betaFeedback

WorksMathematics

Phase transitions and sharp localisation in anomalous localised resonance

Abstract

We study how detuning and absorption determine the strength and spatial extent of anomalous localised resonance in a two-dimensional quasistatic coated structure with shell permittivity −1+cℓp+iℓ-1+c\ell^p+i\ell, where ℓ>0\ell>0 is the loss and c≠0c\ne0, p>0p>0 are fixed. Detuning can carry the material path through narrow resonances, producing intermittent rather than sustained power growth. For an explicit family of noncircular shells and a smooth exterior dipole source, both fixed, we classify the dissipated power under a quantitative smallness condition on the deformation. Depending on the detuning exponent and conformal source distance, the power diverges along the entire path, exhibits divergent peaks with vanishing gaps, or decays, including the threshold cases. We also identify the sharp conformal boundary beyond which the field, normalized by the square root of the dissipated power, converges locally to zero along every divergent-power sequence. For resonant sources, tangential reference crossings yield an explicit nonzero limit for the exterior gradient energy divided by the dissipated power at the critical radius.

Introduction

Plasmonic structures can concentrate electromagnetic fields near their interfaces through the excitation of surface modes. In the quasistatic regime, anomalous localised resonance describes a singular form of this response: as material loss vanishes, increasingly fine spatial oscillations produce large fields in a bounded region, while the response farther away remains controlled. This separation between local amplification and exterior behavior underlies its connection with cloaking and superlensing [16, 14]. Related near-field enhancement has been observed experimentally through plasmon-assisted transmission across silver films [9] and sub-diffraction-limited imaging with a silver superlens [8]. These experiments demonstrate the physical relevance of surface-mode amplification and its sensitivity to absorption, rather than the zero-loss limit studied here.

In this paper, we consider a two-dimensional coated structure whose normalized core and exterior permittivities are equal to one, while the shell permittivity is sℓ=−1+cℓp+iℓs_\ell=-1+c\ell^{p}+i\ell, with fixed c≠0c\ne0 and p>0p>0. The positive imaginary part ℓ\ell represents material loss and enters the dissipated power through the absorption density ℓ∣∇uℓ∣2\ell\lvert\nabla u_\ell\rvert^2, where uℓu_\ell denotes the complex electrostatic potential and −∇uℓ-\nabla u_\ell is the corresponding electric field. The real increment cℓpc\ell^{p} represents detuning from the reference value −1-1. Thus, pp controls whether detuning is smaller than, comparable to, or larger than loss. We call the curve ℓ↦sℓ\ell\mapsto s_\ell in the complex permittivity plane the material path. As ℓ\ell decreases to zero, this curve approaches −1-1 from the upper half-plane, with cc fixing the sign of the real detuning and pp determining the manner of approach. In particular, for 0<p<10<p<1 the approach is tangential to the real axis. Physically, this path describes a family of shell materials with progressively weaker absorption, whose real permittivity is simultaneously tuned towards −1-1, while the geometry and source remain fixed. It specifies how closely the shell is tuned as its damping is reduced: the real part controls proximity to the surface-mode resonances, whereas the imaginary part limits their amplification. Thus, the exponent pp specifies the balance between these two changes, rather than a change in the shape of the coating. This is a prescribed relation between material parameters, not necessarily a frequency sweep in a single material. Our question is how this relative rate determines both the growth of the dissipated power and the spatial extent of the resonant response for a prescribed source.

The dependence on the material path is particularly important when detuning dominates loss. As the shell approaches its limiting permittivity, the path can pass through narrow resonant windows separated by much weaker responses. Reducing loss need not, therefore, imply sustained power growth. The source position also matters, since it determines how strongly the increasingly fine modes are excited. Sensitivity to relative dissipation rates has been established for slab configurations [15], while related competition between material dispersion and damping appears in [1]. Here we vary the real detuning within the shell and retain a fixed source and geometry throughout the limit.

The mathematical analysis of this phenomenon has linked dissipated-power blow-up to the spectrum of a two-interface Neumann–Poincaré-type operator [3], with further developments concerning spectral accumulation, nonradial configurations and complementary media [4, 12, 17]. We work with the noncircular family generated by Ψ(w)=w+εw−2\Psi(w)=w+\varepsilon w^{-2} from the annulus ρ<∣w∣<1\rho<|w|<1, and a smooth exterior dipole source. Under an explicit condition relating the fixed deformation to ρ\rho and pp, we obtain a complete classification in terms of the detuning exponent and conformal source distance. The dissipated power either diverges along the entire loss path, exhibits divergent peaks separated by vanishing responses, or tends to zero. The classification includes the threshold cases and distinguishes persistent absorption from intermittent excitation of narrow resonances.

