Gunnarsson O., Schonhammer K.
The Anderson model at zero temperature is studied as a function of the f-level position epsilon f and the f level-conduction electron hopping matrix element V. The f-f Coulomb interaction U is assumed to be finite, and double occupancy of the f level is taken into account. For a large value of the f-level degeneracy Nf, there is an important asymmetry between f0 and f2 configurations. Even for 'symmetric' parameters, 2 epsilon f+U=2 epsilon F=0, the f2 weight is much larger than the f0 weight if V is small. The effect of this asymmetry on other properties is studied for Nf to infinity . The static susceptibility is primarily determined by the f0 weight, while the shape of the valence photoemission spectrum close to the Fermi energy epsilon F also has an important dependence on the f2 weight. The valence photoemission spectrum can have a pronounced two-peak character, with one peak close to epsilon f and a second structure close to epsilon F. For epsilon f well below epsilon F ('spin-fluctuation' limit) the weight of the second structure can be strongly enhanced compared to the U= infinity limit, and its shape and position depends on the conduction density of states. This structure can therefore have a peak below epsilon F. The bremsstrahlung isochromat spectroscopy spectrum shows an f1 peak with an energy separation from epsilon F which is determined by the 'Kondo' temperature. The tail of this peak contributes to the structure in the valence photoemission spectrum below epsilon F. Ground-state properties are calculated variationally, treating 1/Nf as a small parameter. A new technique for performing these calculations is developed. This technique makes it possible to include such a large basis set that accurate results are obtained for the ground-state energy and the f-level occupancy in the limit Nf=1. To calculate the spectra the authors introduce a time-dependent method which facilitates the inclusion of f2 configurations in the valence photoemission spectrum.
Max-Planck Institut für Festkörperforschung; Postfach 80 06 65 D-70506 Stuttgart |