1. Introduction
The emergence of powerful electromagnetic attack weapons poses a significant risk to sensitive electronic devices and photoelectric detection systems, as they are vulnerable to high-power electromagnetic attacks [
1,
2,
3]. On the other side, as the application of technology deepens, frequency band resources are gradually becoming scarce. This scarcity leads to increased signal interference between different devices, and intensifies the competition for spectrum resources, which has now reached a white-hot stage. The photoelectric detection system requires robust electromagnetic protection materials that possess not only high shielding and ultra-wideband protection against strong electromagnetic attacks but also excellent infrared transmittance. This is essential for ensuring effective signal transmission and meeting the demands of high-quality communication and imaging. Electromagnetic shielding technology aims to mitigate the negative effects of electromagnetic interference (EMI) by utilizing effective shielding measures. Its purpose is to safeguard against potential harm caused by electromagnetic waves and to explore the use of shielding materials or methods to block or weaken the propagation of electromagnetic signals between a protected area and the outside world [
4,
5,
6]. Filters are one of the many devices used for selecting frequency bands. They work by allowing electromagnetic waves of specific wavelengths to pass through. Traditional filters come in two main structures: a grid structure [
7,
8] and a metal-based structure [
9,
10]. The current transparent electromagnetic shielding materials, such as metal oxide conductive film and metal nanowire film materials, face a challenge in achieving high infrared transmittance and microwave shielding efficiency simultaneously. This is due to the mutual restriction between their infrared transmittance and conductivity. On the other hand, electromagnetic shielding materials that rely on metal grids face the challenge of limited shielding bandwidth due to high-pass filtering characteristics. This can result in a reduction of shielding effectiveness (SE) in the high-frequency region. Currently, most of the electromagnetic shielding materials that use metal grids employ a multilayer film system in combination with the grid. The multilayer film system utilizes materials that meet the demands for both high transmittance and light filtering. Additionally, the grid structure effectively shields microwave band signals. Furthermore, when utilizing metal-based conductive electromagnetic shielding materials, such as metal-based conductive transparent film and metal nanowire film, the electromagnetic energy is primarily shielded through either electromagnetic reflection or absorption. However, this process can result in a more intricate electromagnetic environment, leading to potential issues such as electromagnetic interference and radiation heating. Currently, it is of the utmost importance to discover a suitable structure or material that can effectively address the layer adhesion and diffraction issues in multilayer film and grid structures, respectively.
Surface plasmons (SPs) possess exceptional characteristics, including subwavelength size, high electric field localization, and local field enhancement [
11,
12]. By etching micro/nano structures of varying sizes onto a metal surface, the surface plasmons can enhance the optical transmission (EOT) and filtering effect of specific wavelengths [
13,
14,
15,
16]. Simultaneously, the conductive structure created by the remaining metal film after etching forms a continuous shielding layer. This layer achieves high shielding and ultra-wideband effects in the radar band, effectively addressing the limitations of traditional strong electromagnetic protection materials.
In this article, the circle hole array structure has been designed and optimized by a full wave simulation. Firstly, a filter, which consists of a circle hole array with a period of 4 μm and a circle hole radius of 1.5 μm, with a transmittance of more than 85% and an electromagnetic SE of more than 50 dB in the 3.7–4.8 μm band, is designed in the middle-infrared band. At the same time, a filter, which consists of a circle hole array with a period of 8 μm and a circle hole radius of 3.5 μm, with a transmittance of more than 90% and an electromagnetic SE of more than 45 dB in the 8–12 μm band, is designed in the long infrared band by optimizing the parameters of the geometric structure. It is worth noting that the optical performance and electromagnetic shielding performance of the medium wave filter were verified through experiments. Moreover, the corresponding color is extracted by inverting its transmission spectrum in the visible light band into the CIE1931 color co-ordinates, and its type and performance directly correspond one-to-one to realize the visual identification of filter. Finally, through the method of the cascade theory, a large-thickness substrate is fitted on the side of the ultra-thin metal film as a load, which effectively solves the problem of the large-thickness substrate not being able to be meshed under the simulation software.
2. Design and Modeling
2.1. Structure Model
When considering the impact of the polarization direction of an incident wave on a structure, it is advisable to focus on a circle hole structure that boasts complete symmetry. This type of structure is the ideal subject for research purposes.
