THE 3-PARTICLE LADDER APPROXIMATION AS A NEW APPROACH TOWARDS A GENERAL-THEORY OF ELECTRON-CORRELATION EFFECTS IN CVV AUGER-ELECTRON AND APPEARANCE-POTENTIAL SPECTROSCOPY

Due to their sensitivity to electron-correlation effects, CVV Auger-electron (AES) and appearance-potential spectroscopy (APS) can provide useful information on the electronic structure of solids. Correlations among the valence-band electrons (VV correlations) as well as correlations between the valence-band and the core electrons (CV correlations) are responsible for a variety of effects. Strong VV correlations are well known to give rise to sharp satellites in the spectra, which are related to localized two-hole (electron) final states. On the other hand, the screening of the core-hole potential in the initial state for AES, the sudden response of the valence-band electrons after the destruction of the core hole, and, for APS, the scattering of the valence-band electrons at the core hole are all consequences of CV correlations. Up to now, however, little is known about the combined influence of both types of correlations on the spectra. We present a new theoretical approach that refers to the general case of a model system with arbitrary band-filling and arbitrary strengths of VV as well as CV correlations. Remaining restrictions and simplifications concerning the degeneracy of the valence band, the transition matrix elements, etc. can be improved systematically. Of course, this generality can only be achieved at the expense of inevitable approximations in the theoretical formulation. The AES and APS intensities are given by properly defined three-particle Green functions, which are determined by use of a diagrammatic vertex-correction method that is based on the three-particle ladder approximation, which is the main idea of our approach. It is a direct generalization of the two-particle ladder approximation, which in the past has been applied for the calculation of two-particle Green functions that are related to the AES and APS intensities, if CV correlations can be neglected.