X-ray Photoelectron Spectroscopy (XPS) is a quantitative analytical method to determine the chemical composition of a sample as well as several electronic properties, e.g. the oxidation state of an element and the binding energies of each of its electrons. To do so, photoelectrons are analysed in terms of their kinetic energy with respect to the X-ray radiation that is used for the experiment.^[1]

Investigated samples are up to some square millimetres of surface area with an analysis depth of around 10 nm.^[2] XPS is a method in which the sample is not destroyed.

Even though the phenomenon was discovered already in early 20th century, the technique is commercially used only since the seventies.

Principle  

X-ray radiation is high energetic electromagnetic radiation of frequencies in a range between 10¹⁶ Hz and 10¹⁹ Hz and a wavelength between 10^-8 nm and 10^-11 nm. When focussed on mater, the electrons of atoms interact with the electromagnetic field of the radiation and can be ejected. Electrons that are ejected due to photons are called photoelectrons and they cause electron holes to be created at their original orbitals. When these holes are occupied by electrons from higher energy levels, either X-ray fluorescence radiation is caused or Auger electrons are ejected radiationless (see Atomic Emission Spectroscopy).^[1] Figure 1 illustrates those processes.

Figure 1. Interaction of X-ray photons with electrons. Electrons that are ejected by electromagnetic radiation are called photoelectrons. If the resulting electron holes are filled, photons or Auger electrons can be released (Figure: Igor Graf von Westarp).

Applying energy conservation, by determining the kinetic energy of photoelectrons the binding energy can be calculated according to the following equation:^[3]

E_r = E_kin+E_b+φ

Where E_r=hν is the energy of the X-ray radiation, E_kin is the kinetic energy of the detected photoelectron, E_b the binding energy of the photoelectron and φ the work function of the spectrometer which leads to energy losses during the detection. With XPS, all electrons can be investigated in terms of their binding energy and related to their specific orbitals. Additionally, the binding energy of core electrons provides information about the oxidative state, the effective charge and the chemical environment of an atom.^[3]

In contrast to that, during Ultraviolet photoelectron spectroscopy (UPS) weaker radiation is used and hence only valence electrons are analysed.^[3]

In XPS analysis, the intensity (counts) is plotted versus either the kinetic energy of electrons or their binding energy (see Figure 2). The signals correspond to orbitals which the photoelectrons were ejected from. Based on the intensity of signals the composition of a sample can be calculated. To do so, the areas of each element's signals are compared. It is worth noting that the exact binding energy of an electron varies depending on the chemical environment (Chemical shift).^[3] Hence, XPS reveals information about e.g. hybridisation, coordinating elements and coordination number. Also the oxidative state and the charge can be derived from XPS data.

Fig. 2a: XPS spectrum of a gold surface. The binding energy increases with decreasing distance between orbital and atom core. The step function is due to bremsstrahlung that ejects photoelectrons of various kinetic energy. Impurities of carbon and oxygen can be seen. A more detailed explanation of the effect of bremsstrahlung can be found lateron. Fig. 2b: High resolution of carbon impurity peak reveals several compounds to be responsible for the signal. (Figure: A. Al-Ajlony, A. Kanjilal, S. Harilal, A. Hassanein^[4])

In Fig. 2 a XPS spectrum is displayed, in which some orbitals lead to two signals, e.g. Au4d_5/2 and Au 4d_3/2. To assign a sample to certain elements, this so called "spin orbital splitting" is a very useful fingerprinting property. The energetic distance between two signals of a splitted orbital is characteristic for elements. Additionally, to identify the correct element, the ratio of peak areas of an element's orbitals is investigated. Further investigations of the analysed gold surface have shown that the carbon signal can actually be drawn back to four different impurities: C-C, C-H, C=O and COOH fragments.^[4]

Experimental performance

The sample is placed in a chamber of ultra-high vacuum in order to avoid collisions of photoelectrons with gas phase molecules.^[1] The pressure is in a range of 10^-10 to 10^-7 mbar. Commonly used X-ray radiations are Magnesium K_α and Aluminium K_α beams with energies of 1253.6 eV and 1486.6 eV, respectively. K_α means, that electrons fall from the L level to an electron hole in the K level and release during this process energy that is equal to the energy gap between those orbitals. To obtain monochromatic radiation (only one certain wavelength) sometimes also synchrotrons are used. Alternatively, a monochromator is used in which the radiation is reflected according to Bragg’s law:

nλ=2d∙sin(Θ)

The desired monochromatic radiation can be obtained in a certain angle Θ that depends on the used wavelength λ and on the distance d between atomic layers of the monochromator.

Once photoelectrons are ejected from the sample, their kinetic energy needs to be detected. To do so, an electric field can be applied that forces the electrons to change their direction, whereby electrons with low kinetic energy are stronger affected by the field. Only electrons with a certain velocity can reach the detector where they are counted, the other electrons collide with the walls. After the electrons were counted, the electric field can be changed and the ratio of electrons with another energy is counted.^[1] A summarizing scheme can be found in Figure 3.

