ABINIT Discussion Forums

I have been trying to calculate phonon band structures for Si and Ge using the Troullier-Martins pseudo potentials. The results I get agree largely with
what is published in the literature (for example, Wei and Chou, PRB 50, 1994, 2221), but there are some small differences. First, the lowest bands
which are exactly the same in the paper between the L and Gamma points are slightly different in my calculation (both for Si and Ge). Second,
there is an additional "peak" between the W and L points which is not present in the paper. I have varied the number of q vectors, but the results
are almost identical. I have attached the input for Si and the resulting band structure plots.
Any hint on what might be wrong is appreciated.

Jörg

jrh wrote:First, the lowest bands
which are exactly the same in the paper between the L and Gamma points are slightly different in my calculation (both for Si and Ge)

You mean : they should be degenerate. Yes, indeed. I suspect that nband 6 is the cause of the problem,
perhaps there is a degeneracy between the 6th and 7th band in some of the k point you are sampling.
Could you try simply to use nband 4 ?

Best,
X

I have tried your suggestion, but it has made things much worse. I have attached a pdf file containing the phonon band structure I am trying to
reproduce and the results of some of the attempts I have made. The closest I can come to the published results is when I run with a 12x12x12
grid of q vectors, but even there I still have differences between W and L as marked in the pdf file. Reducing the number of bands result in
a complete catastrophe.
I am not sure if the observed behavior has something to do with the way the q vectors are choosen. I am using the following procedure:
- I run an Abinit calculation with kptopt 1, ngkpt 8 8 8, nshiftk 1, shiftk 0.0 0.0 0.0
- I take the k points printed in the output of this run and copy them to the qpt keyword in the response function input where I use
nkpt 0, kptopt 3, ngkpt 8 8 8 for the k point
which is as I understand the procedure you have suggested in the second Abinit response function tutorial. Maybe the q vectors have to
be choosen differently. Therefore, I would appreciate any explanation and how this is supposed to be done in Abinit.

Thanks for your help,
Jörg

Perhaps a minor point. If you do not specify the shiftk explicitly it will be nshiftk 1 shiftk 0.5 0.5 0.5, which will probably break symmetry for an FCC system. Set this to the usual nshiftk 4 shiftk 0.5 0.5 0.5 0.5 0 0 0 0.5 0 0 0 0.5, or use a shift of 0 0 0 to check.

matthieu