Figure 1 depicts a unit structure diagram, where the transparent substrate (made of sapphire material with a loss tangent value of 0.0004 in the CST studio suite simulation software material library used for simulation) is represented in blue, and the metal material (described by the modified Drude model [
17,
18] to express its relative dielectric constant) is shown in yellow. In this subject paper, we use
Px and
Py to represent the structure periods along the
X- and
Y-axes. Then, a metal thin film, with a thickness of
h1, is situated on top of a substrate with
h2 thickness. One way to minimize the impact of the polarization direction of the incoming electromagnetic wave is by creating a circle hole that goes through the entire metal film layer. In this setup, the incident light travels perpendicular to the surface of a metal film, while the polarization direction of the electric field is in the
X-axis. The simulation of the light propagation of the subwavelength metal holes is achieved through the use of the finite difference time domain (FDTD) method. Among them, the transmittance
T:
T =
Pout(
λ)
/Pin(
λ), where
Pin(out) is the power flux through the metal membrane [
19].
2.2. Structural Color Theory
According to color science [
20,
21], any color in nature can be displayed as a combination of the three primary colors of red, green, and blue in a specific proportion. Measuring the color corresponding to the simulated spectrum can predict the range of color gamut that can be covered by the designed structure, so as to provide guidance for optimizing the structural parameters to obtain the widest color gamut. At the same time, it can also identify the types and corresponding performances of each structure and device according to the color index. In the second-field-of-view XYZ system, if the object is in the relative spectral power distribution of
D(
λ), the transmittance at different wavelengths is
T(
λ); the
X,
Y, and
Z corresponding to the transmission color of the object can be calculated according to the following formula [
22,
23]:
where the wavelength integration range shall be within the visible light band (400–700 nm);
I(
λ) is the spectrum of the standard illuminant incident light source (D65) closest to sunlight;
T(
λ) is the transmission spectrum;
x’(
λ),
y’(
λ), and
z’(
λ) represent the stimulus value of the three primary colors;
Y represents not only the stimulus value of the green primary color to the human eye, but also the brightness of the object color; and
k is the normalization coefficient, and its definition value is: adjust the
Y stimulation value of the perfect reflector (reflectivity: 1) or ideal transmissive material (transmittance: 1) to 100, so the calculation formula of
k is:
In order to express the relative proportion of the stimulus amount of the three primary colors
X,
Y, and
Z in the total amount of
X + Y + Z, the international lighting committee introduced the chromaticity co-ordinates x, y, and z to represent the proportion of the primary colors and formulated the chromaticity diagram. For nonluminous objects, the calculation formula of the chromaticity co-ordinates x, y, and z is:
(x, y) is the chromaticity co-ordinate of the transmission spectrum; thus, we only need to substitute the simulated or experimental transmission spectrum into the formula for calculation, and the color of the obtained structural color pixel can be accurately and quantitatively described by the tristimulus value, that is, the chromaticity value X, Y, and Z of the color presented by the object. Meanwhile, in the chromaticity diagram, the physical meaning of the X-axis co-ordinate is to reflect the relative proportion of the red primary color; the physical meaning of the Y-axis co-ordinate is to represent the relative proportion of the green primary color; and the point outside the horseshoe shape represents the chromaticity of the color that does not exist physically. We know that different colors have different chromaticity co-ordinates, and they occupy different positions in the chromaticity diagram. A group of chromaticity co-ordinates (x, y) in the figure represent common characteristics of colors with the same tone and saturation but different brightness. Devices or structures that have a transmission or reflection spectrum can have their colors accurately displayed through spectral inversion in the visible light band. Essentially, each device or structure has its own unique spectral distribution, resulting in distinct color forms.
2.3. Two-Port Transmission Line Theory
For microwave networks with arbitrary ports, we can characterize them using the
Z,
Y, and
S parameters [
24]. However, it is important to note that many microwave networks actually comprises multiple two-port networks. In this case, a 2 × 2 transmission matrix (also known as an ABCD matrix) can be used to define the situation of each dual-port network.
The details are as follows:
The cascading of a network can be represented as:
Furthermore, convert the parameters of the network to:
As shown above, V1 is the total voltage flowing into the port, and V2 is the total voltage flowing out of the port; I1 is the total current flowing into the port, and I2 is the total current flowing out of the port; S11 is the input reflection coefficient passing through the port network, S22 is the output transmission coefficient passing through the port network, S12 is the feedback coefficient passing through the port network, and S21 is the forward transmission coefficient passing through the port network; and Z0 is the characteristic impedance.
Through measurement or theoretical analysis, a description of the ‘black box’ effects on these dual ports can be obtained. The characteristic of the transition section can be expressed through the network parameters (Z, Y, S, or ABCD) of the two-port network. Generally speaking, in practical applications, simulating the model structure of thick transparent substrates or thin films (relative to the response band) may result in an excessive program load due to the need for excessive grid partitioning. However, the transmission line cascade theory calculation is a useful method for obtaining the characteristic parameters of a single thin layer material through calculation. By cascading these parameters according to the requirements, this method can effectively solve the problems mentioned above.
