X-ray photoelectron spectroscopy (XPS) performed with soft X-ray laboratory sources is well known under the name ESCA (electron spectroscopy for chemical analysis). At synchrotron radiation sources, XPS is also well established, using harder X-rays and detecting electrons with kinetic energies up to several keV. Now, at beamline ID32, we have extended photoelectron spectroscopy into the 10 keV range.

The advantages to using high-energy XPS are many: an almost complete elimination of surface effects when studying material properties; the availability of monochromatic X-rays with 10 - 50 meV resolution; the possibility of combining photoemission with X-ray diffraction and standing waves; the accessibility of buried interfaces to photoemission studies. Of special interest is the study of valence band states and core states with low binding energy, since they carry information on the chemical and electronic properties of the material. Photoelectrons interact strongly with the atoms of the material from which they are produced, rapidly losing their characteristic kinetic energy. Thus, if a material is investigated by "normal" XPS, mostly its surface, rather than its bulk properties, is revealed. High kinetic energy electrons interact less strongly, hence travel further before being scattered, and as a consequence they carry with them information about chemical and electronic properties of the bulk.

The fundamental challenge at these high photon energies is that the absorption of the X-rays by these valence band and shallow core states decreases at a far faster rate than signal is gained by detecting more material from deeper in the sample. Our knowledge about cross sections and the escape depths of high-energy photoelectrons is exclusively based on theoretical calculations [1], which need to be tested experimentally.

 

Fig. 160: Mounting of the pre-lens on the commercial PHI analyser.

 

Our approach was pragmatic since we wanted first to establish sublevel cross sections and gain experimental experience before embarking on the purchase or development of new instrumentation. So we extended the kinetic energy range of our existing electron analyser and built a refocusing lens, which allows the retardation of electrons emerging from a sample biased at up to +10 kV relative to ground (Figure 160). The electrons are handed over at ground potential to the "Omega" retarding lens (0 to -4.5 keV) of a Perkin Elmer PHI Model 10-360 hemispherical analyser equipped with a 16-channel detection system. The pre-retarding lens was modelled with the SIMION code.

 

Fig. 161: The faster decrease of the cross section for the 4f than for the other levels with increasing photon energy is also predicted by theory.

 

Experiments were carried out at beamline ID32, using X-rays from a Si(111) monochromator (10-4 resolution) and a Si(444) post-monochromator (40 meV resolution at 8 keV) for some measurements. We performed a variety of measurements on Au, graphite and YBa2Cu3O7, a 90 K superconductor, up to 14.5 keV (Figure 161). We succeeded in measuring the Au 5d, 5p, 4f and 5s cross-sections between 5 and 13.5 keV on an absolute scale comparing them to the established cross section of C1s (Figure 162). A preliminary evaluation gives cross sections two to ten times higher than predicted by theory [1]. The count rates will allow valence band investigations in further detail in this energy range.

 

Fig. 162: Overview of the full energy range of the Au levels from 2p to the valence band in one spectrum.

 

The best resolution obtained was limited to 250 meV at the Au Fermi edge at room temperature, determined by the noise and stability of our power supplies. However, given the inherent simplicity of the setup, we consider the resolving power of 32000 remarkable. We know that our initial experimental situation is not ideal and we can envisage a few simple technical improvements; such as focussing of the X-ray beam onto the sample and power supply improvements which would increase our electron energy resolution. This would open up the possibilities for new experiments in bulk systems, interfaces and material science.

References
[1] J.H. Scofield, Lawrence Livermore Report UCRL-51326 (1973).

Authors
S. Thiess (a), B. Cowie (a), C. Kunz (b), T.-L. Lee (a), M. Renier (a) and J. Zegenhagen (a).
(a) ESRF
(b) Universität Hamburg (Germany)