X–ray photoelectron spectroscopy (XPS)(also known as electron spectroscopy for chemical analysis –ESCA) is a technique mainly used for the quantitative measure- ment of the composition and chemical state of the surface region of a sample. XPS involves directing collimated, monochromatic x–rays at a sample to excite photo- electrons, with kinetic energy Ek, from the filled core–level states of the atoms. A
proportion of the subsequent photoelectrons emitted from the surface are collected by a hemispherical or cylindrical analyser.
Each element has its own electronic configuration – hence characteristic binding energies Eb of the electrons, which enables them to be identified. The number of
photoelectrons detected is proportionate to the number of atoms of that species in the sample area probed, enabling calculation of the chemical composition. Following Einstein’s relationship for the photo–electric effect:
Eb =hν−φ−Ek (8)
one can see the binding energy (Eb) of the detected photoelectron can be deduced
if the work function (φ– the energy required for a photoelectron to escape from the Fermi level to the vacuum) and initial x–ray photon energy (hν) is known. Therefore, the photoelectron energy spectrum shows the density of occupied states shifted by
hν. This equation applies to a single electron in orbit of an atomic nucleus. In reality the calculation of Ek requires the inclusion of many–body effects from the
other bound electrons and conduction electrons in the solid.
The number of electrons detected – the intensity of the spectral line – can be used to calculate the chemical composition in the sample area probed. Spectral lines from Auger emission and final state effects (shake–up, shake–off peaks and plasmon losses) are also seen. Spin–orbit splitting is observed forp,d and f core levels ( L–S coupling, j =l+s). The ratio of the electrons in each split level is almost element independent enabling easier identification of lines with the same orbital momentum
quantum number.
Bonding to surface atoms and different chemical species in the surface can be determined from the chemical shift of peaks. For example, the removal of an electron to form a bond leaves the other electrons in a more positive potential (there is less core-screening and hence more repulsion). As a result the core–binding energies increase on the order of ∼eV.
Typically, the photoelectrons detected have escaped into the vacuum from the first 10 – 20nm of the sample. The x-rays penetrate a greater depth into the sample (on the order of µms) but the resulting photoelectrons are attenuated by inelastic interactions in the material and recaptured. The surface signal to bulk signal can be improved by using grazing incidence between the surface and analyser. The detected photoelectrons at this angle will have travelled a greater lateral distance through the sample on average therefore decreasing the signal from deeper layers of the material.
Higher resolution and more bulk sensitivity can be gained by using synchrotron radiation, this technique is termed hard x–ray photoelectron spectroscopy (HAX- PES). A standard laboratory x–ray source is the dual anode configuration. Electrons produced by a hot filament are attracted towards a water cooled, metallic coated target anode by a high voltage. The high energy electrons collide with anode atoms exciting core–level electrons. Subsequently electrons from less bound outer elec- tronic levels relax to fill the empty core–level state emitting x–rays. The two most common soft x–ray source anodes are Al and Mg. Both metals have dominant Kα
emission doublet line (FWHM∼0.7 – 0.8 eV) [127]. There are other less intense core emission lines on a continuous Bremstrahlung radiation background. Such an un- monochromated x–ray source leads to satellite peaks in the photoemission spectrum from a surface. These peaks may overlap with other emission peaks thus inhibiting peak identification and quantification. An Al Kα ( hν = 1486.6 eV) x–ray beam
can be made more monochromatic by being diffracted from the (1010) planes of a quartz crystal.
Hemispherical analysers
Figure 28 shows a basic schematic of a concentric hemispherical analyser which internally consists of two concentric stainless steel hemisphere electrodes. Electro- static transfer lenses are placed before the analyser to reduce the angular spread of the incident electrons. The voltage of the electrostatic transfer lenses is varied
so that only electrons with a certain kinetic energy pass into the analyser. The constant potential applied between the electrodes acts on the charged particles en- tering the analyser through the entrance slit bending their projectile path towards the detector. The hemisphere shape of the analyser ensures the beam is focused both parallel and perpendicular to the exit slit. Jost correctors are also placed in the entrance and exit planes to reduce the field distortion there.
The absolute energy resolution ∆E is given by:
∆E ≈ w 4dV2 i L2V c (9)
where w is the exit slit width, d is the distance between the electrode plates, Vi is
the potential related to the pass energy of the electrons, L is the path length of the electrons and Vc is the potential between the plates. This means that the larger the
electrons’ path (i.e. size of the analyser) relative to the entrance and exit slit size, the greater the resolution.
The electron detector is usually an electron channel multiplier. This increases the amount of detected electrons; initial electrons collide with the walls of the multiplier to produce cascades of secondary electrons. Hundreds of channel multipliers are placed in an array to make a channel plate, which can produce gains in the detected electron signal up to 104.
Photoemission peaks
The spectral lines from photoemission are broadened due to several factors which can be described by a sum of Gaussian and Lorentzian formulae (SGL). The relative resolution of the electron analyser, phonon broadening, the energy width of the initial x–ray used to excite the photoelectrons and temperature dependent effects contribute to a Gaussian peak shape. While the effects of the life–time broadening of the core–level hole state are best represented by a Lorentzian distribution. The choice of background correction algorithm used also influences the shape of the peak to be fitted. The equation fitted to peaks to quantify their area is given by:
SGL(x, F, E, m) = (1−m)exp(−4ln2(x−E) 2
F2 ) +m(1 + 4
(x−E)2
F2 ) (10)
Figure 28: Schematic cross–section of a concentric hemispherical analyser for the detection of analysis electrons emitted from a sample.
determines the intermixing, so that m=1 is a pure Lorentzian and m=0 is a pure Gaussian curve.
Quantification assumes that the surface concentration of a particular atom is directly proportional to the peak intensity and that there is a homogeneous distri- bution of atoms in the surface. Photoemission peaks from metals have an asymme- try on the higher binding energy side that must be accounted for. This is due to final state effects where the core-hole is screened allowing small excitations above the Fermi energy in the continuum of conduction electrons. A Doniach Sunjic line shape can be fitted to account for this asymmetry. The asymmetry index in this
formula is given by:
α= 1− HW HMr
HW HMl
(11)
where the half at width-half maximum (HWHM) on the left (l) and right (r) are measured. The CasaXPS software is used to fit the components to quantify peaks [128].