[QE-users] convergence issue in DFT_U+V calculation
Dr. K. C. Bhamu
kcbhamu85 at gmail.com
Wed Dec 1 11:33:20 CET 2021
Dear Dr. Matteo,
Thank you very much for the quick response.
Yes, I am doing vc-relax in between each step. Doing vc-relax without
adopting geometry information from the previous run. I am only supplying
the HP file containing U+V obtained from the previous run for the vc-relax
for the next step.
Fe1 (octahedral coordination) atom has a mixed charge state (Fe3+ and Fe2+).
The magnetic moment of Fe1( Fe3+) cancels out with Fe2 (tetrahedral
coordination; all Fe3+) and the resultant magnetic moment comes from Fe1
(2+) atoms.
According to Ref. [1]
<https://www.sciencedirect.com/science/article/abs/pii/S0039602812000544>,
I set all Fe1 as up and all Fe2 as down spins.
Do you think my input still has some issues?
-------------------
I am providing some additional information here.
Additional information:
Ref. [1]
<https://www.sciencedirect.com/science/article/abs/pii/S0039602812000544>
Just below Fig. 3 it is mentioned that "For Fe3O4 having ferrimagnetic
property, the spin of the octahedral and tetrahedral Fe was set as up and
down, respectively."
Relevant para from Ref. [2]
<https://link.springer.com/article/10.1134%2FS1027451010030079> is as below:
Fe2+ cations in the face-centered cubic lattice formed by oxygen anions
occupy octahedral sites, while Fe3+ cations are distributed between
octahedral and tetrahedral sites. The trivalent Fe3+ ion is in the state
with a spin S = 5/2 and zero orbital moments. The bivalent Fe2+ ion has a
spin of 2, and its contribution comes to ~4μB unless one considers a
certain possible contribution from the orbital moment [2]. The measured
value of the magnetic moment for magnetite is 4.1μB. The magnetic moments
of Fe3+ ions are antiparallel to each other, and the observed moment is
determined by Fe2+ ions only (Fig. 1).
Thanks and regards
Bhamu
On Wed, Dec 1, 2021 at 2:17 PM Matteo Cococcioni <matteo.cococcioni at unipv.it>
wrote:
> the fourth iteration does not look too good to me as the U for Fe2 is
> suddenly going from 4.31 to 0.7.
> Fe3O4 is a difficult system of its own. are you sure that all the Fe1 are
> of the same kind (3+)? The system is for sure
> a mixed valence one but I don't remember how this matches its magnetic
> structure. Are you relaxing the structure in between?
> Maybe because of the structure changes hp is not recognizing the neighbors
> it had before for each Fe.
>
> Regards,
>
> Matteo
>
> Il giorno mer 1 dic 2021 alle ore 06:07 Dr. K. C. Bhamu <
> kcbhamu85 at gmail.com> ha scritto:
>
>> Dear Iurii,
>> This is from the calculation without nbnd in QE input file (QE input and
>> HP input are supplied at the end of this email).
>> I am trying to get a converged value of U+V. Up to four steps, all the
>> calculations seem to be fine (other than the converged value of U+V). But
>> in the 5th step, I see that my calculations are not converging.
>>
>> I am getting this message:
>> atom # 5 q point # 3 iter # 180
>> chi: 1 -83.8242814567 residue: 103.6191835332
>> chi: 2 -83.8242814548 residue: 103.6191835384
>> chi: 3 ************** residue: 381.0626501229
>> chi: 4 ************** residue: 326.7198566114
>> chi: 5 317.5473216386 residue: 329.6382092529
>> chi: 6 ************** residue: 580.5784405595
>> chi: 7 9.0496504843 residue: 1.9763844786
>> chi: 8 -6.2884312048 residue: 21.8495062869
>> chi: 9 9.0496505082 residue: 1.9763845088
>> chi: 10 -6.2884312084 residue: 21.8495062832
>> chi: 11 10.2338160751 residue: 7.3557134894
>> chi: 12 -5.1687829951 residue: 22.9825921172
>> chi: 13 6.7503553298 residue: 7.1360232476
>> chi: 14 -3.9982889995 residue: 27.0388948104
>> Average number of iter. to solve lin. system: 23.6
>> Total CPU time : 63284.6 s
>> ###################
>> atom # 5 q point # 3 iter # 274
>> chi: 1 ************** residue: **************
>> chi: 2 ************** residue: **************
>> chi: 3 ************** residue: **************
>> chi: 4 ************** residue: **************
>> chi: 5 ************** residue: **************
>> chi: 6 ************** residue: **************
>> chi: 7 ************** residue: 78.8265104616
>> chi: 8 190.5140692168 residue: **************
>> chi: 9 ************** residue: 78.8264974402
>> chi: 10 190.5140679621 residue: **************
>> chi: 11 ************** residue: **************
>> chi: 12 544.1915326984 residue: **************
>> chi: 13 ************** residue: **************
>> chi: 14 ************** residue: **************
>>
>> U+V value from four completed calculations is given below:
>>
>>
>> site n. type label spin new_type new_label Hubbard U (eV)
>> 1 1 Fe1 1 1 Fe1 5.1791
>> 5 2 Fe2 -1 2 Fe2 5.0962
>> 7 3 O 0 3 O 7.1203
>>
>> site n. type label spin new_type new_label Hubbard U (eV)
>> 1 1 Fe1 1 1 Fe1 5.6013
>> 5 2 Fe2 -1 2 Fe2 4.3164
>> 7 3 O 0 3 O 7.5401
>>
>> site n. type label spin new_type new_label Hubbard U (eV)
>> 1 1 Fe1 1 1 Fe1 5.6013
>> 5 2 Fe2 -1 2 Fe2 4.3164
>> 7 3 O 0 3 O 7.5401
>>
>> site n. type label spin new_type new_label Hubbard U (eV)
>> 1 1 Fe1 1 1 Fe1 4.4843
>> 5 2 Fe2 -1 2 Fe2 0.7259
>> 7 3 O 0 3 O 7.3480
>>
>>
>> I do not think changing "conv_thr_chi" will help here as the number is
>> still oscillating in integer values.
>>
>> What suggestions do you have for this situation?
>>
>>
>> HP input file:
>>
>> &inputhp
>> prefix = 'pwscf',
>> niter_max=300
>> outdir = './temp/',
>> nq1 = 2, nq2 = 2, nq3 = 2,
>> conv_thr_chi = 1.0d-5,
>> find_atpert = 1
>> docc_thr=1.d-3
>> !disable_type_analysis=.true.,
>> /
>>
>> QE input file:
>> &CONTROL
>> calculation = 'vc-relax'
>> restart_mode='from_scratch',
>> etot_conv_thr = 1.4000000000d-04
>> forc_conv_thr = 1.0000000000d-04
>> outdir = './temp/'
>> prefix = 'pwscf'
>> pseudo_dir = '~/PPs/jana_2/'
>> tprnfor = .true.
>> tstress = .true.
>> verbosity = 'high'
>> /
>> &SYSTEM
>> degauss = 0.022
>> ecutrho = 320
>> ecutwfc = 40
>> occupations = 'smearing' , smearing = 'mp'
>> ibrav = 0
>> nat = 14
>> nosym = .false.
>> ntyp = 3
>> nspin=2
>> starting_magnetization(1) = 0.8
>> starting_magnetization(2) = -0.8
>> starting_magnetization(3) = 0.0
>>
>> lda_plus_u = .true.,
>> lda_plus_u_kind = 2,
>> U_projection_type = 'ortho-atomic',
>> !Hubbard_V(1,1,1) = 1.d-8
>> !Hubbard_V(5,5,1) = 1.d-8
>> !Hubbard_V(7,7,1) = 1.d-8
>> Hubbard_parameters = 'file'
>>
>>
>> /
>> &ELECTRONS
>> conv_thr = 1.0000000000d-06
>> electron_maxstep = 400
>> mixing_beta = 1.5000000000d-01
>> /
>> &IONS
>> ion_dynamics = 'bfgs'
>> /
>> &CELL
>> /
>>
>> ATOMIC_SPECIES
>> Fe1 55.845 Fe.pbe-nd-rrkjus.UPF
>> Fe2 55.845 Fe.pbe-nd-rrkjus.UPF
>> O 15.9994 O.pbe-rrkjus.UPF
>> ATOMIC_POSITIONS (crystal)
>> Fe1 0.0000000000 -0.0000000000 -0.0000000000
>> Fe1 0.5000000000 0.0000000000 0.0000000000
>> Fe1 0.9999999892 0.5000000216 -0.0000000000
>> Fe1 0.9999999965 0.9999999956 0.5000000115
>> Fe2 0.3749993721 0.3749993349 0.3750019210
>> Fe2 0.6250006225 0.6250006532 0.6249981020
>> O 0.2454868415 0.2454868399 0.2454949640
>> O 0.7545131845 0.7545131098 0.7545050345
>> O 0.7635313546 0.2454868399 0.2454949640
>> O 0.2364686713 0.7545131098 0.7545050345
>> O 0.2454868307 0.7635313746 0.2454949640
>> O 0.7545132523 0.2364686217 0.7545050345
>> O 0.2454908593 0.2454908222 0.7635274591
>> O 0.7545090629 0.7545091741 0.2364725394
>>
>> CELL_PARAMETERS (angstrom)
>> 5.944682649 0.000000000 0.000000000
>> 2.972341325 5.148246192 0.000000000
>> 2.972341325 1.716082064 4.853885701
>>
>> K_POINTS automatic
>> 7 7 7 0 0 0
>>
>>
>> Regards
>> Bhamu
>>
>> _______________________________________________
>> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
>> users mailing list users at lists.quantum-espresso.org
>> https://lists.quantum-espresso.org/mailman/listinfo/users
>
>
>
> --
> Matteo Cococcioni
> Department of Physics
> University of Pavia
> Via Bassi 6, I-27100 Pavia, Italy
> tel +39-0382-987485
> e-mail matteo.cococcioni at unipv.it <lucio.andreani at unipv.it>
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20211201/f10d7e03/attachment.html>
More information about the users
mailing list