Prediction of the physico-chemical properties of nitroaromatic compounds using QSPR models

摘要

The evaluation of the explosibility of chemical substances essentially relies on the use of experimental tests according to international regulatory schemes (UN Manual of Tests and Criteria of the Recommendations on the transport of dangerous goods, regulation (EC) N°440/2008). The recent evolution of the European regulatory framework related to chemicals (REACH, CLP) implies a quantity of works incompatible with a complete systematic experimental characterization of hazardous properties (for reason of time, cost or availability of products). For example, more than 140 000 existing substances could be under concern in the registration process of REACH. Moreover, the explosive intrinsic property of a substance ranks at the top of physico-chemical hazards that may be feared from the use of a given chemical. So, the development of methods allowing the identification of this hazardous property on the basis of chemical structures is of great interest not only for existing substances but also at the R&D stage in the development of new products. For these reasons, INERIS, in collaboration with Chimie ParisTech, develops new predictive models (alternative or complementary to experimental approaches) for the evaluation of explosibility properties of hazardous substances. The development of these predictive models lies on an original method combining the statistical tools used in quantitative structure property relationship method (QSPR), i.e. multilinear regressions, principal component analyses or decision trees, with quantum chemical calculations. The proposed contribution presents such structure-property analyses for the development of models dedicated to the prediction of two properties of potentially explosive nitroaromatic compounds (heat of decomposition and electric spark sensitivity) taking into account the requirements for their use within a regulatory context, including not only their statistical validations but also their chemical interpretations.