Expand this Topic clickable element to expand a topic
Skip to content
Optica Publishing Group

Robust and non-robust bound states in the continuum in rotationally symmetric periodic waveguides

Open Access Open Access

Abstract

A fiber grating and a one-dimensional (1D) periodic array of spheres are examples of rotationally symmetric periodic (RSP) waveguides. It is well known that bound states in the continuum (BICs) may exist in lossless dielectric RSP waveguides. Any guided mode in an RSP waveguide is characterized by an azimuthal index m, the frequency ω, and Bloch wavenumber β. A BIC is a guided mode, but for the same m, ω and β, cylindrical waves can propagate to or from infinity in the surrounding homogeneous medium. In this paper, we investigate the robustness of nondegenerate BICs in lossless dielectric RSP waveguides. The question is whether a BIC in an RSP waveguide with a reflection symmetry along its axis z, can continue its existence when the waveguide is perturbed by small but arbitrary structural perturbations that preserve the periodicity and the reflection symmetry in z. It is shown that for m = 0 and m ≠ 0, generic BICs with only a single propagating diffraction order are robust and non-robust, respectively, and a non-robust BIC with m ≠ 0 can continue to exist if the perturbation contains one tunable parameter. The theory is established by proving the existence of a BIC in the perturbed structure mathematically, where the perturbation is small but arbitrary, and contains an extra tunable parameter for the case of m ≠ 0. The theory is validated by numerical examples for propagating BICs with m ≠ 0 and β ≠ 0 in fiber gratings and 1D arrays of circular disks.

© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement

1. Introduction

A bound state in the continuum (BIC) is a trapped or guided mode with a frequency in the frequency interval of radiation modes [110]. Mathematically, a BIC corresponds to an embedded eigenvalue in a related continuous spectrum [1115]. In recent years, BICs have become a topic of significant research interest in the photonics community [710]. Photonic BICs have been found in many different structures including periodic structures sandwiched between two homogeneous media [1521], waveguides with lateral leakage channels [2224], layered anisotropic media [2527], and rotationally symmetric periodic (RSP) waveguides such as fiber gratings and one-dimensional (1D) arrays of dielectric spheres [2830]. A number of important applications like lasing [31], sensing [32,33], filtering [34] and switching [35], have been demonstrated experimentally [710].

In structures consisting of ordinary lossless dielectric material, BICs can only exist when the structure is unbounded in at least one spatial direction. An RSP waveguide is periodic along its axis (the $z$ axis in this paper) and rotationally symmetric about the same axis [2830,3638]. A guided mode of an RSP waveguide is associated with a Bloch wavenumber $\beta$ (for the $z$ direction) and an azimuthal index $m$, so that its electric field can be written as ${\boldsymbol{\Phi} }(\rho,z) e^{ i ( m \phi + \beta z)}$, where $(\rho, \phi, z)$ are the cylindrical coordinates and ${\boldsymbol{\Phi} }$ is periodic in $z$. A BIC is a special guided mode above the light line. Specifically, if the RSP waveguide is surrounded by air, then the free space wavenumber $k$ and Bloch wavenumber $\beta$ of a BIC must satisfy $k > |\beta |$. Corresponding to the cases $\beta =0$ and $\beta \neq 0$, the BIC is a standing wave and a propagating mode, respectively. For $m=0$, the BIC is either transverse-magnetic (TM) or transverse-electric (TE) polarized. If the RSP waveguide has a reflection symmetry in $z$, a typical standing wave with $m=0$ is symmetry-protected, but there also exist standing waves unprotected by symmetry [6,36]. Propagating BICs with $m=0$ have been found in many RSP waveguides [28,30,36]. For $m\neq 0$, the TE and TM components of any guided mode in an RSP waveguide are fully coupled. Standing waves with a nonzero $m$ have been found in a number of RSP waveguides such as 1D arrays of dielectric spheres or circular disks [28,36] and periodic nanowires [38]. It has been shown that propagating BICs with $m\ne 0$ exist in 1D arrays of dielectric spheres [29].

Existing studies have revealed that many BICs are robust, in the sense that they continue to exist when the structure is perturbed [3947]. When the structure has a relevant symmetry, for example the in-plane inversion symmetry for periodic structures sandwiched between two homogeneous media, the symmetry-protected BICs are clearly robust with respect to small but arbitrary symmetry-preserving structural perturbations [5,11,12,19,22]. If the perturbation breaks the relevant symmetry, the BIC is usually (but not always) destroyed and becomes a resonant mode with a finite quality factor ($Q$ factor) [48,49]. Some BICs unprotected by symmetry can also be robust, but symmetry still plays a key role [39,40]. For example, in periodic structures with the in-plane inversion symmetry, some propagating BICs are robust with respect to small structural perturbations that preserve the inversion symmetry [39,40,45,47]. For some BICs, a topological charge can be defined using the far field polarization vector of the surrounding resonant modes, and the conservation of topological charge implies robustness [39,40]. On the other hand, mathematical theories on the robustness of BICs in periodic structures sandwiched between homogeneous media and in waveguides with lateral leakage channels have been established [4547]. In these works, the existence of BICs in perturbed structures is proved constructively using power series in a small parameter $\delta$ related to the amplitude of the perturbation.

In this paper, we analyze the robustness of nondegenerate BICs in lossless dielectric RSP waveguides with a reflection symmetry in $z$. Specifically, we assume $k$ and $\beta$ of the BIC satisfy $|\beta | < k < 2\pi /L - |\beta |$, where $L$ is the period of the RSP waveguide. This condition ensures that the zeroth diffraction order is the only propagating order. It is shown that with respect to structural perturbations that preserve the reflection symmetry, generic BICs with $m=0$ (including both standing waves and propagating modes) are robust, and those BICs with $m\ne 0$ are not robust. The theory for BICs with $m=0$ can be relatively easily developed following the approach of Ref. [45]. A non-robust BIC with $m\ne 0$ is usually destroyed by a structural perturbation, even when the perturbation is symmetric in $z$, but the BIC can nevertheless persist under some special structural perturbations. We show that if the structural perturbation contains a tunable parameter, a generic non-robust BIC with $m \ne 0$ can continue to exist when the parameter is properly tuned. This implies that in any symmetric RSP waveguide with two real structural parameters, a generic non-robust BIC with $m\ne 0$ exists as a curve in the parameter plane.

The rest of this paper is organized as follows. In section 2, we recall some facts about BICs in RSP waveguides and introduce some useful scattering solutions. The theory is developed in section 3 mainly for the case of $m\ne 0$, but we also include a brief discussion for the case of $m=0$. To show the continual existence of generic non-robust BICs under small structural perturbations with a tunable parameter, we determine the BIC in the perturbed structure and the parameter value together, using power series in $\delta$, where $\delta$ is the amplitude of the perturbation. In section 4, we illustrate our theory by numerical examples including new propagating BICs with $m\ne 0$ in 1D arrays of circular disks and fiber gratings. The paper is concluded with a brief discussion in section 5.

2. Background

We consider an isotropic and lossless RSP waveguide that is periodic in $z$ with period $L$, has a reflection symmetry in $z$, and is surrounded by air. A particular example is a periodic array of circular disks shown in Fig. 1 below. The dielectric function $\varepsilon$ satisfies

$$\varepsilon(\rho,z)=\varepsilon(\rho,z+L),\;\varepsilon(\rho,z)=\varepsilon(\rho,-z),$$
and $\varepsilon (\rho,z)=1\text { for }\rho >R$, where $(\rho, \phi, z)$ are the cylindrical coordinates. In a non-magnetic structure without sources, the time-harmonic electric field ${\textbf{E}}$ satisfies the vector wave equation
$$\nabla\times\nabla\times{\textbf{E}}(\rho,z)-k^2\varepsilon{\textbf{E}}(\rho,z)=0,$$
where $k=\omega /c$ is the free space wavenumber, $\omega$ is the angular frequency, and $c$ is the speed of light in vacuum. The magnetic field ${\textbf{H}}$ scaled with the free space impedance could be obtained from $\nabla \times {\textbf{E}}= ik {\textbf{H}}$. The time dependence is assumed to be $\exp (-i\omega t)$. Since the structure is rotationally symmetric, we consider a Bloch mode ${\textbf{E}}={\boldsymbol{\Phi} }(\rho,z)\exp [i(m\phi +\beta z)]$, where $m$ is the azimuthal index, $\beta$ is the Bloch wavenumber, ${\boldsymbol{\Phi} }(\rho,z)$ is the electric mode profile and it is periodic in $z$ with period $L$. In terms of ${\boldsymbol{\Phi} }$, Eq. (2) can be written as
$$\mathcal{L}{{\boldsymbol{\Phi}}}:=(\nabla_c+i\beta{\hat z})\times(\nabla_c+i\beta{\hat z})\times{\boldsymbol{\Phi}}(\rho,z)-k^2\varepsilon{\boldsymbol{\Phi}}(\rho,z)=0,$$
where $\hat {z}$ is the unit vector in the $z$ direction, $\nabla _c$ is an operator defined implicitly by
$$\nabla\times\left[{\textbf{f}}(\rho,z)e^{i(m\phi+\beta z)}\right]=\left[ ( \nabla_c+ i \beta \hat{z}) \times{\textbf{f}}(\rho,z)\right]e^{i(m\phi+\beta z)}$$
for any differentiable vector function ${\textbf{f}}$.

 figure: Fig. 1.

