## Abstract

The Zak phase and topological plasmonic Tamm states in plasmonic crystals based on periodic metal-insulator-metal waveguides are systematically investigated. We reveal that robust topological interfacial states against structural defects exist when the Zak phase between two adjoining plasmonic lattices are different in a common band gap. A kind of efficient admittance-based transfer matrix method is proposed to calculate and optimize the configuration with inverse symmetry. The topologically protected states are favorable for the spatial confinement and enhancement of electromagnetic fields, which open a new avenue for topological photonic applications.

© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement

## 1. Introduction

An important aspect in the development of nano-optics is analogous studies of photons with the electrons in solid-state materials. Topological insulator, a compelling and peculiar material that possesses non-trivial topological bounded states protected by time-reversal symmetry, has drawn considerable attentions due to its elusive properties of one-way spin-polarized transport [1]. Over the past decades, a number of topological states have been successfully emulated in photonic systems, such as magneto-optical photonic crystals [2], coupled resonator optical waveguides [3,4], photonic Floquet topological insulators [5], topological crystalline insulators [6,7], topological insulator lasers [8–11]. Moreover, topological states in plasmonic systems also attract growing interests in the last few years. The topological plasmonic interfacial modes, as a successful extension of the famous Su-Schrieffer-Heeger (SSH) model for polyacetylene-type solitons to plasmonic nanoparticles chains [12–15] and plasmonics waveguide arrays [16–18], exist when the nearby structures possess different Zak phases. Nowadays, more complex plasmonic structures composed of metallic patterns [19,20] and graphene superlattices [21,22] are demonstrated to construct topologically protected one-way edge states. The nontrival topologically protected states have opened up a fascinating avenue for topological photonic applications.

On the other hand, plasmonic crystals (PCs), which can be recognized as photonic crystals composed of plasmonic materials, can also induce an energy band structure analogous to superlattices and photonic crystals. Meanwhile, the electromagnetic surface states such as Tamm states would exist due to the strong Bragg scattering [23]. Tamm states, one of the surface Bloch states localized at the interfaces of truncated crystalline materials, have been sufficiently studied in superlattices [24], photonic crystals [25] and photonic lattices [26]. Unlike conventional Tamm states, the Plasmon Tamm states (PTSs) formed by MIM Bragg reflector (BR) and metal exist in subwavelength unit cells [27,28]. Attributing to strong confinement of surface plasmon polaritons(SPPs), the electromagnetic fields can be enhanced by three orders of magnitude. However, the high index core is required to contact with the metal end according to the general phase-matching condition in these configurations, as a result, the Zak phase can assume any value as a function of the band index due to the lack of spatial inverse symmetry [29]. Fortunately, some research groups have found that the topologically protected interfacial state can exist in 1D photonic crystals [30–32]. Nevertheless, large scale and relatively high operation light intensity limit the applications of traditional 1D photonic crystals in nanoscale all-optical circuits, such as fluorescence enhancement and bistable switches. Inspired by these works in 1D photonic crystals, a topologic plasmonic Tamm states (TPTSs) can be designed in the plasmonic materials with spatial inverse symmetry to overcome these drawbacks of photonic crystals.

In this paper, analogous to the SSH model in polyacetylene, the TPTSs are proposed in connected plasmonic crystals (PCs). The topological transition points are found in PCs and regarded as parent states to construct a topologically protected interfacial state via geometrical parameters tuning. Regarding to the symmetric PCs, an efficient admittance-based transfer matrix method (TMM) is proposed to calculate and optimize the structure, we find that the symmetric PCs can be regarded as homogenous layer with effective phase delay and equivalent characterised admittance. Through calculating the reflective spectra by the proposed method, the TPTSs are found to be robust against the inserted defect of the PCs. Our proposed TPTSs provide an efficient way to trap and enhance the plasmon wave, which makes it a promising optical device for integrated photonic circuits and able to provide significant applications, such as all-optical bistable switches [33], harmonic wave generation [34], and fluorescence enhancement [35].

## 2. Results and discussion

Our proposed scheme composed of two connected PCs, i.e., PC1 and PC2, is indicated in Fig. 1(a). The PCs consist of A/B/A layered metal-insulator-metal (MIM) waveguides with different unit cells as marked by the dashed frame in the inset of Fig. 1(a). The dielectric core has a width of *w* = 40 nm. Here, the cores of layer A and B are chosen as air and SiO_{2}, respectively. And the thicknesses of layer A and B are defined as 2*d _{A}*

_{1},

*d*

_{B}_{1}in PC1 and 2

*d*

_{A}_{2},

*dB*

_{2}in PC2, respectively. And the periods of PC1 and PC2 are Λ

_{1}= 2

*d A*

_{1}+

*d B*

_{1}and Λ

_{2}= 2

*d*

_{A}_{2}+

*d*

_{B}_{2}. For the sake of simplicity, the metal is silver characterized by a Drude dielectric function $\u03f5\left(\omega \right)={\u03f5}_{\infty}-{\omega}_{p}^{2}/({\omega}^{2}+\mathrm{i}\gamma \omega )$ for optical communication frequencies, with

*ϵ*

_{∞}= 3.7,

*ω*= 9.1 eV, and

_{p}*γ*= 0.018 eV, and the refractive index of SiO

_{2}is set as 1.5. The plasmon wave propagates along

*z-*direction in MIM waveguide.

