Numerical Verification of a Polarization-Insensitive Electrically Tunable Far Infrared Band-Stop Meta-Surface Filter

Abstract

Tunable filters have many potential applications in diverse fields, including high-capacity communications, dynamic beam shaping and spectral imaging. Although providing a high-performance solution for actively tunable devices, metasurface combined with tunable materials faces the great challenges of limited tuning range and modulation depth. Here, we propose a far-infrared tunable band-stop filter based on Fabry-Perot (FP) resonators and graphene surface plasmons. By switching the wavelength of the critical coupling condition of the filter via the gate voltage applied on graphene, achieving the dynamically tunable band-stop filtering at the central wavelengths from 12.4 μm to 14.1 μm with a modulation depth of more than 99%. Due to the symmetry of the proposed meta-atoms, the filter is insensitive to the polarization direction of the incident light. And it can realize more than 85% filtering efficiency within 60° angle of incidence around the vertical direction. By adjusting the geometry of the meta-atoms structure, it is feasible to move the operational range from the near-infrared to terahertz bands.

1. Introduction

Optical filters are integral components as the function of light modulation to be applied in various areas such as high-capacity communications, dynamic beam shaping and spectral imaging [1,2]. With the development of meta-optics, the performance of meta-surface filters has been dramatically developed to replace possibly the traditional devices including film, grating and the tunable liquid crystal [3]. However, once the meta-atom structure is prepared, its operating wavelength is determined to weaken the application of the filter. The optical properties of tunable materials can be changed through external excitations such as light, electricity and heat. Tunable materials and functional metasurfaces is expected to dynamically control the wavefront of light at sub-wavelength scales. In recent years, a large number of tunable metamaterial devices based on tunable materials such as graphene, GST, VO₂, and liquid crystals have emerged [3,4,5,6,7,8,9,10]. For tunable filters, the optical responses of metasurfaces dynamically can be controlled to achieve a tunable filtering in a certain wavelength range. These tunable filters overcome the limitations faced by conventional filters and improve the ability to analyze and process spectral information. Moreover, the tunable meta-surface devices are significant factors in enhancing the modulation speed, extending the modulation range, and reducing the weight of the system.

Graphene as a two-dimensional material, exhibits superior electronic and optical properties and stability [11,12], as well as high carrier mobility and a voltage-tunable optical response [13,14,15]. In graphene, incident light can stimulate electronic transitions, the inter-band and intra-band transitions of graphene are determined by the Fermi energy level E_F. It is possible to adjust the Fermi energy level of graphene through the use of chemical doping or the gating voltage to control the carrier concentration on the graphene. The unique electronic properties of graphene make it a promising platform to build active plasmonic devices for a wide wavelength range from near-infrared to terahertz. In 2011, Ju et al. demonstrated for the first time that the optical properties of graphene could be modulated by applying a voltage to the material [16]. In recent years, graphene-based active tuning devices have been researched widely, including amplitude modulation [17,18,19], tunable filtering [20,21,22,23,24,25], and phase modulation [26,27]. These devices represent a significant advancement over their predecessors.

Many researchers have utilized phase change materials (VO₂, GST, etc.) and graphene to design the tunable wavelength filters. Matthew N. Julian’s group proposed a GeSbTe-based tunable filter with all-optical modulation with the transmittance of 70% [6], which is difficult to achieve a precise control of the filter due to the limitation of the modulation method. VO₂-based tunable filters also have the problems of low modulation depth and complicated modulation method [28]. Jiang et al. designed a graphene-based terahertz filter that allows active tuning of the transmission wavelength from 19.6 μm to 22 μm. However, the filter has some drawbacks of limited modulation depth and incidence angle constraint [20]. Wang et al. proposed a graphene-based large-spectrum-range and narrow-bandwidth infrared filter, which can be tunable modulated from 10 μm to 15 μm. However, the limitations of restriction on the direction of incident-light polarization and the complexity of fabrication impose application restrictions on the filter [22]. As a result, although they have been improved relative to conventional filters, these tunable filters still have some shortcomings, such as a restricted modulated spectrum range, polarization sensitivity, the limited angle of incidence and complex fabrication and processing procedures.

