[Wien] L2main - QTL-B error for the bcc Fe wire

[Peter Blaha]

>    I am trying to calculate a bcc Fe wire along c-axis with lattice
> parameters a=b=4*2.86A, c=2.86A(The lattice constant is 2.86 A for bcc
> Fe).  However, the "L2main error, QTL-B" problem occurs in the first
> iteration. I have read all the digests and the FAQs about QTL-B which
> given by Prof. P. Blaha et al., recently. I still can not solve the
> problem, even though I have tried to adjust the energy parameters
> according to the FAQ(I read the case.scf2 and case.helpXXX files. It was
> found that s, p and d states show ghost bands and cause the problem. I
> have changed the energy parameters in the case.in1c file). Could someone
> help me to check the problem? We are very appreciated your help and
> advice. Here I attach the case.struct and case.in1c files.

This is a very unusual structure and I doubt very much that this is
what you actually want to simulate. Have a look with xcrysdens to
this structure.

For these reasons the Fe atoms come very close and you are forced to use
very small RMTs.
For these small spheres lstart gives you a "core-leakage" error when you
use just the default -6.0 Ry.
So you must include also the Fe 3s state by lowering E-cut to -8.

When you do this, an extra LO for Fe-3s is generated and the calculation
runs without problems.

Without this extra 3s-LO you run into problems with the 4s orbital, since
it is not properly orthogonalized to the 3s and this leads to the ghostbands.

Regards
                                      P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW: http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------