[CPMD-list] Questions from a newbie: test on Argon crystal (fwd)

[Juerg Hutter]

Hi

from your input I see the following problems:

1) your using a FCC cell with a latice constant of 5 Angstrom.
   In the newest version you should change the input to
   CELL
      5.0 1.0 1.0 0.0 0.0 0.0
   For this cell, the nearest neighbor distance is 5/SQRT(2).
   Have you taken this into account?
2) You haven't correctly specified the DFT functional. Correct is
   &DFT
     FUNCTIONAL REVPBE
   &END
   So most likely, all calculations were done with the default (LDA),
   which is known to severely overbind.
3) For production runs I would tighten the convergence criteria for
   the wavefunctions.

regards

Juerg

> Dear all,
>
> I run several tests on argon Fcc crystal by different functionals in
> CPMD. pseudo potentials were created from Jurg's parkage. Here are
> calculation results. It doesn't look reasonable. The
> energy is too large.
> As a comparison, I also listed Castep results as comparison. The
> relative energy in table are in kcal/mole. Ar-Ar distance is in Angstrom.
> Anyone can give some clue what is wrong? I also attached an example of the
> input file below. Thanks a lot.
>
> Regards,
>
> Weiqiao
>
> TM pseudo potential
>       CPMD                Castep
> Ar-Ar  BLYP  PBE     BP    GGA(PW91)
> 3.00 302.53 302.62 299.19 16.23
> 3.20 185.26 185.31 182.23  6.13
> 3.40 110.93 110.96 108.12  2.82
> 3.60 64.46  64.47  61.81   0.30
> 3.72 46.17  46.18  43.59  -0.76
> 3.80 36.00  36.01  33.46  -1.27
> 4.00 18.99  18.99  16.53  -1.82
> 4.50  2.36   2.36   0.00  -0.58
> 5.00  0.00   0.00   0.00   0.00
>
> &CPMD
>   OPTIMIZE WAVEFUNCTION
>   TIMESTEP
>    25
>   COMPRESS WRITE32
>   CONVERGENCE
>    5.D-5 1
> &END
> &SYSTEM
>   ANGSTROM
>   POINT GROUP
>     AUTO
>   SYMMETRY
>     2
>   CELL
>      5.0 1.0 1.0 0.5 0.5 0.5
>   CUTOFF
>     70.0
>   SCALE
>    TESR
>      6
>    KPOINTS MONKHORST-PACK fULL
>    4 4 4
> &END
> &ATOMS
> *Ar_revPBE_MT KLEINMAN-BYLANDER
>  LMAX=P
> 1
>    0 0 0
> &END
> &DFT
>   REVPBE
> &END
>
>
>
> _______________________________________________
> CPMD-list mailing list
> CPMD-list at cpmd.org
> http://cpmd.org/mailman/listinfo/cpmd-list
>