Charged systems - deprotonation energy calculation in zeolites

[lux]

Hi,

I'm having a problem with the deprotonation energies(E = E(charged zeolite)-E(neutral zaolite with H+)) in zeolites, specifically with the charged zeolite. I'm calculating the -1 charged zeolite cell (19x14x15 Angstroms). After the optimization run I calculated Charge+dipol+quadrupole corrections according to http://cms.mpi.univie.ac.at/vasp/vasp/node152.html.
I tried to compare the deprotonozation energies with the literature - CPMD or various QM/MM approaches (QM_POT) and I found out that the deprot. energies from VASP (compared to both CPMD or QM(turbomole)/MM(gulp))are about 7-8 eV lower (the charged system being overstabilized). Although it seems that when the deprot. energies for various zeolite samples are compared, the relative differences between the deprot. energies are the same in VASP or QM/MM...
I came across the Forum thread http://cms.mpi.univie.ac.at/vasp-forum/ ... php?3.1514 stating that the might be a problem in the PSCENC energy contribution stating that "The current implementation will lead to errors in charged systems since the compensating background charge density interacts with the actual coulombic potential, not the pseudopotential."
I compared the PSCENC energies (single point, same geometry) for charged and neutral systems (both having NELECT = 1152)

neutral:
Free energy of the ion-electron system (eV)
---------------------------------------------------
alpha Z PSCENC = 1070.73716217
Ewald energy TEWEN = -16341.81642500
-1/2 Hartree DENC = -50450.23163310
-V(xc)+E(xc) XCENC = 3481.42889801
PAW double counting = 49167.78582241 -48102.90314419
entropy T*S EENTRO = 0.00000000
eigenvalues EBANDS = -9920.60016069
atomic energy EATOM = 69370.24228363
---------------------------------------------------
free energy TOTEN = -1725.35719675 eV

charged:

Free energy of the ion-electron system (eV)
---------------------------------------------------
alpha Z PSCENC = 1069.13659442
Ewald energy TEWEN = -16286.89957237
-1/2 Hartree DENC = -50514.76297214
-V(xc)+E(xc) XCENC = 3481.08295147
PAW double counting = 49159.52047497 -48094.61371448
entropy T*S EENTRO = 0.00000000
eigenvalues EBANDS = -9892.27473825
atomic energy EATOM = 69357.76365097
---------------------------------------------------
free energy TOTEN = -1721.04732540 eV

The difference is about 1.6 eV, but still I'm missing about 5 eV.
What might be the other sources of discrepancies between CPMD(QM/MM) and VASP deprot. energies?

Thanks a lot for your help

[lux]

Hi again,

I've done some testing regarding the deprotonation energies of mine zeolite H-FER. I tried the cluster calculation using both VASP and other programs like Molpro, Gaussian, Turbomole and Siesta. All latter programs gave approximately same deprotonation energy ~14 eV, while VASP gave the deprot. energy of ~1eV.
At the same time it seems that the problem seems to be with the calculation of the charged system - I've tried to calculate the extension of the O-H bond (in the limit leading to the deprotonation energy) and compared the energy changes(for various O-H bond lengths - zero net charge of the cell) calculated by VASP and Molpro. This time the energy changes were approx. the same.
Does anyone has an idea what might be behind this discrepancies??

Thanks a lot for help