oil concentrations in the sediments (Coil). In Figure 1, some examples of the resulting log-log curves are presented. For experiments with fresh oil (Figure 1a), nominal oil concentrations were used (NaN3 had been added to prevent biodegradation), but for weathered sediments, actual oil concentrations were measured and used to construct Figure 1b. Latter concentrations measured only 13-50% of the initial (spiked) concentrations, clearly pointing to degradation and/ or loss of oil from the systems. Log Kd-log Coil plots show that at low oil concentrations Kd values generally are relatively independent of Coil, but at higher concentrations, Kd values increase gradually. In many cases, however, also a decrease in Kd values is observed at oil concentrations of 100-1000 mg of oil/kg of sediment. This phenomenon is most pronounced for the three-ring PAHs (phenanthrene and anthracene). Because standard deviations (SDs) of POMSPE measurements are very small (2, 20), these Kd decreases most probably are not caused by an artifact. Triplicate determinations at Coil ) 0 and within-system duplicate determinations for oil-containing sediments confirmed the high reproducibility of the method for the current systems (averaged SDs of 0.04 and 0.03 log unit, respectively). In addition, the analytical error for the present experiments was even lower (i.e., averaged value of 0.05 log unit) than the error reported in ref 20 because SDs for POM-water distribution coefficients (KPOM) appeared to be much lower

FIGURE 1. Logarithmic sediment-water distribution coefficients (log Kd) for different PAH/oil combinations [squares for Phen-d10 (BIL), diamonds for Flu-d10 (BIL), triangles for BaA-d12 (ACL), and circles for BghiPe-d12 (DMA)] as a function of the logarithmic oil concentration (log Coil) in (a) unweathered (freshly oil-spiked) sediments and (b) weathered sediments. Dashed lines represent modeled sorption behavior (see Results and Discussion section) according to Kd ) fOCKOC [with fOC values measured as total organic carbon by elemental analysis (includes oil) and KOC data from Table 1]. Dotted lines represents sorption modeled using Kd ) fOCKOC + foilKoil (eq 1; see text for further explanation). in saltwater as compared to freshwater (see Table 1). It should be noted, however, that log Kd-log Coil curves for weathered sediments are much less smoother than those for freshly spiked sediments (see Figure 1). Most certainly, this is not due to analytical inaccuracy (see above) but to a larger heterogeneity of the weathered sediments that must have resulted from differences in degree of weathering among systems. As will be discussed later on, differences in weathering state will have repercussions on the extent of sorption. Considering the above and the fact that a Kd decrease at log Coil ) 2-3 is observed for most oil/PAH combinations, we conclude that this phenomenon has a nonanalytical mechanistic basis and should be interpreted as the result of competitive sorption effects. Probably, oil is able to occupy sorption sites in sedimentary organic matter, which as a consequence are no longer available for deuterated PAHs (present at only 1-4 mg/kg in these systems). Some of these sites may be specific ones to which only a fraction of the oil (i.e., specific oil constituents) can sorb since the oil VOL. 37, NO. 22, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5199

TABLE 1. Logarithmic Organic Carbon-Water (log KOC) and Oil-Water (log Koil) Distribution Coefficients (L/kg) for Unweathered and Weathered Sediments unweathered sediments compda Phen-d10 Ant-d10 Flu-d10 BaA-d12 BkF-d12 BghiPe-d12

