[Wien] [SPAM?] Should the valence electrons configuration of charge transfer insulators be changed ?

[Peter Blaha]

I'll add a few statements about core-EELS:

1) Core hole: In principle we want to simulate the excitation of ONE 
core electron into the conduction band. Thus one should create a big 
supercell (as big as possible, at least 64 atoms) and put a full core 
hole (I guess this was NOT yet mentioned, but is the most important 
point of the discussion !!!!!). This hole will be partially screened, 
and with our limited supercell size and the static DFT approximation, 
this screening could be incomplete and thus one sometimes uses 
"empirically" 1/2  or no core hole (in particular for metals) at all.
This is an often used method, but of course it is no longer "ab initio".

2) Slaters transistion state is a well known concept to calculate the 
XPS binding energy of a core state, where you would remove the excited 
electron from the system (it comes out and goes to the detector). It has 
NOTHING to do in EELS , where the excited electron stays in the system 
(except if you would attempt to calculate the absolute energy of an edge).

3) excited electron: In principle it is clear that the excited electron 
should go into a dipole allowed conduction band state. However, we have 
NO MEANS to select such a state and the electron will go into the first 
empty states in the system in a scf procedure.If we feel that this state 
is not the state where it would go in experiment, it is better to put 
the electron into the "background" charge (mixer). E.g in NiO the O-1s 
electron should go into a O-2p state. However, the first conduction 
bands are Ni-d states in the supercell calculation and thus adding an 
electron to the valence electrons is not appropriate. In the case of 
cuprates, I'd probably add it to the valence, since the "hole" state is 
a mixture of Cu-d-x2-y2 - O-2p and thus at least partly it is ok to put 
the electron into it. In any case, I'd do the calculation with both, 
adding the electron to valence or to background.

4) spin state: It is of course clear, that the photon does not change 
the spin state of the excited electron.In a spin-polarized calculation 
when you put the electron into the valence, it is usually obeyed anyway, 
because the missing core electron of "spin-up" will lower the potential 
of spin-up and the electron will go into the spin-up conduction bands, 
preserving the total spin of the system.
However, correlations within the conduction bands could change this 
anyway, because the "other electrons" could react on the presence of an 
additional spin-up electron.This is in particular true for correlated TM 
oxides. And if you use the background-option, the spin.state is not 
defined anyway, since the background option cannot be done spin-selective.

In non-spinpolarized calculations it should not really matter.



Am 17.11.2019 um 14:58 schrieb 丁一凡:
> As we all know, DFT deals with the system in the ground state. When 
> dealing with the charge transfer insulator system, can I modify the 
> valence electronic configuration after initialization and before SCF and 
> EELS (Electron Energy Loss Spectroscopy) calculations ?
> 
> The Cu-based high temperature superconducting (HTSC) oxides are known to 
> be insulators of a charge-transfer type, with the charge-transfer (CT) 
> gap originating from the energy difference between the O(2p) and the 
> Cu(3dx2-y2) orbitals. Before calculating EELS of Cu-based HTSC oxides, 
> will it make the result reasonable if their valence electron 
> configuration is changed ? For example, we remove one oxygen 2p electron 
> and add one electron in Cu 3d orbit. Just like the treatment of core 
> hole effect. For a “core-hole” calculation we will edit super.inc and 
> remove one core electron from the desired atom and state (1s or 2p, 
> ...). Then we add the missing electron either in super.inm (background 
> charge) or super.in2 (add it to the valence electrons).
> 
> This problem haunts me for several weeks, and my question is still 
> unsolved after consulting the previous mailing list. Any comment(s) 
> would be highly appreciated. Thanks in advance!
> 
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at:  http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
> 

-- 
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-165300             FAX: +43-1-58801-165982
Email: blaha at theochem.tuwien.ac.at    WIEN2k: http://www.wien2k.at
WWW: 
http://www.imc.tuwien.ac.at/tc_blaha-------------------------------------------------------------------------