Page 1 of 1

Artifactual charged states when using LAECHG tag?

Posted: Fri Sep 10, 2010 5:54 pm
by camilo
Hi,

I am using vasp.4.6.35 3Apr08, and am running a test calculation on a single water molecule:

O H
1.000000000
15.000000000 0.000000000 0.000000000
0.000000000 15.000000000 0.000000000
0.000000000 0.000000000 15.000000000
1 2
CARTESIAN
7.500000000 7.907680000 7.500000000
8.291224000 7.296160000 7.500000000
6.708776000 7.296160000 7.500000000


The INCAR is:

SYSTEM = Water
NWRITE = 4
ISMEAR = 1
SIGMA = 0.005
ENCUT = 700.0
PREC = HIGH
VOSKOWN = 1
EDIFF = 1.0E-05
ISYM = 0
IBRION = 2
NSW = 100
POTIM = 0.50
LAECHG = .TRUE.

And the POTCAR I am using is the following:

TITEL = PAW_GGA O_h 04May1998
TITEL = PAW_GGA H_h 04May1998


The standard output of the code has a few remarks in it concerning the number of electrons in the system, and the final output has the following:

augmentation electrons 5.951938207320587E-002
soft electrons 7.94117696309909
augmentation electrons 5.951938207320587E-002
soft electrons 7.94117696309909
augmentation electrons 5.951938207320587E-002
soft electrons 7.94117696309909
augmentation electrons 5.951938207320587E-002
soft electrons 7.94117696309909
total electrons 8.00069634517230
total electrons 8.00069634517230
total electrons 8.00069634517230
total electrons 8.00069634517230

(I am assuming that there is one copy of each statement per CPU, as this job was performed on 4 CPUs, using MPICH2)

My concern is that the total number of electrons exceeds the number of valence electrons that H2O ought to have, which is 8.000.

Any ideas as to why this is the case, and if this is actually a bug in the code?

Camilo

Artifactual charged states when using LAECHG tag?

Posted: Sun Sep 26, 2010 1:07 pm
by admin
you are correct, your system should have exactly 8 valence electrons. Please make suer that you used 1 k-point only for the molecule in the box
concerning the INCAR file, please add
ISMEAR = 0
SIGMA = 0.001
to make sure that the states of the discrete line spectrum of the molecule are sharp.
concerning the sdout file: please make sure that you used a parallel executable