PBE atomization energy of O2

[marcindulak]

Hi,

i have a problem calculating atomization energies with vasp,
for example I get 6.955 eV for PBE O2 compared to reference 6.214
from DOI:10.1063/1.1926272.
E0(O)= -.15275132E+01, E0(O2)= -.10010293E+02.
What is wrong with my input files?
I'm using the PAW_PBE O_h 20Dec2001 potential,
vasp.5.2.11 18Jan11 complex.

Best regards,

Marcin

INCAR O

Code: Select all

INCAR O
 SIGMA = 0.000000
 ENCUT = 1000.000000
 NELECT = 6.000000
 ENAUG = 1500.000000
 PREC = high
 ALGO = Fast
 NBANDS = 6
 ISMEAR = -1
 NELM = 300
 NUPDOWN = 2
 FERDO = 1.00 1.00 0.00 0.00 0.00 0.00 
 FERWE = 1.00 1.00 1.00 1.00 0.00 0.00 
 LASPH = .TRUE.
 LSCALAPACK = .FALSE.
 ISPIN = 2
 MAGMOM = 1*2.0000 

POSCAR O

Code: Select all

 O 
 1.0000000000000000
    15.0000000000000000    0.0000000000000000    0.0000000000000000
     0.0000000000000000   15.0000000000000000    0.0000000000000000
     0.0000000000000000    0.0000000000000000   15.0000000000000000
   1
Cartesian
  7.5000000000000000  7.5000000000000000  7.5000000000000000

INCAR O2

Code: Select all

INCAR O2
 SIGMA = 0.000000
 ENCUT = 1000.000000
 NELECT = 12.000000
 ENAUG = 1500.000000
 PREC = high
 ALGO = Fast
 NBANDS = 9
 ISMEAR = -1
 NELM = 300
 NUPDOWN = 2
 IDIPOL = 4
 FERDO = 1.00 1.00 1.00 1.00 1.00 0.00 0.00 0.00 0.00 
 FERWE = 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.00 0.00 
 DIPOL = 0.5000 0.5000 0.5000 
 LDIPOL = .TRUE.
 LASPH = .FALSE.
 LSCALAPACK = .FALSE.
 ISPIN = 2
 MAGMOM = 2*1.0000 

POSCAR O2

Code: Select all

 O 
 1.0000000000000000
    15.0000000000000000    0.0000000000000000    0.0000000000000000
     0.0000000000000000   15.0000000000000000    0.0000000000000000
     0.0000000000000000    0.0000000000000000   16.2459559999999996
   2
Cartesian
  7.5000000000000000  7.5000000000000000  8.7459559999999996
  7.5000000000000000  7.5000000000000000  7.5000000000000000

IBZKPT

Code: Select all

Automatically generated mesh
       1
Reciprocal lattice
    0.00000000000000    0.00000000000000    0.00000000000000             1

[alex]

Check the DOS if the occupation is what you would expect in both O and O2.

[marcindulak]

Thanks, yes, the energy levels for O look wrong, occupations right:

Code: Select all

 spin component 1

 k-point   1 :       0.0000    0.0000    0.0000
  band No.  band energies     occupation 
      1     -25.2091      1.00000
      2     -10.2088      1.00000
      3     -10.2088      1.00000
      4     -10.2088      1.00000
      5      -0.2964      0.00000
      6       0.2488      0.00000

 spin component 2

 k-point   1 :       0.0000    0.0000    0.0000
  band No.  band energies     occupation 
      1     -21.9620      1.00000
      2      -7.1753      1.00000
      3      -7.1753      0.00000
      4      -7.1753      0.00000
      5      -0.2202      0.00000
      6       0.4133      0.00000

I use ferwe/ferdo keywords to get rid of leaking electrons:
http://cms.mpi.univie.ac.at/vasp-forum/ ... c.php?3.92
What is the right input file to calculate an atom in vasp?
How this has been done in DOI:10.1063/1.1926272?

Marcin

[marcindulak]

Let me answer myself:the problem is in the cell dimensions for the atom -
it must have a broken symmetry. In fact DOI:10.1063/1.1926272 mentions the cell
they used.

Marcin