Integrated Experimental and Software Methods for Non-Targeted Analyses Investigation of Vehicle-Derived Chemicals and Their Transformation Products

Abstract

Water pollution is a significant environmental issue that can yield detrimental impacts on human health and ecosystem. Compounding the basic environmental pollution problem is the fact that many chemicals can undergo various transformation reactions under environmental conditions to form structurally similar transformation products (TPs). Notably, some TPs may contribute significantly to the potential for adverse effects in environmental systems while their characteristics and transformations are not fully understood. For example, among various pollutant classes, vehicle-derived chemicals and their TPs are a growing concern, representing compounds that are concurrently abundant (by mass use), widespread, and poorly characterized or identified. For instance, our group recently identified 6PPD (N (1,3-dimethylbutyl)-N’-phenyl-p-phenylenediamine), a ubiquitously used tire rubber antioxidant, would be transformed under environmental conditions into 6PPDQ, a toxicant which is responsible for the coho salmon (Oncorhynchus kisutch) acute mortality observed for several decades in the Pacific Northwest. Therefore, there is a pressing need to chemically characterize and to investigate the environmental fate and transport of these classes of vehicle derived chemicals and their TPs for comprehensive environmental risk assessment, remediation and policy making. Problematically, identifying and tracking specific groups of pollutants or contaminant sources is often challenging due to the complex chemical matrices present in surface waters and the relatively narrow detection capabilities of traditional analytical methods (e.g., targeted analytical methods via low-resolution mass spectrometry) that focus on limited numbers of pre-defined analytes. Non-targeted analysis (NTA) can potentially address this challenge by coupling high-resolution mass spectrometry (HRMS) instrumentation with data science approaches for data analysis to better identify or quantify unknown pollutants or contamination sources in complex mixtures. However, the development of HRMS data processing workflows is still in a relatively early stage, is often labor intensive, and can lack well-established protocols and integrated data processing capabilities, especially for open-source software, for environmental data and systems. To address the challenges above, this thesis communicates the development of improved integrated experimental and open-source software methods for NTA using HRMS instrumentation and capabilities. Subsequently, these methodologies were deployed for the chemical characterization of vehicle-derived chemicals, focusing on the industrial antiozonant 6PPD widely used in tire rubbers and related transformation products. Chapter 1 presents an introduction to these systems. Chapters 2-4 communicate the analysis and characterization of 6PPD and related TPs. 6PPD TPs include 6PPD-quinone (6PPDQ), which is an emerging contaminant that was previously identified as the “primary causal toxicant” for acute mortality events in stormwater-exposed coho salmon. Given its extraordinarily high toxicity, interest in 6PPDQ properties and fate extends from local to global scales. Experiments were conducted to (a) investigate the physiochemical properties of 6PPDQ; (b) measure 6PPD ozonation dynamics and quantify known TPs (i.e., 6PPDQ); (c) prioritize other potential 6PPD TPs with HRMS-based NTA; and (d) evaluate the environmental fate of 6PPD and formation of TPs under varied environmental conditions (ozone exposure, aerobic and anaerobic conditions) while quantifying 6PPD and TPs with liquid chromatography-tandem mass spectrometry (LC MS/MS). Chapters 5-6 focus on method development for HRMS NTA, including workflow development for NTA data processing and development and optimization of source identification and apportionment methodologies. Chapter 1 provides an introduction and explains the research goals of the thesis. Chapter 2 discusses the investigation of the physiochemical properties of 6PPDQ, including logKow, solubility, leaching potentials, sorption potentials and aqueous stability. We focused on reporting chemical characteristics relevant to the fate and transport of the recently discovered environmental toxicant 6PPDQ. The aqueous solubility and octanol–water partitioning coefficient (logKow) for 6PPDQ were measured as 38 ± 10 μg L−1 and 4.30 ± 0.02, respectively. Within the context of analytical measurement and laboratory processing, sorption to various laboratory materials was evaluated, indicating that glass was largely inert but loss of 6PPDQ (including non-recoverable mass) to other material types was common, including materials commonly found in the laboratory. Aqueous leaching simulations from tire tread wear particles (TWPs) indicated short-term release of ∼5.2 μg 6PPDQ per gram TWP over 6 h under flow through conditions. Aqueous stability tests observed a slight-to-moderate loss of 6PPDQ over 47 days (26 ± 3% loss) for pH 5, 7, and 9. These measured physicochemical properties suggest that 6PPDQ is generally poorly soluble (almost surprisingly so) but fairly stable over short time periods (~5% loss over 3d) in simple aqueous systems. 6PPDQ can also leach readily from TWPs for subsequent environmental transport, posing high potential for adverse effects in local aquatic environments. Chapter 3 evaluated the transformation kinetics of 6PPD degradation and 6PPDQ formation using heterogeneous 6PPD ozonation in the gas phase. We investigated TP formation occurring during heterogeneous reaction of gas-phase ozone with 6PPD; exposures included both pure 6PPD solids and TWP rubber particles that contained 6PPD. Oxidative transformation occurred during these ozonation conditions (inlet O3 concentration ∼360 ppbv), with up to 81% of 6PPD mass reacting over 6 h. Conversion of 6PPD to 6PPDQ was confirmed at a 9.7% molar yield for pure 6PPD solid and 0.95% molar yield for 6PPD present within TWPs, representing likely minima over these time scales and conditions. Leveraging HRMS suspect screening approaches, we identified 19 probable 6PPD-derived TPs in both ozonated 6PPD and TWP samples, underscoring