Quantum-geometry-driven Mott transitions and magnetism

Abstract: Quantum geometry quantifies how the single-particle Bloch wavefunction changes in phase and amplitude across the Brillouin Zone. In multi-orbital systems where bands have strongly mixed orbital composition, quantum geometry plays a vital role in determining the ground state and low-energy properties of interacting electronic systems. In this work, we show that Mott metal-insulator transitions, as well as transitions between different magnetic orders within the Mott insulating phase, can be driven by the quantum geometry of the underlying Bloch band, thereby providing a mechanism complementary to conventional bandwidth-tuned Mott transitions. By studying the Kane-Mele-Hubbard model using exact diagonalization, we demonstrate that in in half-filled and topologically-trivial bands, quantum geometric properties of the Bloch states alone can act as a tuning knob for Mott metal-to-insulator and affect the competition between ferromagnetism and antiferromagnetism. We show that both transitions may be heuristically understood via non-local Coulomb scattering in a basis of exponentially localized Wannier functions. These results highlight the role of quantum geometry beyond topological settings as a governing principle for conventional Mott and magnetic physics in multi-orbital and moiré materials.

• The paper demonstrates that Bloch-wavefunction quantum geometry independently tunes Mott metal–insulator transitions by limiting Wannier localization, reducing effective on-site repulsion, and enhancing density-assisted hopping.
• Exact-diagonalization studies of Kane–Mele and BHZ models find transitions among metallic, ferromagnetic Mott, and antiferromagnetic Mott phases, with geometry potentially enabling a direct ferromagnet-to-metal transition.
• A Wannier-based Schrieffer–Wolff analysis shows that geometry-controlled direct exchange competes with superexchange, explaining magnetic transitions and highlighting effects missing from single-band Hubbard models.

Overview

This paper by Ding and Claassen establishes that the quantum geometry of Bloch wavefunctions — independent of bandwidth and filling — can act as a tuning knob for Mott metal-insulator transitions (MITs) and for ferromagnet–antiferromagnet (FM–AFM) transitions within the Mott insulating phase. The central claim is that in topologically trivial, half-filled bands with strong orbital mixing, quantum geometry sets a lower bound on Wannier localization; imperfectly localized orbitals reduce the effective on-site repulsion and generate density-assisted hopping and nearest-neighbor exchange, thereby modifying both charge and spin ordering. This provides a mechanism complementary to conventional bandwidth-tuned Mott transitions (2602.22548).

The study is motivated by two-dimensional moiré materials, where band-geometric effects are pronounced and where correlated insulating states have been observed whose conventional Hubbard-model descriptions may be incomplete. Importantly, the authors restrict their analysis to unobstructed, topologically trivial bands ($U, W \ll \Delta$, where $\Delta$ is the gap to remote bands), distinguishing their scenario from topological-band settings that generically exhibit flavor ferromagnetism and quantum anomalous Hall physics.

Model and methodology

The starting point is a time-reversal-symmetric two-band tight-binding model of spinful fermions with a purely local, intra-orbital Hubbard interaction $U$. The kinetic Hamiltonian has the generic form $$h_s(\mathbf{k}) = \mathbf{a}(\mathbf{k}) \cdot \boldsymbol{\sigma}$$, possessing $\mathbb{Z}_2 \times U(1)$ spin symmetry. To disentangle quantum geometric effects from dispersion effects, the authors employ a controlled deformation procedure: they flatten the bands while preserving the Bloch eigenvectors exactly, then reintroduce a small controllable dispersion via a nearest-neighbor hopping $t_W$ between Wannier orbitals of the lower band. This leaves only two independent scales in the problem: $U/t_W$ and a dimensionless parameter controlling the momentum-space structure of the Bloch wavefunctions.

Two limiting cases anchor the analysis. In the flat-band limit ($t_W = 0$), the projected interaction is positive semidefinite and fully polarized Slater determinants are zero-energy ground states, yielding a ferromagnetic Mott insulator (Mott FM) analogous to flat-band ferromagnets studied by Tasaki and Mielke. In the "trivial quantum geometry" limit, the form factors reduce to those of a single-band Hubbard model, which is generically a paramagnetic metal at small $U/t_W$ and an antiferromagnetic Mott insulator (Mott AFM) at large $U/t_W$. Between these limits, transitions among Mott FM, Mott AFM, and metallic phases are expected.

Numerically, the authors solve the band-projected Hamiltonian using exact diagonalization (ED) on hexagonal 12d4 Betts clusters for the Kane-Mele model (with $\Delta$0, focusing on the trivial regime $\Delta$1), and on square 10h3 Betts clusters for the BHZ model as a cross-check. Charge gaps are diagnosed through four complementary criteria: direct thresholding of the finite-size charge gap, linear extrapolation of $\Delta$2 from fits to $\Delta$3, twisted-boundary-condition averaging, and an analytical estimate based on renormalized interaction parameters.

