Dear Alex and all ATAT users,
I’m trying to generate sqs for bcc random solid solution(Nb-Zr-Ni) with mcsqs, Niobium atoms are partially replaced by Zirconium and Nickle atoms .
The following is my rndstr.in:
3.305 3.305 3.305 90 90 90
0.5 -0.5 0.5
0.5 0.5 -0.5
-0.5 0.5 0.5
0 0 0 Nb=0.407407407 Zr=0.296296296 Ni=0.296296296
My sqscell.out is :
1
3 0 0
0 3 0
0 0 3
I generated the clusters with typing :
corrdump -l=rndstr.in -ro -noe -nop -clus -2=6 -3=4
mcsqs -rc -n=54
Thus, I got the bestsqs.out after a few seconds.
Then , I type :
str2cif <bestsqs.out >bestsqs.cif
I got a BCC ternary structure .
Now, I have some questions :
1M-cM-^@M-^AIs my rndstr.in right? The '3.305' is the lattice constant of Nb.
2M-cM-^@M-^AI don't know how to type the -2=? -3=? ...-n=?
3M-cM-^@M-^AI need to add a hydrogen in the tetrahedron position of the SQS, so I hope I can get the BCC structure with the mcsqs(although I know -rc will reduce the optimal point ),I want to know if the structure I got with -rc(the BCC ternary alloy) is reasonable.I need to calculate the dissolution and diffusion of the BCC ternary alloy after hydrogen with vasp.
4M-cM-^@M-^ADoes the bestsqs.out take only several seconds? Does the bestsqs.out contain only one structure ? Is the structure best?
Your input files look correct.
It is possible that the code finds a solution after a few seconds.
You could try to increase the cutoff for -2=… and -3=… and you will see that mcsqs will take longer (it may even never stop, which is normal - in that case you just stop it when the solution is sufficient good for your purposes).
The only secure way to tell if the sqs is good enough is to try different ranges and see that the calculated energy converges. But sometimes you just pick the largest cell you can run and increase the -2= and -3= until mcsqs cannot find any better solutions.
Dear Alex,
Thanks for your reply.
I increase the -2=…,but I don’t know the reference standard,also in the mcsqs.log there is the word "correlation mismatch"instead of the "perfect match" .
You don’t need to have a "perfect match" in the log file. If you specify many correlations to match, this may never happen. But that’s ok, at any given time mcsqs reports the best sqs found so far.
The similar questions with yours I got are that how to deal with -2=…-3=…-4= and is it reasonable to extend the 333 cell to 336 cell to get different impurity concentrations (by interstitial atom).
Have you got any progress ?
Yes, you can generate SQS with interstitials, but you have to make sure if its plausible for the system you consider. (Often, the interstitials are rare because they are enegetically costly.)
Yes, you can extend the cell by modifying your sqscell.out file (with the -rc option) or increasing the number of atoms (-n option, no -rc option).