(181l) Micelle/Water Partition Coefficients Using COSMO-RS: Conformational Study

In the field of chemical engineering and life sciences, the
prediction of phase equilibria in complex multicomponent systems is of growing
interest. One of the most popular and effective methods here is the COSMO-RS
model, which allows for the predictions based only on the chemical structure of
the system components. As shown by our group, the results depend strongly on
the choice of conformations. For small molecules, sets of minimum energy conformations
from the common conformational analysis made in vacuum give adequate results.
Large flexible molecules however may have a huge number of conformations and
the prediction results vary significantly for different conformations. Therefore,
the reliable conformations of large flexible molecules, which afford
reproducible results and lead to stable a-priori predictions on phase
equilibria, should be identified.

In this work, the partitioning of small solutes (alkanes, alcohols,
etc.) in the micellar solution of the non-ionic surfactant Triton X-100 (TX100)
in water is studied as an example. Micelles hereby are treated as a macroscopic
phase being in equilibrium with the aqueous surrounding (pseudo-phase
approach), so that the partitioning of a solute between these coexisting
pseudo-phases is determined by the thermodynamic equilibrium and thus the micelle/water
partition coefficient (K^MW) can be calculated based on the solute
limiting activity coefficients in the two pseudo-phases. For any small solute molecule with the limited number of
conformations, the same set of the minimum energy conformations was used
in all calculations. However, both pseudo-phases also
contain large flexible amphiphilic molecules, for which adequate conformations
have to be found. The conformations of TX100 were generated using two
methods: the force-field-based energy-minimization method (hyperchem HC) and
the condensed phase molecular dynamics (MD, Gromacs). The HC conformational
search was carried out in vacuum, while the MD simulations were made in water
and in octanol (the model solvent for micellar pseudo-phase) to account for the
solvent effects. The conformations obtained by both methods were then used to
calculate K^MW of the solutes using COSMO-RS and their influence on
the latter was studied.

The figure shows the probability distribution of the
logarithmic K^MW of octanol calculated by COSMO-RS for different single
conformations. The maxima observed in the distributions indicate that some of
the conformations occur more often than the others and thus will lead to the
more reproducible values of K^MW, i.e., are reliable for K^MW.
However, using this method to identify reliable conformations, great
calculation efforts have to be made. Thus, it would be advantageous to be able
to select such conformations directly from the conformational space obtained from
MD or HC, even before performing the time-consuming DFT/COSMO geometry
optimization calculations. Such a selection should be based on physical
parameters. In this study, we consider two of them, the radius of gyration and conformation
specific energy as well as their combination. It has been shown that if the
selection is made from the 20% most probable conformations of any of criteria,
the outliers in the probability distributions of K^MW are avoided. If
both criteria are considered simultaneously, a single surfactant conformation can
be identified that leads to the most probable result for K^MW. Thus,
the probability-based methodology leads to reliable and stable a-priori
predictions of solute partitioning in micellar systems as well as of phase
equilibria in general.

We thank DFG (MO 2199/1-1) for the financial support.