Environ. Sci. Technol. 2005, 39, 3688-3694
Sorption to Black Carbon of Organic Compounds with Varying Polarity and Planarity G E R A R D C O R N E L I S S E N , * ,† JORIS HAFTKA,‡ JOHN PARSONS,‡ AND O ¨ RJAN GUSTAFSSON† Department of Applied Environmental Science, Stockholm University, 10691 Stockholm, Sweden, and Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, Nieuwe Achtergracht 166, 1018 WV Amsterdam
It is becoming increasingly clear that the products of incomplete combustion (soot and charcoal, collectively termed black carbon or BC) can be responsible for as much as 80-90% of the total sorption to sediments of aromatic, planar, and hydrophobic compounds such as polycyclic aromatic hydrocarbons or planar polychlorinated biphenyls. In the present study, it was investigated whether a nonpolar aliphatic compound (hexachloroethane) and three nonplanar bipolar compounds with different functional groups [free electron pairs but no aromatic ring (butylate) or free electron pairs and an aromatic ring (diuron, atrazine)] would also show strong and nonlinear sorption to a BC-enriched sediment. At a concentration of 1 ng/L, the extent of elevated BC sorption compared to total organic carbon (TOC) sorption increased in the order atrazine < hexachloroethane < butylate < diuron. Rationalization of the differences between the sorbates was attempted in terms of dispersive and steric effects. This study shows that the effects of strong BC sorption apply to a broader range of organic contaminants than previously thought, and the results will aid in a better understanding of BC sorption mechanisms and improved fate modeling of contaminants in the environment.
Introduction Pyrogenic carbon particles such as soot and charcoal (collectively termed black carbon or BC) are ubiquitous in the aquatic environment (1-5) and frequently make up 1-20% of total organic carbon or TOC in sediments (2-5). The sorption to pure BC materials such as diesel soot, coal soot, and wood charcoal has been observed to be up to 101000 times stronger than the sorption to “other” nonpyrogenic OC in soils and sediments (2, 4, 6-14), for planar aromatic compounds such as polycyclic aromatic hydrocarbons (PAHs; 6-10), planar polychlorinated biphenyls (PCBs; 9, 11), polychlorinated dibenzo-p-dioxins and -furans (PCDD/Fs; 12), and polybrominated diphenyl ethers (PBDEs; 12). As a result, BC can dominate total sorption of such organic pollutants, with up to 80-90% of sorbate adsorbed to BC (6, 10, 15, 16). As the sorption to BC is often strongly nonlinear * Corresponding author phone: +47-22023082; fax: +47-22230448; e-mail:
[email protected]. Present address: Norwegian Geotechnical Institute, P.O. Box 3930, Ullevaal Stadium, N-0806 Oslo, Norway. † Stockholm University. ‡ University of Amsterdam. 3688
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 10, 2005
[Freundlich nonlinearity coefficient as low as 0.50-0.70 (6, 7, 10, 13-15, 19, 40)], the effect of BC is strongest at low (nanogram per liter and less) sorbate concentrations (15). BC sorption was observed to be approximately 1 order of magnitude less strong for nonplanar ortho-substituted PCBs than for corresponding planar ones with similar hydrophobicity (8, 11, 14). Thus, BC sorption strength appears to be dependent on a combination of dispersive interactions and separation distance (steric effects). The BC itself can probably be regarded as a “super-PAH” with its aromatic and condensed character rendering it well-polarizable and capable of strong dispersive interactions. Strong dispersive interactions occur between BC and nonpolar, readily polarizable aromatic compounds such as the previously mentioned groups. However, these interactions are strongly attenuated when the separation distance becomes larger, such as for nonplanar PCBs. It remains unclear whether nonpolar aliphatic compounds and mono/dipolar compounds with functional groups will show similarly strong sorption to environmental BC. Indications that this is probably the case include the observations that (i) the nonplanar pesticide diuron shows strong and nonlinear charcoal sorption (13, 17), (ii) chlorinated aliphatics show strong and nonlinear coal sorption (18, 19), and (iii) many pesticides are strongly bound by activated carbon (e.g., 20-23). It is unresolved whether such compounds also show strong sorption to environmental BC (i.e., field samples enriched in BC relative to other C forms). Therefore, the sorption to a sediment enriched in BC was studied for four compounds of similar MW but with different functionalities and thus different polarizability, planarity, and hydrophobicity. Through the comparison of these structurally different compounds, our understanding of the mechanisms involved in sorption to environmental BC should increase.
