Chris Rodgers, St Johns College, Oxford

Easter 2002

Essays in Inorganic Chemistry

Photoelectron Spectroscopy
2000/Q13 A: Give an account of the application of photoelectron spectroscopy (PES) in characterising the electronic structure of inorganic compounds in the gas and solid phase. Include in your coverage a discussion of the utility of varying the excitation photon energy in UV-PES.

Outline
Intro - explain photoelectric effect. - Koopmans theorem. - Selection rules.

Experimental apparatus. UV-PES - look at valence orbitals. - use for gases and solids. - fine structure effects. - Ne isoelectronic series. - d-orbital character XPES mainly used for studies of core electron studies. ESCA. Talk about chem. Shift. Intervalence cpds. Mention associated methods of Auger and BIS

Introduction
Photoelectron spectroscopy (PES) is conceptually a very simple and direct way of looking at electron binding energies in a substance. The process at the heart of the experiment is the photoemission (light-induced emission) of an electron from the substance in question. This photoelectric effect was first explained by Einstein in his famous 1905 paper (which was, incidentally, one of the landmark pieces of evidence for the quantization of energy and for the emergent quantum theory). Einstein formulated the process as follows. Xi + h X+j + e(a one-photon, one-electron process.)

A species, X, in state i (initial) absorbs exactly one photon of energy h. It then loses this excess of energy by releasing one of its electrons, with a certain kinetic energy (Ek) and leaving the species as a positive ion, in a (possibly excited) state j. If we apply the law of the conservation of energy, we see that: Energy in = h = Energy out = Iij + Ek , where Iij is the amount of energy required to just convert Xi into X+j. So, using this relation we can perform a spectroscopy by shining a light of some frequency, , onto our sample and then studying the kinetic energy of the electrons released. Alternatively, one could vary the frequency of the incident light and see what frequency is required to produce electrons of a specific energy. We then build up a spectrum showing the different Iij values for our substance. Consider a boron atom. It has an energy level diagram like this:

Page 1 or 6.

Chris Rodgers, St Johns College, Oxford

1 2 3

Easter 2002

Looking at this simple diagram, we would expect there to be three possible peaks in the photoelectron spectrum. That is, a peak corresponding to ionisation by process 1, 2 or 3, each of which gives a different X+j state and has a different binding energy. A very useful approximation is Koopmans approximation (or theorem). This states that: Iij = -j where the initial state is the electronic ground state, and where j is the orbital energy for the orbital from which we ionise the electron (orbital j). This amounts to saying that the rest of the orbitals are frozen, so that they do not react to removing an electron from another orbital. Obviously, this is not really the case, but in practice, the approximation is often sufficiently accurate to allow us to assign photoelectron spectra on the basis of orbital energy level diagrams. In most cases, the remaining orbitals will contract on ionisation. This gives increased electron repulsion, but it means that the exchange stabilisation energy also increases. These effects largely cancel one-another. Note that this expression only includes the final state. Assuming we start with our species in the ground state (which, considering a Boltzmann energy distribution, will be the case) then the nature of the photoelectron spectrum we produce depends entirely upon the final state energy levels. Koopmans approximation is saying that the spacing of these final state energy levels is not much different to that in the ground state. However, any fine structure which develops in the spectrum will arise from splitting of the final state energy levels. Finally, we should note that there are no selection rules in PES. However, given that this is a one-electron, one-photon process we expect that the only excited states which will normally be produced are those arising from the removal of one electron from one of the orbitals in the ground state. That is, we do not usually expect to see states where there are excitations of other electrons at the same time as the photoelectron leaves. (However, these energy-sharing processes do occur and give rise to shake-up peaks and other distortions. Their low probability means that they may be dealt with as an aspect of fine structure. In the solid state, however, phonon excitations are widespread and distort the shape of the PE spectra obtained.)

