My Community

Dear Support,

I am resubmitting a post that I submitted previously (https://vasp.at/forum/viewtopic.php?p=18825#p18825). An updated version of it is given below:

I have a couple of questions about correcting the energy of a surface or slab model with a net dipole moment along the surface normal direction.

When performing slab calculations, a post hoc correction to the energy of a slab model with a net dipole moment along the surface normal can be made by adding a corrective term to the computed energy. In VASP, this correction is done using the VASP keyword IDIPOL = 1, 2, or 3. For example, if the surface normal is parallel to the c lattice vector, then the correction is done using IDIPOL = 3. A self-consistent correction to the potential of the system can also be made by setting the VASP keywords IDIPOL, LDIPOL, and DIPOL. However, I would like to ask about the correction to the energy.

My understanding is that the dipole moment of a periodic system is dependent on the choice of origin. If I am correct, then the size of the correction to the energy is also dependent on the choice of origin. My questions are:

1) What is the theoretical justification or basis for this correction if the correction is dependent on the choice of origin?

2) As discussed in the VASP manual under the DIPOL keyword (https://www.vasp.at/wiki/index.php/DIPOL#), to make the correction to the energy, the origin is chosen to include the plane of maximum charge density parallel to the surface, unless it is explicitly set using the DIPOL keyword. Why is this choice of origin made? Is it based on some convention?

Thanks,

Yves

Dear Yves,

Regarding your first question: the correction is only origin dependent for charged systems, i.e. where a monopole is present. This is not the case for dipoles.

For the second question: VASP will add a sawtooth potential in region in the cell. The optimal place to do so is where the charge density is minimal. The origin dictates where there is charge density (DIPOL) and then VASP adds c/2 to that and places the saw tooth potential at that point. If the origin is not set, VASP uses the point as in the method you described.

Kind regards,
Pedro

Dear Pedro,

Thanks for your reply.

I agree with you that the dipole moment of a neutral system is not origin dependent. Thus, the dipole moment of a slab in a super cell is not origin dependent, and the correction to the energy can be made.

Also, I agree with you that, to make the dipole correction to the energy, VASP will add a single sawtooth or ramp potential corresponding to a planar dipole layer in a part of the super cell [see Neugebauer and Scheffler, Phys. Rev. B 46, 16067 (1992) and Bengtsson, Phys. Rev. B 59, 12301 (1999)]. A jump in the total potential is introduced where this potential is added. Thus, it needs to be added in the middle of the vacuum region, where the charge density is minimal. Because it needs to be added there, the origin is set to be the center of mass of the slab.

Thanks for your help,

Yves