Problems in lanthanoid with 4f electrons in the valence
Posted: Tue Mar 04, 2014 10:01 am
Dear VASP Admin,
Just recently I am trying to perform calculations for a system consisting of 4f-electrons in some Lanthanoid elements. For the case of Ce and Pr there were no problems. However, when the systems consist of Sm and Gd including their 4f-electrons in the valence the calculations encounter some errors. Usually I use either POTCARs generated from VASP4 or VASP5. For most of my calculations I always use GGA-PBE POTCARs. I worked successfully when I consider 4f-electrons of Ce and Pr but keep failing when I consider Sm and Gd (I am not sure though what will happen with other La-elements). Here are some examples of errors I have encountered when I calculated Sm2O3 system:
1. When I used PAW-PBE for Sm, no matter what GGA-tag (PS,PE,P91,LDA) I used it always give me this error
N E dE d eps ncg rms rms(c)
DAV: 1 0.129150048634E+05 0.12915E+05 -0.25783E+05 2080 0.228E+03
DAV: 2 0.402956447822E+04 -0.88854E+04 -0.76319E+04 2096 0.647E+02
DAV: 3 0.448774649753E+03 -0.35808E+04 -0.28584E+04 2136 0.291E+02
Error EDDDAV: Call to ZHEGV failed. Returncode = 9 2 16
2. Then I tried to change the POTCAR and used the one in VASP4, the PW91 one. No matter what GGA-tag I used, the calculations always converge with dubious results like this (i.e. magnetization reached 160):
N E dE d eps ncg rms rms(c)
DAV: 1 -0.802645254495E+03 -0.36124E-01 -0.51472E+00 2288 0.410E+00 0.160E+00
DAV: 2 -0.802615948335E+03 0.29306E-01 -0.45901E-01 2784 0.190E+00 0.686E-01
DAV: 3 -0.802610139594E+03 0.58087E-02 -0.66102E-02 3072 0.554E-01 0.259E-01
DAV: 4 -0.802610546022E+03 -0.40643E-03 -0.16988E-02 3088 0.217E-01 0.897E-02
DAV: 5 -0.802610815386E+03 -0.26936E-03 -0.45137E-03 3016 0.130E-01 0.815E-02
DAV: 6 -0.802610909447E+03 -0.94062E-04 -0.97667E-04 2352 0.687E-02
4 F= -.80261091E+03 E0= -.80261091E+03 d E =-.178514E-02 mag= 159.9999
trial-energy change: -0.001785 1 .order -0.003528 -0.013557 0.006500
step: 0.5398(harm= 0.6674) dis= 0.00151 next Energy= -802.612545 (dE=-0.342E-02)
bond charge predicted
N E dE d eps ncg rms rms(c)
DAV: 1 -0.802620885791E+03 -0.10070E-01 -0.10789E+00 2272 0.185E+00 0.746E-01
DAV: 2 -0.802614290817E+03 0.65950E-02 -0.94821E-02 2752 0.861E-01 0.321E-01
DAV: 3 -0.802612995251E+03 0.12956E-02 -0.13564E-02 3088 0.262E-01 0.134E-01
DAV: 4 -0.802612991134E+03 0.41176E-05 -0.35061E-03 2680 0.103E-01 0.452E-02
DAV: 5 -0.802613054309E+03 -0.63176E-04 -0.12707E-03 2560 0.785E-02 0.420E-02
DAV: 6 -0.802613065230E+03 -0.10921E-04 -0.22910E-04 1712 0.438E-02
5 F= -.80261307E+03 E0= -.80261307E+03 d E =-.394093E-02 mag= 159.9999
curvature: -0.41 expect dE=-0.312E-03 dE for cont linesearch -0.416E-04
trial: gam= 0.00000 g(F)= 0.451E-03 g(S)= 0.314E-03 ort =-0.143E-02 (trialstep = 0.898E+00)
search vector abs. value= 0.765E-03
reached required accuracy - stopping structural energy minimisation
writing wavefunctions
In these calculations, I have tried to also modified and playing around with the parameters in INCAR file like set VOSKOWN also playing with MAGMOM and etc.
And to emphasize the matter, these kinds of problems never occur in Ce and Pr systems i.e. CeO2, Pr-doped ceria, and Pr2O3. Note that I also use U correction on Sm 4f-electrons since this is the purpose of doing this calculation.