log KOWb

log KPOMc

log KOC

4.6 4.6 5.2 5.9 6.2 6.9

3.30 ( 0.03 3.55 ( 0.04 3.80 ( 0.02 4.58 ( 0.02 4.92 ( 0.02 4.69 ( 0.04

4.68 ( 0.05 5.71 ( 0.04 5.34 ( 0.04 6.28 ( 0.04 6.84 ( 0.04 7.23 ( 0.03

weathered sediments

log Koild

log Koild

DMA

ACL

BIL

log KOC

DMA

ACL

BIL

5.15 ( 0.05 5.11 ( 0.06 6.39 ( 0.02 6.97 ( 0.06 7.01 ( 0.05

nd 5.28 ( 0.06 5.43 ( 0.09 6.37 ( 0.08 7.01 ( 0.05 7.16 ( 0.02

nd 5.11 ( 0.09 5.42 ( 0.07 6.29 ( 0.03 6.81 ( 0.07 6.98 ( 0.12

5.19 ( 0.05 5.49 ( 0.04 5.63 ( 0.04 6.55 ( 0.03 7.09 ( 0.04 7.40 ( 0.06

6.86 ( 0.11 6.92 ( 0.02 6.51 ( 0.01 7.24 ( 0.08 6.96 ( 0.01 7.10 ( 0.07

6.90 ( 0.10 6.96 ( 0.24 7.07 ( 0.16 7.41 ( 0.05 7.36 ( 0.11 7.38 ( 0.10

6.87 ( 0.10 6.94 ( 0.22 6.74 ( 0.12 7.22 ( 0.29 6.89 ( 0.07 6.86 ( 0.10

nde

a Explanation of compound abbreviations: Phen, phenanthrene; Ant, anthracene; Flu, fluoranthene; BaA, benz[a]anthracene; BkF, benzo[k]fluoranthene; BghiPe, benzo[g,h,i]perylene. b Log KOW values were adopted from ref 2. c POM-water distribution coefficients are needed to calculate sediment-water distribution, see ref 20. d Standard deviations relate to n ) 2, 3, or 4. e Not determined [phenanthrene undetectable due to coelution with (light) oil constituents].

content is at least 25-1000 times as high as the deuterated PAH concentration. Competitive effects may however also be nonspecific, not involving specific sites, but only occurring because large molecules from the oil lower the accessibility of the organic matrix by blocking pores or covering surfaces. The fact is that these effects cause overall sorption of PAHs to reduce with a factor of up to approximately 3 (see Figure 1). The existence of competitive effects in the current systems implies that plasticizer effects (18) of oil either (a) do not exist or occur, (b) are overruled by competitive effects, or (c) only appear at higher concentrations, that is, above 1000 mg/kg, initiating the curve-up of log Kd-log Coil plots (see Figure 1). If plasticizer effects do exist, it is questionable in which concentration range they will occur. For polar solvents, it has been hypothesized that competitive and plasticizer or “solvation” effects can co-occur, but at low concentrations, solvation effects will dominate, whereas at high concentrations competitive effects are favorable (25). However, oil is not a polar solvent, and solvation of organic matter will probably take place at much higher concentrations because these solvation effects by hydrophobic oil constituents may act according to another mechanism. After all, oil cannot hydrate polar moieties of organic matter like polar solvents (25) creating new sorption sites for PAHs at, for example, disrupted H-bonding sites. Instead, oil will sorb at the same hydrophobic sites as PAHs do, and increasing the sorption capacity of the organic matter can only be accomplished by exposing hydrophobic interior sorption sites by swelling. Swelling may only occur at high oil concentrations when the hydrocarbon mixture starts to act as a solvent, swelling the organic matter like a sponge or turning it into a “gel” (26). However, oil-induced swelling may be strongly limited because the process is inversely related to the molar volume of the “solvent” (24, 27), which in case of (in particular weathered) oil is very large. At high oil concentrations, oil will also start to form separate phases (14) (i.e., separate droplets or films attached to sediment particles). These phases will supply additional sorption pools for PAHs and therefore will also cause overall sorption to increase. Hence, the current setup cannot distinguish between the two sorption-enhancing effects (plasticizer and separate sorption phase effects). However, as stated above, in particular for weathered oils plasticizer effects are unlikely considering the oil’s large molar volume and the branched structure of its molecules. Moreover, the existence of separate phases in the present experiments was confirmed by UV microscopy (365 nm), which revealed the presence of oil droplets for all oil types at concentrations roughly above 3000 mg/kg (data not shown). Therefore, for the sediment used in the current experiments, the concentration above which oils form a significant amount of separate phases (the critical separate phase concentration; CSPC) lies between 1000 and 3000 mg/ kg. The exact concentration cannot be deduced from the 5200

