Sediment beds at many sites are contaminated by diverse hydrophobic organiccompounds (HOCs) such as PAHs, PCBs, and pesticides. In an effort to manage such contaminated solids, regulators try to use data on the sediment concentrations (Csed) of these HOCs. The key parameter allowing use of an HOC's sediment concentration is that compound's solid-water sorption coefficient, Kd. However, using the current practice of estimating Kd from a sediment's organic carbon content (foc) and the compound's organic-carbon normalized sorption coefficient (Koc) has been found to be very inaccurate for estimating HOC mobility and bioavailability at many sites. We now believe this inaccuracy derives in large part from ignoring the sorption of HOCs to black carbons (BCs) like soots and chars in the sediment. Hence, it was the overarching goal of this research to improve our fundamental understanding of organic chemical sorption to BCs in sediments, while at the same time providing a practical means for estimating this interaction quantitatively.
Adsorption isotherms were determined for structurally diverse arrays of probesorbates to two representative BCs, a NIST diesel soot and a wood char. Since the char particles could be readily settled, sorption testing was done using batch methods. However, since the soot was very finely divided, this BC was coated on clean sand and this aggregate was used as a stationary phase in liquid chromatography frontal analysis of each sorbate run at a series of increasing concentrations. The resultant concentration-dependent Kd values were analyzed for the importance of various intermolecular interactions by fitting the data for each BC sorbent to find a polyparameter linear free energy relationship (ppLFER).
All of the sorbate-sorbent combinations fit by Freundlich isotherms very well. Sorption to the wood char Kd data for the diverse sorbate set also fit the following ppLFER:
log Kd,char = [(4.03±0.14) + (-0.15±0.04) log ai ] • V + [(-0.28±0.04) log ai ] • S + (-5.20±0.21) • B(N=128, R2=0.98, SE=0.41)
where V (in cm3 mol-1/100) is McGowan’s characteristic volume, S is the polarity/polarizability parameter, B is hydrogen basicity, and ai is the sorbate activity. This fitted expression implies that sorption to the char was strongly encouraged by London dispersive forces, was strongly discouraged by moieties that hydrogen bond with water by accepting protons, and was modestly encouraged by sorbate polarity (note log ai is always a negative number).
In contrast, sorption to the diesel soot fit a similar ppLFER:
log Kd,soot (L/kg) = (4.26±0.11) • V + [(-0.30±0.02) log ai )] • E + (-3.16±0.15) • B + (-1.95±0.10)(N=86, R2=0.96, SE=0.16)
While similar, sorption to soot proved weaker than char sorption, even when sorption coefficients were normalized to the sorbents' surface areas.
One important application of such ppLFERs is predicting Kd of other compounds. The soot ppLFER successfully predicted the Kd of phenanthrene and dimethyl phthalate to soot, butatrazine was generally overestimated, especially at lower activities. We suspect the nonplanarnature of atrazine caused this (all the training sorbates were planar). We also examined the accuracy of our ability to estimate Kd values for PAHs and PCBs on sediments and soils.Generally, predicted values of phenanthrene agreed well with reported ones; however, ppLFER-based estimates for larger HOCs compared much less favorably. We suspect that larger sorbates cannot access the same microporous surface areas as done by our sorbate training set.
Our results showed that one can develop fundamental understandings of sorption to black carbon surfaces using diverse sorbates and ppLFER synthesis of the results. And use of such ppLFER fits with independent measures of the black carbon contents of sediments and soils reasonably estimated Kd values reported by other investigators for phenanthrene. Hence, this approach appears promising for real world samples.
However, extrapolations well outside the tested ppLFER parameter space were not accurate. We suspect two steric issues came into play: (a) non-planar sorbates interact with the BC surfaces less effectively than our planar sorbate probes, and (b) larger HOCs may not enjoy as much access to highly sorptive sites in microporous regions of BCs as our somewhat smaller probe sorbates, thereby reducing their sorption coefficients. Future work needs to examine these effects.
Finally, our experience with sorption to diesel soot and wood char indicates that these BCs exhibit different affinities for like sorbates. As a result, application of Kd,soot or Kd,char values will require information on the nature of the BC in any soil or sediment of interest. Hence, we need to develop an analytical methodology that can accurately quantify soots, chars, and any other BC types (e.g., coal dust) in natural solids.