A robust model for estimating thermal conductivity of liquid alkyl halides
Thermal conductivity is an essential thermodynamic property in chemical engineering application. As a result, estimating the thermal conductivity of organic compounds is of significance in industry production. Alkyl halides are important organic intermediates and raw materials, but little investigations have been performed to estimate their thermal conductivity. In this study, the structures of compounds were optimized in Gaussian 09W and molecular descriptors were extracted by Dragon software. Finally, we developed a 6-descriptor linear quantitative structure–property relationship (QSPR) model to estimate the thermal conductivity of alkyl halides using the genetic function approximation (GFA) method. Validation proved that the developed model had goodness-of-fit, robustness and predictive ability. The r2pred and root-mean-square error (RMSEP) of prediction set for the model were equal to 0.9745 and 0.0035, respectively. Meanwhile, the applicability domain was visualized by means of the Williams plot. This study provides a new model for estimating the thermal conductivity of this important class of chemicals.