Photodegradation of micropollutants in water by UV/H2O2 and UV/persulfate

• Main Author: Zhang, Yiqing
• Other Authors: Lim Teik Thye
• Format: Theses and Dissertations
• Language: English
• Published: 2019
• Subjects: DRNTU::Engineering::Civil engineering
• Online Access: https://hdl.handle.net/10356/92074; http://hdl.handle.net/10220/48548
• Institution: Nanyang Technological University

The emerging micropollutants have posed serious threat to the ecological environment and human beings. Among them, cytostatic drugs and antibiotics have raised great public concern due to their frequent usage, high persistence, and biological stability towards conventional water treatment processes. Therefore, this study focused on the photodegradation of cytostatic drugs and amoxicillin using UV, UV/H2O2 and UV/persulfate (S2O82-, PS) processes. In addition, quantitative structure–property relationship (QSPR) method was applied to model the photodegradation efficiency of the emerging micropollutants by UV/H2O2 and UV/PS processes. In the first part of this study, direct and indirect UV photolysis were applied to investigate the degradation effect of eight cytostatic drugs using a pseudo first-order kinetic model. Six cytostatic drugs were degraded by less than 10% at UV dose of 400 mJ·cm-2, indicating the ineffectiveness of UV photolysis towards most of the investigated cytostatic drugs. The influence of water matrix components was investigated, including natural dissolved organic matter (DOM), bicarbonate (HCO3-), nitrate (NO3-), chloride (Cl-), and sulfate (SO42-). Treated water from a water treatment plant and secondary effluent from a wastewater treatment plant were employed as natural waters to evaluate the potential application of the processes. The primary photodegradation byproducts of cytostatic drugs were identified to investigate the degradation mechanism of the processes. In the second part of this study, the reaction kinetics, degradation pathways, and antibacterial activity of UV/H2O2 and UV/PS systems were compared using amoxicillin as a typical antibiotic. The extensive use of non-metabolized amoxicillin has led to the contamination of the aquatic environment. UV irradiation alone shows negligible effect on amoxicillin degradation; while the degradation efficiency of amoxicillin increases significantly with the addition of oxidant due to the generation of radical species. The second-order rate constant of amoxicillin with HO• and SO4•- are 3.9 × 109 M-1 s-1 and 3.5 × 109 M-1 s-1, respectively. In the UV/PS system at neutral pH, the contributions of UV, HO•, and SO4•- for amoxicillin degradation are 7.3%, 22.8%, and 69.9%, respectively. Based on the experimental evidence substantiated with theoretical calculations, the degradation pathways of amoxicillin in the UV/H2O2 and UV/PS systems were proposed, including hydroxylation (+16 Da), hydrolysis (+18 Da), and decarboxylation (-44 Da). The frontier electron density of amoxicillin was calculated to predict the susceptible regions for HO• and SO4•- attack. The antibacterial activity of amoxicillin solution decreases significantly after applying UV/H2O2 or UV/PS processes. In the third part of this study, QSPR method was developed to model the second-order hydroxyl radical and sulfate radical rate constants ( and ) for a wide range of micropollutants. The modeling was conducted by a sequential process of data collection, descriptor filtration, model optimization, and model validation. 128 micropollutants were collected from the published papers and randomly divided into the training, validation, and test set. 5286 descriptors were generated from the software to develop the model. Multiple linear regression (MLR) models show low predictivity in this study. In comparison, artificial neural network (ANN) and deep neural network (DNN) models can predict the dataset accurately. Three descriptor filtration approaches were evaluated and compared, including non-dimension-reduction approach, correlation analysis approach, and principal component analysis approach. The most predictive models were validated using application domain and visualized in the Williams plot. With outliers identified and removed, the and models developed in this study show good robustness and high reliability, indicating that ANN and DNN methods can be used to evaluate the degradation of micropollutants by HO• and SO4•- in water. The selected descriptors for model indicate three reaction pathways for HO• oxidation with chemicals, namely hydrogen atom abstraction, radical addition, and electron transfer. In comparison, the electrophilic radical SO4•- tends to react with electron-rich moiety groups directly through electron-transfer mechanism.