In this study, we present a tunable band-stop filter by incorporating a widely tunable metasurface on a graphene to generate an ultrathin modulator with a modulation depth of more than 99% from 12.4 μm to 14.1 μm (the central wavelength range). The isotropic nature of the meta-atoms structure enables the device to demonstrate polarization-insensitive performance, as well as maintaining a filtering efficiency of more than 85% across a wide incidence angle range of 60°. On the consequence, the fabrication errors are analyzed to demonstrate the feasibility and stability of preparation. Furthermore, by adjusting the dimensions of the unit, it is possible to move the filter operating range to other domain such as the near-infrared and terahertz, therefore provide efficient solutions for reconfigurable flat optics and optoelectronics.

2. Design of Tunable Filter Structure

The proposed tunable filter consists of an aluminum substrate, a silicon dioxide layer, and an electrically tunable metasurface comprising plasmonic structures on graphene, as shown in Figure 1. The period of the unit structure is 2 μm, the thickness of the aluminum substrate is 500 nm, and the thickness of the dielectric layer SiO₂ is 400 nm, as illuminated in Figure 1c,d. The metal substrate and graphene, functioning as two mirrors form a Fabry-Perot (FP) cavity. In this resonator, the critical coupling conditions can be realized at a dielectric layer thickness d much smaller than the wavelength [18], which makes it possible to absorb electromagnetic waves of a specific wavelength. In our proposed model, the conductivity of the metasurface when the asymmetric Fabry Perot cavity reaches the critical coupling condition is a function of the relative dielectric thickness. The relative dielectric thickness is defined as d₁, d₁ = nd/λ₀, where n is the refractive index of the dielectric layer, d is the thickness of the dielectric layer, and λ₀ is the wavelength of the incident wave. In our tunable filter model, n = 1.45, d = 400 nm. To absorb electromagnetic waves in the wavelength range of λ₀ = 12–14 μm, the relative dielectric thickness d₁ is approximately 0.041–0.048 (far less than 1). We can adjust the structural dimensions on the metasurface to obtain the surface conductivity required for critical coupling conditions.

The monolayer graphene is a bandgap-free semiconductor in which the conduction and valence bands meet at the Dirac point [29,30]. There is a linear energy-momentum dispersion relation near the Dirac point, which can be modeled as a massless Dirac fermion [31]. Graphene can produce surface plasmon polaritons (SPP) in the infrared spectrum [32]. Additionally, the surface conductivity of graphene is susceptible to fluctuations contingent upon the Fermi energy level, thereby enabling the modulation of its surface conductivity through chemical doping or the application of an external electric field [33,34]. The surface conductivity of graphene can be represented by the Kubo formula [35]:

$$\begin{array}{l}{\sigma }_s(\omega ,{\mu }_c,\tau ,T)=-{\displaystyle \frac{i{e}^2(\omega +i{\tau }^{-1})}{\pi {ħ}^2}}\\ \left[{\displaystyle {\int }_{-\infty }^{+\infty }\frac{\left|\epsilon \right|}{{(\omega +i{\tau }^{-1})}^2}}{\displaystyle \frac{\partial f_d(\epsilon )}{\partial \epsilon }}d\epsilon -{\displaystyle {\int }_0^{+\infty }\frac{\partial f_d(-\epsilon )-\partial f_d(\epsilon )}{{(\omega +i{\tau }^{-1})}^2-4{(\epsilon /ħ)}^2}}d\epsilon \right]\end{array}$$

(1)

where f_d = 1/(1 + exp[(ε − μ_c)/(k_BT)]) is the Fermi-Dirac distribution, k_B is the Boltzmann constant, ε is energy, μ_c is the chemical potential of graphene (Fermi energy level, E_F), e is the electron charge, T is the temperature, ħ is the reduced Planck constant, τ = μE_F/(ev_f²) is the carrier relaxation time as a function of the carrier mobility, μ = 200,000 cm²/(V·s) is the carrier mobility [36,37]. The first term in (1) corresponding to the intra-band transition process can be expressed as:

$${\sigma }_{\mathrm{intra}}=i{\displaystyle \frac{e^2{k}_{B}T}{\pi {\hslash }^2(\omega +i{\tau }^{-1})}}\left[{\displaystyle \frac{{\mu }_c}{k_{B}T}}+2\ln \left(\exp \left(-{\displaystyle \frac{{\mu }_c}{k_{B}T}}\right)+1\right)\right]$$

(2)

