Graphite interlayer distance
Task
In this example you will determine the interlayer distance of graphite in the stacking direction using the method of Tchatchenko and Scheffler to account for van der Waals interactions.
Semilocal DFT at the GGA level underestimates long-range dispersion interactions. This problem causes a bad overestimation of graphite lattice in the stacking direction: 8.84 Å (PBE) vs. 6.71 Å (exp).
In this example, the dispersion correction method of Tchatchenko and Scheffler is used to cope with this problem.
Input
POSCAR
graphite 1.0 1.22800000 -2.12695839 0.00000000 1.22800000 2.12695839 0.00000000 0.00000000 0.00000000 7.0 4 direct 0.00000000 0.00000000 0.25000000 0.00000000 0.00000000 0.75000000 0.33333333 0.66666667 0.25000000 0.66666667 0.33333333 0.75000000
INCAR
IVDW = 20 LVDW_EWALD =.TRUE. NSW = 1 IBRION = 2 ISIF = 4 PREC = Accurate EDIFFG = 1e-5 LWAVE = .FALSE. LCHARG = .FALSE. ISMEAR = -5 SIGMA = 0.01 EDIFF = 1e-6 ALGO = Fast NPAR = 2
KPOINTS
Monkhorst Pack 0 gamma 16 16 8 0 0 0
Running this example
To run this example, execute the run.sh bash-script:
#
# To run VASP this script calls $vasp_std
# (or posibly $vasp_gam and/or $vasp_ncl).
# These variables can be defined by sourcing vaspcmd
. vaspcmd 2> /dev/null
#
# When vaspcmd is not available and $vasp_std,
# $vasp_gam, and/or $vasp_ncl are not set as environment
# variables, you can specify them here
[ -z "`echo $vasp_std`" ] && vasp_std="mpirun -np 8 /path-to-your-vasp/vasp_std"
[ -z "`echo $vasp_gam`" ] && vasp_gam="mpirun -np 8 /path-to-your-vasp/vasp_gam"
[ -z "`echo $vasp_ncl`" ] && vasp_ncl="mpirun -np 8 /path-to-your-vasp/vasp_ncl"
#
# The real work starts here
#
rm results.dat
for d in 6.5 6.6 6.65 6.7 6.75 6.8 6.9 7.0
do
cat>POSCAR<<!
graphite
1.0
1.22800000 -2.12695839 0.00000000
1.22800000 2.12695839 0.00000000
0.00000000 0.00000000 $d
4
direct
0.00000000 0.00000000 0.25000000
0.00000000 0.00000000 0.75000000
0.33333333 0.66666667 0.25000000
0.66666667 0.33333333 0.75000000
!
$vasp_std
cp OUTCAR OUTCAR.$d
energy=$(grep "free ene" OUTCAR.$d|awk '{print $5}')
echo $d $energy >> results.dat
done
The optimal length of the lattice vector c normal to the stacking direction is determined in a series of single point calculations with varied value of c (all other degrees of freedom are fixed at their experimental values).
The computed c vs. energy dependence is written in the file results.dat and can be visualized e.g. using xmgrace.
The optimal value can be obtained using the attached utility (python with numpy or Numeric is needed):
./utilities/fit.py results.dat
This should yield:
200 iterations performed
Ch-square: 4.30305519481e-09
---------
E0(eV): -37.433456779
d0(A): 6.65603352689
The computed value of 6.66 Å agrees well with experiment (6.71 Å).