should we include the magnetic properties in the calculation of the phonon spectrum?

[Kun Tao]

Dear everyone,

Now, I am trying to calculate the phonon spectrum of a superconductor. My questions is the following: should we include the magnetic properties in the calculation of the phonon spectrum?

The difference between the tow incar files are the following :

< ISPIN = 2
< MAGMOM = 58*1 24*0 16*0 32*1 49*1 8*1
---
> # ISPIN = 2
> # MAGMOM = 58*1 24*0 16*0 32*1 49*1 8*1
41,44c41,44
< AMIX = 0.005
< BMIX = 0.0001
< AMIX_MAG = 0.01
< BMIX_MAG = 0.0001
---
> # AMIX = 0.005
> # BMIX = 0.0001
> # AMIX_MAG = 0.01
> # BMIX_MAG = 0.0001
69,75c69,75
< LDAU=.TRUE. # AFLOW LSDA+U
< # LDAU_SPECIES=B Lu # LDAU species
< LDAUL=1 1 1 2 0 0 # l-quantum number for which the on site interaction is added (Default 2) automatic LDAUL tab le
< LDAUU=0 0 0 7.0 0 0 # UEFF parameter. Automatic LDAUU table
< LDAUJ=0 0 0 0.7 0 0 # J parameter (if used). Automatic LDAUJ table
< LDAUTYPE=2 # Type of LDA+U.
< LDAUPRINT=0 # Controls verbosity of the L(S)DA+U module. (Default 0) # AFLOW LSDA+U
---
> # LDAU=.TRUE. # AFLOW LSDA+U
> # LDAU_SPECIES=B Lu # LDAU species
> # LDAUL=1 1 1 2 0 0 # l-quantum number for which the on site interaction is added (Default 2) automatic LDAUL tab le
> # LDAUU=0 0 0 7.0 0 0 # UEFF parameter. Automatic LDAUU table
> # LDAUJ=0 0 0 0.7 0 0 # J parameter (if used). Automatic LDAUJ table
> # LDAUTYPE=2 # Type of LDA+U.
> # LDAUPRINT=0

But, when I check the output files, found a huge difference:

540 f = 1.332527 THz 8.372512 2PiTHz 44.448303 cm-1 5.510889 meV
541 f = 1.091772 THz 6.859807 2PiTHz 36.417600 cm-1 4.515209 meV
542 f = 0.751457 THz 4.721542 2PiTHz 25.065901 cm-1 3.107777 meV
543 f = 0.590428 THz 3.709772 2PiTHz 19.694574 cm-1 2.441817 meV
544 f = 0.206928 THz 1.300167 2PiTHz 6.902377 cm-1 0.855786 meV
545 f/i= 0.789976 THz 4.963567 2PiTHz 26.350768 cm-1 3.267080 meV
546 f/i= 1.321425 THz 8.302757 2PiTHz 44.077986 cm-1 5.464976 meV
547 f/i= 1.756884 THz 11.038827 2PiTHz 58.603334 cm-1 7.265890 meV
548 f/i= 2.537332 THz 15.942524 2PiTHz 84.636267 cm-1 10.493563 meV
549 f/i= 2.649730 THz 16.648746 2PiTHz 88.385482 cm-1 10.958407 meV
550 f/i= 2.992821 THz 18.804450 2PiTHz 99.829768 cm-1 12.377318 meV
551 f/i= 3.980351 THz 25.009284 2PiTHz 132.770219 cm-1 16.461415 meV
552 f/i= 4.427076 THz 27.816142 2PiTHz 147.671371 cm-1 18.308923 meV
553 f/i= 4.602357 THz 28.917464 2PiTHz 153.518113 cm-1 19.033827 meV
554 f/i= 4.844559 THz 30.439260 2PiTHz 161.597081 cm-1 20.035492 meV
555 f/i= 5.425162 THz 34.087297 2PiTHz 180.963914 cm-1 22.436674 meV
556 f/i= 5.852109 THz 36.769888 2PiTHz 195.205350 cm-1 24.202388 meV
557 f/i= 6.428802 THz 40.393351 2PiTHz 214.441731 cm-1 26.587396 meV
558 f/i= 6.492420 THz 40.793078 2PiTHz 216.563813 cm-1 26.850500 meV
559 f/i= 7.372367 THz 46.321947 2PiTHz 245.915681 cm-1 30.489670 meV
560 f/i= 7.936976 THz 49.869490 2PiTHz 264.749007 cm-1 32.824705 meV
561 f/i= 10.803572 THz 67.880844 2PiTHz 360.368359 cm-1 44.679998 meV