Let us start by considering the propagating behavior of SPPs in a homogeneous MIM waveguide. According to the transmission line theory, the characteristic admittance of MIM waveguides can be expressed as:

*w*is the width in

*y*-direction,

*ϵ*

_{0}

*ϵ*is the permittivity of the dielectric core,

_{d}*d*denotes the unit length in

_{x}*x-*direction, and

*β*denotes the propagation constants of the MIM structure, which can be solved analytically through the dispersion relations of the odd TM mode (

*E*is odd) [36]: − tanh(

_{z}*k*/2) =

_{d}w*k*/

_{m}ϵ_{d}*k*, where

_{d}ϵ_{m}*ϵ*

_{m}_{,}

*are the dielectric constants, and${k}_{m,d}=\sqrt{{\beta}^{2}-{\u03f5}_{m,d}{k}_{0}^{2}}$ denote the transverse propagation constants of the metal and the dielectric, respectively.*

_{d}With the characteristic admittance, the band structure of a binary PC shown in Fig. 1(a) can be easily found from the effective phase delay in a unit cell (see methods)

*δ*= 2

_{A}*β*,

_{A}d_{A}*δ*=

_{B}*β*and

_{B}d_{B}*δ*=

*K*Λ are the actual phase delay in slab A, slab B and a unit cell, while

*η*and

_{A}*η*are the characteristic admittance of slab A and B.

_{B}*K*and Λ = 2

*d*+

_{A}*d*are the Bloch wavevector and period of the PC, respectively. In order to construct a topological nontrivial interface state, one needs to make sure the band topologies of the two PCs are different inside a common band gap. A commonly used way to achieve this goal is tuning the system parameters across a topological transition point. Therefore, the topological phase transition states are usually used as parent states to construct topologically protected edge states. The critical parameters of topological phase transition point in PCs could be adjusted to the crossing of band structure according to the condition:

_{B}It is worth noting that the optical loss should be omitted when calculating the band structure in order to realize isolated energy bands. For the telecom optical wavelength at 1550 nm (≈0.8 eV), we have *β _{A}* = 5.891

*µm*

^{−1}and

*β*= 8.868

_{B}*µm*

^{−1}, respectively. According to the band crossing theory [30], the band

*m*≡

*m*

_{1}+

*m*

_{2}and band

*m*− 1 will cross at the frequency $\tilde{\omega}$ satisfying ${\delta}_{A}\left(\tilde{\omega}\right)+{\delta}_{B}\left(\tilde{\omega}\right)=m\pi $. When the gap is open, the frequency derived by ${\delta}_{A}\left(\tilde{\omega}\right)+{\delta}_{B}\left(\tilde{\omega}\right)=m\pi $ labels the midgap positions of the

*m*th gap. At the lowest crossing frequency, the dependence of the band structure on the thickness of slab A is shown in Fig. 1(b) for fixed

*δ*+

_{A}*δ*= 3

_{B}*π*. One can find that there are four crossing points at the operation frequency. At the crossing points, (

*δ*,

_{A}*δ*) equals (0, 3

_{B}*π*), (

*π*, 2

*π*), (2

*π*,

*π*), and (3

*π*, 0), respectively. Among them, the first and last crossing points are meaningless because the length of the slab A or B is zero. Noting that an inverted band gap appears when the parameters went through either of the remaining crossing points [1,37], therefore the two remaining crossing points are assumed to be topological transition points (labeled by TP

_{I}and TP

_{II}), which can be regarded as the 1D counterpart of Dirac points. The critical parameters are ${d}_{A}^{\mathrm{I}}=\pi /2{\beta}_{A}=266.6\phantom{\rule{0.2em}{0ex}}\text{nm}$, ${d}_{B}^{\mathrm{I}}=2\pi /{\beta}_{B}=708.5\phantom{\rule{0.2em}{0ex}}\text{nm}$ at TP

_{I}and ${d}_{A}^{\text{II}}=\pi /{\beta}_{A}=533.3\phantom{\rule{0.2em}{0ex}}\text{nm}$, ${d}_{B}^{\text{II}}=\pi \text{/}{\beta}_{B}=354.3\phantom{\rule{0.2em}{0ex}}\text{nm}$ at TP

_{II}, respectively. As a result, sin

*δ*= sin

_{A}*δ*= 0 and cos

_{B}*δ*= cos

*δ*cos

_{A}*δ*= −1 are fulfilled at the crossing points. Accordingly, the band 2 and band 3 will cross at the boundary of the Brillouin zone. It is expected that the PC constructed from the parameters sandwiched by the two topological phase points is topologically distinct from the remanning parameter space.

