My dearest Marcos and Andrei

You are both right, as usual. But there is more than it seems.

There used to be an issue with the file periodic_table.f in SIESTA.
When we were producing http://arxiv.org/abs/cond-mat/0611624v1 and
http://arxiv.org/abs/cond-mat/0610879v2 ,  we realized that even a
single atom of Pt wasn't getting magnetized, no matter how big the
DM.InitSpin you put, and actually SIESTA gave a warning like
     .        'initdm: WARNING: atom ', atom(iat),
     .        ' has a closed-shell and cannot be polarized'
Therefore, SIESTA ignored the .psf electronic config and used the one
in periodic_table.f . So we needed to tweak the config for Pt and Pd.

May one of you, or the gentleman who launched the question, try with a
finite value of DM.InitSpin, and see if still gets this warning? That
would be a warning to modify their periodic_table.f . A better
solution in the long term would be to get the info from the psf, since
it may vary (for instance, diferent atomic configurations).

Regards

Lu

On Fri, Dec 2, 2011 at 07:22,  <[email protected]> wrote:
> Thanks Marcos,
> you are right.
> Then, it must be the dimer issue which kills the magnetism?
> It would make sense to stop after some few iterations
> and have a look at the magnetic moments
> (and at the structure, e.g. the XV file, as well)
>
> Best regards
>
> Andrei
>
>
>> Hi Andrei,
>>
>> Just one comment. If I'm not mistaken, the default spin polarization for
>> Siesta, when you don't specify anything, is that one atom is initialized
>> with spin up and the other down, in this case - an antiferro
>> configuration.
>> The final spin of the system could be non zero in this case, I suppose.
>>
>> Furthermore, I was checking something on Pt and Pd some days ago on the
>> siesta mailing list, to see what people had done, and it seems that Pt is
>> a
>> bordeline case for magnetism, when it comes to wires. Results would differ
>> with the use of GGA and LDA - check work by Anna Delin and Erio Tosatti on
>> PRL, as well as the associated comment by Simone Alexandre and Jose Soler,
>> and the reply to the comment by the authors of the paper.
>>
>> I just gave it a bird's eye look on the subject, and I recommend a more
>> careful look at the siesta mailing list and the corresponding papers.
>>
>> Best regards,
>>
>> Marcos
>>
>> On Fri, Dec 2, 2011 at 8:44 AM, <[email protected]> wrote:
>>
>>> Dear isivkov,
>>>
>>> You use
>>> SpinPolarized             T    #default
>>> but I don't see if you ever made spin up different
>>> from spin down (by InitSpin or otherwise).
>>> By default, they start equal and remain equal.
>>>
>>> Moreover:
>>> I don't see that you defined
>>> AtomicCoordinatesFormat
>>> so it must be Bohr by default? -
>>> then you probably have Pt2 dimers at 1 Bohr distance,
>>> separated by 20 Ang.
>>>
>>> In addition, i seems weird
>>> to construct an empty box of 120 Ang size
>>> (in X and Y dimensions), but this has nothing to do
>>> with the magnetism issue.
>>>
>>> Best regards
>>>
>>> Andrei Postnikov
>>>
>>>
>>> > Hello all.
>>> >
>>> > I decided to calculate a single atom of platinum, Not exactly single,
>>> but
>>> > chain of platinum with distance between atoms 10 A. I think it is like
>>> a
>>> > single atom. My .fdf file is here below. It is very strange, that
>>> platinum
>>> > is non-magnetic (as you can see from output file below), wile this is
>>> the
>>> > single atom. Atom of platinum must have magnetic moment.
>>> >
>>> > My input file
