Prediction of thermal stability of some reactive chemicals using the QSPR approach

The reactivity hazard of reactive chemicals has been reported as one of the main causes for fire and explosion in process industries. The detected exothermic onset temperature (T_o) is one of the most important thermal stability parameters for risk assessment and safe management of reactive chemicals. The quantitative structure-property relationship (QSPR) methodology was applied to predict the T_o of 63 nitroaromatic compounds and 16 organic peroxides, respectively, from only their molecular structures. Various kinds of molecular descriptors were employed to characterize the molecular structures of reactive chemicals. The genetic algorithm combined with multiple linear regression (GA-MLR) is employed to select optimal subsets of descriptors that have significant contribution to the T_o property for nitroaromatic compounds and organic peroxides, respectively, to construct accurate and economic models to predict their thermal stability. The best resulted models for nitroaromatic compounds and organic peroxides are both five-parameter multilinear equations, with the coefficient of determination (R²) being 0.738 and 0.988, and the cross-validation coefficient (Q_LOO²) being 0.715 and 0.963, respectively. Model validations were performed to check the robustness, stability, and predictivity of the presented models. The results showed that both models are valid and with internal or external predictivities. Compared with the existing QSPR models, the proposed models are more accurate, computationally inexpensive, and simpler to apply. The proposed study can provide a new way to predict the T_o of reactive chemicals.