_{B}Based on the aforementioned theory, the parameters of PC1 are chosen as *d _{B}*

_{1}= 2

*d*

_{A}_{1}, i.e.,

*d*

_{B}_{1}= 3

*π*/(

*β*+

_{A}*β*) = 638.6 nm and

_{B}*d*

_{A}_{1}=319.3 nm without loss of generality as marked in Fig. 1(b). The band gap of PC1 is 0.038 eV. Considering the band edge depends linearly on the thicknesses near the topological transition points, one can easily obtain the parameters

*d*

_{A}_{2}=210 nm and

*d*

_{B}_{2}=738.8 nm for PC2 to possess the same band gap width. Apparently, the band gap is shared by PC1 and PC2, which serves as a good platform to support edge states. When PC1 and PC2 are connected, for example, on the left side of a boundary is the semi-infinite PC1 while on the right side is the semi-infinite PC2. This photonic boundary is translation invariant along the

*x-*direction and its corresponding band structure is given in Fig. 1(c). An edge state expanding from

*k*= 0 to

_{x}*π*/

*a*appears in the common gap. Figure 1(d) plots the amplitude of magnetic fields of one representative edge state with normal incident wave (

*k*= 0). The magnetic field localizes around the boundary and decays exponentially away from the boundary.

_{x}The Zak phase [29], as an analogy of the Berry phase [38], has been used to classify the band topology for studying their edge states. For each isolated band *n*, the Zak phase can be defined as the formula:

*u*

_{n}_{,}

*(*

_{q}*z*+ Λ) =

*u*

_{n}_{,}

*(*

_{q}*z*) is the periodic Bloch magnetic field eigenfunction of a state on the

*n*th band with wave vector

*q*, i.e.,

*H*

_{y}_{;}

_{n}_{,}

*=*

_{q}*u*

_{n}_{,}

*(*

_{q}*z*) exp (i

*qz*). The function

*u*

_{n}_{,}

*(*

_{q}*z*) can be obtained analytically from the admittance based transfer-matrix method (see methods). Due to the inversion symmetry of the structure, the Zak phase is quantized at either 0 or

*π*when the origin coincides with either of the inversion centers. In the paper, the boundary of the unit cell is chosen as the origin for calculating Zak phases.

It is generally known that the reflectivity satisfies *r*_{left}*r*_{right} = 1 for the presence of an interface state between two adjacent semi-infinite structures, which indicates that the sign of reflection phases for the left and right parts must be opposite. If there is no band crossing, the Zak phases are related to the signs of the reflection phases *φ*^{(}^{n}^{)}, i.e.,

*n*th gap are connected, it is expected that there is an edge state localized near the interface in the

*n*th gap. In Figs. 2(a) and 2(b), we show the band structures of two PCs from the 1st to the 4th gaps with normal incident plasmon wave, i.e.,

*k*= 0. For convenience, the tuning process ${d}_{A1}\to {d}_{A}^{\mathrm{I}}\to {d}_{A2}$ is considered. The Zak phases of bands 0 and 1 remain unchanged while the Zak phases of bands 2 and 3 switch, which comes from a corresponding sign change for the reflection phases in the 3th gap after the perturbation. Thus the PCs undergo a topological phase transition in the parameter tuning process. To determine the location of the interface state, we recall the general condition

_{x}*r*

_{left}

*r*

_{right}= 1, which is governed by

*φ*

_{PC1}= −

*φ*

_{PC2}in our configuration. In the symmetric configuration, the reflectivity of

*N*layer ABA structure can be reduced as (see methods)

*η*

_{0}=

*η*

_{N}_{+1}is the admittance of input and output MIM waveguide,

*Nδ*is the total phase delay in the structure. And

*η*is the equivalent admittance of unit cell, is derived from

*M*

_{12}or

*M*

_{21}:

The calculated reflection phases are shown in Fig. 2(c). One can find that there is only one interface state in the third gap when the photon energy is 0.8 eV. It is worth mentioning that a symmetric unit cell can always be regarded as a homogenous layer with effective phase *δ* and the equivalent admittance *η*, which lays the foundation for the further calculations of composite structures.

The calculation of Zak phase by Eq. (4) needs to integrate the Berry connection over the whole Brillouin zone, which is very tedious and cumbersome. Fortunately, the Zak phase in an isolated band can be also determined by the symmetries of the eigenstates at high symmetry points in Brillouin zone, i.e., *K* = 0 and *K* = ±*π*/Λ. Considering that the amplitude of the Bloch magnetic fields for the band-edge states at the origin (*z* = 0) is either zero or maximum, one can divide the edge states into two categories: antisymmetric states with zero amplitude (marked by red circles) or symmetric states with maximum amplitude (marked by indigo circles). The Bloch magnetic fields of the eight band edge states sandwiched by band 2 and band 3 are calculated by the admittance based transfer-matrix method (see methods). Figure 2(d) shows the functions |*u _{n}*

_{,}

*(*

_{q}*z*= 0)| for the four edge states in PC1 at points K, L, M, and N labeled in Fig. 2(a) and in PC2 at points O, P, Q, and R labeled in Fig. 2(b). Due to the mode orthogonality, one can find that the symmetry of the two edge states sandwiching a gap is always different. According to the admirable result from Kohn [39] and Zak [29] for 1D systems, the Zak phase is zero if the two edge states have the same symmetry in the

*n*th band, otherwise, the Zak phase is

*π*. From the eigenfunctions, one can find the edge states K, L, N, O, Q, and R are antisymmetric states, yet the remaining states P and M are symmetric states. Thus the values of the Zak phases are 0(

*π*) for band 2(3) in PC1, while they switch in PC2, indicating the appearance of a topological phase transition.

