Publication Date

2024

Document Type

Thesis

Committee Members

Audrey E. McGowin, Ph.D. (Advisor); Willie F. Harper, Jr., Ph.D. (Committee Member); Steve R. Higgins, Ph.D. (Committee Member)

Degree Name

Master of Science (MS)

Abstract

Per- and polyfluoroalkyl substances (PFAS) are of great interest recently because some members of this class exhibit human health risks at extremely low levels, making them the current subject of proposed regulation and policy. Treatment of PFAS contaminated wastewater often is performed through adsorptive processes, e.g., with activated carbon. Due to complexities with PFAS laboratory analysis, assessing treatment efficacy through direct measurement of PFAS is expensive and time consuming, potentially delaying the design and implementation of site-specific treatment systems. This project identified and characterized potential suitable simulants for PFAS during wastewater treatment for which analysis can be timely and economical. This work investigated the suitability of three non-hazardous food dyes as PFAS simulants during wastewater treatment with powdered activated carbon (PAC). The PAC was analyzed before and after PFAS exposure using Scanning Electron Microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS) to reveal fluorine and PFAS functional groups attached to the surface. Data from batch experiments with PAC revealed Freundlich constants (Kd) and pseudo first-order adsorption rate constants (k) for adsorption of allura red, tartrazine, indigo carmine, PFOA, PFOS, and several other PFAS species commonly found in aqueous film-forming foam. The food dyes were found to have Kd and k values comparable to PFOA, PFOS, and other PFAS species. An Artificial Neural Network (ANN) was developed to estimate the ability of the food dyes to serve as simulants for PFAS by correlating up to sixteen molecular descriptors with publicly-available Kd values for PFAS. Based on experimental observations and ANN predictions, all three food dyes were found to be poor simulants for PFBA, PFBS, and PFHpA (i.e. food dyes Log Kd > PFAS Log Kd), but conservative simulants for PFOA, PFOS, and other PFAS species (i.e. food dyes Log Kd < PFAS Log Kd). To this author’s knowledge this is the first research to combine the use of machine learning and experimental observations to characterize PFAS simulants.

Page Count

168

Department or Program

Department of Earth and Environmental Sciences

Year Degree Awarded

2024

Download


Share

COinS