Experimental Procedures Materials. Diuron (DIU, C9H6ON2Cl2; >98%), atrazine (ATR, C8H13N5Cl; >99%), hexachloroethane (HCE, C2Cl6; 99%), butylate (BUT, C11H23ONS; >98%), and silica (Kieselguhr) were obtained from Sigma-Aldrich, Sweden. Oasis HLB extraction cartridges (30 mg of Oasis resin, cartridge volume 20 mL) were obtained from Waters, Sweden. Polyoxymethylene (POM) was obtained in 0.5 mm thick sheets from Vink Kunststoffen BV, The Netherlands. Solvents (acetone, hexane, methanol; all glass-distilled Burdick and Jackson quality) were obtained from Fluka, Sweden. Test Compounds. First, hexachloroethane (HCE; Figure 1a) was selected because it is a nonpolar (zero dipole moment) nonaromatic compound that is capable of only van der Waals interactions with the BC. Second, the herbicide butylate was studied (BUT, S-ethyl diisobutylthiocarbamate; Figure 1b), a nonplanar, nonaromatic monopolar compound that can exhibit dispersive interactions through its free electron pairs on three functional groups (N, O, and S). Third, the herbicide diuron was used [DIU, 3-(3,4-dichlorophenyl)-1,1-dimethylurea; Figure 1c] because it is a nonplanar, bipolar aromatic compound for which strong charcoal sorption has been shown (13, 17). Fourth, the herbicide atrazine was selected [ATR, 2-chloro-4-(ethylamino)-6-(isopropylamino)-1,3,5-triazine; Figure 1d], because it is a nonplanar aromatic compound possessing a heterocyclic aromatic ring. The physicochemical properties of the studied compounds are shown in Table 1. Sediment. Sediment used was from Ketelmeer (KET), a freshwater lake in The Netherlands (52° 36′ N, 5° 45′ E; 10.1021/es048346n CCC: $30.25
2005 American Chemical Society Published on Web 04/07/2005
TABLE 1. Selected Properties of the Compounds Studieda HCE BUT DIU ATR PHE
MW
log KOW, L/L
S, mg/L
H, Pa‚m3/mol
MV, Å3
MSA, Å2
D, debyes
avg r, Å3
∆Eplanar, kcal/mol
237 217 233 216 178
3.8 4.2 2.6 2.7 4.6
50 45 42 33 1
250 0.56 1 × 10-4 3 × 10-4 4
457 768 651 684 189
303 470 414 440 373
0.001 10.2 12.1 7.3 3.1
16.0 25.4 22.6 22.1 24.6
-b 1.1 0.31 ∼0c 0c
a Molecular weight (MW) (36), log K OW (36), solid solubility in water at 25 °C (S) (36), Henry’s law constant (H) (36), molecular volume (MV), molecular surface area (MSA), dipole moment (D), average polarizability (R), and energy difference between minimal energy and “forced-planar” structure (∆Eplanar) are presented for hexachloroethane (HCE), butylate (BUT), diuron (DIU), atrazine (ATR), and phenanthrene (PHE). See text for the calculation of MV, MSA, D, R and ∆Eplanar. b Not calculated because of impossibility to attain a planar structure. c Lowest-energy structure is (near-)planar.