In experiments, the TPTSs shown in Fig. 1(a) can be excited by impinging surface plasmon polaritons(SPPs) from either side of the MIM waveguides. When the incident SPPs are injected from the left side, the energy is delivered across the PC1 and confined at the interface between PC1 and PC2 simultaneously. Due to the different roles PC1 and PC2 play in exciting the topological edge states, the numbers of periods *N*_{1} and *N*_{2} are usually unequal. In the process, PC2 acts purely as a reflector, one can choose a large number, e.g., *N*_{2} = 8 to maximally reflect the incoming SPPs. However, PC1 works as an energy delivery as well as a reflector. To achieve the balance between energy confinement and Ohmic loss, *N*_{1} can not be too small or too large. The optimized number of periods can be derived from the minimum of reflection spectra. When TPTSs are excited, electromagnetic fields are trapped and significantly Ohm damped. Therefore, a minimum in reflection spectrum reveals the excitation of TPTSs. The reflection of the system PC1/PC2 can be expressed as functions of the phase delay(*δ*_{1}, *δ*_{2}), characteristics admittance(*η*_{1}, *η*_{2}) and layer numbers(*N*_{1}, *N*_{2}) find that the optimized layer number of PC1 is *N*_{1} = 4. In Figs. 3(a)-3(c), the reflection spectra of 4 unit cells of PC1, 8 unit cells of PC2, and their composites are plotted, where band gaps 2 and 3 are included. The spectra are calculated by TMM and confirmed by finite element method (marked by crosses). There is a sharp dip with a minimum of 0.6% and FWHM of 24.1 nm near the photon energy *E*=0.8 eV, which is in the 3rd gap of PC1 and PC2 simultaneously. As a contrast, no edge mode exists in the 2nd band despite there is similar reflectance of PC1 and PC2 in gaps 2 and 3, this can be interpreted directly via the topologies below gap 2 are the same between PC1 and PC2. In Fig. 3(b), one can see that the spectra is flat with reflectivity over 75% near the operation frequency, indicating that 8 unit cells of PC2 are good enough to act as a well-behaved reflector in the MIM waveguide. However, the spectrum is dispersive with reflectivity of only 47% near the operation frequency in Fig. 3(a), indicating the balance of the two functions of PC1. The TPTS has a ignorable blue-shift(~ 1.2 nm) compared with the designed operation wavelength attributing to the imperfect reflector of PC1, which makes it a well-behaved narrow-band absorber at designed wavelength.

In Figs. 3(d) and 3(e), we plot the field amplitude distribution at the resonant photon wavelength of 1550 nm. Both electric field |*E*| and magnetic field |*H*| are normalized to incident wave. The electromagnetic fields are trapped and enhanced near the interface between PC1 and PC2. The insets show the distributions of |*E*| and |*H*| along the center of MIM waveguide. The fields in PC1 and PC2 are approximately symmetric with the PC1/PC2 interface. Specifically, one can find that the symmetric axis of electric field is in PC1, while it is in PC2 for magnetic field. Due to the larger Ohm loss in PC2, the field in PC1 is larger than its symmetric positions in PC2. As a result, the maximal values of electric and magnetic fields are always in PC1. The maximal |*E*| and |*H*| in PC1 occur on the left and right of the first *ϵ _{A}*/

*ϵ*interface near the PC1/PC2 interface with an enhancement factor of 2.7 and 3, respectively. Moreover, the position for each local maximum of |

_{B}*E*| is also related to local minimum of |

*H*| and vice versa. The phenomenon originates from the simple relation

*r*= −

_{E}*r*, so there is a phase different of

_{H}*π*between reflected electric and magnetic fields. As a result, constructive interference for one quantity will indicate destructive interference for the other quantity.

Since the Zak-Berry phase is a ’global’ or ’topological’ property of the entire band, the edge mode protected by Zak phase is expected robust against structural perturbations. Next, we introduce the geometric perturbations by inserting a MIM waveguide between PC1 and PC2. For simplicity, the core of the defective MIM waveguide is chose as air. In our configuration, the defect is equivalent to prolong the thickness of the part A connecting PC1 and PC2. The thickness of slab A sandwiched by slab B without perturbation at the interface is *d _{A}*

_{1}+

*d*

_{A}_{2}. After the defect is inserted, the total thickness of slab A increases to

*d*=

*d*

_{A}_{1}+

*d*

_{A}_{2}+ Δ

*d*. The reflectance spectra for Δ

*d*= 0 nm, 100 nm, 300 nm, and 450 nm are shown in Fig. 4(a). The TPTS red-shifts as the increase of Δ

*d*. Nevertheless, a new resonance dip appears when Δ

*d*≳ 300 nm, which certifies that the interface state always exists in the considered band gap. Incidentally, we find the field spot of protected states are always in the defect layer, which make it a promising all-optical bistable switch [33]. The dependence of the resonance photon energy of edge mode on the thickness of defect is shown in Fig. 4(b). From calculating the phase delay in defect layer Δ

*δ*=

*β*Δ

_{A}*d*, one can find that the second mode appears if

*π*/2 ≤ Δ

*δ*≤ 3

*π*/2. Actually, this is a periodic function of Δ

*δ*with a period of

*π*. Specially, when Δ

*δ*=

*nπ*,

*n*∈

*N*

^{+}, the resonance frequency will come back to the starting position.

