My Community

I'm fairly new to VASP and have encountered problems when trying to perform a relaxation on NiO. The lattice vectors shrink too much and the pressure reaches -242.95 kB. I've searched for a solution and changed my INCAR file several times. Has anyone else had problems like this or have any suggestions?


Here is one of my INCAR files:

SYSTEM = NiO

PREC = Medium
IBRION = 2
NSW = 0
ISIF = 2
NELM = 200
EDIFF = 1.0e-06
ENCUT = 400
ISTART = 0
ICHARG = 2
LWAVE = .True.
LCHARG = .True.
IALGO = 38
LREAL = .False.
ISMEAR = 0
SIGMA = 0.01
GGA = PE
NPAR = 4

I'm not sure about this, because I also new to VASP, but I experiences the same thing like you did when I used ISIF=2, but by the time I change it to ISIF=0 everything seems to be okay, because I only needed h ionic relaxations

Section 7.6 explain about this one and they say that this is can also happening because of your precision is medium, because they need ENCUT = 1.3*default, so may be high is much more better, but I'm not sure because I also still learning about this,

ISIF = 3

I have tried using ISIF = 3, but it reduces the lattice parameters of my cell (NiO has same structure as NaCl) by a factor of 2. I want to relax the parameters, but I still need them to agree somewhat with experimental observations.
ISIF = 0 is similar to ISIF = 2, but it does not calculate the stress tensor, which can be used to check that the cell is truly relaxed.
I just tried to use a larger cell (2x2x4) with the same lattice parameter, but the same thing happened and my parameters were decreased by a factor of about 2. I understand that NiO is a magnetic material; would that cause a problem in my relaxation calculations?

Hi nate84,

you've hit two traps:
- plain DFT does not work with strongly correlated systems like NiO,
- 400eV as energy cutoff is too small for lattice optimisation (Pulay stress). You may come around this if you use a fitting technique.

Cheers,

alex