FIGURE 1. Structures of the studied compounds hexachloroethane (HCE), butylate (BUT), diuron (DIU), and atrazine (ATR). sampled layer 10-50 cm) that is the first major sedimentation area of one of the major European rivers, River Rhine. The characteristics of the currently used batch of KET sediment have been presented in refs 10 and 15 and include dry weight (47.2%), TOC (5.51%), BC (0.72%), TOC:TON (total organic nitrogen) atomic ratio (21), BC:BN (black nitrogen) atomic ratio (40), ratio of fossil fuel to biomass-derived BC (80:20, as revealed by 14C dating of the BC), and total native PAH and PCB contents [∼40 mg/kg and ∼0.15 mg/kg dry weight (dw), respectively]. Chemothermal Oxidation at 375 °C to Enrich in BC. The 375 °C chemothermal oxidation (CTO-375) method employed in this work has been thoroughly tested with negative and positive standards (e.g., 2, 24, 25) and provides BC results in sediments that are geochemically consistent (e.g., 26-28). Limitations of the CTO-375 method include (i) particles with a high relative content of nitrogen may char to artifactually form BC during CTO-375 combustion (24, 29) and (ii) it may remove also some less-condensed pyrogenic constituents formed at lower combustion temperatures (e.g., 8, 31, 32). During combustion, the native PAHs and OM are thermally desorbed and/or combusted (e.g., 33, 34). Hence, to “harvest” combusted sediment for the present sorption study, dry sediment (10 mg in Ag capsules) was combusted at 375 ( 2 °C during 18 h under a constant air flow of 200 mL/min. A small size of a well-pulverized sample was employed to optimize access of oxygen throughout the sediments, which is important to prevent charring (2, 24). In this study, the combustion method was scaled up by combusting 36 (10 mg sized) samples in each run and doing several runs to obtain enough material for the sorption experiments. The BC:BN ratio of 40 (n ) 10) for the KET
sediment suggests that the abovementioned charring of any N-rich organic matter was negligible in this study. Sorption Experiments. Sorption to KET sediment before and after combustion was measured at a wide concentration range (approximately a factor of 1000). Separate experiments were performed for each compound. All sorption experiments were performed at neutral pH (pH 7). HCE and BUT. For HCE and BUT, the sorption experiments were done by the polyoxymethylene solid-phase extraction (POM-SPE) method developed by Jonker and Koelmans (9) with slight modifications (10). In short, a POMwater-sediment system is shaken until equilibrium is reached. POM is a pure polymer that serves as a nondepletive extraction agent that is added in order to quantify the freely dissolved aqueous concentration (CW) of BUT and HCE. One can deduce CW from the sorbate contents in POM with concentration-independent POM-water distribution ratios. Aqueous sorbate stock solution (0.05-50 mL; prepared by adding 50 µL/L of methanol containing one of the test compounds), 300-500 mg of dry sediment, 0-50 mL of distilled water, 100 mg of NaN3 (biocide and ionic strength regulator), and 10 mg of HgCl2 (biocide) were combined in 50 mL all-glass flasks. A strip of POM was also added (500 mg for HCE, 800 mg for BUT; POM size dependent on expected sorption properties of the compound). The flasks were horizontally shaken at 180 rpm at 24.5 ( 0.2 °C for 70 days. An equilibrium time of ∼20-25 days was found to be sufficient for spiked phenanthrene (PHE) in both original and combusted KET sediment (10), so the presently employed shaking time of 70 days probably sufficed for the establishment of equilibrium for the presently used compounds as well. After equilibration, the isotherm ranges were 0.1-100 µg/L (BUT) and 0.05-50 µg/L (HCE). Thus, the upper isotherm points represented aqueous concentrations below the maximum solubilities by about a factor of 1000 (BUT) and 10 000 (HCE). Sorption coefficients were not corrected for the “salt effect” caused by the biocide concentration, as this salt correction due to the molar salt concentration (0.03 M) would be on the order of 0.01 log unit (35). Because of the high Henry’s law coefficient (H) of HCE [around 250 Pa‚m3/mol (36); Table 1], evaporation losses from the system were anticipated for this compound. Therefore both POM and sediment were analyzed, in exactly the same manner as in refs 10 and 14. BUT is much less volatile [H ) 0.56 Pa‚m3/mol (36); Table 1] and thus only slight evaporation losses were suspected, which was confirmed by high recoveries of blanks in the absence of sediment (recoveries 91% ( 14%; n ) 3; concentration range 10-200 µg/L BUT). Therefore, for BUT only POM but not the sediment was analyzed. Because care was taken that >95% of the total BUT in the system was in the sediment, evaporation losses influenced the measured KOC values by maximally 0.05 log unit. POM strips were “cold”-extracted by horizontal shaking (180 rpm; 24.5 °C) with hexane (for HCE) or acetone (for BUT); the selection of solvents was based on pilot extraction/ VOL. 39, NO. 10, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3689