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 22, 2003

sorption data in Figure 1 since it may be obscured by both plasticizer and competitive sorption effects that will also exist at very high oil concentrations. From sorption experiments with polychlorinated biphenyls (PCBs) and hydraulic oilcontaminated soils, Sun and Boyd (14) deduced a CSPC of ca. 1000 mg of oil/kg of soil. In addition, oil degradation experiments with an oil-contaminated soil performed by De Jonge et al. (28) pointed to a CSPC of ca. 4000 mg/kg. Although these values are close to our current range, it should be stressed that the CSPC is dependent on the organic carbon content (14) and possibly also on the type of organic carbon. Separate phases will only appear once the organic carbon is “saturated” with oil, and the amount of oil needed for this saturation varies with the amount (and type) of organic carbon. Therefore, CSPCs should be normalized to the sediment’s organic carbon content. For the above-mentioned values, normalization results in thresholds of 11 and 15% of oil in soil organic carbon (w/w; for refs 14 and 28, respectively) and 10-30% of oil in sedimentary organic carbon for the present experiments. Remarkably, these percentages are in close agreement, and it seems plausible to suppose that the CSPC for oils approximates 15% on an organic carbon weight basis. In other words, below 15% of oil in organic carbon, oil will be sorbed into the organic matrix, and above this percentage oil will start to behave as an additional sorption phase, (strongly) increasing overall sorption of HOCs. In addition, above the CSPC, both microbial degradation (28) and toxicity of the oil itself may increase; the latter probably not only caused by an increase in “chemical availability” but also in “physical availability” (i.e., “smothering effects”; 29). Oil-Water Distribution. When calculating overall sorption of HOCs to field sediments containing residual oil, it is important to know the exact oil-water distribution behavior of the chemicals, in particular for sediments containing oil concentrations above the CSPC (see below). For fresh oils, oil-water distribution coefficients (Koil) can be measured directly using batch equilibrium experiments (5-7) or, preferably, using the slow-stirring method (30). However, for weathered oils, measuring these values is problematic due to difficulties in obtaining pure weathered oil phases. Oil has artificially been weathered for use in, for example, toxicological studies by heating fresh oil (31), and weathered oil has been extracted from field samples using organic solvents (14). It will be clear that both methods have important drawbacks and that the use of weathered oils obtained by these methods therefore results in values that should only be considered as a first rough approximation. The current experiments, however, offer the possibility to calculate Koil values for PAHs since above the CSPC overall sorption can be described by (13, 14):

Kd ) fOCKOC + foilKoil

(1)

