since a while we have a ghost line from our Mg x-ray source, which I cannot identify yet. Any idea where this could come from? Attached is a Au reference sample, measured with a standard Mg x-ray source.
is there any Ni present in the tube; e.g. close to the filament, so that gradually Ni is slightly evaporated or sputtered onto the x-ray target in the course of time?
It seems that in the course of time the tube spectrum might gradually show up Ni L-alpha at about 851,5eV *
So please imagine that Ni L-alpha is present in the spectrum and calculate the kinetic energies of the photoelectrons according to the binding energies of Au5d, Au 5p and Au 4d given in your diagram. As next step 'ignore' the excitation via Ni-L-alpha and assume that these photoelectons are due Mg K-alpha (1253,6eV*) excitation, calculate the hypothetic binding energies and mark these positions in your diagram. You will find excellent agreement with the positions of your unknown ghost peaks.
1253,6eV (Mg K-alpha) - 851,5eV (Ni L-alpha) = 402,1eV; your measured peak shift given above is 402,7eV; perfect!
So please figure out from where any Ni contamination of your x-ray target may arise.
Good luck and best regards
G.M.
*) The x-ray emission line energies are taken from the 'X-Ray Data Booklet':
Hello Thorsten, the ghost peaks are probably from copper. See, for example, here http://xps-cmmp.blogspot.com/2014/06/ghost-peaks.html: "The Mg anode has it's own potential set of ghost peaks. O Kα at + 728.7 eV, Cu Lα at + 323.9 eV, and Al Kα at -233.0 eV. "
Mg anode is tipically a Mg film deposed on a water cooled copper part. So, it's not surprising to see these ghost peaks.
But unfortunately the spectral data of Cu L-alpha (929,7eV) do not match here.
Then the ghost peakshift will be 1253,6eV - 929,7eV= 323,9eV*; and that is far from the observed shift of 402 eV of the above diagram. Maybe there is some kind of adhasive/intermediate coating of Ni to the copper part. In the course of time the Mg layer may be degraded; e.g. thinned by evaporation (melting point 650°C) and/or cracks show up.
*) that is the value (potential set of ghost peak) for Cu L-alpha mentioned above by you.