[SIESTA-L] Pseudopotential and basis set for Ag

Tue, 20 Jun 2017 13:03:50 -0700 

Dear SIESTA users,
I've used pseudopotentials (PP) and basis sets (BS) from an article published in 2015 (Systematic pseudopotentials from reference eigenvalue sets for DFT calculations, url : http://www.sciencedirect.com/science/article/pii/S0927025614007940#f0005 ) 


In table 1, one finds parameters (radius cut-off values and core  
corrections) to be passed on to atom, to create norm-conserving  
troullier-martins PPs.
In table 2 there are cut-off parameters for BSs that were found while  
optimizing the PP/BS choice.



In my calculations I am using Ag, with a PBE functional. Here are the  
two rows pertaining to Ag, from table 1 and table 2:

PP : Ag (47)    5s14d10 2.19(5s), 2.83(5p), 2.40(4d), 2.82(4f), 0.010, 0.000
BS : Ag (47)    5.777, 4.704(5s), 3.637, 3.137(4d), 6.514(5p)

Here is my PAO.Basis block:
%block PAO.Basis
Ag    4
n=5    0    2
  5.777   4.704
n=5    1    1
  6.514
n=4    2    2
  3.637   3.137
n=4    3    1
  2.79
%endblock PAO.Basis


I have three questions :

1. Relating to the PP : My input contains the parameters indicated by  
the article, but the output indicates slightly different parameters  
(2.17 instead of 2.19, 2.82 instead of 2.83, 2.40, 2.79 instead of  
2.82). I suppose that this is normal, and shouldn't affect my  
calculations. Am I correct?



2. Relating to the BS: The article doesn't provide any cut-off radius  
for the 4f orbital, and siesta complains if I don't provide it (since  
the PP is generated with the 4f information). I didn't know what to  
put in, so I set it to 2.79 arbitrarily. I suppose that I should vary  
this parameter and verify that I get decent results for an Ag crystal.  
However, the resulting lattice parameter after a CG MD run is quite  
close to the one provided in the article : 4.049 Ang instead of 4.014  
Ang (article value). Furthermore, the band structure I obtain is the  
same as in the article. Do you think I should spend time and  
ressources trying to fix this discrepancy?



3. I will be studying Ag nanoparticles, and have read an article  
(Optimal strictly localized basis sets for noble metal surfaces, url:  
https://journals.aps.org/prb/pdf/10.1103/PhysRevB.79.075441)  
discussing the use of fictitious diffuse s orbitals on surface atoms  
(which comprise most of the nanoparticle in small cases). How does one  
include such an orbital? Do I simply add the lines

    ' n=6 0 1
        4.76 '
to my Basis block?


I realize that these questions may be a bit vague, but I would  
appreciate any feedback, comments, or hints/tips.


Thanks a bunch,
Nicolas Duchêne
Masters Student in Applied Science, Polytechnique Montréal.
























					Responder a