PARENT SESSION

1I - Physicochemical properties
Poster Hall
8:30 AM - Monday, 28 April 2003
Chair: Finizio, A.¹, ¹

(MOP/49) Estimation of the soil sorption coefficient from chemical structure.

Schüürmann, Gerrit¹, Ebert, Ralf-Uwe¹, Huuskonen, Jarmo², Kühne, Ralph¹, ¹ Department of Chemical Ecotoxicology, UFZ Centre for Environmental Research Leipzig-Halle, Leipzig, Germany² Uudenkaupungin Uusi Apteekki, Uusikaupunki, Finland

ABSTRACT- In environmental fate modeling, the partitioning of chemicals between soil/sediment and water may be estimated from the soil sorption coefficient K_oc. In contrast to other compound properties such as water solubility or vapour pressure, there are much less experimental data available. Furthermore, currently available log K_oc estimation methods are somewhat limited in performance and application range. The paper provides a new technique for the calculation of log K_oc from molecular structure, based on experimental data from literature for 571 organic compounds that cover a wide range of chemical classes. Model fit was confined to 80% of these data (457 compounds), whilst 20% (114 chemicals) served as test set to evaluate the prediction capability of the new model. The final model contains 30 descriptors: molecular weight, 2 topological descriptors, 26 substructure correction factors, and a general regression constant. The resultant standard error is about 0.5 logarithmic units. With the test set, very similar statistics are achieved, which confirms the prediction power of the new calculation scheme. Comparison with the performance of existing methods that involve log K_ow, topological indices and fragments demonstrates a superiority of the newly derived model, which holds true particularly also for the test set that was not part of the regression.

Key words: Estimation method , Soil sorption coefficient , Organic compounds