[Pw_forum] DFT+U: different oxidation states on the same element (charge ordering)

Mehmet Topsakal mtopsaka at umn.edu
Mon Jun 9 12:22:45 CEST 2014


I think the problem might be related with iron ultrasoft pseudopotential.

As in page 40 of this pdf :

http://www.vasp.at/vasp-workshop/slides/pseudopp2.pdf

Low-spin case (NM) is okay. But for high-spin iron, energies are too
different.

VASP use PAW potential. So you should switch to PAW with QE.

Hope this helps.
Dear Matteo,
thanks a lot for the reply.
I tried also using occupations="smearing" but the situation is the same. I
get a difference of 0.3 electrons between the iron atom Fe1 and the other
three Fe2.
Is there anything else that I can try? Could this behaviour due to the
differences between VASP (that you used in the refererence article) and
Quantum Espresso?
Best regards,
Mauro.



2014-06-06 16:55 GMT+02:00 Matteo Cococcioni <matteo at umn.edu>:

>
> Dear Mauro,
>
> one thing I notice is that you are using occupation fixed. is there a
> reason for that? this, I think, might actually cause the result you are not
> happy about.
>
> Matteo
>
>
> On Fri, Jun 6, 2014 at 4:39 PM, Mauro Sgroi <
> maurofrancesco.sgroi at gmail.com> wrote:
>
>> Dear all,
>> I'm trying to study the phase separation in LixFePO4 using DFT+U.
>> The reference article that I'm using is PRB 69, 201101 (R) 2004, "Phase
>> separation in LixFePO4 induced by correlation effects, Zhou et al.
>>
>> The authors compare, using VASP, standard DFT with DFT+U and obtain the
>> correct description of the material with the latter approach. In particular
>> they found that is possible to describe the formation of two type of ion,
>> Fe2+ and Fe3+, according to to the different position in the unit cell in
>> the partially lithiated structures. With GGA all Fe ions have the same
>> occupancies, regardless the position in the cell and the fact that they are
>> more or less near a lithium ion.
>> To obtain this result with DFT+U they had to force the breaking of the
>> symmetry of the structure and, I imagine, they set the initial occupations
>> in some way to force the desired electronic configuration.
>>
>> I'm trying to do the same, differentiating the Fe atoms, breaking the
>> symmetry and using starting_ns_eigenvalue to force the initial occupations.
>> My problem is that at the end of the calculation I got always the same
>> occupations on the 4 Fe atoms in the structure (more or less, the
>> difference could be 0.3 electrons).
>>
>> I'm using lda_plus_u_kind=0, but also using the other approach was not
>> beneficial.
>>
>> My input file is below.
>> Have you any suggestion?
>>
>> Thanks a lot in advance and best regards,
>> Mauro Sgroi.
>>
>> &control
>>     calculation='scf',
>>     pseudo_dir = "/usr2/sgroi/DATABASE/ESPRESSO",
>>     prefix='lifepo4_scfU',
>>     outdir='/home/dati/espresso/temp',
>>  /
>>  &system
>>     ibrav=8, celldm(1)=19.748515, celldm(2)=0.586193, celldm(3)=0.441599,
>> nat=27, ntyp=5, nosym=.true.
>>     ecutwfc=40, ecutrho=400, nbnd=130, occupations='fixed',  nspin=2,
>> tot_magnetization=17,
>>     lda_plus_u=.true., Hubbard_U(2)=4., Hubbard_U(3)=4.475,
>>     starting_ns_eigenvalue(5,2,2)=0,
>>     starting_ns_eigenvalue(4,2,2)=0,
>>     starting_ns_eigenvalue(3,2,2)=0,
>>     starting_ns_eigenvalue(2,2,2)=0,
>>     starting_ns_eigenvalue(1,2,2)=0,
>>     starting_ns_eigenvalue(5,2,3)=1,
>>     starting_ns_eigenvalue(4,2,3)=0,
>>     starting_ns_eigenvalue(3,2,3)=0,
>>     starting_ns_eigenvalue(2,2,3)=0,
>>     starting_ns_eigenvalue(1,2,3)=0
>>  /
>>  &electrons
>>      conv_thr=1.0d-8, electron_maxstep=100, mixing_beta=0.5
>>  /
>>
>> ATOMIC_SPECIES
>>  Li 6.941   Li.pbe-s-van_ak.UPF
>>  Fe1 55.847  Fe.pbe-sp-van_ak.UPF
>>  Fe2 55.847  Fe.pbe-sp-van_ak.UPF
>>  P  15.9994 P.pbe-van_ak.UPF
>>  O  30.9737 O.pbe-van_ak.UPF
>>
>> ATOMIC_POSITIONS {crystal}
>> Fe1      0.281537138   0.250000000   0.981851206    1   0   1
>> P        0.103893879   0.250000000   0.415586970    1   0   1
>> O        0.119764467   0.250000000   0.744164738    1   0   1
>> O        0.447356144   0.250000000   0.201257756    1   0   1
>> O        0.187375840   0.060350405   0.283173864
>> Li       0.500000000   0.000000000   0.500000000    0   0   0
>> Fe2      0.211199573   0.750000000   0.474632077    1   0   1
>> P        0.405092969   0.750000000   0.922566801    1   0   1
>> O        0.401164833   0.750000000   0.251505445    1   0   1
>> O        0.041727760   0.750000000   0.686207216    1   0   1
>> O        0.336134397   0.956851552   0.790300073
>> Li       0.000000000   0.500000000   0.000000000    0   0   0
>> Fe2      0.719627852   0.750000000   0.020422522    1   0   1
>> P        0.902369089   0.750000000   0.575773635    1   0   1
>> O        0.896492242   0.750000000   0.246982466    1   0   1
>> O        0.540334658   0.750000000   0.790913565    1   0   1
>> O        0.840097513   0.539754373   0.720678002
>> Li       0.500000000   0.500000000   0.500000000    0   0   0
>> Fe2      0.786894270   0.250000000   0.525238732    1   0   1
>> P        0.588956858   0.250000000   0.078832190    1   0   1
>> O        0.587126133   0.250000000   0.749458661    1   0   1
>> O        0.966454798   0.250000000   0.315230191    1   0   1
>> O        0.654364669   0.453679217   0.216461848
>> O        0.822813149   0.944845411   0.700568536
>> O        0.653714964   0.047069624   0.216100948
>> O        0.172021112   0.453011228   0.263457462
>> O        0.335493673   0.543326856   0.788919121
>>
>> K_POINTS {automatic}
>> 2 4 8 1 1 1
>>
>>
>> _______________________________________________
>> Pw_forum mailing list
>> Pw_forum at pwscf.org
>> http://pwscf.org/mailman/listinfo/pw_forum
>>
>
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
>


_______________________________________________
Pw_forum mailing list
Pw_forum at pwscf.org
http://pwscf.org/mailman/listinfo/pw_forum
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20140609/b9f97ecd/attachment.html>


More information about the users mailing list