Influence of chemical structure of organic micropollutants on the degradability with ozonation

•Qualitative interpretation of influencing structures on degradability by ozonation.•Less polarizable molecules are degraded faster within ozonation.•Heteroatoms with lone e− contributing to resonance increase reactivity by ozonation.•Number of fused rings increase degradability by ozonation.•Substructures such as formamide group lower reactivity towards ozone. The increasing environmental problems due to various organic micropollutants in water cause the search of suitable additional water treatment methods. Gaining experimental data for the large amount and variety of pollutants would consume a lot of time as well as economic and ecologic resources. An alternative approach is predictive quantitative structure-property relationship (QSPR) modeling, which establishes a correlation between the structural properties of a molecules with a biological, physical, or chemical property. Therefore, in this study, QSPR modeling has been conducted using extensive validation techniques and statistical test to investigate the structural influence on the degradability of organic micropollutants with ozonation. In contrast to most of the other studies, the underlying dataset - rate constants for 92 organic molecules - were obtained under standardized conditions with defined experimental parameters. QSPR modeling was executed using a combination of the software PaDEL for descriptor calculation and QSARINS for the modeling process respecting all five OECD-requirements for applicable QSAR/QSPR-models. The final model was selected using a multi-criteria decision-making tool to evaluate the model quality based on all calculated statistical quality parameters. The model included 10 selected descriptors and fingerprints and showed good regression abilities, predictive power, and stability (R² = 0.8221, CCCtr = 0.9024, Q²loo = 0.7436, R²ext = 0.8420, Q²F1 = 0.8104). The applicability domain of the QSPR model was defined and an interpretation of selected model descriptors has been connected to previous experimental studies. A significant influence of the interpretable descriptors was put into experimental context and compared with previous studies and models. For example, the molar refractivity as a measure of size and polarizability of a molecule and the occurrence of important substructures such as a formamide group seem to decrease the removal rate constant. The contribution of lone electrons entering into resonance as well as the occurrence of fused rings were identified as i