My Community

Dear all,

I have a question on the frequence calculation with VASP.
First, I optimized the slab with INCAR1 (relaxed the two top layers of the 4-layer slab and the adsorbed probe molecule)
Then, i calculated the frequence of the probe molecule with INCAR2.(only relaxed the probe molecule and the atom of the adsorption site of the slab) the problem is that there are always several f/i in the OUTCAR reslut.
could anyone help to solve this proble?
Thanks in advance!!
the INCAR1 for structure optimal as listed :
SYSTEM = XXX
ENCUT = 400
ISTART = 0 ; ICHARG = 2
ISMEAR = 0
LREAL = Auto
dynamic:
INIWAV = 1
ALGO = Fast
IBRION = 1
NSW = 400
POTIM = 0.1
ISIF = 2
GGA = 91
VOSKOWN = 1
PREC = Accurate

the INCAR2 for frequence calculation as listed:
SYSTEM = XXX
ENCUT = 400
ISTART = 0 ; ICHARG = 2
ISMEAR = 0
LREAL = Auto
dynamic:
INIWAV = 1
ALGO = Fast
IBRION = 5
NSW = 400
POTIM = 0.01
ISIF = 2
GGA = 91
VOSKOWN = 1
PREC = High
NFREE = 2
EDIFF = 0.000001

your input to calculate the vibrational frequencies looks ok, so the reason for obtaining imaginary frequencies most probably is that
-- either the input geometry does not correspond to the real equilibrium structure (i.e. it is not the global minimum, but a saddle point on the potential energy hypersurface).
-- or that (one of) the ionic steps to calculate the Hessian matrix is not fully conveged electronically.

Please therefore check
--) the OUTCAR file of run 1 to see if the forces (especially on the adsorbate molecule and the topmost layers of the substrate are below 0.02 eV/Ã…) If this is not the case, please do a further relaxation run, setting
EDIFF= 1e-06 EDIFFG= -0.01
--) the total energies of run 2: the total energy of the starting geometry (1st step) has to be lower than the total energies of all displacement steps. Furthermore, all steps have to be fully converged electronically, i.e.: please check if the convergence limit of 1e-06 is reached or if just the forces are calculated after 60 electronic steps have been done. (please increase NELM if 60 steps are not sufficient to reach electronic convergence) It may also be necessary to decrease EDIFF to 1e-08 in case that the minimum is very shallow.