4f heavy fermion photoelectron spectra do not exhibit the Kondo scale

Abstract

It has been the authors contention for some time that the Single Impurity Anderson Model (SIAM), as extended by Gunnarsson and Schonhammer (GS), or the non-crossing approximation (NCA), does not correctly describe the 4f photoelectron spectra of heavy fermions. Recently, they have concentrated on Yb heavy fermions since in these materials the Kondo resonance (KR) is fully occupied and thus accessible via photoemission. In particular, they have repeatedly pointed out that the width, position, spectral weight, lineshape, and temperature dependence of the features assumed to be the KR and its sidebands, are nearly independent of the Kondo temperature, T{sub K}, while at the same time bearing a striking resemblance to the simple 4f core level spectra of pure Yb metal, or of Lu in isostructural Lu compounds. It is important to resolve these issues in view of the fundamental nature of the problem. Here, the authors chose to test the bulk vs. surface hypothesis by performing measurements on YbCu{sub 2}Si{sub 2} and YbAl{sub 3} single crystals at hv {approx} 120 eV (UPS) and hv {approx} 1,500 eV(XPS) to see if the n{sub f}, hole occupancy, values increase markedly at XPS energies as the electron escape depth increases by about a factor of 3--5. Measurements were performed at both 300K and 20K using single crystals cleaved in-situ, with photoelectrons collected in normal emission for maximum bulk sensitivity. UPS measurements were performed at NSLS and the University of Wisconsin SRC, while XPS measurements were done at the University of Minnesota. The UPS, ultraviolet photoelectron spectra, and the L{sub III} edge x-ray absorption and photoemission measurements are in fundamental disagreement.