Based on the VASP wiki example in this link
Task: Relaxation of the internal coordinates, volume and cell shape in cubic diamond Si. Also see the discussion on energy vs volume, volume relaxations and Pulay stress
There are several ways in which the equilibrium volume can be determined
The first approach was done in the previous example cd Si. Here, the second approach will be used. While it’s fast, one needs to be careful regarding the accuracy and consider the Pulay stress.
First, copy the example folder which contains the VASP input files
cp -r /software/sse/manual/vasp/training/ws2022/cd_Si_vol_relax .
cd cd_Si_vol_relax
and copy the latest POTCAR file for Si:
cp /software/sse/manual/vasp/POTCARs/PBE/2015-09-21/Si/POTCAR .
POSCAR
cubic diamond Si
5.5
0.0 0.5 0.5
0.5 0.0 0.5
0.5 0.5 0.0
Si
2
Direct
-0.125 -0.125 -0.125
0.125 0.125 0.125
INCAR
System = diamond Si
ISMEAR = 0
SIGMA = 0.1
ENMAX = 240
IBRION = 2
ISIF = 3
NSW = 15
EDIFF = 0.1E-04
EDIFFG = -0.01
EDIFFG=-0.01, if negative value, stop criterium for ionic relaxation if forces are smaller than | EDIFFG |
KPOINTS
k-points
0
Monkhorst Pack
11 11 11
0 0 0
Start the volume relaxation calculation by
sbatch run.sh
Monitor how the structural relaxation progresses with
cat OSZICAR
Check the relaxed final structure
cat CONTCAR
Compare with the EOS lattice parameter obtained in the previous example and from the VASP wiki a= 5.4687 Å
Note that one doesn’t use the total energy or DOS from a structural relaxation. For that, CONTCAR should be copied to POSCAR for a separate caclulation with no structural relaxation.
Check the convergence of the total energy and forces using p4vasp
p4v &
by selecting Convergence > Energy or > Forces.
Repeat the volume relaxation calculation in a new folder, but this time start from the relaxed structure CONTCAR
, by copying it to POSCAR. Does the structure change, if so, how much?
Repeat the volume relaxation calculation from scratch with higher accuracy, e.g. set ENCUT to 1.3 x ENMAX (grep ENMAX POTCAR) and PREC=Accurate in INCAR. Also, set a smaller value for EDIFF. Any difference in the lattice parameter a?
Guides, documentation and FAQ.
Applying for projects and login accounts.