US EPA

3400664 

Journal Article 

Modeling the reactivities of hydroxyl radical and ozone towards atmospheric organic chemicals using quantitative structure-reactivity relationship approaches 

Gupta, S; Basant, N; Mohan, D; Singh, KP 

2016 

Yes 

Environmental Science and Pollution Research
ISSN: 0944-1344
EISSN: 1614-7499 

23 

14 

14034-14046 

English 

27040550 

10.1007/s11356-016-6527-2 

WOS:000379553500040 

The persistence and the removal of organic chemicals from the atmosphere are largely determined by their reactions with the OH radical and O3. Experimental determinations of the kinetic rate constants of OH and O3 with a large number of chemicals are tedious and resource intensive and development of computational approaches has widely been advocated. Recently, ensemble machine learning (EML) methods have emerged as unbiased tools to establish relationship between independent and dependent variables having a nonlinear dependence. In this study, EML-based, temperature-dependent quantitative structure-reactivity relationship (QSRR) models have been developed for predicting the kinetic rate constants for OH (kOH) and O3 (kO3) reactions with diverse chemicals. Structural diversity of chemicals was evaluated using a Tanimoto similarity index. The generalization and prediction abilities of the constructed models were established through rigorous internal and external validation performed employing statistical checks. In test data, the EML QSRR models yielded correlation (R (2)) of ≥0.91 between the measured and the predicted reactivities. The applicability domains of the constructed models were determined using methods based on descriptors range, Euclidean distance, leverage, and standardization approaches. The prediction accuracies for the higher reactivity compounds were relatively better than those of the low reactivity compounds. Proposed EML QSRR models performed well and outperformed the previous reports. The proposed QSRR models can make predictions of rate constants at different temperatures. The proposed models can be useful tools in predicting the reactivities of chemicals towards OH radical and O3 in the atmosphere. 

Hydroxyl radical; Ozone; Organic chemicals; Chemical reactivities; Ensemble machine learning