Just recently I am trying to perform calculations for a system consisting of 4f-electrons in some Lanthanoid elements. For the case of Ce and Pr there were no problems. However, when the systems consist of Sm and Gd including their 4f-electrons in the valence the calculations encounter some errors. Usually I use either POTCARs generated from VASP4 or VASP5. For most of my calculations I always use GGA-PBE POTCARs. I worked successfully when I consider 4f-electrons of Ce and Pr but keep failing when I consider Sm and Gd (I am not sure though what will happen with other La-elements). Here are some examples of errors I have encountered when I calculated Sm2O3 system:
1. When I used PAW-PBE for Sm, no matter what GGA-tag (PS,PE,P91,LDA) I used it always give me this error
N E dE d eps ncg rms rms(c)
DAV: 1 0.129150048634E+05 0.12915E+05 -0.25783E+05 2080 0.228E+03
DAV: 2 0.402956447822E+04 -0.88854E+04 -0.76319E+04 2096 0.647E+02
DAV: 3 0.448774649753E+03 -0.35808E+04 -0.28584E+04 2136 0.291E+02
Error EDDDAV: Call to ZHEGV failed. Returncode = 9 2 16
2. Then I tried to change the POTCAR and used the one in VASP4, the PW91 one. No matter what GGA-tag I used, the calculations always converge with dubious results like this (i.e. magnetization reached 160):
N E dE d eps ncg rms rms(c)
DAV: 1 -0.802645254495E+03 -0.36124E-01 -0.51472E+00 2288 0.410E+00 0.160E+00
DAV: 2 -0.802615948335E+03 0.29306E-01 -0.45901E-01 2784 0.190E+00 0.686E-01
DAV: 3 -0.802610139594E+03 0.58087E-02 -0.66102E-02 3072 0.554E-01 0.259E-01
DAV: 4 -0.802610546022E+03 -0.40643E-03 -0.16988E-02 3088 0.217E-01 0.897E-02
DAV: 5 -0.802610815386E+03 -0.26936E-03 -0.45137E-03 3016 0.130E-01 0.815E-02
DAV: 6 -0.802610909447E+03 -0.94062E-04 -0.97667E-04 2352 0.687E-02
4 F= -.80261091E+03 E0= -.80261091E+03 d E =-.178514E-02 mag= 159.9999
trial-energy change: -0.001785 1 .order -0.003528 -0.013557 0.006500
step: 0.5398(harm= 0.6674) dis= 0.00151 next Energy= -802.612545 (dE=-0.342E-02)
bond charge predicted
N E dE d eps ncg rms rms(c)
DAV: 1 -0.802620885791E+03 -0.10070E-01 -0.10789E+00 2272 0.185E+00 0.746E-01
DAV: 2 -0.802614290817E+03 0.65950E-02 -0.94821E-02 2752 0.861E-01 0.321E-01
DAV: 3 -0.802612995251E+03 0.12956E-02 -0.13564E-02 3088 0.262E-01 0.134E-01
DAV: 4 -0.802612991134E+03 0.41176E-05 -0.35061E-03 2680 0.103E-01 0.452E-02
DAV: 5 -0.802613054309E+03 -0.63176E-04 -0.12707E-03 2560 0.785E-02 0.420E-02
DAV: 6 -0.802613065230E+03 -0.10921E-04 -0.22910E-04 1712 0.438E-02
5 F= -.80261307E+03 E0= -.80261307E+03 d E =-.394093E-02 mag= 159.9999
curvature: -0.41 expect dE=-0.312E-03 dE for cont linesearch -0.416E-04
trial: gam= 0.00000 g(F)= 0.451E-03 g(S)= 0.314E-03 ort =-0.143E-02 (trialstep = 0.898E+00)
search vector abs. value= 0.765E-03
reached required accuracy - stopping structural energy minimisation
writing wavefunctions
In these calculations, I have tried to also modified and playing around with the parameters in INCAR file like set VOSKOWN also playing with MAGMOM and etc.
And to emphasize the matter, these kinds of problems never occur in Ce and Pr systems i.e. CeO2, Pr-doped ceria, and Pr2O3. Note that I also use U correction on Sm 4f-electrons since this is the purpose of doing this calculation.