This article reviews different formulations of the thermodynamic cycles used for the prediction of pK a values, their advantages, and disadvantages with special emphasis on the limitations resulting from the necessity of gas-phase calculations, which allow introducing some difficult cases that motivated alternative strategies. Before introducing the protocols that do not consider gas-phase calculations, the two current opinions available in the literature on the debate about the correct formalism for the calculation of free energies in solution are briefly introduced. Then, the isodesmic proton exchange reaction in solution is reviewed by analyzing its performance on difficult cases for thermodynamic cycles such as carbon acids and amino acids. The pK a values predicted by the isodesmic reaction for common acid species are also reviewed to compare their accuracy results in relation with those of thermodynamic cycles. Linear regressions between experimental pK a values and the calculated free energies obtained with the isodesmic reaction provide expressions for the dependence of the error in the calculated pK a s on the pK a difference between the studied acid and the reference species. Finally, it is shown that linear regressions correct the calculated free energies of the isodesmic reaction, when high constant precision is required in a large pK a range. The deviations from the expected behavior are equivalent to those reported previously for different pK a calculation protocols and are determined by the inaccuracies of continuum solvent models on the interactions with ionic species.

No abstract