Closed xiki-tempula closed 3 years ago
You can use any QM program you want for the DFT part, that's independent of crest
.
The only thing to look out for is whether you are able to apply the frequency scaling factor and rotor cut-off in the thermostatistical calculation. xtb
's and crest
's thermo
implementation can do that, but only in Turbomole's/xtb
's vibspectrum
format.
You might even use other functionals if you wish to do so and determine appropriate frequency scaling factors.
@pprcht Thank you for the comment. I'm trying to use ORCA for this calculation.
Based on my understanding, the S = S_msRRHO + S_conf
.
The crest standalone could give the S_conf at GFN2-xTB or GFN-FF level (or other xtb supported level).
The S_RRHO is the original Total entropy correction
scaled according to Truhlar (https://comp.chem.umn.edu/freqscale/version3b2.htm) and applying the Grimme's entropy interpolation between harmonic and free-rotor approximations.
I wonder if this is correct? I'm aware of the recent development of the Single Point Hessian but I cannot find a tool for applying it to orca.
yes, crest
gives you only S_conf , which is a combination of Gibbs-Boltzmann conformational entropy S'_conf and a |S|_msRRHO average calculated at some GFN level (i.e., S_conf = S'_conf + |S|_msRRHO). These terms are entirely included in the S_conf crest
gives you, and no manual adjustments are required here.
For the absolute entropy (S = S_msRRHO + S_conf) you then need then S_msRRHO for the lowest conformer at DFT level, for which the frequencies are scaled and the interpolation is applied, as you already mentioned. The B97-3c ord B3LYP scaling factors we applied are in the Chem. Sci. paper rather than in Truhlars database because we newly determined them for this study. In ORCA the scaling factor can be easily modified, but I'm not completely sure sure about the interpolation. I know the interpolation is implemented in ORCA with a default threshold (35 cm-1). Apparently this can be modified in ORCA's %freq
input block with the keyword CutOffFreq
, but please look at the ORCA manual again.
If that doesn't work, a workaround would be to take the frequencies from the ORCA calculation and write a vibspectrum
file with it, which then could use xtb
's or crest
's thermo
implementation.
This is unrelated to the Single Point Hessian procedure. SPH is used within the crest
calculations, so it doesn't influence your DFT calculation.
@pprcht Thank you. I just want to check if my understanding is correct and S_msRRHO
doesn't contain anything else.
I usually use https://github.com/bobbypaton/GoodVibes or http://sobereva.com/soft/shermo/ for scaling and Grimme's entropy interpolation.
@pprcht Thanks for addressing this question.
Another short question. The output is:
------------------------------------------
| FINAL MOLECULAR ENTROPY AT T= 298.15 K |
------------------------------------------
Sconf = 1.296712
+ δSrrho = -0.029233
----------------------------------------
= S(total) = 1.267479
H(T)-H(0) = 0.126621
G = -0.377899 (-T*S)
----------------------------------------
= G(total) = -0.251278 (H-T*S)
Cp(total) = 0.166227
So the S_conf would be = S(total) = 1.267479
, I wonder what is the unit? Is it in in Hartree (T= 298.15 K) or cal/molK?
The unit is not explicitly stated and I think number seems to ba a big too big for hatree.
The entropies and heat capacities are in cal/molK, the enthalpy and free energy are in kcal/mol. A bit further up there should be a note about that (below one of the tables with the temperature dependencies). I'll add a note for this printout again.
I was reading the recent paper of "Calculation of absolute molecular entropies and heat capacities made simple" and the higher level DFT calculations are done using TURBOMOLE. Given that TURBOMOLE is commercial software and might not be available to anyone. I wonder if it is possible to use support ORCA? Since it also has the level of B97-3c and of course B3LYP-D3(BJ)/def2-TZVP with RIJCOSX support.