betaFeedback

WorksPhysics and Astronomy

Acoustics of a partially partitioned narrow slit connected to a half-plane: case study for exponential quasi-bound states in the continuum and their resonant excitation

Abstract

Localised wave oscillations in an open system that do not decay or grow in time, despite their frequency lying within a continuous spectrum of radiation modes carrying energy to or from infinity, are known as bound states in the continuum (BIC). Small perturbations from the typically delicate conditions for BIC almost always result in the waves weakly coupling with the radiation modes, leading to leaky states called quasi-BIC that have a large quality factor. We study the asymptotic nature of this weak coupling in the case of acoustic waves interacting with a rigid substrate featuring a partially partitioned slit – a setup that supports quasi-BIC that exponentially approach BIC as the slit is made increasingly narrow. In that limit, we use the method of matched asymptotic expansions in conjunction with reciprocal relations to study those quasi-BIC and their resonant excitation. In particular, we derive a leading approximation for the exponentially small imaginary part of each wavenumber eigenvalue (inversely proportional to quality factor), which is beyond all orders of the expansion for the wavenumber eigenvalue itself. Furthermore, we derive a leading approximation for the exponentially large amplitudes of the states in the case where they are resonantly excited by a plane wave at oblique incidence. These resonances occur in exponentially narrow wavenumber intervals and are physically manifested in cylindrical-dipolar waves emanating from the slit aperture and exponentially large field enhancements inside the slit. The asymptotic approximations are validated against numerical calculations.

I Introduction

In an open system, unforced wave oscillations normally decay with time even in the absence of dissipation. This happens because of radiation damping, whereby local oscillations couple to waves propagating to infinity. There are rare counter examples, however, of localised time-harmonic waves that co-exist, yet do not couple with outgoing propagating waves. These are called bound states in the continuum (BIC), embedded trapped modes, edge states or dark states; they were first described in quantum mechanics by von Neumann and Wigner [1, 2] and since established as a general wave phenomenon with potential applications across wave physics, especially in acoustics, water waves and photonics [3–6]. The terminology alludes to the fact that BIC occur at single isolated frequency eigenvalues embedded in the continuous frequency intervals in which waves are allowed to propagate to and from infinity; they are distinguished from conventional trapped modes occurring outside of these frequency intervals, wherein the system is effectively closed.

Since BIC do not couple with outgoing waves, their quality factor is infinite. Furthermore, they cannot be observed in the far field. Conversely, time-reversal symmetry implies that BIC also do not couple with incoming waves and so cannot be excited by incident radiation. BIC only exist for idealised systems and under contrived conditions; small perturbations, say involving geometry, material properties, dissipation or nonlinearity, almost always lead to coupling with incoming and outgoing waves. BIC then become leaky states with high quality factors (i.e., long life times relative to the period of oscillation) that can be resonantly excited by incident radiation, albeit slowly, giving rise to narrowband resonances that can be observed in the far-field and exhibit significant field enhancements in the near-field. Leaky states derived from BIC are called asymptotic- or quasi-BIC, indicating their asymptotic approach to BIC as some perturbation parameter, say ϵ\epsilon, vanishes. Notably, the latter constitutes a singular asymptotic limit, in which, for example, the quality factor diverges. To date, the asymptotic approach of quasi-BIC to BIC has been theoretically studied mainly in the context of embedded guided modes in periodic structures [14, 15, 16, 17] [7, 8, 9, 10, 11, 12, 13], in which case perturbation theory [18, 19, 20, 21, 22] reveals algebraically singular quality factors scaling like 1/ϵs1/\epsilon^{s}, where ss is a positive integer (typically s=2s=2, 4 or 6) and ϵ\epsilon represents the perturbation of the Bloch wavenumber from its value at which an embedded guided mode exists.

In this paper we use singular perturbation methods [23] to asymptotically study a simple setup that supports quasi-BIC whose quality factor diverges exponentially as a perturbation parameter ϵ\epsilon vanishes. We also study the exponentially resonant excitation of these states by incident radiation. The setup is shown in Figure 1. We consider planar acoustic waves in the domain exterior to a flat rigid substrate into which a perpendicular rectangular slit is embedded. Attached to the bottom of the slit is a rigid partition dividing a lower part of the slit into two channels of equal width. We denote the partition length by aa, the slit length by (1+l)a(1+l)a, where l>0l>0 is a parameter, and the slit width by 2ϵa2\epsilon a. The thickness of the partition is neglected. The leaky states, i.e., quasi-normal modes [24], of this configuration are governed by an eigenvalue problem where there is no incident field and the eigenfunction satisfies an outward-radiation condition at large distances (Figure 1a). The wavenumber, denoted k/ak/a so that kk is dimensionless, serves as the eigenvalue. Since the system is open, kk is in general complex valued. Writing k=k′+ik′′k=k^\prime+ik^{\prime\prime}, where k′k^\prime and k′′k^{\prime\prime} are real, the quality factor is asymptotically Q=k′/∣2k′′∣Q=k^\prime/\lvert2k^{\prime\prime}\rvert for small k′′/k′k^{\prime\prime}/k^\prime [25]. The excitation of leaky modes is studied by considering an associated scattering problem where the wavenumber, denoted κ/a\kappa/a, is real and there is an incident field, which we take to be a plane wave incoming at an angle α\alpha to the partition (Figure 1b).

Eigenvalue and scattering problems for a partially partitioned slit channel connected to a half-plane

Figure 1. (a) Eigenvalue problem governing the acoustic leaky states of a partially partitioned slit channel connected to a half plane. (b) Associated scattering problem with an incident plane wave.

To see why this setup supports quasi-BIC, consider the semi-infinite geometry shown in Figure 2. This geometry is just the partitioned slit shown in Figure 1, infinitely extended away from the partition rather than connecting to a half-plane. A more relevant perspective is that the finite slit domain degenerates to the semi-infinite one in the limit ϵ→0\epsilon\to0 if the position along the slit is held fixed. The degenerate geometry is a special case of the more general obstacle-in-channel class of configurations that have been extensively searched for BIC in acoustics, water waves and photonics [26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38] [3, 4, 5]. In particular, the semi-infinite geometry shown in Figure 2 is known to support BIC due to symmetry. Indeed, in light of the domain’s symmetry, eigenfunctions are either symmetric or anti-symmetric with respect to the partition. For k<π/2ϵk < \pi/2\epsilon, however, only the fundamental, transversally uniform and hence symmetric, waveguide mode is propagating in the unified part of the channel. Accordingly, the eigenvalue problem restricted to anti-symmetric eigenfunctions in this wavenumber regime effectively corresponds to a closed system, in which case discretely distributed real-valued eigenvalues are to be expected. As depicted in Figure 2, the BIC consist of standing waves at dimensionless wavenumbers k≈(1/2+n)πk \approx(1/2+n)\pi, corresponding to an integer n+1n+1 number of quarter-wavelengths fitted along the partition, with order ϵ\epsilon corrections associated with end effects at the edge of the partition and with the eigenfunctions decaying exponentially on a length scale ϵa\epsilon a away from the partition. A rigorous and constructive proof of existence employing a modified residue-calculus method was provided by Evans, Linton and Ursell [30] (for a more general class of geometries including the case of an off-centred thin partition, which remarkably can also support BIC). The same paper also presents an intuitive derivation of accurate approximate formulae to be discussed later. Returning to the finite-slit geometry of Figure 1, it is plausible that the same picture applies, only that

A semi-infinite slit channel with a finite partition, showing standing waves, evanescent decay, and the propagation of only symmetric waves

Figure 2. A semi-infinite slit channel with a finite partition supports anti-symmetric bound states in the continuum (BIC). The anti-symmetric leaky states of the setup in Figure 1a that approach the BIC of the semi-infinite geometry as ϵ→0\epsilon\to0 are called quasi-BIC.

now propagating waves are ‘turned-on’ at the aperture by the eigenfunction’s evanescent tail, which has decayed exponentially a distance lal_a at a rate of order 1/ϵa1/\epsilon_a. This, in turn, suggests leakage to infinity and corrections to the standing waves in the vicinity of the partition that are exponentially small in ϵ/l\epsilon/l.

Besides being a case study for the asymptotic properties of quasi-BIC, the present study may also be relevant to ‘metamaterial’ modelling and design, where cavities and channels supporting resonances are commonly used in order to manipulate waves; in particular, substrates and plates featuring narrow slits have been extensively used to study the phenomenon of extraordinary transmission [39, 40, 41, 42, 43, 44, 45, 46]. These studies exploit the conventional standing-wave resonances of a slit cavity, which are associated with cylindrical monopole-like radiation emitted from the slit aperture and quality factors and field enhancements that are algebraically large in the cavity aspect ratio. The asymptotic properties of such resonances have been studied extensively using various methods [47, 48, 49, 50, 51, 52, 53, 54, 55]. In this context, the present work shows that adding a partition to a slit introduces an extra set of resonances, which lie in the same frequency regime but are distinct in that they exhibit exponentially large quality factors and field enhancements and that cylindrical dipole-like radiation is emitted from the slit aperture. This added freedom could be used to explore novel BIC-based metamaterial phenomena [5, 6], in particular using the analytical results developed herein.

