defect induced quadrupole moment in the presence of a dipole moment

[david6477]

Hello All,

I'm working with a charged defect and am trying to calculate the necessary corrections. The defect cell (GaAs with an As vacancy at the origin with a charge of +3) was first relaxed without any correction parameters in the INCAR, and then a static calculation was done with IDIPOL = 4. Here is the output in the OUTCAR:

DIPCOR: dipole corrections for dipol
direction 1 min pos 16, direction 2 min pos 16, direction 3 min pos 16,
dipolmoment 113.134608 113.134608 113.134615 electrons x Angstroem
Tr[quadrupol] -764.936125

energy correction for charged system 15.955710 eV
dipol+quadrupol energy correction 711.636033 eV
added-field ion interaction 0.000000 eV (added to PSCEN)

I am following the procedure outlined in the VASP manual on http://cms.mpi.univie.ac.at/vasp/vasp/node152.html. I know that I must correct the dipol+quadrupol term (I have already done an IDIPOL = 4 and DIPOL = 0 0 0 calculation of the pristine bulk cell) such that I correct only for the defect induced quadrupole moment, but that the dipole term needs no correction (outside of a scaling by the dielectric constant).

The manual says to run a calculation on a free ion in a cell of the same dimension as the defect cell with the same charge state to help determine the quadrupole moment. I have tried a few ions, but each has a dipole moment in the OUTCAR:

DIPCOR: dipole corrections for dipol
direction 1 min pos 75, direction 2 min pos 75, direction 3 min pos 43,
dipolmoment 2.700443 2.700443 11.346195 electrons x Angstroem
Tr[quadrupol] -49.076840

energy correction for charged system 15.955710 eV
dipol+quadrupol energy correction -0.076900 eV
added-field ion interaction 0.000000 eV (added to PSCEN)

I assume this means that -0.076900 eV is not the energy associated with the monopole-quadrupole interaction energy, and won't help me determine the monopole-quadrupole interaction energy of my defect cell, but I could be wrong. This brings me to my question: is it necessary for the free ion calculation to not have a dipole moment in order to get the correct quadrupole term? If so, does anyone have any quidelines for choosing the right ion to use?

Thank you all for your time!
David

[david6477]

I think I have figured out the answer. It occurred to me, right after submitting my original question, that I may not have had the ion at the origin of the cell. I checked the POSCAR, found out I hadn't put the ion at the origin; so I put the ion at the origin, and reran the calculation. I also added DIPOL = 0 0 0 to the INCAR. Now, there is no dipole moment in the OUTCAR.

I also ran a calculation without the DIPOL tag, but with the ion at the origin. There was still a dipole moment in the OUTCAR, so it seems that the DIPOL tag is the important thing to specify, as well as putting the "defect" at the origin, as the VASP maual says.

Thanks,
David