We also determine how far this amplification remains visible after the field is divided by the square root of the dissipated power. The sharp conformal observation radius is rc=ρ−1/(2min⁡(p,1))r_c=\rho^{-1/(2\min(p,1))}: beyond its image under Ψ\Psi, the normalized field converges locally to zero along every divergent-power sequence. In the resonant source range, this boundary cannot be moved closer to the shell. For tangential reference crossings, we further compute an explicit nonzero limit of the exterior gradient energy divided by the dissipated power at rcr_c. Local decay away from the critical curve, therefore, coexists with residual integrated energy near it. These conclusions concern normalized fields for a prescribed source, rather than invisibility of arbitrary objects.

Our coefficient formulation is motivated by the use of Faber polynomials and Grunsky coefficients in closely related transmission problems. In [10] construct geometry-adapted harmonic bases are constructed and layer-potential operators are represented by matrices with Grunsky coefficients, while in [5], corresponding geometric multipole expansions are developped. Particularly close to our setting is the work carried in [6], which treats the Neumann–Poincaré spectrum between two conformal level curves. We use this framework to encode the noncircular core response and the source in compatible coordinates. A related modal analysis of field concentration appears in [2], where spherical harmonics and radial Bessel functions are used to describe non-singular hotspots between high-index particles. Here, the concentration instead arises from vanishing loss in a fixed geometry. The difficulty here is to control a forced response as loss vanishes on one fixed shell, rather than to determine spectral locations alone. Geometric coupling must remain controlled even near increasingly narrow resonances. Uniform comparisons with a circular auxiliary problem resolve the power growth, and weighted comparisons retain the outward attenuation needed for sharp localisation. Our analysis is quasistatic. Finite-frequency resonance and limiting absorption involve separate issues addressed in [19, 20, 13, 11, 18].

The paper is organised as follows. In § 2, we formulate the transmission problem and state the main results. Sections 3 and 4 establish its boundary-integral and coefficient representations. In Sections 5 and 6, we derive the uniform comparison and prove the power classification. Sections 7 and 8 establish the weighted comparison, sharp localisation and endpoint limit. Section 9 illustrates these results numerically. Detailed supporting proofs are collected in the appendices.

Notation. Let A(ℓ),B(ℓ)≥0A(\ell),B(\ell)\ge0 for ℓ>0\ell>0. As ℓ↓0\ell\downarrow0, we use the conventions

A≲B⟺∃C,ℓ0>0:A(ℓ)≤CB(ℓ)(0<ℓ<ℓ0),A\lesssim B \quad\Longleftrightarrow\quad\exists C,\ell_0>0:\quad A(\ell)\le CB(\ell)\quad(0<\ell<\ell_0),
A≍B⟺A≲B and B≲A.A\asymp B \quad\Longleftrightarrow\quad A\lesssim B\text{ and }B\lesssim A.

Thus, A≍BA\asymp B means that c0B(ℓ)≤A(ℓ)≤CB(ℓ)c_0B(\ell)\le A(\ell)\le CB(\ell) for some 0<c0≤C<+∞0<c_0\le C<+\infty and all sufficiently small ℓ>0\ell>0. The constants are independent of ℓ\ell and, for estimates involving mode indices, uniform over the indices in the stated range. They may depend on fixed parameters and fixed observation radii. The same conventions apply to other limiting variables, including sequences as n→+∞n\to+\infty. If BB is positive in the limiting regime, we write

A∼B⟺lim⁡ℓ↓0A(ℓ)B(ℓ)=1.A\sim B \quad\Longleftrightarrow\quad\lim_{\ell\downarrow0}\frac{A(\ell)}{B(\ell)}=1.

Problem and main results

Let us start by formulating the transmission problem, specify the fixed geometry and source, and state the power and localisation results.

Problem setting

Let Ωi\Omega_i and Ωe\Omega_e be bounded simply connected domains in R2\mathbb{R}^2 with real-analytic Jordan boundaries, with Ωi‾⊂Ωe\overline{\Omega_i}\subset\Omega_e. We write

Γi=∂Ωi,Γe=∂ΩeandΩs=Ωe∖Ωi‾.\Gamma_i=\partial\Omega_i,\qquad\Gamma_e=\partial\Omega_e\qquad\text{and}\qquad\Omega_s=\Omega_e\setminus\overline{\Omega_i}.

The unit normals νi\nu_i and νe\nu_e point out of the bounded domains Ωi\Omega_i and Ωe\Omega_e, respectively. In particular, νi\nu_i points into the shell, and the outward shell normal on Γi\Gamma_i is −νi-\nu_i. For each interface, the symbols ++ and −- denote limits from the exterior and interior of its bounded domain, taken along this same normal.

Figure 1. The core Ωi\Omega_i, shell Ωs\Omega_s and exterior, with their normalized relative permittivities. The interfaces are Γi\Gamma_i and Γe\Gamma_e, and both normals point out of their bounded interiors.

