My Community

Hi All,
I am attempting to calculate the phonons for U₃Si₂ using the GGA (PBEsol)+U functional. The optimized lattice constants with U=1.7 are:

a = b = 7.295 Å
c = 3.857 Å,
which is in comparison to experimental data:
a = b = 7.315 Å
c = 3.893 Å.
Below, I have provided the INCAR, POSCAR-unitcell files, and KPOINTS that I used for the calculations at the end of this message.

For the phonon calculations, I employed a 2x2x4 supercell with five displacements generated by the Phonopy code. However, the calculations are not converging. I attempted changing ALGO = Normal to ALGO = VeryFast and ISMEAR = 2 to ISMEAR = 3. I also increased NELM to 150, but these adjustments did not resolve the issue.

The total drift forces for the atomic displacements vary significantly from one displacement to another. For example:

For displacements 1 and 2, the total drift force is < 1e-4 eV/Å,
For displacement 3 and 5, it is around 1e-2 eV/Å,
For displacement 4, it is approximately 1e-3 eV/Å.
Additionally, I encountered the following warning messages:
WARNING in EDDRMM: call to ZHEGV failed, returncode = 6 3 79. and
This one also
-----------------------------------------------------------------------------
| |
| W W AA RRRRR N N II N N GGGG !!! |
| W W A A R R NN N II NN N G G !!! |
| W W A A R R N N N II N N N G !!! |
| W WW W AAAAAA RRRRR N N N II N N N G GGG ! |
| WW WW A A R R N NN II N NN G G |
| W W A A R R N N II N N GGGG !!! |
| |
| The electronic self-consistency was not achieved in the given |
| number of steps (NELM). The forces and other quantities evaluated |
| might not be reliable so examine the results carefully. If you find |
| spurious results, we suggest increasing NELM, if you were close to |
| convergence or switching to a different ALGO or adjusting the |
| density mixing parameters otherwise. |
| |
-----------------------------------------------------------------------------

INCAR
# SCF input for VASP
# Note that VASP uses the FIRST occurence of a keyword
SYSTEM = U3Si2_Force calculations

PREC = Accurate
# ENCUT = 400
IBRION = -1
NSW = 0 ! changed from 100
ISIF = 2
ALGO = VeryFast !Normal (blocked Davidson)
NELM = 150
NELMIN = 2
EDIFF = 1.0e-06 ! was 1e-5
EDIFFG = -0.001
VOSKOWN = 1
NBLOCK = 1

ISPIN = 2
GGA = PS #PBEsol
LDAU = .TRUE.
LDAUTYPE = 2 # 2: simplified rotationally invariant (U_J): Dudarev et al
LDAUL = 3 -1 #l-quantum number, -1 means no on-site interaction is added
LDAUU = 1.7 0 # Effective on-site Coulomb Interactions (eV)
LDAUJ = 0 0 # Effective on-site exchange Interactions (eV)
LDAUPRINT = 0 # 1 write the spin-up and spin down matrix to OUTCAR
LMAXMIX = 6 # 4 for d-elements, 6 for f-elements

INIWAV = 1
ISTART = 0
ICHARG = 2
LWAVE = .FALSE.
LCHARG = .FALSE.
ADDGRID = .FALSE.
ISMEAR = 3 ! was 2 was 1
SIGMA = 0.2
LREAL = .FALSE.

NCORE = 4 ! # of nodes
KPAR = 2 ! mpi-rank =16 , NKPOINTS =32 ==> 2 groups, each kpoint on 8 cores

POSCAR-unitcell
U3Si2
1.000000000000000
5.8984426312529790 4.2918888570923786 -0.0195012428659402
-4.2918888270434019 5.8981701645018036 -0.0599561637662552
-0.0103135917650089 0.0316970283885453 3.8564718251165897
U Si
6 4
Direct
0.1854642126921195 0.6854642126921189 0.5000000000000000
0.8145357873078811 0.3145357873078805 0.5000000000000000
0.3145357873078805 0.1854642126921195 0.5000000000000000
0.6854642126921189 0.8145357873078811 0.5000000000000000
0.0000000000000000 0.0000000000000000 -0.0000000000000000
0.5000000000000000 0.5000000000000000 -0.0000000000000000
0.3858851229787523 0.8858851229787528 -0.0000000000000000
0.6141148770212472 0.1141148770212475 0.0000000000000000
0.1141148770212475 0.3858851229787523 -0.0000000000000000
0.8858851229787528 0.6141148770212472 0.0000000000000000

KPOINTS
Automatic mesh
0
Monkhorst Pack
4 4 4
0. 0. 0.

Thank you in advance for your suggestions!
Iyad

Dear Iyad,

thank you for reaching out to us via the official VASP forum.

I am happy to try myself and help you with this specific problem. Would you mind to also send me your POTCAR. Possibly as full set of input files INCAR, POSCAR, KPOINTS, POTCAR in one zip / tar archive? That would help a lot. The non-convergence of the SCF calculation could also due to the choice of the POTCAR.

Best regards,
Alex

Dear Alex,

Thank you so much for willingness to help.

Attached are VAP input files. You will find two INCAR files. One I used for optimization and the other for single point calculations.

Please let me know if you have any questions.

Thank you again,
Iyad

Hi Iyad,

I managed to converge your calculation by changing ALGO=Normal. Note that the RMM algorithm VeryFast can have trouble to converge for LDAU and magnetic system (more degrees of freedom). For this size of system the DAV algorithm is a good choice I would say; the calculation should still be reasonably fast. I also noticed that you did not specify the MAGMOM tag. Please note that this might be a poor choice and slow down convergence. I see that your system becomes Ferromagnetic. Please check if this is what you want. Otherwise set the MAGMOM tag to control the magnetic order.

Let me know if this helps you to convergence the calculations.

Best regards,
Alex

Thank you, Alex.

I will let you know how the calculations proceed.

Best Regards,
Iyad