Расчёт сольватации в Gaussian 09

[azeif]

Коллеги, добрый день!

Нужно посчитать свободную энергию сольватации в Gaussian 09. В Gaussian 03 эта величина осязаемая и выдаётся непосредственно в выходном файле (deltaG(solv)), в Gaussian 09 я так понимаю, что нет.
Сейчас я считаю так:
расчёт в вакууме:
#P B3LYP/6-31G(d) gfprint freq=noraman opt
расчёт в воде:
#P B3LYP/gen gfprint freq=noraman opt
scrf(solvent=water,read)
...
dis rep cav

В рез-те получаю параметр Sum of electronic and thermal free energies в конце лог-файла, который равен ΔG (прав ли я?). Затем вычитаю это ΔG для воды из ΔG для вакуума.

Вопрос: учитываются ли электростатические и неэлектростатические сольватационные вклады в этом ΔG? Или они учитываются только в величине E(RB3LYP)? Просто результаты получаются очень странные, и они не согласуются с расчётами той же системы в G03. Есть ещё параметр Dovacuum, но я не понял, как он работает - в выходном файле всё равно нет отдельной энергии сольватации.

Спасибо!

[YuraM]

Good evening,

usually the effects from solvation are included via the single points only.
Regarding your question -
In principle you are right, your approach is correct, but the problem is that the polarizable continuum models are to the best of my knowledge not parametrized to include the contributions from rotation/vibrations. I think to evaluate the solvation energy in G 09 you need to perform to calculations:
1) SCF in gas phase
2) SCF in the solvent

Then, to get the solvation free energy you need to subtract former from the latter:
grep 'SCF Done' *out

DeltaGsolv = E (SCF, solv) - E(SCF, gas)

Usually thus obtained numbers are reasonable. Could you be specific and say why you think your number is not reasonable? What solvent?
Also, despite I adore the PCM solvation model, I have to admit that SD from Truhlar sometimes work better, if you specify that one - no need to include Disp, REp, Cav - it is already in SMD although it is introduced via atomic surface tensions.

Good luck with your research!

[azeif]

YuraM, thank you so much!

I have contacted Gaussian help and they've adviced me to use SMD too. It seems that in Gaussian 09 the only reliable way to account for solvation deltaG is by SMD. So I will use it. Interestingly, I have found an article where G09 solvation deltaG's were calculated without non-electrostatic parameters, but authors compared them with SMD (which includes non-electrostatic) and with experimental ones. It was very hard for me to figure out that non-electrostatic parameters are by default omitted in Gaussian.

According to my own investigations, E (SCF, solv) - E(SCF, gas) is quite similar to dG (SCF, solv) - dG(SCF, gas) and thus I hope PCM and frequency analysis works together. It is more comfortable for me to use deltadeltaG to get deltaG of solvation thatn deltadeltaE)