Quantum-geometric Mott transition

At fixed $\Delta$4, decreasing the sublattice mass parameter $\Delta$5 — which increases orbital mixing and delocalizes the Wannier functions — drives the system from a Mott insulator toward a metal. All four MIT criteria produce qualitatively consistent phase boundaries, though they differ in detail; the authors acknowledge that pinpointing the exact transition location in finite-size ED is challenging. The observed MIT boundary has positive slope and approaches the known triangular-lattice Hubbard result ($\Delta$6 from DMRG studies) in the large-$\Delta$7 limit, providing a consistency check against established results.

The physical mechanism is transparent in the Wannier basis: quantum geometry controls the minimal achievable Wannier spread, and delocalized orbitals simultaneously (i) reduce the effective on-site repulsion $\Delta$8, (ii) introduce density-assisted hopping $\Delta$9 that broadens the effective bandwidth to approximately $U$0, and (iii) generate nearest-neighbor repulsion $U$1. An analytical estimate of the MIT location based on the triangular-lattice critical ratio $U$2 agrees fairly well with the numerical boundary. Notably, at strong quantum geometric distortion, the phase diagram suggests a direct Mott FM-to-metal transition without an intervening Mott AFM phase — contrary to expectations from the single-band Hubbard model.

Magnetic order and FM-AFM transitions

Within the Mott insulating phase, the ground state exhibits out-of-plane $U$3 ferromagnetism at small $U$4 and in-plane $U$5 Néel antiferromagnetism at larger $U$6 or smaller $U$7. Spin correlations computed in the $U$8 sector are nearly isotropic but show slightly stronger in-plane correlations peaked at $U$9, identifying the Mott AFM region. The FM-AFM boundary corroborates recent proposals that quantum geometry contributes to magnetic transitions in narrow bands [Repellin et al., Hu et al., Oh et al.], and extends prior continuum-model observations by explicitly separating quantum geometric from bandwidth effects.

The mechanism is again captured quantitatively via short-range Wannier-basis matrix elements. Performing a Schrieffer-Wolff transformation to second order in $$h_s(\mathbf{k}) = \mathbf{a}(\mathbf{k}) \cdot \boldsymbol{\sigma}$$0 yields an effective nearest-neighbor XXZ spin model with exchange

$$h_s(\mathbf{k}) = \mathbf{a}(\mathbf{k}) \cdot \boldsymbol{\sigma}$$1

where $$h_s(\mathbf{k}) = \mathbf{a}(\mathbf{k}) \cdot \boldsymbol{\sigma}$$2 is ferromagnetic direct exchange arising from Wannier overlap, and the superexchange term includes contributions from both bare hopping and density-assisted hopping. Delocalization increases the ferromagnetic direct exchange, which competes with antiferromagnetic superexchange controlled by $$h_s(\mathbf{k}) = \mathbf{a}(\mathbf{k}) \cdot \boldsymbol{\sigma}$$3 and $$h_s(\mathbf{k}) = \mathbf{a}(\mathbf{k}) \cdot \boldsymbol{\sigma}$$4. The sign change of $$h_s(\mathbf{k}) = \mathbf{a}(\mathbf{k}) \cdot \boldsymbol{\sigma}$$5 tracks the numerically determined FM-AFM boundary well in both the Kane-Mele and BHZ models, indicating that quantum-geometry-driven magnetic transitions are generic rather than model-specific.

Limitations and open questions

Several caveats qualify these results. First, all ED calculations are performed on finite Betts clusters (12d4 and 10h3), and the precise locations of both the MIT and the FM-AFM transition carry finite-size uncertainty; four different MIT criteria yield slightly different boundaries. Second, a remnant insulating region with weak, ambiguous spin correlations resists identification; the authors note it could correspond to a chiral spin liquid — consistent with DMRG findings in the triangular-lattice Hubbard model — but this remains inaccessible at their cluster sizes. Third, the analytical estimates rely on truncating the Wannier-basis interaction to local and nearest-neighbor terms, justified by exponential decay but not systematically controlled. Fourth, in the BHZ model no metallic phase was observed in the studied parameter range, which the authors attribute to exact $$h_s(\mathbf{k}) = \mathbf{a}(\mathbf{k}) \cdot \boldsymbol{\sigma}$$6 Fermi-surface nesting of the square-lattice dispersion; observing the quantum-geometric MIT there required artificially breaking nesting with next-nearest-neighbor hopping. Finally, whether spectral or transport signatures can distinguish quantum-geometry-driven from bandwidth-driven Mott transitions experimentally remains an open question the paper raises but does not resolve.

Conclusion

This work demonstrates that Bloch-state quantum geometry constitutes an independent control parameter for Mott physics in topologically trivial bands, distinct from bandwidth and doping. By tracking the evolution of projected Coulomb matrix elements in a basis of exponentially localized but poorly localized Wannier functions, the authors provide both numerical evidence (via ED on the Kane-Mele-Hubbard and BHZ-Hubbard models) and semi-analytical understanding of geometry-driven MITs and FM-AFM transitions. The results imply that quantitative analyses of Mott physics in multi-orbital and moiré systems should account for the full Bloch wavefunction structure beyond effective local Hubbard models, and suggest that quantum geometry may stabilize proximate phases such as chiral spin liquids through longer-range exchange interactions induced by imperfect Wannier localization.