>>> > ==================================================================
>>> > #
>>> >
>>> -----------------------------------------------------------------------------
>>> > # FDF for Pt bulk
>>> > #
>>> > # LDA
>>> > # Scalar-relativistic pseudopotential with non-linear partial-core
>>> > correction
>>> > #
>>> > #
>>> >
>>> -----------------------------------------------------------------------------
>>> >
>>> > #### Cu bulk #######
>>> >
>>> > SystemName      PtLinChain
>>> > SystemLabel     PtLinChain           # Short name for naming files
>>> >
>>> > # Output options
>>> >
>>> > WriteCoorStep          true
>>> >
>>> > WriteMullikenPop       1
>>> >
>>> > # Species and atoms
>>> >
>>> > NumberOfSpecies        1
>>> > NumberOfAtoms          2
>>> >
>>> > %block ChemicalSpeciesLabel
>>> >   1  78  Pt.LDA
>>> > %endblock ChemicalSpeciesLabel
>>> >
>>> >
>>> >
>>> > %block PAO.Basis
>>> >    Pt.LDA 2 split 0.00  # Species label, number of l-shells
>>> >    n=6 0 2 P 1      # n, l, Nzeta, Polarization, NzetaPol
>>> >    0.00 0.00        # 0.0 => default [6.982  5.935 \n 1.000  1.000]
>>> >    n=5 2 2          # n, l, zeta
>>> >    0.00 0.00
>>> > %endblock PAO.Basis
>>> >
>>> >
>>> >
>>> > LatticeConstant      10  Ang
>>> >
>>> > %block LatticeVectors
>>> >   12.000  0.000  0.000
>>> >   0.000  12.000  0.000
>>> >   0.000  0.000   2.000
>>> > %endblock LatticeVectors
>>> >
>>> > %block kgrid_Monkhorst_Pack
>>> >   1  0  0  0
>>> >   0  1  0  0
>>> >   0  0  8  0
>>> > %endblock kgrid_Monkhorst_Pack
>>> >
>>> >
>>> > XC.functional         LDA           # Exchange-correlation functional
>>> > XC.authors            CA            # Exchange-correlation version
>>> >
>>> > MeshCutoff           150 Ry        # Mesh cutoff. real space mesh
>>> >
>>> > # SCF options
>>> > MaxSCFIterations      200           # Maximum number of SCF iter
>>> > DM.MixingWeight       0.02           # New DM amount for next SCF
>>> cycle
>>> > DM.Tolerance          1.d-4         # Tolerance in maximum difference
>>> >                                     # between input and output DM
>>> > DM.UseSaveDM          true          # to use continuation files
>>> > DM.NumberPulay         5
>>> >
>>> > Diag.DivideAndConquer  .false.
>>> > SolutionMethod         diagon       # OrderN or Diagon
>>> > ElectronicTemperature  25 meV       # Temp. for Fermi smearing
>>> >
>>> > SpinPolarized             T    #default
>>> >
>>> > # MD options
>>> > #MD.TypeOfRun           cg           # Type of dynamics:
>>> >
>>> > #MD.UseSaveCG         .true.          # for restarting
>>> > #MD.UseSaveXV         F               # atomic coords
>>> >
>>> > #MD.NumCGsteps          0            # Number of CG steps for
>>> >                                     #   coordinate optimization
>>> > #MD.MaxCGDispl          0.05 Ang      # Maximum atomic displacement
>>> >                                     #   in one CG step (Bohr)
>>> > #MD.MaxForceTol         0.005 eV/Ang  # Tolerance in the maximum
>>> >                                     #   atomic force (Ry/Bohr)
>>> >
>>> > # Atomic coordinates
>>> > AtomicCoordinatesFormat ScaledCartesian
>>> >
>>> > # %block Zmatrix
>>> > #cartesian
>>> > #1 0.0000  0.0000  0.0000  1 1 1
>>> > #1 0.3535  0.3535  0.5000  1 1 1
>>> > #1 0.0000  0.0000  1.0000  1 1 1
>>> > #1 0.3535  0.3535  1.5000  1 1 1
>>> > #1 0.0000  0.0000  2.0000  1 1 1
>>> > #1 0.3535  0.3535  2.5000  1 1 1
>>> > #1 0.0000  0.0000  3.0000  0 0 0
>>> > #1 0.3535  0.3535  3.5000  0 0 0
>>> > #1 0.0000  0.0000  4.0000  0 0 0