Fig. 1. A periodic array of circular disks surrounded by air. The period of the array is $L$. The radius, thickness and dielectric constant of the disks are $R$, $d$ and $\varepsilon _1$, respectively.

Download Full Size | PDF

If $m=0$, the electromagnetic field can be decoupled into TE (${\textbf{H}}^e=H_\rho \hat {\rho }+H_z\hat {z},\;{\textbf{E}}^e=E_\phi \hat {\phi }$) and TM (${\textbf{E}}^h=E_\rho \hat {\rho }+E_z\hat {z},\;{\textbf{H}}^h=H_\phi \hat {\phi }$) polarized fields [36], where $\hat {\rho }$ and $\hat {\phi }$ are the unit vectors in the radial and azimuthal directions, respectively. If $m\neq 0$, the TE components

$${\textbf{H}}^e=\left[ \begin{array}{c} \displaystyle H_\rho^e\\ H_\phi^e\\ H_z\\ \end{array} \right],\quad {\textbf{E}}^e=\left[ \begin{array}{c} \displaystyle E_{\rho}^e\\ \displaystyle E_\phi^e\\ 0\\ \end{array} \right]$$
and TM components
$${\textbf{E}}^h=\left[ \begin{array}{c} E_\rho^h\\ E_\phi^h\\ E_z\\ \end{array} \right],\quad {\textbf{H}}^h=\left[ \begin{array}{c} H_\rho^h\\ H_\phi^h\\ 0\\ \end{array} \right]$$
are fully coupled.

When $\rho >R$, the electric field of any Bloch mode satisfying an outgoing radiation condition can be expanded as [30]

$${\textbf{E}}={\textbf{E}}^e+{\textbf{E}}^h=e^{i(m\phi+\beta z)}\sum_p \left[c_p^e{\textbf{e}}_p^e(\rho)+c_p^h{\textbf{e}}_p^h(\rho)\right]e^{i2\pi p z/L},$$
where $p$ is an integer index, $c_p^e$ and $c_p^h$ are $p$-th coefficients, ${\textbf{e}}_p^e(\rho )$ and ${\textbf{e}}_p^h(\rho )$ are given by
$${\textbf{e}}_p^e(\rho)=k\left[ \begin{array}{c} \displaystyle \frac{m}{\rho} H^{(1)}_m(\kappa_p\rho)\\ \displaystyle i\kappa_pH'^{(1)}_m(\kappa_p\rho)\\ \displaystyle 0\\ \end{array} \right],\quad {\textbf{e}}_p^h(\rho)=\left[ \begin{array}{c} \displaystyle \hat{\beta}_p\kappa_pH'^{(1)}_m(\kappa_p\rho)\\ \displaystyle im\frac{\hat{\beta}_p}{\rho} H^{(1)}_m(\kappa_p\rho)\\ \displaystyle -i\kappa_p^2H^{(1)}_m(\kappa_p\rho)\\ \end{array} \right].$$
In the above, $\hat {\beta }_p=\beta +{2\pi p}/{L}$, $\kappa _p=\sqrt {k^2-\hat {\beta }_p^2}$, $H^{(1)}_m$ is the $m$-th order Hankel function of the first kind, and $H'^{(1)}_m$ is the derivatives of $H^{(1)}_m$. If $\beta =0$, then ${\textbf{e}}_0^h(\rho )=[0;0;-ik^2H^{(1)}_m(k\rho )]$. In addition, we have
$${\textbf{e}}_p^e(\rho)\sim \frac{1}{\sqrt{\rho}}{\textbf{d}}_p^ee^{i\kappa_p\rho},\;{\textbf{e}}_p^h(\rho)\sim \frac{1}{\sqrt{\rho}}{\textbf{d}}_p^he^{i\kappa_p\rho},\text{ as }\rho\rightarrow\infty,$$
where ${\textbf{d}}_p^e$ and ${\textbf{d}}_p^h$ are constant vectors. If $k$ is real and $k^2-\hat {\beta }_p^2<0$, then $\kappa _p$ is pure imaginary and the corresponding cylindrical wave is evanescent. As $\rho \rightarrow \infty$, ${\boldsymbol{\Phi}}$ has the asymptotic form
$$\sqrt{\rho}{\boldsymbol{\Phi}}\sim\sum_{\kappa_p\geq 0}\left(c_p^e{\textbf{d}}_p^e+c_p^h{\textbf{d}}_p^h\right)e^{i\kappa_p\rho}e^{i2\pi p z/L}.$$

A guided mode with a real $k$ has a mode profile satisfying $\sqrt {\rho }{\boldsymbol{\Phi} }\rightarrow 0$ as $\rho \rightarrow +\infty$. In general, guided modes exist below the light line ($k<|\beta |$) such that all $\kappa _p$ are pure imaginary. A BIC is a special guided mode above the light line (with a real $k$ and a real $\beta$ satisfying $k>|\beta |$). In this paper, we focus on BICs satisfying $|\beta | < k < 2\pi /L - |\beta |$. This condition implies that $\kappa _0>0$ and all other $\kappa _p$ ($p\ne 0$) are pure imaginary. Consequently, for such a BIC, the coefficients $c_0^e$ and $c_0^h$ must be zero.

In the rest of this section, we discuss some properties of the BICs and introduce related scattering solutions. The BICs are assumed to be nondegenerate. The scattering solutions are needed in the following section for analyzing the robustness of BICs. Since the RSP waveguide has a reflection symmetry in $z$ and no material loss, the BICs and the scattering solutions can be scaled to have a symmetry related to parity-inversion and time-reversal operations. To simplify the presentation, we define two operators ${\mathcal {P}}$ and ${\mathcal {T}}$ by

$${\mathcal{P}}{\textbf{f}}=\left[ \begin{array}{c} \displaystyle f_\rho(\rho,-z)\\ f_\phi(\rho,-z)\\ -f_z(\rho,-z)\\ \end{array} \right],\quad {\mathcal{T}}{\textbf{f}}=\left[ \begin{array}{c} \displaystyle \overline{f}_\rho(\rho,z)\\ \displaystyle -\overline{f}_\phi(\rho,z)\\ \overline{f}_z(\rho,z)\\ \end{array} \right],$$
where ${\textbf{f}} = [ f_\rho ; f_\phi ; f_z]$ is an arbitrary 3D vector function. It is easy to verify that
$${\mathcal{P}\mathcal{T}}=\mathcal{T}\mathcal{P},\quad\mathcal{P}^2=\mathcal{T}^2=\mathcal{I},$$
where $\mathcal {I}$ is the identity operator. For operator $\mathcal {L}$ defined in Eq. (3), we can show that
$${\mathcal{P}\mathcal{T}}\mathcal{L}=\mathcal{L}{\mathcal{P}\mathcal{T}},$$
and consequently, we can scale the BIC such that ${\mathcal {P}}{\mathcal {T}}{\boldsymbol{\Phi} }={\boldsymbol{\Phi} }$. In other words, the components of ${\boldsymbol{\Phi}}$ are either ${\mathcal {P}\mathcal {T}}$-symmetric or anti-${\mathcal {P}\mathcal {T}}$-symmetric [50]. If $\beta =0$, we have
$${\mathcal{T}}\mathcal{L}=\mathcal{L}{\mathcal{T}},$$

${\mathcal {T}}{\boldsymbol{\Phi} }={\boldsymbol{\Phi} }$, and ${\mathcal {P}}\mathcal {L}=\mathcal {L}{\mathcal {P}}$. This implies that ${\boldsymbol{\Phi}}$ satisfies either ${\mathcal {P}}{\boldsymbol{\Phi} }=-{\boldsymbol{\Phi} }$ or ${\mathcal {P}}{\boldsymbol{\Phi} }={\boldsymbol{\Phi} }$. More specifically, $\Phi _z$ is either even or odd in $z$ and the other components have the opposite parity in $z$.

A resonant mode is associated with a complex $k$ and a real $\beta$, and satisfies an outgoing radiation condition. A BIC can be regarded as a special resonant mode with a real $k$, a real $\beta$, and an infinite $Q$ factor, where $Q=-0.5\mbox {Re}(k)/\mbox {Im}(k)$ [51]. When $\beta =0$, $\Phi _z$ of a nondegenerate resonant mode is either odd or even in $z$. Therefore, either $c_0^h$ or $c_0^e$ must be zero. In other words, a resonant mode with $\beta =0$ radiates out power in a single polarized radiation channel.