cleanup experiments with several solvents. HCE was extracted from the sediment by hexane/acetone/water reflux, where 0.05-2 g of sediment, 15-40 mL of water, 15 mL of acetone, and 50 mL of hexane were brought in 300 mL Erlenmeyer flasks and refluxed for 6 h under vigorous magnetic stirring. Internal standards (200 ng of 2-PCB in 50 µL of hexane for HCE, 200 ng of pyrene-d10 in 50 µL of acetone for BUT; internal standard selection on the basis of similar H of test compound and internal standard) were added before POM and sediment extraction. After extraction, the hexane and acetone phases were volume-reduced in a vacuum centrifuge; cleaned up by activated Cu (sulfate removal), a silica microcolumn (silica deactivated with 10% water), and Na2SO4 (removal of water); and once more volume-reduced in a vacuum centrifuge. Quantification of HCE and BUT was done on a GC-MS equipped with an SGE BPX 5 fused silica capillary column (column 30 m × 0.25 mm; film 0.25 µm) and a Fisons FI MD 800 mass spectrometer operated in electron impact mode (EI+, 70 eV) and single ion monitoring data acquisition. Detection limits in water for POM and GC-MS analysis were around 10 ng/L (BUT) and 0.5 ng/L (HCE). Mass balances for HCE were 26% ( 11% (n ) 12), so the suspected evaporation losses were observed. Because sorbate contents were measured in both sediment and POM, this did not influence our reported sorption results. DIU and ATR. As pilot experiments showed that DIU and ATR did not sorb to POM sufficiently strongly, these compounds were subjected to “conventional” batch equilibrations in the absence of POM. With the exception of the addition of POM, the setup of the experiments was exactly similar to that of BUT and HCE. After equilibration, the suspensions were centrifuged (1500 rpm, 20 min), after which the aqueous phase was extracted and analyzed for ATR and DIU. It was assumed that the amount of ATR and DIU sorbed to dissolved organic matter (DOM) (and thus the “particle concentration effect”) was insignificant because of the limited hydrophobicity of these compounds [literature log KOC ) 2.6 (36) for both ATR and DIU]. Isotherms were measured over concentration ranges of 0.02-15 mg/L (DIU) and 0.03-3.6 mg/L (ATR); these are factors of 3 (DIU) and 10 (ATR) below aqueous solubilities (36). Because of the low volatility of ATR and DIU (H ) 3 × 10-4 and 1 × 10-4 Pa‚m3/mol, respectively; Table 1), no evaporation losses were expected. This was confirmed by high recoveries of blanks in the absence of sediment for ATR [134% ( 20% (one standard deviation; n ) 3; concentration range 0.5-5 mg/L ATR)] and DIU [119% ( 25%; n 3 ; concentration range 3-18 mg/L DIU)]. Therefore, only the water but not the sediment was analyzed. Because of the photosensitivity of especially DIU, care was taken to shield the samples from direct light during equilibration and handling. Oasis HLB extraction cartridges (60 mg of resin in 20 mL cartridges) were used for the extraction of the aqueous phase. Procedures were similar to the ones in ref 37. The columns were preconditioned by the elution of 3 × 5 mL of methanol followed by 3 × 5 mL of distilled water. A 10 mL portion of the aqueous phase was extracted by an Oasis cartridge, along with 100 µL of internal standard (10 mg/L) in water. ATR was used as an internal standard for DIU, and vice versa. Subsequently the column was eluted with 10 mL of methanol. The recoveries of the column extractions were tested with aqueous standard solutions and turned out to be 98% ( 7% (n ) 3) for ATR and 103% ( 5% for DIU (n ) 3). The methanol volumes were reduced in a vacuum centrifuge, and analyses were performed without further cleanup. Quantification was done by means of HPLC (Merck Hitachi LaChrom equipped with a L-7400 UV detector), on a MSC18 5A12 Spherisorb C18-5 µm microcolumn (150 × 2.0 mm; Chrom Tech AB, Sweden) with methanol-water gradi3690
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 10, 2005
ent elution. UV wavelengths were 225 nm for ATR and 251 nm for DIU. Detection limits were around 10 µg/L (ATR) and 20 µg/L (DIU). Molecular Modeling. Molecular structures, atomic charges, and properties were calculated with the program Hyperchem (38). Molecular structures were initially geometrically optimized in Amber 99 force field [Polak-Ribiere optimization with a gradient criterion (RMS) of 0.01 kcal/mol], subsequently optimized semiempirically in PM3 software [restricted Hartree-Fock (RHF), convergence limit ) 0.001, iteration limit ) 50, RMS ) 0.01 kcal/mol] to obtain atomic charges. Finally, the stable geometries were found with the Amber 99 force field, and values for the dipole moment (D) and the energy difference between the minimal-energy structure and a structure in which the molecule was planar (∆Eplanar) could be calculated (presented in Table 1). Descriptors derived from the geometry-optimized structuress molecular volume (MV), molecular surface area (MSA), and polarizability (R)swere calculated by QSAR Properties package included in Hyperchem. For comparison, values for phenanthrene (PHE) are also presented in Table 1. In our modeling of the interaction parameters, we did not take into account the effect of the hydration sphere on the HOC sorption to the BC surface.
