부산대학교 도서관

Due to mobility and specific toxic actions of the ionic and non-ionic micropollutants in the aquatic environment, the micropollutants should be removed before being discharged. Predicting the adsorption properties of micropollutants onto carbon-based materials, i.e., activated charcoal (AC), multi-walled carbon nanotubes (MWCNTs), and activated carbon fiber (ACF), and understanding the adsorption mechanisms are important for assessing the environmental fate and transport of micropollutants and adsorbents.In Chapter 3, the QSAR model development and validation methods were introduced. High-performance liquid chromatography (HPLC) analysis methods for micropollutants were described. Physicochemical properties and linear free energy relationship (LFER) descriptors of micropollutants including pharmaceuticals and personal care products (PPCPs), endocrine-disruption chemicals (EDCs), and other neutral emerging compounds were summarized. Characterization of AC, MWCNTs, and ACF were performed by N2 adsorption and desorption isotherms and zeta-potential analysis.In Chapter 4, the adsorptive interactions between cationic pharmaceuticals and AC were investigated. Isotherm experiments were performed and then plotted by the Langmuir model to determine the adsorption affinity (Kad,n) and capacity (qm). Afterwards, to interpret the adsorption behaviors, two simple prediction models were developed based on quantitative structure-activity relationships (QSAR). In the modeling, molecular weight (MW), polar surface area (PSA), and octanol-water partitioning coefficient (log P) were used as model parameters. In the results, the combinations of these three parameters could predict the adsorption affinity and capacity in R2 of 0.85 and 0.80, respectively. The robustness of models was validated by leave-one-out cross-validation (Q2LOO) and the estimated Q2LOO values were 0.60 and 0.55 for the adsorption affinity and capacity, respectively, which are higher than the acceptability of standard i.e., 0.5.In Chapter 5, adsorptive interactions between anionic pharmaceuticals of nonsteroidal anti-inflammatory drugs (NSAIDs) and AC were estimated by batch isotherm experiments, and the interactions were modeled based on QSAR concept. Experimental results showed that AC had a high qm (0.38-0.67 mmol g-1) and Kad,n (14.03-930.8 L mmol-1) for the selected NSAIDs. In QSAR modeling, LFER descriptors of excess molar refraction (E), dipolarity/polarizability (S), and Coulombic interactions of anions (J−) were highly related to log qm, and the combination of the three terms could predict log qm in R2 of 0.97 and SE of 0.015 log unit. In the case of Kad,n, only single B term showed a good correlation with log Kad,n in R2 of 0.81. Additionally, the combination of hydrogen bonding acceptors (HBAs) and molar volume (MV), which are easily calculable parameters, could also derive good predictability in R2=0.81 and SE=0.26 log unit. Afterwards, validation of the QSAR models based on the leave-one-out cross-validation method (Q2LOO) showed that the models were acceptable.In Chapter 6, thirty compounds constituting cationic, anionic, and neutral were used, and the effects of diverse intermolecular interactions were captured by isotherm fitting of the Langmuir model. The predicted equation of distribution coefficient (Kd), adsorption capacity (qm), and adsorption affinity (Kad,n) of activated charcoal (AC) for compounds was developed. For the Kd values, an activity-dependent LFER model was developed, and could predict log Kd within R2 of 0.85. The determined system parameters and their statistical results revealed that the molecular volume (V) of compounds encouraged the adsorption from water to AC, whereas the hydrogen basicity (B) discouraged such adsorption. In the cases of qm and Kad,n, AC had relatively good capacities for neutrals, which were higher than those for cations and anions. Whereas, AC had stronger adsorption affinities with anions, compared to those for neutrals and cations. The phenomena were explained by the developed LFER models, in which the p-/n-electron interaction (E) made a great positive contribution to increase the adsorption capacity and affinity, while Coulombic interactions (J+ and J) demonstrated a negative trend. Finally, the developed models can predict the qm and Kad,n in R2 of 0.86 and 0.76, respectively.In Chapter 7, the adsorption affinities of thirty-one micropollutants in ionic and non-ionic forms onto non-functionalized and functionalized MWCNTs with -COOH, -OH, and -NH2 functional groups were examined. Then, the property of each of adsorbents was predicted based on LFER model. In experimental results, it showed that both non-functionalized and functionalized have high adsorption affinities for ionic and neutral compounds, but the property was dependent on the functionalities on the surface of MWCNTs. In the modeling analysis, it was shown that the LFER model can predict the adsorption affinities in high accuracy of R2 in range of 0.81-0.91. Based on the developed LFER models, the adsorption mechanisms and the contribution of individual intermolecular interaction to the overall adsorption were interpreted. Molecular interactions induced by molecular volume (V) and hydrogen-bonding basicity (B) have predominant contributions to the adsorption on non-functionalized MWCNTs, while in the case of functionalized MWCNTs, Coulombic interactions by the charges (J+ and J–) are important factors.In Chapter 8, the adsorption properties of ACF were characterized using batch isotherm experiments and LFER modeling analysis. For the experiments, adsorption affinities of thirty-five chemicals, i.e., pharmaceuticals and endocrine-disrupting chemicals in cationic, anionic, and neutral forms, on ACF were estimated. Afterwards, the experimental data were used as dependent variables in LFER modeling. Three isolated models for each of chemical species, i.e., cations, anions, and neutral compounds, and a comprehensive model for whole dataset were developed. The LFER results revealed that the models for anionic and neutral compounds have high predictabilities in R2 of 0.97 and 0.96, respectively, while that for cations has slightly lower R2 of 0.72. In the comprehensive model including cationic, anionic, and neutral compounds, the accuracy of it is 0.81. From the developed LFER model based on the whole dataset, the adsorption mechanisms of ACF for the selected substances could be interpreted, in where the terms of hydrophobic interaction (V), hydrogen bonding basicity (B), and anionic Coulombic force (J–) of the compounds were identified as predominant interactions with ACF. Finally, the adsorption interactions between micropollutants and carbon-based materials were clearly clarified based on the experimental methods and modeling analysis in this thesis.