## 3. Conclusion

A TPTS configuration is proposed by two adjacent semi-infinite MIM PCs. By using the proposed admittance based TMM, we find that the unit cell of symmetric PC can be regarded as a homogeneous layer with effective phase delay and equivalent admittance. On this basis, we have studied the topological properties of PCs and revealed the relationship among their topologies, reflective phase and phase delay in each layer. As such, their topology and the associated edge (interface) states are extremely robust against the structural perturbations. Such topologically protected localized states can be used as a generalization of plasmonic Tamm states, which makes it a promising optical device for integrated photonic circuits and able to provide significant applications, such as optical nonlinear enhancement.

## 4. Methods

#### 4.1. Admittance based transfer matrix method

In the layered periodic MIM structure, the electric field of transversal magnetic (TM) wave can be expressed by incident and reflected waves as:

*l*th layer, respectively. According to the continuous boundary condition of the electromagnetic field at the interface

*j*, the field can be written as:

If the A/B and B/A interface in a unit cell shown in Fig. 1(b) are marked by green number ’1’ and ’2’ from left to right, the electric field in the beginning and end of the *N*th unit cell satisfy:

*l*) and right(

*r*) limit of the related interfaces. After some mathematics, one could arrive at

*T*=

_{A}E_{N}*MT*

_{A}E_{N}_{+1}, where

One can find the expression of transmission matrix of unit cell is the same as the transmission matrix of a homogeneous layer. Thus the unit cell can be regarded as ’homogenous’ with equivalent phase delay and admittance. As a result, the phase delay of the unit cell represents the dispersion relation of the PCs, i.e., cos *δ* = cos (*K* Λ).

When the PCs are composed of *N* periods, the total transmission can be written as:

In the same manner, one can conclude that:

Expanding the expression, we have

After some mathematics, one could arrive at

An important case is that when *η*_{0} = *η _{N}*

_{+1}, because of ${\widehat{M}}_{11}={\widehat{M}}_{22}=\mathrm{cos}(N\delta )$ and $\eta {\widehat{M}}_{12}={\widehat{M}}_{21}/\eta =-\mathrm{i}\mathrm{sin}(N\delta )$, the reflectivity can be reduced as

When the structure is composite of two PCs, one can also regard the two structures with two separated homogeneous layers: $M={M}_{N1}{M}_{N2}=\left[\begin{array}{cc}{\tilde{M}}_{11}& {\tilde{M}}_{12}\\ {\tilde{M}}_{21}& {\tilde{M}}_{22}\end{array}\right]$, where

*N*

_{1,2},

*δ*

_{1,2}and

*η*

_{1,2}are the layer numbers, phase delay and equivalent characteristic admittance of PC1 and PC2, respectively. Similarly, when an extra MIM waveguide is inserted into the PCs, the transition matrix must be substituted by

*M*=

*M*

_{N}_{1}

*M*

_{d}M_{N}_{2}, where ${M}_{d}=\left[\begin{array}{cc}\mathrm{cos}\mathrm{\Delta}\delta & -\text{i/}{\eta}_{d}\mathrm{sin}\mathrm{\Delta}\delta \\ -\mathrm{i}{\eta}_{d}\mathrm{sin}\mathrm{\Delta}\delta & \mathrm{cos}\mathrm{\Delta}\delta \end{array}\right]$ is the transmission matrix of the defect layer. Finally, the reflectance of the composite structure can be expressed as:

#### 4.2. Bloch function

We start from the expression *T _{A}E_{N}* =

*MT*, thus we have

_{A}E_{N+1}*M*=

^{E}E_{N}*E*

_{N}_{+1}with the definition ${M}^{E}\equiv {T}_{A}^{-1}{M}^{-1}{T}_{A}$. If the periodic boundary condition is applied to a unit cell, the Bloch electric field satisfies

*M*= exp(i

^{E}E_{N}*K*Λ)

*E*. This is a typical eigenvalue problem, and the eigenfunction at origin can be chosen as ${E}_{0}\left(\begin{array}{c}1\\ [\mathrm{exp}(\mathrm{i}K\Lambda )-{M}_{11}^{E}]/{M}_{12}^{E}\end{array}\right)\equiv \left(\begin{array}{c}1\\ \Gamma \end{array}\right)$,where

_{N}It is worth mentioning that ${M}_{12}^{E}$ and $\mathrm{exp}(\mathrm{i}K\Lambda )-{M}_{11}^{E}$ equal to zero simultaneously at the frequency point $\tilde{\omega}$ at which $\mathrm{sin}({\delta}_{B}(\tilde{\omega}))=0$. If one isolated band (excluding the 0th band) contains these frequency points , then the Zak phase of the band must be *π* in the configuration.