Building on previous singular-perturbation analyses of resonant phenomena involving narrow slits [51, 52, 53, 54, 56], we shall use the method of matched asymptotic expansions [23] to construct approximations for the wave field by systematically relating asymptotic expansions whose domain of validity spatially overlap. Furthermore, as in [51, 52, 56], we shall use conformal mappings [57] to solve certain ‘inner’ problems that arise as part of the analysis, in this case associated with regions near the partition edge and slit aperture. The present problem differs from those previous singular-perturbation analyses (as well as other asymptotic approaches to slit resonances [50] and quasi-BIC for periodic media [18, 19, 20, 21, 22]), however, in that exponentially small terms must be calculated. In particular, k′′k^{\prime\prime} is exponentially small and beyond all orders of the expansion for kk in powers of ϵ\epsilon. Nonetheless, we shall see that a leading approximation for k′′k^{\prime\prime} can be derived based on knowledge of the leading terms in the expansions for kk and the eigenfunction, though including regions away from the partition where the latter is exponentially small. This is done using a reciprocal relation which represents a balance between the rate of change of energy of the state and the energy loss by radiation to infinity. The idea of isolating exponentially small terms using appropriate integral balances is commonly used in scenarios involving exponentially weak radiation losses [58]. In the context of the scattering problem, we use a similar approach to calculate the exponential asymptotics describing the unconventional resonant excitation of the quasi-BIC based on the asymptotic approximations derived for the eigenvalue problem.

The paper is structured as follows. In §II, we formulate the eigenvalue problem governing the leaky states of the configuration (Figure 1a). In §III, we analyse the anti-symmetric states of the configuration in the limit ϵ→0\epsilon\to0. Up to exponentially small terms, they are localised to the vicinity of the partition, where they appear identical to the BIC of the semi-infinite configuration (Figure 2). The analysis in this section describes the eigenfunctions in those regions and furnishes a real-valued expansion for kk (we suffice with calculating the first two terms). In §IV, we extend the asymptotic description to regions away from the partition, where the eigenfunctions are exponentially smaller in magnitude. We then use the reciprocal relation from §II to calculate an exponentially small leading approximation for k′′k^{\prime\prime}. In §V, we formulate the scattering problem (Figure 1b) and derive an associated reciprocal relation. We then use the eigenfunction asymptotics and the latter reciprocal relation to find asymptotic approximations describing the exponentially resonant excitation of the quasi-BIC. We give concluding remarks in §VI.

Throughout the paper, we will compare our asymptotic formulae with numerical results. These were obtained by numerically solving both the eigenvalue problem formulated in §II and the scattering problem formulated in §V based on the formulation and approximation of integral equations which arise from representing solutions using eigenfunction expansions and Fourier transforms in finite and semi-infinite domains, respectively. Details of the numerical schemes are provided as supplementary material [59].

Eigenvalue problem

Geometry and problem formulation

It is convenient to normalise lengths by aa and work with the rescaled wavenumber eigenvalue k/ak/a, where k=k′+ik′′k = k^{\prime} + \mathrm{i}k^{\prime\prime} as in the introduction. Figure 3a shows a dimensionless sketch of the geometry and defines Cartesian coordinates (x,y)(x,y) and polar coordinates (r,ϕ)(r,\phi). We assume the convention where the wave field varies with time tt like exp⁡(−ikt/a)\exp(-ikt/a). In the absence of any source of energy, an outward-radiating wave field cannot grow in time; thus k′′≤0k^{\prime\prime} \le0. The reduced wave field, or eigenfunction, obtained by suppressing the time dependence is denoted by uu. It is determined up to a multiplicative constant and taken to be dimensionless without loss of generality.

Dimensionless sketch of the geometry and asymptotic regions in the limit $\epsilon \ll 1$

Figure 3. (a) Dimensionless sketch of the geometry. (b) Asymptotic regions in the limit ϵ≪1\epsilon\ll1.

With these conventions, the eigenvalue problem consists of the Helmholtz equation

∇2u+k2u=0,(1)\nabla^{2}u+k^{2}u=0, \tag*{(1)}

the Neumann boundary condition

∂u∂n=0(2)\frac{\partial u}{\partial n}=0 \tag*{(2)}

and an outward-radiation condition as r→∞r \to\infty. The latter condition is implemented by requiring that the eigenfunction’s far field can be expanded using Hankel functions of the first kind.

In the next two sections, we study the above eigenvalue problem in the limit ϵ≪1\epsilon\ll1. Specifically, our interest is in the family of quasi-BIC discussed in the introduction, for which the eigenfunctions are anti-symmetric, i.e., odd in xx, and for which k≃1k \simeq1, where

≃\simeq hereafter means ‘of asymptotic order’. The latter condition implies that, for small ϵ\epsilon, anti-symmetric waves are evanescent in the unified portion of the slit channel. Given the anti-symmetry of the eigenfunctions, it will sometimes be convenient to consider one half of the fluid domain (x>0x > 0, or x<0x < 0) using the Dirichlet condition

u=0atx=0(y>−l).(3)u = 0 \quad\text{at} \quad x = 0 \quad(y > -l). \tag*{(3)}

Reciprocal relation

Critical to our analysis is an exact reciprocal relation obtained by applying Green’s second identity to the field uu and its complex conjugate u∗u^*,

∮∂Dλ(u∗∂u∂n−u∂u∗∂n) dl+4ik′k′′∫Dλ∣u∣2 dA=0,(4)\oint_{\partial D_\lambda} \left( u^* \frac{\partial u}{\partial n} - u \frac{\partial u^*}{\partial n} \right) \, dl + 4i k' k^{\prime\prime} \int_{D_\lambda} |u|^2 \, dA = 0, \tag*{(4)}

where dldl and dAdA are infinitesimal length and area elements, respectively, and we choose the domain DλD_\lambda as the union of the slit region and the subset r<λr < \lambda of the exterior half-plane, for some λ>0\lambda> 0 (see Figure 3a). With that choice, and using the Neumann condition (2), identity (4) can be written

k′′=i4k′λ∫0π(u∗∂u∂r−u∂u∗∂r)r=λ dϕ∫Dλ∣u∣2 dA.(5)k^{\prime\prime} = \frac{i}{4k'} \frac{ \lambda\int_0^\pi\left( u^* \frac{\partial u}{\partial r} - u \frac{\partial u^*}{\partial r} \right)_{r=\lambda} \, d\phi}{ \int_{D_\lambda} |u|^2 \, dA}. \tag*{(5)}

Note that k′′k^{\prime\prime} cannot depend upon λ\lambda, which is arbitrary. This identity represents a balance between the time rate of change of the state’s energy, which is proportional to k′′k^{\prime\prime}, and the energy lost by radiation to infinity. It will be used in §IV D to calculate an asymptotic approximation for k′′k^{\prime\prime}.

Apparent bound states in the continuum

Figure 3b portrays the asymptotic regions which are relevant to the analysis of the anti-symmetric states in the limit ϵ≪1\epsilon\ll1. As we shall see, the eigenfunctions are strongly localised to the partition-channel and partition-edge regions. In fact, our analysis in the present section will confirm that, up to exponentially small orders, the eigenfunctions involve only those regions and are therefore apparently equivalent to the BIC of the semi-infinite configuration discussed in the introduction. The unified-channel, aperture and exterior regions, where the eigenfunctions are exponentially smaller than in the partitioned-channel and partition-edge region, will be considered in §IV.

Partitioned channel

Without loss of generality, let the eigenfunction uu be of order unity in the partitioned-channel region. To analyse that region, we define the stretched coordinates x=ϵXx=\epsilon X and define u(x,y;ϵ)=U(X,y;ϵ)u(x,y;\epsilon)=U(X,y;\epsilon), with −1−l<y<−l-1-l<y<-l and, focusing on the right-hand side of the partitioned-channel region, 0<X<10<X<1. The eigenfunction UU satisfies

∂2U∂X2+ϵ2(∂2U∂y2+k2U)=0,(6)\frac{\partial^2 U}{\partial X^2}+\epsilon^2\left(\frac{\partial^2 U}{\partial y^2}+k^2U\right)=0, \tag*{(6)}

the Neumann boundary condition

∂U∂X=0atX=0,1(7)\frac{\partial U}{\partial X}=0\quad\text{at}\quad X=0,1 \tag*{(7)}

and, naively, the Neumann boundary condition

∂U∂y=0aty=−1−l.(8)\frac{\partial U}{\partial y}=0\quad\text{at}\quad y=-1-l. \tag*{(8)}

Strictly speaking, the latter condition should be verified by matching to an ‘end’ region y+1+l≃ϵy+1+l\simeq\epsilon. Nonetheless, we shall assume that (8) holds at least up to exponentially small orders for reasons noted at the end of this section. Additionally, conditions at y→−ly\to-l are to be derived by asymptotic matching with the partition-edge region.

The wavenumber is expanded as

k=k0+ϵk1+⋯asϵ→0.(9)k=k_0+\epsilon k_1+\cdots\quad\text{as}\quad\epsilon\to0. \tag*{(9)}

Similarly, the eigenfunction is expanded as

U(X,y;ϵ)=U0(X,y)+ϵU1(X,y)+ϵ2U2(X,y)+ϵ3U3(X,y)+⋯asϵ→0.(10)U(X,y;\epsilon)=U_0(X,y)+\epsilon U_1(X,y)+\epsilon^2U_2(X,y)+\epsilon^3U_3(X,y)+\cdots\quad\text{as}\quad\epsilon\to0. \tag*{(10)}

At orders unity and ϵ\epsilon, (6) gives

∂2U0∂X2=0,∂2U1∂X2=0.(11)\frac{\partial^2 U_0}{\partial X^2}=0,\qquad\frac{\partial^2 U_1}{\partial X^2}=0. \tag*{(11)}

Together with the respective orders of the Neumann conditions (7), we find

U0(X,y)=U0(y),U1(X,y)=U1(y).(12)U_0(X,y)=U_0(y),\qquad U_1(X,y)=U_1(y). \tag*{(12)}

At order ϵ2\epsilon^2, (6) gives

∂2U2∂X2+d2U0dy2+k02U0=0.(13)\frac{\partial^2 U_2}{\partial X^2}+\frac{d^2U_0}{dy^2}+k_0^2U_0=0. \tag*{(13)}

Integration with respect to XX and using the Neumann conditions (7) at the respective order, we find the one-dimensional Helmholtz equation

d2U0dy2+k02U0=0.(14)\frac{d^2U_0}{dy^2}+k_0^2U_0=0. \tag*{(14)}

One boundary condition for (14) is obtained from the leading-order balance of the Neumann condition (8),

dU0dy=0aty=−1−l.(15)\frac{dU_0}{dy}=0 \quad\text{at}\quad y=-1-l. \tag*{(15)}

A second boundary condition is obtained by anticipating that the eigenfunction is asymptotically small in the partition-edge region. (As noted in §III B, assuming otherwise leads to a contradiction in the partition-edge region.) That implies the matching condition

U0=0aty=−l.(16)U_0=0 \quad\text{at}\quad y=-l. \tag*{(16)}

Solving the eigenvalue problem (14)–(16) gives the leading-order eigenvalues,

k0=(12+n)π,n=0,1,2,…,(17)k_0=\left(\frac{1}{2}+n\right)\pi,\quad n=0,1,2,\ldots, \tag*{(17)}

and corresponding eigenfunctions

U0=cos⁡(k0(y+1+l)),(18)U_0=\cos\left(k_0(y+1+l)\right), \tag*{(18)}

where without loss of generality we choose the constant of proportionality to be unity. For simplicity we keep the mode number nn implicit in our notation for the wavenumber eigenvalue and eigenfunction.

