unreliable work function versus k point mesh

[naoum_bacalis]

Dear Forum;

For the clean W(001) surface I did a series of single point LDA
calculations (starting from scratch), in a GGA relaxed structure
(obtained by revised PBE functional). This was a study of the electron
work-function with respect to the k point mesh, expecting to converge in
increasing the number of irreducible k points.

As you'll see, in the attached figure, the work-function against the
number of irreducible k points (parameter NKPTS in OUTCAR) shows a
haphazard behavior, rendering these single point LDA calculations
completely unreliable!

Why?

Thank you, looking forward to your answer at your earliest convenience.

Best Regards,

Naoum Bacalis

[ferenc_karsai]

The work function can be a tricky thing and too many parameters are involved in the calculations. You also gave us very little information, so it's hard to say what went wrong.
If the other parameters are not fully converged than it can be that you don't see monotonic behaviour of the work function with respect to the k-mesh.
Here are some possible things you might consider:
-) Is the slab large enough? This is especially important for your vacuum reference energy, but it can also influence your Fermi levels. Both entering in the calculation of the work function.
-) Is the plane wave cut off large enough?
-) You maybe need to use PREC=Accurate