The second term corresponds to the inter-band transition:

$${\sigma }_{\mathrm{inter}}=i{\displaystyle \frac{e^2}{4\pi \hslash }}\ln \left[{\displaystyle \frac{2\left|{\mu }_c\right|-\hslash (\omega +i{\tau }^{-1})}{2\left|{\mu }_c\right|+\hslash (\omega +i{\tau }^{-1})}}\right]$$

(3)

The two transition processes together determine the surface conductivity of graphene. In our proposed model, graphene metasurfaces and aluminum form parallel plate capacitors, and the carrier concentration in graphene is n₀ = ε₀ε_dV_g/(d·e) [21], where d is the thickness of the dielectric layer, and V_g is the voltage between the two plates of the capacitor, ε_d is the relative permittivity of the dielectric layer. The equations show that the carrier concentration on graphene is related to the applied voltage and the thickness of the dielectric layer, as well as the dielectric constant of the dielectric layer. The variation of the carrier concentration is also related to the value of the Fermi energy level. The relationship between the chemical potential and carrier concentration (n₀) of graphene is E_F = ħv_f(πn₀)^1/2, where v_f = 1 × 10⁶ m/s is the Fermi speed [38]. Consequently, the Fermi energy levels of graphene can be modified by applying a voltage across the ends of a parallel capacitor to modulate the optical properties of the device.

3. Results and Discussion

From the above, the resonator can absorb electromagnetic waves of specific wavelengths when the critical coupling condition of FP resonator is reached. Furthermore, the bias voltage applied to graphene can regulate the Fermi energy level of graphene and, consequently, the effective refractive index of the device. This indicates that the gating voltage can be employed as a means of controlling the center wavelength of the band-stop filter. To validate this process, the proposed filter model was verified using finite element method (FEM). This was done by irradiating vertically incident TE polarized light on a periodically arranged meta-atom structure. Set the numerical grid for aluminum metal to be refined, and set the numerical grid for the dielectric layer, gold, and air to be more refined. In the model, the input port is 300 nm away from the upper surface of graphene, and the medium between graphene and the input port is air. The output port is located on the lower surface of aluminum. The refractive index of SiO₂ in the far-infrared band is shown in Figure 2a, where it is taken as 1.45. In our simulation, we simulated gold and aluminum based on the measured refractive index, and the specific material parameters are shown in

Table 1. The refractive index and thickness of each material in the model.

Material	Au	Al	SiO2	Graphene
Refractivity	Measured	Measured	1.45	Kubo model
Thickness	60 nm	500 nm	400 nm	0.7 nm

.

The resulting reflectance spectrum of the device among the voltage from 5 V to 30 V are shown in Figure 2. As the critical coupling condition of the resonator is reached, the electromagnetic wave of a certain wavelength is absorbed, forming the effect of band-stop filtering. In order to realize the effect of active modulation, we use the simpler method to dynamically shape the reflection light by applying a gating voltage to modulate the Fermi energy level. As the voltage is increased from 5 V to 30 V, a change in the conductivity on the surface of the device results in a blue-shift of the center wavelength of the band-stop filter, which corresponds to a change (0.048–0.12 eV) in the Fermi energy level of graphene (see Figure 2b). By increasing the voltage, the center wavelength of the band-stop moves from 14.1 μm to 12.4 μm with a tuning range of 1.7 μm. At the center wavelength of the band-stop, the resonance absorption is the strongest to restrain the reflectivity. The half-height width of the filter is about 1.5 μm. In contrast, the reflectivity of the other wavelengths is almost unaffected.