For later reference, we note the behaviour

U0∼−σnk0(y+l)asy→−l,(19)U_0\sim-\sigma_n k_0(y+l)\quad\text{as}\quad y\to-l, \tag*{(19)}

where we define σn=(−1)n\sigma_n=(-1)^n. We also note that substitution of the one-dimensional Helmholtz equation (14) into (13) and using the order ϵ2\epsilon^2 balance of the Neumann conditions (7) shows that U2(X,y)=U2(y)U_2(X,y)=U_2(y).

Partition edge

To analyse the partition-edge region we employ the strained coordinates X=x/ϵX=x/\epsilon and Y=(y+l)/ϵY=(y+l)/\epsilon and define u(x,y;ϵ)=V(X,Y;ϵ)u(x,y;\epsilon)=V(X,Y;\epsilon), with (X,Y)(X,Y) in the domain shown on the right-hand side of Figure 4. The eigenfunction VV satisfies

Mapping from the upper half of an auxiliary complex $\zeta$ plane to the domain in the complex $Z$ plane corresponding to the partition-edge region

Figure 4. Mapping from the upper half of an auxiliary complex ζ\zeta plane to the domain in the complex ZZ plane corresponding to the partition-edge region.

∂2V∂X2+∂2V∂Y2+ε2k2V=0,(20)\frac{\partial^2 V}{\partial X^2}+\frac{\partial^2 V}{\partial Y^2}+\varepsilon^2 k^2 V=0, \tag*{(20)}

the Neumann boundary condition

∂V∂N=0,(21)\frac{\partial V}{\partial N}=0, \tag*{(21)}

where ∂/∂N=n^⋅(e^x∂/∂X+e^y∂/∂Y)\partial/\partial N=\hat{\mathbf{n}}\cdot(\hat{\mathbf{e}}_x\partial/\partial X+\hat{\mathbf{e}}_y\partial/\partial Y), and conditions to be derived by matching with the partitioned- and unified-channel regions. In particular, we anticipate that VV, up to exponential orders in ϵ\epsilon, decays exponentially as Y→+∞Y\to+\infty. We also enforce that VV is odd in XX.

The behaviour (19) suggests V≃ϵV\simeq\epsilon. Accordingly, VV is expanded as

V(X,Y;ϵ)=ϵV0(X,Y)+⋯asϵ→0.(22)V(X,Y;\epsilon)=\epsilon V_0(X,Y)+\cdots\quad\text{as}\quad\epsilon\to0. \tag*{(22)}

From (20) and (21), V0V_0 satisfies Laplace’s equation

∂2V0∂X2+∂2V0∂Y2=0(23)\frac{\partial^2 V_0}{\partial X^2}+\frac{\partial^2 V_0}{\partial Y^2}=0 \tag*{(23)}

and the Neumann condition

∂V0∂N=0,(24)\frac{\partial V_0}{\partial N}=0, \tag*{(24)}

respectively. The anticipated decay towards the unified-channel region implies

V0→0as Y→+∞(∣X∣<1),(25)V_0 \to0 \quad\text{as } Y \to+\infty\quad(|X| < 1), \tag*{(25)}

while matching the partitioned-channel expansion (10) to order unity with the partition-edge expansion (22) to order ϵ\epsilon, using (19), yields the condition

V0∼−σnk0Yas Y→−∞(0<X<1).(26)V_0 \sim-\sigma_n k_0 Y \quad\text{as } Y \to-\infty\quad(0 < X < 1). \tag*{(26)}

The corresponding condition for −1<X<0-1 < X < 0 follows from V0V_0 being odd in XX.

To solve the above potential problem, we introduce the complex variable Z=X+iYZ = X + iY, the auxiliary complex variable ζ=ξ+iη\zeta= \xi+ i\eta and the mapping [57]

Z=1+iπ[log⁡(ζ−1)+log⁡(ζ+1)](27)Z = 1 + \frac{i}{\pi} [\log(\zeta- 1) + \log(\zeta+ 1)] \tag*{(27)}

depicted in Figure 4 from the upper half of the ζ\zeta plane to the fluid domain of the partition-edge region. It is useful to derive the asymptotic behaviours of the mapping as ζ\zeta approaches infinity as well as the singular points ζ=±1\zeta= \pm1. As ζ→∞\zeta\to\infty in the upper half-plane, we find

X=1−2θπ+O(1ζ2),Y=2ln⁡∣ζ∣π+O(1ζ2),(28)X = 1 - \frac{2\theta}{\pi} + O\left(\frac{1}{\zeta^2}\right), \qquad Y = \frac{2\ln|\zeta|}{\pi} + O\left(\frac{1}{\zeta^2}\right), \tag*{(28)}

where θ=arg⁡(ζ)\theta= \arg(\zeta) with 0≤θ≤π0 \le\theta\le\pi. As ζ→1\zeta\to1 in the upper half-plane, we find

X=1−φπ+O(ζ−1),Y=1πln⁡∣ζ−1∣+ln⁡2π+O(ζ−1),(29)X = 1 - \frac{\varphi}{\pi} + O(\zeta- 1), \qquad Y = \frac{1}{\pi}\ln|\zeta- 1| + \frac{\ln2}{\pi} + O(\zeta- 1), \tag*{(29)}

where φ=arg⁡(ζ−1)\varphi= \arg(\zeta- 1) with 0≤φ≤π0 \le\varphi\le\pi. The corresponding behaviour as ζ→−1\zeta\to-1 readily follows from the symmetry of the mapping about the η\eta axis, with Z(±ξ,η)=±X+iYZ(\pm\xi,\eta) = \pm X + iY.

Let V0(X,Y)=v(ξ,η)V_0(X,Y) = v(\xi,\eta). The harmonic function v(ξ,η)v(\xi,\eta) is anti-symmetric about the η\eta axis, decays as ∣ζ∣→∞|\zeta| \to\infty in the upper half-plane and satisfies a Neumann condition on η=0\eta= 0 except at the singular points ζ=±1\zeta= \pm1, where (26) and (29) together give

v∼∓σnk0πln⁡∣ζ∓1∣as ζ→±1.(30)v \sim\mp\frac{\sigma_n k_0}{\pi} \ln|\zeta\mp1| \quad\text{as } \zeta\to\pm1. \tag*{(30)}

The latter condition, which effectively prescribes opposite monopoles at ζ=±1\zeta= \pm1, can also be derived by matching fluxes between the partitioned-channel partition-edge regions. The solution to the problem governing vv is readily seen to be

v=−σnk0π(ln⁡∣ζ−1∣−ln⁡∣ζ+1∣).(31)v = -\frac{\sigma_n k_0}{\pi}(\ln|\zeta- 1| - \ln|\zeta+ 1|). \tag*{(31)}

The limit

β=lim⁡Y→−∞{V0+σnk0Y},(32)\beta= \lim_{Y \to-\infty} \{V_0 + \sigma_n k_0 Y\}, \tag*{(32)}

taken in 0<X<10<X<1, will be required at the next order of the analysis. Using (29) and (31), we find

β=σnk0ln⁡4π.(33)\beta= \frac{\sigma_n k_0 \ln4}{\pi}. \tag*{(33)}

Another output of the solution (31) is the behaviour

V0∼2σnk0πsin⁡πX2exp⁡(−πY2)as Y→+∞(∣X∣<1),(34)V_0 \sim\frac{2\sigma_n k_0}{\pi}\sin\frac{\pi X}{2}\exp\left(-\frac{\pi Y}{2}\right) \quad\text{as } Y\to+\infty\quad(|X|<1), \tag*{(34)}

which is derived using (28). We will use this behaviour in §IV to match with the unified-channel region. For the time being, it serves to confirm that the eigenfunction is exponentially small in the unified-channel, aperture and exterior regions.

We remark in passing that if we had not assumed in §III A that U0(−l)=0U_0(-l)=0, then VV would be order unity; a leading approximation would satisfy the same problem as V0V_0 above, only that it would require to approach the non-zero constants ±U0(−l)\pm U_0(-l) as Y→−∞Y\to-\infty, respectively in 0<±X<10<\pm X<1. The latter problem, however, does not possess a solution which is why we rejected this possibility.

