you write: "Is it right that the calibration should be done according to zero count should be Ef=0 binding energy?"

answer: maybe, maybe no. Much depends on what your sample is (polycristalline, textured, single crystalline), what you know about it (band structure, band gap, n- or p-doped). From what you write, I reckon it's a semiconductor.

This may become a difficult task, for several reasons. But that also depends on the level of accuracy you want to achieve. What is the goal and what is current practice in the lab you're working at? (so what's the opinion of the pro's around you?)

If you have a single crystal, you can do ARUPS and determine the band dispersions. It is then tempting to associate the peak at smallest binding energy you can find with the VBM. However, with standard He lab sources, you don't actually know whether your spectrum picks the right momentum vector along the surface normal.

To give you an idea how this could be done with a single crystal, you can have a look in the paper I linked below. It's not recent and measured with "old style" instrumentation (channeltron detectors in analysers, no 2D detectors). But it should be instructive. You will notice that measurements with synchrotron radiation (variable photon energy) were required to assign the VBM. (Note that newer, work with better resolution yields a different conclusion about the location of the VBM).

When measuring a polycristal (truly random) you may run into the problem to have too little intensity near the VBM to actually find it in your spectrum. But you can do cross-checks. If you know the band gap of your material, then the VBM cannot be off by more than that from the Fermi edge position of a metallic sample (gold or silver, preferably) measured under the same consitions. That would give you a first indication. If you know the character of doping, then you also know, whether you expect the VBM or the CBM nearer to the Fermi level (btw.: do you know, whether you have Ohmic contact to your sample? If it's a Schottky contact there may occur extra shifts, rendering the latter conclusions invalid!)

With XPS, you have much worse energy resolution. Therefore the intensity will usually extend over the Fermi level and it is hard to know what to do.

OK, I'll leave it at that for the moment.

P.S.: why is your question tagged with "SPM" and "XAS"?

http://journals.aps.org/prb/abstract/10.1103/PhysRevB.55.10400

Article Occupied and unoccupied electronic band structure of WSe_ {2}