Next, we construct scattering solutions with the same symmetry properties as the BICs. We start with a scattering solution excited by the following TE cylindrical incident wave

$${\textbf{E}}^{\text{i},e}=\left[{\mathcal{P}\mathcal{T}} {\textbf{e}}_0^e(\rho)\right] e^{i(m\phi+\beta z)}.$$
Notice that ${\textbf{e}}_0^e(\rho )$ in Eq. (6) represents an outgoing wave and ${\mathcal {P}\mathcal {T}} {\textbf{e}}_0^e(\rho )$ in Eq. (14) represents an incoming wave. If ${\boldsymbol{\Psi} }^{e}\exp [i(m\phi +\beta z)]$ is a scattering solution associated with the above incident wave, $({\boldsymbol{\Psi} }^{e}+C{\boldsymbol{\Phi} }) \exp [i(m\phi +\beta z)]$ for any constant $C$, is also a solution for the same incident wave. This implies that scattering solutions for the same $k$ and $\beta$ of a BIC are not unique. But when $\rho \rightarrow \infty$, all scattering solutions have the same asymptotic form
$$\sqrt{\rho}{\boldsymbol{\Psi}}^e\sim \left(s^{ee}{\textbf{d}}_0^e+s^{eh}{\textbf{d}}_0^h\right) e^{i\kappa_0\rho}+\left( {\mathcal{P}\mathcal{T}} {\textbf{d}}_0^e\right) e^{{-}i\kappa_0\rho},$$
where $s^{ee}$ and $s^{eh}$ are scattering coefficients. If $\beta \neq 0$, we can construct a new scattering solution with the profile $\hat {\boldsymbol{\Psi} }^e:={\mathcal {P}}{\mathcal {T}}{\boldsymbol{\Psi} }^e+{\boldsymbol{\Psi} }^e$. It corresponds to the incident wave
$$\begin{aligned} \hat{\boldsymbol{\Psi}}^{\text{i},e} & =(1+\overline{s}^{ee})\left[{\mathcal{P}\mathcal{T}} {\textbf{e}}_0^e(\rho)\right]+\overline{s}^{eh}\left[{\mathcal{P}\mathcal{T}}{\textbf{e}}_0^h(\rho)\right], \end{aligned}$$
and satisfies ${\mathcal {P}}{\mathcal {T}}\hat {\boldsymbol{\Psi} }^e=\hat {\boldsymbol{\Psi} }^e$. Similarly, starting from a scattering solution excited by a TM cylindrical incident wave, we can construct a new scattering solution with profile $\hat {\boldsymbol{\Psi} }^h$ satisfying ${\mathcal {P}\mathcal {T}}\hat {\boldsymbol{\Psi} }^h=\hat {\boldsymbol{\Psi} }^h$.

For $\beta =0$, since $\mathcal {P}{\textbf{e}}_0^e={\textbf{e}}_0^e$ and $\mathcal {P}{\textbf{e}}_0^h=-{\textbf{e}}_0^h$, the incident wave in Eq. (14) can be reduced to

$${\textbf{E}}^{\text{i},e}=\left[\mathcal{T} {\textbf{e}}_0^e(\rho)\right] e^{im\phi}.$$
As $\rho \rightarrow \infty$, $\mathcal {P}{\boldsymbol{\Psi} }^e$ satisfy
$$\sqrt{\rho}\mathcal{P}{\boldsymbol{\Psi}}^e\sim \left(s^{ee}{\textbf{d}}_0^e-s^{eh}{\textbf{d}}_0^h\right) e^{ik\rho} + \left( \mathcal{T} {\textbf{d}}_0^e\right) e^{{-}ik\rho}.$$
It is easy to verify that $\mathcal {P}{\boldsymbol{\Psi} }^e$ is also a diffraction solution profile for the same TE cylindrical wave (17). Hence, we have a scattering solution with a profile $\hat {\boldsymbol{\Psi} }^e:=\left [{\boldsymbol{\Psi} }^e+\mathcal {P}{\boldsymbol{\Psi} }^e\right ]/2$ satisfying $\mathcal {P}\hat {\boldsymbol{\Psi} }^e=\hat {\boldsymbol{\Psi} }^e$. Similarly, we can construct a scattering solution with a profile $\hat {\boldsymbol{\Psi} }^h$ satisfying $\mathcal {P}\hat {\boldsymbol{\Psi} }^h=-\hat {\boldsymbol{\Psi} }^h$. In addition, we can scale these profiles such that ${\mathcal {T}}\hat {\boldsymbol{\Psi} }^l=\hat {\boldsymbol{\Psi} }^l$, $l\in \{e,h\}$.

In summary, if $\beta \ne 0$, we can scale the BIC such that ${\mathcal {P}\mathcal {T}}{\boldsymbol{\Phi} }={\boldsymbol{\Phi} }$, and construct scattering solutions satisfying ${\mathcal {P}}{\mathcal {T}}\hat {\boldsymbol{\Psi} }^l=\hat {\boldsymbol{\Psi} }^l$, $l\in \{e,h\}$. If $\beta =0$, we can scale the BIC such that $\mathcal {T}{\boldsymbol{\Phi} }={\boldsymbol{\Phi} }$ and $\mathcal {P}{\boldsymbol{\Phi} }={\boldsymbol{\Phi} }$ or $\mathcal {P}{\boldsymbol{\Phi} }=-{\boldsymbol{\Phi} }$. We can also construct diffraction solutions satisfying ${\mathcal {P}}\hat {\boldsymbol{\Psi} }^e=\hat {\boldsymbol{\Psi} }^e$, ${\mathcal {P}}\hat {\boldsymbol{\Psi} }^h=-\hat {\boldsymbol{\Psi} }^h$ and $\mathcal {T}\hat {\boldsymbol{\Psi} }^l=\hat {\boldsymbol{\Psi} }^{l}$, $l\in \{e,h\}$.

3. Theory

In this section, we analyze the robustness of BICs in RSP waveguides. It is assumed that an RSP waveguide with dielectric function $\varepsilon _*({\rho,z})$ supports a nondegenerate $\mathrm {BIC}$ with free space wavenumber $k_*$, Bloch wavenumber $\beta _*$, and electric field ${\textbf{E}}_*({\textbf{r}})={\boldsymbol{\Phi} }_*({\rho,z})\exp {[i (m\phi +\beta _* z)]}$. The RSP waveguide is periodic in $z$ with period $L$, is symmetric in $z$ and surrounded by air. The dielectric function $\varepsilon _*({\rho,z})$ satisfies Eq. (1). As in section 2, we assume $k_*$ and $\beta _*$ of the BIC satisfy

$$|\beta_*| < k_* < \frac{2\pi}{L} - |\beta_*|,$$
so that the only radiation channels are associated with the zeroth diffraction order, i.e., $p=0$. It is also assumed that the BIC and the related scattering solutions ${\textbf{E}}_*^{l}={\boldsymbol{\Psi} }_*^l\exp {[i (m\phi +\beta _* z)]}$ for $l \in \{e, h\}$ satisfy the scaling and symmetry properties discussed in section 2. When $\beta _* \ne 0$, these conditions are simply ${\mathcal {P}}{\mathcal {T}}{\boldsymbol{\Phi} }_*={\boldsymbol{\Phi} }_*$ and ${\mathcal {P}}{\mathcal {T}}{\boldsymbol{\Psi} }_*^{l}={\boldsymbol{\Psi} }_*^{l}$, $l\in \{e,h\}$. In addition, we normalize the BIC and shift the scattering solutions so that
$$\frac{1}{L^2}\int_{\Omega}\varepsilon_*|{\boldsymbol{\Phi}}_*|^2d\Omega=1,\;\int_{\Omega}\varepsilon_*\overline{\boldsymbol{\Phi}}_*\cdot {\boldsymbol{\Psi}}_*^ld\Omega=0,\;l\in\{e,h\},$$
where $\Omega =\{(\rho,z);\rho >0,|z|<L/2\}$, $d\Omega =\rho dz d\rho$, and $\overline {\boldsymbol{\Phi} }_*$ is the complex conjugate of ${\boldsymbol{\Phi} }_*$.

We are concerned with the robustness of generic nondegenerate BICs with respect to small but arbitrary structural perturbations that preserve the periodicity and reflection symmetry in $z$. Existing numerical results suggest that the BICs with $m=0$ are robust with respect to such perturbations [30,36]. This is indeed true and can be proved rigorously following the approach of Ref. [45]. This means that for any real function $F(\rho,z)$ satisfying Eq. (1) and $F(\rho,z) = 0$ for $\rho > R$, and for any small real number $\delta$, the perturbed RSP waveguide with the dielectric function $\varepsilon (\rho,z) = \varepsilon _* (\rho,z) + \delta F(\rho, z)$, has a BIC near the original one in the unperturbed structure. On the other hand, as the numerical examples have suggested [36], the BICs with $m\ne 0$ are nno-robust with respect to symmetric and periodic perturbations. For arbitrarily chosen $F$ and $\delta$, there is usually no BIC (near the original BIC) in the perturbed structure. However, for specially chosen perturbations, there could still be a BIC in the perturbed structure. Following the approach first developed in Ref. [52], we consider a perturbed RSP waveguide with the dielectric function

$$\varepsilon(\rho,z)=\varepsilon_*(\rho,z)+\delta F(\rho,z)+\gamma G(\rho,z),$$
where $F$ and $G$ satisfy Eq. (1) and vanish for $\rho > R$, $\delta$ is a small but arbitrary real number, and $\gamma$ is a tunable parameter. In the following, we show that for any $\delta$, any $F$, and a generic $G$, if $\gamma$ is properly chosen, the perturbed waveguide has a BIC near the original one with the same nonzero $m$. For simplicity, we focus on the case of $m\ne 0$ and briefly discuss the $m=0$ case at the end of this section.

