Solid and gas phase energies for Acetic Acid

[CStar]

I calculated the solid phase energy of Acetic Acid (CH3COOH) to be -46.899 eV per Acetic Acid. I got this value by performing an energy minimization for a box containing four AA molecules, with initial lattice parameters and atomic positions as given by (Acta. Crystallogr. 11, 484). I allowed the box size to change during the optimization, and the length of a increased by ~2 Angstroms, while b and c did not change too much (<0.5 Angstroms). Is it strange that the value of a should be 2 Angstroms different from the experimentally-determined value? The correct space group was retained throughout the calculation.

I then calculated the vapor phase energy of Acetic Acid by placing an isolated molecule in a cubic cell of increasing side length (up to 40 Angstrom), and again doing energy minimization. The strange thing is that the gas phase energy converged to -46.745, only ~0.2 eV larger than for the solid phase. How come there is such a small energy difference for these two cases? I imagined that it would be much greater, since at T=0 the material would be found in the solid form, which does not melt until T=290 K Am I missing something?

[alex]

Missing van der Waals interactions in DFT is the answer you are looking for. This is also responsible for not getting the lattice parameter in stacking direction right. However, 2 Angstroem are a little much. How did you do it? ENCUT = ?, How many runs to get remainders of Pulay stress minimized?

alex

[CStar]

Thanks for the reply Alex. I had not really considered the role of vdW interactions. The 2 Angstrom difference still seems large though.

I did not specify ENCUT in the INCAR, but I set PREC=high, so which I think should be OK (as I understand it, that means the calculation will be done with ENMAX and ENAUG at 130% x the recommended values). I didn't minimize the Pulay stress - the results I got were the result of just one cell optimization. Since it seems like the result I am after (i.e. the energy difference between the solid and vapor phases) can't be obtained without including vdW interactions I guess I'll leave it at that.

I have another quick question though. I am looking to find the deprotonation energy for acetic acid within PAW-PBE, so I need the Energy of the H+ ion. I wonder is this just the number EATOM at the top of the POTCAR file for H, i.e. 12.4884 eV? I found the difference in energy for H and H+ using Gaussian with the PBE functionals to be 13.6 eV, but is this number different in VASP for some reason I am not thinking of? In the context of my calculations 12.4884 makes more sense, but it seems strange that I get a different number with the other code using the same functionals.

[alex]

Hi there again,

the 2 \AA might relate to your non proper cell optimisation. DFT will get you some binding though.

ENCUT should always been specified. Use about twice the number recommend as minimum for cell optimisations, e.g. 800eV in your case. Repeat the optimisation 3 times to get well converged results. Do remove old WAVECARs!.

About the energy of charged systems, there was a discussion here some years ago (with vasp 4.6): H+ is E=0 by definition (no electron).
You have also to take into account the energy of the PAW part to calculate the complete total energy in eV. It should be of the size of a Gaussian calculation.
Since there is no screening of the slowly decaying Coulomb interactions between the periodic images, you'll have practically no convergence. In short: forget about it.

Btw. CPMD has this screening and can do the job.

Hth

alex