ATAT has been used by Zaddach et al. to produce the special quasirandom structure of high entropy alloys of Ni-Fe-Cr-Co-Mn (JOM, Vol. 65, No. 12, 2013).
Is it possible to use mcsqs code to create Mg based quaternary and quinary alloys. They are not necessarily high entropy alloys and may include constituent elements with having 1% at. or less than 1% at. compositions.
The resulting SQS may be used to find the phase diagrams and stacking fault energies.
Thanks
mcsqs should be able to handle this - but keep in mind that <1% composition means a supercell of at least 100 atoms and probably more since other constraints need to be satisfied for a good SQS!
thaks!
Dear ATATers,
I am trying to use corrdump to generate a clusters.out file (to be used as input to mcsqs).
1- I wonder what to set -2=… and -3=…
corrdump -l=randstr.in -ro -noe -nop -2=10 -3=10 -clus
I have set them randomly to 10. I wonder what is the best setting? Apparently I cannot just ignore them.
2- What is the typical duration time for corrdump to run. After almost 30 minutes it was killed and only a sym.out file was generated.
Thanks
Is there any way to solve the below problem other than increase the number of atoms in the supercell?
mcsqs -n=32 -l=randstr.in -cf=clusters.out
Impossible to match point correlations due to incompatible supercell size.
Thanks
There is no "best" setting for the range of correlations that have to match the disordered states.
Larger numbers give you the potential to reach a higher accuracy, but the resulting SQS need to be larger. "Best" depends on your trade-off between accuracy and computational cost.
You should read the original references on SQS if you want to understand better what these correlations range mean:
In terms of time, it usually takes a few minutes to run corrdump (but 10 is probably a rather long range (too long), which implies hundreds of clusters, which could take a long time…)
I suggest you start small (with -2 only and set it to the nearest neighbor distance) and see how long it takes. Gradually increase it as needed.
The message: "Impossible to match point correlations due to incompatible supercell size." Just tells you that give the concentration you gave as input, it is impossible to even contruct a single supercell of the size you request that would have that composition. For instance, if you have compositions like A=0.70,B=0.25,C=0.05 on one site, you’d need a multiple of 20 atoms per supercell.
Many thanks
You mentioned:
"For instance, if you have compositions like A=0.70,B=0.25,C=0.05 on one site, you’d need a multiple of 20 atoms per supercell."
Has this "multiples of 20 atoms" been inferred from dividing 100 by the least atomic % of the constituent element?
e.g. if A=0.70,B=0.28,C=0.02, then
the least at. % is 2percent, and a multiple of 50 atoms per supercell is needed?
yes. Express compositions as fractions, find least common denominator.