Systematic Evaluation of a Cluster-Continuum-Model Workflow to Compute the Free Energies of Solvation of Ions in Different Solvents

A partially automatized computational protocol for the construction of embedded microsolvated clusters based on a combination of the quantum cluster growth algorithm, DFT reoptimization, higher-level energy computations, conformer selection based on free energy, embedding into advanced implicit solvation models like COSMO-RS, and averaging over a range of cluster sizes has been validated and applied to calculate single-ion Gibbs free energies of solvation in different solvents. For aqueous solution, excellent agreement with a widely accepted single-ion scale based on the cluster pair approximation (CPA) is found, to within the error margins of that scale, but without anchoring to one particular ion like the latter. Similar calculations show that the TATB assumption underlying the second widely used single-ion scale is not justified, shifting the TATB scale even further from the CPA scale. Applications of the cluster-continuum method to single-ion values in acetonitrile show that here the CPA-based proton value may be less settled than with water, methanol, or DMSO, and it deserves closer examination. Finally, the solvation of the fluoride ion is compared in a variety of different solvents (water, acetonitrile, methanol, DMSO, diethyl ether, and benzene). While water and methanol provide the most negative solvation free energies, acetonitrile, methanol, and DMSO are not far behind, and even the relatively nonpolar diethyl ether and benzene exhibit appreciable stabilization. This is due to significant charge-assisted C-H···F hydrogen bonds to the highly compact fluoride ion in all of the formally aprotic solvents, involving in some cases more than one proton from a given solvent molecule in a chelate binding mode.