Predicting explosibility properties of chemicals from quantitative structure-property relationships

National audienceQuantitative Structure-Property Relationship (QSPR) type methods have been up to now mainly devoted to biological, toxicological applications but their use to predict physico-chemical properties is a growing interest. In this context, an original approach associating QSPR methods and quantum chemical calculations for the prediction of chemicals explosibility properties is presented here. Indeed, the new European regulation of chemicals named REACH implies the new assessment of a tremendous number of substances for their hazardous properties. But, the complete characterization of toxicological, ecotoxicological, and physico-chemical hazards at experimental level is incompatible with the imposed calendar of REACH. Hence, there is a real need in evaluating capabilities of alternative methods for assessing hazardous properties as a screening process. This contribution focuses on models that have been established to predict accurately the thermal stability and electric spark sensitivity of a series of potentially explosive nitroaromatic molecules. Descriptors related to their molecular structure (topological, geometrical, electronic, quantum chemical), partially obtained from density functional theory (DFT) calculations, were computed and statistical analyses (multilinear regressions) were performed to link the adequate molecular descriptors with the experimental properties. These first results coupling theoretical calculations and QSPR methods open new perspectives for the prediction of other physico-chemical propertie