Results and Discussion TOC and BC Sorption. POM-water distribution ratios (KPOM) of HCE and BUT were needed to calculate CW from the concentrations in the POM after sorption. These were experimentally determined. Log KPOM was 2.21 ( 0.08 for BUT (n ) 3) and 2.61 ( 0.03 for HCE (n ) 3). All sorption isotherms were fitted with the Freundlich equation:
CS ) fXKF,XCWnF,X
(1)
where X denotes either BC or TOC. CS is the sorbed concentration on a whole sediment basis (micrograms per kilogram dry weight ), fX is the fraction of BC or TOC, KF,X is the Freundlich sorption coefficient for TOC or BC [(µg/kgX)/ (µg/L)nF], nF,X is the Freundlich sorption nonlinearity parameter, and CW is the aqueous concentration (micrograms per liter). The TOC in the original sediment before combustion showed rather linear isotherms for all four chemicals (nF,TOC ) 0.87-0.99; Figure 2 and Table 2). In a recent paper (10) it was shown that “environmental” sorption to BC (i.e., in the presence of OM and native compounds) is weaker than “intrinsic” BC sorption (i.e., to “clean” environmental BC after combustion). For PHE-d10, the difference between the environmental KF,BCenv (in original KET sediment) and the intrinsic KF,BCint (in combusted sediment) was a factor of 9 (10). This effect was illustrated by plotting CS against CW for total sorption in the original sediment and the BC part of this total sorption calculated on the basis of the sorption isotherm for combusted sediment (see below). The OC- and BC-inclusive dual-mode sorption isotherm is (4-7, 10, 14-16, 26, 27)
CS ) fOCKOCCW + fBCKF,BCCWnF,BC
(2)
where fOC and fBC are the geosorbent fractions of OC and BC, respectively (fTOC ) fOC + fBC), and KOC is the “BC-free” OCwater partition coefficient. The last term in eq 2 represents the BC contribution to total sorption; the first term is the contribution of the other OC. Ideally, one would use KF,BCenv and not KF,BCint in eq 2. For the calculation of KF,BCenv from KF,BCint, an iterative procedure (10) was used where the following three-step cycle was repeated until a constant KF,BCenv was obtained: (i)
FIGURE 2. TOC-normalized Freundlich sorption isotherms for original Ketelmeer (KET) sediment (]) and BC-normalized Freundlich isotherms for combusted KET sediment (OM and all native sorbates removed; [) for HCE (top left), BUT (top right), DIU (bottom left), and ATR (bottom right). Lines were obtained by linear regression.
TABLE 2. Logarithmic Freundlich Sorption Coefficients of the Test Compounds with TOC and BC in Lake Ketelmeer (KET) Sediment
HCE BUT DIU ATR PHEc
log KF,TOC, (µg/kgTOC)/(µg/L)nF
nF,TOC
r2
log KF,BCint a, (µg/kgBC)/(µg/L)nF
nF,BC
r2
log KBCint at 1 ng/L, L/kg
log KF,BCenv, (µg/kgBC)/(µg/L)nF
4.22 ( 0.03 3.84 ( 0.12 2.9 ( 0.2 2.48 ( 0.16 5.05 ( 0.14
0.99 ( 0.04 0.92 ( 0.11 0.87 ( 0.08 0.99 ( 0.06 0.89 ( 0.05
0.994 0.949 0.963 0.984 0.991
4.63 ( 0.08 5.03 ( 0.09 5.3 ( 0.2 3.1 ( 0.3 5.62 ( 0.04
0.60 ( 0.08 0.87 ( 0.08 0.50 ( 0.08 0.97 ( 0.12 0.54 ( 0.02
0.925 0.970 0.882 0.946 0.976
5.8 5.4 6.8 3.2 7.0
4.4 4.5 4.4 3.1 4.9
a K int is the “intrinsic” Freundlich BC sorption coefficient for BC in combusted sediment without PAHs and OM. b K env is the environmental F,BC F,BC Freundlich BC sorption coefficient before combustion, in the presence of native PAHs and OM. Values are calculated from the data (see text).c From ref 10.
fBCKF,BCintCwnF,BC was subtracted from the total CS (eq 2) for the isotherm points where BC contributed 70% of total sorption at 1 ng/L. The importance of BC for HCE, BUT, and DIU sorption increased with decreasing concentrations (90-100% at 1 pg/L) but diminished at higher ones (
