Phase diagram of electronic systems with quadratic Fermi nodes in $2<d<4$: $2+\epsilon$ expansion, $4-\epsilon$ expansion, and functional renormalization group

TL;DR

This paper investigates the stability of quadratic Fermi node systems in 2<d<4 dimensions, using epsilon expansions and functional renormalization group methods, predicting a topological Mott insulator phase for certain materials.

Contribution

It introduces a combined epsilon expansion and functional renormalization group approach to determine the critical number of bands for instability in quadratic Fermi node systems.

Findings

Critical number of bands N_c = 64/(25 ε^2) near two dimensions.

N_c = 1.86 in three dimensions, above the physical N=1.

Prediction of a topological Mott insulator as the ground state.

Abstract

Several materials in the regime of strong spin-orbit interaction such as HgTe, the pyrochlore iridate Pr$_2$Ir$_2$O$_7$, and the half-Heusler compound LaPtBi, as well as various systems related to these three prototype materials, are believed to host a quadratic band touching point at the Fermi level. Recently, it has been proposed that such a three-dimensional gapless state is unstable to a Mott-insulating ground state at low temperatures when the number of band touching points $N$ at the Fermi level is smaller than a certain critical number $N_c$. We further substantiate and quantify this scenario by various approaches. Using $\epsilon$ expansion near two spatial dimensions, we show that $N_c = 64/(25 \epsilon^2) + O(1/\epsilon)$ and demonstrate that the instability for $N < N_c$ is towards a nematic ground state that can be understood as if the system were under (dynamically generated) uniaxial strain. We also propose a truncation of the functional renormalization group equations in the dynamical bosonization scheme which we show to agree to one-loop order with the results from $\epsilon$ expansion both near two as well as near four dimensions, and which smoothly interpolates between these two perturbatively accessible limits for general $2<d<4$. Directly in $d=3$ we therewith find $N_c = 1.86$, and thus again above the physical $N=1$. All these results are consistent with the prediction that the interacting ground state of pure, unstrained HgTe, and possibly also Pr$_2$Ir$_2$O$_7$, is a strong topological insulator with a dynamically-generated gap -- a topological Mott insulator.