Higher-order analysis

Returning to the right-hand side of the partitioned-channel region, we find from the order ϵ3\epsilon^3 balance of (6) that U1U_1 satisfies

d2U1dy2+k02U1+∂2U3∂X2=−2k0k1U0.(35)\frac{d^2U_1}{dy^2}+k_0^2U_1+\frac{\partial^2U_3}{\partial X^2}=-2k_0k_1U_0. \tag*{(35)}

Integration with respect to XX, together with the respective order of the Neumann boundary conditions (7), we find the inhomogeneous one-dimensional Helmholtz equation

d2U1dy2+k02U1=−2k0k1U0.(36)\frac{d^2U_1}{dy^2}+k_0^2U_1=-2k_0k_1U_0. \tag*{(36)}

This equation is supplemented by the Neumann condition

dU1dy=0aty=−1−l,(37)\frac{dU_1}{dy}=0 \quad\text{at} \quad y=-1-l, \tag*{(37)}

which follows from the order ϵ\epsilon balance of the Neumann condition (8), and the condition

U1=βaty=−l,(38)U_1=\beta\quad\text{at} \quad y=-l, \tag*{(38)}

which follows from matching the partition-edge expansion (22) and the partitioned-channel expansion (10) to order ϵ\epsilon, using (32).

The problem for U1U_1 is the same as the singular homogeneous problem for U0U_0, with the addition of forcing terms. A solution to the former problem therefore exists only under special conditions on those forcing terms. Subtracting the complex conjugate of (14) multiplied by U1U_1 from (36) multiplied by U0∗U_0^*, where an asterisk denotes complex conjugation, and using integration by parts and the boundary conditions (15), (16), (37) and (38), we find the solvability condition

βdU0∗dy∣y=−l=2k0k1∫−1−l−l∣U0∣2 dy.(39)\left.\beta\frac{dU_0^*}{dy}\right|_{y=-l}=2k_0k_1\int_{-1-l}^{-l}|U_0|^2\,dy. \tag*{(39)}

Substituting (17) and (18), we find the first wavenumber correction

k1=−k0ln⁡4π.(40)k_1=-k_0\frac{\ln4}{\pi}. \tag*{(40)}

Combining (17) and (40), the approximation for the wavenumber can be written

k−k0k0∼−ϵln⁡4πasϵ→0.(41)\frac{k-k_0}{k_0}\sim-\epsilon\frac{\ln4}{\pi}\quad\text{as}\quad\epsilon\to0. \tag*{(41)}

This approximation is compared with numerical results in Figure 5. We also note that (41) is asymptotically consistent with the higher-order approximations in implicit form obtained by Evans, Linton and Ursell [30] for the BIC of the semi-infinite slit geometry seen in Figure 2. We note that the approximations in [30] are derived intuitively by alluding to exact Wiener-Hopf solutions to the Helmholtz equation for a semi-infinite geometry corresponding to our partition-edge region.

Relative shift of the wavenumber as a function of $\epsilon$

Figure 5. The relative shift of the wavenumber kk from its leading approximation k0k_0 as a function of ϵ\epsilon, for the fundamental anti-symmetric mode n=0n = 0 for which k0=π/2k_0 = \pi/2. Solid curve: asymptotic approximation (41). Symbols: numerical results for the indicated values of ll.

There is no conceptual difficulty in extending the analysis of the present section, involving solely the partitioned-channel and partition-edge regions, to higher orders. In particular, it is clear that to arbitrary power of ϵ\epsilon the partitioned-channel eigenfunction remains transversally uniform (thus justifying the use of the Neumann boundary condition (8)) and that the partition-edge eigenfunction decays exponentially, though with compressibility effects entering at order ϵ3\epsilon^3 as a perturbation. Working through these details would furnish an explicit, real-valued, infinite asymptotic expansion in powers of ϵ\epsilon for kk. Instead, we next derive a leading-order approximation for k′′k^{\prime\prime} which is beyond all orders of the expansion for kk.

Quasi-bound states in the continuum

Unified channel

In this section we extend the asymptotic description of the anti-symmetric modes into the unified-channel and exterior regions, where the eigenfunction exponentially small. Together with the reciprocal identity (5), this will allow us to calculate a leading approximation for k′′k^{\prime\prime}. We start with the unified-channel region, where we employ the coordinates (X,y)(X,y), with −l<y<0-l<y<0 and ∣X∣<1|X|<1, and define u(x,y;ϵ)=W(X,y;ϵ)u(x,y;\epsilon)=W(X,y;\epsilon). The eigenfunction WW satisfies the equation

∂2W∂X2+ϵ2∂2W∂y2+ϵ2k2W=0,(42)\frac{\partial^2 W}{\partial X^2}+\epsilon^2\frac{\partial^2 W}{\partial y^2}+\epsilon^2 k^2 W=0, \tag*{(42)}

the Neumann conditions

∂W∂X=0at X=±1(43)\frac{\partial W}{\partial X}=0\quad\text{at }X=\pm1 \tag*{(43)}

and conditions as y→−ly\to-l and y→0y\to0 to be derived from asymptotic matching with the partition-edge and aperture regions, respectively. We also enforce that WW is odd in XX.

Like the partitioned-channel region, the unified-channel region corresponds to a channel of width ≃ϵ\simeq\epsilon and length ≃1\simeq1. In contrast to the partitioned-channel region, however, the absence of the partition in the unified-channel together with the anti-symmetry of the eigenfunction precludes wave propagation. Rather, the eigenfunction in the unified-channel region is, to leading-order, non-uniform in the transverse direction, attenuating exponentially away from the partition edge. An asymptotic expansion in this region can be derived using a WKB ansatz [23]; alternatively, it can be extracted from an exact separation-of-variables series solution to the Helmholtz equation (42) in the rectangular domain corresponding to the unified-channel region. Given (34) and the need to match with the partition-edge region, we find the leading-order approximation

W(X,y;ϵ)=ϵ2σnk0πsin⁡πX2exp⁡{−π2ϵ(y+l)}+⋯as ϵ→0.(44)W(X,y;\epsilon)=\epsilon\frac{2\sigma_n k_0}{\pi}\sin\frac{\pi X}{2}\exp\left\{-\frac{\pi}{2\epsilon}(y+l)\right\}+\cdots\quad\text{as }\epsilon\to0. \tag*{(44)}

We note that the above approximation is harmonic even though the unified-channel region is not subwavelength.

Aperture

We next consider the aperture region bridging the unified-channel and exterior regions. To study this region, we employ the strained coordinates XX and Y~=y/ϵ\widetilde{Y}=y/\epsilon and define u(x,y;ϵ)=S(X,Y~;ϵ)u(x,y;\epsilon)=S(X,\widetilde{Y};\epsilon). The domain of XX and Y~\widetilde{Y} is as shown on the right-hand side of Figure 6. In that domain, the eigenfunction SS satisfies the Helmholtz equation

Mapping from the upper half of an auxiliary complex $\zeta$ plane to the domain in the complex $\widetilde{Z}$ plane corresponding to the left-hand side of the aperture region.

Figure 6. Mapping from the upper half of an auxiliary complex ζ\zeta plane to the domain in the complex Z~\widetilde{Z} plane corresponding to the left-hand side of the aperture region.

∂2S∂X2+∂2S∂Y~2+ϵ2k02S=0,(45)\frac{\partial^2 S}{\partial X^2}+\frac{\partial^2 S}{\partial\widetilde{Y}^2}+\epsilon^2 k_0^2 S=0, \tag*{(45)}

the Neumann condition

∂S∂N=0(46)\frac{\partial S}{\partial N}=0 \tag*{(46)}

and far-field conditions to be derived by matching with the unified-channel and exterior regions. We also enforce that SS is odd in XX.

The form of the unified-channel expansion (44) implies that SS should be expanded as

S=ϵexp⁡(−πl2ϵ)S0(X,Y~)+⋯as ϵ→0.(47)S=\epsilon\exp\left(-\frac{\pi l}{2\epsilon}\right)S_0(X,\widetilde{Y})+\cdots\quad\text{as }\epsilon\to0. \tag*{(47)}

From (1), S0S_0 satisfies Laplace’s equation

∂2S0∂X2+∂2S0∂Y~2=0,(48)\frac{\partial^2 S_0}{\partial X^2}+\frac{\partial^2 S_0}{\partial\widetilde{Y}^2}=0, \tag*{(48)}

while (2) gives the Neumann condition

∂S0∂N=0.(49)\frac{\partial S_0}{\partial N}=0. \tag*{(49)}

With regards to matching with the exterior region, the anti-symmetry of the eigenfunction implies the attenuation condition

S0→0asX2+Y~2→∞(Y~>0).(50)S_0 \to0 \quad\text{as}\quad X^2+\widetilde{Y}^2\to\infty\quad(\widetilde{Y}>0). \tag*{(50)}

Indeed, the far-field of an anti-symmetric harmonic function in a half-plane domain either grows algebraically or decays algebraically. The former possibility is non-sensical as it would imply that the exterior eigenfunction is asymptotically larger than the aperture eigenfunction; indeed, only a decaying aperture eigenfunction is consistent with the exterior eigenfunction being outward-radiating. A last condition on S0S_0 is obtained by matching the leading orders of the unified-channel expansion (44) and the aperture expansion (47). This gives

S0∼σn2k0πsin⁡πX2exp⁡{−π2Y~}asY~→−∞(∣X∣<1).(51)S_0\sim\sigma_n\frac{2k_0}{\pi}\sin\frac{\pi X}{2}\exp\left\{-\frac{\pi}{2}\widetilde{Y}\right\}\quad\text{as}\quad\widetilde{Y}\to-\infty\quad(|X|<1). \tag*{(51)}

To solve for S0S_0 we introduce the complex variable Z~=X+iY~\tilde{Z}=X+i\tilde{Y}, the auxiliary variable ζ=ξ+iη\zeta=\xi+i\eta and the mapping [57]

