Trying to quantify the elemental composition in some XPS spectra I found some inconsistencies between the sensitivity factors.
I generally use the ASF (Atomic Sensitivity Factors) from Handbook of Photoelectron Spectroscopy (10.1002/sia.740030412). But, as I needed the ASF for some secondary peak, which are not present in the handbook, I searched for new sources... And I fell down the rabbit hole.
Beside the fact that the names change from sources: ASF, SF, RSF, σ... Probably with some differences that I didn't catch.
The numbers may vary a lot.
I compared two sources that do not take into consideration the source energy: the ASF of the handbook ( https://i0.wp.com/xpslibrary.com/wp-content/uploads/2019/08/PHI-54-deg-RSF-list.jpg ) and the values obtained by Scofield, which seems to be used by CasaXPS ( https://xpslibrary.com/wp-content/uploads/2019/09/IP-e-SF-Periodic-Table-Aug-30-2019-v3-1.pdf ). And two sources that instead gives different values with different source energy: Electra ( https://vuo.elettra.eu/services/elements/WebElements.html ) and AMRSF ( https://www.npl.co.uk/getattachment/research/surface-tech/test/xpsamrsf.pdf ).
For example, normalising to 1 the value of C 1s the factors for Au 4f are:
Handbook: 21.11
Scofield: 17.12
Electra @Al: 18.37
Electra @Mg: 18.74
AMRSF @Al: 19.28
AMRSF @Mg: 20.14
With differences up to 23%!!!
Which should be used?
Should I conclude that XPS is not really quantitative?
XPS itself is definitely quantitative, but there are always limitations to methods.
In the case of the sensitivity factors, these involve a cross section for the primary excitation of the electron. Depending on the source, these are calculated/determined in different manners. Scofield and Yeh used differently modeled Hartree-Fock calculations with different basis sets - and like always in quantum chemistry, different basis sets lead to different results.
You are actually adressing a problem which bothers quite a number of people in the community, though. At last year's IOP Dielectrics there was quite the discussion why there is not routine code implemented in the standard QC software to calculate individual cross sections for a compound - after all, the core electrons you investigate here may be polarized, differently attracted or repelled etc. all of which would change the matrix elements involved here.
You could ask the authors of this page: https://a-x-s.org/research/cross-sections/ for deeper information, they were among the spokespeople at that conference discussion.
In Addition to what @Jürgen Weippert has already mentioned, there are additional corrections of the photoionization cross section, e.g. due to the polarization of the X- Ray source and the geometry of the set-up, See e.g. Reilman et al, or Bautista, If I remember correctly, both authors have works that appeared in different contexts such as astronomy. And Not only surface Science. At least You will get some ideas for the observed variations. Best regards, Dirk
Jürgen Weippert, Dirk Luetzenkirchen-Hecht, thank you for your answers.
Mine was a provocative statement: I too believe that XPS is quantitative.
I understand your arguments, and think they are important considerations for spectroscopists specifically studying XPS.
But for practical use, for those who use this technique as a further characterization of their materials, do you have a suggestion on which set to use?
Another aspect to consider when talking about the sensitivity factors in XPS is the transmission function of your analyzer. This can vary quite a bit for different setups, but even for the same setup different settings can yield different transmission functions. Your best option would be to use the sensitivity factors given by the manufacturer of your analyzer (if provided). You can also determine the transmission function of your analyzer yourself, see e.g. https://public.wsu.edu/~scudiero/documents/Calibration.pdf
I would say the most important part is accurately reporting in your publication what you used, because then other people else who wish to compare their results to yours will be able to compare stoichiometries in a fully quantitative manner.
Unfortunately, this is often not done in the proper manner. Many publications are accepted either without any cross section/sensitivity factor included whatsoever, or only the cross section and none of the other factors are mentioned.
If you see a publication from someone else and they didn't report anything, they probably used Scofield, but of course that creates an avoidable insecurity.
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts