I'm wondering if anyone else had encountered similar problem before. Any
suggestions will be greatly appreciated.
I have been using the QHA code and our own code to calculated the atomic
projected phonon DOS using the phonon eigenvectors from QE phonon
calculation. Both methods gave very similar results, so I have no doubt
about that part.
For two different structures, ScF3 (cubic) and ZrO2 (monoclinic), the
projected phonon DOS didn't have the correct ratio between species. For
example, one will expect that oxygen will have twice as much contribution as
from Zr in ZrO2, but the actual ratio is about 1.4 instead. Each atom in
unit cell is supposed to have 3 degrees of freedom but Zr ended up with 3.7
while oxygen is about 2.6. (It is strange that the total degrees of freedom
still add up to 9, as expected.
I has used the our code on the eigenvectors from VASP calculation and it
gives the correct ratio. (I even adapted the VASP eigenvector format to be
used with QHA code. The results are good too.) It seems the eigenvectors
from QE may have some problem or maybe there is some factors to them? I'm
pretty confident about the convergence of the calculations. It is also
unlikely the convergence of calculation will give that much difference
because it should not change the phonon DOS conservation anyway.
California Institute of Technology
MC 138-78 1200 E California Blvd
Pasadena, CA 91125
-------------- next part --------------
An HTML attachment was scrubbed...