Unable to converge the Single point calculation using Dy potential(f in valence)

[shelaka_15]

Dear All
I am facing issue in convergence in one of my system having Dy atom. I am trying to do a single point calculation but in output I do not see the energy lowering down as can be seen in the snippet below. If I take Dy_3 pseudopotential with f inside the core, it works fine, but I do not want f in core hence I was trying with Dy pseudopotential having f as valence. Kindly help.
This is my INCAR:

Code: Select all

Electronic structure
  PREC    = NORMAL
  LREAL   = Auto
  ENCUT   = 520
  ISYM    = 0
  ISPIN   = 2
Electronic minimization
  IALGO   = 48
  EDIFF   = 1e-6
  NELM    = 2000
  ISMEAR  = 0
  SIGMA   = 0.05
  MAGMOM  = 1*6 130*0 30*0
  NUPDOWN = 5
  LORBIT  = 11
Density functional
  GGA     = PE
  IVDW    = 12
Hubbard correction
  LDAU    = .TRUE.
  LDAUPRINT = 0
  LDAUTYPE  = 2
  LDAUL = 3 -1 -1
  LDAUJ = 0 0 0
  LDAUU = 6 0 0
Ionic relaxation
  ISIF    = 3
  EDIFFG  = -0.05
  NSW     = 0
  IBRION  = -1
  POTIM   = 0.1
Parallelization
  NSIM    = 4
  NPAR    = 2
  LPLANE  = .TRUE.
  LSCALU  = .FALSE
Other parameters

A part of outvasp file

Code: Select all

       N       E                     dE             d eps       ncg     rms          rms(c)
RMM:   1    -0.127822545279E+04   -0.12782E+04   -0.27219E+01  2512   0.339E+00    0.159E+00
RMM:   2    -0.127742295452E+04    0.80250E+00   -0.22758E+01  2481   0.305E+00    0.246E+00
RMM:   3    -0.127695982433E+04    0.46313E+00   -0.11985E+01  2500   0.210E+00    0.209E+00
RMM:   4    -0.127717504884E+04   -0.21522E+00   -0.10854E+01  2429   0.177E+00    0.165E+00
RMM:   5    -0.127729372327E+04   -0.11867E+00   -0.13374E+01  2526   0.203E+00    0.157E+00
WARNING in EDDRMM: call to ZHEGV failed, returncode =   8  4    152
RMM:   6    -0.127768995125E+04   -0.39623E+00   -0.18054E+01  2473   0.210E+00    0.992E-01
RMM:   7    -0.127816396620E+04   -0.47401E+00   -0.26527E+01  2138   0.245E+00    0.223E+00
RMM:   8    -0.127675707087E+04    0.14069E+01   -0.19610E+01  2124   0.282E+00    0.138E+00
RMM:   9    -0.127773638163E+04   -0.97931E+00   -0.20195E+01  2357   0.224E+00    0.564E-01
RMM:  10    -0.127830844383E+04   -0.57206E+00   -0.27889E+01  2453   0.250E+00    0.182E+00
RMM:  11    -0.127777624084E+04    0.53220E+00   -0.29644E+01  2244   0.298E+00    0.289E+00
RMM:  12    -0.127754432564E+04    0.23192E+00   -0.27615E+01  2087   0.291E+00    0.546E-01
RMM:  13    -0.127729886974E+04    0.24546E+00   -0.16502E+01  2344   0.243E+00    0.978E-01
RMM:  14    -0.127732017042E+04   -0.21301E-01   -0.13956E+01  2239   0.199E+00    0.677E-01
RMM:  15    -0.127782474788E+04   -0.50458E+00   -0.20145E+01  2478   0.232E+00    0.561E-01

[manuel_engel1]

In order to assist you, I need some additional information. Could you please provide all relevant input and output files according to the Forum Posting Guidelines.

[alex]

Hi,

if you starting a new project, it's maybe also a good idea to look into the new generation of PAW (version 6.4) released recently.
I know from my earlier days that these f-states are notorious cases.

Good luck there!

alex

[yong_han]

POTCAR for Dy (including f) is problematic even for energy convergence. Instead, Dy_3 can work well, but you need to check the results to be consistent with your expected properties. Also see the last paragraph of APPENDIX B in Phys. Rev. Mater. 5 (2021) 074004.