Dirac fermions in non-Hermitian magnetic fields: Zero modes and index theorem

In the study of quantum materials, researchers often focus on how external forces shape the behavior of electrons. In systems like graphene, electrons behave like massless, relativistic particles known as Dirac fermions. Strong magnetic fields can create a specialized set of energy states called the "zeroth Landau level." These states are topologically protected. This means they remain stable against local disturbances.

Until now, this stability was understood primarily through Hermitian physics. In Hermitian systems, energy levels are real and predictable. However, a new frontier is emerging in "non-Hermitian" (NH) physics. This field describes systems that exchange energy with their environment. Such systems lead to complex energy values. A new paper explores a specific question: what happens when the magnetic field itself is non-Hermitian? The authors propose a theoretical construction where such fields bind particles into localized "left" or "right" eigenmodes (mathematical vectors that solve the energy equation from different sides).

The limits of traditional magnetic coupling

In standard condensed matter physics, magnetic fields are introduced via a gauge potential. This potential shifts the momentum of particles. When dealing with Dirac fermions in a plane, this coupling follows the Aharonov-Casher index theorem (ACIT). This theorem provides a mathematical guarantee. It states that the number of zero-energy states equals the amount of enclosed magnetic flux. This relationship is a cornerstone of the quantum Hall effect in graphene.

The current framework relies heavily on Hermiticity. In a Hermitian system, wavefunctions must be "normalizable." This means the particle's probability density stays within a finite region of space. Standard magnetic fields create stable, degenerate manifolds of these states. However, they cannot easily account for the unique behaviors seen in open quantum systems. There has been a gap in understanding how mass-like, anti-Hermitian operators could interact with these fermions.

Constructing the non-Hermitian magnetic field

The authors propose a Lorentz-invariant non-Hermitian Dirac theory. Instead of simply adding a magnetic field to a standard Hamiltonian (the operator representing total energy), they introduce a masslike anti-Hermitian term. Their approach follows a three-step construction:

1. Defining the Operator: They start with a standard Dirac Hamiltonian ($H_{Dir}$) and add a masslike term ($\alpha MH_{Dir}$). Here, $\alpha$ dictates the strength of the non-Hermiticity. This ensures the theory remains Lorentz invariant.

2. Coupling to Gauge Fields: They allow the non-Hermitian parameter to vary spatially. By substituting momentum $k$ with a vector potential $A(r)$ in the anti-Hermitian term, they treat the non-Hermiticity as a "non-Hermitian magnetic field" (NHMF).

3. Generating Zero Modes: This spatial modulation creates a similarity transformation. As shown in Equation 3 of the paper, the resulting Hamiltonian is related to the original one through a space-dependent scaling factor, $\chi(r)$.

This mechanism changes the nature of zero-energy states. In non-Hermitian systems, the "left" and "right" eigenvectors are no longer identical. Depending on the sign of the spatial modulation $\chi(r)$, the system hosts either localized right eigenmodes or left eigenmodes .

Evidence from the honeycomb lattice

The authors anchored their continuum model using microscopic simulations of graphene’s honeycomb lattice. They implemented two distinct types of NHMFs:

• Type-I NHMFs: These are realized through a sublattice-independent on-site potential that varies across the lattice .

The authors report that for a positive modulation, the "left" zero-energy eigenmodes are sharply localized in the bulk. Meanwhile, the "right" eigenmodes reside at the boundaries [Figure 2(a, b)]. * Type-II NHMFs: These arise from imaginary next-nearest-neighbor hopping amplitudes . The numerical results show a different pattern. Both left and right eigenmodes are peaked near the boundary. However, they do not sit exactly on the edges [Figure 2(c, d)].

The authors also confirm that the "index theorem" holds in this regime. They report that the number of near-zero-energy modes ($N_0$) scales linearly with the number of enclosed magnetic flux quanta ($N$). This confirms that the relationship between topology and flux survives in non-Hermitian physics. The authors even extended this logic to a ${10, 3}$ hyperbolic lattice .

Constraints of the masslike approximation

The scope of this paper is intentionally limited. The authors focus exclusively on "masslike" non-Hermitian gauge fields. These fields are designed to vanish as momentum $k$ approaches zero. Consequently, the model may not capture all possible non-Hermitian interactions.

The distinction between "normalizable" and "non-normalizable" modes is vital for experimentalists. In non-Hermitian systems, a mode might be mathematically valid but physically difficult to observe. A non-normalizable mode lacks a finite integrated probability density. The paper notes that for any given field direction, only one type of mode is self-normalizable. This means experimentalists must precisely tune the field. Doing so ensures they probe the states that are actually physically observable.

The verdict: A new toolkit for topological design

The existence of zero-energy NH flat bands offers several possibilities. It provides a foundation for "non-Hermitian magnetic catalysis." This is a process where zero-energy states could drive the formation of new topological phases.

Is this ready for the lab? The authors suggest that engineered quantum systems are the best testing grounds. These include topolectric circuits, photonic lattices, and optical lattices of neutral atoms. These platforms allow for the precise, non-reciprocal hopping required to simulate Type-I and Type-II fields. The research moves non-Hermitian physics toward becoming a predictive tool for designing new topological materials.