How to find a good SQS structure for Al-Co-Cr-Ni-Fe systems

Alloy Theoretic Automated Toolkit (ATAT)
1

Dear Members in ATAT forum
I would like to generate BCC sqs structure with Al-Co-Cr-Ni-Fe systems. Here are the input file parameters.
SQS 1
rndstr.in
1 1 1 90 90 90
0.5 0.5 0.5
0.5 -0.5 0.5
0.5 0.5 -0.5
0 0 0 Al=0.45,Fe=0.1,Cr=0.1,Co=0.25,Ni=0.1
Runjob file
#!/bin/bash
#SBATCH --time=268:00:00
#SBATCH -p long
#SBATCH --ntasks-per-core=8
#SBATCH -J mcsqs

./corrdump -l=rndstr.in -ro -noe -nop -clus -2=1; ./getclus
while true; do
if [ -f clusters.out ]; then
for (( id=0 ; id<8 ; id++ )); do
./mcsqs -n=80 -wr=20 -wn=0.75 -wd=1 -ip=$id &
done
wait
break
else
sleep 5
fi
done

When I run the job file for 3 days, it cannot find a SQS structure with good correlation functions. The objective functions stop at -19 and not drop anymore.

Objective_function= -19.991618
Correlations_mismatch= 0.015581 -0.012904 0.003125 0.007102 0.001276 -0.020408 0.000349 0.001606 -0.003046 0.009661 0.022707 0.029062 0.008333 -0.005058 0.010731 0.004221 -0.003727 -0.002915 -0.004492 0.002811
Objective_function= -19.992368
Correlations_mismatch= 0.034543 -0.003420 -0.002734 -0.012500 0.005437 -0.007266 -0.009259 -0.005637 0.004869 0.001594 0.000644 0.017463 -0.007292 0.007587 -0.009553 -0.006332 -0.001398 -0.001685 -0.003446 -0.008155
Objective_function= -19.992532
Correlations_mismatch= 0.002941 -0.006743 0.007031 0.018340 0.001276 -0.008870 0.000349 0.006434 0.006922 0.009661 -0.009773 0.010295 0.016146 0.011171 0.005522 -0.007178 0.002096 -0.012102 -0.005138 -0.002397
Objective_function= -19.993217
Correlations_mismatch= -0.002458 -0.006258 -0.002734 0.004114 -0.002788 0.003002 0.023933 -0.007699 -0.007153 -0.013619 -0.007783 0.000787 -0.004688 0.002957 -0.015101 0.006113 -0.000233 0.001533 0.019106 0.002601
Objective_function= -19.993266
Correlations_mismatch= -0.001888 -0.002151 0.022656 0.003480 0.011758 0.000314 -0.013626 0.019075 -0.000508 -0.000820 -0.007783 -0.005735 -0.004688 0.011171 0.006412 -0.007178 -0.002562 0.001533 -0.001354 -0.008495
Objective_function= -19.993730
Correlations_mismatch= -0.004872 0.001957 -0.004687 0.004264 0.006105 -0.008386 0.000349 -0.001378 0.006138 -0.002980 -0.016211 -0.002997 -0.002083 0.017094 -0.000236 0.017912 -0.006056 0.004752 0.003077 -0.017472
Objective_function= -19.993825
Correlations_mismatch= -0.011192 -0.007527 0.007031 0.000157 0.003278 -0.001740 -0.001398 0.001036 -0.001777 -0.004059 -0.016211 0.005864 -0.002083 0.016448 -0.024628 -0.000856 0.000932 0.004752 -0.002400 -0.013914

I just wonder if there are some problems about my input file or this composition is difficult to generate?

2

Yes, this is a difficult composition, here is why.
To match the right frequency of Al-Fe pairs, you would need a fraction 0.45*0.1=9/200 of the pairs to be of Al-Fe type. Since in bcc there are 4 nn pairs per atom, this means your cell must be a multiple of 50 atoms.

To match the right frequency of Al-Co pairs, you would need a fraction 0.45*0.25=9/80 of the pairs to be of Al-Co type. This means your cell must be a multiple of 20 atoms.

The smallest multiple of 50 and 20 is 100… and I have not checked the other pairs…

3

Thank you very much!

4

There are 6 pairs/atom in fcc. Yes 200 is not divisible by 6. Then write:
9/200 = 27/600. And then 600/6=100, so your cell must be a multiple of 100.

Do the same for 9/80= 27/240. And 240/6=40, so your cell must be a multiple of 40.

The smallest multiple of 40 and 100 is 200.

This 0.45 composition is what is limiting you.

5

Thank you Alex
How long it will take for generating a 200 atoms SQS with mcsqs code?

6

Dear Axel
Another questions, if I want to generate the fcc random supercell with the same composition, what’s the smallest multiple should I get? it seems that there are 3 nn pairs per atom in fcc lattice, but 200/3 is not an integer.

7
8

dear nanyanshouhu
I have a question,can we communicate with Wechat or e-mail?
wechat:13704800632
qqmail:[email protected]