Z~=2iπ(1+ζ)1/2+iπlog⁡(1+ζ)1/2−1(1+ζ)1/2+1(52)\tilde{Z}=\frac{2i}{\pi}(1+\zeta)^{1/2}+\frac{i}{\pi}\log\frac{(1+\zeta)^{1/2}-1}{(1+\zeta)^{1/2}+1} \tag*{(52)}

from the upper half of the complex ζ\zeta plane to the left-hand side of the aperture domain (see Figure 6). As ζ→0\zeta\to0 in the upper half-plane,

X=−θπ+o(1),Y~=1πln⁡∣ζ∣+2π(1−ln⁡2)+o(1).(53)X=-\frac{\theta}{\pi}+o(1),\qquad\tilde{Y}=\frac{1}{\pi}\ln|\zeta|+\frac{2}{\pi}(1-\ln2)+o(1). \tag*{(53)}

where θ=arg⁡ζ\theta=\arg\zeta with 0≤θ≤π0\leq\theta\leq\pi. As ζ→∞\zeta\to\infty in the upper half-plane,

Z~=2iπζ1/2+o(1ζ1/2).(54)\tilde{Z}=\frac{2i}{\pi}\zeta^{1/2}+o\left(\frac{1}{\zeta^{1/2}}\right). \tag*{(54)}

Let S0(X,Y~)=s(ξ,η)S_0(X,\tilde{Y})=s(\xi,\eta). The harmonic function s(ξ,η)s(\xi,\eta) satisfies a decay condition as ∣ζ∣→∞|\zeta|\to\infty in the upper half-plane, a Neumann condition on the negative ξ\xi axis, a Dirichlet condition on the positive ξ\xi axis and the matching condition

s∼−4k0σnsin⁡(θ/2)πe∣ζ∣1/2as ζ→0(55)s\sim-\frac{4k_0\sigma_n\sin(\theta/2)}{\pi e|\zeta|^{1/2}}\quad\text{as }\zeta\to0 \tag*{(55)}

in the upper half-plane, which is derived from (51) and (53). It is easy to see that the solution is identical to that asymptotic behaviour, i.e.,

s=−4k0σnsin⁡(θ/2)πe∣ζ∣1/2.(56)s=-\frac{4k_0\sigma_n\sin(\theta/2)}{\pi e|\zeta|^{1/2}}. \tag*{(56)}

The solution can also be viewed as being proportional to Im⁡(1/ζ1/2)\operatorname{Im}(1/\zeta^{1/2}).

In preparation for matching with the exterior region, we use (54) to extract from the solution (56) the far-field behaviour

S0∼8k0σncos⁡ϕπ2eR~as R~→∞,(57)S_0\sim\frac{8k_0\sigma_n\cos\phi}{\pi^2 e\tilde{R}}\quad\text{as }\tilde{R}\to\infty, \tag*{(57)}

where R~2=X2+Y~2\tilde{R}^2=X^2+\tilde{Y}^2 and ϕ\phi is the angle measured counter-clockwise from the XX direction (i.e., identical to the angle defined in §II). The behaviour (57) represents a potential dipole.

Exterior

Consider now the exterior region, where the unstrained coordinates (x,y)(x,y), alternatively (r,ϕ)(r,\phi), are held fixed and the eigenfunction is denoted, as in the exact problem formulation, by uu. The exterior problem consists of the Helmholtz equation (1) in the half-plane y>0y>0, with a far-field outward-radiation condition, a Neumann condition at y=0y=0 (x≠0x\ne0) and conditions as r→0r\to0 to be derived from matching with the aperture region.

In particular, the aperture expansion (47) together with the dipolar behaviour (57) suggests expanding the exterior region as

u(r,ϕ;ϵ)=ϵ2exp⁡(−πl2ϵ)u0(r,ϕ)+⋯as ϵ→0.(58)u(r,\phi;\epsilon)=\epsilon^2\exp\left(-\frac{\pi l}{2\epsilon}\right)u_0(r,\phi)+\cdots\quad\text{as }\epsilon\to0. \tag*{(58)}

From (1), u0u_0 satisfies the Helmholtz equation

∇2u0+k02u0=0,(59)\nabla^2u_0+k_0^2u_0=0, \tag*{(59)}

while (2) gives the Neumann condition

∂u0∂y=0at y=0(x≠0).(60)\frac{\partial u_0}{\partial y}=0\quad\text{at }y=0\quad(x\ne0). \tag*{(60)}

Using (57) to match the leading orders of the aperture expansion (47) and exterior expansion (58), we find the singular condition

u0∼8k0σncos⁡ϕπ2ϵ1ras r→0.(61)u_0\sim\frac{8k_0\sigma_n\cos\phi}{\pi^2\epsilon}\frac{1}{r}\quad\text{as }r\to0. \tag*{(61)}

Lastly, we have that u0u_0 satisfies an outward-radiation condition as r→∞r\to\infty.

The solution is readily seen to be

u0=4ik02σnπϵH1(1)(k0r)cos⁡ϕ,(62)u_0=\frac{4ik_0^2\sigma_n}{\pi\epsilon}H_1^{(1)}(k_0r)\cos\phi, \tag*{(62)}

where H1(1)H_1^{(1)} denotes the first-order Hankel function of the first kind [60]. The far-field of u0u_0 can be extracted from (62) using the large-argument asymptotic behaviour of the Hankel function [60]. This gives the dipolar radiation field

u0∼21/2k03/2σnϵ−iπ/4cos⁡ϕπ3/2ϵr1/2eik0ras r→∞.(63)u_0\sim\frac{2^{1/2}k_0^{3/2}\sigma_n\epsilon^{-i\pi/4}\cos\phi}{\pi^{3/2}\epsilon r^{1/2}}e^{ik_0r}\quad\text{as }r\to\infty. \tag*{(63)}

We stress that the exterior expansion is derived in the limit ϵ→0\epsilon\to0 with rr fixed. It clearly breaks down at large distances r≃1/ϵr\simeq1/\epsilon, where the wavenumber correction k1k_1 becomes important to leading order. At larger and larger distances higher corrections of the wavenumber become important; eventually, at exponentially large distances r≃1/∣k′′∣r\simeq1/|k''|, the imaginary part of the wavenumber enters the leading-order balance and the eigenfunction begins to grow exponentially, as one would expect for a quasi-normal mode [24].

Imaginary part of the wavenumber

We are now ready to calculate a leading-order approximation for k′′k^{\prime\prime}. To this end, we employ the reciprocal relation (5) with 1≪λ≪1/ϵ1 \ll\lambda\ll1/\epsilon. Then (58) together with (63) gives

λ∫0π(u∗∂u∂r−u∂u∗∂r)r=λ dϕ∼32ik04π2ϵ2ϵ4exp⁡(−πlϵ)as ϵ→0.(64)\lambda\int_{0}^{\pi} \left(u^* \frac{\partial u}{\partial r} - u \frac{\partial u^*}{\partial r}\right)_{r=\lambda} \,d\phi\sim\frac{32 i k_0^4}{\pi^2 \epsilon^2}\epsilon^4 \exp\left(-\frac{\pi l}{\epsilon}\right) \quad\text{as } \epsilon\to0. \tag*{(64)}

The area integral in (5) is dominated by the partitioned-channel region. Thus, using (10) and (18) we find

∫Dλ∣u∣2 dA∼2ϵ∫−1−l−l∣U0∣2 dy=ϵas ϵ→0.(65)\int_{D_\lambda} |u|^2\,dA \sim2\epsilon\int_{-1-l}^{-l} |U_0|^2\,dy = \epsilon\quad\text{as } \epsilon\to0. \tag*{(65)}

With (64) and (65), the reciprocal relation (5) yields

k′′∼8k03ϵ3ϵ2π2exp⁡(−πlϵ)as ϵ→0.(66)k^{\prime\prime} \sim\frac{8k_0^3\epsilon^3}{\epsilon^2\pi^2}\exp\left(-\frac{\pi l}{\epsilon}\right) \quad\text{as } \epsilon\to0. \tag*{(66)}

As expected, k′′k^{\prime\prime} is seen to be exponentially small, the order of exponential smallness being proportional to l/ϵl/\epsilon. In Figure 7, the asymptotic approximation (66) is compared with numerical results.

Plots of the imaginary part of the wavenumber

Figure 7. The imaginary part k′′k^{\prime\prime} of the wavenumber kk for the fundamental anti-symmetric mode n=0n=0. Left: ratio of the asymptotic approximation (66) for k′′k^{\prime\prime} to numerically computed values as a function of ϵ\epsilon and for the indicated values of ll. Right: k′′k^{\prime\prime} and ϵ\epsilon scaled so as to collapse numerical data for different ll on a single analytical curve implied by (66).

Resonant excitation by oblique plane wave

Scattering problem

In order to illustrate the significance of the quasi-BIC studied in the preceding sections, we next consider the associated scattering problem described in the Introduction where these states are resonantly excited — it involves the same geometry and physical assumptions as for the eigenvalue problem formulated in §II, only with a real dimensionless wavenumber κ\kappa and a plane wave at oblique incidence α\alpha as shown in Figure 1b. Our analysis of this scattering problem will focus on the resonant excitation of the quasi-BIC for small ϵ\epsilon and will accordingly build on the results already obtained for the eigenvalue problem.

We denote by vv the wave field normalised by the amplitude of the incident plane wave. It satisfies the Helmholtz equation

∇2v+κ2v=0,(67)\nabla^2 v+\kappa^2v=0, \tag*{(67)}

the Neumann condition

∂v∂n=0(68)\frac{\partial v}{\partial n}=0 \tag*{(68)}

and an outward radiation condition as r→∞r\to\infty in the exterior half-plane, on v−viv-v_i, where

vi=exp⁡{iκ(xsin⁡α−ycos⁡α)}(69)v_i=\exp\{i\kappa(x\sin\alpha-y\cos\alpha)\} \tag*{(69)}

is the incident plane wave.

