[Wien] Choosing proper Emin in electron density plot

[Peter Blaha]

 From " Chemistry" it should be clear, that only Ti 3s,3p are 
"semicore", C 2s belongs already to "valence" states.
Thus you should put   -1 as EMIN to truncate the semicore for plotting.

I admit, the comment in the scf file:
 > In TiC.scf, there is also "Energy to separate semicore and
 > valencestates:  0.25298".
is misleading. I should change it to something like "low and high energies"

 > This value is consistent with above reasoning.
No, because C-2 is valence.

 > -3.4310763 Ti-3s                correct
 > -1.6693347 Ti-3p                correct
 > -0.1792457 C-2s                 correct
 >  0.7811822 C-2p                 NO

You can easily assign low-E states, because they have mostly "pure" 
character. It does not really make sense in most cases to assign the 
"mixed band states", since they will be in general hybridized (Ti-3d + 
C-2p) and have a large bandwidth (you were looking on just one k-point, 
for others, this energy can be quite different).

> Dear Wien2k users
> There are some discussions about setting Emin in case.in2 when plotting
> the electron density on the mailinglist early this year. Some good hints
> were given. However, clearer conclusion may be more informative especially
> for new users, e.g., me. I want to take the TiC as example to work out
> the right way by following those hints. Please give your advice.
> In the TiC.scf file, I find the eigenvalues:
> -3.4310763   -1.6693347   -1.6693347   -1.6693347   -0.1792457
>  0.7811822    0.7811822    0.7811822    0.8523448    0.8523448
>  0.8523448    0.8563761    0.8563761    1.4909371    1.7608187
> The multiplicities for both Ti and C are 1. So I assign the states as:
> -3.4310763 Ti-3s
> -1.6693347 Ti-3p
> -0.1792457 C-2s
>  0.7811822 C-2p
> The 5 values around 0.85*** are close but not same, since they are
> forming band due to bonding effects in solid. They corresponds to the
> partially-filled valence states Ti-3d.
> The Fermi level is at 0.73236 just slightly below the C-2p which is not
> full. So it seems above assignment is correct. An Emin set between
> -0.1792457~0.7811822 should be able to exclude all semicore states in Ti
> and C.
> In TiC.scf, there is also "Energy to separate semicore and valencestates:
> 0.25298". This value is consistent with above reasoning.
> However, the value in the UG is -1.0. Why this value? Is there any other
> consideration I missed? Will a deeper Emin just include a constant
> part of electron density from semicore states and won't cause big error if
> we only want to know the difference between different plotting?
> Please give your advice, whatever it is empirical, technical or
> theoretical.
> Thanks
> 
> Shu Miao
> 
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien

-- 

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