Crystal structure optimization with ISIF=3

[okuno]

Dear Vasp Users.

I want to optimize crystal structure not only the atomi positions, but also
crystal volume and shape (lattice parameter).

Because I want to compare the mimimum enegy of rhombohedral
and tetragonal structure. I had calculated the rhombohedral
energy mimumum and tetragonal mimumum with ISIF=3.

But the calculated results is insonsistent to experiments
, the crystal structure I had calculated is rhombohedral structure in
experiments but the results of calculation was that the tetragonal
structure had lower energy than tetragonal one.

I had sets PREC=High in the calculation.

If we set ISIF=3, can we get the minimum of energy with lattice volume and
lattice shape (lattice parameter)?

Or if we want to get the mimimum of energy with lattice volume and lattice shape,
is it safe to find the minimum of energy with canging lattice parameter with fixed volume, and then get the plot of mimimum energy with volume.

Sincerely.

[admin]

please note that optimizing the structure with ISIF has to be done very carefully:
1) in addition to PREC=high, also set ENCUT = 1.3*max(ENMAX) of the PPs (or even higher)
2) you have to check that the k-meshes are really converged (for each of the structures)
3) to adjust the basis sets to the changed volumes/shapes of the cells, always pre-converge the cells first and re-start from the last CONTCAR (WITHOUT using WAVECAR) once again
4) make sure you use the approriate BZ-integration algorithm , according to the electronic structure of your system (ISMEAR, SIGMA)
5) set the convergence criteria accurate enough (EDIFF, EDIFFG): EDIFF should be at least 2 orders of magnitude smaller than the expected differences of the total energies of the structures
6) compare the sigma-->0 energies.