Local Defects and the Topology of the Haldane Model

Technical Summary: Local Defects and the Topology of the Haldane Model

Problem Statement
While the tenfold classification of topological phases provides a robust framework for understanding bulk properties of non-interacting fermionic systems, the interplay between bulk topology and local defects remains a complex area of study. Specifically, in the Haldane model—a paradigmatic Chern insulator characterized by an integer topological invariant ($C = \pm 1$)—the nature of localized states induced by point defects (vacancies) is not fully understood in relation to the bulk invariant. The central question addressed is whether the existence and protection of defect-induced states depend solely on the defect configuration or if they are coupled to the global bulk topology. Furthermore, distinguishing these topological defect states from trivial local perturbations (such as adatoms) requires identifying specific, measurable signatures beyond simple energy localization.

Methodology
The authors investigate the Haldane model on a honeycomb lattice, introducing two types of point defects:

1. Vacancies: Modeled by removing a lattice site and all associated hopping bonds.
2. Adatoms: Modeled by adding an infinite on-site potential at a lattice site, effectively decoupling it without removing bonds.

The study employs a combination of analytical and numerical techniques:

• Analytical Derivation: The authors utilize the Weyl symbol (phase-space symbol) of the position-dependent Hamiltonian near Dirac points to derive the effective continuum theory. They further employ the Chern-Simons invariant on a doubled Hilbert space to establish the topological classification of the defect configuration.
• Numerical Diagnostics: Using tight-binding parameters ($t_1=1, t_2=0.1, \phi=\pi/2$), the authors perform exact diagonalization on finite systems to compute energy spectra, wavefunctions, induced charges, and probability currents.
• Wavefunction Filtering: To visualize topological features, the authors interpolate discrete eigenstates to a smooth real-space function, Fourier transform them, and filter around the $K$ and $K'$ valleys to isolate intervalley interference patterns.

Key Contributions and Results

1. The Bulk-Defect Coupling Formula
The primary theoretical result is the derivation of a relation coupling the bulk Chern number ($C$) and the defect configuration to a $\mathbb{Z}_2$ invariant ($\nu$):
$$\nu = C \cdot m \pmod 2$$
where $m = N_A - N_B$ is the net sublattice imbalance of the vacancy configuration ($N_A$ and $N_B$ are the number of vacancies on sublattices A and B, respectively).

• Implication: According to the Atiyah-Singer index theorem, $\nu$ determines the number of protected zero-energy modes modulo 2.
  • If $|N_A - N_B|$ is odd ($\nu=1$), a single protected zero-energy mode exists.
  • If $|N_A - N_B|$ is even ($\nu=0$), no protected zero mode exists; defect states hybridize and split away from zero energy.
    This establishes that the topological protection of a defect state is not intrinsic to the defect alone but is contingent on the bulk topology and the specific sublattice arrangement of the defects.

2. Three Independent Experimental Signatures
The paper identifies and verifies three distinct signatures that distinguish topological vacancy-induced states from trivial adatom defects:

• Wavefunction Dislocations:
  Interference between the $K$ and $K'$ valleys creates oscillatory wavefronts. A defect with non-zero phase winding forces a mismatch in these wavefronts, appearing as a "dislocation" in the real-space wavefunction.

  • Result: Single vacancies ($\nu=1$) exhibit a single dislocation. Configurations with an even number of dislocations (e.g., two vacancies on the same sublattice or one on each) correspond to the trivial sector ($\nu=0$). Adatoms show no dislocations. This provides a direct real-space visualization of the $\mathbb{Z}_2$ invariant.

• Fractional Charge Accumulation:
  The induced charge $Q$ is defined as the difference in accumulated charge density between the defective and clean systems.

  • Result: For topological vacancies ($\nu=1$), the charge saturates to a fractional value of $Q = -e/2$. For trivial adatoms ($\nu=0$), the net accumulated charge is zero. This offers a quantitative, experimentally measurable criterion for distinguishing defect types.

• Current Reversal:
  The probability current circulating around the defect is analyzed.

  • Result: In the clean Haldane model ($C=+1$), chiral edge states circulate counterclockwise. Vacancy-induced zero modes ($\nu=1$) circulate clockwise (opposite to the edge current). Remarkably, this current reversal persists even for hybridized modes in the $\nu=0$ case (two vacancies), indicating that the local topological character of the vacancy imprints a specific current direction regardless of whether the state is pinned at zero energy. This is described as the tight-binding analogue of current reversal produced by a vortex in a $p$-wave superconductor.

Significance and Claims
The paper claims that the coupling formula $\nu = C \cdot m \pmod 2$ serves as an organizing principle linking bulk and defect topology. The significance of the work lies in:

1. Unification: Demonstrating that bulk and defect topologies are not independent but coupled, with the defect's topological protection determined by the bulk invariant and sublattice imbalance.
2. Differentiation: Providing a rigorous set of three independent diagnostics (dislocations, fractional charge, current reversal) that definitively separate topological vacancy states from trivial local perturbations.
3. Generalizability: The authors suggest that this bulk-defect coupling principle is a general consequence of the tenfold classification, applicable to any system where a bulk $\mathbb{Z}$ invariant coexists with a defect $\mathbb{Z}_2$ invariant (specifically drawing an analogy to class-D systems like $p+ip$ superconductors).

The authors note that these signatures are in principle measurable in existing experimental platforms, such as ultracold fermionic atoms in optical lattices, photonic arrays, and engineered honeycomb lattices on metal surfaces, where single-site control is achievable.