how are the DOS energies referenced?

[mhyman]

I'm doing density of states calculations to study the bonding during adsorption. But it seems as though the energies reported in the DOSCAR file don't have a meaningful reference. But I would like to see how the DOS changes during the adsorption process.

For instance, H2S adsorbed on Pd(111).

The lowest energy s orbital shifts by ~2.5 eV from the gas phase molecule to the adsorbed molecule. I don't believe that there really should be a shift, since most of the bonding should take place with the p orbitals of S. Thus, I should be able to line up the DOS plots before and after adsorption, and the lowest energy orbitals should line up.

Am I wrong? Or is the energy not aribtrary at all?

[admin]

usually there are 2 ways to align the energy scales of a DOS
1) shift them with respect to the Fermi level (E_fermi=0). This is the standard method.
2) for slab and adsorption calculations you may align the Es with respect to the "vacuum level"(=0): In a repeated slab calculation, the zero-point of the energy scale can be estimated from the constant - potential region in the center of the vacuum (of course, only if the vacuum between the slabs is thick enough) by setting LVTOT=true in INCAR and plotting the planar-averaged potentials along the surface normal (vtotav.f, file LOCPOT).
3) I would not align them with respect to the S-s-band minimum, because this might already undergo a certain shift. (low-lying Pd
core states might be appropriate, but they are not available straightforwardly from VASP.

[Modey3]

Admin,

Is it true that the DOS of two different systems, for instance an Fe (100) and an Al (100) slab, cannot be compared because there is a constant energy shift in the eigenvalue scale between the two? If that's the case. the two methods you described above to align the DOS scales won't work because there is an unknown shift between the two energy scales. In what situations will that unknown shift not be present?

Best Regards

modey3