The transmission spectrum of the cascade calculation and direct simulation under different conditions are compared and analyzed as shown in
Figure 2. For a single sapphire transparent substrate in
Figure 2a, the cascade calculation results of different thicknesses are in good agreement with the direct simulation results. Obviously, the transmission spectrum shows periodic fluctuations. This phenomenon can be explained by the etalon effect [
25]:
Herein,
a and
b are the absorption coefficient and the thickness of the sample, respectively.
R is the reflectivity of the one-sided reflection;
θ =
4πnb/
λ0;
λ0 is the wavelength of the radiation in the vacuum, and
n is the refractive index. As the wavelength increases,
θ changes, leading to periodic changes in cos
θ, ultimately leading to periodic oscillations in transmittance. In
Figure 2c, we used a Fourier-transform infrared spectrometer to measure the infrared transmittance of the sapphire substrate in the 2.5–6 μm range. Firstly, when the wavelength is greater than 5 μm, the transmittance significantly decreases due to the material characteristics of sapphire: the transmittance range of the sapphire material in the infrared band is between 2–6 μm. At the same time, the measured transmittance distribution spectrum also shows obvious oscillation characteristics, but, due to the dense oscillation period, the curve distribution is relatively smooth compared to
Figure 2a. The difference is that, for metal materials only (see
Figure 2b), the cascade calculation results produce a slight discrepancy compared with the direct simulation result, but this will not affect the maximum amplitude and trend of the spectral distribution. This difference is attributed to the strong coupling between electrons and photons in metallic materials. In the previous work [
26], cascade calculations were carried out for metal films with different thicknesses. The research results indicate that the number of cascade layers of metal films directly impacts their similarity to direct simulation results. Furthermore, the equivalent circuit model calculation reveals that the deposition of metal film on a transparent substrate results in coupling to some degree, with the strength of this coupling being dependent on the thickness of the substrate. In general, reducing the response frequency results in an improvement in the accuracy of the cascade calculation. As a result, for the subsequent direct simulation, we aim to address the issue of excessive mesh generation leading to lengthy simulation times. To achieve this, the following direct simulation only models and simulates a single metal film that has been etched with micro/nano structures. As shown in
Figure 2c, the sapphire substrate has a relatively high transmittance at 3–5 μm.
3. Results and Discussion
Detectors and sensors operating in the medium and long infrared bands are widely utilized in photoelectric detection systems [
27,
28,
29]. Specifically, medium infrared detectors and sensors typically operate in the 3.7–4.8 μm range [
30,
31], while long infrared devices operate in the 8–12 μm range [
32,
33].
The geometric parameters of the circle hole structure were optimized under different metal thin film materials (Au, Ag, and Al), as shown in
Figure 3. Upon analysis, it has been discovered that the transmission spectrum remains largely consistent between 3.7–4.8 μm when the thickness of the metal film (
h1) is altered from 50 nm to 300 nm. Meanwhile, it is noted that, when the radius (
r) of a circle hole changes from 1.3 μm to 1.7 μm, the transmission peak remains stationary, but the peak width increases. From the previous work [
34], in a square array, if
εd and
εm have been determined, then the period
P of the structure is one of the main factors controlling the position of the transmission peak when the incident light is perpendicular to the metal film. When
P (4 μm) remains constant, only the radius
r can be altered and the transmission peak will remain unchanged. However, if
P increases (meaning that the proportion of the circle hole structure in the same area decreases), the electric field localization on the metal surface becomes stronger. This results in an enhanced coupling of the charge density on the circle wall, causing the effective refractive index
neff to increase. As a result, the transmission peak will experience a red-shift. It is important to note that the intensity of the transmission peak decreases as a result of the duty cycle reduction.
The change of structure has a nonnegligible effect on the infrared transmission characteristics [
35,
36,
37]. In cases where the polarization effect of the incident light is weaker, other structures besides the circle hole structure may be present. The
Figure 4 provides the transmission spectrum of four different structures (all with the same duty cycle) in the middle infrared band. Additionally, the electric field distribution of the corresponding structure at the wavelength with the maximum transmission peak intensity is included for a more in-depth analysis of the physical mechanism. It is clear that the full width at half maximum (FWHM) of the transmission peak is wider when the periodic structure is connected as a whole (circle and cross-shaped) than when it is not connected (cylinder and hollow cross-shaped) as shown in
Figure 4a. Moreover, the connected periodic structure exhibits a broader response range in the infrared band. As a continuous conductor structure, the electromagnetic SE in the radar band is higher. Additionally, as depicted in
Figure 4b–e, the charge distribution is more localized for cylinder and hollow cross-shaped structures, resembling metal nanoparticles. As known in reference [
38], metal nanoparticles in close proximity to one another are impacted by external light fields. The dipole radiation field of one nanoparticle can disrupt the dipole radiation field of the other, altering the force of free electrons and subsequently changing the resonance frequency and distribution of electron clouds. On the other hand, the interaction between metal nanoparticles is more significant in terms of frequency selection, resulting in a narrower FWHM of the transmission peak. Meanwhile, the cylindrical structure and the circular hole structure are complementary structures, and their light propagation paths are completely opposite. In a circular hole structure, the propagation path of light is within the circular hole, and the charge distribution is more compact compared to the entire periodic structure. However, the light propagation path of the cylindrical structure is between adjacent cylinders. Compared to the circular hole structure, the path of light propagation is wider, and the charge distribution is sparser. The enhancement effect of the local field is not as strong as that exhibited by the circle hole structure. Therefore, as shown in
Figure 4a, the peak width of the transmission spectrum of the cylindrical structure is smaller than that of the circular hole structure. Similarly, the cross-shaped and hollow cross-shaped structures are the same.
