[Wien] Reference energy for eigenvalues

[Peter Blaha]

> In my case, the difference of Fermi energies is a too rough estimate.  
> However, what works probably fine is taking one of the deep core  
> states (say S1) as a reference when comparing two calculations,  
> provided this atom present in both calculations.
> 
> For example in trigonal Selenium (semiconductor):
> 
> VOLUME = 552.24497
> E_FERMI = 0.27371 Ry
> E_1S = -916.580021885 Ry
> 
> VOLUME = 535.67766
> E_FERMI = 0.24834 Ry
> E_1S = -916.601436004 Ry
> 
> The difference between the Fermi energies (the top of the valence  
> band) is 0.02537 Ry. The difference between 1S eigenvalues is 0.02141  
> Ry. Can attribute 0.00396 Ry (= 0.02537 - 0.02141) to the deformation  
> potential of the valence band edge? If so, I get correct at least an  
> order of magnitude for the deformation potential.

As long as you compare Se under pressure, this should be fine, since the 1s level
will hardly change. The procedure may fail when you compare Se with ZnSe or
similar, since ionicity will change the 1s core level.

> Sorry, I did not express myself clearly. -2.30 and -0.73 Ry are  
> eigenvalues of 3d-As and 3d-Ga states in GaAs. They are exactly the  
> same in 2- and 8-atom cell. This was my point to say that, although,  
> UST0 changes eigenvalues stay the same.

Identical eigenvalues show that your calculation is ok.

> What about units of UST0? I looked at REAN3.f, and it seems that UST0  
> is just a first Fourier coefficient, which should contain the average  
> value. Could it be that UST0 is the average potential multiplied by  
> the number of points in Fourier expansion?

I can't remember exactly the definition of UST0. One needs to dig into the code.
But I'd assume it is related to the average potential in the interstitial,
eventually multiplied by the unitcell volume ? It should be in Ry units.

However, I'd expect that when you change computational or structural parameters
(eg. GMAX in case.in2, or RMTs or unit cell volume) this energy zero will change,
but you don't know if this is a computational or physical change.
(when going from 2 to 8 atom cells, this should be "computational", while when you
change volume,....


-- 

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