In the first period, we consider the electromagnetic field along the central axis(y = 0), the field in slab A is straightforward: *E* (*z* ) = exp(i*β _{A}z* ) + Γ exp(−i

*β*) and $\frac{{d}_{x}}{w}{H}_{x}(z)={\eta}_{A}\mathrm{exp}(\mathrm{i}{\beta}_{A}z)-{\eta}_{A}\Gamma \mathrm{exp}(-\mathrm{i}{\beta}_{A}z)$. Specially, the Bloch magnetic fields in origin satisfy

_{A}z*u*(

*z*= 0) ∝ 1 − Γ. Due to the relation ${T}_{A}{E}_{0}={M}_{A}{T}_{B}{E}_{1}^{r}$, the field at the right limit of interface labeled ’1’ satisfies ${E}_{1}^{r}=\left(\begin{array}{c}{E}_{1r}^{+}\\ {E}_{1r}^{-}\end{array}\right)={T}_{B}^{-1}{M}_{A}^{-1}{T}_{A}{E}_{0}$. From which one can get the electromagnetic field in slab B:

## Appendix Topological plasmonic Tamm states supported by HfO_{2}/SiO_{2}/HfO_{2} waveguides

In the main paper, we have explored the plasmonic crystals (PCs) with air/SiO_{2}/air cores, indicating *ϵ _{A} < ϵ_{B}*, this hypothesis is very compatible with the current fabrication process. But in principle, the case

*ϵ*is also available such as adopting the PCs with HfO

_{A}> ϵ_{B}_{2}/SiO

_{2}/HfO

_{2}cores. In the same manner, to verify the existence of such an topological interface states, we consider two adjoining one dimensional PCs with parameters across the topological transition point, and a silver-air-silver waveguide are adopted as the input and output waveguides just as the case in the main paper. For the telecom optical wavelength at 1550 nm (≈0.8 eV), we have

*β*= 11.884

_{A}*µm*

^{−1},

*β*= 8.868

_{B}*µm*

^{−1}and

*β*

_{0}=

*β*

_{N}_{+1}= 5.891

*µm*

^{−1}, respectively. Specifically, TP

_{II}, i.e., ${\delta}_{A}(\tilde{\omega})=2\pi $ and ${\delta}_{B}(\tilde{\omega})=\pi $ are adopted due to

*β*; thus ${d}_{A}^{\text{II}}=2\pi /{\beta}_{A}/2=264.4\phantom{\rule{0.2em}{0ex}}\text{nm}$ and ${d}_{B}^{\text{II}}=\pi /{\beta}_{B}=354.3\phantom{\rule{0.2em}{0ex}}\text{nm}$, respectively. The parameters of PC1 are chosen as

_{A}> β_{B}*d*

_{B}_{1}= 2

*d*

_{A}_{1}, i.e.,

*d*

_{B}_{1}= 3

*π*/(

*β*+

_{A}*β*) = 454.2 nm and

_{B}*d*

_{A}_{1}=227.1 nm without loss of generality. The band structure and Zak phase of plasmonic crystals consist of HfO

_{2}/SiO

_{2}/HfO

_{2}cores are shown in Fig. 5. The Considering the band edge is linear dependent on thicknesses near the topological transition points, one can easily obtain the parameters ${d}_{A2}=2{d}_{A}^{\text{II}}-{d}_{A1}=151.4\phantom{\rule{0.2em}{0ex}}\text{nm}$ and

*d*

_{B}_{2}=302.8 nm for PC2 to share the same band gap with PC1. For convenience, the tuning process ${d}_{A1}\to {d}_{A}^{\text{II}}\to {d}_{A2}$ is considered. The Zak phases of bands 0 and 1 remain unchanged while the Zak phases of bands 2 and 3 switch, which comes from a corresponding sign change for the reflection phases in the 3th gap after the perturbation. Thus the PCs undergo a topological phase transition in the parameter tuning process. Compared with the PCs with air/SiO

_{2}/air cores, the Zak phase of the 0th and 1st band is

*π*(0) rather than 0(

*π*). The reason for this is that the Zak phase of the 0th band is charactered by $\mathrm{exp}(\mathrm{i}{\theta}_{0}^{\text{Zak}})=\mathrm{sgn}[1-{\eta}_{A}^{2}/{\eta}_{B}^{2}]$, due to

*η*, one can obtain ${\theta}_{0}^{\text{Zak}}=\pi $ immediately. Due to the sign of reflection phase in gap 2 is the same as discussed in main paper, thus the Zak phase of the 0th and 1st band must be opposite with the case

_{A}> η_{B}*η*. Moreover, the Zak phase of the band 2 is different attributing to the topological phase transition, thus a topologically protected interface state exists at the interfaces between PC1 and PC2. For comparison, the reflection spectra of the configurations are shown in Fig. 6. One can find a similar TPTS at considered wavelength in the configurations.

