Single crystal antiferromagnetic USb2 was studied at 15K by angle-resolved photoemission with an overall energy resolution of 24 meV. The measurements unambiguously show the dispersion of extremely narrow bands situated near the Fermi level. The peak at th
Fermi edge remains the same with respect to the other valence band features and one observes the predictable exponential decay of the entire valence band with PES mean free path as Ar is covering the surface.
Normal emission spectra presented in Fig. 6 provide additional evidence that the near E F peak derives from bulk crystal states. The data were taken for photon energies ranging from 17.54 eV to 34 eV. A dispersion of around 10 meV of the sharp near-E F peak may be seen. Dispersion perpendicular to the surface is evidence that the very sharp peak near the Fermi edge is not a surface state. This also means that USb2 is not of purely 2D electronic structure but has a 3D character which couples weakly to the in-plane features, and thus requires treatment as a 3D material in reciprocal space. This is in contrast to the Fermi surface proposed in [27-28] which is decidedly 2D.
Measurements near the U5d → U5f absorption edge (Fig.5) were done at 15K. The main resonance in uranium compounds is split into two because of the large spin-orbit splitting of the U5d shell. Consequently, there is no clean anti-resonance in uranium compounds, but just a minimum of the resonances from each of the cores. The PES spectra taken near the maximum of the resonance (hν=108 eV) shows the U5f enhancement mainly in the binding energy range between 300 meV and the Fermi edge. The spectra measured near the valley between the 3/2 and 5/2 resonances (hν = 102 eV) shows the main photoemission structure to be between 300 meV and 500 meV, but we can also notice a smaller structure near the Fermi edge. These results show that the B structure, which is also observed in PES spectra taken at lower photon energies, consists of hybridized 5f and conduction band electrons. The peak A is primarily of 5f origin but contains a non negligible contribution from the conduction electrons.
Resonant photoemission measurements confirm the conclusion about the 5f hybridization. The small intensity difference between on- and off-resonance spectra is indicative of 5f-conduction band hybridization, but also that the valley at 102 eV is not a true antiresonance, but just a minimum between the two main resonances at 108 eV and 98 eV. Photoemission spectra and calculations for uranium compounds [29] show larger
9