6
$\begingroup$

I did an optimization for a monolayer with GGA (550 eV ecut, 12 12 1 k-mesh) with ISIF=3 and IBRION=2.

My original POSCAR was:

5.2916193008         0.0000000000         0.0000000000
0.0000000000         5.2916193008         0.0000000000
0.0000000000         0.0000000000        20.0097122192

After optimization, my CONTACR is this:

 5.2753749526661906    0.0000000006352637    0.0000000700426255
-0.0000000006164817    5.2753749536648096   -0.0000000640637957
 0.0000002625200859   -0.0000002421078825   19.9912853387061134

You can see my lattice vectors (off-diagonal) has some non-zero values now and vacuum size is reduced a bit.

Do you think my CONTCAR is okay for further calculations, or should I worry about that?

NOTE: I did not fix any axis for the optimization.

$\endgroup$
1
  • 2
    $\begingroup$ For a 2D system, you can't use ISIF=3. $\endgroup$
    – Jack
    Aug 22 at 13:55
6
$\begingroup$

You used ISIF = 3, which tells VASP you want to relax the atomic positions and the cell shape/volume. It looks like the cell shape/volume has changed slightly during the relaxation process, as you noted. Based on this information, the input lattice constants yield a structure with higher energy than those in the converged CONTCAR. It seems like VASP is doing precisely what you asked it to do.

On a related note, if you have a monolayer with lots of vacuum in the $z$-dimension, it does not seem wise to relax all three lattice constants. You probably want to keep the vacuum dimension fixed.

$\endgroup$
2
  • $\begingroup$ Thanks, Sir for a prompt response. I have updated the question for an extended query. $\endgroup$
    – astha
    Aug 22 at 6:50
  • 2
    $\begingroup$ @astha I suggest you ask a new question if you have a new question. You can provide a link to this question there if you feel it's appropriate. $\endgroup$ Aug 22 at 13:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.