ABINIT Discussion Forums

Posted: **Sat Jun 19, 2010 5:04 pm**

I attempted to calculate the hydrogen energy levels, by plotting the bands of a lone hydrogen atom in a large box. I expected the lowest band to lie at -13.6eV but it was actually at around -6eV. The band is flat, as it should be, so I believe it is well converged. Anyone have any suggestions for what I am doing wrong?

I am using the Troullier-Martins LDA psuedopotential and my input file is as follows:

Code: Select all

#########
#hydrogen
#########
chkprim 0     # the unit cell is not primitive so do not check

ndtset 2

##########################################################################
#Spin Info taken from .../tutorial/Input/t15.in.
##########################################################################
nsppol1 2          # Allow a spin-polarized calculation
occopt1 2          # Allow occupation numbers to be set by hand
nband1  1 1        # Number of bands for spin up and spin down
occ1    1.0  0.0   # Occupation numbers for spin up state and spin down state.
                  # The spin up state is occupied, not the spin down state.
spinat1 0.0 0.0 1.0  # Gives an initial estimation of the spin on the atom (easy)

############################
#Definition of the unit cell
############################
acell    3*30

rprim    1.0  0.0  0.0
            0.0  1.0  0.0
            0.0  0.0  1.0

#############################
#Definition of the atom types
#############################
ntypat 1          # There is only one type of atom
znucl 1           # The keyword "znucl" refers to the atomic number of the 
                  # possible type(s) of atom.
                         
########################
#Definition of the atoms
########################
natom 1           # There is one atom
typat 1           # It is hydrogen.

xred
1/2 1/2 1/2

######################################
#Definition of the planewave basis set
######################################
ecut 40.0

###########################################
#Definition of the k-point grid # First Run
###########################################
kptopt 1          # Option for the automatic generation of k points, taking
                  # into account the symmetry
ngkpt  1 1 1

################################
#Definition of the SCF procedure
################################
nstep 30          # Maximal number of SCF cycles
toldfe 1.0d-8     # Will stop when, twice in a row, the difference 
                  # between two consecutive evaluations of total energy 
                  # differ by less than toldfe (in Hartree) 
diemac 1.0        # Hydrogen requires a number between 1 and 3


########################
#Plot the band structure
########################
iscf2   -2
getden2   -1
nband2   10
kptopt2   -1

ndivk2   10
kptbounds2   0.0  0.0  0.0                   # Gamma point
             1.0  1.0  0.0                   # Gamma point in another cell.
tolwfr2  1.0d-12
enunit2  1

Posted: **Sat Jun 19, 2010 10:05 pm**

The major problem is with the exchange-correlation functional ...
None of the LDA or GGAs is self-interaction free and exact for the Hydrogen case.
By contrast, for the hydrogen atom, using -1/r, that is
the so called Fermi-Amaldi correction (ixc=20), is the exact XC functional.

Actually, this example is treated in test/v1 #25 (see tests/v1/Input/t25.in,
as well as the documentation in tests/v1/README ). There, one uses the bare
hydrogenic potential (instead of a pseudopotential), and not sufficiently
large values of acell and ecut (for testing purposes), and you will see
that the 1s-2s splitting is obtained as 0.368 Ha, instead of the exact 0.375 Ha.
Simply increasing the ecut and acell will give a better (closer to the exact value) splitting.
One is able to obtain a good value for the ionisation potential, but the acell
has to be quite large.

So, using ixc=20, wth the base hydrogen nuclei potential, with a high ecut and large acell
will give you the value of 0.5 Ha.

X.

Posted: **Sun Jun 20, 2010 3:56 pm**

Thanks gonze, that makes sense. I have one further question. If the band structure for a lone hydrogen is so far off, am I able to have any faith in the band structure for a hydrogen compound, such as hydrogen on a silicon substrate?

Posted: **Tue Jun 22, 2010 3:58 pm**

Self interaction tends to be a smaller effect in larger systems. H is the extreme case : 50% correction for self interaction energy.

There are, of course, leftovers, and errors which can be treated with other methods (GW, hybrids, SIC...) in particular the band gap discontinuity.

matthieu

ABINIT Discussion Forums

hydrogen energy levels

hydrogen energy levels

Re: hydrogen energy levels

Re: hydrogen energy levels

Re: hydrogen energy levels