May I regard this as converged? (ZBRENT: fatal error in bracketing)

[murphy]

Dear VASP users,

I am calculating defect energy for semiconductors.

I encountered the problem of
"ZBRENT: fatal error in bracketing
please rerun with smaller EDIFF, or copy CONTCAR
to POSCAR and continue" for almost every calculation.

But I checked the energy convergence for one calculation
(by the command "grep 'free ' OUTCAR |nl") as follows, I found that the
energy is almost unchanged. May I regard this as converged already?
But what happened with the higher energy in the last cycle?
May I take -243.92600636 eV as the converged energy?
58 free energy TOTEN = -243.92597172 eV
59 free energy TOTEN = -243.92599208 eV
60 free energy TOTEN = -243.92600048 eV
61 free energy TOTEN = -243.92600408 eV
62 free energy TOTEN = -243.92608763 eV
63 free energy TOTEN = -243.92602029 eV
64 free energy TOTEN = -243.92600508 eV
65 free energy TOTEN = -243.92600630 eV
66 free energy TOTEN = -243.92600639 eV
67 free energy TOTEN = -243.92600640 eV
68 free energy TOTEN = -243.92600622 eV
69 free energy TOTEN = -243.92600593 eV
70 free energy TOTEN = -243.92600626 eV
71 free energy TOTEN = -243.92600580 eV
72 free energy TOTEN = -243.92600615 eV
73 free energy TOTEN = -243.92600636 eV
74 free energy TOTEN = -243.92498274 eV

The important INCAR settings are as follows:
EDIFF = 0.0001
EDIFFG = -0.001
IBRION=2
POTIM=0.5
NSW= 200
ISIF = 3
NELMDL = 6
LREAL = Auto
NBANDS=720
ENCUT = 350
ENAUG = 500
IALGO = 48

I really do need your help!
Thank you all!

Sincerely,
Murphy

[kambiz]

Hi,

Why dont you take the CONTCAR and re-run with IBRION = 1? It usually doesnot take much

Cheers

[murphy]

Thank you kambiz,

Sorry for not seeing your reply until now. Can you explain more in detail why should I use IBRION = 1 in the next run?

And I found that when this problem occurs, usually the forces are not converged, though the energy is seemingly converged.

Looking forward to your reply!