Hello everyone, I wanted to know the proper workflow of DFT calculations using VASP, from optimization to band structure calculations, as I am new to DFT. Kindly tell me if I am anywhere wrong in the following steps.
1. We take the cif/POSCAR file from websites like ICSD or Materials project and do geometric optimization. We choose a proper k-mesh to sample the k-space. We should put ISIF = 3, i.e., full relaxation (atomic position, cell shape, and volume).
2. Use the CONTCAR from the relaxation step as the new POSCAR for SCF calculations.
3. Perform a single-point energy calculation (SCF) with a stricter EDIFF value than in the relaxation step.
4. The CHGCAR and WAVECAR files generated from the SCF calculation should be carried forward for DOS and Band structure calculations. For DOS, a denser k-mesh should be chosen to sample the whole Brillouin zone.
5. Again, for band structure calculation, the same CHGCAR and WAVECAR files should be used, and a high-symmetry path should be chosen in the KPOINTS file.
Doubts
1. If the file is taken from the Materials Project website, is it already relaxed, or do we need to do geometric optimization?
2. If the EDIFF tag in the INCAR file for ionic relaxation is 10e-5, should it be lesser(like 10e-6) for scf, DOS, and band structure calculations?
3. For DOS calculation, should ISMEAR = -5 or 0? What are the disadvantages of using ISMEAR=0 for DOS?
4. For a calculation starting from scratch, ICHARG = 0 or 2? In VASPWiki, it is written that ICHARG=2 for ISTART=0 (default), but ICHARG=2 freezes the charge density, right? And if I need to continue a calculation, i.e., CHGCAR and WAVECAR files are available, but convergence is not reached, then should I use ICHARG=1 and ISTART=1?
Are there any missing steps in the above workflow, as there can be seen variations in the bandgap values and DOS plots when compared to journals?