I can be observed there are a lot of imaginary phonon model with include the magnetic properties. However, when I switch off the magnetic properties I the incar file I obtained the following results:

540 f = 1.895375 THz 11.908994 2PiTHz 63.222909 cm-1 7.838644 meV
541 f = 1.869243 THz 11.744800 2PiTHz 62.351230 cm-1 7.730570 meV
542 f = 1.862656 THz 11.703410 2PiTHz 62.131499 cm-1 7.703327 meV
543 f = 1.844790 THz 11.591160 2PiTHz 61.535581 cm-1 7.629442 meV
544 f = 1.816160 THz 11.411268 2PiTHz 60.580566 cm-1 7.511036 meV
545 f = 1.765935 THz 11.095695 2PiTHz 58.905240 cm-1 7.303322 meV
546 f = 1.752883 THz 11.013690 2PiTHz 58.469887 cm-1 7.249345 meV
547 f = 1.713983 THz 10.769275 2PiTHz 57.172330 cm-1 7.088468 meV
548 f = 1.686510 THz 10.596652 2PiTHz 56.255904 cm-1 6.974846 meV
549 f = 1.654800 THz 10.397415 2PiTHz 55.198184 cm-1 6.843705 meV
550 f = 1.641529 THz 10.314030 2PiTHz 54.755506 cm-1 6.788820 meV
551 f = 1.567347 THz 9.847932 2PiTHz 52.281071 cm-1 6.482029 meV
552 f = 1.552726 THz 9.756067 2PiTHz 51.793374 cm-1 6.421562 meV
553 f = 1.523356 THz 9.571526 2PiTHz 50.813673 cm-1 6.300095 meV
554 f = 1.467319 THz 9.219439 2PiTHz 48.944504 cm-1 6.068347 meV
555 f = 1.409594 THz 8.856741 2PiTHz 47.018995 cm-1 5.829614 meV
556 f = 1.290476 THz 8.108300 2PiTHz 43.045645 cm-1 5.336982 meV
557 f = 0.992179 THz 6.234046 2PiTHz 33.095537 cm-1 4.103325 meV
558 f = 0.589789 THz 3.705752 2PiTHz 19.673233 cm-1 2.439171 meV
559 f/i= 0.513992 THz 3.229504 2PiTHz 17.144912 cm-1 2.125699 meV
560 f/i= 0.530329 THz 3.332156 2PiTHz 17.689873 cm-1 2.193266 meV
561 f/i= 0.578556 THz 3.635177 2PiTHz 19.298561 cm-1 2.392717 meV

Although there is still some tiny imaginary phonon models, but they are quite small more or less it can be acceptable. Now, I don't understand why they differs so much? Any suggestions will be appreciated!

The following it the input file for the calculations :

SYstem

ISTART = 0
NWRITE = 2 : verbosity
ENCUT = 520
ADDGRID = .TRUE.

