[Wien] total energy of indevudual atom for cohesion energy

[Peter Blaha]

In order to calculate transition metal atoms you must

a) use a spin-polarized calculation
b) use TEMP or GAUSS with a fairly large broadening
   (As you can easily check eg. for Fe using lstart and different
case.inst files, the Fe atom CANNOT be calculated properly within DFT.
This means: when you put 2 e into the 4s levels, their energy raises above 
the 3d levels. Thus in the next iteration one should move one 4s electron 
into the the 3d levels. If one does, the potential changes and the 4s 
level comes down again, thus should be doubly occupied. This scenario 
continues to infinity and that's what happens in your calculation.

A small mixing is necessary, but NOT sufficient in these cases, since the 
valence density will always change by ONE electron (and fractional 
occupation is not possible with TETRA and one k-point).

Using the broadening means that both, the 3d AND the 4s levels can be 
partly occupied and thus a stable situation can eventually be reached
(again from atomic tests we know that for Fe around 4s1.6 (or similar) the
3d and 4s levels are equal).


> First, i would like to thank Dr Stefaan Cottenier, for his replay and help, and Dr L. D. Marks for his helpful notes and suggestions.
>   But, i do the calculation for the Fe and Ti and also for Zr (for ZrNiH enthalpy) no convergence (between 300 and 400 iterations). I have save calculations after evry 50 iteration (save_lapw case1 in xterm) i have use 0.004 (mixing).
>   Thank you again
> 
>  		
> ---------------------------------
> Do you Yahoo!?
>  Get on board. You're invited to try the new Yahoo! Mail Beta.


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