Prediction of the temperature dependency of Henry's law constant from chemical structure

A new model to estimate the temperature dependency of Henry's law constant in water for organic compounds from the two-dimensional structure is presented. Air/water partition enthalpies of 456 chemicals were fitted to 46 substructural parameters with a squared correlation coefficient r2 of 0.81 and a standard error of 7.1 kJ/mol. The compound set covers various organic compound classes with the atom types C, H, N, O, F, Cl, Br, I, and S. Application of the model together with experimental data for 25 degrees C to a set of 462 compounds with 2119 experimental Henry's law constants at temperatures below 20 degrees C yields a predictive squared correlation coefficient q2 of 0.99 and a standard error of 0.21 logarithmic units. The prediction capability is further evaluated using cross validation and permutation.