Fermi Velocity Spectrum and Incipient Magnetism in TiBe2

Abstract

We address the origin of the incipient magnetism in TiBe$_2$ through precise first principles calculations, which overestimate the ferromagnetic tendency and therefore require correction to account for spin fluctuations. TiBe$_2$ has sharp fine structure in its electronic density of states, with a van Hove singularity only 3 meV above the Fermi level. Similarly to the isovalent weak ferromagnet ZrZn$_2$, it is flat bands along the K-W-U lines of hexagonal face of the fcc Brillouin zone make the system prone to magnetism, and more so if electrons are added. We find that the Moriya $B$ coefficient (multiplying $\frac{\omega}{q}$ in the fluctuation susceptibility $\Delta \chi(q,\omega)$) is divergent when the velocity vanishes at a point on the Fermi surface, which is very close (3 meV) to occurring in TiBe$_2$. In exploring how the FM instability (the $q$=0 Stoner enhancement is $S\approx 60$) might be suppressed by fluctuations in TiBe$_2$, we calculate that the Moriya A coefficient (of $q^2$) is negative, so $q$=0 is not the primary instability. Explicit calculation of $\chi_o(q)$ shows that its maximum occurs at the X point $(1,0,0)\frac{2\pi}{a}$; TiBe$_2$ is thus an incipient {\it anti}ferromagnet rather than ferromagnet as has been supposed. We further show that simple temperature smearing of the peak accounts for most of the temperature dependence of the susceptibility, which previously had been attributed to local moments (via a Curie-Weiss fit), and that energy dependence of the density of states also strongly affects the magnetic field variation of $\chi$.