Peak Asymmetry in Conducting Materials

Peak Asymmetry

Quick Overview

Typically, spectral peaks from photelectron emission exhibit a symmetrical lineshape, representing a range of energies characterised by the full-width at half maximum. For example, the oxygen peak in figure 1.

Figure 1: (Left) Symmetric O 1s XPS peak and (Right) Asymmetric Ni 2p XPS peak

Metals and other conductive materials such as graphene, however, may in fact produce asymmetric peaks (such as the nickel peak in figure 1). This is the result of final state effects within the metal, caused by the conduction band.

What Causes Peak Asymmetry?

When we think about metals/conductive materials we think of the energy levels as bands. The valence band (equivalent to the HOMO) in nickel consists of the 4s and 3d electrons. The LUMO is the conduction band, a continuum of unfilled molecular energy levels (figure 2).

Figure 2: Nickel valence energy levels and conduction band

Figure 3: Shake-up processes in conducting material

We have discussed shake-up peaks in another section, wherein a photoemission can excite the resulting ion to an energy state slightly above the ground state, typically a few eV. When our LUMO is a conduction band, however we have a continuum of possible shake-up structures which may leave the ion in a variety of states (figure 3).

The result of this is that the final peak possesses an asymmetric character, due to a multitude of final states each having lost a different amount of kinetic energy (and hence increasing effective binding energy of the photoemission).

The degree of asymmetry depends heavily on the density of states (DOS) near the Fermi level. High DOS means there is more of a continuum for shake-up processes.

Figure 4: Comparison of asymmetry of Pt vs Au and DOS near the Fermi level

This is best visualised by looking across the first row of the transition metal series – as we fill the d-band we see more and more symmetry in our resulting photoemission peaks.

Figure 5: Asymmetry and Fermi DOS in first row TMs.

How Can We Account For This?

There exist a number of useful lineshapes in CasaXPS and other software we can use to fit these asymmetries.

The LA lineshape

The most useful lineshape in data description is known as the LA lineshape, which takes the general form of LA (α, β, m).

α and β modify the tail shape at the higher and lower binding energy side of the peak respectively, with an increase in value reducing the spread of the tail.

m may be modified between 0 and 499 and controls the width of the Gaussian function to be convoluted with the Lorentzian function to create the peak.

The LF function

The LF function is identical to the LA function except it includes an additional parameter, it takes the general form:

LF(α, β, w, m)

Where w may force the high binding energy tail towards the background to improve the fit.

The Doniach-Sunjic lineshape

The GL-modified DS lineshape uses the general form:

DS(α, n)GL(X)

Where α is an asymmetry parameter and n (0-499) controls convolution width. Modification of the GL parameter controls tail spread on both energy sides.

This line shape represents a very good theoretical evaluation of asymmetric XPS peaks, however experimentally it often fails due to a lack of limits.