Predicting the Reactivity of Micropollutants with Ozone by Modeling their Reaction Rate Constants

摘要

The second-order rate constants of micropollutants are valuable information for assessing micropollutant removal from drinking water by ozonation. A Quantitative Structure Property Relationship model using a combination of piecewise linear regression and linear discriminant analysis (PLR-LDA) was developed to predict rate constants from structural characteristics. With a pre-defined breakpoint (logko3 = 2.00 M~(-1)s~(-1)), the PLR model using average molecular weight (AMW) and number of phenolic group (nArOH) as molecular descriptors shows very good results in terms of fitting to the training and the validation set. The PLR model explains more than 96% of the variation of the training set (n = 22, experimental data), and over 85% of the external validation set (n = 33, literature data). A discriminant analysis was carried out to develop a classification function for grouping compounds into one of the two groups, highreactive and low-reactive compounds. This function shows great classification ability in both training set and validation set. Overall, the PLR-LDA approach provides the means to model the ozone rate constants of various structural diverse compounds. These modeled rate constants can be used as an initial assessment to screen numerous organic micropollutants for their treatability with molecular ozone.