about the formation energy calculated by vasp

[utimasa]

Hi, every one,
I have a question want to consult. I'm using vasp to calculate the formation energy of N doped semiconductor, but the total energy difference of neutral states and charged states seems unreasonable. It's too large.
A 3*3*3 supercell were used. One O atom was substituted by N atom.
All the atoms positions were relaxed.
For neutral states, the total energy of the doped system is -648.481449. However, for minus one (-1) states, the total energy of the doped system is -643.658924.

For minus one (-1), I just set the parameter NELECT to be one more electron larger than the neutral states and the ions were relaxed with the lattice parameter unchanged. I noted the fermi level of the neutral system is 4.6823 eV.

In all our calculations ,PBE pseudopotentials and GGA+U were used.



I don?t know why there is so large energy difference, about 4.82eV, between the neutral states and (-1) charged states. I'm puzzled by this question for a long time. If you have any idea , could you give me some advice? I am great grateful for your help.



look forward to your reply. thanks in advance.



PS, INCAR file is as follows:

NWRITE = 2 # verbosity write-tag(how much is written)
PREC = Normal
ISTART = 0 # startjob 0=new 1=cont 2=samecut
ISPIN = 2 # spin polarized calculation (2=yes,1=no)
INIWAV = 1 # initial electr wf 0=lowe 1=rand
ENCUT = 450.0 eV
EDIFF = 1.0E-06
NSW = 0 # number of steps for ionic upd
IBRION = -1 # ionic relaxation 0=MD 1=quasi-New 2=CG
ISIF = 2
ISMEAR = 0 # partial occupancies -5=Blochl(energy) -4=tet -1=fermi 0=gaussian >0 MP(force)
SIGMA = 0.05
LORBIT = 11 # 0 simple, 1 ext, 2 COOP (PROOUT)
# NELECT = 702.0000
#plus U

LDAU=.TRUE.
LDAUTYPE=2
LDAUL= -1 -1 -1 2
LDAUU=0.00 0.00 0.00 5.2
LDAUPRINT=2
LMAXMIX=4

<span class='smallblacktext'>[ Edited ]</span>

[tlchan]

You can calculate the formation energy as a function of unit cell size, and see if your result varies. If yes, there may be interactions between the charged unit cells. You can refer to the VASP manual for details ( here and here ).