Copyright © 2005 Casa Software Ltd. http://www.casaxps.com
X-ray Photoelectron Spectroscopy (XPS) involves irradiating a sample with X-rays of a characteristic energy and measuring the flux of electrons leaving the surface. The energy spectrum for the ejected electrons is a combination of an overall trend due to transmission characteristics of the spectrometer, energy loss processes within the sample and resonance structures that derive form electronic states of the material under analysis. The instrumental contribution is an unwelcome fact of the measurement process, but the background and resonance peaks offer information about the sample surface.
Figure 1: Wide scan energy spectrum acquired using a Silver anode on a Kratos Axis Ultra
An energy spectrum from a clean gold sample using monochromatic silver X-rays is shown in Figure 1. The peaks in the spectrum correspond to two distinct processes, namely, electrons ejected due to the photoelectric effect and emission due to the Auger effect.
Photoemission results in electrons of a given kinetic energy where the recorded energy is determined from a combination of the binding energy of the electron (with respect to some molecular form for each element) and the characteristic energy of the X-ray source. The dependence of the energy spectrum on the anode used to produce the X-rays means that the photoelectric lines move in position when the same sample is analyzed but using a different anode to produce the X-rays. Since these lines provide the bulk of the chemical information found in the spectra, and it is this chemical information that makes XPS such a powerful technique, it has become common practice to display XPS spectra using binding energy for the abscissa. The kinetic energy scale is reported relative to the photon energy of the excitation source and so photoelectric line positions with respect to a binding energy scale become independent of the X-rays used to excite the sample, while Auger line positions are invariant with respect to the X-ray anode only when plotted against a kinetic energy scale.
Figure 2: Clean Gold sample measured using a monochromatic Al X-rays.
The Auger electrons are emitted with kinetic energies that are only dependent on the electronic state of the element responsible for the ejected electron. That is to say, unlike the photoelectric lines, changing the X-ray characteristic energy does not alter the position of the Auger lines in the recorded spectra with respect to a kinetic energy scale. Auger electrons are produced by auto-ionization. The collision processes involved with XPS result in electronic excited states with energies that lie above the ionization threshold. Some of these states are meta-stable with respect to radiative transitions within the neutral atom, but may couple with the continuum states of the ionic form to produce a radiation-less transition where the excess energy is transferred to the emitted electron. For this reason an electron spectrum may include Auger features that, when view using a binding energy scale, appear to move as a function of the X-ray source. In reality the Auger peaks always appear at the same kinetic energy and it is the photoelectric lines that move when a different excitation source is used.
Figure 3: X-rays from Cr and Al are both passed through a quartz filter. Photoelectric lines due to excitation from the Cr K beta and Al K alpha X-ray lines are visible in this Gold spectrum.
Although binding energy is the natural scale to use from a chemical perspective, the kinetic energy of an electron is more significant from an instrumental point of view. Most notably the measured intensity of a peak is very dependent on the kinetic energy of the electron. The instrumental transmission characteristics change with kinetic energy and these variations can be quite dramatic. The spectrum in Figure 1 has been recorded using a monochromatic silver X-ray source with photon energy about 2984.2 eV. The intensities of the 4f, 4d, 4p and 4s photoelectric lines are very different from the same peaks measured with an Al X-ray source (1486.6 eV) Figure 2. The drop in transmission efficiency of the analyzer for larger kinetic energies influences the change in intensity. However the relative size of these lines when excited by the different photon energies is thought to be a consequence of the differences in photo-ionization cross-sections for these transitions, which also contributes to the reduction in the signal for the Ag X-ray induced photoelectric lines compared to the Al K-Alpha. Note that in both Figure 1 and Figure 2 the abscissa is kinetic energy to emphasize the different energy of the electrons form these same states, but excited with different photon energies.
Electrons generated by either the photoelectric effect or the mechanisms described by Auger, leave the surface with a characteristic energy provided they have not undergone some energy loss process. The probability of these electrons emerging from the surface without some energy loss is related to the inelastic mean-free-path, which is in turn a function of the kinetic energy of the ejected electrons. The sampling depth for a given photoelectric line recorded with different X-ray anodes is therefore different, while the sampling depth for the same Auger line is unaffected by similar considerations.
Photoelectric lines for a given X-ray anode are governed by the same sampling depth variation as a function of the ejected electron energies. Variations in the relative intensities for lines from a given element may exist due to differences in the depths sampled by the individual lines.