In order to show the existence of a BIC in the perturbed RSP waveguide [with the dielectric function given in Eq. (21)] for a properly chosen $\gamma$, we construct the BIC and find $\gamma$ together through power series in $\delta$. Let ${\boldsymbol{\Phi}}$, $k$, $\beta$ be the electric field profile, free space wavenumber and Bloch wavenumber of the desired BIC, we expand these quantities and $\gamma$ as

$${\boldsymbol{\Phi}}={\boldsymbol{\Phi}}_*+\delta{\boldsymbol{\Phi}}_1 +\delta^2{\boldsymbol{\Phi}}_2 +\cdots,$$
$$k=k_*+\delta k_1 +\delta^2k_2 +\cdots,$$
$$\beta=\beta_*+\delta\beta_1 +\delta^2\beta_2 +\cdots,$$
$$\gamma=\delta \gamma_1 +\delta^2\gamma_2 +\cdots.$$
For each $j \ge 1$, we need to show that $k_j$, $\beta _j$ and $\gamma _j$ can be solved and they are real, and ${\boldsymbol{\Phi} }_j$ can be solved and it satisfies
$${\boldsymbol{\Phi}}_j(\rho,z)={\boldsymbol{\Phi}}_j(\rho,z+L),\quad{\mathcal{P}}{\mathcal{T}}{\boldsymbol{\Phi}}_j={\boldsymbol{\Phi}}_j,\quad \sqrt{\rho}{\boldsymbol{\Phi}}_j\rightarrow 0\text{ as }\rho\rightarrow\infty.$$

Inserting Eqs. (21)–(25) into Eq. (3) and comparing the different orders of $\delta$, we obtain

$$\mathcal{L}_*{{\boldsymbol{\Phi}}}_*=0,$$
$$\mathcal{L}_*{{\boldsymbol{\Phi}}}_j={\textbf{f}}_j:=\beta_j\mathcal{B}_*{\boldsymbol{\Phi}}_*+\left(2k_*k_j\varepsilon_*+k_*^2\gamma_jG\right){\boldsymbol{\Phi}}_*+\boldsymbol{\Theta}_j(\rho,z),\quad j\geq 1,$$
where $\mathcal {L}_*$ and $\mathcal {B}_*$ are operators satisfying
$$\mathcal{L}_* {\textbf{w}} =(\nabla_c+i\beta_*{\hat z})\times(\nabla_c+i\beta_*{\hat z})\times {\textbf{w}} -k_*^2\varepsilon_*{\textbf{w}},$$
$$\mathcal{B}_* {\textbf{w}} ={-}i\left[\hat{z}\times(\nabla_c+i\beta_*\hat{z})\times {\textbf{w}} +(\nabla_c+i\beta_*\hat{z})\times\hat{z}\times {\textbf{w}} \right],$$
for any vector function ${\textbf{w}}$. The function $\boldsymbol {\Theta }_j(\rho,z)$ depends on all previous iterations and is given in the Appendix. Equation (27) is the original vector wave equation satisfied by the BIC. Equation (28) is a vector wave equation with a source term given by the right hand side ${\textbf{f}}_j$. It can be argued rigorously that the necessary and sufficient conditions for Eq. (28) to have a solution ${\boldsymbol{\Phi} }_j$ satisfying $\sqrt {\rho }{\boldsymbol{\Phi} }_j\rightarrow 0$ as $\rho \rightarrow \infty$ are
$$\int_\Omega\overline{{{\boldsymbol{\Phi}}}}_*\cdot{\textbf{f}}_jd\Omega=0,$$
$$\int_\Omega\overline{{{\boldsymbol{\Psi}}}}_*^l\cdot{\textbf{f}}_jd\Omega=0,\quad l\in\{e,h\}.$$
Equation (29) is the orthogonality between the source term ${\textbf{f}}_j$ and the BIC. It ensures that the source term will not excite the BIC, so that the inhomogeneous Eq. (28) can have a bounded solution. In general, the solution satisfies an outgoing radiation condition as $\rho \to \infty$. Equation (30) ensures that the solution decays to zero rapidly, i.e., $\sqrt {\rho } {\boldsymbol{\Phi} }_j \to 0$ as $\rho \to \infty$.

Equations (29) and (30) give rise to a linear system

$$\left[ \begin{array}{ccc} a_{11} & a_{12} & a_{13}\\ a_{21} & a_{22} & a_{23}\\ a_{31} & a_{32} & a_{33} \end{array} \right]\left[ \begin{array}{c} k_j\\ \beta_j\\ \gamma_j\\ \end{array} \right]=\left[ \begin{array}{c} d_{1j}\\ d_{2j}\\ d_{3j} \end{array} \right],$$
where
$$a_{11}=2k_*\int_\Omega\varepsilon_*\overline{{{\boldsymbol{\Phi}}}}_*\cdot{{\boldsymbol{\Phi}}}_*d\Omega=2k_*L^2,$$
$$a_{12}=\int_\Omega\overline{{{\boldsymbol{\Phi}}}}_*\cdot\mathcal{B}_*{{\boldsymbol{\Phi}}}_*d\Omega,$$
$$a_{13}=k_*^2\int_\Omega G \overline{{{\boldsymbol{\Phi}}}}_*\cdot{{\boldsymbol{\Phi}}}_*d\Omega,$$
$$d_{1j}={-}\int_{\Omega} \overline{\boldsymbol{\Phi}}_* \cdot \boldsymbol{\Theta}_j d\Omega.$$
The entries $a_{2l}$, $a_{3l}$ for $l=1,2,3$ and $d_{2j}$, $d_{3j}$ are defined similarly as $a_{1l}$ and $d_{1j}$ with $\overline {\boldsymbol {\Phi }}_*$ replaced by $\overline {\boldsymbol {\Psi }}^e_*$ and $\overline {\boldsymbol {\Psi }}^h_*$, respectively. It is easy to verify that all entries of the $3 \times 3$ coefficient matrix in Eq. (31) are real. In particular, due to the orthogonality condition in Eq. (20), we have $a_{21}= a_{31} = 0$. In the Appendix, it is shown that $\boldsymbol {\Theta }_j$ satisfies ${\cal PT} \boldsymbol {\Theta }_j = \boldsymbol {\Theta }_j$. This implies that ${\mathcal {P}\mathcal {T}}{\textbf{f}}_j={\textbf{f}}_j$, and thus the right hand side of Eq. (31) is real. Clearly, if $a_{22}a_{33}-a_{23}a_{32}\neq 0$, the coefficient matrix is invertible, and we can find real solutions for $k_j$, $\beta _j$ and $\gamma _j$. To ensure this condition is satisfied, we impose some conditions on the BIC and the perturbation profile $G$. For the BIC, the required condition is that $a_{22}$ and $a_{32}$ are not both zero, namely,
$$\left( \int_\Omega \overline{\boldsymbol{\Psi}}_*^e \cdot {\cal B}_* {\boldsymbol{\Phi}}_* d\Omega, \int_\Omega \overline{\boldsymbol{\Psi}}_*^h \cdot {\cal B}_* {\boldsymbol{\Phi}}_* d\Omega \right) \ne (0, 0).$$

We call a BIC satisfying the above condition a generic BIC. For the perturbation profile $G$, we require that $(a_{23}, a_{33})$ is not a multiple of $(a_{22}, a_{32})$, namely,

$$\left( \int_\Omega G \overline{\boldsymbol{\Psi}}_*^e \cdot {\boldsymbol{\Phi}}_* d\Omega, \int_\Omega G \overline{\boldsymbol{\Psi}}_*^h \cdot {\boldsymbol{\Phi}}_* d\Omega \right) \ne C \left( \int_\Omega \overline{\boldsymbol{\Psi}}_*^e \cdot {\cal B}_* {\boldsymbol{\Phi}}_* d\Omega, \int_\Omega \overline{\boldsymbol{\Psi}}_*^h \cdot {\cal B}_* {\boldsymbol{\Phi}}_* d\Omega \right)$$
for any constant $C$. This condition on $G$ depends on the BIC and the related scattering solutions. We call a perturbation profile satisfying this condition a generic profile.

Consequently, if the BIC is generic and the perturbation profile $G$ is generic, we can solve $k_j$, $\beta _j$ and $\gamma _j$ from Eq. (31). This implies that conditions (29) and (30) are satisfied, thus Eq. (28) has a solution ${\boldsymbol{\Phi} }_j$ satisfying condition (26). In particular, the fact that ${\cal PT} {\textbf{f}}_j = {\textbf{f}}_j$ implies ${\cal PT} {\boldsymbol{\Phi} }_j = {\boldsymbol{\Phi} }_j$. Moreover, since for any constant $C$, ${\boldsymbol{\Phi} }_j+C{\boldsymbol{\Phi} }_*$ is also a solution, we can define a unique solution satisfying the orthogonality condition:

$$\int_{\Omega}\overline{\boldsymbol{\Phi}}_j\cdot {\boldsymbol{\Phi}}_*d\Omega=0.$$
Therefore, if the powers series (22)–(25) are convergent, we have a BIC in the perturbed RSP waveguide for $\gamma$ given by Eq. (25). Since $\delta$ is small but arbitrary, the BIC is in fact a continuous family depending on $\delta$, and $\gamma$ varies with $\delta$ continuously. Clearly, the BIC corresponds to a curve in the $\delta$-$\gamma$ plane. In general, if a symmetric RSP waveguide is given by two real structural parameters, a generic BIC with $m\ne 0$ corresponds to a curve in the plane of these two parameters.