In the same vein, the structures of the material and geometric parameters are optimized for the long infrared band. Unsurprisingly, the thickness of the metal thin film does not significantly impact the transmittance, as seen in
Figure 5 in the medium infrared band. Meanwhile, increasing the radius does not shift the transmission peak but does result in a noticeable increase in the FWHM. Additionally, the period
P changes in a linear fashion, as predicted in the literature [
34].
In summary, the medium infrared band filter can be optimized through the careful consideration of material types, geometric parameters, and practical application. The recommended parameters for this filter include using Au as the metal material, a thickness of 0.1 μm, a circle radius of 1.5 μm, and a period of 4 μm. For the long infrared band filter, the recommended parameters also include using Au as the metal material, a thickness of 0.1 μm, a circle hole radius of 3.5 μm, and a period of 8 μm.
To facilitate the identification of device performance based on its geometric parameters, we utilized the structural color theory to invert the transmission spectrum of filters from two distinct working bands in visible light (as depicted in
Figure 6a) into the CIE1931 color space. From there, we extracted the corresponding color modules for easy visual discernment. The color effects are displayed in
Figure 6b,c. By using inversion measures, the performance changes of the designed filter can be easily identified with the naked eye. This means that under normal lighting conditions, filters with different performance capabilities can be distinguished by their color differences. In
Figure 6d, the electromagnetic EMI SE analysis of the two filters in the radar band of 1–18 GHz shows that the results exceed 45 dB. This suggests that the continuous metal film structure can achieve ultra-high efficiency in electromagnetic shielding.
4. Manufacture and Measurement
To further verify its feasibility, a medium wave filter with high transmittance and ultra-wideband electromagnetic shielding was manufactured, with the following parameters: P = 4 μm; the radius r = 1.5 μm; the substrate material h2 = 0.5 mm; and the surface layer is coated with gold with a thickness h1 = 100 nm.
This study utilized mature processes of ultraviolet (UV) photolithography [
39,
40] to prepare the samples for the fabrication of metallic micro/nano structures. Other methods such as nano-imprinting [
41] and laser direct-writing technology [
42] can also be used for this purpose. The process flow, as shown in
Figure 7a, begins with the deposition of a layer of metal film consisting of 30 nm Cr (as an adhesive layer) and 100 nm Au onto a clean substrate surface. Next, a layer of photoresist is spin-coated onto the metal surface and stoved. The stepper process is then used for exposure and development, followed by IBE etching of the sample. Finally, the remaining photoresist is removed. SEM images of the micro/nano structure are presented in
Figure 7b,c.
The transmission spectrum under different conditions are presented in
Figure 7d. The brown curve represents the actual measurement result. After comparing the cascaded calculation and fitting results, it was discovered that the transmission curve distribution remains relatively consistent within the 3–5 μm band. Further, through analysis, it was found that, due to the errors in exposure time and etching time during the preparation process, the metal linewidth of some of the obtained structures was larger than that of the simulation, and the structural period also experienced errors due to changes in linewidth. Therefore, the transmission peak undergoes a peak width decrease phenomenon at 3–5 μm, which is consistent with the results in
Figure 3b,c. Furthermore, during the etching process, due to the small thickness of the metal film, it is difficult to control the etching time, resulting in uneven pits on the substrate surface after etching, thereby reducing the overall transmittance of the sample.
The rectangular waveguide is connected to the coaxial line through the waveguide coaxial converter and connected to the vector network analyzer to collect
S parameters as shown in the
Figure 8a. According to the measured result in
Figure 8b, the average EMI SE of 1.7–18 GHz is 45 dB. It follows that, combined with
Figure 6d and
Figure 8b, the dependence of EMI SE on the cycle plays a dominant role, indicating that the subwavelength structure is conducive to reducing the dependence of EMI SE on the frequency, especially the electromagnetic shielding stability in the high-frequency range. It cannot be ignored that, during the measurement process, there are errors such as the stability of the connection between the waveguide and the sample, as well as manual operations, which result in certain fluctuations in the final measurement results, but they are basically consistent.