Strongly Correlated f-Electron Systems: A PES Study

Abstract

The term heavy fermions refers to materials (thus far only compounds with elements having an unfilled 4f or 5f shells) whose large specific heat {gamma}-values suggest that the conduction electrons at low temperatures have a very heavy effective mass. Magnetic susceptibility measurements, {chi}, generally yield a Curie-Weiss behavior at high temperatures with a well developed moment, which would be consistent with localized behavior of the f-electrons. Thus, the f-electrons appear to behave as non-interacting single impurities at elevated temperature. Below a characteristic Kondo temperature, T{sub K}, the susceptibility levels off or even decreases. This is interpreted as a compensation of the f-moment by the ligand conduction electrons that are believed to align anti-parallel to form a singlet state and has led to the widespread use of the Anderson Impurity Hamiltonian and the Single Impurity Model (SIM). Weak hybridization with these conduction electrons yields a narrow, highly temperature dependent, DOS at the Fermi energy, often referred to as the Kondo resonance (KR). At still lower temperatures it is generally agreed that in stoichiometric compounds a lattice of these singlet states finally results in extremely narrow bands at the Fermi energy, whose bandwidth is of the order k{sub B}T{sub K}. Clearly coherentmore » bands cannot form above T{sub K} owing to the narrow width. A model for periodic Kondo systems will inevitably have to include the lattice. Preliminary PAM calculations indicate that this inclusion yields results differing qualitatively, rather than just quantitatively, from the SIM predictions. The photoemission data on single crystal heavy fermions are consistent with the following PAM predictions: (1) the temperature dependence of the KR is much slower than expected from the SIM; indeed, it is primarily7 due to broadening and Fermi function truncation; (2) the spectral weight of the KR relative to the localized 4f feature (not discussed here) is much larger than the SIM expectations (equivalently, n{sub f} values are far too small); (3) the KR and its sidebands does not lose spectral weight with T, but rather only broadens; (4) f-electrons in both Ce and U systems form narrow bands already far above T{sub K} (the jury is still out for Yb systems); (5) the width of these bands is much larger than k{sub B}T{sub K}; (6) f-character is obtained in only some regions of the Brillouin zone; i.e., momentum dependence of the KR above T{sub K}. While the PAM seems to predict the correct trends, they have no reason yet to rule out other models, such as those of Liu and Sheng and Cooper. Such discrimination may occur when the models develop sufficiently to allow real system calculations.« less