formation of diverse TPs from this antioxidant compound. By screening environmental samples, nine 6PPD-derived TPs were subsequently detected within roadway runoff to communicate environmental relevance. The data confirms that when tire rubber antioxidants react with ozone, as intended, they form and release various TPs to surrounding environments with high potential to impair water quality. Extending from Chapter 3, Chapter 4 evaluates 6PPD fate and transformation under additional model environmental conditions. Different reaction conditions were tested to comparatively investigate 6PPD transformation processes, including gas phase ozonation, gas phase and aqueous phase aerobic exposures. The gas phase ozonation experiments included conditions that were not evaluated in Chapter 3. During these experiments, 6PPD mass reacted progressively during ozone exposure under tested conditions (93±2% 6PPD degraded during 2.5 h; inlet [O3] = 360 ppbv). The aqueous phase ozonation experiment determined the apparent reaction rate constant of 6PPD with molecular O3 as (1.18±0.16)×106. Additionally, 6PPD environmental transformation under aerobic conditions in gaseous and aqueous phases suggested 6PPD is prone to transformation in water (half-life: 10.0 h) and fairly stable under gas phase aerobic exposures if they lack or only contain trace molecular ozone (half-life: 69.3 days). The formation of 6PPD TPs was evaluated during all these exposure conditions. Among all the tested conditions, 6PPDQ formation was only observed from 6PPD in gaseous systems where ozone was clearly present. The results suggested that 6PPD-gaseous O3 interaction is the primary formation pathway of 6PPDQ under ambient environmental conditions, while the environmental fate, transport, and potential effects of other, still poorly understood 6PPD-TPs merit further investigation. Chapter 5 describes the development of an open-source Python package Mass-suite (MSS) as a full, expandable open-source HRMS data analysis toolbox with multiple NTA data processing capabilities. MSS provides flexible, user-defined workflows for HRMS data processing and analysis, including both basic functions (e.g., feature extraction, data reduction, feature annotation, data visualization, and statistical analyses) and advanced exploratory data mining and predictive modeling capabilities that are not provided by currently available open source software (e.g., unsupervised clustering analyses, a machine learning-based source tracking and apportionment tool). As a key advance, most core MSS functions are supported by machine learning algorithms (e.g., clustering algorithms and predictive modeling algorithms) to facilitate function accuracy and/or efficiency. MSS reliability was validated with mixed chemical standards of known composition, with 99.5% feature extraction accuracy and ~52% overlap of extracted features relative to other open-source software tools. Example user cases of laboratory data evaluation are provided to illustrate MSS functionalities and demonstrate platform reliability. MSS expands available HRMS data analysis workflows for water quality evaluation and environmental forensics and is readily integrated with existing database and screening platforms. Though fully described in Chapter 5, applications of some MSS functions are also described in Chapters 3 and 4 to facilitate data analysis of complex chemical mixtures. Chapter 6 focused on the development and optimization of HRMS data processing workflows with machine learning methodologies. Different algorithms and methodologies were trained with ~550 data points and optimized for source identification and apportionment predictions. Both qualitative and quantitative models were leveraged to facilitate the data analysis and enable insight into NTA data tools that do not rely on feature identification. Dot product-logistic regression (100% Acc.) and XGBoost classification model (93.3% Acc. on training data) were trained for source identification. For source quantification method development, various strategies were evaluated to mitigate the potential risks of overfitting derived from relatively scarce sample data points (n=~550) versus substantial detection numbers of unique chemical features (n=~3000). The methods tested include feature selection (XGBoost), feature extraction (similarity scores and principal component analysis) and regularization regressions (lasso, ridge and elastic net regressions). The results suggested that regularization models outperformed other models on source apportionment predictions (with the highest R2 scores of predicted and actual value), while PCA (Principal Component Analysis) dimension reduction coupled with SVM (Support-Vector Machine) regression is a strong candidate data processing strategy that merits further training and optimization efforts for source apportionment applications. These research findings provide valuable insights for researchers, policymakers, and regulatory agencies in understanding and mitigating the impact of emerging contaminants on human and environmental health. Leveraging the use of targeted analysis approaches, the transformation dynamics and the environmental fate and transport of 6PPD and 6PPD-derived TPs were further characterized by these studies. On the other hand, the data processing workflow described here provides a workflow and developmental template for future non targeted analysis investigations of complex chemical mixtures in aquatic environments, especially for source tracking studies. Future investigations of other related emerging contaminants (i.e., especially other PPD family chemicals) and their transformation products are essential for improving the overall risk assessment of these types of pollutants in aquatic environments. Further development and optimization of these analytical workflows, along with more sample collection and analysis efforts, is needed to fully optimize potential HRMS-NTA workflows for contamination source identification and quantification for different source and background matrices. In conclusion, this research provides critical information to develop management strategies for emerging contaminants derived from 6PPD used in vehicle tires, as well as develop novel HRMS methodologies for source tracking studies.