X-ray photoelectron spectroscopy: A perspective on quantitation accuracy for composition analysis of homogeneous materials

We present a perspective on the use of XPS relative peak intensities for determining composition in homogeneous bulk materials. Nonhomogenous effects, such as composition variation with depth or severe topography effects (e.g., in nanoparticles), are not discussed. We consider only the use of conventional laboratory-based instruments with x-ray sources, Alk(alpha) or Mgk(alpha). We address accuracy (not precision, which is much more straightforward) using relative sensitivity factors, RSFs, obtained either empirically from standards (e-RSF) or from the use of theoretical cross sections, sigma, (t-RSF). Issues involved are (1) the uncertainty of background subtraction of inelastically scattered electrons, (2) the accuracy of the RSFs, and (3) the role of XPS peak satellite structure, which affects both (1) and (2) above. The XPS of materials tends to fall into two broad classes: where the signals being used for quantification are "main" peaks, which are narrower and more symmetric, followed by a relatively low background with only weak satellite structure and where the "main" peaks are broader and often asymmetric, followed by backgrounds that are higher and have a stronger satellite structure. The former generally will yield better accuracy, more easily, than the latter. The latter comprises all compounds containing elements with open valence shell electrons. These are mostly the 3d, 4d, and 5d transition metals, the lanthanides, and the actinides. Compounds involving only the first row elements, Li to F, where the 1s binding energy is used for quantitation, are those where the best accuracy is potentially achievable. We specifically address the issue of long-claimed serious discrepancies between e-RSFs and t-RSFs, which were interpreted as indicating calculated sigma s, used as a parameter in t-RSFs, were seriously in error. We conclude this claim to be untrue and that, if done correctly, there is no disagreement between the two approaches within the limits of accuracy claimed. Finally, we suggest protocols for rapid element composition analysis by obtaining relative XPS signal intensities using only low energy resolution.