A discussion of approaches for fitting asymmetric signals in X-ray photoelectron spectroscopy (XPS), noting the importance of Voigt-like peak shapes

Although the fundamental, theoretical peak shape in X-ray photoelectron spectroscopy (XPS) is Lorentzian, some Gaussian character is observed in most XPS signals. Additional complexity in the form of asymmetry is also found in many XPS signals, which requires more advanced peak shapes than the traditional, symmetric Voigt and Gaussian-Lorentzian sum and product (pseudo-Voigt) functions. Here, we discuss the merits and disadvantages of four approaches that have been used to introduce asymmetry into XPS peak shapes: addition of a decaying exponential tail to a symmetric peak shape, the Doniach-Sunjic peak shape, the double-Lorentzian, DL, function, and the LX peak shapes, which include the asymmetric Lorentzian (LA), finite Lorentzian (LF), and square Lorentzian (LS) functions. The Doniach-Sunjic peak shape is the only asymmetric, synthetic peak that has a theoretical basis. However, it has an infinite integral, which makes it problematic in quantitative work. The mathematical bases for the LX and DL peak shapes are discussed, and practical examples of their use in peak fitting are presented. The case is made for the Voigt function being the most appropriate function for XPS peak fitting, in general, which suggests that a modified Voigt function may be the most reasonable for fitting asymmetric XPS signals. The LX and DL functions include convolution with a Guassian, which, with the exception of the LS function, makes them Voigt-like functions. The sources of asymmetry and its complexity are discussed. It is emphasized that not every asymmetric spectrum should be fit with an asymmetric peak shape.