# How to calculate Special Quasirandom Structure with local ordering?

I have a supercell that has 12 layers, each with four atoms. I want to generate an SQS with local ordering at layer numbers 6, 7 and 8. Rest of the layers should be disordered.
I followed the procedure "How to use SQS for disordered materials." but I get the following error:

Impossible to match point correlations due to incompatible supercell size.


The input code (I am considering only the pairs):

mcsqs -2=4
mcsqs -rc


The rndstr.in file(A,B,C,D can be any metallic element):

6.92820300      0.00000000      0.00000000
2.30940100      6.53197300      0.00000000
0.00000000      0.00000000  33.94112500
1   0   0
0   1   0
0   0   1
1   0.5 1   A=.25,B=.25,C=.25,D=.25
1   1   1   A=.25,B=.25,C=.25,D=.25
0.5 1   1   A=.25,B=.25,C=.25,D=.25
0.5 0.5 1   A=.25,B=.25,C=.25,D=.25
0.25    0.25    0.916667    A=.25,B=.25,C=.25,D=.25
0.75    0.25    0.916667    A=.25,B=.25,C=.25,D=.25
0.75    0.75    0.916667    A=.25,B=.25,C=.25,D=.25
0.25    0.75    0.916667    A=.25,B=.25,C=.25,D=.25
1   1   0.833333    A=.25,B=.25,C=.25,D=.25
1   0.5 0.833333    A=.25,B=.25,C=.25,D=.25
0.5 1   0.833333    A=.25,B=.25,C=.25,D=.25
0.5 0.5 0.833333    A=.25,B=.25,C=.25,D=.25
0.75    0.25    0.75    A=.25,B=.25,C=.25,D=.25
0.25    0.75    0.75    A=.25,B=.25,C=.25,D=.25
0.25    0.25    0.75    A=.25,B=.25,C=.25,D=.25
0.75    0.75    0.75    A=.25,B=.25,C=.25,D=.25
0.5 0.5 0.666667    A=.25,B=.25,C=.25,D=.25
0.5 1   0.666667    A=.25,B=.25,C=.25,D=.25
1   1   0.666667    A=.25,B=.25,C=.25,D=.25
1   0.5 0.666667    A=.25,B=.25,C=.25,D=.25
0.25    0.25    0.583333    B
0.75    0.25    0.583333    C
0.75    0.75    0.583333    C
0.25    0.75    0.583333    D
1   1   0.5 B
1   0.5 0.5 C
0.5 1   0.5 D
0.5 0.5 0.5 A
0.25    0.75    0.416667    B
0.25    0.25    0.416667    D
0.75    0.75    0.416667    A
0.75    0.25    0.416667    A
0.5 0.5 0.333333    A=.25,B=.25,C=.25,D=.25
0.5 1   0.333333    A=.25,B=.25,C=.25,D=.25
1   1   0.333333    A=.25,B=.25,C=.25,D=.25
1   0.5 0.333333    A=.25,B=.25,C=.25,D=.25
0.25    0.25    0.25    A=.25,B=.25,C=.25,D=.25
0.75    0.75    0.25    A=.25,B=.25,C=.25,D=.25
0.75    0.25    0.25    A=.25,B=.25,C=.25,D=.25
0.25    0.75    0.25    A=.25,B=.25,C=.25,D=.25
0.5 0.5 0.166667    A=.25,B=.25,C=.25,D=.25
0.5 1   0.166667    A=.25,B=.25,C=.25,D=.25
1   0.5 0.166667    A=.25,B=.25,C=.25,D=.25
1   1   0.166667    A=.25,B=.25,C=.25,D=.25
0.25    0.75    0.083333    A=.25,B=.25,C=.25,D=.25
0.75    0.75    0.083333    A=.25,B=.25,C=.25,D=.25
0.75    0.25    0.083333    A=.25,B=.25,C=.25,D=.25
0.25    0.25    0.083333    A=.25,B=.25,C=.25,D=.25


Please suggest if I am doing something wrong, or is there anything else further which I need to specify to the ATAT MCSQS code?

• Could you post the piece of input code that you think is giving you trouble? I imagine that would be required to debug this Jun 4 '20 at 16:30
• +1. However you ask us to: "Please suggest if I am doing something wrong", but how can we suggest what you did wrong if we do not even know what you did? You then ask "or is there anything else further which I need to specify to the ATAT MCSQS code?" but how can we tell you what else to do if we do not even know what you have done at all? Jun 4 '20 at 16:44
• Can you add the contents of the rndstr.in file you used for this calculation? In my experience with mcsqs, I've noticed it requires more than the # of atoms you specify in the rndstr.in file to generate possible structures when you aren't making a completely random structure (i.e. when you have certain sites with order). Jun 7 '20 at 18:51
• @HitanshuSachania rndstr.in file is added which has local ordering. Jun 8 '20 at 8:55
• Thank you Sufyan. That took a while. Took so long that the question now got closed and cannot be answered, I have now voted to re-open it. Jun 8 '20 at 14:56

Often we are hung up on the idea of making sure we generate an SQS of a size (i.e. # of atoms) that matches the composition of our system. For instance, in the case of $$A_3B$$ we would generally take multiples of 4 as the # of atoms. This isn't always enough, especially in your case. Since you want to make a cell that is ordered in certain layers and disordered in others, you are limiting yourself in terms of # of atoms required to generate the SQS.
This 'integer # of atoms and corresponding composition' constraint also applies to sub-lattices in your cell. Sometimes the overall cell might comply with this constraint but certain symmetry equivalent sites may not. I don't know how exactly does one determine sublattices in a partially ordered cell but corrdump seems to be the way out - read here.