My Community

Dear Vaspusers

I did a calculation of single Pt atom in a large cell (box size 15 A,
1-kt, PAW91, GGA91) with LORBMOM and LORBIT. LSORBIT tag
turn on in the INCAR file.

However, the values of orbital moments and spin moments
I get are not consistent with empirical Hund's rule.
The Pt PP file has valence configuration 5d96s1 ie PP's are
already generated in the atomic ground state configuration and
not in bulk configuration. So I should get

spin moment on Pt atom = 2 bohr magnetons as there are two
unparied electrons each spin 1/2 and electron g factor is 2.

orbital moment = 2 bohr magneton as L=2 for d-orbital.

since the LNONCOLLINEAR tag is automatically on, I get
three components of orbital moments and three components
of spin moments printed in the OUTCAR file.

when I square each component, add it up and then take the
square root I get

spin-moment on Pt sphere : 0.989 bohr magneton
orbital moment on Pt sphere: 1.05 bohr magneton
spin-moment on the cell : 1.61 bohr magneton

I expected the spin-moment on the cell to be exactly 2 as per
Hund's rule if
we take into account the arbitrariness in first two values associated with the sphere size.

The sign seems to be correct for more than half filled atomic
shell in this case both L and S points in the same direction when I look into the sign of each orbital and spin components.

But the spin-moment magnitude is wrong.

My INCAR file:

ISPIN=2
PREC=high
NELMIN=5
NELM=70
EDIFF=1E-05
LSORBIT=.TRUE.
LORBIT= 11
LORBMOM=.TRUE.
ISMEAR =0
SIGMA =0.001
LREAL=F
GGA=91
VOSKOWN=1


Since PP is already generated in the ground state, Do I have
to still follow the suggested receipe for the free atom
ie either NUPDOWN = number of unpaired electrons in the atom
or use FERWE and FERDO and occupy the states according to
the Hund's rule,
ISMEAR=-2 to keep the occupancies fixed
SIGMA= very small say 0.00001
LDIAG= .FALSE. to keep the ordering of the eigenstates fixed.

to do the free atom calculation and to get the correct spin-moment on the cell.

How to get the correct orbital moment on the cell.

sahu

[ Edited Wed Jul 13 2005, 11:17AM ]

Hund's rule strictly applies only for free atoms
in spherical symmetry, which exactly defined degeneracies
and l,m character of the levels.
in ANY bulk calculation (including calculations of free atoms within a bulk-like box-shaped unit cell),
this symmetry is broken by the
--choice of a unit cell and furthermore,
--GGA tends to break it in addition. (this is a general feature of GGA)
(have a look at the levels in your OUTCAR file!)
This is not a feature of VASP, it is common to all DFT solid state
methods.

Hi

thanx for the reply.

However, my question is not yet answered.

The PAW PP's for Pt used for this calculation is already
in the atomic ground state valence configuration.

Then the
question is whether the prescribed procedure for calculating free atom in a large box ie

ISMEAR=-2 to keep the occupancies fixed
SIGMA= very small say 0.00001 eV
LDIAG= .FALSE. to keep the ordering of the eigenstates fixed

etc.

are still needed for the free atom calculation in a large
box to get close to atomic moments(spin & orbital)
and close to atomic energies.

sahu

Hi

If you have any suggestion to my query above, pl.let me
know.

sahu

the calculation of free atoms with VASP is discussed in the
online manual and in the VASP workshop file handsonI.pdf and HandsonI.tgz
http://cms.mpi.univie.ac.at/vasp/vasp/node198.html
http://cms.mpi.univie.ac.at/vasp-workshop

If you do a free atom calculation, first have a look at the potential of the
atom, the eletronic configuration can be seen from POTCAR and from the V_RHFIN files. the POTCARs are generated for the electronic BULK ground state for all atoms
1) If the electronic configuration corresponds to the electronic configuration of the free atom (level- and spin occupancies), you just have to
-- reduce SIGMA to avoid smearing
-- set ISPIN = 2 (if there are unpaired electrons)
(this would be the case for Pt, the VASP configuration is 6s1, 5d9, like
the atomic ground state of Pt)

2) usually, d metals have s1 d(x+1) configuration in bulk as opposed to the
the s2 d(x) configuration of free atoms in the gas phase.
in that case, futher tags have to be set in INCAR,

ISPIN = 2
FERWE = (set occupancies for spin up and down states according to
FERDO = Hund's rule, NBANDS numbers have to be given each)
ISMEAR = -2 (keep occupancies fixed)
LDIAG = .False.
SIGMA = 0.001

for GGA potentials, choose a box which is slightly non-cubic e.g. 10.0, 10.25, 10.5