N-Nitrosamines make up a class of carcinogenic industrial pollutants that lack well-characterized physicochemical properties. Classical approaches to determine octanol-water partition coefficient (Kow) values are laborious, slow, and challenged by experimental error. Alternative methods include quantum chemical estimation (e.g., COSMO-RS), quantitative structure-property relationship (QSPR) models, and high-performance liquid chromatographic (HPLC) measurements; however, systematic compound-by-compound comparisons of these methods for chemical classes remain lacking. This study evaluates the performance of four methods (shake-flask, HPLC retention time, QSPR, and COSMO-RS estimation) for the log Kow determination. Shake-flask measurements for N-nitrosodiemthylamine (-0.54), N-nitrosomorpholine (-0.54), N-nitrosopiperidine (0.64), and N-nitrosodibutylamine (2.54) were compared to previously reported values, where the omission of quality control procedures (i.e., mutual solvent saturation and sufficient equilibration time) led to variations in measurements up to 0.64 log units. Among alternative methods, the COSMO-RS calculation in this study performed the best, relative to direct experimental measurement, with a root mean absolute error (RMSE) of 0.12 and improved accuracy compared to previous estimations. QSPR determination was comparable to that of COSMO-RS (RMSE of 0.14). Two methods of HPLC determination demonstrated the worst performance (RMSEs of 0.27 and 0.45). This study highlights the weaknesses in using the presented HPLC methods for compound classes that include polar molecules, demonstrates improved performance of theoretical calculations, and reports partitioning data for known (n = 8) and recently characterized (n = 7) N-nitrosamines found in the environment.
Bugher et al. (Wed,) studied this question.