On the prediction by density functional theory of entropies in solution within implicit solvation models.

Context: Entropies are fundamental contributions to Gibbs energies that carry important chemical information, in particular when investigating reaction mechanisms. However, evaluating them in solution is far from being straightforward. In this paper, we focus on its evaluation within the framework of implicit solvation models. To this aim, successive corrections (with increased complexity) involving only contributions available from any standard quantum chemistry code and macroscopic solvent properties are built and assessed by comparison to more than one hundred experimental entropy values measured in a liquid phase. It turns out that significant improvement with respect to the standard ideal gas approximation can be achieved at an almost negligible computational cost, affording a robust and transferable predictive model.

Methods: DFT calculations with the ADF software at the PBE or PBE0/TZ2P level of theory with COSMO solvent model. Python scripts for regressions.