The theory above is developed for BICs with $m\ne 0$ and $\beta _* \ne 0$, and it remains valid for standing waves with $m \ne 0$ and $\beta _*=0$. As discussed earlier, we can assume the BIC with $\beta _* = 0$ and the related scattering solutions satisfy ${\cal P} {\boldsymbol{\Phi} }_* = \pm {\boldsymbol{\Phi} }_*$, ${\cal T} {\boldsymbol{\Phi} }_* = {\boldsymbol{\Phi} }_*$, ${\cal P} {\boldsymbol{\Psi} }^e_* = {\boldsymbol{\Psi} }^e_*$, ${\cal P} {\boldsymbol{\Psi} }^h_* = -{\boldsymbol{\Psi} }^h_*$, ${\cal T} {\boldsymbol{\Psi} }^l_* = {\boldsymbol{\Psi} }^l_*$ for $l \in \{e, h\}$, and Eq. (20). For the case when the BIC satisfies ${\cal P} {\boldsymbol{\Phi} }_* = {\boldsymbol{\Phi} }_*$, we have $a_{21}=a_{31}=0$ as before and also $a_{33}=d_{3j}=0$. For this case, a generic BIC with $\beta _*=0$ is defined as one satisfying $a_{32} \ne 0$, i.e.,

$$\int_\Omega \overline{\boldsymbol{\Psi}}_*^h \cdot {\cal B}_* {\boldsymbol{\Phi}}_* d\Omega \ne 0,$$
and a perturbation profile $G$ is called generic, if $a_{23} \ne 0$, namely,
$$\int_\Omega G \overline{\boldsymbol{\Psi}}_*^e \cdot {\boldsymbol{\Phi}}_* d\Omega \ne 0.$$
It is clear that the third equation in system (31) gives $\beta _j = 0$, and $k_j$ and $\gamma _j$ can be solved from
$$\left[ \begin{array}{cc} a_{11} & a_{13}\\ 0 & a_{23} \end{array} \right]\left[ \begin{array}{c} k_j\\ \gamma_j\\ \end{array} \right]=\left[ \begin{array}{c} d_{1j}\\ d_{2j} \end{array} \right].$$

The case when the BIC satisfies ${\cal P} {\boldsymbol{\Phi} }_* = - {\boldsymbol{\Phi} }_*$ is similar. Therefore, generic standing waves with $m \ne 0$ are not robust with respect to symmetric structural perturbations, but they can continue to exist if a structural parameter is properly tuned. Notice that such standing waves are not symmetry protected. If ${\boldsymbol{\Phi} }_*$ satisfies ${\cal P} {\boldsymbol{\Phi} }_* = {\boldsymbol{\Phi} }_*$, the $z$ components of the BIC, $E_z$ and $H_z$, are even and odd in $z$, respectively. A generic symmetric perturbation will turn this BIC to a resonant mode with $c_0^h \ne 0$ and $c_0^e=0$. In some sense, such a standing wave is partially symmetry protected.

For the case of $m=0$, any nondegenerate BIC has a definite polarization. If we proceed as above, the second or third rows of both the coefficient matrix and the right hand side of system (31) are zero. This implies that $\gamma _j$ can be chosen arbitrarily. In other words, $\gamma$ is a free parameter. As the existence of the BIC can be proved for $\gamma =0$, that is, for dielectric function given by $\varepsilon (\rho,z) = \varepsilon _* (\rho,z) + \delta F(\rho, z)$, the BIC is robust with respect to symmetric structural perturbations. Assuming the BIC and the related scattering solutions satisfy the same scaling and normalization conditions used above, we only need the additional assumption

$$\int_\Omega \overline{\boldsymbol{\Psi}}_*^l \cdot {\cal B}_* {\boldsymbol{\Phi}}_* d\Omega \ne 0,$$
where $l$ is chosen so that ${\boldsymbol{\Psi} }_*^l$ has the same polarization as ${\boldsymbol{\Phi} }_*$. The above is used as a definition for a generic BIC with $m=0$. The robustness theory for BICs with $m=0$ is similar to the theory developed in Ref. [45] for 2D structures with 1D periodicity.

In previous studies [45,53], the robustness of BICs in periodic structures sandwiched between two homogeneous media have been thoroughly analyzed. For 2D structures that are invariant in $x$, periodic in $y$, bounded in $z$, and symmetric in both $y$ and $z$, any guided mode has a wave vector $(\alpha, \beta )$ with $\alpha$ and $\beta$ associated with $x$ and $y$, respectively. It is known that in such a 2D structure with 1D periodicity, whether $\alpha$ is zero or not, generic nondegenerate BICs (with a frequency in the range of a single propagation diffraction order) are always robust [45,53]. A symmetric RSP waveguide is analogous to such a 2D structure, with the cylindrical coordinates $\rho$, $\phi$, $z$ correspond to $z$, $x$ and $y$, respectively, $m$ corresponds to $\alpha$, and $\beta$ remains fixed. However, as shown in this section, BICs in RSP waveguides exhibit different robustness for $m=0$ and $m\ne 0$, respectively. This is related to the fact that $m$ is a fixed integer that cannot be varied when the structure is perturbed, while $\alpha$ (for BICs in 2D structures with 1D periodicity) is a solution parameter that can be varied.

4. Numerical examples

In this section, we present numerical examples to illustrate our theory. Since there are many numerical examples for BICs with $m=0$ or $m\ne 0$ but $\beta =0$ [2830,3638], we concentrate on the case $m \ne 0$ and $\beta \ne 0$. According to the theory developed in section 3, such a BIC in an RSP waveguide is non-robust, but can often be found by tuning a single structural parameter. The numerical examples in this section are calculated using a numerical modal method based on piecewise polynomials of $z$ [54,55]. The method is similar to the widely used Fourier modal method, also called rigorous coupled-wave analysis (RCWA), and it has been validated by reproducing the examples in Refs. [36] and [30].

First, we consider a periodic array of circular dielectric disks shown schematically in Fig. 1. The disks are surrounded by air. Their radius, thickness, and dielectric constant are $R$, $d$ and $\varepsilon _1$, respectively. For fixed $d=L/2$ and $\varepsilon _1 = 3$, we look for a propagating BIC with $m=1$ by varying the radius $R$. For $R\approx 1.8335L$, we obtain a BIC with $k_* \approx 0.4936 (2\pi /L)$ and $\beta _* \approx 0.3678 (2\pi /L)$. In Fig. 2, we show the six field components of the BIC in the $xz$ plane for three periods in $z$. Due to a scaling, the electromagnetic field is dimensionless, but consistency among all six components is maintained. From Fig. 2, it is clear that the amplitudes of field components decay rapidly away from the core. This BIC with $m=1$ is non-robust with respect to structural perturbations that preserve the reflection symmetry in $z$. For fixed $d=L/2$ and $\varepsilon _1=3$, when $R$ is moved away from $1.8335L$, the BIC is destroyed, so that all resonant modes for $\beta$ near $\beta _* = 0.3678 (2\pi /L)$ have a finite $Q$ factor. In Figs. 3(a) and (b), we show the imaginary part of the normalized frequency $\omega L/(2\pi c) = kL/(2\pi )$, and the $Q$ factor [$Q = -0.5 \mbox {Re}(\omega )/\mbox {Im}(\omega )$] of the resonant modes for $R$ near $1.8335L$ and $\beta$ near $\beta _*$. Similarly, if $d=L/2$, $R$ is fixed at $1.8335L$, and $\varepsilon _1$ is moved away from $3$, the BIC is destroyed. The corresponding $\mbox {Im}(k)$ and $Q$ factor of the resonant modes are shown in Figs. 3(c) and (d).

 figure: Fig. 2.

Fig. 2. Scaled field patterns on the $xz$ plane of a propagating BIC with $m=1$ in a periodic array of circular disks. $H_x$, $H_y$ and $H_z$ are scaled magnetic field components multiplied by the free space impedance.

Download Full Size | PDF

 figure: Fig. 3.

Fig. 3. Imaginary part of free space wavenumber $k$ and $Q$ factor of resonant modes near a BIC in a periodic array of circular disks. The BIC is obtained for dielectric constant $\varepsilon _1 = 3$ and radius $R = 1.8335L$. Results for $R$ near $1.8335L$ are shown in panels (a) and (b). Those for $\varepsilon _1$ near $3$ are shown in panels (c) and (d).

Download Full Size | PDF

Next, we consider a fiber grating consisting of three dielectric media with dielectric constants $\varepsilon _1$, $\varepsilon _2$ and $\varepsilon _3$, respectively. One period of the structure contains one segment of $\varepsilon _1$ medium with length $2d_1$, two segments of $\varepsilon _2$ medium with length $d_2$, and one segment of $\varepsilon _3$ medium with length $d_3$. The radius $R$ of the fiber grating is independent of $z$. If we choose the center of the segment with dielectric constant $\varepsilon _3$ as the origin, the dielectric function $\varepsilon (\rho,z)$ has a reflection symmetry in $z$ and is periodic in $z$ with period $L=2d_1 + 2d_2 + d_3$. The fiber grating is surrounded by air, thus $\varepsilon (\rho,z) = 1$ if $\rho > R$. In Fig. 4, we show one period of the fiber grating for $-L/2 < z < L/2$. Instead of a single segment of length $2d_1$, this unit cell contains two segments of $\varepsilon _1$ medium with length $d_1$. For fixed $R=L$, $d_1=d_2=L/8$, $d_3 = L/2$, $\varepsilon _2 = 3$ and $\varepsilon _3 = 3.4$, we search for a BIC with $m=1$ by varying $\varepsilon _1$. A BIC with $k_* \approx 0.6466 (2\pi /L)$ and $\beta _* \approx 0.1282 (2\pi /L)$ is obtained at $\varepsilon _1 \approx 3.5565$. The field components of this BIC are shown in Fig. 5. The non-robustness of the BIC is verified in Fig. 6, where the $Q$ factor of the resonant mode with fixed $\beta = \beta _*$ is shown as a function $\varepsilon _1$. Clearly, as $\varepsilon _1$ moves away from 3.5565, the BIC is destroyed and becomes a resonant mode with a finite $Q$ factor.

 figure: Fig. 4.

