Due to the large number of different surfactant and solute molecules used in micellar solutions, it is important to have predictive models for micelle-water partition equilibrium. The relative new method COSMOmic, an extension of COSMO-RS, can predict thermodynamic data by taking the anisotropy of a system into account. For this purpose, it needs the structure of the system in atomistic resolution. In previous studies, atomic distributions were obtained from self-assembly molecular dynamics (MD) simulations. In this work MD simulations starting from pre-assembled micelles are used to receive atomic distributions. These simulations are more efficient than self-assembly simulations with respect to system size and simulation time and allow the detailed study of micelle size/structure influences on solute partitioning. For the first time a wide range of aggregation numbers for different types of surfactants are investigated. More than 90 different micelles structures were obtained for six different surfactants: SDS (sodium dodecyl sulfate), CTAB (cetyltrimetylammonium bromide), C12E10, Brij35 (C12E23), Triton X-114 and Triton X-100 with initial aggregation numbers between 20 and 160. These were investigated with respect to micelle size, micelle shape, surfactant concentration and simulation time. To examine the prediction quality of COSMOmic, experimental data was collected and 36 micelle-water partition coefficients for C12E10, Triton X-100 and Brij35 were measured. In total more than 370 experimental partition coefficients were used to validate the calculations. The results show that the calculations are in good agreement with the experimental data and that they are better than predictions from methods that do not take the micelle structure into account (pseudo phase approach). In this context, small spherical micelles are proven to be more efficient while larger more ellipsoidal micelles show a higher probability to result in outliers. Therefore, this article provides a guideline for the development of averaged atomic distributions as input for COSMOmic.
Read full abstract