It is convenient to decompose the total field as

v=va+vs,(70)v=v_a+v_s, \tag*{(70)}

in which

va=2cos⁡(κycos⁡α)exp⁡{iκxsin⁡α}(71)v_a=2\cos(\kappa y\cos\alpha)\exp\{i\kappa x\sin\alpha\} \tag*{(71)}

is the ambient field, namely the sum of the incident field viv_i and the reflection from the flat substrate in the absence of the slit. The ambient field satisfies the Helmholtz equation (67) with a Neumann condition on y=0y=0. The scattered field vsv_s satisfies the Helmholtz equation (67), a radiation condition as r→∞r\to\infty in the exterior half-plane and an inhomogeneous Neumann condition.

The incident field viv_i can be decomposed into parts which are symmetric and anti-symmetric with respect to the yy axis, respectively. The total field vv and ambient field vav_a can be similarly decomposed and there is no coupling between the symmetric and anti-symmetric fields. Thus, for normal incidence the fields are perfectly symmetric thence the quasi-BIC, which are anti-symmetric, cannot be excited. We therefore restrict our attention to incidence angles in the range 0<α<π/20 < \alpha< \pi/2.

Reciprocal relation

We shall employ a reciprocal relation between the scattering problem and the eigenvalue problem of §II. To derive it, we apply Green’s second identity to v∗v^*, the complex conjugate of the total field, and an eigenfunction uu over the domain Dλ\mathcal{D}_{\lambda} defined in §II B (Figure 3a). Together with (1), (2), (67) and (68), we find

(κ2−k2)∫Dλv∗u dA−∫0πλ(v∗∂u∂r−u∂v∗∂r)r=λ dϕ=0.(72)(\kappa^2-k^2)\int_{\mathcal{D}_{\lambda}}v^*u\,\mathrm{d}A-\int_0^\pi\lambda\left(v^*\frac{\partial u}{\partial r}-u\frac{\partial v^*}{\partial r}\right)_{r=\lambda}\,\mathrm{d}\phi=0. \tag*{(72)}

Substituting the decomposition (70), we can write (72) in the form

(κ2−k2)∫Dλvs∗u dA−∫0πλ(vs∗∂u∂r−u∂vs∗∂r)r=λ dϕ=F,(73)(\kappa^2-k^2)\int_{\mathcal{D}_{\lambda}}v_s^*u\,\mathrm{d}A-\int_0^\pi\lambda\left(v_s^*\frac{\partial u}{\partial r}-u\frac{\partial v_s^*}{\partial r}\right)_{r=\lambda}\,\mathrm{d}\phi=\mathcal{F}, \tag*{(73)}

where we define the ‘forcing’ term

F=∫0πλ(va∗∂u∂r−u∂va∗∂r)r=λ dϕ−(κ2−k2)∫Dλva∗u dA,(74)\mathcal{F}=\int_0^\pi\lambda\left(v_a^*\frac{\partial u}{\partial r}-u\frac{\partial v_a^*}{\partial r}\right)_{r=\lambda}\,\mathrm{d}\phi-(\kappa^2-k^2)\int_{\mathcal{D}_{\lambda}}v_a^*u\,\mathrm{d}A, \tag*{(74)}

which is linear in the ambient field.

Resonance

In analysing the scattering problem, we shall restrict our attention to studying the resonant excitation of the anti-symmetric quasi-BIC in the limit ϵ≪1\epsilon\ll1. By resonant excitation we mean that throughout the domain the scattered field can be expanded as

vs=Au+⋯as ϵ→0,(75)v_s=\mathcal{A}u+\cdots\quad\text{as }\epsilon\to0, \tag*{(75)}

where A\mathcal{A} is a complex amplitude which remains to be determined and is expected to depend on the geometric parameters ϵ\epsilon and ll as well as the mode number nn. Given that k′′k'' is exponentially small for the quasi-BIC, we expect such resonance excitation to occur in exponentially narrow wavenumber intervals. That is, for κ\kappa such that

κ−k′=o(ϵm)asϵ→0(76)\kappa-k'=o\left(\epsilon^m\right)\quad\text{as}\quad\epsilon\to0 \tag*{(76)}

for all real mm. Our analysis will confirm the existence of these resonance intervals and uncover their precise exponential scaling.

We need to asymptotically estimate the integrals appearing in (73) and (74) in the limit ϵ→0\epsilon\to0, with κ\kappa satisfying (76) and for some mode number nn. As in §IV D, we choose 1≪λ≪1/ϵ1\ll\lambda\ll1/\epsilon. Then, using (75), leading-order approximations for the two integrals in (73) follow from (64) and (65), respectively, up to the multiplicative factor A∗\mathcal{A}^{*}. Together with (66) and (76), these results give

∫Dλvs∗u dA∼ϵA∗,∫0πλ(vs∗∂u∂r−u∂vs∗∂r)r=λ dϕ∼−4ik0k′′×ϵA∗asϵ→0.(77)\int_{\mathcal{D}_{\lambda}}v_s^{*}u\,dA\sim\epsilon\mathcal{A}^{*},\qquad\int_{0}^{\pi}\lambda\left(v_s^{*}\frac{\partial u}{\partial r}-u\frac{\partial v_s^{*}}{\partial r}\right)_{r=\lambda}\,d\phi\sim-4ik_0k^{\prime\prime}\times\epsilon\mathcal{A}^{*}\quad\text{as}\quad\epsilon\to0. \tag*{(77)}

Moving to the integrals appearing in (74), we note that the area integral is order unity, hence negligible relative to (77) since for (75) to be valid in the exterior region A\mathcal{A} must be asymptotically large. In that case, F\mathcal{F} is dominated by the first integral in (74), which can be approximated using (58), the exterior expansion for uu. Thus,

F∼ϵ2exp⁡(−πl2ϵ)×∫0πλ(va∗∂u0∂r−u0∂va∗∂r)r=λ dϕasϵ→0.(78)\mathcal{F}\sim\epsilon^{2}\exp\left(-\frac{\pi l}{2\epsilon}\right)\times\int_{0}^{\pi}\lambda\left(v_a^{*}\frac{\partial u_0}{\partial r}-u_0\frac{\partial v_a^{*}}{\partial r}\right)_{r=\lambda}\,d\phi\quad\text{as}\quad\epsilon\to0. \tag*{(78)}

To evaluate the latter integral, we use Jacobi-Anger expansions [60] to write the ambient field (71) as the Fourier series

va=2J0(κr)+4∑m=1∞ΛmJm(κr)cos⁡mϕ,(79)v_a=2J_0(\kappa r)+4\sum_{m=1}^{\infty}\Lambda_mJ_m(\kappa r)\cos m\phi, \tag*{(79)}

where JmJ_m denotes the Bessel function of order mm and

Λm=χmcos⁡mα+i(1−χm)sin⁡mα,(80)\Lambda_m=\chi_m\cos m\alpha+i(1-\chi_m)\sin m\alpha, \tag*{(80)}

χm\chi_m being unity for mm even and zero for mm odd. Since u0u_0 involves only the cos⁡ϕ\cos\phi Fourier mode, and given the orthogonality properties of the Fourier modes, only the term 4Λ1J1(κr)cos⁡ϕ4\Lambda_1J_1(\kappa r)\cos\phi in vav_a contributes to the integral on the right-hand side of (78), with (80) giving Λ1=isin⁡α\Lambda_1=i\sin\alpha. Using (63) and (76), in conjunction with the large-argument behaviour of J1J_1 [60], we find

F∼16ik02σnsin⁡απϵϵ2exp⁡(−πl2ϵ)asϵ→0.(81)\mathcal{F}\sim\frac{16ik_0^2\sigma_n\sin\alpha}{\pi\epsilon}\epsilon^2\exp\left(-\frac{\pi l}{2\epsilon}\right)\quad\text{as}\quad\epsilon\to0. \tag*{(81)}

Substituting (77) and (81) into the reciprocal relation (73), we find the requisite asymptotic amplitude-wavenumber relation

A=−8iϵσnk0sin⁡απϵexp⁡(−πl2ϵ)κ−k′−ik′′.(82)\mathcal{A}=-\frac{8\mathrm{i}\epsilon\sigma_n k_0\sin\alpha}{\pi\epsilon}\frac{\exp\left(-\frac{\pi l}{2\epsilon}\right)}{\kappa-k'-\mathrm{i}k''}. \tag*{(82)}

In the denominator, it is not allowed to assume a leading approximation for k′k', or even any finite number of terms in its expansion; this is because κ\kappa and k′k' cancel out up to the exponentially small ϵ3exp⁡(−πl/ϵ)\epsilon^3\exp(-\pi l/\epsilon) order of k′′k''. Since va≃1v_a\simeq1 in the exterior region, and given (82) and the scaling of k′′k'', the resonance condition (75) requires

κ−k′=O(ϵ3e−πlϵ).(83)\kappa-k'=O\left(\epsilon^3\mathrm{e}^{-\frac{\pi l}{\epsilon}}\right). \tag*{(83)}

This refines our initial estimate (76) for the resonance regime, in which (75) holds with A\mathcal{A} given by (82). Note that in that regime the scaling of A\mathcal{A} is exponentially large:

A≃1ϵ2exp⁡(πl2ϵ).(84)\mathcal{A}\simeq\frac{1}{\epsilon^2}\exp\left(\frac{\pi l}{2\epsilon}\right). \tag*{(84)}

In particular, at κ=k′\kappa=k', namely at resonance, A\mathcal{A} attains the value (cf. (66))