Using this equation, Koil values can be calculated for the three (sometimes two or four) highest oil concentration data points on the basis of (a) experimentally determined values for Kd, fOC (at Coil ) 0), and foil, and (b) KOC values deduced from Kd values measured just below the CSPC (using fOC values determined at Coil ) 0). The latter is done to account for competitive and/or plasticizer effects that may also be present at the highest oil concentrations. It should be mentioned though that using either these KOC values or ones determined at Coil ) 0 results in highly comparable log Koil values for both fresh and weathered oils, with differences between both calculation methods always being less than 0.2 log unit (data not shown). The resulting two-four individual sorbate specific values were averaged, and the results are presented in Table 1 together with accompanying SDs. In addition, KOC values () Kd/fOC for Coil ) 0) for all chemicals are given in this table. Note that Koil values for phenanthrene-d10 sorption to fresh oil are missing. This is due to chromatographic problems caused by oil constituents from fresh oil, which made it impossible to quantify phenanthrene-d10 in this case. Many interesting observations can be made from Table 1. First of all, Koil values for different types of fresh oils appear to be almost identical (not significantly different based on t-tests). This was against expectations because the oils used have different chemical compositions (22) and physical characteristics (like color and viscosity). However, Ortiz et al. (7) also reported similar values for distribution of three low molecular weight PAHs over different types of fresh NAPLs [oil, grease, paraffin, petrolatum (i.e., Vaseline)] and water. Apparently, viscosity and detailed composition do not influence partitioning of PAHs into fresh oils, and the extent of this process solely depends on the chemicals’ hydrophobicities. The latter is illustrated by the fact that log Koil values for fresh oils are proportional to log KOW values (see Table 1 and Figure 2a). Log Koil values for fresh oils, however, appear to be somewhat (though not significantly) higher than log KOW values. Furthermore, they are statistically indistinguishable (t-test) from both log KOC and log KOM values [i.e., organic matter (OM)-water distribution coefficients, calculated assuming that organic matter contains 58% of carbon; data not shown] for unweathered sediments. Logically, these data point to highly comparable hydrophobicities of fresh oils and sedimentary organic matter and a somewhat lower hydrophobicity of octanol as compared to these oils. The observed small difference between octanol and oils is consistent with reports in the literature (5-7) and can be explained by the somewhat more polar nature of octanol due to its hydroxyl group. The fact that log KOC values are slightly higher than log KOW values (see Table 1), however, is in contrast with the organic matter partitioning paradigm (32). This observation may point to the presence of adsorption domains in the sediment (i.e., “glassy” organic matter or soot(-like) particles), but it should also be stressed that KOC values were determined in saltwater using a technique that does not suffer from artifacts related to the presence of DOC. This may result in higher (true) distribution coefficients as compared to literature KOC values, which apply to freshwater systems and which are usually measured with classical batch equilibrium methods (20, 33). For weathered systems, oil-water distribution of PAHs also appears to be relatively independent of the type of oil (see Table 1), although differences among Koil values for different oils are larger than for fresh oils. Apparently, different oils are weathered similar, leading to matrixes with comparable sorption characteristics for PAHs. These ultimate weathered matrixes show a most remarkable phenomenon: As compared to unweathered conditions, log Koil values for low molecular weight (three- or four-ring) PAHs have increased spectacularly. For higher molecular weight PAHs,

FIGURE 2. Effects of weathering on (a) oil-water distribution (log Koil) and (b) natural organic carbon-water distribution (log KOC) of PAHs. Distribution coefficients are plotted as a function of octanolwater partition coefficients (log KOW). Open symbols are for unweathered oil/organic carbon; solid symbols are for weathered oil/organic carbon. Dotted and dashed lines represent regression curves for unweathered and weathered systems, respectively. Solid lines are the 1:1 relationships. however, changes are negligible. As a result, the slope of the log Koil-log KOW plot has decreased from almost unity (0.99) to only 0.13 upon weathering (see Figure 2a). Although the latter slope does significantly differ from zero (F-test), sorption of the presently studied six PAHs to these weathered oil residues seems to be rather independent of sorbate characteristics and roughly quantifiable by a constant (averaged) log Koil of 7.0 ( 0.24. Furthermore, the results from Table 1 demonstrate that the absolute affinity of low molecular weight PAHs for weathered oil is extremely high. On a mass basis, affinities of the first three PAHs for weathered oils are even higher than for most soot-like materials (2), which are now generally being considered as “super sorbents” (34). Analogous to soot, this extreme sorption to weathered oil probably has a “steric” origin. As a result of the weathering process, the amount of polar functional groups will be minimal, which creates a hyper-hydrophobic environment that does not bear any resemblance to octanol and in which sorption of PAH molecules is most favorable. However, apparently only the low molecular weight PAHs can gain by this, probably because the large(r) molecules of high molecular weight PAHs are unable to penetrate the extremely dense “asphaltenic” matrix of branched hydrocarbons that VOL. 37, NO. 22, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5201