Let ℓ>0\ell>0, c∈R∖{0}c\in\mathbb{R}\setminus\{0\}, and p>0p>0. Set

sℓ=−1+cℓp+iℓandaℓ(x)={1,x∈Ωi∪(R2∖Ωe‾),sℓ,x∈Ωs.s_\ell=-1+c\ell^p+i\ell\qquad\text{and}\qquad a_\ell(x)= \begin{cases} 1, & x\in\Omega_i\cup(\mathbb{R}^2\setminus\overline{\Omega_e}),\\ s_\ell, & x\in\Omega_s. \end{cases}

The function aℓa_\ell is the spatially varying relative permittivity, normalized by the positive exterior permittivity, and sℓs_\ell is its value in the shell. The core and exterior are lossless and have the same normalized permittivity. With the time convention e−iωte^{-i\omega t}, the imaginary part Im⁡sℓ=ℓ>0\operatorname{Im}s_\ell=\ell>0 describes absorption in the shell. We call ℓ\ell the loss parameter and dℓ=Re⁡sℓ+1=cℓpd_\ell=\operatorname{Re}s_\ell+1=c\ell^p the real detuning, as it measures the displacement of the real shell permittivity from −1-1, the reference contrast in the quasistatic resonance model [3].

Remark 2.1. The exponent pp compares these two scales, since

lim⁡ℓ→0∣dℓ∣ℓ=lim⁡ℓ→0∣c∣ℓp−1={0,p>1,∣c∣,p=1,+∞,0<p<1.\lim_{\ell\to0}\frac{|d_\ell|}{\ell}=\lim_{\ell\to0}|c|\ell^{p-1}= \begin{cases} 0, & p>1,\\ |c|, & p=1,\\ +\infty, & 0<p<1. \end{cases}

Thus, detuning is smaller than loss for p>1p>1, comparable to loss for p=1p=1, and larger than loss for p<1p<1. In the last case the permittivity approaches the real axis tangentially, allowing the material path to resolve narrow resonant peaks.

We consider a fixed real source f∈C∞(R2∖Ωe‾)f\in C^\infty(\mathbb{R}^2\setminus\overline{\Omega_e}), separated from Γe\Gamma_e, satisfying ∫R2f=0\int_{\mathbb{R}^2}f=0. The complex electrostatic potential uℓu_\ell, whose negative gradient is the electric field, solves

{∇⋅(aℓ∇uℓ)=fin R2,uℓ(x)=O(∣x∣−1)as ∣x∣→+∞.(1)\begin{cases} \nabla\cdot(a_\ell\nabla u_\ell)=f & \text{in }\mathbb{R}^2,\\ u_\ell(x)=O(|x|^{-1}) & \text{as }|x|\to+\infty. \tag*{(1)} \end{cases}

Equivalently, the governing transmission problem is

{Δuℓ=0in Ωi∪Ωs,Δuℓ=fin R2∖Ωe‾,uℓ∣+−uℓ∣−=0on Γi∪Γe,sℓ∂∂νiuℓ∣+=∂∂νiuℓ∣−on Γi,∂∂νeuℓ∣+=sℓ∂∂νeuℓ∣−on Γe,uℓ(x)=O(∣x∣−1)as ∣x∣→+∞.(2)\begin{cases} \Delta u_\ell=0 & \text{in }\Omega_i\cup\Omega_s,\\ \Delta u_\ell=f & \text{in }\mathbb{R}^2\setminus\overline{\Omega_e},\\ u_\ell|_+-u_\ell|_-=0 & \text{on }\Gamma_i\cup\Gamma_e,\\ s_\ell\frac{\partial}{\partial\nu_i}u_\ell|_+=\frac{\partial}{\partial\nu_i}u_\ell|_- & \text{on }\Gamma_i,\\ \frac{\partial}{\partial\nu_e}u_\ell|_+=s_\ell\frac{\partial}{\partial\nu_e}u_\ell|_- & \text{on }\Gamma_e,\\ u_\ell(x)=O(|x|^{-1}) & \text{as }|x|\to+\infty. \tag*{(2)} \end{cases}

The solution is required to belong to Hloc1(R2)H^1_{\mathrm{loc}}(\mathbb{R}^2) with ∇uℓ∈L2(R2)\nabla u_\ell\in L^2(\mathbb{R}^2). Since sℓ≠0s_\ell\ne0, the shell equation in (2) is equivalent to the divergence-form equation there.

Figure 2. The conformal annulus and its physical image under Ψ\Psi. Only the annulus and exterior are mapped as Ψ\Psi is not used to parametrize the core interior.