Fig. 4. One period of a fiber grating consisting of segments with dielectric constants $\varepsilon _1$, $\varepsilon _2$ and $\varepsilon _3$.

Download Full Size | PDF

 figure: Fig. 5.

Fig. 5. Scaled field patterns on the $xz$ plane of a propagating BIC with $m=1$ in a fiber grating. $H_x$, $H_y$ and $H_z$ are scaled magnetic field components multiplied by the free space impedance.

Download Full Size | PDF

 figure: Fig. 6.

Fig. 6. Quality factor of resonant modes in a fiber grating for fixed Bloch wavenumber $\beta = \beta _* = 0.1282 (2\pi /L)$ and dielectric constant $\varepsilon _1$ near 3.5565. A BIC with $m=1$ and Bloch wavenumber $\beta _*$ is obtained at $\varepsilon _1 = 3.5565$.

Download Full Size | PDF

To verify the theory developed in section 3, we introduce two real parameters $\delta$ and $\gamma$ by

$$\varepsilon_3 = 3.4 + \delta, \quad \varepsilon_1 = 3.5565 + \gamma.$$
To follow the notations used in section 3, we can define perturbation profiles $F$ and $G$, such that $F=1$ (or $G=1$) in the medium with dielectric constant $\varepsilon _3$ (or $\varepsilon _1$) and $F=0$ (or $G=0$) otherwise. Since the small $\delta$ and $\gamma$ correspond to structural perturbations that preserve the reflection symmetry in $z$, the BIC should exist continuously with $\delta$, provided that $\gamma$ is properly chosen as a function of $\delta$. This is confirmed by the numerical results shown in Fig. 7. Fig. 7(a) shows a curve in the $\delta$-$\gamma$ plane on which the BIC exists continuously. Figure 7(b) shows the corresponding free space wavenumber $k$ and Bloch wavenumber $\beta$ of the BIC as $\delta$ is varied and $\gamma$ follows $\delta$ as in Fig. 7(a). The asterisks correspond to the particular BIC for $\delta =\gamma =0$.

 figure: Fig. 7.

Fig. 7. Parametric dependence of a BIC in a fiber grating. Panel (a) shows a curve in the parameter plane on which the BIC exists continuously. Panel (b) shows $k$ and $\beta$ of the BIC as functions of parameter $\delta$.

Download Full Size | PDF

5. Conclusion

Existing works on BICs in RSP waveguides are mostly concerned with specific structures, such as fiber gratings and 1D arrays of spheres or circular disks. There are also numerical studies about the dependence of BICs on specific structural parameters, such as the radius of the spheres and the dielectric constant of the disks. The theoretical analysis of this paper is developed for BICs in a general RSP waveguide with a reflection symmetry in $z$. The concept of robustness is defined with respect to small but arbitrary perturbations that preserve the periodicity and reflection symmetry in $z$. It is shown that generic BICs with azimuthal index $m=0$ are robust, and those with $m\ne 0$ are non-robust, but can nevertheless continue to exist if a parameter in the perturbation is properly tuned. The theory is developed by proving the existence of a BIC in the perturbed structure mathematically. It is applicable to BICs with free space and Bloch wavenumbers satisfying condition (19). In addition, the BICs are assumed to be generic, as defined by condition (36) in general and condition (38) or (41) for special cases. For the case of non-robust BICs with $m\ne 0$, the continual existence is only proved when the perturbation profile $G$ associated with the tunable parameter $\gamma$ is generic, as defined by condition (37) or (39).

The robustness theory developed in this paper can be compared with the theory for BICs in symmetric 2D structures (with 1D periodicity) given by a dielectric function $\varepsilon (y,z)$ that is independent of $x$, periodic in $y$, bounded in $z$, and symmetric in both $y$ and $z$. The cylindrical coordinates $\rho$, $\phi$ and $z$ for describing RSP waveguides correspond to $z$, $x$ and $y$, respectively. In such a 2D structure, a generic BIC with wave vector $(\alpha, \beta )$ and a frequency in the band for single propagating diffraction order, is robust with respect to all symmetry-preserving perturbations. This is true for both $\alpha = 0$ and $\alpha \ne 0$. Notice that $\alpha$ corresponds to $m$ for BICs in RSP waveguides, but unlike the fixed integer $m$, $\alpha$ is a changeable solution parameter for BICs in 2D structures. Consequently, BICs in 2D structure with $\alpha \ne 0$ and those in RSP waveguides with $m\ne 0$ exhibit distinct robustness properties.

Appendix

In this appendix, formulas and symmetries of $\boldsymbol {\Theta }_j(\rho,z)$ are given. Let

$$K_0=k_*^2,\quad K_l=2k_*k_l+\sum_{q=1}^{l-1}k_qk_{l-q},\quad l\geq 1,$$
then $\boldsymbol {\Theta }_j(\rho,z)$ could be written as
$$\boldsymbol{\Theta}_j(\rho,z)=\sum_{l=1}^{j-1}\left[(V_l+\mathcal{D}_l){\boldsymbol{\Phi}}_{j-l}+\beta_l \hat{z}\times \beta_{j-l} \hat{z}\times {\boldsymbol{\Phi}}_*\right]+W_j{\boldsymbol{\Phi}}_*,\;j\geq1,$$
where,
$$V_l=W_l+2 k_* k_l \varepsilon_*+k_*^2\gamma_lG,$$
$$W_l=\sum_{q=1}^{l-1}\left(k_q k_{l-q} \varepsilon_*+K_q\gamma_{l-q}G\right)+K_{l-1}F,$$
$$\mathcal{D}_l=\beta_l\mathcal{B}_*+\sum_{q=1}^{l-1}\beta_l\beta_{l-q}\hat{z}\times\hat{z}\times{\cdot}.$$
For any 3D vector function ${\textbf{f}}$, it is easy to verify that ${\mathcal {P}\mathcal {T}} \mathcal {B}_*{\textbf{f}}=\mathcal {B}_*{\mathcal {P}\mathcal {T}}{\textbf{f}}$ and ${\mathcal {P}\mathcal {T}}\left [\hat {z}\times {\textbf{f}}\right ]=-\hat {z}\times {\textbf{f}}$. If ${\boldsymbol{\Phi} }_l$ satisfies ${\mathcal {P}}{\mathcal {T}}{\boldsymbol{\Phi} }_l={\boldsymbol{\Phi} }_l$ for $0<l<j$, we then can conclude that $\boldsymbol {\Theta }_j$ satisfies ${\mathcal {P}}{\mathcal {T}}\boldsymbol {\Theta }_j=\boldsymbol {\Theta }_j$ as well.

Funding

Research Grants Council of Hong Kong Special Administrative Region, China (CityU 11304619).

Disclosures

The authors declare no conflicts of interest.

Data availability

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.

References

1. J. von Neumann and E. Wigner, “Über merkwürdige diskrete eigenwerte,” Phys. Z. 30, 465–467 (1929).

2. L. Fonda and R. G. Newton, “Theory of resonance reactions,” Ann. Phys. 10(4), 490–515 (1960). [CrossRef]  

3. F. H. Stillinger and D. R. Herrick, “Bound states in the continuum,” Phys. Rev. A 11(2), 446–454 (1975). [CrossRef]  

4. D. C. Marinica, A. G. Borisov, and S. V. Shabanov, “Bound states in the continuum in photonics,” Phys. Rev. Lett. 100(18), 183902 (2008). [CrossRef]  

5. E. N. Bulgakov and A. F. Sadreev, “Bound states in the continuum in photonic waveguides inspired by defects,” Phys. Rev. B 78(7), 075105 (2008). [CrossRef]  

6. Y. Plotnik, O. Peleg, F. Dreisow, M. Heinrich, S. Nolte, A. Szameit, and M. Segev, “Experimental observation of optical bound states in the continuum,” Phys. Rev. Lett. 107(18), 183901 (2011). [CrossRef]  

7. C. W. Hsu, B. Zhen, A. D. Stone, J. D. Joannopoulos, and M. Soljačić, “Bound states in the continuum,” Nat. Rev. Mater. 1(9), 16048 (2016). [CrossRef]  

8. K. Koshelev, G. Favraud, A. Bogdanov, Y. Kivshar, and A. Fratalocchi, “Nonradiating photonics with resonant dielectric nanostructures,” Nanophotonics 8(5), 725–745 (2019). [CrossRef]  

9. S. I. Azzam and A. V. Kildishev, “Photonic bound states in the continuum: from basics to applications,” Adv. Opt. Mater. 9(1), 2001469 (2021). [CrossRef]  