Figure 2c,d show the reflectance and absorptance of the tunable filter at different chemical potential of graphene, respectively. The lowest point of the reflectance shifts from 14.1 μm at 0.048 eV to 12.3 μm at 0.12 eV with the voltage increase in which the absorption rate is above 99%. The electromagnetic wave with the wavelength of 13.3 μm is completely absorbed in the case of the chemical potential is 0.069 eV. This phenomenon is caused by the fact that the stronger resonant absorption occurs at the chemical potential is 0.069 eV, and the electric field strength on the surface of the device is the highest at a voltage of 10 V, as shown in Figure 3, when the incident light wavelength is 13.3 μm and the chemical potential of graphene increasing from 0.048 eV to 0.12 eV. From the reflection and absorption spectra of the filter, it can be seen that there is an obvious peak at 12.7 μm when a voltage of 5 V is applied. To explain this phenomenon, we calculated the energy of a photon with a wavelength of 12.7 μm as ħω = 1.5652 × 10⁻²⁰ J. The Fermi energy level (E_F) when a 5 V voltage is applied can also be seen from the Figure 2 as 0.048 eV = 7.69 × 10⁻²¹ J. Comparing the Fermi level and photon energy, it can be found that ħω ≈ 2E_F. When a voltage of 5 V is applied, which means the Fermi level is 0.048 eV, the incident photons at 12.7 μm are the boundary value between interband transitions and interband transitions (ħω = 2E_F). The incident photons with wavelengths less than 12.7 μm are mainly interband transitions (ħω > 2E_F), while those with wavelengths greater than 12.7 μm are mainly intraband transitions (ħω < 2E_F). Therefore, an obvious peak is generated at 12.7 μm. After calculating the Fermi level at a voltage of 10 V (0.069 eV = 1.1 × 10⁻²⁰ J) and comparing it with the energy of photons at a wavelength of 9 μm (ħω = 2.19 × 10⁻²⁰ J), it can be found that ħω ≈ 2E_F still satisfied. Therefore, a peak also appears at the 9 μm wavelength when a voltage of 10 V is applied, as shown in Figure 2c,d. Since the magnitude of the electric field is an indication of the resonance absorption strength, and reflects the absorption rate of electromagnetic waves by the device. Consequently, the electric field strengths on the graphene and the metasurfaces at voltages of 5 V, 15 V, and 20 V are observed to be lower than those at a voltage of 10 V. On the other hand, we simulate the electric field distribution for different wavelengths at the voltage of 10 V, as shown in Figure 4. Comparing the electric fields in the XY and YZ planes at 11.3 μm, 13.3 μm, and 15.3 μm, it can be observed that the device exhibits the greatest resonance at the wavelength of strongest absorption, which corresponds to the lowest reflectivity of 13.3 μm. The resonances are mainly concentrated in the spacing part of the unit structure. At wavelengths of 11.3 μm and 15.3 μm, the electric field strength is markedly diminished, indicating a relatively weak resonance. Therefore, the reflectivity exceeds 90%. This is because the distance between the unit structures is 100 nm, which is much smaller than the operational wavelength from 12.4 μm to 14.1 μm, and highly enhanced light-graphene interactions take place between the meta-atom spacers. The conductivity of the metasurface on graphene is much larger than the graphene layer alone, making the critical coupling conditions are more easily reached [18,39]. And the surface conductivity of graphene is highly sensitive to the concentration of graphene carriers, thereby enabling the control of the optical response of the metasurface through the application of an appropriate voltage.

Given that the resonance primarily occurs within the gap between the two units, it is valuable to investigate the impact of the cell spacing for the overall performance of the filter. The reflectance at cell spacing W_p from 70 nm to 130 nm is simulated without the change of all other parameters at a voltage of 10 V, as depicted in Figure 5. It can be observed that the cell spacing has a significant impact on the wavelength of resonance absorption, which tends to shorten gradually with increasing spacing. Additionally, it can be seen that the center wavelengths of reflectance are basically linear with the center wavelength, where the center wavelength exhibits a blue shift of approximately 0.2–0.3 μm for every 10 nm increase of cell spacing, as illustrated in Figure 5b. While the cell spacing has no impact on the filtering efficiency, which is consistently above 99%. The operating range of the filter is notably sensitive to the spacing width. This is due to the fact that the resonance primarily occurs within the gap between the two cells. This also reflects that metasurfaces on graphene are more likely to reach critical coupling conditions compared to pure graphene, and due to the highly enhanced interaction between light and graphene in the unit gaps of metasurfaces, the conductivity of graphene metasurfaces is closely related to the carrier concentration of graphene and the cell spacing. We can control the optical response of devices by controlling the carrier concentration in graphene or adjusting the cell spacing on the metasurface.

