One ordered sublattice, one disordered in mcsqs

Alloy Theoretic Automated Toolkit (ATAT)
1

I am attempting to use mcsqs to generate a SQS of a perovskite which has an A-site cation sublattice ordered between two species, a single species occupying the B-site, and an anion sublattice that is disordered between two species.

The two pieces of relevant information I’ve found are

  1. The help for corrdump, which says "When only one symbol is listed, this site is ignored for the purpose of calculating correlations, but not for determining symmetry."

  2. An excerpt from Efficient Stochastic Generation of Special Quasirandom Structures which says "All symmetry equivalent sites must also have the same occupations. To override this requirement, `dummy’ species (with zero occupation) that differ in the two sublattices must be included."

In my first attempt, I described two different A-site cations as shown

3.933290 0.000000 0.000000
0.000000 3.933290 0.000000
0.000000 0.000000 7.565200
1.000000 0.000000 0.000000
0.000000 1.000000 0.000000
0.000000 0.000000 1.000000
0.500000 0.000000 0.250000 O=0.83333,F=0.16667
0.500000 0.000000 0.750000 O=0.83333,F=0.16667
0.500000 0.500000 0.000000 O=0.83333,F=0.16667
0.500000 0.500000 0.500000 O=0.83333,F=0.16667
0.000000 0.500000 0.250000 O=0.83333,F=0.16667
0.000000 0.500000 0.750000 O=0.83333,F=0.16667
0.000000 0.000000 0.000000 Ba
0.000000 0.000000 0.500000 Y
0.500000 0.500000 0.250000 Fe
0.500000 0.500000 0.750000 Fe

However, it ignored the cations altogether when finding clusters, only considering anion-anion interactions. I then attempted

0.000000 0.000000 0.000000 Ba=1,Y=0
0.000000 0.000000 0.500000 Y=1,Ba=0
0.500000 0.500000 0.250000 Fe=1,La=0
0.500000 0.500000 0.750000 Fe=1,La=0

using "dummy" species to override the symmetry requirement. This failed with the error message
"Please drop irrelevant species when site is fully occupied with one species", so I tried

0.000000 0.000000 0.000000 Ba=1,Mn=0
0.000000 0.000000 0.500000 Y=1,La=0
0.500000 0.500000 0.250000 Fe
0.500000 0.500000 0.750000 Fe

which failed with the same message. What is the correct way to specify this structure so that it recognizes the A-site (Ba and Y) is ordered, and considers Ba-O, Ba-F, Y-O, and Y-F clusters to be four separate categories?

2

Your first attempt is the correct way to proceed:

3.933290 0.000000 0.000000
0.000000 3.933290 0.000000
0.000000 0.000000 7.565200
1.000000 0.000000 0.000000
0.000000 1.000000 0.000000
0.000000 0.000000 1.000000
0.500000 0.000000 0.250000 O=0.83333,F=0.16667
0.500000 0.000000 0.750000 O=0.83333,F=0.16667
0.500000 0.500000 0.000000 O=0.83333,F=0.16667
0.500000 0.500000 0.500000 O=0.83333,F=0.16667
0.000000 0.500000 0.250000 O=0.83333,F=0.16667
0.000000 0.500000 0.750000 O=0.83333,F=0.16667
0.000000 0.000000 0.000000 Ba
0.000000 0.000000 0.500000 Y
0.500000 0.500000 0.250000 Fe
0.500000 0.500000 0.750000 Fe

The fact that the code does not least anion-cation interaction is completely normal. In the cluster expansion formalism, when the cation sublattice is not disordered, the anion-cation interaction can be fully accounted for by the point interaction on the anion sublattice.

The requirement that the symmetrically equivalent sublattices have the same composition is not a problem in your case as the code already sees different species on the different sites (Ba not = Fe, for instance), so they are not considered equivalent.

Thanks for pointing out a glitch in the manual/code, indeed there is a check for zero occupation that prevents doing what is said in the paper. I’ll fix that - although it is a rare situation.