10. S. Joseph, S. Pandey, S. Sarkar, and J. Joseph, “Bound states in the continuum in resonant nanostructures: an overview of engineered materials for tailored applications,” Nanophotonics 10(17), 4175–4207 (2021). [CrossRef]  

11. A.-S. Bonnet-Bendhia and F. Starling, “Guided waves by electromagnetic gratings and non-uniqueness examples for the diffraction problem,” Math. Meth. Appl. Sci. 17(5), 305–338 (1994). [CrossRef]  

12. D. V. Evans, M. Levitin, and D. Vassiliev, “Existence theorems for trapped modes,” J. Fluid Mech. 261, 21–31 (1994). [CrossRef]  

13. S. P. Shipman and S. Venakides, “Resonance and bound states in photonic crystal slabs,” SIAM J. Appl. Math. 64(1), 322–342 (2003). [CrossRef]  

14. R. Porter and D. V. Evans, “Embedded Rayleigh-Bloch surface waves along periodic rectangular arrays,” Wave Motion 43(1), 29–50 (2005). [CrossRef]  

15. R. F. Ndangali and S. V. Shabanov, “Electromagnetic bound states in the radiation continuum for periodic double arrays of subwavelength dielectric cylinders,” J. Math Phys. 51(10), 102901 (2010). [CrossRef]  

16. C. W. Hsu, B. Zhen, J. Lee, S.-L. Chua, S. G. Johnson, J. D. Joannopoulos, and M. Soljačić, “Observation of trapped light within the radiation continuum,” Nature 499(7457), 188–191 (2013). [CrossRef]  

17. Y. Yang, C. Peng, Y. Liang, Z. Li, and S. Noda, “Analytical perspective for bound states in the continuum in photonic crystal slabs,” Phys. Rev. Lett. 113(3), 037401 (2014). [CrossRef]  

18. E. N. Bulgakov and A. F. Sadreev, “Bloch bound states in the radiation continuum in a periodic array of dielectric rods,” Phys. Rev. A 90(5), 053801 (2014). [CrossRef]  

19. Z. Hu and Y. Y. Lu, “Standing waves on two-dimensional periodic dielectric waveguides,” J. Opt. 17(6), 065601 (2015). [CrossRef]  

20. X. Gao, C. W. Hsu, B. Zhen, X. Lin, J. D. Joannopoulos, M. Soljačić, and H. Chen, “Formation mechanism of guided resonances and bound states in the continuum in photonic crystal slabs,” Sci. Rep. 6(1), 31908 (2016). [CrossRef]  

21. L. Yuan and Y. Y. Lu, “Propagating bloch modes above the lightline on a periodic array of cylinders,” J. Phys. B 50(5), 05LT01 (2017). [CrossRef]  

22. A.-S. Bonnet-Bendhia and F. Mahé, “A guided mode in the range of the radiation modes for a rib waveguide,” J. Opt. 28(1), 41–43 (1997). [CrossRef]  

23. S. Weimann, Y. Xu, R. Keil, A. E. Miroshnichenko, A. Tünnermann, S. Nolte, A. A. Sukhorukov, A. Szameit, and Y. S. Kivshar, “Compact surface fano states embedded in the continuum of waveguide arrays,” Phys. Rev. Lett. 111(24), 240403 (2013). [CrossRef]  

24. C.-L. Zou, J.-M. Cui, F.-W. Sun, X. Xiong, X.-B. Zou, Z.-F. Han, and G.-C. Guo, “Guiding light through optical bound states in the continuum for ultrahigh-Q microresonators,” Laser Photonics Rev. 9(1), 114–119 (2015). [CrossRef]  

25. J. Gomis-Bresco, D. Artigas, and L. Torner, “Anisotropy-induced photonic bound states in the continuum,” Nat. Photonics 11(4), 232–236 (2017). [CrossRef]  

26. S. Mukherjee, J. Gomis-Bresco, P. Pujol-Closa, D. Artigas, and L. Torner, “Topological properties of bound states in the continuum in geometries with broken anisotropy symmetry,” Phys. Rev. A 98(6), 063826 (2018). [CrossRef]  

27. S. Mukherjee, J. Gomis-Bresco, P. Pujol-Closa, D. Artigas, and L. Torner, “Angular control of anisotropy-induced bound states in the continuum,” Opt. Lett. 44(21), 5362–5365 (2019). [CrossRef]  

28. E. N. Bulgakov and A. F. Sadreev, “Light trapping above the light cone in one-dimensional arrays of dielectric spheres,” Phys. Rev. A 92(2), 023816 (2015). [CrossRef]  

29. E. N. Bulgakov and A. F. Sadreev, “Propagating bloch bound states with orbital angular momentum above the light line in the array of dielectric spheres,” J. Opt. Soc. Am. A 34(6), 949–952 (2017). [CrossRef]  

30. X. Gao, Z. Bo, M. Soljačić, H. Chen, and C. W. Hsu, “Bound states in the continuum in fiber bragg gratings,” ACS Photonics 6(11), 2996–3002 (2019). [CrossRef]  

31. A. Kodigala, T. Lepetit, Q. Gu, B. Bahari, Y. Fainman, and B. Kanté, “Lasing action from photonic bound states in continuum,” Nature 541(7636), 196–199 (2017). [CrossRef]  

32. S. Romano, A. Lamberti, M. Masullo, E. Penzo, S. Cabrini, I. Rendina, and V. Mocella, “Optical biosensors based on photonic crystals supporting bound states in the continuum,” Materials 11(4), 526 (2018). [CrossRef]  

33. R. E. Jacobsen, A. Krasnok, S. Arslanagic, A. V. Lavrinenko, and A. Alu, “Boundary-induced embedded eigenstate in a single resonator for advanced sensing,” ACS Photonics 9(6), 1936–1943 (2022). [CrossRef]  

34. X. Cui, H. Tian, Y. Du, G. Shi, and Z. Zhou, “Normal incidence filters using symmetry-protected modes in dielectric subwavelength gratings,” Sci. Rep. 6(1), 36066 (2016). [CrossRef]  

35. S. Han, L. Cong, Y. K. Srivastava, B. Qiang, M. V. Rybin, A. Kumar, R. Jain, W. X. Lim, V. G. Achanta, S. S. Prabhu, Q. J. Wang, Y. S. Kivshar, and R. Singh, “All-dielectric active terahertz photonics driven by bound states in the continuum,” Adv. Mater. 31(37), 1901921 (2019). [CrossRef]  

36. E. N. Bulgakov and A. F. Sadreev, “Bound states in the continuum with high orbital angular momentum in a dielectric rod with periodically modulated permittivity,” Phys. Rev. A 96(1), 013841 (2017). [CrossRef]  

37. Z. F. Sadrieva, M. A. Belyakov, M. A. Balezin, P. V. Kapitanova, E. A. Nenasheva, A. F. Sadreev, and A. A. Bogdanov, “Experimental observation of a symmetry-protected bound state in the continuum in a chain of dielectric disks,” Phys. Rev. A 99(5), 053804 (2019). [CrossRef]  

38. S. Kim, K.-H. Kim, and J. F. Cahoon, “Optical bound states in the continuum with nanowire geometric superlattices,” Phys. Rev. Lett. 122(18), 187402 (2019). [CrossRef]  

39. B. Zhen, C. W. Hsu, L. Lu, A. D. Stone, and M. Soljačić, “Topological nature of optical bound states in the continuum,” Phys. Rev. Lett. 113(25), 257401 (2014). [CrossRef]  

40. E. N. Bulgakov and D. N. Maksimov, “Bound states in the continuum and polarization singularities in periodic arrays of dielectric rods,” Phys. Rev. A 96(6), 063833 (2017). [CrossRef]  

41. E. N. Bulgakov and D. N. Maksimov, “Topological bound states in the continuum in arrays of dielectric spheres,” Phys. Rev. Lett. 118(26), 267401 (2017). [CrossRef]  

42. W. Liu, B. Wang, Y. Zhang, J. Wang, M. Zhao, F. Guan, X. Liu, L. Shi, and J. Zi, “Circularly polarized states spawning from bound states in the continuum,” Phys. Rev. Lett. 123(11), 116104 (2019). [CrossRef]  

43. T. Yoda and M. Notomi, “Generation and annihilation of topologically protected bound states in the continuum and circularly polarized states by symmetry breaking,” Phys. Rev. Lett. 125(5), 053902 (2020). [CrossRef]  

44. D. A. Bykov, E. A. Bezus, and L. L. Doskolovich, “Bound states in the continuum and strong phase resonances in integrated gires-tournois interferometer,” Nanophotonics 9(1), 83–92 (2020). [CrossRef]  

45. L. Yuan and Y. Y. Lu, “Bound states in the continuum on periodic structures: perturbation theory and robustness,” Opt. Lett. 42(21), 4490–4493 (2017). [CrossRef]  

46. L. Yuan and Y. Y. Lu, “On the robustness of bound states in the continuum in waveguides with lateral leakage channels,” Opt. Express 29(11), 16695–16709 (2021). [CrossRef]  

47. L. Yuan and Y. Y. Lu, “Conditional robustness of propagating bound states in the continuum in structures with two-dimensional periodicity,” Phys. Rev. A 103(4), 043507 (2021). [CrossRef]  

48. 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(19), 193903 (2018). [CrossRef]  

49. 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(4), 043827 (2020). [CrossRef]  

50. C. M. Bender and S. Boettcher, “Real spectra in non-hermitian hamiltonians having ${\mathcal {P}\mathcal {T}}$ symmetry,” Phys. Rev. Lett. 80(24), 5243–5246 (1998). [CrossRef]  

51. H. A. Haus, Waves and Fields in Optoelectronics (Prentice-Hall, Inc., Englewoord Cliffs, New Jersey, 1984).

52. L. Yuan and Y. Y. Lu, “Parametric dependence of bound states in the continuum on periodic structures,” Phys. Rev. A 102(3), 033513 (2020). [CrossRef]  

53. L. Yuan, X. Luo, and Y. Y. Lu, “Parametric dependence of bound states in the continuum in periodic structures: Vectorial cases,” Phys. Rev. A 104(2), 023521 (2021). [CrossRef]  

54. D. Song, L. Yuan, and Y. Y. Lu, “Fourier-matching pseudospectral modal method for diffraction gratings,” J. Opt. Soc. Am. A 28(4), 613–620 (2011). [CrossRef]  

55. G. Granet, “Spectral element method with modified Legendre polynomials for modal analysis of lamellar gratings,” J. Opt. Soc. Am. A 38(1), 52–59 (2021). [CrossRef]  

Data availability

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.