Experimental
The essential apparatus is as follows: PES instruments consist of an X-ray or UV source, an energy analyzer for the photoelectrons, and an electron detector. The analysis and detection of photoelectrons requires that the sample be placed in a high-vacuum chamber. This has several effects: - It removes adsorbed gases from the sample. - It prevents contaminants (e.g. pump grease) from adsorbing onto the sample. - It prevents electrical arcing and / or high voltage breakdown. - It increases the mean free path for electrons, ions and photons. Since the photoelectron energy depends on the incident radiations energy, the excitation source must be monochromatic. The energy of the photoelectrons is analyzed by an electrostatic analyzer, and the photoelectrons are detected by an electron multiplier tube or a multichannel detector such as a microchannel plate. When applied to solids, PES is a surface technique. This is because, although the incident radiation may penetrate quite far into the bulk of the sample, the photoelectrons which emerge have a mean free path of only a few . Hence, any photoelectrons Page 2 or 6.

Energy 0 2p 2s 1s

Chris Rodgers, St Johns College, Oxford Easter 2002 emitted from atoms further into the sample than this will probably suffer inelastic (energy-losing) collisions on the way out of the sample. Hence, the signal from these photoelectrons will be degraded very rapidly with increasing depth. Computer

Hemispherical Energy Analyzer

Outer Sphere Magnetic Shield

System

Inner Sphere

Analyzer Control

Electron Optics X-ray Source

Lenses for Energy Adjustment (Retardation)

Multi-Channel Plate Electron Multiplier

Resistive Anode
Encoder
Position Computer

Lenses for Analysis Area Definition

Sampl e

Position Sensitive Detector (PSD)

5 4 . 7

Position Address Converter

PES was developed in two forms (X-PES and UV-PES) by independent workers. The main difference is in the source of radiation used to illuminate the sample.

UV-PES
In UV-PES, the sample is irradiated with light in the vacuum ultraviolet region of the spectrum. Typically, this light is from a gas-discharge tube inside the apparatus and the energies involved are: He I: 21.22eV He II: 40.81eV Ne I: 16.7eV (doublet) Ne II: 26.9eV (doublet)

These energies are of the same order of magnitude as the binding energies for valence shell electrons, but are too low to ionise from the core orbitals (e.g. 1s in the diagram above). This means that UV-PES gives us information about only the valence orbitals in the sample. However, despite the energy limitation, UV-PES has strong advantages in many cases. The probability of photoemission varies with frequency. This can be characterized by the absorption cross-section for photoemission. It turns out that this varies approximately as -3.5. Hence, excitation in the UV region is much easier. In addition, the flux of photons produced by these discharge lamps is very intense in comparison to the X-ray sources in X-PES. These factors mean that UVPES is more sensitive. It is therefore suited to working with dilute samples, such as gases (using differentially-pumped apparatus to obtain the low pressures required to make PES work, whilst still allowing there to be sufficient gas in the sample chamber) or to adsorbates on solid surfaces. UV-PES is also advantageous when working with delicate samples, as there is less chance of radiation damage. Finally, using UV-PES and varying the irradiating light allows one to see the frequencydependence of the various PES peaks. This frequency dependence is particularly important in solid-state work. It can be shown on a quantum mechanical basis that the photoemission cross-section depends on the size, number of (angular) nodes and localization of the orbital in question. This means that by repeating the measurement of the PE spectrum with many different exciting radiation energies, one can identify the d-orbital character of the orbitals involved. This is useful for assigning bands in solid-state work. It is also an important test of the bonding model which is being used to explain the band structure of the solid in question. It is found that the photoemission cross-section for 3d or 4d electrons increases as the exciting energy rises above the threshold value, while that for other orbitals falls off. The series of spectra to the left are from a lutetium compound and show the variations of photoemission cross-section with energy being used to identify the orbitals involved. Page 3 or 6.

Chris Rodgers, St Johns College, Oxford

Easter 2002

UV-PES of solids is also useful for distinguishing between localized and band-structured solids. So, one can tell a localized solid e.g. Li2SO4 apart from a delocalised solid e.g. metallic Hg. Again, this presents a test for metallic bonding models.

Some simple examples of UV-PES are provided in the problems. For example, it is instructive to look at the spectrum of Ne. This shows two sets of peaks, those from the 2s and 2p levels. We can then compare Ne with a series of isolectronic hydrides formed by partitioning off one proton at a time from the Ne nucleus. Thus we form the series, Ne, FH, OH2, NH3, CH4. The spectra we see agree very well with a molecular orbital description of these molecules. (See 1999/Aii for more details.) This is one of the most important results of PES that it confirms the validity of the molecular orbital description for electrons in molecules, and it does not agree with the alternative valence bond, etc theories.

