Energy and Environmental Research Center, University of North Dakota, 15 North 23rd Street, Grand Forks, North Dakota 58203, United States.
Environ Sci Technol. 2011 Sep 1;45(17):7365-71. doi: 10.1021/es200802j. Epub 2011 Aug 5.
More than 1900 sediment-water partitioning coefficients were measured for 58 polychlorinated biphenyl (PCB) congeners in 53 historically contaminated sediments collected from 10 urban and rural waterways in the United States and Canada. Freely dissolved porewater concentrations were determined using passive sampling with polyoxymethylene. Measured total organic carbon (TOC)/water partitioning coefficients, K(TOC), ranged from one to nearly three orders-of-magnitude higher than typical literature values based on spiking experiments and model predictions. Although total PCB concentrations ranged from 0.08 to 194 mg/kg, the more highly contaminated sediments showed only slightly lower K(TOC) values than less-contaminated sediments. No correlation was observed between log K(TOC) values and sediment TOC, black carbon (BC), or BC/TOC fractions (r(2) typically <0.1). Utilizing a two-carbon model incorporating anthropogenic BC did not improve predictions over a one-carbon TOC model. A comparison of models recently validated for field data showed that a coal-tar poly parameter linear-free energy relationship (PP-LFER) and a Raoult's Law model were successful at predicting average log K(TOC) values, without the need for any calibration or fitting (within a factor of 10 more than 90% of the time, and within a factor of 30 more than 99% of the time). Predictions were further improved by the introduction of a Weathering Factor (WF) that accounts for the relative depletion of lower molecular weight congeners due to weathering. Highly weathered sediments (with a WF near 1) tended to follow the coal-tar PP-LFER and Raoult's Law model the closest. Less-weathered sediments (with WF ≪ 1) sorbed less than predicted by these models. Noncalibrated WF inclusive coal-tar PP-LFER and Raoult's Law models performed as well or better than a quantitative-structure activity relationship (QSAR) model calibrated specifically to the data. These recommended partitioning models here can readily be used for all 209-PCB congeners.
对来自美国和加拿大 10 条城市和农村水道的 53 个历史污染沉积物中 58 种多氯联苯 (PCB) 同系物的 1900 多个分配系数进行了测量。使用聚甲醛的被动采样测定了自由溶解的孔隙水浓度。测量的总有机碳 (TOC)/水分配系数 K(TOC) 比基于添加实验和模型预测的典型文献值高出一个到近三个数量级。尽管总 PCB 浓度范围为 0.08 至 194mg/kg，但受污染程度较高的沉积物的 K(TOC) 值仅略低于受污染程度较低的沉积物。没有观察到 log K(TOC) 值与沉积物 TOC、黑碳 (BC) 或 BC/TOC 分数之间的相关性 (r²通常<0.1)。利用包含人为产生的 BC 的两碳模型并没有提高对单碳 TOC 模型的预测。对最近针对现场数据进行验证的模型进行比较表明，煤焦油多参数线性自由能关系 (PP-LFER) 和 Raoult 定律模型成功地预测了平均 log K(TOC) 值，而无需进行任何校准或拟合(在 90%以上的时间内，误差在 10 倍以内，在 99%以上的时间内，误差在 30 倍以内)。通过引入风化因子 (WF) 可以进一步提高预测结果，该因子考虑了由于风化而导致的低分子量同系物相对耗尽的情况。高度风化的沉积物 (WF 接近 1) 往往最接近煤焦油 PP-LFER 和 Raoult 定律模型。风化程度较低的沉积物 (WF≪1) 的吸附量低于这些模型的预测值。未校准的 WF 包含的煤焦油 PP-LFER 和 Raoult 定律模型的表现与专门针对该数据进行校准的定量结构活性关系 (QSAR) 模型一样好或更好。这里推荐的分配模型可以很容易地用于所有 209-PCB 同系物。
