[Wien] About antiferromagnetism calculation

[Peter Blaha]

I can hardly understand how a ferromagnetic solution can be obtained with
runafm   (except you have some unphysical or incorrect inclmcopy file.

Of course the calculations may converge to a non-magnetic solution instead
of the desired AFM. This could also depend on starting moments (inst file), 
mixing parameter (inm) or k-mesh.

In certain cases, convergence of a magnetic system with runsp might be
very difficult and the magnetic moment may only change by 0.001 uB /iteration
and one "declares" the calculations for converged although they are not.


> I have sometimes encoutered this type of problem : starting from antiparallel spins and using the "runafm" script may lead to a non-magnetic or to a ferromagnetic solution, even though the ground state turns out to be the antiferromagnetic one, at last (from the Wien calculations). 
> For example, this was the case of TeCuO3 (Chem. Mater. 17, 4350, 2005). 
> 
> The trick is then to : 
> - check the evolution of the magnetic moments along the iterations, 
> - when close to the expected values, stop the "runafm" calculation,
> - restart with "runsp" until convergence.
> 
> or, run the AFM calculation with runsp only (i.e. with spins flipped but without clmcopy)


                                      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/
--------------------------------------------------------------------------