I understand that my question is similar to what Mythreyi asked here: Generating special quasirandom structures (SQS) for surface calculations?, but based on the rules this website specifies, I thought it was better to start a new question.
My research is on gas-surface reactions. I generated SQS using ATAT mcsqs, and optimized the structure using VASP. I then built the surface using VASP optimized bulk structure, added the vacuum space so that I can study gas-surface reaction, and tried to optimize the structure again with VASP, but it failed. It seems this failure supports Mythreyi's concern.
I wonder if anyone had ever successfully developed the surface for SQS.
Edit: Hello Nike, thank you for your comments. I have some updates. When I tried to optimize the geometry the first time, I allowed the atom position, Cell shape, and Cell volume to change, and VASP failed to converge. I, then, performed the following three restricted calculations, and all three calculations were converged.
Allow atom position and cell volume to change. E0 = -436.77914 eV after 18 calculation cycles
Allow only atom position to change. E0 = -436.77971 eV after 18 calculation cycles
Allow atom position and cell shape to change. E0 = -436.82406 eV after 22 calculation cycles
The total energies for #1 and #2 were similar. #3 was different from the other two, but it had a lower energy.
I greatly appreciate it if anyone can comment on these results, such as which one is the most reasonable choice.
I'll post my inputs and possibly outputs later.