Ultrafast modulation of graphene can be realized due to its 200,000 cm²/(V·s) high carrier mobility at room temperature. But during the growth and transfer process of graphene, impurities and defects will be generated in graphene, which will lead to a decrease in the carrier mobility of graphene. Therefore, we simulated the reflectivity of the filter under different carrier mobilities, as shown in Figure 6. The filtering efficiency slightly decreased at carrier mobility of 10,000 cm²/(V·s) and 20,000 cm²/(V·s), reaching 94% and 96% respectively, without affecting the overall filtering performance. The thickness and distribution of graphene can also affect the performance of tunable devices [40]. We simulated the effect of graphene thickness on the filter, and the results are shown in Figure 6b. It can be seen that single-layer graphene exhibits the same effect as double-layer and three-layer graphene. During the processing, the stacking of multi-layer graphene may be easier to maintain its own photoelectric properties compared to single-layer graphene, and it is easier to achieve the desired performance.

The filtering performances of filter under different linear polarization directions of incident light and under oblique incidence are also significant indexes of the filter. Filters that are polarization insensitive and able to maintain high efficiency over a wide range of angles of incidence have greater applicability. In order to verify the polarization performance, we simulate the reflection spectra when the polarization direction of the incident light is gradually shifted with the step of 30° from parallel to the x-axis to parallel to the x-axis again. The reflectance spectra at 0°, 30°, 60°, 90°,120°,150°and 180° are identical, as illustrated in Figure 7a, which demonstrates the polarization-insensitivity performance of the proposed structure. By analyzing the electric field distribution on the metasurface with the alteration of the polarization direction from the x-axis direction to the y-axis direction, it can be observed that the resonance undergoes a gradual shift from the y-axis direction to the x-axis direction, as depicted in Figure 7b–e. This suggests that the incident wave resonates exclusively in the cell spacing perpendicular to its own polarization direction. Due to the incident light with any polarization direction can be regarded as the vector sum of two components in the horizontal and vertical directions, the device has the same resonant intensity for electromagnetic waves with any polarization direction, exhibiting polarization insensitive characteristics.

In addition to exhibiting insensitivity to the polarization of the incident wave, it is also important absorbing character of the different incidence angles. Fixing voltage V_g = 10 V, we simulate the reflectivity and electric field distribution under different oblique incidence conditions, as shown in Figure 8. As illustrated in Figure 8a, the reflectivity exhibits a slight increase of only 1.3%, when the incident wave transitions from vertical incidence to oblique incidence with an angle of 30°. In the case of oblique incidence at an angle of 60°, the depth of modulation is still able to reach more than 85%. The filtering efficiency decreases to 28.2% when the incident wave is obliquely incident at 85°. We analyze the electric field distribution with the change of the incidence angle, as shown in Figure 8b–d. When the incident plane wave is incident obliquely at an angle of 60°, Figure 8b shows that the electric field strength on the surface of the device exhibits a reduction to 87% in comparison with that at vertical incidence. Additionally, when the incident plane wave is incident obliquely at an angle of 85°, the electric field strength on the metasurface decreases to about 30% of that at vertical incidence, which is similar to the alteration of the modulation depth. The proposed structure displays remarkable robustness with regard to the angle of incidence of the plane wave, it is capable of achieving filtering efficiency over 85% within the range of 60°.

Compared with previous designs, our design has a filtering efficiency of over 99% and a tuning range of 1.7 µm, and can move the overall working range by adjusting the structural dimensions. In contrast, it also has polarization insensitivity and higher stability under oblique incidence. And our design can tolerate relatively high manufacturing errors and has great stability. The performance comparison between our proposed filter and previous filters is shown in the

Table 2. Comparison between the proposed tunable filter and previous works.

	Proposed Model	Filter in [22]	Filter in [20]	Filter in [6]
Polarization insensitive	Yes	No	No	Yes
Oblique incidence restriction	85% (60°)	95% (60°)	20% (60°)	None
Filtering efficiency	99%	96.7%	95%	70%
tuning range	1.7 μm	5 μm	2.3 μm	0.5 μm

. It can be seen that our proposed filter can simultaneously have polarization insensitivity and the highest filtering efficiency, and has relatively high efficiency under oblique incidence. Accordingly, our proposed tunable filter model exhibits excellent performance in terms of polarization insensitivity and wide-angle incidence adaptation. This enables its use in a range of operating conditions and makes it a promising candidate for spectral filtering applications.

