My Community

Hello
I calculate dissociation of H2O molecule adsorbed on Si(001)_2x1 surface : Molecular dynamics.
I have three questions.

One, This is my INCAR file. Is this correct?

system = a-SiO2

ALGO = FAST
ISMEAR = -2
SIGMA = 0.05
LREAL = T
PREC = High
ISTART = 0
!EDIFF = 1e-8
!EDIFFG = 1e-7
!NBANDS = 30
TEBEG = 1916.67
ENCUT = 500
LWAVE = .FALSE.
LCHARGE = .FALSE.
LVTOT = .FALSE.

Ionic relaxation
NSW = 1000
ISIF = 2
IBRION = 0
NFREE = 10 ! number of DIIS vectors to save
POTIM = 1

Especially I don't know, how select TEBEG and TEEND.
In advance I use methods that " Tsoll x (Nions-1)/Nions "

Two, I apply velocity to H2O. But the job is not calculate. So, I see the stdout.log
and I find error.
"ERROR FEXCP : supplied Exchange-correletion table is too small, maximal index: 21749"
What is means?

Three, I calculate MD(Molecular Dynamics). But always expose same warning
Warning

"VASP found 36 degrees of freedom
the temperature will equal 2*E(kin)/(degrees of freedom)
this differs from previous releases, where T was 2*E(kin)/(3NIONS).
The new definitions is more consistent"
What is means?

1) in the INCAR file, please replace ISMEAR=-2 by 0 or 1. The occupancies of the levels should not be held fixed in an MD run.
TEBEG and TEEND have to be set if you do simulated annealing of your system, otherwise please just set TEBEG..
2) the FEXCP error is probably due to too small interatomic distances. please have a look at either (if the warning appears in the very first ionic step) the OUTCAR file (data following the 'nearest neighbor table' line) or to the input geometry of the last step ( also seen from OUTCAR, or from XDATCAR). unreasonable distances in higher ionic steps are mostly due to
-- too large relaxation steps (POTIM)
-- electronically non-converged results before the forces are calculated and hence unreliable forces (please note that unless NELM is explicitely set, the forces are calculated after 60 electronic steps, irrespective of whether the convergence criterium is reached or not).
3) this warning is not actually a warning (in the sense that sth is going wrong), but simply tells you that the way T is calculated was changed (as compared to vasp.3) in all versions up from vasp.4.4