Is there any definitive evidence that implicit solvation methods are more accurate than explicit solvation methods for free energy calculations?

Question

I am looking for real data that concludes that implicit solvent models are definitively better than explicit solvent models for solvation free energy calculations. The motivation is a lovely reviewer who has said this. Evidence that they are not statistically different is also valid.

Answer 1

In principle, you would expect the exact opposite (i.e. explicit better than implicit) simply because an explicit model more closely matches the reality of molecules interacting with other molecules rather than some sort of continuum.

In practice, you generally can't achieve full explicit solvation or even a single solvation shell and are limited to working with only a few solvent molecules. While this does allow you to treat some of the specific solute-solvent interactions, the ad-hoc inclusion of some interactions and not others can be problematic. This could potentially lead to a more uniform implicit model performing better for some systems (or all systems if you aren't careful about the arrangement of solvent molecules, conformational averaging, etc).

However, the general consensus in the literature seems to be that, when they can be done, explicit models are the way to go. This 2017 JCTC paper [1] comparing implicit, explicit, and experimental solvation free energies of organic molecules in organic solvent might be of interest to you. They found that all the implicit models they tested were in worse agreement with experiment than an explicit model, in some cases substantially worse.

1. J. Zhang, H. Zhang, T. Wu, Q. Wang, and D. van der Spoel J. Chem. Theory Comput. 2017, 13, 3, 1034–1043 DOI: 10.1021/acs.jctc.7b00169

Answer 2

From what I have read, the performance of implicit models depends on a combination of the method used (Poisson-Boltzmann vs. Generalized-Born vs. another analytical method); and the size, shape, and species of the solute (ie: protein vs. ion). I think the best way to answer this would be to search for prior work on different solvation methods applied to the solute of interest (or a highly-similar one).

Answer 3

Highly depends on what your system of interest is.

(i) For molecular and ionic systems, check papers by Chris Cramer and Donald Truhlar.

(ii) Another paper of interest for heterogeneous catalysis could be the below paper. Basically for nonpolar/weakly polar solute implicit solvation is pretty good. However, the solvation energies can start to deviate for polar solutes (especially when solute can form H bonds with water). Well, the deviation here is based on comparison with explicit solvation, which in themselves can be wrong. So, one of the biggest problems is lack of experimental standards - I believe the closest water adsorption on heterogeneous surface experiments are calorimetric experiments by Charlie Campbell in U Washington(?). I think they did experiments on dissociated water adsorption on Pt(111). I can dig it up if you are interested in it. https://chemistry-europe.onlinelibrary.wiley.com/doi/abs/10.1002/cphc.201700162