make up the unresolved complex mixture. Consequently, sorption becomes independent of hydrophobicity, an effect that is illustrated in Figure 2a and that previously has also been observed for PAH sorption to some soot-like sorbents (in particular (char)coal and activated carbon) (2). Although so much is certain that the chemical composition of the oils changed upon weathering [lower boiling-point fractions and resolved (linear) alkanes disappeared whereas higher boilingpoint fractions relatively increased in abundance (22)], to our knowledge there is no literature available on “pore sizes” in weathered oil residues, which makes the above “size exclusion explanation” strictly hypothetical. Sun and Boyd (14) have reported residual (hydraulic PCB) oil phases from field soils to be 38 times as effective as soil organic carbon in sorbing 2-chlorobiphenyl. However, their KOC value for this chemical was taken from the literature and therefore does not necessarily represent the site-specific value. A comparison of the authors’ data with KOW therefore is more illustrative and yields a factor of only 3.5 (i.e., Koil ) 3.5KOW). The present data show values for sediment that are much more extreme: Sorption of phenanthrene-d10 to weathered oil is 50 times as strong as sorption to (weathered) sedimentary organic carbon and even 200 times as high as partitioning into octanol. Finally, “weathering effects” are also observed for the KOC data in Table 1. After approximately 2 yr at 20 °C, KOC values for PAHs have increased significantly (paired t-test) by 0.20.5 log unit (with the exception of anthracene’s KOC that decreased by 0.2 log unit and for which we have no explanation). The resulting shift of the log KOC-log KOW curve is presented in Figure 2b. These results indicate that despite the increase and unlike the situation for oils, KOC values for weathered sediments are still proportional to KOW. Most probably this implies that the polarity and/or structure of organic matter have not changed that drastically as the polarity and/or structure of oils. Possibly, oils are degraded differently and/or much easier by the existing sedimentary bacteria as compared to natural sedimentary organic matter. Because sediments were “washed” during the 2-yr weathering period, it is also possible that the sedimentary organic matter has not been degraded at all, but that its decrease in polarity is due to extraction of the more polar constituents by water. Because the present data set concerns added chemicals equilibrated with sediments for only 4 weeks, it demonstrates that (a) “aging” of contaminants in the field (i.e., increased sorption with increased contact time) may not only be explained by slow diffusion and subsequent sequestration in the organic matter or mineral matrix (35) but also by an increase in sorption affinity for organic matter (as suggested before in refs 36 and 37) or oil in time, caused by compositional changes of these sorption domains and that (b) it does not take decades of contact time (i.e., “aging”) to obtain extremely strong sorption; a conclusion previously also drawn on the basis of sorption data for soot-like materials (2, 3). Implications for Estimating Sorption to Field Sediments. Sorption of HOCs to sediment is generally quantified by the chemicals’ KOC. This organic carbon-normalized sedimentwater distribution coefficient is one of the main parameters in the equilibrium partitioning theory (EPT), which forms the basis of most integrated HOC fate and effect models (38). KOC values are still being used for risk assessment purposes on the assumption that these compound specific values are constant, even though research during the past two decades has identified several factors that contradict a constant KOC approach (3). Our current results demonstrate that also the presence of residual oil may have important implications for the applicability of EPT. In Figure 3, KOC values (calculated separately for all sediments by dividing Kd values as depicted in Figure 1 by TOC fractions detected using elemental analysis for the 5202

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 22, 2003

FIGURE 3. Logarithmic organic carbon-normalized distribution coefficients (log KOC) for different PAH/oil combinations [squares for Phen-d10 (BIL), diamonds for Flu-d10 (BIL), triangles for BaA-d12 (ACL), and circles for BghiPe-d12 (DMA)] as a function of logarithmic oil concentration (log Coil) in (a) unweathered sediments and (b) weathered sediments. Dotted lines indicate fixed compound specific KOC values (determined at Coil ) 0). sediment of concern) are plotted as a function of the oil concentration in sediment. This figure shows that a constant KOC is reasonably acceptable for sediments containing fresh oil (Figure 3a) and for higher molecular weight PAHs in sediments with weathered oil (Figure 3b). However, for low molecular weight PAHs in weathered sediments, KOC values are strongly dependent on Coil (Figure 3b). Apparently, normalization to total organic carbon as detected by elemental analysis cannot account for the observed enhanced sediment-water distribution. For the oil concentration range applied in this study, KOC values vary with as much as almost 2 log units for phenanthrene-d10. Therefore, for sediments with high, weathered oil contents, KOC values may severely be underestimated when neglecting the presence of oil. Furthermore, for both fresh and weathered systems, normalizing to total organic carbon cannot account for competitive sorption effects; therefore, within particular oil concentration ranges, true KOC values are lower than anticipated. Finally, for sediments contaminated with fresh oils, small but significant increases in KOC values at the highest oil concentrations are observed (Figure 3a). Since Koil values

