Modeling the pH and temperature dependence of aqueousphase hydroxyl radical reaction rate constants of organic micropollutants using QSPR approach

Abstract

Designing of advanced oxidation process (AOP) requires knowledge of the aqueous phase hydroxyl radical (OH) reactions rate constants (k ), which are strictly dependent upon the pH and temperature of the medium. In this study, pH- and temperature-dependent quantitative structure-property relationship (QSPR) models based on the decision tree boost (DTB) approach were developed for the prediction of k of diverse organic contaminants following the OECD guidelines. Experimental datasets (n = 958) pertaining to the k values of aqueous phase reactions at different pH (n = 470; 1.4 × 10 to 3.8 × 10 M s) and temperature (n = 171; 1.0 × 10 to 2.6 × 10 M s) were considered and molecular descriptors of the compounds were derived. The Sanderson scale electronegativity, topological polar surface area, number of double bonds, and halogen atoms in the molecule, in addition to the pH and temperature, were found to be the relevant predictors. The models were validated and their external predictivity was evaluated in terms of most stringent criteria parameters derived on the test data. High values of the coefficient of determination (R ) and small root mean squared error (RMSE) in respective training (> 0.972, ≤ 0.12) and test (≥ 0.936, ≤ 0.16) sets indicated high generalization and predictivity of the developed QSPR model. Other statistical parameters derived from the training and test data also supported the robustness of the models and their suitability for screening new chemicals within the defined chemical space. The developed QSPR models provide a valuable tool for predicting the OH reaction rate constants of emerging new water contaminants for their susceptibility to AOPs.

QSAR model and microscopic mechanism analysis of dye removal by coagulation of aluminum chloride under alkaline conditions

“…Frontiers in Environmental Science frontiersin.org both >0.5, the developed model indicated robust and predictive (Gramatica, 2007). Finally, Y-randomization and applicability domain tests were performed for the optimal model to verify its stability and credibility (Melagraki & Afantitis, 2014;Gupta & Basant, 2017;Ortiz et al, 2017).…”

Do We Still Need a Laboratory to Study Advanced Oxidation Processes? A Review of the Modelling of Radical Reactions used for Water Treatment

“…between contaminants and hydroxyl radicals [51][52][53], chlorine, chlorine dioxide, ferrate [49], ozone [48,49], Fenton process [54], tetra-amido macrocyclic ligand/H 2 O 2 [55], peroxone [56] or electro-peroxone [47]. Modelling the kinetic rate constants for the reactions of •OH with contaminants was also presented on pH-and temperature-dependent models based on the decision tree boost [57].…”

Supervised Machine Learning Algorithms for Predicting Rate Constants of Ozone Reaction with Micropollutants

“…In this regard, the QSPR method based on the concept that the molecular properties/activities are determined by their structures has been gaining popularity . Thus, far, numerous QSPR models have been developed for evaluating the aqueous phase hydroxyl radical involved reaction rate constants, ,− while QSPR models built for calculating the ozonation second-order rate constant ( k O3 ) are relatively few. ,, Recently, our group proposed a method used combinational quantum chemical and norm descriptors to develop a QSPR model for k • OH prediction, the model showed very promising results, which provides a good thought to develop models for evaluating the k O3 values.…”