> Dear Andrei Postnikov, > Thank you very much. > I want to know why the two figures are so different (I expected them > almost > be superposition) after shift the fermi level. > Even 1e is added to O2 molecule, it is not reasonable that the 2p orbitals > of O2, O2- is so different. > How does siesta find the systems fermi level
Dear Fen Hong, there are two different questions: 1. How does siesta find the systems fermi level? - has been indeed discussed before; it is SOMEHOW calculated with respect to SOMEHOW averaged potential, but I don't think the absolute number has any useful meaning. For any practical use, you'd compare either (differences of) total energies, or energies of peaks aligned to the Fermi level (i.e., chemical potentials). A priori you shouldn't expect any specific relation between Fermi levels in two DIFFERENT systems anyway. Now you wonder why the absolute Fermi levels are SO different? - because one of your systems is charged, so that (somehow, somewhere) the energy is added of the order of electrostatic energy of elementary charges separated by your cell size. Since such energy would diverge when summed up over infinite "crystal" of your molecules, it is somehow corrected for in Siesta (the story about Madelung terms, and that charged systems can only be correctly treated in cubic cells). 2. If we align two spectra somehow, say by positions of their 2s levels, why do their 2p do not match? - Again, because the systems are different, and the energy of 2p shell with 4 electrons cannot be the same as with 4.5 electrons. To get an idea, look at results of (all-electron) single atom calculation (in the pseudopotential part, before doing any Siesta) and at corresponding orbital energies. These are just technical issues, not physical ones. Concerning the latter, I do not address the question how good/bad a DFT calculation on O2- molecule can in principle be; quite probably an established knowledge already exists on this subject. Best regards Andrei Postnikov
