Correct binding energy (BE) spectra referencing of insulating samples remains the major challenge in modern XPS analyses. Ar 2p signal of implanted Ar is sometimes used for this purpose. The method relies upon the assumption that chemically inert species such as noble gas atoms would be ideally suited as other factors affecting core level peak positions (such as chemical bonding) can be excluded. Here, we present a systematic study on the Ar 2p referencing method applied to a wide range of thin film sample materials of metals, nitrides, carbides, and borides. All specimens exhibit a well-defined Fermi edge, which serves as an independent internal reference for Ar 2p spectra of in-situ implanted Ar. Ar 2p3/2 binding energy is shown to vary by as much as 5.1 eV between samples. This is more than typical chemical shifts of interest, which obviously disqualifies Ar 2p referencing. The BE of the Ar 2p peaks shows a strong correlation to the number of valence electrons available for screening, implying that the polarization energy has a major role for the observed large spread of Ar 2p3/2 BE values. In several cases of single-phase films, an additional Ar 2p doublet is observed with the Ar 2p3/2 BE referenced to the vacuum level higher than the gas phase value of 248.6 eV, which is tentatively assigned to the formation of Ar-N and Ar-C complexes stabilized by Van der Waals forces. Ar implantation into two-phase samples, exemplified here by phase-segregated NiCrC/a-C:H and nanocomposite c-TiN/SiNx thin films, leads to complex Ar 2p spectra, which further demonstrates unreliability of the referencing method. The firm conclusion of the study is that the Ar 2p3/2 peak from implanted Ar is not a remedy for the charge referencing problem.
Read full abstract