PREC = accurate : accuracy
NELM = 500 : maximum number of SCF iterations
NELMIN = 8 : minimal number of SCF iterations
NELMDL = -6 : number of non-SCF iterations
EDIFF = 1E-8 : criterion for SCF convergence
ISMEAR = 0 : Methfessel-Paxton smearing
SIGMA = 0.05 : sigma for smearing

GGA = PS
GGA_COMPAT = .FALSE.
VOSKOWN = 1

# IALGO = 48
ALGO = All

# ISPIN = 2
# MAGMOM = 58*1 24*0 16*0 32*1 49*1 8*1

# LNONCOLLINEAR= .TRUE.

LORBIT = 11

# AMIX = 0.005
# BMIX = 0.0001
# AMIX_MAG = 0.01
# BMIX_MAG = 0.0001

LWAVE = .FALSE.
# LCHARG = .FALSE.

LREAL = Auto
####################################
# LDA+U
# LDAU=.TRUE. # AFLOW LSDA+U
# LDAU_SPECIES=B Lu # LDAU species
# LDAUL=1 1 1 2 0 0 # l-quantum number for which the on site interaction is added (Default 2) automatic LDAUL tab le
# LDAUU=0 0 0 7.0 0 0 # UEFF parameter. Automatic LDAUU table
# LDAUJ=0 0 0 0.7 0 0 # J parameter (if used). Automatic LDAUJ table
# LDAUTYPE=2 # Type of LDA+U.
# LDAUPRINT=0 # Controls verbosity of the L(S)DA+U module. (Default 0) # AFLOW LSDA+U

####################################
# Ionic Relaxation
NSW = 1 : maximum number of ionic steps
EDIFFG = -0.5E-2 : maximal residual force in eV/Ang
IBRION = 8 : quasi-Newton optimization
ISIF = 2 : optimize ions, cell shape and volume

With regards,
Kun Tao

[Kun Tao]

By the way, could anyone give me some suggestions about how to remove the tiny imaginary phonon model？ In fact, I only used 1*1 unitcell to calculate the phonon because the unitcell is too large, for example a and b is about 21 \AA. Any suggestions will be appreciated.

With regards,
Kun Tao

[Kun Tao]

Dear everyone,

Attached below are two phonon spectrum. Moreover,could anyone give me some suggestions why I obtained so many flat phonon model?

With regards,
Kun Tao

[pedro_melo]

Dear Kun Tao,

If your system is magnetic, then this should be accounted for in your simulations. From your post I do not understand if you optimised your cell and atomic positions with or without considering magnetism.

Did you use lattice parameters and atomic coordinates obtained from a previous relaxation step that had no magnetism included? If so I advise you to instead consider magnetism in all steps, even during relaxation.

Kind regards,
Pedro

[Kun Tao]

Dear Pedro,

In fact, I have optimized my cell and atomic positions, taking magnetism into account in all of my calculations. However, for the phonon calculations, I tried with two different input files: one that incorporates magnetism and another that does not. Additionally, I have thoroughly checked all the input files, including the INCAR, KPOINTS, POSCAR, and POTCAR files, and confirmed that they are identical except for the variations detailed in the aforementioned INCAR files.

Could you give me some suggestions about why they result in such big differences?

With regards,
Kun Tao

[pedro_melo]

Dear Kun Tao,

OK, then please post your input files (INCAR,POSCAR,POTCAR,KPOINTS), and the OUTCAR for each run. When doing it, please keep the magnetic and non-magnetic cases separate.

Kind regards,
Pedro

[Kun Tao]

Dear Pedro,

Is it possible for you to give me your private email? Thank you very much! My email is: taokun@lzu.edu.cn

With regards,
Kun Tao

[pedro_melo]

Dear Kun Tao,

There are several reasons as to why we prefer that questions and their respective answers be posted here. By doing so, the VASP forum acts as a repository of knowledge with all the issues experienced by our users and their respective solutions. In several cases it can happen that another user has already experienced the same issue as you and is able to provide an answer.

Not only that, but having differentiated treatment could lead to all users requesting that their issues be treated as confidential, thus flooding our email accounts with similar requests. This would invalidate the purpose of the forums and make more difficult for several users to find answers to their problems.

If you have issues regarding confidentiality of your work, I suggest that you recreate the problem on a different system that you are able to share.

Kind regards,
Pedro