Environ. Sci. Technol. 2000, 34, 5144-5151
Quantification of the Soot-Water Distribution Coefficient of PAHs Provides Mechanistic Basis for Enhanced Sorption Observations THOMAS D. BUCHELI AND O ¨ RJAN GUSTAFSSON* Institute of Applied Environmental Research (ITM), Stockholm University, 10691 Stockholm, Sweden
There is an increasing recognition of the necessity to consider the heterogeneity of geosorbents, and in particular the condensed carbon facies fraction, to improve prediction of hydrophobic pollutant phase speciation. Field observations of much elevated organic-carbon normalized distribution coefficients (Koc) of PAHssrelative to predictions from bulk organic-matter partitioning modelss have been suggested to be explainable by soot sorption. To afford testing of this hypothesis, we here report on the sootwater distribution coefficients (Ksc) for a series of PAHs (naphthalene (NP), fluorene (FL), phenanthrene (PH), and pyrene (PY)) using diesel particulate matter (NIST standard reference material SRM-1650) as model soot sorbent. Specifically adapted batch and column experiments yielded average log Ksc values of 5.23, 5.40, 5.82, and 6.59 (batch) and 4.63, 6.03, 6.62, and 7.03 (column) for NP, FL, PH, and PY, respectively (all data in [Lw/kgsc]). The obtained values are 35-250 times higher than respective Koc predictions and are considerably closer to theoretically estimated sootwater distribution coefficients. Our data are among the highest solid-water distribution coefficients of an environmentally relevant sorbent ever reported and lend direct empirical support of active soot sorption as a viable explanation to the enhanced PAH partitioning. Sorption kinetics on the hours-days time scale and similarity of external geometric and BET surface areas suggest that interaction sites are largely restricted to the outer surface of the soot. The constrained Ksc values facilitate prediction of speciation and bioavailable exposures of PAHs in aquatic and sedimentary environments.
Introduction Knowledge of the phase distribution of toxic and xenobiotic chemicals such as polycyclic aromatic hydrocarbons (PAHs) is a fundamental prerequisite for comprehending and predicting their environmental fate and behavior, including their bioavailable exposures. The field of hydrophobic sorption is currently undergoing a paradigm shift with respect to the conceptualization and model formulations of both kinetic and equilibrium aspects (e.g., refs 1 and 2). The wellestablished organic-matter partitioning model (3, 4) has been found to underestimatesby several orders of magnitudes the solid-water distribution coefficient (Kd) of PAHs in a broad * Corresponding author phone: ++46 8 674 73 17; fax: ++46 8 674 76 38; e-mail:
[email protected]. 5144
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 24, 2000
range of real field situations (5-8; reviewed in 2). Such poor speciation predictability has a proportional effect on our (in)ability not only to estimate direct bioavailabilities but also to anticipate all other speciation-dependent fate processes of PAHs. It was early on speculated that such elevated Kd values of PAHs reflected some permanent occlusion of PAHs during their coproduction with soot by incomplete combustion (5, 9, 10). However, the elevated observed organic-carbon normalized partition coefficients appear to follow trends in physicochemical properties of the PAHs such as their hydrophobicities as opposed to their relative production yields. This has led to the proposition that PAHs are actively partitioning with some supersorbent (11), which very well may be highly aromatic residues of incomplete combustion (soot) known to be ubiquitous in marine sediments (e.g., refs 12-15). It has consequently been hypothesized that Kd may be more accurately predicted by expanding the organicmatter partitioning model by considering additional association also with the soot-carbon fraction of the solid phase (16)
Kd ) foc‚Koc + fsc‚Ksc
(1)
where foc and fsc are the fractions of organic carbon and soot carbon, respectively, in the solid matrix [kg organic or soot carbon/kg total solid], and Koc and Ksc are the organic-carbon and soot-carbon normalized distribution coefficients, respectively [(mol/kg organic or soot carbon)/(mol/L solution)]. By applying this expanded partitioning model (eq 1), with explicit measures of fsc and crude approximations of Ksc either from theoretical arguments (16) or by using activated-carbon distribution coefficients (17, 18), the actual field-observed distributions of planar aromatic compounds have been much better explained than by assumption of partitioning solely with bulk organic matter (11, 19, 20). Other condensed carbon matrices such as coal or charcoal residues in aquifer systems have recently been inferred to have a similar enhanced sorption capacity (21-23). However, there are considerable differences in physicochemical properties between soot and these types of particulate matter in terms of, for instance, surface area, elemental composition, and functional groups. The soot fraction (fsc) can be isolated and quantified using a selective thermochemical oxidation method, where more amorphous natural organic matter is removed by thermal oxidation in air (375 °C for 24 h), and inorganic carbonates are released via in situ acidification (11). The fsc is found to make up 2-30% of total organic carbon in coastal sediments (13-15). Hence, for the soot sorbent subfraction to be dominating over the natural organic matter as a carrier of PAHs, Ksc must be at least an order of magnitude larger than Koc. Direct experimental determinations of Ksc, necessary for explicit testing of the soot “super-sorbent” hypothesis, have not yet been conducted. This paper reports on the quantification of Ksc for a set of two to four ringed PAHs (naphthalene: NP, fluorene: FL, phenanthrene: PH, and pyrene: PY) using a model soot material. Diesel particulate matter (standard reference material SRM-1650) from the U.S. National Institute of Standards and Technology (NIST, Gaithersburg, MD) was selected, as it represents one of the environmentally most relevant anthropogenic types of soot (12, 24). Moreover, a significant portion of its solid phase consists of soot carbon (see below), and concentrations of natively bound PAHs are reported in the NIST certificate (25). Quantification of Ksc was achieved with both batch and column experiments that 10.1021/es000092s CCC: $19.00
2000 American Chemical Society Published on Web 11/16/2000
were specifically designed for distribution experiments with highly hydrophobic and strongly sorbing matrices such as soot.
