I tried to make the supercell as cubic as possible, thus the dimensions 1x3x2.
It is a slightly distorted near-cubic parallelepiped.
I should point that the experiments were done by implanting 111In in the Ga2O3
samples, which decays to 111Cd.
This means that the position of the probes is given
Cd 2+ has a full d shell and thus the contribution of the d-electrons to
the EFG should be small. (Try an EFG anylysis).
LDA+U can be important for partially filled d-shells (Ti-Ni), but in any
case, I'd try GGA+U, but for sure also the other position.
PS: Run x nn after relaxations and
One more thought:
How is the time delay for In--Cd decay and EFG measurement by PAC ???
Is the Cd already at a relaxed position or still in the In-position ?
You could try to make a structural relaxation for the In-impurity, and
then calculate the EFG at Cd in the In position without
The size should be big enough, at least for a first calculation.
Is the supercell nearly cubic, i.e. similar length in all dimensions ?
Is there only one Ga site ?
How are the distances after relaxation ?
Would it correspond to a more or less axial symmetry or is x,y,z very
different and
Dear Prof Blaha,
The biggest supercell I used for Ga2O3 was 1x3x2, which has 120 atoms.
Then I changed one of the Ga atoms to Cd and relaxed the atomic positions.
Since GGA+U or hybrids won’t affect the EFG and ETA in this case, do you think
I should try bigger supercells?
Thank you very much
No, I don't think that for this system GGA+U or hybrids are appropriate.
Is your supercell large enough (~100 atoms) ??
Did you relax the atomic positions ??
Maybe the Cd sits in a different position ??
Am 02.02.2015 um 20:09 schrieb Marcelo Barbosa:
Dear all,
I need to calculate the EFG and
Dear all,
I need to calculate the EFG and ETA in a Cd impurity in Ga2O3 but using the
usual GGA method gives me an ETA near 1 when experimentally it is known to be
between 0.0 and 0.1.
I would like then to ask what is the best method to get good EFG and ETA values
in such a case:
- use GGA+U
7 matches
Mail list logo