for fresh oils are statistically indistinguishable from KOC values, this observation is rather unexpected. However, it should be recalled that Koil values are presented on a mass basis, in contrast to KOC values, which are based on carbon content. Because oil is composed of more than carbon alone, this might explain the upward curvature in Figure 3a. On the other hand, the curvature might also (partly) result from an artifact. Total organic carbon contents at high fresh oil concentrations may have been underestimated due to loss of volatile oil constituents during freeze-drying and drying at 60 °C as part of the descaling procedure. For field sediments, which usually will contain weathered oil, this possible artifact most likely will be of no importance because volatile oil fractions will have disappeared as a result of the weathering process (16). Because KOC values determined in this study (see Table 1) are significantly higher than literature KOC values used in fate models, application of the latter ones to estimate overall sediment-water distribution of PAHs in the presently used sediments would lead to a considerable underestimation. However, since organic carbon-normalized distribution coefficients for sediments containing oil are dependent on residual oil content, models using the KOC values from Table 1 will also fail to correctly describe sorption, in particular for sediments contaminated with weathered oil. This is illustrated in Figure 1 by dashed lines, representing Kd ) fOCKOC, with fOC values measured as total organic carbon by elemental analysis (including oil-carbon), and KOC values from Table 1. Addition of a term representing sorption to separate oil phases (according to eq 1, with fOC values measured at Coil ) 0, and KOC values derived from Kd values just before the CSPCssee previous section) results in much better fits, although the improvement logically is restricted to higher oil concentrations and competitive sorption effects still are not accounted for (see dotted lines in Figure 1). In summary, the applicability of EPT in predicting sorption of PAHs to oil-contaminated sediments largely depends on the chemical, the right choice of KOC values, the weathering state of oil, and the sedimentary oil concentration. The oil concentration above which the hydrocarbon mixture forms a separate phase (CSPC) and above which significant underestimation of overall sorption will occur when neglecting the presence of oil is for its part dependent on system characteristics (i.e., sediment organic carbon content). Fortunately, affinities of PAHs for different types of oils seem to be highly comparable, which somewhat simplifies the complex situation. Fact is, analogous to soot, that neglecting the presence of oil during risk assessment of PAH-contaminated sediments may confront water managers with unnecessarily high remediation cost.

Acknowledgments The research described in this paper was supported by and performed within the Research Center on Soil, Sediment, and Groundwater Management and Remediation, WUR/ TNO. We would like to thank Gulf Oil (Nigtevecht, The Netherlands) and the Dutch Shell refinery (Pernis, The Netherlands) for donating DMA and Arabian Crude oil, respectively, and H. Zweers for performing the GC-FID analyses.

Literature Cited (1) Luthy, R. G.; et al. Environ. Sci. Technol. 1997, 31, 3341. (2) Jonker, M. T. O.; Koelmans, A. A. Environ. Sci. Technol. 2002, 36, 3725.