Experimental Section Materials. Sodium sulfate anhydrous (> 99%), hexane (p.a.) and toluene (p.a.) were from Merck (Darmstadt, Germany), while methanol (MeOH) was purchased from BDH Laboratory Supplies (Poole, England). The soot standard (diesel particulate matter, SRM-1650) was obtained from the NIST (Gaithersburg, MD). Silica60 (70-230 mesh) and thiourea (> 99%) were provided from Merck (Darmstadt, Germany). The PAHs were purchased from Sigma (St. Louis, MO). Deuterated analogues (d8-NP (99%), d10-FL, d10-PH, and d10-PY (all 98%)) used as internal standards were obtained from Cambridge Isotope Laboratories (Andover, MA). Soot Surface Analysis. The BET (26) surface area, pore volume and pore size distribution of SRM-1650 were determined with nitrogen adsorption measurements at 77 K on an Accelerated Surface Area and Porosimetry system, ASAP 2010 (Micromeritics Instrument Corporation, Norcross, GA), after degassing of the sample at 373 and 573 K. Batch Experiments. The small average particle size (e.g., ref 27) and the extreme hydrophobicity of soot render common aqueous suspension batch experiments (e.g., ref 28) inappropriate for the determination of soot-water distribution coefficients, because subsequent phase separation cannot easily be achieved. “Half-cell” systems, where the first compartment, containing both the solid and aqueous phase, stands in contact via air with the second one, containing only the aqueous phase, might overcome these limitations (29). We choose a “beaker in jar” constellation, where a 1 L glass beaker was placed into a 3 L glass jar with a hermetic ground-glass lid. The beaker itself as well as the space between the beaker and the jar can thus hold two distinct and physically disconnected water reservoirs. Exchange of solutes between the two reservoirs takes place via the remaining air volume. To constrain PAH equilibration kinetics and recoveries, inner water reservoirs of soot-free systems were spiked at 14-30 µg/L, which where then analyzed over time in the outer water phase. Batch sorption experiments were conducted both with single compound systems (NP, PH, and PY at 25 °C, and for PH also at 7 °C), as well as with mixed solute systems to investigate potential competitive effects. All experiments were performed with 800 mL deionized water in each of the two water compartments. The sorbate solutions for each isotherm experiment were prepared externally in one pool, and then distributed to the various inner beakers. Aliquots of these sorbate solutions were also stored separately to allow investigation of recovery and mass balance. In the single solute sorption experiments, initial (equimolar) concentrations in the inner water reservoir were 19, 26, and 30 µg/L, for NP, PH, and PY, respectively. This corresponded to onefifth of the aqueous solubility of PY and much less for the others. No PAHs were added to the outer water phase. Soot was added to the inner water phase at concentrations from 1.7 to 19.0 mg/L, which caused a decrease of the initial aqueous phase concentration of 26-60% for NP, 29-81% for PH at 25 °C, 28-82% for PH at 7 °C, and 56-97% for PY at the final sampling day. In most cases, the sorbent predominantly distributed at the air-water interface but occasionally also dispersed in the water phase. Additional experiments were performed with a PAH mixture containing 14 µg/L each of NP, FL, PH, and PY, at both 25 and 7 °C. Soot concentrations of 2.5-4.4 mg/L led to a reduction of the solutes’ aqueous fraction of 17-93%. Individual sorption isotherm points were determined for six different PAH-tosoot ratios for single compound experiments and three different ratios for the PAH mixture. To account for possibly
desorbing natively soot-bound analytes, control experiments with no PAHs added were conducted at both temperatures. After the completion of the initial spiking, the jars were tightly closed, stored in the dark, and continuously and gently agitated (horizontal shaker at approximately 30 rpm). At certain time intervals, water samples were collected from the outer water phase and the control solutions and analyzed for PAHs. Solute sorption was calculated using mass balance calculations that included losses due to air-water equilibrium (