A(r)=−σnπϵsin⁡αϵ2k02exp⁡(πl2ϵ).(85)\mathcal{A}^{(r)}=-\frac{\sigma_n\pi\epsilon\sin\alpha}{\epsilon^2 k_0^2}\exp\left(\frac{\pi l}{2\epsilon}\right). \tag*{(85)}

The resonance condition (75), together with the asymptotic amplitude-wavenumber relation (82) and the matched asymptotic expansions for the quasi-BIC eigenfunctions allows calculating a leading-order asymptotic approximation for the scattered field in all regions. In particular, from (10) and (18) we see that ∣A∣|\mathcal{A}| directly corresponds to the magnitude of the standing wave in the partitioned channel, which is accordingly exponentially large at resonance. Furthermore, using (58) and (62) we see that in the exterior region the scattered field possesses the asymptotic approximation

vs∼32k03ϵ3exp⁡(−πl/ϵ)sin⁡απ2ϵ2(κ−k′−ik′′)H1(1)(k0r)cos⁡ϕas ϵ→0.(86)v_s\sim\frac{32k_0^3\epsilon^3\exp(-\pi l/\epsilon)\sin\alpha}{\pi^2\epsilon^2(\kappa-k'-\mathrm{i}k'')}H_1^{(1)}(k_0r)\cos\phi\quad\text{as }\epsilon\to0. \tag*{(86)}

At resonance, substituting (66) shows that (86) reduces to

vs(r)∼−4isin⁡αH1(1)(k0r)cos⁡ϕas ϵ→0.(87)v_s^{(r)}\sim-4\mathrm{i}\sin\alpha H_1^{(1)}(k_0r)\cos\phi\quad\text{as }\epsilon\to0. \tag*{(87)}

Notably, the magnitude 4sin⁡α4\sin\alpha of the scattered dipole at resonance is independent of both ll and ϵ\epsilon. The fact that the scattering magnitude at resonance approaches a constant as ϵ→0\epsilon\to0, namely as the scatterer size vanishes, means that the scatterer becomes increasingly efficient in that limit.

We finish this section by comparing the above asymptotic approximations with numerical results. In our numerical scheme, the field in the right-hand side, say, of the partitioned portion of the slit is expanded in terms of eigenfunctions, the fundamental eigenfunction being the xx-independent standing wave cos⁡(κ(y+1+l))\cos(\kappa(y + 1 + l)); at a quasi-BIC resonance, this fundamental eigenfunction dominates the numerical solution. From (75) and (18), the corresponding asymptotic approximation at resonance for the scattered (and total) field in the partitioned-channel region is A(r)cos⁡(k0(y+1+l))\mathcal{A}^{(r)}\cos(k_0(y + 1 + l)). Given (76), it is accordingly meaningful to compare A(r)\mathcal{A}^{(r)} directly with the coefficient multiplying the fundamental eigenfunction in the numerical scheme. This comparison is shown in Figure 8. Instead comparing A(r)\mathcal{A}^{(r)} with the numerical wave field averaged over the right-hand side of the bottom boundary of the slit gives results which are indistinguishable from those presented.

The amplitude at resonance for incidence angle $\alpha = \pi/4$

Figure 8. The amplitude at resonance A(r)\mathcal{A}^{(r)} for incidence angle α=π/4\alpha= \pi/4. Left: ratio of the asymptotic approximation (85) to numerically computed values as a function of ϵ\epsilon and for the indicated values of ll. Right: A(r)\mathcal{A}^{(r)} and ϵ\epsilon scaled so as to collapse numerical data for different ll on a single analytical curve implied by (85).

Furthermore, we can extract from our numerical scheme the coefficients cmc_m in a conventional multipole expansion of the scattered far-field,

vs=∑m=0∞cmHm(1)(κr)cos⁡mφ,(88)v_s = \sum_{m=0}^{\infty} c_m H_m^{(1)}(\kappa r) \cos m\varphi, \tag*{(88)}

where Hm(1)H_m^{(1)} denotes the mm-th order Hankel function of the first kind and cmc_m are diffraction coefficients. We refer to c0c_0 and c1c_1 as the monopole and dipole coefficients, respectively. Our asymptotic results imply that near the quasi-BIC resonances the dipolar term in (88) is dominant, with an asymptotic approximation for c1c_1 following from (86). As shown in Figure 9, we have confirmed numerically that ∣c1∣/sin⁡α\lvert c_1\rvert/\sin\alpha approaches 4 at resonance as predicted by (87). For small ϵ\epsilon, the remaining coefficients are, as expected, small when evaluated at the quasi-BIC resonances.

Magnitude at resonance of the dipole coefficient $c_1$

Figure 9. The magnitude at resonance of the dipole coefficient c1c_1 in the multipole expansion (88) of the scattered wave for incidence angle α=π/4\alpha= \pi/4, plotted as a function of ϵ\epsilon. Symbols: numerical values for indicated values of ll. Solid line: the ll- and ϵ\epsilon-independent asymptotic value 4 (cf. (87)).

Concluding remarks

Our analysis of the setup shown in Figure 1 has exclusively focused on the anti-symmetric quasi-BIC states and their resonant excitation. To put these resonances in a wider perspective, we plot in Figure 10 the first two diffraction coefficients in the multipole expansion (88), calculated numerically, as a function of the wavenumber κ\kappa, for α=π/4\alpha= \pi/4, l=1l = 1 and ϵ=0.5\epsilon= 0.5. The quasi-BIC resonances correspond to the extremely narrow resonance peak in the anti-symmetric cylindrical-dipolar response. They are closely followed by anti-resonances, a signature of so-called Fano-type resonances associated with the interaction between an isolated quasi-bound mode and a continuous-spectrum of propagating modes [61]; the anti-resonances occur outside the exponentially narrow resonance interval and are therefore not captured by our asymptotic analysis. The symmetric cylindrical-monopolar response is seen to exhibit less narrowband resonances; these correspond to the conventional standing-wave resonances of a slit channel open on one side, which are not perturbed at all by the presence of the thin partition. These resonances occur at wavenumbers κ/a≈(π/(1+l))(1/2+nˉ)\kappa/a \approx(\pi/(1+l))(1/2+\bar{n}), where nˉ=0,1,2,…\bar{n}=0,1,2,\ldots. Closely related resonances have been studied asymptotically [50, 51, 52, 53], showing, up to logarithmic modulations, order ϵ\epsilon wavenumber corrections associated with end effects, order ϵ\epsilon bandwidth (i.e., order 1/ϵ1/\epsilon quality factor) and order 1/ϵ1/\epsilon field enhancement inside the slit.

Numerically calculated monopole and dipole coefficients in the multipole expansion

Figure 10. Numerically calculated monopole and dipole coefficients in the multipole expansion (88) for the scattered field in the exterior region, for α=π/4\alpha= \pi/4, ϵ=0.5\epsilon= 0.5 and l=1l = 1. A moderate value of ϵ\epsilon is chosen so as to be able to depict the exponentially narrow quasi-BIC resonance.

The nnth anti-symmetric, i.e., quasi-BIC, and nˉ\bar{n}th symmetric, i.e., conventional resonances can be made to approximately coincide by tuning the geometric parameter ll to 2(nˉ−n)/(1+2n)2(\bar{n}-n)/(1+2n). (A more precise condition could be derived by including ϵ\epsilon-dependent wavenumber corrections associated with end effects.) This may be of interest in order to achieve multi-polar radiation in the exterior region. For l≃1l \simeq1, the field inside the slit under such simultaneous excitation would be dominated by the exponentially excited anti-symmetric state, notwithstanding the symmetric and anti-symmetric states radiating waves of comparable magnitude in the exterior region. Note that simultaneously exciting the fundamental symmetric and anti-symmetric states (n=nˉ=0n=\bar{n}=0) requires ll near zero; our analysis however fails for l=O(ϵ)l=O(\epsilon) in which case the partition-edge, unified-channel region and aperture regions merge, thereby eliminating the evanescent decay underpinning the exponential nature of the quasi-BIC. That regime could be investigated using a methodology similar to that used herein; we would still expect quasi-BIC derived from the BIC of the associated semi-infinite geometry, however with the quality factor diverging algebraically rather than exponentially as ϵ→0\epsilon\to0.

Alternatively, an exponential quasi-BIC exhibiting multi-polar radiation could be achieved by displacing the partition away from the centreline of the slit. As long as the partition’s thickness can be neglected, there are still perfect BIC in the associated semi-infinite geometry [30]. The evanescent tail of these states, however, would now have both symmetric and anti-symmetric components, which would ‘turn on’ cylindrical monopole and dipole waves, respectively, at the slit aperture. It is possible that the mixture of radiation modes could be controlled through the lateral displacement of the partition.

We conclude by proposing several additional extensions. Thus, it would be relatively straightforward to adapt the methodology in the paper to study quasi-BIC resonances in closely related geometries, including transmission scenarios involving partitioned slits that are open at both ends, analogous three-dimensional configurations, multiple interacting partitioned slits and scaling up to metasurfaces. Physically, nonlinear and dissipation effects are both likely to be present within the partitioned portion of the slit given the exponential enhancement of the field there. In particular, the exponentially large quality factors that we have calculated are radiative quality factors, namely with dissipation neglected. Maximal dissipation is known to occur, however, under critical-coupling conditions, where the radiative and dissipative quality factors are comparable, suggesting that exponential quasi-BIC are sensitive to exponentially small loss levels. Dissipation could be included from first principles following the analysis in [53] of thermal and viscous boundary-layer effects on the symmetric resonances of a slit channel in a transmission scenario.

Acknowledgements

Acknowledgements. RP is grateful for the financial support provided by EPSRC under grant number EP/V04740X/1.

