[Wien] EFG: theory Vs experiment for a case

[Peter Blaha]

Most important is the minimization of the forces !!!!

When Hf substitutes Ti, the O-atoms around it will slightly relax; but 
in particular when you simulate a O-vacancy, unrelaxed calculations are 
meaningless.   The EFG is usually EXTREMELY sensitive to the exact 
positions of atoms. (and I don't understand your argument about "basic 
lattice was already optimized ?? In cubic SrTiO3 there is NOTHING to 
optimize.)

I definitely would go to a "cubic supercell". Use first a 2x2x2 cell, 
and once you have results, you check your supercell size with 3x3x3.

Forget "correlation", at least at the moment.

You may reduce RMT of Sr, but don't reduce Ti (Hf) RMTs. Otherwise you 
would have to increase RKMAX ....

You can reduce the cpu-time a lot by doing the relaxation of forces 
first with RKmax 5.5 (or 6) and only VERY few k-points, and only once it 
is relaxed, increase RKmax and k-mesh, continue the relaxation (if 
necessary - when forces are bigger again).

As was mentioned before, you can already see after   x symmetry  if the 
asymmetry parameter eta is zero (only a "2 0" term in the LM list) or 
can be non-zero (2 2 and/or 2 1 terms).

I'd expect that for eta=0.5 you would need a "complicated" defect (low 
symmetry). As mentioned before, this defect would need to be 
energetically favorable, as compared to others.

I don't know if one can trust eta from the PAC results ?


>                             Calculation: STO structure is optimized and 
> minimized. Created supercell 3x2x2 as it allows me to use 5 processor 
> with 8 gb ram (with 3x3x3, calc. restricted to only 2 processors due to 
> RAM limitation. So long scf run time. i will try now) . Problem of 
> breaking symmetry did not strike me at the time of calculation. Replaced 
> one Ti by Hf. Then supercell calculation was done for oxygen vacancies 
> in first cordination and fourth nearest neighbour cordination. The efg 
> matched for the later case (oxygen vacancy at fourth nearest neighbour) 
> but not assymetry parameter. Minimization of forces on supercell was not 
> done ( i thought basic lattice unit was already optimized).
>           What i understood from previous replies of this thread is 
> 1.Must do 3x3x3 supercell calculation, 2. Electron electron correlation 
> can also be included for improvement of calculation, 3. Need to check 
> calculation with reduced RMT of Sr and Ti.
> 
> To what % agreemnet wrt experimental efg data, efg values from 
> theoretical calculation can be accepted.
> Thanks Dr. Cottenier for the suggestion, i had already subscribed to HF 
> course A (ID: iak). I will subscribe to Course B after finishing course A.
> 
> Thanks,
> A. Kumar
> 
> 
> On Wed, Dec 19, 2018 at 12:08 AM Ashwani Kumar <ashwani.ism at gmail.com 
> <mailto:ashwani.ism at gmail.com>> wrote:
> 
>     hi,
>         thanks for reply. the assymetry parameter, (Vxx-Vyy)/Vzz,  is
>     zero (wien2k calculation) whereas i got 0.52 from TDPAC (Time
>     dependent perturbed angular correlation) spectroscopy for a SrTiO3
>     (STO) defect structure. EFG component is -1.63 x10^21 V/m2 (wien2k,
>     lapw2 -efg) and i obtained 1.69 x 10^21 V/m2 (calculated from TDPAC
>     results). STO has cubic lattice so no efg and no assymetry 
>     parameter (for both wien2k and TDPAC) but defect STO structure
>     showed very pure hyperfine interactions with assymetry parameter :
>     0.52. So i am not having confidence over my wien2k calculation because :
>     1. i am getting assymetry parameter =0
>     2. Negative EFG which i understood from previous answers that
>     negative sign indicates the rate of decrease of z-component of EF
>     wrt to distance.
> 
>     Am i missing something?,
>     Calculation parameters are :Supercell (3x2x2) STO with doped Hf
>     atom, PBE, KGEN:200, rkmax: 7.0,
> 
>     thanks,
>     A. Kumar
> 
>     On Fri, Dec 14, 2018 at 11:02 PM Ashwani Kumar
>     <ashwani.ism at gmail.com <mailto:ashwani.ism at gmail.com>> wrote:
> 
>         Hi,
>            i have calculated EFG  the defect structure of crystalline
>         system from experimental data from PAC spectroscopy. Then using
>         WIEN2K (crystal structure--> supercell--> defect introduced),
>         EFG is calculated.
>         Exper. Calculated : 1.69 x 10^21 V/m2 whereas wien2k
>         calculation: -1.66 x 10^21 V/m2 (crude value still have to do
>         lapw2 -efg) on probe atom. Wien2k calculation shows negative
>         value. is there any significance of the negative sign.
> 
>         thanks,
>         A. Kumar
> 
> 
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at:  http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
> 

-- 

                                       P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-165300             FAX: +43-1-58801-165982
Email: blaha at theochem.tuwien.ac.at    WIEN2k: http://www.wien2k.at
WWW:   http://www.imc.tuwien.ac.at/TC_Blaha
--------------------------------------------------------------------------