[Pw_forum] Inconsistent U(Fe,oct) and U(Fe,tet) in Fe3O4

[Matteo Cococcioni]

dear Xu

looks like that the calculation in one of the cells you are using (or maybe
both) is not converged with the size of the cell.
The linear response calculation of U is only sensitive to the distance and
the number of neighbors. As long as the supercell you are using guarantees
that each Fe (of each type) has the correct number of neighbors at the
correct distance, the result should be the same.

Matteo

On Tue, Feb 3, 2015 at 2:55 AM, <huangxu1706 at sina.com> wrote:

> Hi,
>
> I'm calculating the Hubbard U for Fe (octahedral) and Fe (tetrahedral) in
> bulk Fe3O4. The ultrasoft pseudopotential (USPP) I'm using now was
> downloaded from QE website called "Fe.pbe-nd-rrkjus.UPF". I'm using
> QE-5.0.2-pgi-13.4
>
> But I got inconsistent U(Fe,oct) and U(Fe,tet) values when I use different
> type of cell to do the linear perturbation. When I use cubic cell (56 atoms
> in total, 16 Fe_oct, 8 Fe_tet and 32 O), I got U(Fe,oct) < U(Fe,tet). When
> I use hexagonal cell (42 atoms in total, 12 Fe_oct, 6 Fe_tet and 24 O), or
> rhombohedral primitive cell (14 atoms in total, 4 Fe_oct, 2 Fe_tet and 8
> O), I got U(Fe,oct) > U(Fe,tet).
>
> I changed the CELL_PARAMETERS to make the Fe_oct and Fe_tet in cubic cell
> have the same Fe-O bond directions as those in hexagonal or rhombohedral
> cell. But still, it gave U(Fe,oct) < U(Fe,tet).
>
> I don't know if the problem is from the USPP or from the linear
> perturbation theory itself implemented in QE. People keep treating these
> two types of Fe the same in magnetite Fe3O4 studies. I read papers from
> Emily Carter's group using different U(Fe,oct) and U(Fe,tet). But their U
> values were not derived by calculation or by Fe3O4 itself, but from
> empirical combination obtained by FeO and Fe2O3. I got stuck with those
> inconsistent U values of Fe3O4 from QE and have no idea about how to deal
> with it.
>
> Thank you,
> Xu Huang
>
>
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20150203/49ed277d/attachment.html>