Formatted Title
A Robust Approach to Interpreting Changes in Profiles when Applying Statistical Fingerprinting to PFAS
Background/Objectives
Background/Objectives. Statistical fingerprinting is routinely applied to chemical mixtures such as PCCD/Fs, PCBs, and PAHs, using methods that are widely published and routinely tested in the courtroom. These methods provide a means to explore sources and define the footprints associated with these sources. However, there are serious challenges associated with the application of traditional fingerprinting methods to PFAS. There is increasing awareness that multiple processes, many of which are not considered when fingerprinting other compound classes, can be responsible for changes on the PFAS profiles. Understanding these processes and how they manifest as changes in the PFAS profiles is critical in understanding and identifying PFAS sources and linking these sources to environmental impacts. These processes can include (1) changes in the mixture of PFAS used at facilities in response to changing market conditions or participation in the PFAS stewardship program, (2) the transformation of precursors to PFAS due to degradation processes, (3) changes in the relative concentrations of different PFAS as a groundwater plume migrates due to differences in the retardation rates, and (4) the unique properties of surfactants and their ability to self-aggregate into micelles and other structures and interact with surfaces and other compounds in solution, affecting PFAS fate and transport processes. We explored these different processes and developed strategies that can be used to aid in the interpretation of changes in PFAS profiles.
Approach/Activities
Approach. We used publicly available data and simple simulations to demonstrate how different processes can result in changes in PFAS profiles and developed strategies to help identify the underlying processes when interpreting the results of a PFAS forensic analysis. These strategies include statistical data analysis methods, graphical presentation of the profiles, geospatial analysis of the distribution of these profiles, and the development of other lines of evidence, such as the semi-automated review of site documentation, aerial deposition modeling, and groundwater modeling.
Results/Lessons Learned
Results. We developed strategies that can be used to interpret the results of a forensic analysis and improve the understanding of PFAS sources and fate and transport processes. Chief among these is the development of hypotheses and alternative hypotheses that address the likely impacts of the different processes that can result in changes in the profiles and provides a framework for robust hypothesis testing. We also characterized how these different processes manifest themselves in PFAS profiles. We demonstrate that traditional forensics statistical analysis methods, such as cluster analysis, principal components analysis and receptor models; and less common methods, such as UMAP, can provide useful lines of evidence to differentiate between different processes and evaluate hypotheses. The results indicate that although these processes present complications that can add uncertainty to understanding of sources and sinks, the careful application of our approach provides a means to test for the presence of these processes and develop a robust understanding of sources and receptors.