🔢 mathematics

Trapped modes in electromagnetic waveguides

This paper establishes the existence of electromagnetic trapped modes (L² solutions) in three-dimensional unbounded domains composed of a resonator and semi-infinite waveguides, demonstrating that both homogeneous and locally perturbed non-homogeneous geometries can support such eigenvalues, some of which arise from mechanisms unique to Maxwell's equations and absent in scalar Laplacian problems.

Original authors: Anne-Sophie Bonnet-Ben Dhia, Lucas Chesnel, Sonia Fliss

Published 2026-06-05

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Anne-Sophie Bonnet-Ben Dhia, Lucas Chesnel, Sonia Fliss

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a giant, endless hallway made of perfect mirrors. This is an electromagnetic waveguide. Usually, if you shout (send a light wave) down this hallway, the sound travels forever, bouncing off the walls but never stopping. It escapes to infinity.

However, this paper asks a tricky question: Can we design a hallway where the sound gets "stuck" in a specific room, never escaping to infinity, even though the hallway continues forever?

In physics, these stuck waves are called "trapped modes." The authors of this paper are mathematicians who specialize in proving that such "ghost rooms" can exist for light (Maxwell's equations), not just for sound or water waves.

Here is a breakdown of their findings using simple analogies:

1. The Problem: Why is trapping light so hard?

Think of a standard hallway. If you drop a ball, it rolls away. In math, a famous rule (Rellich's theorem) says that in an open, empty space, waves must eventually escape. You can't trap them in a simple box in the middle of a field.

But waveguides are different. They are like long, narrow tubes. At certain frequencies, the tube acts like a filter. It only lets certain "shapes" of waves pass through. The authors show that if you build a special "dead end" or a "bump" in this tube, you can create a room where a wave vibrates happily but cannot find the exit.

2. The Two Main Strategies

The authors used two different "tools" to prove these trapped waves exist, depending on the shape of the hallway.

Strategy A: The "Lego" Approach (Separation of Variables)

Imagine a hallway that is perfectly straight and uniform, like a long Lego brick.

The Trick: If the hallway is shaped just right, you can mathematically "slice" the 3D problem into a 2D puzzle and a 1D line.
The Result: They found that if you take a 2D shape (like an L-shaped room) that traps sound, and you stretch it into a 3D tube, you can create a "ghost room" for light.
The Analogy: It's like taking a flat piece of paper with a hole in it, rolling it into a tube, and realizing the hole creates a pocket where a marble can spin forever without falling out.
Key Finding: They proved that in these specific, symmetrical shapes, you can have an infinite number of these trapped frequencies. Some are "low" frequencies (easy to catch), and some are "high" frequencies hidden inside the range of normal traveling waves.

Strategy B: The "Test Drive" Approach (Min-Max Principle)

Sometimes the hallway is too weird to slice up (like a 6-way intersection or a tripod shape). You can't use the "Lego" trick.

The Trick: Instead of solving the whole puzzle, they built a "test wave." Imagine they created a fake wave that looks like it wants to be trapped. They calculated its energy.
The Logic: If they can show that this test wave has less energy than the minimum energy required to escape, then a real trapped wave must exist. It's like proving a ball will stay in a bowl because the bowl is deeper than the hill it needs to climb to escape.
The Challenge: Light waves have a strict rule: they cannot "pile up" (they must be divergence-free). Making a test wave that fits this rule is like trying to fill a bucket with water without it overflowing or leaking. The authors had to be very clever, adding "correction terms" to their test waves to make them fit the physics rules.
Key Finding: They proved that if the "dead end" room is big enough, or if the hallway branches out in specific shapes (like an L-shape or a 6-way cross), trapped waves are guaranteed to exist.

3. The "Magic" of Topology (The Donut Effect)

One of the most surprising findings involves the shape of the hallway's cross-section.

The Analogy: Imagine a hallway that is a solid square vs. a hallway that is a square with a hole in the middle (like a donut).
The Finding: If the hallway has a hole (is not "simply connected"), light can travel in a special "TEM mode" that exists at any frequency.
The Result: By putting a "dead end" room with a hole in it at the end of a straight tube, they showed you can trap light even if the tube is very short. This is a trick that works for electromagnetism but does not work for sound or water waves. It's a unique superpower of light in these specific geometries.

4. Changing the "Air" (Material Perturbations)

So far, we assumed the hallway is empty air. What if we change the material inside?

The Idea: Imagine filling part of the hallway with a special glass that slows down light (changing the dielectric constant).
The Finding: The authors proved that if you increase the "slowness" of the material in a specific local area, you can trap light.
The Catch: It's not just about making the material "denser." The math shows that the shape of the change matters. If you change the material in a way that satisfies a specific inequality (essentially, if the material slows light down enough in a specific pattern), a trapped mode appears.

5. The "Ghost" in the Machine (Embedded Eigenvalues)

Usually, trapped waves are "low energy" and easy to find. But the authors also found ways to create "ghost" waves that are embedded in the stream of normal traveling waves.

The Analogy: Imagine a busy highway. Usually, a car (wave) drives past. An "embedded" trapped mode is like a car that is driving at the exact same speed as the traffic but is somehow stuck in a parallel lane that no one else can see, never merging or exiting.
How they did it: They used symmetry. If you build a hallway that is perfectly symmetrical (a mirror image on the left and right), you can force a wave to vibrate in a way that it cancels itself out at the "mirror line." This prevents the wave from leaking out, trapping it inside the symmetry.
Why it matters: These are unstable. If you nudge the wall slightly, the ghost wave disappears or turns into a "resonance" (a wave that stays for a long time but eventually leaks). This is the basis for "Fano resonances," which create sharp spikes in how light reflects, useful for sensors.

Summary

This paper is a mathematical proof that you can build electromagnetic cages for light.

Geometry is key: By bending, branching, or adding "dead ends" to a waveguide, you can trap light.
Topology matters: Holes in the waveguide allow for new types of trapping that don't exist in sound or water.
Materials matter: Changing the material properties locally can also trap light.
Symmetry is a shield: Perfect symmetry can hide waves inside the flow of normal waves.

The authors didn't just say "it's possible"; they provided the mathematical blueprints and the specific shapes (like L-shapes, X-shapes, and tripods) where these "ghost rooms" for light are guaranteed to exist.

Technical Summary: Trapped Modes in Electromagnetic Waveguides

Problem Statement
This work investigates the existence of electromagnetic trapped modes in three-dimensional unbounded domains. These domains are modeled as the union of a bounded resonator and one or several semi-infinite waveguides. The mathematical formulation involves Maxwell's equations with perfect electric conductor (PEC) boundary conditions ( $E \times \nu = 0$ ) and source-free conditions. The primary objective is to prove the existence of non-trivial $L^2$ solutions (trapped modes) to the eigenvalue problem:
$\text{curl curl } E = \lambda E, \quad \text{div } E = 0 \quad \text{in } \Omega, \quad E \times \nu = 0 \quad \text{on } \partial\Omega.$
These solutions correspond to eigenvalues of the self-adjoint Maxwell operator $A$ . The spectrum of $A$ consists of an essential spectrum $\sigma_{ess}(A)$ , corresponding to propagating modes, and a point spectrum $\sigma_p(A)$ . The authors distinguish between eigenvalues in the discrete spectrum ( $\sigma_d(A)$ ), which lie below the essential spectrum, and embedded eigenvalues ( $\sigma_{emb}(A)$ ), which lie within the essential spectrum.

Methodology
The paper employs spectral theory for self-adjoint operators, specifically the min-max principle, to establish the existence of eigenvalues. The core strategy involves constructing test fields $E \in X_N(\Omega)$ (the space of divergence-free vector fields with vanishing tangential trace) such that the Rayleigh quotient is strictly less than the lower bound of the essential spectrum.

The authors utilize two main approaches depending on the geometry:

Exact Construction via Separation of Variables: In geometries allowing complete separation of variables (e.g., product domains), exact trapped modes are constructed directly from eigenfunctions of 2D Dirichlet or Neumann Laplacians.
Variational Approach (Min-Max Principle): For more complex geometries where separation of variables is not globally possible, the authors construct ad hoc test fields. A significant technical challenge addressed is ensuring the divergence-free condition ( $\text{div } E = 0$ $div E = 0$ ) for test fields that are not compactly supported or are constructed by patching solutions. This is achieved by:
- Extending eigenmodes of bounded subdomains by zero (when boundary conditions permit).
- Adding gradients of scalar potentials to non-divergence-free fields to correct the divergence while maintaining boundary conditions.
- Utilizing slowly decaying functions multiplied by fundamental modes (TE, TM, or TEM) to handle local material perturbations.

Key Contributions and Results

Characterization of the Essential Spectrum: The paper clarifies that for a homogeneous waveguide with a simply connected cross-section $S$ , $\sigma_{ess}(A) = [\lambda_N; +\infty)$ , where $\lambda_N$ is the first positive eigenvalue of the Neumann Laplacian in $S$ . If $S$ is not simply connected, the existence of Transverse Electro-Magnetic (TEM) modes implies $\sigma_{ess}(A) = [0; +\infty)$ .
Existence in Separable Geometries:
- Dirichlet Construction: For waveguides formed by extruding a 2D domain $\Omega_{2D}$ (with Dirichlet eigenvalues) along a finite interval, the authors prove that the 3D operator possesses an unbounded sequence of eigenvalues. Depending on the dimensions, these can be discrete or embedded.
- Neumann Construction: Similarly, for domains extruded from 2D Neumann eigenfunctions, the authors construct families of embedded eigenvalues.
Existence in Non-Separable Geometries:
- Resonator-Based Designs: The authors prove that if a resonator $R$ is a product domain (e.g., a cylinder or a cuboid) attached to a straight waveguide, discrete spectrum exists provided the resonator's dimensions are sufficiently large. This is shown by comparing the eigenvalues of the resonator's TE, TM, or TEM modes against the cut-off frequency of the waveguide.
- Complex Topologies (6-Legs and Tripode): The paper establishes the existence of discrete spectrum in "6-legs" (X-shaped) and "Tripode" (L-shaped) 3D geometries. This requires sharp estimates of Rayleigh quotients using test fields derived from 2D Dirichlet eigenfunctions of the corresponding planar cross-sections. The authors note that while numerical approximations of the 2D eigenvalues are used to verify the strict inequality required by the min-max principle, the estimates are sufficiently robust to confirm existence.
- Non-Monotonicity: The work demonstrates that the spectrum of the Maxwell operator is not monotonic with respect to geometric inclusion (i.e., $\Omega_1 \subset \Omega_2$ does not imply $\inf \sigma(A_1) \leq \inf \sigma(A_2)$ ), a property distinct from scalar Laplacians.
Material Perturbations:
- The authors show that local perturbations of the dielectric permittivity ( $\varepsilon$ ) and magnetic permeability ( $\mu$ ) can support trapped modes.
- They derive criteria (involving integrals of $\varepsilon, \mu$ against eigenfunctions) under which increasing $\varepsilon$ and/or $\mu$ above unity creates discrete spectrum.
- A specific approach using the magnetic operator ( $H$ ) allows for symmetric treatment of $\varepsilon$ and $\mu$ , proving that if $\varepsilon \geq 1$ and $\mu \geq 1$ (with at least one strictly greater than 1 in a set of non-empty interior), discrete spectrum exists.
Embedded Eigenvalues via Symmetry:
- The paper explains how to construct homogeneous waveguides supporting embedded eigenvalues (Bound States in the Continuum) by exploiting symmetries. By considering a symmetric domain where the cross-section is not simply connected (allowing $\sigma_{ess} = [0, \infty)$ ) but the half-domain is simply connected, one can lift a discrete eigenvalue from the half-domain to an embedded eigenvalue in the full domain by enforcing specific parity conditions on the field components.

Significance and Claims
The authors emphasize that while the literature on trapped modes for scalar problems (acoustics, quantum mechanics) is rich, results for vector problems in electromagnetism are scarce. This work fills that gap by providing rigorous existence proofs for Maxwell's equations in various 3D configurations.

Key distinctions highlighted by the authors include:

Vector Specificity: Some mechanisms proposed, particularly those relying on TEM modes in non-simply connected domains or the specific behavior of vector fields under geometric inclusion, have no equivalent for scalar Dirichlet or Neumann Laplacians.
Stability: The paper notes that while discrete eigenvalues below the essential spectrum are stable, embedded eigenvalues are generally unstable under geometric perturbations, often turning into complex resonances (quasi-normal modes), which relates to the Fano resonance phenomenon.
Mathematical Interest: The existence of trapped modes is crucial for the mathematical analysis of scattering problems, as their presence renders the scattering problem ill-posed at specific frequencies and prevents classical local energy decay.

The paper concludes by listing open questions, such as the existence of trapped modes in waveguides with exterior bumps (a known result for scalar Dirichlet problems) and the potential for proving non-existence results for Maxwell's operator in specific geometries where scalar operators have no discrete spectrum.