Fine structure
Fine structure is obviously very important, especially because it allows us to assign the peaks in a PE spectrum to particular energy levels and because it gives us more information about the species when its in these levels. There are several causes of fine structure in PES. 1. Vibrational fine structure.

As in other forms of electronic spectroscopy, we see vibrational fine-structure within the electronic energy levels. The Frank-Condon principle says that electronic transitions must not involve any change in the position of the atoms. This is because an electronic transition is much faster than the motion of the nuclei, so they are effectively frozen in place during the transition. Therefore, the excited molecule may well be formed with the nuclei in positions away from their new equilibrium positions. If this is the case, then a quantum mechanical description is to say that we have excited some quanta of a vibration in the system. The probabilities of transition to the different vibrational levels within a given electronic state are given by the Frank-Condon factors that is by the overlap integrals between the vibrational wave function in the ground and excited states. This means that when the equilibrium geometry changes little during the transition, the v=0 peak is expected to dominate and a very short vibrational progression will be seen. However, when there is a substantial change in geometry, there will be a long vibrational progression. We expect that ionisation from a non-bonding orbital will not change the equilibrium geometry much so it should give a short vibrational progression. Whereas ionisation from a strongly bonding or anti-bonding orbital will change it substantially, so there will be a long vibrational progression. 2. Spin-orbit splitting. The spin and orbital angular momenta in a species will interact via a magnetic spin-orbit coupling interaction. This takes the form: = L.S . Hence, when we have states with orbital angular momentum, they will be split into various components depending on the relative orientation of the spin and orbital angular momentum vectors. This can be seen in the spectra of the halogen acids HF, HCl, HBr, HI where the spin-orbit splitting gets larger as the atomic number increases. (This is because rises in this way.) Shake-up and shake-off peaks. It was stated above that the photoelectric effect only involves changes to one photon and one electron. However, this is not necessarily true. It is possible that as the photoelectron is being ejected, it may give up some of its energy to excite another electron in the molecule either to a higher bound state (shakeup) or to ionise it (shake-off). This effect gives rise to a series of peaks to high binding energy of the main peak. These peaks make reliable interpretation of the spectra more difficult. Jahn-Teller (non-linear molecules) and Renner-Teller (linear molecules) effects. The Jahn-Teller theorem states that "any non-linear molecular system in a degenerate electronic state will be unstable and will undergo distortion to form a system of lower symmetry and lower energy thereby removing the degeneracy", while the related Renner-Teller theorem applies the same sort of result to linear molecules. This lifting of degeneracy occurs via the excitation of a vibrational mode which acts to lower the symmetry of the molecule at an instant in time, although the average Page 4 or 6.

3.

4.

Chris Rodgers, St Johns College, Oxford Easter 2002 position of the molecule is still the same symmetry. Hence, a single peak may be given complicated fine structure. A good example of this is the t2 peak in methane, which has a very complicated J-T distorted fine structure. 5. Inversion doubling. This occurs in molecules like NH3 or OH2 which have inversion symmetry through some barrier. In the vibrational progression of the vibration which causes the molecule to invert, there will be a sudden halving of the spacing of the peaks as the barrier is crossed. This is due to the molecule escaping from the local potential well around one of the two forms and being free to cross the inversion barrier. Phonon excitations. While a photoelectron is being ejected from a solid sample, it is also possible for the photoelectron to give up some of its energy to the lattice, exciting a phonon there. These phonon loss peaks tend to be very broad (e.g. 10eV wide) and are common in solids. They occur to high binding energy of the actual peak.

6.

X-PES (ESCA)
The other main technique is X-PES, also known as ESCA (Electron Spectroscopy for Chemical Analysis). In this case, the sample is irradiated with soft (low energy) X-rays. These are generated either by electron bombardment of a metallic target, e.g. Mg K: 1253.6eV Al K: 1486.6eV Y M: 132.3eV Cu K: 8047.8eV

or by using the synchrotron radiation from an electron storage ring in a particle accelerator. Synchrotron radiation has the advantage that it is variable frequency, that it is possible to make it more monochromatic and that it more intense than the Xrays produced by electron bombardment. These energies are well into the core-ionisation region, so that X-PES is mainly looking at states produced by ionising the core electrons in the sample, although it can also be used to study ionisation from valence bands. However, this is less attractive than using UV-PES in these cases because the photoemission cross-sections are much lower in the X-ray region, and the available resolution is significantly lower in X-PES than in UV-PES.

