Angle-resolved photoemission measurements of divalent hexaborides reveals a >1 eV X-point gap between the valence and conduction bands, in contradiction to the band overlap assumed in several models of their novel ferromagnetism. While the global ARPES band structure and gap size observed are consistent with the results of bulk-sensitive soft x-ray absorption and emission boron K-edge spectroscopy, the surface-sensitive photoemission measurements also show a variation with cation, surface and time of the position of the surface chemical potential in the band structure.KEYWORDS: hexaboride, CaB 6 , EuB 6 , angle resolve photemission §1. Introduction Great interest in the divalent hexaborides has been generated recently by the discovery of high Curie temperature weak-moment ferromagnetism (FM) in La-doped CaB 6 1) and by exotic theoretical models to explain the unusual magnetism, e.g. that it represents the ground state of a dilute electron gas 2) or of a doped excitonic insulator.3) The starting point of most thinking about the divalent hexaborides, and central to the excitonic instability model, is the presumed existence of a band overlap between the top of the boron valence states and the bottom of the cation d-conduction band at the Xpoint of the simple cubic Brillouin zone appropriate to these materials. Without such overlap stoichiometric divalent hexaborides would be insulators. Band overlap is predicted by band structure calculations 4, 5) and de Haas van Alphen (dHvA) and Shubnikov de Hass (ShD) data 6) have been interpreted in this semi-metal framework.Our early angle-resolved photoemission spectroscopy (ARPES) studies of EuB 6 and SrB 6 synchrotron measurements predating the FM discovery, showed, contrary to the band-overlap picture, an isolated X-point electron pocket separated from the X-point boron valence band maximum by a gap >1 eV. However, motivated by the prior theory results, by apparently firm evidence for bulk band overlap from dHvA and SdH data discussed below and by certain surface sensitive aspects of our ARPES data, we interpreted 7) the observed gap as a property only of the surface region probed in ARPES. It was reported essentially simultaneously 8) that a new band calculation including a GW self energy predicts CaB 6 to have an X-point bandgap of 0.8 eV, similar to what we measured, thereby motivating us to reexamine our early ARPES results as showing a bulk property.
9)We have recently presented new ARPES measurements and complementary bulk-sensitive soft x-ray emission and absorption 9, 10) that demonstrate conclusively the existence of a bulk band gap at the X-point and a global band structure in good agreement with the GW band calculation. This experimental result rules out all mechanisms for FM requiring an X-point band overlap and also forces the consideration of boron vacancies as the origin of the conduction band carriers observed in nominally stoichiometric divalent material, and possibly 11) also the origin of the magnetic moments. It is important that ARPES data showing onl...