Impacts of Heterogeneous Organic Matter on Phenanthrene Sorption: Different Soil and Sediment Samples HRISSI K. KARAPANAGIOTI,* ,† JEFFREY CHILDS, ‡ AND DAVID A. SABATINI ‡ ICEHT/FORTH, Institute of Chem ical Engineering and High-Tem perature Processes, Foundation for Research and Technology Hellas, Patra, 26500, Greece, and School of Civil Engineering and Environmental Science, University of Oklahoma, Norman, Oklahoma 73019 Organic petrography has been proposed as a tool for characterizing the heterogeneous organic matter present in soil and sediment samples. A new simplified method is proposed as a quantitative means of interpreting observed sorption behavior for phenanthrene and different soils and sediments based on their organic petrographical characterization. This method is tested under singe solute conditions and at phenanthrene concentration of 1 μg/L. Since the opaque organic matter fraction dominates the sorption process, we propose that by quantifying this fraction one can interpret organic content normalized sorption distribution coefficient (K oc ) values for a sample. While this method was developed and tested for various samples within the same aquifer, in the current study the method is validated for soil and sediment samples from different sites that cover a wide range of organic matter origin, age, and organic content. All 10 soil and sediment samples studied had log K oc values for the opaque particles between 5.6 and 6.8. This range of K oc values illustrates the heterogeneity of opaque particles between sites and geological formations and thus the need to characterize the opaque fraction of materials on a site-by- site basis. Introduction Research hasdemonstrated that chemicalphase partitioning into soil or sediment organic matter (1, 2) is a satisfactory model for sorption of nonionic contaminants as these compoundsapproach their water solubility(3-6).However, severalstudies on the sorption ofsingle solutes indicate that, at low relative concentrations, the solute sorption may become nonlinear with much higher sorption coefficients (selected refs 3-13).At lowrelative concentrations,nonlinear sorption is observed and surface adsorption is proposed as the possible sorption mechanism, whereas at high relative concentrationspartitioningisdominant(3, 4). It is also found that in multiple-solute systems,the nonlinear sorption effect ofa given solute can be greatly attenuated because ofsolute competition (3, 8, 12). Several researchers have proposed the existence of multiple organic matter domains (14-17), but only limited effort has attempted to quantify these different domains. Chiou (7)and other researchers suggest that the presence ofsmallamountsofhigh surface area carbonaceousmaterial (e.g.,charcoal)maybe responsible for the observed nonlinear sorption, even when present in small amounts (7, 4, 18, 19). Gustafsson et al. (20) suggested that a soot-carbon fraction in soil organic carbon can also affect soil sorption behavior. Ghosh et al. (21) observed that, although coal comprised only 5% of the sample, it sorbed 62% of the solute added. Thus, the influence of the organic particle fraction (e.g., charcoal or soot) on the contaminant sorption behavior depends on the contaminant concentration range, the amount of the organic particles present, and the concentra- tion of coexisting contaminants. Even considering these factors, it is significant to elucidate the nature of organic particles in soils and sediments with a validated analytical method. Organic petrography has been proposed as such a method for characterizing the organic matter of different samples under the microscope. Previous studies related the types oforganic matter with equilibrium sorption properties (18, 19, 5). Gustafsson et al. (20) predicted composite sorption distribution coefficient (Kd ) values by including both a Kd value due to the soil organic matter and another Kd value due to the soot carbon fraction. Njoroge et al. (22) proposed an additive sorption model that predicted Kd values by assumingthat there are two different types oforganic matter, referred to as shallow-like and deep-like organic material. Karapanagiotiand Sabatini(5) attempted to quantify the cumulative organic content normalized sorption distribution coefficient (Koc)valueforsamplesbyconsideringthesorption properties oforganic matter subgroups within each sample. By summing the products of the subgroup Koc values (Koc OMi) and particle fractions [OMi, estimated based on the relative number of particles (% OMi ) OMi/OMb )], the cumulative Koc value for each sample was estimated using where the indices i and b are used to distinguish individual subgroup and bulksamples,respectively.Thisapproach was used to determine cumulative Koc valuesforsedimentsamples from different depths at a given site. The values predicted were within the 95% confidence interval for the experimen- tally determined log Koc values. Ideally, this Koc prediction method could be extended to Kd prediction.Byusingliterature valuesforthe organiccarbon fraction found in each organic matter subgroup, one could predict the fraction of organic carbon content (f oc) found in each organic matter subgroup (23). However, additional knowledge is required on the sorption properties and organic carbon distribution in organic matter subgroups before this method can be applied. A limitation of the Karapanagioti and Sabatini (5) study is that the samples studied were all from the same site; while the distribution of organic matter subgroups varied between samples (depths), the nature of organic matter in each subgroup was fairly similar. The objectives of the present study were as follows: (1) to characterize organic matter heterogeneity in soils and sediments from different sites,(2)to quantitativelyinterpret Koc values for each soil or sediment based on organic petrography characterization of organic matter subgroups, and (3) to evaluate the ability to extrapolate petrographic and sorption data from one site to soils and sediments from different areas that cover a wide range of organic matter *Corresponding author present address: Department of Marine Sciences, University of the Aegean, Sapphous 5, 81100, Mytilene, Greece. Phone: (+30942) 464763; e-mail: hrissi@yahoo.com. † ICEHT/ FORTH. ‡ University of Oklahoma. K oc ) ∑ % OM i K oc OMi (1) Environ. Sci. Technol. 2001, 35, 4684-4690 4684 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 23, 2001 10.1021/es010654n CCC: $20.00  2001 American Chemical Society Published on Web 10/25/2001