Predicting the Thermal Stability of Nitroaromatic Compounds Using Chemoinformatic Tools

Abstract

In the framework of the European REACH regulation major attention was recently devoted to toxicological and ecotoxicological problems while little attention has been dedicated to other important applications concerning chemical hazards, for instance, explosive properties. In this work different chemoinformatic tools such as partial least squares, multilinear regressions, and decision trees have been used for the development of a novel quantitative structure-property relationships to predict the heat of decomposition of a series of nitroaromatic compounds. Models were conceived in order to follow the regulatory requirements according to OECD principles for the validation of QSAR methods. Three models derived with MLR, PLS and decision tree techniques were developed, validated (internally and externally) and their applicability domains have been defined and analyzed. All models proved to be reliable with remarkable robustness in terms of full cross-validation scheme and showed good predictive power toward the external validation set. These models also present a large applicability domain within nitrobenzene derivatives and are easy to implement and interpret in terms of subjacent mechanisms.

Structure‐Activity Relationship Analysis of the Thermal Stabilities of Nitroaromatic Compounds Following Different Decomposition Mechanisms

“…Furthermore, the best way to evaluate the predictive ability of a model is its validation by new compounds, which do not participate in the modeling process 16,33. Here we use the same training set and validation set as list in Table 4 of the literature 6. The training set samples are used to develop the QSAR model and the validation set is used to measure the predictive ability of the built model.…”

“…The safe usefulness of an explosive can only be appreciated when the properties and the factors affecting them are fully understood. Therefore, it is very important to know the thermal stability properties of materials, that plays a crucial role in the identification of potentially explosive substances 6…”

“…Recently, to obtain more general models, Fayet et al. investigated a sample of 77 nitrobenzene derivatives and published two papers2,6 deeply analyzing corresponding structure‐activity relationships by considering constitutional descriptors, topological indices, geometric descriptors and quantum chemical descriptors obtained by CODESSA 17. From these models, that have good fitting results but lower external predictivity, it is evident that it is very difficult to obtain a general predictive model because of the distinct decomposition mechanisms.…”

Development of validated QSPR models for impact sensitivity of nitroaliphatic compounds

“…Others were extracted from examination of chemical structures or from the works of Kamlet [10], Mullay [12], Badders [16] and Keshavarz [20]: oxygen balance (OB 100 ), molecular electronegativity ( mol ), the ratio of the number of atoms (C, H, O or N) on molecular weight (n C /mw, n H /mw, n O /mw, n N /mw) and the number of specific fragments (as n CNC ). Moreover, descriptors issuing from conceptual DFT [41,42], already successfully used in QSPR models for the prediction of the heat of decomposition of nitroaromatic compounds [26][27][28], were considered. More details about descriptors are available in [40] or in Supporting information (Tables S2 and S3).…”

“…About 400 descriptors were calculated, including quantum chemical ones based on density functional theory (DFT) calculations. These latter were performed to have a better chemical interpretation of the developed models as already demonstrated in previous works for nitroaromatic compounds [26][27][28].…”