A question about SQS

Alloy Theoretic Automated Toolkit (ATAT)
1

Dear Prof. Dr. A. van de Walle,

I use ATAT to get a 16 atom supercell (222) SQS structure. But I don’t think I got the correct result. Could please help me check which step I did wrong? Thank you very much.
The details are as follows:

  1. I prepare rndstr.in (bcc structure with 2 atoms in unit cell, lattice constant a=3.43)

3.43 3.43 3.43 90 90 90

0.5 0.5 0.5

0.5 -0.5 0.5

0.5 0.5 -0.5

0 0 0 A=0.75, B=0.25

  1. corrdump -l=rndstr.in -ro -noe -nop -clus -2=9 -3=6

  2. I prepare supercell.in

2 0 0

0 2 0

0 0 2

  1. cellcvrt -f < rndstr.in | cellcvrt -uss=supercell.in -f | head -3 > sqscell.out

  2. I add number 1 in the first line of sqscell.out, as follows

1

3.43 3.43 3.43

3.43 -3.43 3.43

3.43 3.43 -3.43

But accoring to the manual, the sqscell.out should be given in multiple of the coordinate system, so typically integers…did I get the correct sqscell.out??

  1. mcsqs -n = 16 -rc

In the bestsqs.out file, I found that there are only 8 atoms (but in fact I need 16 atoms!!!), with 6 A and 2 B atoms. What’s the worse, I got the the stucture with very low symmetry, alpha=beta=90, but gamma=120!!! It is hexagonal…It is not what I want…

Could you please give me some tips??? Thank you very much…

2

Your step 4 is not needed. Your sqscell.in should just be the same as your supercell.in but with a 1 on top (assuming you are using version 3.07 or above).
Then run
mcsqs -rc

BTW, why not just run mcsqs -n=16 and forget about specifying your own supercell ? 16 atom is small enough so the automatic supercell generation should be quick.

3

Dear Prof. Dr. A. van de Walle,

Thanks a lot for your reply. I tried your method, and got three bestsqs.out structures, i.e., A_0.25B_0.75, A_0.5B_0.5, and A_0.75B_0.25.

In order to transform the bestsqs files to CIF files, I did like this:
cp bestsqs.out str.out
runstruct_vasp
Then I got POSCAR.relax and POSCAR.static, and open POSCAR.static with vesta.
I found that A_0.75B_0.25 and A_0.25B_0.75 structures are with low symmetry, alpha=gamma=90 degree, but beta is not 90 degree…
Is it correct??? I thought that the supercell should at least keep the symmetry, i.e., alpha=beta=gamma=90 degree…

btw, is there any method which could directly transform the bestsqs.out to CIF files???
According to the manual, str2cif can converts an ATAT structure file to the CIF. But how to ?
I did not find the format of stdin or stdout in the manual…

Thank you very much.

4

It is completely normal for a SQS to have a lower symmetry than the unit cell and for the lattice vectors to relax away from the starting high-symmetry angles.
It is also completely normal for the supercell to be quite different from the unit cell.
(Have a look at the original SQS references.)
The syntax to convert to CIF is simply

str2cif <  bestsqs.out  >  bestsqs.cif

stdin and stdout are the usual UNIX names for standard input and standard output
(google these terms)