Sorption of perfluoroalkyl substances and Fluoroquinolone antibiotics in matrices of environmental interest

Abstract

Perfluoroalkyl substances (PFASs) and fluoroquinolone antibiotics (FQs) are pollutants of environmental interest due to their widespread presence in environmental compartments and their high toxicity and disrupting effects of original ecosystems, respectively. After a contamination event, an evaluation of the potential threat to humans and ecosystems should be performed, a process which is known as risk assessment. To properly evaluate the risk, several data and parameters are required, including the mobility of the contaminant along environmental matrices. The sorption and desorption processes of a pollutant between the solid and liquid phases governs its environmental mobility, and understanding the sorption interaction mechanisms between sorbents and pollutants is important to extrapolate conclusions in other untested but characterized scenarios. If risk assessment studies evidence that intervention actions should be implemented, remediation actions may be taken, as using sorbent materials, such as biochars and activated carbons for contaminated soils and waters. A quantitative parameter used in risk assessment models is the solid-liquid distribution coefficient (Kd). Thus, it is necessary to develop Kd prediction models based on key pollutant and sorbent physicochemical properties descriptive of the interaction mechanisms involved in the sorption process, which in turn may need the creation of robust, critically-reviewed Kd compilation datasets. Current Kd prediction models for PFAS and FQs in soil pure components, soils and/or carbon-rich materials are scarce, have limitations and require a wider range of applicability. Considering this, the main goal of this thesis has been to identify key parameters affecting PFAS and FQs sorption and to develop Kd prediction models in soils and related environmental matrices. Regarding the identification of key parameters governing PFAS sorption in soils and other environmental matrices, it has been shown that sorption of PFAS in soils is largely governed by soil organic carbon content through hydrophobic interactions, increasing with PFAS chain length, although the contribution of the mineral phase can play a significant role in sorption for those soils with a low organic carbon content. Accordingly, a simple Kd prediction model applicable to a wide range of soils and PFAS has been developed. Additionally, a simple Kd prediction model for PFAS in biochars and activated carbons applicable to a wide range of material properties and PFAS has also been developed. In both cases validation procedures confirmed their effectiveness, overcoming previous limitations observed in previous models reported in the literature. Besides, it was revealed that sorption of FQs in both humic acids and phyllosilicate minerals is strong, although highly pH-dependent due to both the speciation of the FQs and the overall surface charge of the phase, and sorption is also dependent on clay nature. On the other hand, sorption of FQs in metal oxides is dependent on the metal nature, but for a specific set of metal oxides sorption increases when increasing mineral surface area. The identification of key properties affecting FQ sorption in pure soil components allowed to deduce the main properties responsible for sorption in soils: it was confirmed that sorption of FQs in bulk soils is pH- dependent, being soil pH, clay and organic matter content key parameters descriptors of sorption, besides demonstrating the sorption analogy of several widely studied FQs. Additionally, several empirical equations have been proposed to predict the Kd of FQs in pure soil components. Specifically, the sorption Kd values in humic acids and different phyllosilicate minerals at different pH values were relatively well described by a second-grade polynomic fitting, and besides, a linear model was proposed to predict Kd values for FQs in metal oxides and (hydro)oxides of Fe at neutral pH according to its surface area. Moreover, an empiric Kd prediction linear model for FQs in soils was developed after multivariate analyses, although it revealed a poor prediction ability when tested against literature data due to the non-linear dependence of Kd as a function of pH. Alternatively, a set of best-estimate Kd values with associated uncertainty were proposed by refining an overall sorption Kd dataset for FQs in soils grouped according to specific key properties through cumulative distribution functions.

L’objectiu principal d’aquesta tesi és l’elucidació de les propietats fisicoquímiques més rellevants que afecten la sorció de substàncies perfluoralquíliques (PFASs) i antibiòtics de tipus fluoroquinolona (FQs) en diferents matrius ambientals (tals com sòls, components orgànics i inorgànics presents en el sòl i materials rics en carboni), així com desenvolupar models per predir- ne el coeficient de distribució sòlid-líquid (Kd) sobre les esmentades matrius. Aquests models poden ser útils per a posteriors estudis de risc ambiental després d’episodis de contaminació. En aquesta tesi s’ha identificat que la matèria orgànica governa la sorció de PFASs en sòls, i s’ha desenvolupat un model de predicció aplicable a PFAS de diferent llargada de cadena a partir de quantificar la sorció sobre els setis presents en les fases orgàniques i inorgàniques del sòl. De manera similar, s’han identificat les principals propietats fisicoquímiques responsables de la sorció de PFASs en materials rics en carboni, i s’ha desenvolupat un model de predicció aplicable a diferents PFAS i materials d’un ampli rang de propietats fisicoquímiques. També es discuteix la sorció de FQs sobre diferents fases pures orgàniques i inorgàniques presents al sòl (substàncies húmiques, òxids metàl·lics i minerals del tipus fil·losilicats), i es proposen diferents models de predicció de Kd basats en relacions univariants (per exemple, funció de pH o de la superfície específica del material) que permeten explicar una bona part de la variabilitat de les dades recopilades de la literatura. Aquest anàlisi permet la deducció de les principals propietats del sòl que afecten la sorció de FQs, on les diferents fases pures poden coexistir i on es poden donar múltiples interaccions a la vegada. En paral·lel, i després de demostrar l’analogia en el comportament de sorció de quatre FQs àmpliament estudiades, s’ha desenvolupat un model de predicció de Kd basat en les propietats del sòl més rellevants en la sorció, però degut a la seva no prou bona qualitat predictiva, alternativament, s’han deduït valors de Kd més probables, amb la seva incertesa associada, a través de funcions de distribució acumulada.