4. Machining Error Analysis and Fabrication Processing

For the metasurface, the size error of the microstructure exerts a notable influence on the operational performance of the model. The processing inevitably introduces a degree of inaccuracy, which potentially affect the working range, the modulation depth and among other factors. Consequently, a high-precision fabrication is essential for the designed performance of these devices. The far-infrared tunable filter model proposed in this paper is a relatively simple structure comprising a metal substrate, a dielectric layer, graphene, and meta-atoms. The period of the unit structure is 2 μm with the linewidths of the microstructures are also on the order of a hundred nanometers, which provides a high degree of stability against processing errors. In order to analyze the machining error tolerance of the proposed structure, the manufacturing errors are within negligible ranges that make the center wavelength shift no more than 0.1 μm and the reduction of the filtering efficiency no more than 1%. Because the influence of the length of gold inside the unit structure (L₁) on the filter corresponds to the influence of the cell spacing, the results are shown in Figure 5. In this section, there are four parameters including the width of the gold (w__Au), the length of the edge of the gold (L₂), the thickness of gold (d__Au) and the thickness of SiO₂ (h_SiO₂) to be analyzed carefully for the filtering performance, as illustrated in Figure 9. When one of the parameters is analyzed, the others are set according to the predefined dimensions. As illustrated in Figure 9c,d, the length of the gold L₂ affects the center wavelength, with an increase of 50 nm resulting in a 0.3 μm shift. The filtering efficiency is also capable of achieving a value exceeding 99%. Therefore, to ensure that the center wavelength remains within a permissible range of 0.1 μm, a processing error of 1000 nm ± 15 nm is permitted. The width of the gold on the cell structure exerts a minor influence on the center wavelength and filtering efficiency, as illustrated in Figure 9e,f. The width of gold increases from 90 nm to 110 nm resulting in the center wavelength decreasing from 13.5 μm to 13.2 μm. So the allowable processing error is 100 ± 10 nm, which basically maintains an accurate working range and modulation depth. As shown in Figure 10a,b, the effect of the gold thickness on the filter is negligible in the range of 50–70 nm, because the filtering efficiency and the center wavelength can be kept unchanged. Finally, we simulated the effect of the thickness error of the dielectric layer. As shown in Figure 10c,d, the results demonstrated that an increase of the dielectric layer thickness from 360 nm to 440 nm leads to a wavelength shift from 13.1 μm to 13.6 μm. Due to the current processing technology being able to accurately control the thickness of the silicon dioxide layer, the thickness (h_SiO₂) of the silicon dioxide layer can be used as a controllable parameter to move the working range of the filter to the target spectrum. For every 40 nm increase in silicon dioxide thickness, the reflection center wavelength can be increased by 0.25 μm. Before preparing the filter, the thickness of SiO₂ can be adjusted to make the working range near the target spectrum. Combined with the voltage regulation ability after preparation, the working range of the filter is greatly expanded, and the flexibility of the filter is improved. In conclusion, it is possible to allow for a manufacturing error of± 10 nm or more for each structural parameter, which greatly improves the processing feasibility to enhance the performance of the tunable filter.

When preparing the proposed filter, 500 nm aluminum was first magnetron sputtering on a silicon substrate, and then 400 nm silicon dioxide was deposited on the aluminum using plasma enhanced chemical vapor deposition (PECVD). Next, a single layer of graphene is deposited onto a silicon dioxide layer through chemical vapor deposition (CVD). The unit structure array on graphene is processed by electron beam evaporation of 60 nm gold on graphene, followed by electron beam lithography (EBL) of the unit structure and subsequent lift-off. Due to the need for multiple processes on graphene during the processing, a protective layer needs to be added between gold and graphene to prevent the performance of graphene from being affected. During the process of peeling off gold, it is necessary to maintain structural integrity while reducing the impact on graphene.

5. Conclusions

A polarization-insensitive tunable filter for the free-space far-infrared light is proposed based on the graphene metasurface. With a gate bias of a few volts to tune the graphene conductivity, the room temperature device achieves high modulation depths of up to 99%. The proposed structure is capable of achieving a tunable center wavelength spanning a range of 12.4 μm to 14.1 μm, with a total variation of 1.7 μm. Due to the symmetry of the proposed structure, the filter is insensitive to the polarization of the incident light. Furthermore, the filtering efficiency of greater than 85% can be achieved within the range of a 120° angle of incidence. In addition, the filter is easy to be manufactured due to its simple structure and high tolerance processing errors. By adjusting the parameters of the unit structure, it is also possible to move the operating range of the filter to the other desired spectrum. These properties indicate that this scheme has a large application prospect in the fields of signal acquisition and processing as well as infrared spectral filtering.