The theory of topological-topological flat bands

Abstract: Electronic flat bands have localized Wannier-like orbitals as zero modes. In the Lieb or the kagome models, the localized orbitals satisfy a topological condition that entails two non-contractible loop eigenstates along $x/y$-axis in real space, and one topological band touching point with other bands in momentum space. In these topological-flat bands, the Bloch state at the touching point is ill-defined, and so is any topological invariant for the entire band. We propose a new topological condition that the loop states in different directions be linearly dependent. Its satisfaction removes the singularity at the band touching point, and enforces nontrivial, well-defined topological invariants. Enforcing the new condition, we obtain topological-topological (top$<sup>2$)-flat bands in 2D and 3D that have nontrivial invariants including the Chern numbers, the $\mathbb{Z}_2$ invariants, and the topological-crystalline invariants. Under small, generic interactions, top$<sup>2$-flat bands flow to correlated topological insulators with a dynamically generated, symmetric mass term; and specially designed interacting models can have top$<sup>2$-flat bands as exact zero modes.

• The paper introduces top²-flat bands that resolve singularities in flat-band models by imposing a linear dependence condition on real-space loop states.
• It provides explicit model realizations in 2D and 3D, demonstrating quantized topological invariants for Chern, ℤ₂, and crystalline phases.
• The analysis shows that interactions can dynamically generate a mass gap, leading to robust interacting topological insulating regimes.

Theoretical Framework for Topological-Topological Flat Bands

Introduction and Motivation

The theory of flat electronic bands has established a critical platform for understanding correlation phenomena due to the quenching of kinetic energy and resultant dominance of interaction effects. In established models—such as Lieb and kagome lattices—compact localized states (CLSs) give rise to flat bands, but topological singularities in those bands, arising from constraints in real-space and momentum-space structures, manifest as ill-defined points that prevent topological invariants from being rigorously assigned across the full Brillouin zone. This fundamentally limits both the mathematical description and physical realization of "topological flat bands."

The present work introduces a refined theoretical construction—termed topological-topological (top$^2$)-flat bands—in which a novel real-space constraint (linear dependence of loop states along different axes) resolves singularities at band touching points. This enables the assignment of nontrivial and well-defined topological invariants within strictly flat bands in translation-invariant and symmetry-rich systems, in both two and three spatial dimensions (2603.24672).

Real- and Momentum-Space Topological Criteria

First Topological Condition: Real-Space Loop States

In traditional flat-band models, CLSs satisfy a condition analogous to a discrete divergence theorem: the sum of states within a region can be equated to the sum over boundary states. For translation-invariant lattices, this yields two non-contractible loop states per dimension, which, through a counting argument, necessarily require the existence of band touching at specific points in momentum space. This underlying band degeneracy renders the Bloch states and associated projectors discontinuous and, critically, the construction of topological invariants on the flat bands impossible.

Second Topological Condition: Linear Dependence and Topological Classification

The central advancement is the imposition of linear dependence between loop states along different directions—for example, in 2D, $\Theta_y = \lambda \Theta_x$ with $\Im \lambda \neq 0$. At the crucial momentum where the standard Bloch state vanishes due to the first condition, the degenerate subspace is now spanned by derivatives of the Bloch states, which are proportional rather than distinct vectors. This modification regularizes the band projector, $P(\mathbf{k})$, across the Brillouin zone, removing discontinuity at the touching point and enforcing a quantized topological invariant (e.g., Chern or $\mathbb{Z}_2$ number) on the flat band itself.

This represents the formal structure of a top$^2$-flat band: a flat band with (i) non-contractible real-space structure, (ii) topological band touching for consistency with no-go theorems, and (iii) well-defined, nontrivial topological class under the revised real-space constraint.

Explicit Model Realizations

Three explicit types of top$^2$-flat bands are constructed and characterized:

• 2D Chern Bands: The imposition of the linear dependence condition yields a flat band with Chern number $\pm1$. The phase winding of Bloch states around the singularity is quantized and nontrivial.
• 2D and 3D $\mathbb{Z}_2$ Topological Bands: Incorporating time-reversal symmetry, simultaneous Kramers-degenerate flat bands are constructed. Appropriate constraints on the paired loop states generate $\mathbb{Z}_2$-nontrivial topological classes in both 2D (quantum spin Hall) and 3D (strong topological insulator) lattice models.
• Topological Crystalline Flat Bands: A layer construction, leveraging the elementary top$\Theta_y = \lambda \Theta_x$0-flat bands, provides exactly flat representations of all symmetry-indicated topological crystalline phases in every wallpaper group (and by extension, 218 space groups in 3D). The theory is extended to topological crystalline insulators beyond layer constructions via a tiling approach.

Interactions and Dynamical Mass Generation

An analytic exploration of the role of interactions on top$\Theta_y = \lambda \Theta_x$1-flat bands demonstrates that—with full band filling—the generic effect of weak interactions is the dynamical generation of a symmetric mass term at the band touching. This opens a many-body gap and drives the system toward an interacting topological insulating regime in the same class as the original free-fermion topological band (e.g., Chern or $\Theta_y = \lambda \Theta_x$2 insulator). The sign of the interaction (e.g., Hubbard attraction or repulsion) and orbital or spin polarization determine the ground state sector. Standard mean-field analysis and renormalization arguments support that the topological sector is robust and interaction-relevant, distinguishing these constructions from prior singular flat bands where the band touching was symmetry-protected and fragile.

Additionally, the existence of fully interacting, non-quadratic parent Hamiltonians retaining flat-band CLSs as exact zero modes is established using a counting argument. For large enough local Hilbert space, the number of available interaction parameters always exceeds the number of fermion constraints, confirming the possibility of constructing strictly flat, topologically nontrivial bands in genuinely interacting systems.

Theoretical and Practical Implications

The work provides a complete and operational theory for constructing dispersionless topological bands equipped with well-defined, quantized invariants in translation-invariant and symmetry-rich lattices. This resolves a central obstacle presented by previous no-go theorems and opens new regimes for the investigation of strongly correlated topological matter, facilitating the design of parent Hamiltonians with flat bands in arbitrary symmetry classes. The extension to topological crystalline insulators, and fully interacting systems, lays the groundwork for engineered quantum phase transitions, symmetry-protected topological states, and potentially for realizing fractionalized topological phenomena and idealized interacting lattice models.

Of note is the theoretical demonstration that any fully flat, symmetry-enforced topological crystalline phase can, in principle, be realized with explicit, local parent Hamiltonians, including genuinely interacting cases—a major advance toward “ideal flat-band” platforms in both solid-state and synthetic quantum materials.

Conclusion

By introducing a second, decisive constraint on the real-space topology of CLSs, this work systematically constructs flat bands with robust, nontrivial topological invariants. The implications encompass the theoretical foundation of flat-band topology, the design of engineered models for correlated topological phases, and the broad expansion of realizable topological crystalline states. This theory provides practical protocols for constructing such bands in both quadratic and purely interacting Hamiltonians, indicating fertile directions for quantum material exploration and strongly correlated electron systems.