References

References

  1. [14]K. Koshelev, S. Lepeshov, M. Liu, A. Bogdanov, and Y. Kivshar, “Asymmetric metasurfaces with high-q resonances governed by bound states in the continuum,” Phys. Rev. Lett. 121, 193903 (2018).DOI
  2. [15]F. Wu, J. Wu, Z. Guo, H. Jiang, Y. Sun, Y. Li, J. Ren, and H. Chen, “Giant enhancement of the Goos-Hänchen shift assisted by quasibound states in the continuum,” Phys. Rev. Appl. 12, 014028 (2019).DOI
  3. [16]D. A. Bykov, E. A. Bezus, and L. L. Doskolovich, “Coupled-wave formalism for bound states in the continuum in guided-mode resonant gratings,” Phys. Rev. A 99, 063805 (2019).DOI
  4. [17]A. C. Overvig, S. C. Malek, M. J. Carter, S. Shrestha, and N. Yu, “Selection rules for quasibound states in the continuum,” Phys. Rev. B 102, 035434 (2020).DOI
  5. [18]L. Yuan and Y. Y. Lu, “Bound states in the continuum on periodic structures surrounded by strong resonances,” Phys. Rev. A 97, 043828 (2018).DOI
  6. [19]L. Yuan and Y. Y. Lu, “Perturbation theories for symmetry-protected bound states in the continuum on two-dimensional periodic structures,” Phys. Rev. A 101, 043827 (2020).DOI
  7. [20]L. Yuan and Y. Y. Lu, “Excitation of bound states in the continuum via second harmonic generations,” SIAM J. Appl. Math. 80, 864–880 (2020).DOI
  8. [21]Z. Hu, L. Yuan, and Y. Y. Lu, “Resonant field enhancement near bound states in the continuum on periodic structures,” Phys. Rev. A 101, 043825 (2020).DOI
  9. [22]Z. Hu, L. Yuan, and Y. Y. Lu, “Resonant field enhancement near bound states in the continuum on periodic structures,” Phys. Rev. A 101, 043825 (2020).DOI
  10. [23]E. J. Hinch, Perturbation Methods (Cambridge University Press, Cambridge, 1991).
  11. [24]E. S. C. Ching, P. T. Leung, A. M. van den Brink, W. M. Suen, S. S. Tong, and K. Young, “Quasinormal-mode expansion for waves in open systems,” Rev. Mod. Phys. 70, 1545 (1998).DOI
  12. [25]R. E. Collin, Foundations for microwave engineering (John Wiley & Sons, 2007).
  13. [26]R. Parker, “Resonance effects in wake shedding from parallel plates: some experimental observations,” J. Sound Vib 4, 62–72 (1966).DOI
  14. [27]R. Parker, “Resonance effects in wake shedding from parallel plates: calculation of resonant frequencies,” J. Sound Vib 5, 330–343 (1967).DOI
  15. [28]N. A. Cumpsty and D. S. Whitehead, “The excitation of acoustic resonances by vortex shedding,” J. Sound Vib 18, 353–369 (1971).DOI
  16. [29]W. Koch, “Resonant acoustic frequencies of flat plate cascades,” J. Sound Vib 88, 233–242 (1983).DOI
  17. [30]D. V. Evans, C. M. Linton, and F. Ursell, “Trapped mode frequencies embedded in the continuous spectrum,” Q. J. Mech. Appl. Math. 46, 253–274 (1993).DOI
  18. [31]P. McIver, C. M. Linton, and M. McIver, “Construction of trapped modes for wave guides and diffraction gratings,” Proc. R. Soc. Lond. A 454, 2593–2616 (1998).DOI
  19. [32]C. M. Linton and M. McIver, “Trapped modes in cylindrical waveguides,” Q. J. Mech. Appl. Math. 51, 389–412 (1998).DOI
  20. [33]M. D. Groves, “Examples of embedded eigenvalues for problems in acoustic waveguides,” Math. Methods Appl. Sci. 21, 479–488 (1998).doi.org/10.1002/(sici)1099-1476(199804)21:6<479::aid-mma950>3.0.co;2-v
  21. [34]E. B. Davies and L. Parnovski, “Trapped modes in acoustic waveguides,” Q. J. Mech. Appl. Math 51, 477–492 (1998).DOI
  22. [35]C. M. Linton, M. McIver, P. McIver, K. Ratcliffe, and J. Zhang, “Trapped modes for off-centre structures in guides,” Wave Motion 36, 67–85 (2002).DOI
  23. [36]Y. Duan, W. Koch, C. M. Linton, and M. McIver, “Complex resonances and trapped modes in ducted domains,” J. Fluid Mech. 571, 119–147 (2007).DOI
  24. [37]P. J. Cobelli, V. Pagneux, A. Maurel, and P. Petitjeans, “Experimental study on water-wave trapped modes,” J. Fluid Mech. 666, 445–476 (2011).
  25. [38]C. V. Sargent and A. J. Mestel, “Trapped modes of the Helmholtz equation in infinite waveguides with wall indentations and circular obstacles,” IMA J. Appl. Math. 84, 312–344 (2019).DOI
  26. [39]J. A. Porto, F. J. Garcia-Vidal, and J. B. Pendry, “Transmission resonances on metallic gratings with very narrow slits,” Phys. Rev. Lett. 83, 2845 (1999).DOI
  27. [40]Y. Takakura, “Optical resonance in a narrow slit in a thick metallic screen,” Phys. Rev. Lett. 86, 5601 (2001).pubmed.ncbi.nlm.nih.gov/11415311
  28. [41]F. J. García-Vidal, H. J. Lezec, T. W. Ebbesen, and L. Martin-Moreno, “Multiple paths to enhance optical transmission through a single subwavelength slit,” Phys. Rev. Lett. 90, 213901 (2003).
  29. [42]J. Bravo-Abad, L. Martin-Moreno, and F. J. Garcia-Vidal, “Transmission properties of a single metallic slit: From the subwavelength regime to the geometrical-optics limit,” Phys. Rev. E 69, 026601 (2004).
  30. [43]J. R. Suckling, A. P. Hibbins, M. J. Lockyear, T. W. Preist, J. R. Sambles, and C. R. Lawrence, “Finite conductance governs the resonance transmission of thin metal slits at mi- crowave frequencies,” Phys. Rev. Lett. 92, 147401 (2004).
  31. [44]J. Christensen, L. Martin-Moreno, and F. J. Garcia-Vidal, “Theory of resonant acoustic transmission through subwavelength apertures,” Phys. Rev. Lett. 101, 014301 (2008).
  32. [45]G. P. Ward, R. K. Lovelock, A. R. J. Murray, A. P. Hibbins, J. R. Sambles, and J. D. Smith, “Boundary-layer effects on acoustic transmission through narrow slit cavities,” Phys. Rev. Lett. 115, 044302 (2015).pubmed.ncbi.nlm.nih.gov/26252688
  33. [46]M. Molerón, M. Serra-Garcia, and C. Daraio, “Visco-thermal effects in acoustic metamaterials: from total transmission to total reflection and high absorption,” New J. Phys. 18, 033003 (2016).
  34. [47]P. Joly and S. Tordeux, “Asymptotic analysis of an approximate model for time harmonic waves in media with thin slots,” M2AN Math. Model. Numer. Anal. 40, 63–97 (2006).DOI
  35. [48]P. Joly and S. Tordeux, “Matching of asymptotic expansions for wave propagation in media with thin slots i: The asymptotic expansion,” Multiscale Model. Simul. 5, 304–336 (2006).DOI
  36. [49]P. Joly and S. Tordeux, “Matching of asymptotic expansions for waves propagation in media with thin slots ii: The error estimates,” M2AN Math. Model. Numer. Anal. 42, 193–221 (2008).DOI
  37. [50]J. Lin and H. Zhang, “Scattering and field enhancement of a perfect conducting narrow slit,” SIAM J. Appl. Math. 77, 951–976 (2017).DOI
  38. [51]D. V. Evans, R. Porter, and J. R. Chaplin, “Extraordinary transmission past cylinders in channels,” in Proc 33rd International Workshop on Water Waves & Floating Bodies, Gudel-Plages, 2018.
  39. [52]J. R. Holley and O. Schnitzer, “Extraordinary transmission through a narrow slit,” Wave Motion 91, 102381 (2019).arxiv.org/abs/1812.06753
  40. [53]R. Brandão, J. R. Holley, and O. Schnitzer, “Boundary-layer effects on electromagnetic and acoustic extraordinary transmission through narrow slits,” Proc. R. Soc. Lond. A 476, 20200444 (2020).DOI
  41. [54]L. Chesnel and S. A. Nazarov, “Design of an acoustic energy distributor using thin resonant slits,” Proc. R. Soc. Lond. A 477, 20200896 (2021).
  42. [55]L. Chesnel, J. Heleine, and S. A. Nazarov, “Acoustic passive cloaking using thin outer resonators,” arXiv preprint arXiv:2105.00922 (2021).DOI
  43. [56]O. Schnitzer, “Spoof surface plasmons guided by narrow grooves,” Phys. Rev. B 96, 085424 (2017).
  44. [57]J. W. Brown and R. V. Churchill, Complex variables and applications (McGraw-Hill, 2009).
  45. [58]H. Segur, S. Tanveer, and H. J. Levine, Asymptotics beyond all orders, Vol. 284 (Springer Science & Business Media, 2012).
  46. [59]The supplementary material describes the semi-analytical schemes used to solve the eigenvalue problem formulated in §II and the scattering problem formulated in §V numerically, for the purpose of comparison with the asymptotic approximations derived herein.
  47. [60]M. Abramowitz and I. A. Stegun, Handbook of mathematical functions (Dover New York, 1972).
  48. [61]S. Hein, W. Koch, and L. Nannen, “Fano resonances in acoustics,” J. Fluid Mech. 664, 238–264 (2010).DOI

Paper details

vlnc:W_8k6feamq
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

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

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents