Transformation-optics cloaks bend waves around a hidden region by giving the surrounding shell carefully graded material properties. At the inner surface of an ideal cloak those properties must become infinite or zero. Fields pile up there: an internal source produces an infinitely concentrated surface field, and incoming waves at certain frequencies get trapped. A known fix is to coat that inner surface with a structured "lining" that enforces a special boundary condition and keeps the fields finite. At a single frequency, this lining cuts the leftover scattering by a factor of 40 to 80. Every analysis of such linings, however, assumes one exact frequency. Every physical way of building a boundary condition, whether corrugated metal grooves, resonant cells, or tuned impedance sheets, is itself frequency-dependent. Causality limits how that frequency dependence can behave.
This note proposes to settle a concrete question. Over a band of frequencies, can any passive lining track the boundary condition a cloak needs, and if not, how much bandwidth can it buy? The approach is to compute, for each frequency and each angular pattern of the wave, the surface impedance that would cancel scattering or suppress blow-up at the inner boundary. Surface impedance here means the ratio of tangential electric to magnetic field the surface enforces. We then compare that target curve against what passive surfaces are physically allowed to do. Where the target bends the wrong way, a passive lining cannot follow it. The mismatch sets a floor on scattering that no lining design can beat. The output is a bound, a near-optimal lining that approaches it, and a fair head-to-head against the cheaper ways of avoiding blow-up.
Core Claim
The stabilizing boundary conditions in this literature are written as fixed numbers or fixed operators. Examples are the soft-and-hard surface (SHS) lining in Improvement of cylindrical cloaking with the SHS lining, the tuned Robin parameter $\alpha^{\text{sen}}$ in Cloaking vs. shielding in transformation optics, and the limiting interface law in Two dimensional invisibility cloaking for Helmholtz equation and non-local boundary conditions. An SHS lining is a corrugated conducting surface that lets current flow along only one direction. A Robin parameter is a single number fixing the ratio between a field and its normal derivative at the surface. In reality, each of these conditions is a target trajectory $Z_{\text{req}}(\omega, m)$: the surface impedance needed at angular frequency $\omega$ for the $m$-th cylindrical harmonic, meaning the wave's $m$-th angular Fourier component.
We claim this trajectory is constrained along two axes at once.
Along frequency, a lossless passive reactive surface $Z_s = iX(\omega)$ must obey Foster's reactance theorem:
$$\frac{dX}{d\omega} > 0,$$
with stronger magnitude conditions near resonances. Here $X$ is the surface reactance, meaning the imaginary part of the impedance under the convention where passivity gives a positive slope. If the cloak's own dispersive shell makes $X_{\text{req}}(\omega)$ decrease anywhere across the band, no lossless lining can follow it there. A lossy one can follow only by absorbing power, and absorbed power is exactly what produces the forward "shadow" that Inverse problems, invisibility, and artificial wormholes identifies as a giveaway.
Along angle, the Lassas–Zhou interface law $\kappa \partial_{\tilde r} u + (-\partial_\theta^2)^{1/2} u = 0$ is diagonal in harmonics. It is equivalent to the impedance condition $\kappa\, \partial_{\tilde r} u_m = -|m|\, u_m$, where $u_m$ is the $m$-th angular component of the field and $\kappa$ is the material constant in their theorem. The required impedance therefore grows with $|m|$. A local surface, one whose response at a point depends only on the field at that point, can match it for at most a narrow range of harmonics. The SHS lining's anisotropy is one particular compromise between these demands, and we will quantify how good a compromise it is.
The deliverable quantity is the band-integrated mismatch
$$\mathcal{E} = \min_{Z_s \in \mathcal{P}} \int_{\omega_1}^{\omega_2} \sum_{m} w_m \,\big|Z_s(\omega,m) - Z_{\text{req}}(\omega,m)\big|^2 \, d\omega .$$
In this expression, $\mathcal{P}$ is the set of passive, causal surface impedances realizable by a finite-depth corrugated or patterned lining. The weights $w_m$ reflect how strongly each harmonic is excited by the source or illumination of interest. We will map $\mathcal{E}$ to a lower bound on the band-averaged scattering ratio and on the peak inner-surface field.
What is new is not a better lining at one frequency. It is a causality-based bandwidth limit on boundary stabilization itself, together with a measured comparison against the alternatives.
Research Program
The first step builds the target trajectories. We use the semi-analytic cylindrical-harmonic model of the truncated cylindrical cloak from Improvement of cylindrical cloaking with the SHS lining. The cloak's truncation radius $R$ controls how close it comes to the ideal singular cloak. We replace its constant shell parameters with causal Lorentz and Drude models, the standard physical models of resonant and free-electron frequency response, fitted so that they hit the design values at the center frequency. For each $\omega$ and $m$, we solve for the inner-surface impedance that zeros the $m$-th scattering coefficient. We also solve for the impedance that minimizes the inner-surface field for an internal line source. The first step's output is a set of target curves $Z_{\text{req}}(\omega,m)$ at $R=1.05$ and $R=1.01$, and a simple diagnostic: on which sub-bands they have Foster-forbidden slope.
The second step repeats this for the spherical sensor setting of Cloaking vs. shielding in transformation optics. That paper shows that only a precisely tuned $\alpha^{\text{sen}}$ achieves low scattering while letting low-order fields into the core, and it flags dispersive realization of that impedance as open. We compute $\alpha^{\text{sen}}(\omega)$ under a dispersive shell and test whether its trajectory is passive-realizable.
The third step computes the bound $\mathcal{E}$. We parameterize $\mathcal{P}$ in two ways. The first is a Foster-form rational impedance, meaning a sum of pole terms of the kind any lossless passive network produces, with optional series resistance. The second is a physical corrugated-groove model, whose "hard" behavior holds only near its groove resonance. We minimize $\mathcal{E}$ over each family and translate the residual mismatch into scattering and field-peak floors through the harmonic model.
The fourth step tests whether loss changes the picture for internal sources. The blow-up of electromagnetic fields in 3-dimensional invisibility cloaking for Maxwell's equations proves that in lossless media the inner-surface field of an internal source tends to a delta function on the interface at every frequency, and leaves the lossy case open. We numerically track the peak and width of that surface field layer as the cloak approaches the ideal, with causal loss included, with and without the best lining. This tells us whether loss alone bounds the spike, and at what absorption cost. Abnormal enhancement of electric field inside a thin permittivity-near-zero object in free space shows that dissipation caps field enhancement in near-zero-permittivity media, which motivates expecting a finite cap here too.
All results are verified with full-wave finite-element simulation of finite-length cylinders. This also addresses the infinite-cylinder idealization of the SHS analysis.
Baselines
The naive reference point is the truncated, dispersive cylindrical cloak with no lining and lossless shell, evaluated across the band. At the center frequency it should reproduce the large inner-boundary field and the scattering that the SHS paper improves upon. If our model does not reproduce the reported far-field scattering reductions there, a factor of "more than 40" at $R=1.05$ and "nearly 80" at $R=1.01$ from Improvement of cylindrical cloaking with the SHS lining, the rest of the study is not trusted.
The strongest stabilization baseline is the ideal, non-dispersive SHS condition of that paper applied unchanged at every frequency. It is the published fix, and it represents an upper envelope that no physical lining can reach. We report how fast its advantage disappears once the lining is given realistic groove dispersion. Every lining is evaluated on the same cloak, the same harmonic truncation, the same mesh, and the same band.
The strongest broadband competitor is the scattering-cancellation cloak built from non-resonant closed rings in Low scattering broadband cylindrical invisibility cloak in free-space. It reports a 10% bandwidth (1.99–2.19 GHz) with radar cross section ratio below 0.2 for a 0.2λ conducting cylinder. Radar cross section ratio is the cloaked object's scattering divided by the bare object's. We adopt exactly that metric and threshold and the same 0.2λ object, re-optimize that coating in our solver with the same compute and the same optimization budget we give our lining, and then grow the object to 1λ and 3λ. Beating this baseline means a stabilized transformation cloak with a passive lining holds the ratio below 0.2 over at least 10% bandwidth at an object size where the cancellation coating no longer does, at equal or lower absorbed power. Beating it at 0.2λ is not expected and not claimed.
The key ablation removes the Foster constraint. A black-box optimizer chooses any impedance trajectory, including non-passive ones, under the same budget. The gap between that unconstrained optimum and the best passive lining isolates how much of the bandwidth limit comes from causality rather than from the cloak itself. A second ablation fixes a local isotropic impedance versus allowing per-harmonic, angle-dependent impedance. This isolates the cost of the non-locality identified by Lassas and Zhou.
