Fate of novel brominated flame retardants in aquatic ecosystems: impact of chemical and substrate properties on partitioning

Abstract

Persistent organic pollutants (POPs) are legacy manufactured organic chemicals classified as such due to their longevity and ubiquity in the environment, high bioaccumulation potential, and adverse effects at environmentally relevant concentrations. POPs include pesticides, industrial chemicals and byproducts, flame retardants, and plasticizers, among others. Novel brominated flame retardants (NBFRs) replaced the legacy brominated flame retardants (BFRs) that had been classified as POPS with the intention that these would be less harmful to humans and the natural environment. This dissertation investigates the fate of NBFRs in aquatic ecosystems by measuring their lipophilicity and partitioning to dissolved organic matter (DOM). It also explores the role of chemical and DOM properties in partitioning. ☐ First, octanol-water partition coefficients (KOW) of several NBFRs were measured by reversed-phase high performance liquid chromatography (HPLC) and estimated by computational models. We observed that computational log KOW estimates differed between themselves by 1–3 orders of magnitude. However, the reliability of the computational models could be evaluated by their ability to predict the log KOW of chemicals with similar structures to the NBFRs. The best-performing computational models predicted log KOW for the calibration chemicals close to their known values with root mean square error (RMSE) = 0.224, and for the NBFRs that were close to those we measured (RMSE = 0.334). Additionally, we confirmed previous research that found planar chemicals exhibited different partitioning behavior to the HPLC column than nonplanar chemicals. ☐ Second, we measured the dissolved organic carbon-water partition coefficients (KDOC) of five NBFRs by the solubility enhancement method to dissolved organic matter (DOM) isolated from diverse aquatic and terrestrial sources. We also measured the molecular weight of the DOM by high performance size exclusion chromatography. DOM performs an important role in the fate of organic contaminants in aquatic ecosystems via partitioning processes. DOM can increase chemical mobility, reduce their bioavailability and effective toxicity, and mediate chemical transformation by catalyzing abiotic and photolytic reactions. We observed that linear free energy relationships (LFERs) which predicted KDOC from chemical KOW significantly overpredicted partitioning of NBFRs to aquatic DOM, while poly-parameter (pp-) LFERs predicted KDOC to within an order of magnitude for four of the five NBFRs studied. We attribute the underestimation of the fifth NBFR to a lack of training set compounds in the pp-LFERs that possessed similar molecular structures and physicochemical properties. Additionally, the three nonplanar NBFRs did not partition most strongly to the DOM with the highest molecular weight and aromaticity, as would be expected. This was the first study to measure the KDOC of these NBFRs to natural aquatic DOM, and we observed partitioning behavior that was not easily explained using traditional LFER models. ☐ Third, we expanded our scope to investigate the partitioning behavior of neutral organic chemicals to natural aquatic DOM and Aldrich Humic Acid (AHA), a commercial DOM manufactured from low-grade coal and often used as a substitute for natural DOM in partitioning studies. We collected experimental KDOC data and assembled a database representing 1648 measurements and 319 structurally diverse neutral organic chemicals. LFERs were derived for AHA based on KOW and Abraham parameters with RMSE of 0.650 and 0.561, respectively. In contrast, no single “universal” LFER could predict partitioning to aquatic DOM. We observed that partitioning was dependent on chemical planarity and molar volume, regardless of measurement method or DOM composition. Planar chemical partitioning positively correlated with chemical molar volume, while nonplanar chemical partitioning similarly increased with molar volume until a breakpoint of 198 cm3/mol, after which KDOC did not or only increased slightly with molar volume. Incorporating planarity and molar volume into our pp-LFER allowed the partitioning of all chemicals to be predicted with significantly improved accuracy (RMSE = 0.604) as compared to traditional LFER models. We posit that our observations for the aquatic DOM dataset are due to the molecular weight and molar volume of the larger and sterically hindered nonplanar chemicals approaching that of the DOM itself. At that point, the binding would be better described as association between co-solutes, rather than partitioning. ☐ Overall, these studies provide insight into the fate of novel brominated flame retardants in aquatic ecosystems by measuring their octanol-water and dissolved organic carbon-water partition coefficients. These partition coefficients offer insight into whether or not these NBFRs should be classified as persistent organic pollutants, as high KOW indicates a high bioaccumulation potential, while high KDOC indicates that DOM may act as a “shield” and reduce the bioavailability of the freely dissolved NBFR to aquatic organisms. Finally, the examination of all available data on the partitioning of neutral organic chemicals to DOM revealed that partitioning was dependent on chemical planarity and molar volume. This new LFER may be used to incorporate DOM partitioning into computational models in order to more accurately predict the fate of NBFRs and other neutral organic chemicals in aquatic ecosystems and evaluate their potential as POPs.