My Community

Does it make sense to perform LDA+U/GGA+U calculation on a free, isolated transition metal atom. If I do GGA+U on Ni atom, with the INCAR below, I get positive total energy. What am I doing wrong? (The energy without entropy is also positive)

Thanks,
Zsolt

The INCAR:
System = Ni atom
ISTART = 1
ICHARG = 1
ENCUT = 560
PREC = High

ISPIN = 2
#MAGMOM = 2

ISMEAR = 0
SIGMA = 0.05

#FERWE = 3*1 5*1 1 7*0
#FERDO = 3*1 3*1 2*0 1 7*0

EDIFF = 0.0001
NELM = 440
#AMIX = 0.2 # mixing set manually
#BMIX = 0.0001


LDAU=.True.
LDAUTYPE = 2
LDAUPRINT = 1
LDAUL = 2
LDAUU = 5.1
LDAUJ = 0.0

#LREAL = Auto

LPLANE = .True.
NPAR = 16 ! number of nodes.
NSIM = 4

LSCALU = .FALSE.

LMAXMIX = 4
MAXMIX = 40

ISYM = 0

Hi zsoltrak,

I'd say something went terribly wrong in your calculation. It should become negative. Have you checked, if the occupation is all right?

Cheers,

alex

Thank you for your comment Alex, however, my question is still unanswered: does it make sense to carry out DFT+U calculation on free atoms? Yes, I checked the occupations; in fact I tried numerous fixed occupations and also continuation of the calculation keeping the GGA occupations fixed. Any suggestions...?

Thanks,
Zsolt

Hi Alex and Zsolt,


The positive energy with DFT+U should be fine. Negative energies are typically obtained with the functional via which the pseduopotential was generated. Any modification of this functional somehow such as adding U or superimposing a different one (e.g. AM05 on top of PBE pseudo) will certainly change the absolute total energy of the free atom.

Anyways absolute total energies energies are meaningless.


Mostafa