My Community

Dear VASP community,

I'm trying to make a cell optimization and vould like to find the optimal lattice parameters. I'm using the scheme of set of fixed volume calculation, i.e. make an structure optimization with ISIF = 4 flag for the set of volumes (see Fig. attached)



I encountered a strange situation, that zero external pressure doesn't correspond to the minimum of VASP energy. I've check this approach for "toy case" of bcc-Fe and everything works fine.

My question is what is the correct ground state (zero pressure or minimal energy) OR what I am doing wrong ?
INCAR file is given:

SYSTEM = (Na2 Co3 O6)24 (R-3) ~ 13282 minimized: (Na2 Co3 O6)24 (R-3) PBEsol+U=5 400 5*5*5 LS (l=7/4) (VASP)

PREC = Normal
ENCUT = 400.000
IBRION = 2
NSW = 100
ISIF = 4
NELMIN = 2
EDIFF = 1.0e-05
EDIFFG = -0.02
VOSKOWN = 1
NBLOCK = 1
NWRITE = 1
NELM = 60
ALGO = Normal (blocked Davidson)
GGA = PS
ISPIN = 2
MAGMOM = 16*0 2*0.015 2*-0.015 0.014 -0.014 3*0.017667 3*-0.017667 6*-0.64267 \
6*0.64267 6*-0.0066667 6*0.0066667 6*0.015 6*-0.014667 6*-0.0033333 \
6*0.0033333 6*0.025 6*-0.024333
LDAU = .TRUE.
LDAUTYPE = 2
LDAUL = -1 2 -1
LDAUU = 0 5.0 0
LDAUJ = 0 0 0
LMAXMIX = 4
INIWAV = 1
ISTART = 0
ICHARG = 1
LWAVE = .FALSE.
LCHARG = .FALSE.
ADDGRID = .FALSE.
LHYPERFINE = .FALSE.
ISMEAR = 0
SIGMA = 0.2
LREAL = Auto
RWIGS = 1.54 1.16 0.73

I would recommend fixed cell shape (ISIF=2), as determined in Xray structure analysis,
and fixed NGX,NGY,NGZ grid (taken from the largest cell volume).

Dear admin,
I've performed a set of calculation with ISIF=2 option and NGX=60, NGY=60, NGZ=60 (taken from the largest volume),
however, situation does not change in general ,
energy = {-498.467724, -498.542205, -498.029708962, -496.974405224, \
-495.414677437} eV, whereas pressure = {-37.32, -74.00, -105.54, -132.97, -157.30} kBar
and question remains: why minimum of energy doesn't correspond to the (almost) zero pressure

Hi lysogors,

This is a bit strange for a charge neutral system. Are you sure that convergence was achieved ? I noticed you enforced 2 electronic steps per each ionic step (NELMIN = 2) and I highly doubt that the self-consistency loop will finish in two steps for Co-based system with DFT+U.



Mostafa

Edit for my previous post: I realized that NILMIN is not forcing the electronic steps to 2 but just the minimum of the steps. But still it is worth to check weather convergence was achieved for each calculation.


Mostafa

Dear Mostafa,

Regarding the convergence of calculations - yes, sure I've check it firstly,
here, for example, last lines of OSZICAR (for volume with minimum energy)
...
N E dE d eps ncg rms rms(c)
DAV: 1 -0.498541816338E+03 -0.51417E-05 -0.93081E-04 30088 0.786E-02 0.170E-02
DAV: 2 -0.498541822891E+03 -0.65527E-05 -0.44347E-05 23728 0.544E-02
7 F= -.49854182E+03 E0= -.49852372E+03 d E =-.714289E-03 mag= 0.0009
N E dE d eps ncg rms rms(c)
DAV: 1 -0.498542093815E+03 -0.27748E-03 -0.32077E-02 28832 0.371E-01 0.526E-02
DAV: 2 -0.498542205515E+03 -0.11170E-03 -0.10712E-03 35824 0.110E-01 0.592E-02
DAV: 3 -0.498542207497E+03 -0.19828E-05 -0.12781E-04 26952 0.107E-01 0.425E-02
DAV: 4 -0.498542204614E+03 0.28839E-05 -0.22578E-05 17504 0.544E-02
8 F= -.49854220E+03 E0= -.49852412E+03 d E =-.381723E-03 mag= -0.0001

however, external pressure in OUTCAR file is far from the zero
...
Total -43.28598 -43.28598 -45.60300 0.00000 0.00000 0.00000
in kB -78.55151 -78.55151 -82.75622 0.00000 0.00000 0.00000
external pressure = -79.95 kB Pullay stress = 0.00 kB
...

Yury

Here are some more details:
LREAL=.FALSE. ! .TRUE. can deform forces
PREC=Accurate
EDIFF=1e-6
EDIFFG=-1e-2
ICHARG=0 ! ICHARG=1 can cause undesirable effects (cf. http://cms.mpi.univie.ac.at/vasp/vasp/ICHARG_tag.html )

It is also useful each relaxed structure to restart once more from CONTCAR