_{A}< η_{B}## Funding

National key R & D Program of China (2017YFA0303800, 2017YFA0305100); Program for Changjiang Scholars and Innovative Research Team in University (IRT13_R29); National Natural Science Foundation of China (NSFC) (11604283, 91750204, 11774185, 11374006, 11504184, 61775106); 111 Project (B07013); Tianjin Natural Science Foundation (18JCQNJC02100); Key Research Projects of Henan Provincial Department of Education (16A140048); Fundamental Research Funds for the Central Universities; Nanhu Scholars Program for Young Scholars of XYNU.

## Disclosures

The authors declare that there are no conflicts of interest related to this article.

## References

**1. **B. A. Bernevig, T. L. Hughes, and S.-C. Zhang, “Quantum spin hall effect and topological phase transition in HgTe quantum wells,” Science **314**, 1757–1761 (2006). [CrossRef] [PubMed]

**2. **Z. Wang, Y. Chong, J. D. Joannopoulos, and M. Soljačić, “Observation of unidirectional backscattering-immune topological electromagnetic states,” Nature **461**, 772–775 (2009). [CrossRef] [PubMed]

**3. **M. Hafezi, E. A. Demler, M. D. Lukin, and J. M. Taylor, “Robust optical delay lines with topological protection,” Nat. Phys. **7**, 907–912 (2011). [CrossRef]

**4. **M. Hafezi, S. Mittal, J. Fan, A. Migdall, and J. M. Taylor, “Imaging topological edge states in silicon photonics,” Nat. Photon. **7**, 1001–1005 (2013). [CrossRef]

**5. **M. C. Rechtsman, J. M. Zeuner, Y. Plotnik, Y. Lumer, D. Podolsky, F. Dreisow, S. Nolte, M. Segev, and A. Szameit, “Photonic floquet topological insulators,” Nature **496**, 196–200 (2013). [CrossRef] [PubMed]

**6. **L. Fu, “Topological crystalline insulators,” Phys. Rev. Lett. **106**, 106802 (2011). [CrossRef] [PubMed]

**7. **L.-H. Wu and X. Hu, “Scheme for achieving a topological photonic crystal by using dielectric material,” Phys. Rev. Lett. **114**, 223901 (2015). [CrossRef] [PubMed]

**8. **G. Harari, M. A. Bandres, Y. Lumer, M. C. Rechtsman, Y. D. Chong, M. Khajavikhan, D. N. Christodoulides, and M. Segev, “Topological insulator laser: Theory,” Science **359**, aar4003 (2018). [CrossRef]

**9. **M. A. Bandres, S. Wittek, G. Harari, M. Parto, J. Ren, M. Segev, D. N. Christodoulides, and M. Khajavikhan, “Topological insulator laser: Experiments,” Science **359**, aar4005 (2018). [CrossRef]

**10. **B. Bahari, A. Ndao, F. Vallini, A. El Amili, Y. Fainman, and B. Kanté, “Nonreciprocal lasing in topological cavities of arbitrary geometries,” Science **358**, 636–640 (2017). [CrossRef] [PubMed]

**11. **H. Zhao, P. Miao, M. H. Teimourpour, S. Malzard, R. El-Ganainy, H. Schomerus, and L. Feng, “Topological hybrid silicon microlasers,” Nat. Commun. **9**, 981 (2018). [CrossRef] [PubMed]

**12. **A. Poddubny, A. Miroshnichenko, A. Slobozhanyuk, and Y. Kivshar, “Topological majorana states in zigzag chains of plasmonic nanoparticles,” ACS Photonics **1**, 101–105 (2014). [CrossRef]

**13. **C. W. Ling, M. Xiao, C. T. Chan, S. F. Yu, and K. H. Fung, “Topological edge plasmon modes between diatomic chains of plasmonic nanoparticles,” Opt. Express **23**, 2021–2031 (2015). [CrossRef] [PubMed]

**14. **S. R. Pocock, X. Xiao, P. A. Huidobro, and V. Giannini, “Topological plasmonic chain with retardation and radiative effects,” ACS Photonics **5**, 2271–2279 (2018). [CrossRef]

**15. **D. E. Gómez, Y. Hwang, J. Lin, T. J. Davis, and A. Roberts, “Plasmonic edge states: An electrostatic eigenmode description,” ACS Photonics **4**, 1607–1614 (2017). [CrossRef]

**16. **Q. Cheng, Y. Pan, Q. Wang, T. Li, and S. Zhu, “Topologically protected interface mode in plasmonic waveguide arrays,” Laser & Photonics Rev. **9**, 392–398 (2015). [CrossRef]

**17. **L. Ge, L. Wang, M. Xiao, W. Wen, C. T. Chan, and D. Han, “Topological edge modes in multilayer graphene systems,” Opt. Express **23**, 21585–21595 (2015). [CrossRef] [PubMed]

**18. **H. Deng, X. Chen, N. C. Panoiu, and F. Ye, “Topological surface plasmons in superlattices with changing sign of the average permittivity,” Opt. Lett. **41**, 4281–4284 (2016). [CrossRef] [PubMed]