Why not just…?
Why not just add loss to the cloak and skip the lining? This is the most serious alternative, and it may partly win. Loss is causal, broadband, and very likely bounds the delta-like surface field from The blow-up of electromagnetic fields in 3-dimensional invisibility cloaking for Maxwell's equations. The price is absorption, which produces a detectable forward shadow according to Inverse problems, invisibility, and artificial wormholes. Our fourth step compares loss-only regularization against the best passive lining at matched absorbed-power fraction. If loss-only regularization achieves the same peak-field and scattering performance at equal absorption across the band, we will report that boundary linings are not worth their complexity in the broadband regime. That would be a useful negative result.
Why not just truncate the parameters at finite values? Improvement of cylindrical cloaking with the SHS lining shows that truncation alone still produces inner-boundary blow-up as $R \to 1^+$, and that the lining improves on it by factors of 40–80, though only at one frequency. Approximate Quantum Cloaking and Almost-Trapped States finds that bounded approximate cloaks develop almost-trapped states near specific energies. Over a band, truncation might degrade more gracefully than a narrowband lining, because it is uniformly imperfect rather than tuned to one frequency. The naive baseline tests exactly this by plotting band-averaged scattering for truncation-only against the dispersive lining.
Why not just use non-resonant structures, either closed rings or waveguide-cutoff channels? The closed-ring coating of Low scattering broadband cylindrical invisibility cloak in free-space is broadband but shown only at 0.2λ. Being a passive coating, it has no mechanism for internal sources. The 3D-printed waveguide-channel cloak of Experimental Realization of an Extreme-Parameter Omnidirectional Cloak avoids resonators, but its near-zero permittivity exists only at the waveguide cutoff near 9.8 GHz, and it was not tested with internal sources. Its non-resonance therefore does not remove the frequency dependence that our bound concerns. The scale-up comparison in the Baselines section is what decides whether the closed-ring approach overtakes ours.
Proposed Experiments and Decisive Checks
All experiments run on the semi-analytic harmonic model and are cross-checked by finite-element simulation of finite-length cylinders and of the 3D spherical sensor geometry. We sweep $R \in \{1.05, 1.01\}$, use Lorentz and Drude shell models, and test fractional bandwidths of 2%, 5%, 10%, and 20%. The metrics are four: fractional bandwidth with radar cross section ratio below 0.2 (Xu et al.'s criterion); peak and width of the inner-surface field for a centered and an off-center line source; absorbed-power fraction; and, for the sensor geometry, the cloaking and shielding measures $\Gamma$ and $\Sigma$ of Cloaking vs. shielding in transformation optics as functions of frequency. For the sensor case, no reported result was found on Lacuna for bandwidth of the sensor regime; only the monochromatic scalings $\Gamma = O(\rho^2)$ and $\Sigma \approx 1$ are available. No reported numeric scattering or transmission figure was found on Lacuna for the Zheng et al. cloak beyond its 9.8 GHz operating point.
The study has three decisive checks. First, the unconstrained optimizer must never fall below the passive bound when it is restricted to passive impedances. If it does, the bound is wrong. Second, the predicted Foster-forbidden sub-bands must coincide with where finite-element scattering of the best physical groove lining rises. If they do not coincide, causality is not the binding limit. Third, matched-absorption comparisons against loss-only regularization and against the closed-ring coating at 1λ and 3λ will show whether boundary stabilization keeps any broadband advantage at all.
If time permits, a single 3D-printed metal groove lining inside a truncated microwave cloak would let us measure the predicted bandwidth edge directly. It would use fabrication methods similar to Experimental Realization of an Extreme-Parameter Omnidirectional Cloak, adapted from that paper's channel arrays to corrugated grooves.
References
- Improvement of cylindrical cloaking with the SHS lining
- Cloaking vs. shielding in transformation optics
- Two dimensional invisibility cloaking for Helmholtz equation and non-local boundary conditions
- Inverse problems, invisibility, and artificial wormholes
- The Blow-up of Electromagnetic Fields in 3-Dimensional Invisibility Cloaking for Maxwell's Equations
- Abnormal enhancement of electric field inside a thin permittivity-near-zero object in free space
- Low scattering broadband cylindrical invisibility cloak in free-space
- Approximate Quantum Cloaking and Almost-Trapped States
- Experimental Realization of an Extreme-Parameter Omnidirectional Cloak