Thermodynamics of FeSi

We show that a simple model involving two narrow (\ensuremath{\simeq}500 K) peaks in the density of states (DOS) at the edges of a narrow (\ensuremath{\simeq}1000 K) gap can account for the observed anomalies in the magnetic susceptibility, specific heat, thermal expansion, and elastic response of FeSi. We also show that the resistivity of FeSi, including the metal-insulator ``transition'' at about 300 K, is well described by this model. Recent band-structure calculations, although predicting the correct value for the semiconducting gap, are unable to account for the narrow peaks in the DOS needed to explain the thermodynamics. We conclude that, given the available alternatives, a Kondo insulator description---involving an extreme renormalization of the noninteracting bands---is most appropriate for FeSi.