energy of the TS calculated with NEB is lower!

[ross]

Hi,
Recently,during my NEB calculation,I encountered one serious problem,that is the energy of the intermediate state is lower than the endpoints. I have checked that the POSCAR of the three internediate made by the nebmake.pl are reasonable.And the forces are quite small during the calculation.I notice that the position of the interstitial atom in the TS moved a lot. one example(not converge yet and the forces on the atom become larger):
0.5416666666666640 0.5000000000000000 0.6666666666666640
to
0.5175850415333922 0.5000000000000000 0.6666666666666643
here is my INCAR:
ISTART = 0 job : 0-new 1-cont 2-samecut
ICHARG = 2 charge: 1-file 2-atom 10-const
INIWAV = 1 electr: 0-lowe 1-rand
Ts search
SPRING = -5
LCLIMB = .TRUE.
IMAGES = 3
IBRION = 3
POTIM = 0
LPLANE = .TRUE.
NPAR = 6
LSCALU = .FALSE.
NSIM = 4
IOPT = 3

Electronic Relaxation
PREC = Accurate
ENCUT = 300 eV
GGA = 91 XC : PB/PW/LM/91/PE-PBE/RP
ISPIN = 1
NBANDS = 300 total number of bands including empty bands
IALGO = 38 algorithm
NELM = 60; NELMIN= 0; NELMDL= 3 Number of ELM steps
EDIFF = 1E-05 stopping criterion for ELM
BMIX = 1.0 mixing parameter
ISMEAR = 1
ISYM = 1 symmetry: 0-off 1-on (default for PP) 2-more efficient mem (default for PAW)
TIME = 0.05

Geometric Relaxation
ISIF = 2 relax cell shape, volume, or ions
EDIFFG = -1E-02 stopping criterion for IOM
NSW = 70 number of steps for IOM
Others
LREAL = .TRUE. real space projections, TRUE recomended if atoms are more than 20
#LWAVE=.TRUE. write wave function (default) or not
LCHARG=.TRUE. write charge density (default) or not

DOS related values:
ISMEAR = 1 ; SIGMA = 0.1
EMIN = -15 ; EMAX = 10


thanks for any reply!
ross

[graeme]

Try minimizing the lower energy intermediate structure. It is possible that there is an intermediate minimum between your initial and final states.

[ross]

[quote="graeme"]Try minimizing the lower energy intermediate structure. It is possible that there is an intermediate minimum between your initial and final states.[/quote]
thank you very much graeme. In fact,my two endpoints corresponded two octahedron site of the bcc(the interstitial atom located nearly in the exact posions i.e 0.5,0.5,1), you mean there is a minimum with lower energy compared with the exact octahedron site ?

[graeme]

Yes, that is what I'm suggesting.

Also, make sure that your NEB calculation has the same parameters as the endpoint minimizations.

[ross]

thank you very much ,graeme.I will take a try.