Abstract
We propose the local density approximation+Gutzwiller method incorporating a Green's function scheme to study the topological physics of correlated materials from the first principles. Applying this method to typical mixed valence materials SmB$_6$, we find its nontrivial $Z_2$ topology, indicating that SmB$_6$ is a strongly correlated topological insulator. The unique feature of this compound is that its surface states contain three Dirac cones in contrast to most known topological insulators.
Introduction
Topological insulators (TIs) represent a unique phase of matter characterized by an insulating bulk and conducting surface states protected by time-reversal symmetry. While early discoveries focused on weakly correlated materials like Bi$_2$Se$_3$, there is growing interest in "strongly correlated" TIs. In these systems, electron-electron interactions are significant, leading to complex phenomena such as mixed valency and heavy-fermion behavior.
Samarium hexaboride (SmB$_6$) is a classic mixed-valence compound that has been a subject of intense study for decades. It exhibits a small insulating gap at low temperatures, the origin of which was long debated. Recent theories suggest SmB$_6$ is a "Topological Kondo Insulator," where the gap arises from the hybridization of localized $4f$ electrons and itinerant $5d$ conduction electrons. However, standard computational methods like the Local Density Approximation (LDA) fail to capture the strong local correlations of $4f$ electrons, making a more sophisticated approach necessary.
Methodology: LDA + Gutzwiller
To accurately model the electronic structure of SmB$_6$, we developed a method combining LDA with the Gutzwiller variational approach. This hybrid scheme allows for the treatment of strong local Coulomb interactions among $4f$ electrons while maintaining a first-principles framework. By calculating the single-particle Green’s function at zero frequency, we can determine the $Z_2$ topological index even in the presence of strong correlations. This approach effectively models "quasi-particles"—renormalized electrons that carry the weight of many-body interactions.
Electronic Structure and Band Renormalization
The application of the LDA+Gutzwiller method reveals significant differences from standard LDA predictions. Strong correlations lead to a dramatic narrowing of the $4f$ bands near the Fermi level, a phenomenon known as band renormalization.
As shown in the figure, the LDA+Gutzwiller results (C and D) demonstrate a "squashing" of the bands compared to the standard LDA results (A and B). The calculated quasi-particle weight is approximately 0.18, indicating a high effective mass for the charge carriers. This renormalization leads to an indirect hybridization gap of about 10 meV, aligning closely with experimental transport data. This gap is formed by the inversion of $d$ and $f$ bands at the $X$ points of the Brillouin zone, a key indicator of topological character.
Topological Surface States: The Triple Dirac Cone
The most striking result of our study is the prediction of the surface state structure. The calculated $Z_2$ index (1; 111) confirms that SmB$_6$ is a strong 3D topological insulator. However, unlike typical TIs that feature a single Dirac cone, SmB$_6$ is predicted to have three.
The simulation of the (001) surface clearly shows these three Dirac cones—one at the center ($\bar{\Gamma}$) and two at the boundaries ($\bar{X}$) of the surface Brillouin zone. This unique "triple Dirac cone" structure provides a specific experimental fingerprint. Verification of these states using techniques like Angle-Resolved Photoemission Spectroscopy (ARPES) would provide definitive proof of SmB$_6$'s status as a correlated topological insulator.
Conclusion
Our findings establish SmB$_6$ as a prototypical strongly correlated topological insulator. By employing the LDA+Gutzwiller method, we have successfully modeled the complex interplay between mixed valency, band inversion, and topology. The prediction of three surface Dirac cones distinguishes SmB$_6$ from previously known topological materials and opens new avenues for exploring the physics of interacting topological phases.
Summary updated