Background/Objectives: Polyethylene (PE) passive samplers are increasingly used to estimate the dissolved concentration of hydrophobic organic contaminants (HOCs) in sediment pore water. Because HOCs in pore water rarely reach equilibrium with passive samplers during typical deployments spanning weeks or months, estimating equilibrium pore water concentrations requires interpreting non-equilibrium data.
Most commonly, the release of one or more performance reference compounds (PRCs) is measured after deployment of the samplers and the measured PRC release rate is used to parameterize a mass transport model that is used to infer sampling kinetics. Two classes of mass transport models are typically applied: exponential (i.e., first-order) or diffusion-based. Exponential models are simple and can be deployed in Microsoft Excel or similar. Diffusion-based models are more mechanistic in nature. However, they require advanced software and are difficult to communicate with stakeholders that are not passive sampling experts. The objective of this work was to evaluate the two models in a cross-site field study, one of the largest ever conducted on passive samplers, with respect to performance, accuracy, and efficiency.
Approach/Activities: PE passive samplers preloaded with 11 rare PCB congener PRCs were obtained from a commercial producer (SiREM, Guelph, ON, Canada) and deployed at a number of field sites (three completed, three to six additional expected before January 2017) impacted with PCBs and/or PAHs. The geographic location, sediment chemistry, and hydrodynamic conditions varied widely between sites. Samplers were embedded within the bioactive layer of sediment for 2-4 weeks. PRC depletion and PCB/PAH uptake results were input into an automated data processing and mass transport modeling script written in Python. Estimated equilibrium PE and pore water concentrations were calculated by both models and compared.
Results/Lessons Learned: The two models calculated practically identical estimates with respect to total PCB/PAH pore water concentration. The diffusion model resulted in total PCB/PAH equilibrium pore water concentration estimates that were a statistically insignificant (paired t-test, p<0.05) 3% greater than the exponential model. However, the two models did not estimate the same results on a compound-by-compound basis. In general, the diffusion model predicted faster mass transport of the more hydrophobic HOCs (e.g., nona-PCBs) than the exponential model. Conversely, the diffusion model predicts 5-10% slower mass transport for less hydrophobic PCBs (e.g., tri-PCBs) than the exponential model. The latter effect dominates the former with respect to total PCB/PAH concentrations, as the very hydrophobic compounds do not significantly contribute to the overall pore water concentration. This prediction of slower mass transport results in, perhaps counterintuitively, higher predictions of equilibrium concentrations, as larger correction factors are applied to the non-equilibrium field data.
The data show that the two models will yield functionally identical pore water estimates under real-world conditions. Therefore, properties such as simplicity (of both operation and communication), cost, and time become deciding factors. The results of this study suggest that, for most practitioners, the simpler exponential model is more suitable.