**19. **F. Gao, Z. Gao, X. Shi, Z. Yang, X. Lin, H. Xu, J. D. Joannopoulos, M. Soljačić, H. Chen, L. Lu, Y. Chong, and B. Zhang, “Probing topological protection using a designer surface plasmon structure,” Nat. Commun. **7**, 11619 (2016). [CrossRef] [PubMed]

**20. **X. Wu, Y. Meng, J. Tian, Y. Huang, H. Xiang, D. Han, and W. Wen, “Direct observation of valley-polarized topological edge states in designer surface plasmon crystals,” Nat. Commun. **8**, 1304 (2017). [CrossRef] [PubMed]

**21. **D. Jin, T. Christensen, M. Soljačić, N. X. Fang, L. Lu, and X. Zhang, “Infrared topological plasmons in graphene,” Phys. Rev. Lett. **118**, 245301 (2017). [CrossRef] [PubMed]

**22. **D. Pan, R. Yu, H. Xu, and F. J. García de Abajo, “Topologically protected dirac plasmons in a graphene superlattice,” Nat. Commun. **8**, 1243 (2017). [CrossRef] [PubMed]

**23. **M. Kaliteevski, I. Iorsh, S. Brand, R. A. Abram, J. M. Chamberlain, A. V. Kavokin, and I. A. Shelykh, “Tamm plasmon-polaritons: Possible electromagnetic states at the interface of a metal and a dielectric bragg mirror,” Phys. Rev. B **76**, 165415 (2007). [CrossRef]

**24. **I. Tamm, “On the possible bound states of electrons on a crystal surface,” Phys. Z. Sov. Union **1**, 733 (1932).

**25. **T. Goto, A. V. Dorofeenko, A. M. Merzlikin, A. V. Baryshev, A. P. Vinogradov, M. Inoue, A. A. Lisyansky, and A. B. Granovsky, “Optical tamm states in one-dimensional magnetophotonic structures,” Phys. Rev. Lett. **101**, 113902 (2008). [CrossRef] [PubMed]

**26. **S. Longhi, “Zak phase of photons in optical waveguide lattices,” Opt. Lett. **38**, 3716–3719 (2016). [CrossRef]

**27. **Y. Xiang, P. Wang, W. Cai, C.-F. Ying, X. Zhang, and J. Xu, “Plasmonic tamm states: dual enhancement of light inside the plasmonic waveguide,” J. Opt. Soc. Am. B **31**, 2769–2772 (2014). [CrossRef]

**28. **L. Niu, Y. Xiang, W. Luo, W. Cai, J. Qi, X. Zhang, and J. Xu, “Nanofocusing of the free-space optical energy with plasmonic tamm states,” Sci. Rep. **6**, 39125 (2016). [CrossRef] [PubMed]

**29. **J. Zak, “Berry’s phase for energy bands in solids,” Phys. Rev. Lett. **62**, 2747–2750 (1989). [CrossRef] [PubMed]

**30. **M. Xiao, Z. Q. Zhang, and C. T. Chan, “Surface impedance and bulk band geometric phases in one-dimensional systems,” Phys. Rev. X **4**, 021017 (2014).

**31. **K. H. Choi, C. W. Ling, K. F. Lee, Y. H. Tsang, and K. H. Fung, “Simultaneous multi-frequency topological edge modes between one-dimensional photonic crystals,” Opt. Lett. **41**, 1644–1647 (2016). [CrossRef] [PubMed]

**32. **W. Gao, M. Xiao, B. Chen, E. Y. B. Pun, C. T. Chan, and W. Y. Tam, “Controlling interface states in 1d photonic crystals by tuning bulk geometric phases,” Opt. Lett. **42**, 1500–1503 (2017). [CrossRef] [PubMed]

**33. **Y. Xiang, X. Zhang, W. Cai, L. Wang, C. Ying, and J. Xu, “Optical bistability based on Bragg grating resonators in metal-insulator-metal plasmonic waveguides,” AIP Adv. **3**, 012106 (2013). [CrossRef]

**34. **B. I. Afinogenov, A. A. Popkova, V. O. Bessonov, B. Lukyanchuk, and A. A. Fedyanin, “Phase matching with Tamm plasmons for enhanced second- and third-harmonic generation,” Phys. Rev. B **97**, 115438 (2018). [CrossRef]

**35. **S. D. Choudhury, R. Badugu, and J. R. Lakowicz, “Directing fluorescence with plasmonic and photonic structures,” Accounts Chem. Res. **48**, 2171–2180 (2015). [CrossRef]

**36. **S. A. Maier, *Plasmonics: Fundamentals and Applications* (SpringerUS, 2007).

**37. **L. Xu, H. Wang, Y. Xu, H. Chen, and J. Jiang, “Accidental degeneracy in photonic bands and topological phase transitions in two-dimensional core-shell dielectric photonic crystals,” Opt. Express **24**, 18059–18071 (2016). [CrossRef] [PubMed]

**38. **M. V. Berry, “Quantal phase factors accompanying adiabatic changes,” Proc. Royal Soc. Lond. **392**, 45–57 (1984). [CrossRef]

**39. **W. Kohn, “Analytic properties of Bloch waves and Wannier functions,” Phys. Rev. **115**, 809–821 (1959). [CrossRef]