Cited By

Optica participates in Crossref's Cited-By Linking service. Citing articles from Optica Publishing Group journals and other participating publishers are listed here.

Alert me when this article is cited.


Figures (7)

Fig. 1.
Fig. 1. A periodic array of circular disks surrounded by air. The period of the array is $L$. The radius, thickness and dielectric constant of the disks are $R$, $d$ and $\varepsilon _1$, respectively.
Fig. 2.
Fig. 2. Scaled field patterns on the $xz$ plane of a propagating BIC with $m=1$ in a periodic array of circular disks. $H_x$, $H_y$ and $H_z$ are scaled magnetic field components multiplied by the free space impedance.
Fig. 3.
Fig. 3. Imaginary part of free space wavenumber $k$ and $Q$ factor of resonant modes near a BIC in a periodic array of circular disks. The BIC is obtained for dielectric constant $\varepsilon _1 = 3$ and radius $R = 1.8335L$. Results for $R$ near $1.8335L$ are shown in panels (a) and (b). Those for $\varepsilon _1$ near $3$ are shown in panels (c) and (d).
Fig. 4.
Fig. 4. One period of a fiber grating consisting of segments with dielectric constants $\varepsilon _1$, $\varepsilon _2$ and $\varepsilon _3$.
Fig. 5.
Fig. 5. Scaled field patterns on the $xz$ plane of a propagating BIC with $m=1$ in a fiber grating. $H_x$, $H_y$ and $H_z$ are scaled magnetic field components multiplied by the free space impedance.
Fig. 6.
Fig. 6. Quality factor of resonant modes in a fiber grating for fixed Bloch wavenumber $\beta = \beta _* = 0.1282 (2\pi /L)$ and dielectric constant $\varepsilon _1$ near 3.5565. A BIC with $m=1$ and Bloch wavenumber $\beta _*$ is obtained at $\varepsilon _1 = 3.5565$.
Fig. 7.
Fig. 7. Parametric dependence of a BIC in a fiber grating. Panel (a) shows a curve in the parameter plane on which the BIC exists continuously. Panel (b) shows $k$ and $\beta$ of the BIC as functions of parameter $\delta$.

Equations (51)

Equations on this page are rendered with MathJax. Learn more.

ε ( ρ , z ) = ε ( ρ , z + L ) , ε ( ρ , z ) = ε ( ρ , z ) ,
× × E ( ρ , z ) k 2 ε E ( ρ , z ) = 0 ,
L Φ := ( c + i β z ^ ) × ( c + i β z ^ ) × Φ ( ρ , z ) k 2 ε Φ ( ρ , z ) = 0 ,
× [ f ( ρ , z ) e i ( m ϕ + β z ) ] = [ ( c + i β z ^ ) × f ( ρ , z ) ] e i ( m ϕ + β z )
H e = [ H ρ e H ϕ e H z ] , E e = [ E ρ e E ϕ e 0 ]
E h = [ E ρ h E ϕ h E z ] , H h = [ H ρ h H ϕ h 0 ]
E = E e + E h = e i ( m ϕ + β z ) p [ c p e e p e ( ρ ) + c p h e p h ( ρ ) ] e i 2 π p z / L ,
e p e ( ρ ) = k [ m ρ H m ( 1 ) ( κ p ρ ) i κ p H m ( 1 ) ( κ p ρ ) 0 ] , e p h ( ρ ) = [ β ^ p κ p H m ( 1 ) ( κ p ρ ) i m β ^ p ρ H m ( 1 ) ( κ p ρ ) i κ p 2 H m ( 1 ) ( κ p ρ ) ] .
e p e ( ρ ) 1 ρ d p e e i κ p ρ , e p h ( ρ ) 1 ρ d p h e i κ p ρ ,  as  ρ ,
ρ Φ κ p 0 ( c p e d p e + c p h d p h ) e i κ p ρ e i 2 π p z / L .
P f = [ f ρ ( ρ , z ) f ϕ ( ρ , z ) f z ( ρ , z ) ] , T f = [ f ¯ ρ ( ρ , z ) f ¯ ϕ ( ρ , z ) f ¯ z ( ρ , z ) ] ,
P T = T P , P 2 = T 2 = I ,
P T L = L P T ,
T L = L T ,
E i , e = [ P T e 0 e ( ρ ) ] e i ( m ϕ + β z ) .
ρ Ψ e ( s e e d 0 e + s e h d 0 h ) e i κ 0 ρ + ( P T d 0 e ) e i κ 0 ρ ,
Ψ ^ i , e = ( 1 + s ¯ e e ) [ P T e 0 e ( ρ ) ] + s ¯ e h [ P T e 0 h ( ρ ) ] ,
E i , e = [ T e 0 e ( ρ ) ] e i m ϕ .
ρ P Ψ e ( s e e d 0 e s e h d 0 h ) e i k ρ + ( T d 0 e ) e i k ρ .
| β | < k < 2 π L | β | ,
1 L 2 Ω ε | Φ | 2 d Ω = 1 , Ω ε Φ ¯ Ψ l d Ω = 0 , l { e , h } ,
ε ( ρ , z ) = ε ( ρ , z ) + δ F ( ρ , z ) + γ G ( ρ , z ) ,
Φ = Φ + δ Φ 1 + δ 2 Φ 2 + ,
k = k + δ k 1 + δ 2 k 2 + ,
β = β + δ β 1 + δ 2 β 2 + ,
γ = δ γ 1 + δ 2 γ 2 + .
Φ j ( ρ , z ) = Φ j ( ρ , z + L ) , P T Φ j = Φ j , ρ Φ j 0  as  ρ .
L Φ = 0 ,
L Φ j = f j := β j B Φ + ( 2 k k j ε + k 2 γ j G ) Φ + Θ j ( ρ , z ) , j 1 ,
L w = ( c + i β z ^ ) × ( c + i β z ^ ) × w k 2 ε w ,
B w = i [ z ^ × ( c + i β z ^ ) × w + ( c + i β z ^ ) × z ^ × w ] ,
Ω Φ ¯ f j d Ω = 0 ,
Ω Ψ ¯ l f j d Ω = 0 , l { e , h } .
[ a 11 a 12 a 13 a 21 a 22 a 23 a 31 a 32 a 33 ] [ k j β j γ j ] = [ d 1 j d 2 j d 3 j ] ,
a 11 = 2 k Ω ε Φ ¯ Φ d Ω = 2 k L 2 ,
a 12 = Ω Φ ¯ B Φ d Ω ,
a 13 = k 2 Ω G Φ ¯ Φ d Ω ,
d 1 j = Ω Φ ¯ Θ j d Ω .
( Ω Ψ ¯ e B Φ d Ω , Ω Ψ ¯ h B Φ d Ω ) ( 0 , 0 ) .
( Ω G Ψ ¯ e Φ d Ω , Ω G Ψ ¯ h Φ d Ω ) C ( Ω Ψ ¯ e B Φ d Ω , Ω Ψ ¯ h B Φ d Ω )
Ω Φ ¯ j Φ d Ω = 0.
Ω Ψ ¯ h B Φ d Ω 0 ,
Ω G Ψ ¯ e Φ d Ω 0.
[ a 11 a 13 0 a 23 ] [ k j γ j ] = [ d 1 j d 2 j ] .
Ω Ψ ¯ l B Φ d Ω 0 ,
ε 3 = 3.4 + δ , ε 1 = 3.5565 + γ .
K 0 = k 2 , K l = 2 k k l + q = 1 l 1 k q k l q , l 1 ,
Θ j ( ρ , z ) = l = 1 j 1 [ ( V l + D l ) Φ j l + β l z ^ × β j l z ^ × Φ ] + W j Φ , j 1 ,
V l = W l + 2 k k l ε + k 2 γ l G ,
W l = q = 1 l 1 ( k q k l q ε + K q γ l q G ) + K l 1 F ,
D l = β l B + q = 1 l 1 β l β l q z ^ × z ^ × .
Select as filters


Select Topics Cancel
© Copyright 2024 | Optica Publishing Group. All rights reserved, including rights for text and data mining and training of artificial technologies or similar technologies.