Predicting reaction rate constants of ozone with ionic/non-ionic compounds in water

Area of Science:

• Environmental Chemistry
• Water Treatment Technologies

Background:

• Ozonation is crucial for water pollutant mitigation.
• Experimental ozone reaction rate constants (kO3) are scarce for many chemicals.
• Ionizable compounds exhibit varying reactivity with ozone based on their ionization state.

Purpose of the Study:

• To develop predictive models for ozone reaction rate constants (kO3).
• To establish quantitative structure activity relationship (QSAR) models for both non-ionic and ionic species.
• To address the data gap in experimental kO3 values.

Main Methods:

• Developed two QSAR models: one for non-ionic species using Partial Least Squares (PLS) and another for ionic species using Support Vector Machine (SVM).
• Utilized large datasets comprising 324 non-ionic states and 188 ionic states.
• Validated models for fitting, robustness, and predictive performance.

Main Results:

• Achieved high fitting (R²tr > 0.760 for non-ionic, > 0.780 for ionic) and predictive performance (R²ext > 0.760 for non-ionic, > 0.810 for ionic).
• Demonstrated robustness with Q²CUM > 0.700 and a wide applicability domain for both models.
• Identified significant differences in molecular parameters between the non-ionic and ionic models, reflecting distinct reactivity.

Conclusions:

• The developed QSAR models effectively predict kO3 for diverse chemical species.
• Combining the models allows estimation of overall kO3 at specific pH values.
• These predictive tools can enhance pollutant conversion rate assessment in water treatment processes.