Many thanks for your kind reply guiding me for corrections. I will try
according to the suggestions given by you.
I will like ask
I wish to compute the band structure of LaS and LaAs (in first case La is +2
and in the later is +3)
n=La c=car # Symbol, XC flavor,{ |r|s}
0.0 0.0 0.0 0.0 0.0 0.0
11 4 # norbs_core, norbs_valence
6 0 2.00 0.00 # 6s2
6 1 0.00 0.00 # 6p0
5 2 1.00 0.00 # 5d1
4 3 0.00 0.00 # 4f0
3.47 4.13 3.00 2.75 0.00 0.00
It is for La+3. How to define it in pseudo file for La+2 ?
And How to decide, in band structure which particular bands belong to which
shell (s,p,d,f)
Many thanks,
Nidhi
[EMAIL PROTECTED] wrote: You mix up several things; I doubt it will help to
resolve your problems
but let us address them one by one.
Sm valence configuration. 4f states are quite localized and
probably (in reality, not in DFT the calculation) are not any near
to the band gap. If you include them in valence states and in the basis,
your trouble will be not performing the calculation as such,
but their wrong calculated positioning (at the Fermi level).
If you attribute them to core... I don't know there is an easy way
to do this, because the 4f shell is not fully occupied.
Search for previous calculations (any method, with DFT and beyond)
on RE chalcogenides, and on any RE calculations using pseudopotentials.
> Dear users, as we know the valence configuration of Sm is 4f6,6s2. In
> order to combine it with chalcogenides it is necessary to make the net
> ionic charge of Sm to 2, means we have to consider the 4f6 in the core.
Why necessary? The "net ionic charge", whatever its definition, will come
out of your calculation somehow. To begin with, you start from neutral
atoms, and they remain neutral, whether you attribute 4f to the core
or to the valence states...
> When we define the PAO basis set as
>
> %block PAO.Basis
> Sm 2 # Label, l-shells
> n=6 0 2 P 1 # n, l, Nzeta, Polarization, NzetaPol
> 0.000 0.000
> 1.000 1.000
Don't forget to include 5d in the basis; they are IMPORTANT.
> Te 2 # Species label, number of l-shells
> n=5 0 2 P 1 # n, l, Nzeta, Polarization, NzetaPol
> 0.000 0.000
> 1.000 1.000
> n=5 1 2 # n, l, Nzeta
> 0.000 0.000
> 1.000 1.000
> %endblock PAO.Basis
Your Te might be OK (or not); at least no obvious faults.
>
> it will display the following message
> reinit:
> reinit: System Label: SmTe
> -----------------------------------------------------------------------
> initatom: Reading input for the pseudopotentials and atomic orbitals
> ----------
> Species number: 1 Label: Sm Atomic number: 62
> Species number: 2 Label: Te Atomic number: 52
> Ground state valence configuration: 6s02 4f06
> Reading pseudopotential information in formatted form from Sm.psf
> Ground state valence configuration: 5s02 5p04
> Reading pseudopotential information in formatted form from Te.psf
> Bad format of (n), l, nzeta line in PAO.Basis
> Stopping Program from Node: 0
This is probably because you promised 2 functions in the basis block for Sm
but passed only one (6s). Make it consistent.
> If I include the 4f6 in basis set it will make the net charge 8 and behave
> as a semi core.
This "net charge" is not exactly your worry. It simply gives you the number
of electrons provided by the atom in question to the valence band,
in does not yet make from Sm a 8+ ion. Similarly, you can choose
the Te configuration either as 5s2 5p4 5d0 (6 valence electrons)
or 5s2 5p4 4d10 (16 valence electrons), it is still the same atom.
Only, you'll have different number of bands. I repeate, the decision
to put Sm 4f in the core or in the valence is only your - difficult,
but free - choice.
Now we come to Te.
> If I use a already generated pseudo file of Te which include 5s2, 5p4, 4d0
> and 4f0 But how can 4d0 is possible although it contains 10 electrons.
This is a misprint in the head line of the Te pseudo. It was generated
with 4d10 in the core and 5d as valence states. (Ask Eduardo Anglada).
> When I use this file then we get results but band gap in B1 phase
> is ~10eV which is quite far from the expt 0.67eV.
This can be due to anything. (In fact an absence of Sm5d in the basis
is a good candidate). Try to look not only at the band gap value
(which will be wrong anyway) but at the density of states,
positioning of different groups of valence bands. The band structure
of RE chalcogenides is well known.
> Please help me how can i resolve the problem of valence charge .
I don't see there is a problem, in fact. The (technical) problem is -
if you want to remove 4f from the valence - how to declare them as core,
even as this shell is not fully occupied. But by default, you can
go ahead with 4f as valence states (in BOTH basis and pseudopotential).
Then you'll see that the positioning of the 4f is wrong, and start
to think how bad this is for the problem your have to solve,
and what to do about it. This is not a SIESTA problem, but one
which appeared before in other DFT calculations.
Good luck,
Andrei Postnikov
---------------------------------
Meet people who discuss and share your passions. Join them now.
