My Community

Hi all,

I am trying to calculate the formation energy of oxygen vacancies in La2Zr2O7 pyrochlore oxide, and I am having some trouble getting convergence with respect to the plane wave energy cut-off (ENCUT). Does anyone have any experience with the behaviour of this kind of system or the atomic species involved?

The formation energy has been calculated for three different values of ENCUT: 520, 600 and 680eV. It increases with about 0.3eV when going from 520 to 600 and from 600 to 680eV. The difference stems from a decrease in the total energy of about 0.4eV for the pure system and 0.1eV for the defective system. The systems have been calculated both with constant and relaxed volume, and the ENCUT trends are the same for both types of calculations (the difference in formation energy from constant- and relaxed volume-calculations is constant).

I cannot use a higher value of ENCUT without using harder potentials (which I do not wish to do since I want to compare with a large number of other calculations which are already performed).

I use an 88 atom supercell, Monkhorst Pack 3x3x3 k-point sampling and PW91 potentials (LEXCH=91).


I would be most grateful for any ideas on the reason for the convergence problem.

/Joakim

For your formation energies do you recalculate the energy of the O2 molecule for the different cutoff energies?

I think that may be the source of your error.

Yes, I recalculate all energies involved in the reaction (corresponding to the pure and defective system as well as the chemical potential of oxygen and the electrons).

Unfortunatelly the change in the O2 energy is only on the order of 10meV while the changes of the pure and defective oxide is one order of magnitude larger, why the error is not cancelled.
<span class='smallblacktext'>[ Edited Thu Jun 05 2008, 06:39PM ]</span>

Would you mind telling me whether you did convergence tests on cutoff energy and Kponts? I did some calculations in ZrP2O7. What I remember is that the system was very easy to be tilted and couldn't have 90degree even though the system should be cubic. I used ENCUT>800eV to have convergence. I hope that my experience is helpful for you.