>>> > #1 0.3535  0.3535  4.5000  0 0 0
>>> > # %endblock Zmatrix
>>> >
>>> > %block AtomicCoordinatesAndAtomicSpecies
>>> > 0.0000 0.0000 0.000 1
>>> > 0.0000 0.0000 1.000 1
>>> > %endblock AtomicCoordinatesAndAtomicSpecies
>>> >
>>> > #%block GeometryConstraints
>>> > #position from 1 to 4
>>> > #%endblock GeometryConstraints
>>> > =================================================================
>>> >
>>> >
>>> > My output file(parts):
>>> > At first SIESTA takes valence configuration from .psf file,
>>> > as was in .inp file.
>>> > ========================================================
>>> > Reading pseudopotential information in formatted form from Pt.LDA.psf
>>> >
>>> > Pseudopotential generated from an atomic spin-polarized calculation
>>> >
>>> > Valence configuration for pseudopotential generation:
>>> > 6s(1.00,0.00) rc: 2.32
>>> > 6p(0.00,0.00) rc: 2.47
>>> > 5d(5.00,4.00) rc: 1.23
>>> > 5f(0.00,0.00) rc: 2.32
>>> > For Pt.LDA, standard SIESTA heuristics set lmxkb to 3
>>> >  (one more than the basis l, including polarization orbitals).
>>> > Use PS.lmax or PS.KBprojectors blocks to override.
>>> >  Warning: Empty PAO shell. l =           1
>>> >  Will have a KB projector anyway...
>>> > ==========================================================
>>> >
>>> > than SIESTA writes as though it takes another configuration
>>> > with 10 electrons on d-state
>>> >
>>> > ===========================================================
>>> > atm_pop: Valence configuration (for local Pseudopot. screening):
>>> >  6s( 0.00)
>>> >  6p( 0.00)
>>> >  5d(10.00)
>>> > Vna: chval, zval:   10.00000  10.00000
>>> >
>>> > Vna:  Cut-off radius for the neutral-atom potential:   4.341778
>>> > comcore: Pseudo-core radius Rcore=  4.234578
>>> > ============================================================
>>> >
>>> > and than that shows absence of magnetic moment after calculation
>>> >
>>> > ===================================================================
>>> > Species: Pt.LDA
>>> > Atom  Qatom  Qorb
>>> >                6s      6s      5dxy    5dyz    5dz2    5dxz    5dx2-y2
>>> > 5dxy
>>> >                5dyz    5dz2    5dxz    5dx2-y2 6Ppy    6Ppz    6Ppx
>>> >    1  5.000   0.062  -0.006   1.129   1.129   1.130   1.129   1.130
>>> > -0.141
>>> >              -0.141  -0.141  -0.141  -0.141   0.000   0.000   0.000
>>> >    2  5.000   0.062  -0.006   1.129   1.129   1.130   1.129   1.130
>>> > -0.141
>>> >              -0.141  -0.141  -0.141  -0.141   0.000   0.000   0.000
>>> >
>>> > mulliken: Qtot =       10.000
>>> >
>>> > mulliken: Spin DOWN
>>> >
>>> > Species: Pt.LDA
>>> > Atom  Qatom  Qorb
>>> >                6s      6s      5dxy    5dyz    5dz2    5dxz    5dx2-y2
>>> > 5dxy
>>> >                5dyz    5dz2    5dxz    5dx2-y2 6Ppy    6Ppz    6Ppx
>>> >    1  5.000   0.062  -0.006   1.129   1.129   1.130   1.129   1.130
>>> > -0.141
>>> >              -0.141  -0.141  -0.141  -0.141   0.000   0.000   0.000
>>> >    2  5.000   0.062  -0.006   1.129   1.129   1.130   1.129   1.130
>>> > -0.141
>>> >              -0.141  -0.141  -0.141  -0.141   0.000   0.000   0.000
>>> >
>>> > mulliken: Qtot =       10.000
>>> > =========================================================================
>>> >
>>> >
>>> >
>>>
>>>
>>
>



-- 
Lucas Fernández Seivane
UTSA - Department of Physics and Astronomy
+1 210 458 8052
(Formerly Universidad de Oviedo - CINN
http://sites.google.com/site/quevedin/
http://www.researcherid.com/rid/A-8610-2008

Responder a