The ABINIT Users Mailing List ( CLOSED )

["Anglade Pierre-Matthieu" <anglade@gmail.com>]

Hi,

It is very usual to find a slight difference between DFT and experimental lattice parameter and most people accomodate with that. A 2% error as you have is rather large though not uncommon. Do you have special reasons to require a perfect matching between your DFT and exp. lattice parameter ? 

There is many  sources of error with pseudopotentials (assuming your calculation settings are OK). The question you must elucidate is which one is responsible for your problem:
 First an intrinsic one: The pseudopotential may not reproduce accurately the full electron computation on which it is based. In this case you can claim the pseudopotential is not good. It is not impossible that some of Abinit pseudopotentials are not accurate.
Second: the pseudopotential approximation is not accurate for your system and core e- relaxation must be taken into account.
third: Some effects have not been included in the pseudopotential; for instance relativistic effects or spin effects. In this case you must search for pseudopotentials including all those effects.
fourth : the exchange correlation function used to generate the pseudopotential is not accurate enough in your case. 

In order to check what is the source of the discrepancy you can
1) try various pseudopotentials (you already did)
2) compare results with full e- computation using the same XC functional
3) Cheat and use other XC functional than the one originally used for your pseudo
4) After identification of the error sources, you will want to develop a new pseudopotential to fit your requirements. Opium (
http://opium.sourceforge.net/) is a very handy program. You can also use the (probably) more versatile FHI98PP (http://www.fhi-berlin.mpg.de/th/fhi98md/fhi98PP/
). Doing so you will be able to include all important features you miss in avaiable pseudopotentials.


Regards

PMA

On Nov 19, 2007 3:03 PM, Salima Saib < saibsalima@yahoo.fr> wrote:

hello
ScN crystallises in nacl structure, the file input is
included the problem that the pseudopotentials
included in abinit page give large lattice parameter
for the unit cell , the experimental one is 4.5
angstrom
the smallest parameter i got it using the
pseudopotentials of the abinit page, 50sn_pspnc for sn
and 7n_pspnc,and it 4.59 angstrom, and it is so large
the other pseudopotentials of the abinit page give
larger than this value
I ask if any body was working with this Sn
pseudopotential and find a better pseudopotential for
sn,please if is it possible to send me the file.
best regards
salima

optcell 1
ionmov  3
ntime  10
dilatmx 1.1
ecutsm  0.5

kptopt 1          # Option for the automatic
generation of k points, taking
                 # into account the symmetry
nshiftk 4
shiftk  0.5 0.5 0.5
       0.5 0.0 0.0
       0.0 0.5 0.0
       0.0 0.0 0.5
ngkpt  8 8 8          #

 getwfk -1
acell 3*8.5066
rprim  0.0  0.5  0.5
      0.5  0.0  0.5
      0.5  0.5  0.0

ntypat 2       # There is only one type of atom
znucl 50 7
natom 2           #
typat 1 2
xred
  0.0  0.0  0.0  #
  1/2  1/2  1/2  #
ecut 80
nstep 20          # Maximal number of SCF cycles
toldfe 1.0d-6
diemac 14.0

     _____________________________________________________________________________
Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail

-- 
Pierre-Matthieu Anglade