Questions on how to rigorously measure free energies in solution

wham09 · 2026-01-20 03:23:59

Dear Prof. Lu,

I did the following calculations for a ground-state molecule:

calc1. Geometry optimization (PCM model) + frequency calculation (B3LYP-D3(BJ) / def2-SVP, PCM)
calc2. Single point calculation (gas-phase) using calc1 geometry (B3LYP-D3(BJ) / def2-TZVP)
calc3. Single point calculation (gas-phase) using calc1 geometry (M05-2X / 6-31G*)
calc4. Single point calculation (SMD model) using calc1 geometry (M05-2X / 6-31G*, SMD)

From calc2, I obtained E_gas.
From calc3 and calc4, I obtained dG_solv.

If I understood correctly, the free energy is calculated as:
G = G_gas + dG_solv + 1.89
= (E_gas + dG_corr_gas) + dG_solv + 1.89

So I utilized Shermo on calc1 output for ZPVE scaling and msRRHO treatment, and obtained dG_corr_gas.
But here are my questions:

1. Am I correct to put E_gas (from calc2) into the Shermo's E value in the setting.ini file, rather than E_gas + dG_solv?

2. For calc3 and calc4 (using M05-2X functional), should superfine grids be used, as they sometimes should be for M06-2X?

3. In line with Q3, if I choose to do calc3 and calc4 using M06-2X instead, should I use superfine grids consistently?

4. calc1 was done with PCM because the geometry was sensitive to solvation. But then, is the Shermo result using the calc1 output truly the dG_corr_gas?

5. As an alternative, I did another frequency calculation (calc5) in the gas-phase using the calc1 geometry, but when I did this, I observed imaginary frequencies (expectedly). Would using Shermo on the calc5 output be more appropriate than using the calc1 output anyways?

6. Is there a truly ideal solution to Q4 and Q5? I don't care if it is cumbersome, I just want to know.

Thank you very much again for your time.

sobereva · 2026-01-20 05:32:41

1 If you want to obtain G in solvent phase, you should put E_gas + dG_solv into settings.ini of Shermo, otherwise solvent effect will be missing.

2 The default grid (int=ultrafine for G16) is completely adequate.

3 When two energetic results will be compared with each other, or taking their difference, integration grid setting must be exactly the same.

4 It is true (at least for present purpose).

5 calc5 is meaningless. opt and freq tasks must be conducted at the same computational level (including the use of solvation model, which affects potential energy surface).

wham09 · 2026-01-20 05:49:12

I would be really grateful if you could explain the answer to Q1 a little more. The way I understand is that the G in solvent phase is as below:
G_sol = G_gas + dG_solv + 1.89
Because G_gas = E_gas + dG_corr_gas,
G_sol = (E_gas + dG_corr_gas) + dG_solv + 1.89
Here, dG_corr_gas should be the correction for gas-phase G. Then, shouldn’t I actually put E_gas into Shermo to obtain G_gas first, and then manually add dG_solv and 1.89kcal/mol?

sobereva · 2026-01-20 06:00:54

If you include "dG_solv + 1.89" into "E=" in settings.ini, then you do not need to manually add it to G_gas, otherwise you need to manually add them up.

Multiwfn forum

#1 2026-01-20 03:23:59

Questions on how to rigorously measure free energies in solution

#2 2026-01-20 05:32:41

Re: Questions on how to rigorously measure free energies in solution

#3 2026-01-20 05:49:12

Re: Questions on how to rigorously measure free energies in solution

#4 2026-01-20 06:00:54

Re: Questions on how to rigorously measure free energies in solution

Board footer