Nonadditive effects in the partitioning behavior of various aliphatic and aromatic molecules

Abstract

Predicting the partition behavior of bifunctional molecules from their molecular structure is a challenge because the combination of two or more functional groups often has nonadditive effects. Data presented here and in the literature reveal that isomers of bifunctional compounds can exhibit partition constants that differ by several orders of magnitude. These effects are not limited to compounds with intramolecular H-bonds. For aliphatic molecules, large effects are found for diones and diesters but not for dioles. For aromatic molecules, large effects are found for compounds with intramolecular H-bonds and also for p-isomers with a strong delocalisation of pi-electrons over both functional groups as in p-nitroaniline and p-nitrophenol. Interestingly, these nonadditive effects apply not only to the specific interactions but also to the van der Waals interactions of a molecule. This diversity of possible nonadditive effects makes it difficult to predict partitioning of such polyfunctional molecules. Our results suggest that successful models require either an extensive number of correction factors or a quantum chemical approach. Predicting partitioning of bifunctional molecules to environmental systems might face an additional complication. Unlike with solvents, steric limitations of natural phases may prevent them from forming multiple H-bonds with bifunctional molecules. Our experimental results, however, indicate that this situation does not occur, as the H-bond-interaction behavior of bifunctional molecules in humic matter and on quartz was the same as it was in various solvent systems.

Prediction of partition coefficients for complex environmental contaminants: Validation of COSMOtherm, ABSOLV, and SPARC

“…The best prediction accuracy was achieved for PDMS/water (RMSE 1.43 log units) and ethylene glycol/1,2-dichloroethane (RMSE 1.55 log units), whereas the 2 other systems exhibited significantly higher RMSE values. This relatively low performance of SPARC stands in contrast to our earlier experience, in which SPARC had performed similarly well compared with COSMOtherm for systems such as octanol/water and air/water [6,13,37]. The results of the present study suggest that SPARC's prediction accuracy for complex chemicals is substantially lower than that for simple chemicals.…”

“…This relatively low performance of SPARC stands in contrast to our earlier experience, in which SPARC had performed similarly well compared with COSMOtherm for systems such as octanol/water and air/water [6,13,37]. The RMSE values of SPARC were 0.5 log units to 2 log units higher than those from COSMOtherm and ABSOLV, and they also spanned over a larger range.…”

“…Typically, these QSARs assume an additive behavior of contributions that come from the various molecular fragments of the chemical. Because this additivity assumption often does not hold for neighboring functional groups [6], the models need additional empirical correction factors, which must be derived from the calibration data set as well. However, even QSARs for octanol/water partition coefficients (K OW ), for which more than 10 000 experimental data are available, appear not to have enough of those correction factors to perform well as a predictive tool for new chemical structures.…”

Terpene and Terpenoid Emissions and Secondary Organic Aerosol Production

“…2)]. Uncertainties in estimates of descriptors for solutes and stationary phases were obtained from Goss et al (2009) and Poole and Poole (2008a), respectively. The uncertainty in the value of A, which is determined empirically from a plot of secondary retention times vs. secondary retention indices , was chosen to be 0.040 (or approximately 3.0%).…”

“…Secondary OA (SOA) consists of a complex mixture of linear, branched, and cyclic hydrocarbons and mono-and multifunctional organic species. Comparison of the very few measurements of solute descriptors for bifunctional species with values estimated by assuming additive behavior of functional groups indicates much poorer agreement for diols and diones than diesters (Goss et al, 2009).…”