[Pw_forum] Al pseudopotential and elastic properties

[Nicola Marzari]

Dear Joerg,



typically you have to use much higher cutoffs to converge the stress 
matrix, than what is needed to converge the relative shape of the
potential energy surface.

I.e. increasing the cutoff beyond your values would only lower the
potental energy surface, but not really change its shape, while it would
still change the stress matrix.

Of course, it could also be bad luck brought by using mp
smearing, rather than mv smearing. These codes can be quite
sensitive.

:-)

			nicola



On 12/8/11 11:48 AM, Jörg Buchwald wrote:
> Hi,
> does anyone has any experiences with the Al pseudopotentials and the
> elastic constants?
> I was testing Al.pbe-n-van.UPF with which I could reproduce the elastic
> constant and the cohesive energy quite well, but using the output of
> the stress matrix I get a way too high results for C11(>150GPa) at a
> strain of 0.001. Appliying strains about>3%, I get results which are
> quite relastic for C11(90-110 GPa). I thought that this has something
> to do whith the kind of approximation, so I also created a ultrasoft
> (PZ-)LDA potential with the vanderbilt code and tried the
> normconserving Al.pz-vbc.UPF, which gave me similar results.
> I applied the strain by stretching the cell vectors and performing a
> 'relax'-simulation.
> This is my input file for a rel. strain of 0.003:
> ---
>   &control
>      calculation='relax'
>      prefix='Al-test11',
>      pseudo_dir='/home/jbuchw/espresso-4.3.2/pseudo'
>      outdir = '/home/jbuchw/scratch',
>      tstress=.true.
>      disk_io='none'
>      etot_conv_thr=1.0D-10
>      forc_conv_thr=1.0D-10
>   /
>   &system
>      ibrav=  0,
>      celldm(1) =7.6271186
>      nat=  4,
>      ntyp= 1,
>      ecutwfc = 85.0
>      occupations='smearing'
>      smearing='mp'
>      degauss=0.0007
>   /
>   &electrons
>      diagonalization='david'
>      mixing_mode='plain'
>      conv_thr=1.0D-10
>   /
>   &ions
>     bfgs_ndim=3
>   /
>   &cell
>   !  cell_dofree='yz'
>   /
> ATOMIC_SPECIES
>   Al  26.981539  Al.pbe-n-van.UPF
> ATOMIC_POSITIONS crystal
>   Al 0.00 0.00 0.00
>   Al 0.50 0.50 0.00
>   Al 0.00 0.50 0.50
>   Al 0.50 0.00 0.50
> K_POINTS automatic
>   14 14 14 0 0 0
> CELL_PARAMETERS
>     0.9995541042   0.000000000   0.000000000
>     0.000000000   0.996564411   0.000000000
>     0.000000000   0.000000000   0.996564411
> --
> Am I doing something wrong, or does it have something to do with the
> pseudopotential?
> Regards,
> Jörg
>
>
> PS: I also tried to get C11 by the second derivative the Energy fitting
> quadratic function which gave me C11=75.6GPa but this would assume that
> the elastic constant is constant over the strain range(which could be
> showed that's not the case for small strains(<0.01) using the
> numerical second derivative).
>
>
>


-- 

----------------------------------------------------------------------
Prof Nicola Marzari, Chair of Theory and Simulation of Materials, EPFL