Figure 4: Satellite lines appear with higher kinetic energy values than the dominant Al K alpha induced photoelectric line.
The fundamental differences between photoelectric and Auger lines, means that a spectrum can not be properly understood without both the energy range over which the intensities were recorded and the characteristic energy of the excitation source. Unfortunately it is not uncommon for acquisition systems to omit information about the source under the (limiting) assumption that binding energy will be sufficient for XPS spectra.
The spectra shown in Figure1 and Figure 2 are recorded using monochromatic X-ray sources. Accelerating electrons onto an anode material produces a range of X-ray energies characteristic of that material. A typical anode, such as Magnesium or Aluminium, is chosen because of a dominant, strong resonance in the X-ray spectrum and, as luck would have it, X-rays with wavelengths related by integral multiples of the Aluminium K alpha X-ray line (Figure 3) can be filtered via a quartz crystal to produce, so called, monochromatic X-ray sources. One of the advantages in using monochromatic X-rays is that the distribution of the photon energies used in the analysis is narrow compared to the unfiltered X-ray line and therefore improves the resolution of the photoelectric peaks in the XPS spectrum. A further consequence of filtering the X-rays prior to irradiating the sample is that minor resonance lines in the X-ray spectrum are removed from the excitation mechanism. If unfiltered, these minor X-ray lines produce additional photoelectric peaks in the XPS spectrum and these appear at kinetic energies characteristic of the energy separation between the primary X-ray lines. These satellite lines in the XPS spectrum (Figure 4) actually have the same binding energy as the primary line but without numerically altering the spectrum, this fact is difficult to accommodate using a binding energy scale.
Figure 5: Clean Aluminium Spectrum showing plasmon resonance structures
The background to an XPS spectrum is typically lower when monochromatic sources are used. Unfiltered X-ray sources include photons from Bremsstrahlung and these X-rays excite electronic states out of the reach of the monochromatic source. Electrons generated by highly energetic X-rays undergo inelastic scattering within the surface of the sample and appear as a background at energies within the measurement range.
The X-rays penetrate the surface to depths that exceed those surface layers responsible for photoelectric and Auger peaks in an XPS spectrum. Electrons emitted within the sample may undergo inelastic collisions thus altering the energy of the electron recorded by the detection system. These energy loss processes result in a background of counts that derive from electronic states other than the characteristic energies for the photoelectric lines, but moreover the shape of the background takes on a character determined by the probability distribution for electrons with a given kinetic energy undergoing some modification to their initial value. If this probability distribution also exhibits a resonance type structure, the convolution of a photoelectric peak with this bias in the energy loss distribution may give rise to peak-like structures that may be considered as part of the background. That is to say, peaks may appear in XPS spectra that are actually extrinsic in nature rather than intrinsic to the photo-excitation mechanism.
The gold spectrum in Figure 2 is an example of a material where the energy loss distribution is relatively broad and so the background rises to the lower kinetic energy side of the primary peaks positions, but without a pronounced energy loss shape. This is in contrast to the spectrum shown in Figure 5 where a spectrum from a clean Aluminium sample includes peak shapes between the Al 2s and Al 2p photoelectric lines. These plasmon peaks are associated with energy loss events for electrons from the Al 2p line. In the case of Aluminium the free electrons are constrained to move within energy bands that are characteristic of the material and these material properties influence the shape of the energy loss distribution, namely, scattering of the photoelectric electrons by free electrons with discrete energy bands produces energy loss distributions with relatively narrow structures. The convolution of an intense photoelectric peak with a relatively narrow energy loss distribution results in plasmon structures that could easily be mistaken for a primary line.
A further source for structure in an XPS spectrum is that of two intrinsic energy loss processes. So called, Shake-up peaks appear at energies characteristic of the excited states for an element with respect to the state measured by the zero loss intensity. Some of the photon energy is used to excite the ion out of the zero loss state whilst at the same instant ejecting the photoelectron with the remaining photon energy. Similarly, Shake-off events, where more than one electron is ejected at the time of photo-ionization, may result in broad structures to high binding energies in a spectrum.
Any measurement process requires stability from the system used to measure the signal. Unfortunately, instruments that rely on high voltages will occasionally suffer an unwelcome event during an acquisition. The consequence to an XPS spectrum of such an event is either a loss of signal or spurious counts entering the data. The latter appears as a narrow peak on the spectrum and in the case of multiple detector systems a series of spikes appear, arranged in accordance with the geometry of the analyzer and detectors.