(3) Jonker, M. T. O.; Smedes, F. Environ. Sci. Technol. 2000, 34, 1620. (4) Bucheli, T. D.; Gustafsson, O. Environ. Toxicol. Chem. 2001, 20, 1450. (5) Chen, C. S.-H.; Delfino, J. J.; Rao, P. S. C. Chemosphere 1994, 28, 1385. (6) Lee, L. S.; Hagwall, M.; Delfino, J. J.; Rao, P. S. C. Environ. Sci. Technol. 1992, 26, 2104. (7) Ortiz, E.; Kraatz, M.; Luthy, R. G. Environ. Sci. Technol. 1999, 33, 235. (8) Forst, C.; Schafer, K.; Andl, A.; Stieglitz, L. Chemosphere 1994, 29, 2157. (9) Walter, T.; Ederer, H. J.; Forst, C.; Stieglitz, L. Chemosphere 2000, 41, 387. (10) Zemanek, M. G.; Pollard, S. T.; Kenefick, S. L.; Hrudey, S. E. Environ. Pollut. 1997, 98, 239. (11) Zwiernik, M. J.; Quensen, J. F.; Boyd, S. A. Environ. Sci. Technol. 1999, 33, 3572. (12) Rutherford, P. M.; Gray, M. R.; Dudas, M. J. Environ. Sci. Technol. 1997, 31, 2515. (13) Boyd, S. A.; Sun, S. Environ. Sci. Technol. 1990, 24, 142. (14) Sun, S.; Boyd, S. A. J. Environ. Qual. 1991, 20, 557. (15) Ghosh, U.; Weber, A. S.; Jensen, J. N.; Smith, J. R. Environ. Sci. Technol. 2000, 34, 2542. (16) Gearing, P. J.; Gearing, J. N.; Pruell, R. J.; Wade, T. L.; Quinn, J. G. Environ. Sci. Technol. 1980, 14, 1129. (17) Frysinger, G. S.; Gaines, R. B.; Xu, L.; Reddy, C. M. Environ. Sci. Technol. 2003, 37, 1653. (18) Van Steenwijk, J. M.; Cornelissen, G.; Rigterink, H.; Freriks, I.; Van Noort, P. C. M. In Abstracts of the 214th American Chemical Society Meeting, Las Vegas, NV, September 7-11, 1997; pp 201204. (19) Cornelissen, G.; Van der Pal, M.; Van Noort, P. C. M.; Govers, H. A. J. Chemosphere 1999, 39, 1971. (20) Jonker, M. T. O.; Koelmans, A. A. Environ. Sci. Technol. 2001, 35, 3742. (21) Brils, J. M.; Huwer, S. L.; Kater, B. J.; Schout, P. G.; Harmsen, J.; Delvigne, G. L.; Scholten, M. T. Environ. Toxicol. Chem. 2002, 21, 2242. (22) Jonker, M. T. O.; Brils, J. M.; Sinke, A. J. C.; Murk, A. J.; Koelmans, A. A. Manuscript in preparation. (23) Riegman, R.; Stolte, W.; Noordeloos, A. A. M.; Slezak, D. J. Phycol. 2000, 36, 87. (24) Jonker, M. T. O.; Koelmans, A. A. Environ. Sci. Technol. 2002, 36, 4107. (25) Borisover, M.; Reddy, M.; Graber, E. R. Environ. Sci. Technol. 2001, 35, 2518. (26) Freeman, D. H.; Cheung, L. S. Science 1981, 214, 790. (27) Lyon, W. G.; Rhodes, D. E. Environ. Toxicol. Chem. 1993, 12, 1405. (28) De Jonge, H.; Freijer, J. I.; Verstraten, J. M.; Westerveld, J.; Van der Wielen, F. M. Environ. Sci. Technol. 1997, 31, 771. (29) Lopes, C. F.; Milanelli, J. C.; Prosperi, V. A.; Zanardi, E.; Truzzi, A. C. Mar. Pollut. Bull. 1997, 34, 923. (30) De Bruijn, J.; Busser, F.; Seinen, W.; Hermens, J. L. M. Environ. Toxicol. Chem. 1989, 8, 499. (31) Neff, J. M.; Ostazeski, S.; Gardiner, W.; Stejskal, I. Environ. Toxicol. Chem. 2000, 19, 1809. (32) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M. Environmental Organic Chemistry; John Wiley & Sons: New York, 1993. (33) Means, J. C. Mar. Chem. 1995, 51, 3. (34) Schaefer, A. Environ. Sci. Technol. 2001, 35, 10A. (35) Pignatello, J. J.; Xing, B. S. Environ. Sci. Technol. 1996, 30, 1. (36) Koelmans, A. A.; Gillissen, F.; Makatita, W.; Van den Berg, M. Water Res. 1997, 31, 461. (37) Martinez-Inigo, M. J.; Almendros, G. Commun. Soil Sci. Plant Anal. 1992, 23, 1717. (38) Koelmans, A. A.; Van der Heijde, A.; Knijff, L. M.; Aalderink, R. H. Water Res. 2001, 35, 3517.

Received for review May 5, 2003. Revised manuscript received August 26, 2003. Accepted September 10, 2003. ES0300564

VOL. 37, NO. 22, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5203