Figure 3. Possible experimental setup to investigate binding energies via XPS. After monochromatic X-ray radiation is isolated, it is focussed on the sample and photoelectrons are ejected (displayed with red arrows). Due to an applied external electric field (yellow arrow), only photoelectrons with a certain velocity can be detected (Figure: Igor Graf von Westarp).

Problems, limitations and corrections

In theory, a spectrum would only consist of a few sharp signals that are referred to certain orbitals. If a sample was invetigated twice, the results would be identical. Unfortunately, the experiments reveal that there are several issues that have to be dealt with, before an XPS spectrum can be analysed. Some of the most common problems are listed below.

• Cross section: To eject an electron with X-ray radiation, it is not sufficient to just overcome a certain energy level (binding energy). Additionally, the energy needs to be transferred from the photon to the electron. The quantity that defines this interaction property is the cross section of an element. Due to their very small cross sections with respect to X-ray radiation, hydrogen and helium therefore cannot be detected properly with XPS.^[5]
  Also, valence electrons usually exhibit smaller cross sections than core electrons, which is why they can be detected worse with XPS than with UPS.^[1] That is because in UPS the interaction with low-energy electrons is larger than in XPS.
  
  

• Degradation: The high energetic X-ray radiation can cause the sample to degrade slightly during the experiment.^[6] Hence, results of multiple experiments of the same sample or long-term experiments should be taken with caution.
  
  

• Electron oscillation: Especially in metals, photoelectrons excite collective oscillations of nearby electrons in the conducting band. Due to losses of kinetic energy the analysed electrons appear in spectra at a larger binding energy, which they in fact did not overcome.^[1] This problem unfortunately cannot be corrected properly so far.
  
  

• Inelastic scattering: In general, photoelectrons most commonly interact with several atom layers before they are detected. Due to inelastic scattering the kinetic energy is reduced and, therefore, wrong binding energies are estimated.^[7] The energy is approximated to decrease exponentially with sample depth and hence exponential correction terms can be used to enhance the quality of spectra.^[1]
  
  

• Bremsstrahlung: Many raw data spectra show step functions, if the number of counts is plotted versus binding energy or kinetic energy. That is due to bremsstrahlung which can eject electrons, too, and their excess of energy cannot be predicted. Thus, at every signal that can be correlated to a certain energy level (orbital), the background increases for lower kinetic energies. This increase is caused by counted photoelectrons that have been ejected by bremsstrahlung.^[7] The bremsstrahlung is caused by the radiation source if no monochromator or synchrotron is used that produces radiation of only very precise wavelengths. Additionally, also in the sample X-ray radiation can be released and cause bremsstrahlung. The latter aspect cannot be avoided. To improve the resolution one needs to enlarge the duration of the experiment or, alternatively, increase the intensity of the applied X-ray radiation. Thereby the ratio of signal to background is enhanced.^[1]
  
  
• Smoothen: Usually spectra are smoothened before they get analysed. That simplifies the interpretation, especially if it is done by a computer software.^[7]

Example

XPS analysis can be used for several different applications. An example of the work of W. Jolly et al.^[8] shows how the spectrum analysis reveals the true identity of a NO-ligand.

The ionic complexes K₃[Cr(CN)₅(NO)]·H₂O and Na₂[Fe(CN)₅(NO)]·2 H₂O both were expected to have the same, positively charged NO⁺ ligand. Accordingly, the oxidative state of Fe and Cr are +II and +I, respectively. XPS investigations have shown the following binding energies of ligands:

The CN^- ligands show very similar values for their energy levels, but the energy difference of the NO ligands in the investigated complexes is rather large. Therefore, it is believed that the ligands are not identical. Instead, it is assumed that the difference in binding energies is related to the charge of the ligand and, consequently, to the charge of the central ion. Hence, it is predicted that in Na₂[Fe(CN)₅(NO)]·2 H₂O the ligand is charged positively (NO⁺), but in K₃[Cr(CN)₅(NO)]·H₂O the ligand is actually charged negatively (NO^-) and, concluding, the central metal atom is in an oxidative state of +III.

To further confirm this interpretation an additional metal complex was investigated, namely K₃[Cr(CN)₆].^[9] It turned out that the 3p binding energies of K₃[Cr(CN)₅(NO)]·H₂O were identical to those in K₃[Cr(CN)₆], which is known to be an Cr(III) complex. Summarizing, XPS analysis allowed to characterize the oxidative state in the chromium complex and the wrong charge of the NO-ligand.

In another experiment the surface composition of plasma-treated organic surface is investigated.^[10] Using XPS and additional analytical methods, the carbon atoms are analysed in order to determine the hybridization of a poly-styrene layer. It turned out that stable plasma polymer layers could be obtained using a pulse plasma process. Similar experiments have been performed with other monomers, too, namely ehtylene, acetylene and butadiens. Worse plasma deposition rates have been yielded, as well as different levels in the concentrations of unsaturated carbon species.

References