Let

References

  1. [1]K. Alexopoulos and B. Davies. The effect of singularities and damping on the spectra of photonic crystals. Journal of Mathematical Physics, 65(8), 2024.
  2. [2]K. Alexopoulos, B. Davies, and P. Millien. Non-singular hotspots between closely spaced high-index nanoparticles. arXiv preprint arXiv:2607.24204, 2026.arxiv.org/abs/2607.24204
  3. [3]H. Ammari, G. Ciraolo, H. Kang, H. Lee, and G. W. Milton. Spectral theory of a neumann–poincaré-type operator and analysis of cloaking due to anomalous localized resonance. Archive for Rational Mechanics and Analysis, 208(2):667–692, 2013.arxiv.org/abs/1109.0479
  4. [4]K. Ando and H. Kang. Analysis of plasmon resonance on smooth domains using spectral properties of the neumann–poincaré operator. Journal of Mathematical Analysis and Applications, 435(1):162–178, 2016.DOI
  5. [5]D. Choi, J. Kim, and M. Lim. Geometric multipole expansion and its application to semi-neutral inclusions of general shape: D. choi and j. kim and m. lim. Zeitschrift für angewandte Mathematik und Physik, 74(1):39, 2023.arxiv.org/abs/1808.02446
  6. [6]D. Choi, M. Lim, and S. P. Shipman. Spectral analysis of the neumann–poincaré operator for thin doubly connected domains. SIAM Journal on Mathematical Analysis, 57(3):2210–2228, 2025.arxiv.org/abs/2309.02892
  7. [7]G. Faber. Über polynomische entwickelungen. Mathematische Annalen, 57(3):389–408, 1903.
  8. [8]N. Fang, H. Lee, C. Sun, and X. Zhang. Sub-diffraction-limited optical imaging with a silver superlens. science, 308(5721):534–537, 2005.pubmed.ncbi.nlm.nih.gov/15845849
  9. [9]N. Fang, Z. Liu, T.-J. Yen, and X. Zhang. Experimental study of transmission enhancement of evanescent waves through silver films assisted by surface plasmon excitation. Applied Physics A, 80(6):1315–1325, 2005.DOI
  10. [10]Y. Jung and M. Lim. Series expansions of the layer potential operators using the faber polynomials and their applications to the transmission problem. SIAM Journal on Mathematical Analysis, 53(2):1630–1669, 2021.DOI
  11. [11]H. Kettunen, M. Lassas, and P. Ola. On absence and existence of the anomalous localized resonance without the quasi-static approximation. SIAM Journal on Applied Mathematics, 78(1):609–628, 2018.DOI
  12. [12]R. V. Kohn, J. Lu, B. Schweizer, and M. I. Weinstein. A variational perspective on cloaking by anomalous localized resonance. Communications in Mathematical Physics, 328(1):1–27, 2014.
  13. [13]H. Li and H. Liu. On anomalous localized resonance and plasmonic cloaking beyond the quasi-static limit. Proceedings. Mathematical, Physical, and Engineering Sciences, 474(2218):20180165, 2018.DOI
  14. [14]R. C. McPhedran and G. W. Milton. A review of anomalous resonance, its associated cloaking, and super-lensing. Comptes Rendus Physique, 21(4-5):409–423, 2020.DOI
  15. [15]15 T. Meklachi, G. Milton, D. Onofrei, A. Thaler, and G. Funchess. Sensitivity of anomalous localized resonance phenomena with respect to dissipation. Quarterly of Applied Mathematics, 74(2):201–234, 2016.arxiv.org/abs/1406.7044
  16. [16]16 G. W. Milton and N.-A. P. Nicorovici. On the cloaking effects associated with anomalous localized resonance. Proceedings: Mathematical, Physical and Engineering Sciences, pages 3027–3059, 2006.DOI
  17. [17]17 H.-M. Nguyen. Cloaking via anomalous localized resonance for doubly complementary media in the quasistatic regime. Journal of the European Mathematical Society, 17(6):1327–1365, 2015.arxiv.org/abs/1407.7977
  18. [18]18 H.-M. Nguyen. Limiting absorption principle and well-posedness for the helmholtz equation with sign changing coefficients. Journal de Mathématiques Pures et Appliquées, 106(2):342–374, 2016.arxiv.org/abs/1507.01730
  19. [19]19 H.-M. Nguyen. Cloaking via anomalous localized resonance for doubly complementary media in the finite frequency regime. Journal d’Analyse Matématique, 138(1):157–184, 2019.
  20. [20]20 D. Onofrei and A. E. Thaler. Anomalous localized resonance phenomena in the nonmagnetic, finite-frequency regime. Advances in Mathematical Physics, 2016(1):4156072, 2016.arxiv.org/abs/1605.08954

Paper details

vlnc:W_2j7shua3
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 5

downloads · 1

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

ChatGPT/Codex

View statement
Looking for related papers…
Contents