Comparison of validity and consistency of methods for quantitative XPS peak analysis

Abstract

Abstract In the present work, we have tested the consistency and validity of the Tougaard, Shirley and straight‐line methods for determination of peak intensities in XPS. Intensities were determined from essentially all photoelectron peaks (with Mg Kα and Al Kα radiation) from seven polycrystalline elemental solids (Co, Ni, Cu, Zn, Ag, Pt and Au) and for three sets of polycrystalline alloys (Cu 0.75 Au 0.25 , Cu 0.5 Au 0.5 , Cu 0.25 Au 0.75 , and Co 0.75 Ni 0.25 , Co 0.5 Ni 0.5 , Co 0.45 Ni 0.55 , Co 0.25 Ni 0.75 and Cu 0.55 Ni 0.44 Mn 0.01 ). For the three methods, the consistency of the peak intensity ratios determined under a variation of analyser settings was studied. The validity of the three methods was studied by comparison of peak intensity ratios to two sets of theoretical photoionization cross‐sections (due to Scofield and Band et al. , respectively). The inelastic electron mean free paths were taken from theoretical tabulations when available or from analytical expressions (both due to Tanuma et al. ). The Tougaard method is found to be superior to the Shirley and straight‐line methods, both with respect to consistency and validity. For a selected set of peaks where the error on theory is minimal, the deviation from theory is 3% for the Tougaard method and 33–60% for the Shirley and straight‐line methods. Considering all XPS peaks from seven metals, the root‐mean‐square deviation from theory for the Tougaard method is found to be ∼11% for the ratios of peak intensities to a peak from the same solid and ∼14% for the ratios of peak intensities to the Au 4d peak intensity. These deviations fall within the expected accuracy of theory. The corresponding root‐mean‐square deviations from theory for the Shirley and straight‐line methods are found to be 35% and 25%, respectively. For alloys, the Tougaard method gives the most consistent description of the shake‐up electrons. For the Shirley and straight‐line methods applied to spin‐orbit doublet peaks, the accuracy is in general worse than the above.