[Wien] [Wien2k Users] Solid Solution Generation of A0.60B0.40

[Ghosh SUDDHASATTWA]

Dear Dr.Blaha, 
Thank you very much for your kind words and reply. 
However, I have some issues. 
My system is an alloy system. As mentioned in the TiC example in the UG, we
have to use only a few k points for supercells. 

I have seen that metals generally require 15000-20000 as input in xkgen. 
So a reduction of factor of 8 means 2500 as input in xkgen. 
Am I right? 


-----Original Message-----
From: wien-bounces at zeus.theochem.tuwien.ac.at
[mailto:wien-bounces at zeus.theochem.tuwien.ac.at] On Behalf Of Peter Blaha
Sent: Tuesday, February 16, 2010 11:55 AM
To: A Mailing list for WIEN2k users
Subject: Re: [Wien] [Wien2k Users] Solid Solution Generation of A0.60B0.40

>    1. I have the alloy A0.60B0.40 where A and B both occupy the same
>       atomic positions. So I make the initial case.struct file using A
>       as the only element. (A1, A2.)
>    2. I run the supercell program with let us say, a 32 atom P cell or
>       may be 64 atom cell. I replace some of the A with B so that
>       exactly 60 % of the sites are filled with A and 40 % with B.

Yes, this is very clever. However, it is clear that you should not 
create just ONE of these configurations, but try several different
arrangements. One could be a completely "ordered" one, another one with
certain preferred partial order, others "very disordered". At the end
you may compare total energies and find out if a certain short-range 
order is favorable or not at all,....
In other words, you make "computer experiments" and try to find out more
about the system (maybe more than what is know experimentally).

>    3. I run xsgroup and then copy the file to the initial case.struct
file.

Of course you need init_lapw before the scf!
>    4. I run the SCF normally (without spin orbit or with spin orbit)
>       with 1 k-point.

1 k-point ?  If it is a metal ?

A metallic structure with 2 atoms/cell requires typically about 1000 - 
10000 k-points (and one has to test it!).
If you create a 2x2x2 supercell, the cell has now 16 atoms, you can 
reduce the k-mesh by a factor of 8; for larger supercells even more.

> 
>  
> 
> Am I right?
> 
>  
> 
> Suddhasattwa Ghosh
> 
>  
> 
>  
> 
>  
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
_______________________________________________
Wien mailing list
Wien at zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien