Double-exchange ferromagnetism of fermionic atoms in a $p$-orbital hexagonal lattice

A large class of correlated quantum materials feature strong Hund’s coupling. Yet cold-atom quantum simulators have so far focused primarily on single-orbital Fermi-Hubbard systems near a Mott insulator. Here we show that repulsively interacting fermions loaded into the $p$-bands of a hexagonal lattice offer a unique platform to study the interplay of “Hundness” and “Mottness.” Our theory predicts that the orbital degrees of freedom, despite geometric frustration, produce a rich phase diagram featuring a competing itinerant ferromagnetic (FM) metal and a spin-1 antiferromagnetic (AFM) insulator, with a surprising first-order transition between them controlled by density near half-filling. Ferromagnetism emerges at low fillings from the flat band and persists to stronger interactions and higher fillings via a double-exchange mechanism, where spins align to avoid Hund-rule penalties at the expense of Dirac-fermion kinetic energy. We further argue that the paramagnetic regime is a correlated “Hund metal.” $p$-orbital Fermi gases thus provide an ideal experimental setting to investigate competing exchange mechanisms in multi-orbital systems with coexisting localized and itinerant spins.

💡 Research Summary

The manuscript presents a comprehensive theoretical study of repulsively interacting spin‑½ fermions loaded into the p‑band of a hexagonal optical lattice, a setting that simultaneously hosts flat bands, Dirac cones, and multi‑orbital degrees of freedom. By deriving a Hubbard‑Kanamori‑type interaction from a deep harmonic‑oscillator approximation, the authors identify three key on‑site terms: intra‑orbital repulsion (3U/4), inter‑orbital repulsion (U/4), and a Hund’s coupling term (−U/4) that energetically favors parallel spins on the same site. Recasting the interaction in terms of spin and orbital pseudo‑spin operators makes the competition between spin ordering and orbital ordering explicit.

Using an unbiased mean‑field decoupling that allows for arbitrary spin‑ and orbital‑wave vectors, the authors solve the self‑consistent equations at zero temperature across the full range of chemical potential μ and interaction strength U. The resulting phase diagram in the μ–U plane contains three dominant phases: a paramagnetic metal (PM), a ferromagnetic metal (FM), and a half‑filled antiferromagnetic insulator (AFM). Crucially, long‑range orbital order is absent throughout the diagram because the hexagonal geometry imposes strong geometric frustration on the p‑orbital hopping; this frustration gaps all orbital excitations and forces the system to select spin order.

At low fillings (≈0.25–0.5) and modest U, ferromagnetism originates from the perfectly flat band. The flat band supports compact, localized plaquette states; virtual hopping between these states yields a direct exchange that, together with the on‑site Hund’s term, stabilizes a fully polarized FM metal in accordance with the Mielke‑Tasaki theorem for flat‑band ferromagnetism. As the filling increases beyond the flat‑band regime, carriers are promoted into the dispersive Dirac bands, yet the FM order persists because Hund’s coupling continues to favor alignment of itinerant spins with the already polarized background.

At half‑filling (⟨n⟩ = 2) each lattice site hosts two fermions. Strong on‑site repulsion forces these two particles into a high‑spin (S = 1) configuration via Hund’s coupling. Second‑order perturbation theory in the hopping t shows that virtual processes generate an antiferromagnetic super‑exchange J_AF ∝ t⁴/U³ between neighboring S = 1 moments, leading to a spin‑1 AFM insulator once U exceeds a critical value U_c ≈ 3.25 t. This AFM phase is markedly different from the usual single‑band Hubbard antiferromagnet because the local moments arise from Hund‑induced spin‑1 complexes rather than single‑electron spins.

The most striking result is the discovery of a first‑order transition between the FM metal and the AFM insulator as the chemical potential (or equivalently the particle density) is tuned near half‑filling at strong coupling. The authors demonstrate that both the total magnetization M and the staggered magnetization m_s jump discontinuously, accompanied by a sudden change in the filling ⟨n⟩. The underlying mechanism is a double‑exchange process: in an FM background a doped hole (or electron) can hop without flipping its spin, thereby gaining kinetic energy, whereas in the AFM background the same hop would create a high‑energy intermediate state penalized by the Hund’s coupling J_H = U/4. Consequently, the kinetic energy gain in the FM state outweighs the super‑exchange energy of the AFM state once a finite density of carriers is present, driving the abrupt FM‑AFM transition.

The paper also discusses experimental feasibility. Current optical‑lattice techniques can isolate the p‑band by deepening the lattice potential, ensuring that the s‑band remains inert. Interaction strength U can be tuned via magnetic‑field‑controlled Feshbach resonances, while the filling can be adjusted by controlling the total atom number or by employing a chemical‑potential‑like gradient. Quantum‑gas microscopy would allow site‑resolved detection of spin correlations, enabling direct observation of the predicted FM‑AFM first‑order transition and the associated double‑exchange physics.

In summary, the work identifies four intertwined ingredients that shape the phase diagram: (i) flat‑band‑driven ferromagnetism at low density, (ii) Hund’s‑coupling‑driven double‑exchange ferromagnetism at intermediate density, (iii) geometric frustration that suppresses orbital order, and (iv) a density‑controlled first‑order FM‑AFM transition. By providing concrete predictions for observable quantities and outlining realistic experimental parameters, the study establishes p‑orbital hexagonal lattices as a versatile quantum‑simulation platform for exploring the competition between “Hundness” and “Mottness” that lies at the heart of many strongly correlated materials.