Chemical Analysis (ESCA)

It is possible to record a broad spectrum covering a wide range of ionisation energies using an X-ray source. It is found that the core ionisation energies depend very little on the environment of an atom, but are instead rather characteristic of the element in question. This suggests that X-PES may be used to determine the composition of a solid sample (i.e. For chemical analysis). Thus, X-PES was used to analyse samples of moon rock returned to Earth by the Apollo project. In more down-to-earth applications, it is used to fingerprint the surface sites of catalysts in order to determine how active a given sample will be. This allows a faulty batch of catalyst to be rejected without wasting substrate trying to do the reaction. It was also found, when studying a solution of sodium thiosulphate, that there is a small chemical shift effect in the X-PES spectra. This depends largely on the electron density available at the particular atom in question, and is therefore governed by such factors as oxidation state and inductive effects from neighbouring groups. The chemical shift values show good correlations with kinetic parameters such as Markovnikow addition rates to alkenes and Taft parameters. Thus X-PES allows a nice overlap between a clearly physical MO-type of experiment and the kinetic parameters often used in organic chemistry.. Unlike NMR, X-PES can study any element with core electrons. That is, any element after lithium. This means that it can sometimes be the only available technique. These chemical shift effects can be used productively to study mixed valence compounds, such as Pb3O4 which is actually 2PbO.PbO 2 or Creutz-Taube compounds ([(NH3)5Ru(pyrazine)Ru(NH3)5]5+) , to name but a few. Unfortunately, there are some serious disadvantages to X-PES when used for analysis. One of these is that X-PES is a surface technique. Thus, whilst it is well suited to applications where were only interested in the surface composition, it must be used with great caution for bulk analysis. We need to prepare the sample very carefully so as to have a representative portion on the surface. Ideally, this means that it must be cleaved under the vacuum immediately prior to the experiment (and the surface may still need cleaning). Furthermore, the relatively low resolution and the presence of artefacts such as shake-up and Auger peaks means that great care must be exercised when concluding how many inequivalent environments are represented. In general, NMR represents a much more sensitive and versatile technique for studying mixed valence systems. X-PES also suffers problems in comparing results between different substances. This is because with insulating samples, a surface positive charge can gradually build up. This increases the work which must be done by an escaping electron and effectively alters the energy zero. Great care must be taken to ensure that this does not cause calibration errors. Finally, in terms of its use in analysis, X-PES is dogged by the problem of accurately relating peak heights to substance concentrations. This arises because of the difficulties in calculating photoemission cross-sections and in determining the sizes of extrinsic and intrinsic plasmon loss peaks. Hence, the established techniques for analysis are still very much in use! (e.g. AAS, X-ray fluorescence, neutron activation analysis)

Page 5 or 6.

Chris Rodgers, St Johns College, Oxford

Easter 2002

Related techniques
There is a wealth of information that can be gained by various related spectroscopies. For example: Auger Electron Spectroscopy: this can be used very successfully for surface analysis. The Auger electron is ejected during the decay of the core hole formed initially in X-PES (or in practice by electron bombardment). The electron beam used can be tightly focussed so that a very small area of the surface can be analysed at a time.

Bremsstrahlung isochromat spectroscopy (BIS): this is the opposite of photoelectron spectroscopy. A beam of electrons (of a single, well-defined KE) is directed at the sample. The sample is monitored for emitted EM radiation at some particular frequency while the electron KE is swept through a range of values. BIS allows the unoccupied states in a system to be mapped out in a way which is complementary to PES mapping of the occupied states. EXAFS: this studies interference effects on the edge of the X-ray absorption peaks. It allows the atoms around some particular element to be located, even in the presence of much other disordered bulk. For example, EXAFS is capable of resolving the atoms around the active site in a protein, even when a crystal cannot be grown. This is obviously very useful.

Page 6 or 6.