DIPOL tag meaning

[david6477]

Hello All,

I am trying to understand what the DIPOL tag represents. According to the VASP manual, it says that it should be the center of the cell. It also mentions that "this tag determines as in VASP.3.2 the center of the net charge distribution." I do not see a reason why these should be the same, so what does the DIPOL tag mean? Is it the center of the cell, the center of mass, or the center of the net charge distribution? All three of these phrases are used in relation to DIPOL on this page: http://cms.mpi.univie.ac.at/vasp/vasp/node152.html, so trying to use the manual to settle this question has not helped.

I have found by searching through this forum that I should set LDIPOL=True and DIPOL=(center of cell) to do a charged defect calculation (in this case, a vacancy) of a bulk solid that is symmetrical in all three dimensions. However, the VASP manual says that any calculation using LDIPOL=True is impossible (http://cms.mpi.univie.ac.at/vasp/vasp/node153.html). How can this calculation be done? Should I use IDIPOL=4 (although other posts here say that this is only done for molecules, not solids) and hope for the best?

Thank you very much for your time,
David

[admin]

you are correct, the centers of the cell and of the net charge distribution need not be the same, it depends on how the coorditaes of the atoms in the cell are set up (eg, if a slab stretches between -z/2 and +z/2 or from 0 to z). DIPOL denotes the center of the dipole , which corresponds approximately the origin the cell only (in the slab example above) if the slab is symmetric around 0.
the sentence you mention indicates that you cannot simply take this single number as printed in OUTCAR of one run only, but that you should obtain the correction terms as indicated in the manual (... in order to obtain the correct energy correction one has to proceed as follows:...
please note that in vasp.4.6
--- the dielectric constant of the solid is assumed to be 1, therefore you have to scale the results with epsilon as obtained from literature. This will not be the case with vasp.5.1 anymore)
--- the quadrupole corrections are 1/L^3 (which of course strictly only holds for cubic cells)