problem in calculating polarization with Berry Phase in large slab
Posted: Mon Apr 26, 2010 8:10 am
Dear sir,
Using Berry phase method to calculate polarization of a single unit cell is satisfying.
However, I encountered a problem in calculating polarization for a ferroelectric material with a large supercell, e.g. a 1x1x10 cell.
There are tens of atoms in the cell. When it reach the ferroelectric phase all the atoms remove from their starting position, and hence it is sophistic to choose the relative point, namely the DIPOL-tag, for both of the polarized and non-polarized structures during Berry phase calculation.
I find different choice of relative point turns out large discrepancy in the results even the relative point is chosen with the guarantee that the ions in the distorted and the undistorted structure remain on the same side. (Its influence is in the <ionic term>).
Also the choice of the boundary of the cell also greatly influence the result by yielding different value in <electronic term>
Please give me some guide line in this issue. Thanks a lot.
Using Berry phase method to calculate polarization of a single unit cell is satisfying.
However, I encountered a problem in calculating polarization for a ferroelectric material with a large supercell, e.g. a 1x1x10 cell.
There are tens of atoms in the cell. When it reach the ferroelectric phase all the atoms remove from their starting position, and hence it is sophistic to choose the relative point, namely the DIPOL-tag, for both of the polarized and non-polarized structures during Berry phase calculation.
I find different choice of relative point turns out large discrepancy in the results even the relative point is chosen with the guarantee that the ions in the distorted and the undistorted structure remain on the same side. (Its influence is in the <ionic term>).
Also the choice of the boundary of the cell also greatly influence the result by yielding different value in <electronic term>
Please give me some guide line in this issue. Thanks a lot.