Bio-inspired multispectral camouflage material for microwave, infrared, and visible bands based on single hierarchical metasurface

2.1 Single metasurface for microwave and IR camouflage

Diffuse-scattering metasurfaces can reduce the radar cross section (RCS) in the backward or specular direction by randomly scattering the EM energy in various directions in space [13]. This is an alternative microwave camouflage method for absorption. This mechanism relies only on the phase cancellation between different elements rather than the absorption of incident EM waves. This enables the use of good conductors in metasurface elements. Good conductors also have a low emissivity in the IR band, which allows for IR camouflage. Therefore, we propose a strategy to achieve camouflage in both IR and microwave bands on a single metasurface by designing cross-scale structures. The mechanism of RCS reduction (RCSR) achieved by a diffused scattering metasurface can be described using the antenna array theory. For a ground plane, the array factor (AF) can be represented by Eq. (1) [42]:

where β _x and β _y are the phase shifts between the elements in the x- and y-directions, respectively. d _x and d _y are the spacings between the elements along the x- and y-axes, respectively. M and N are the numbers of elements in the x- and y-directions, respectively. The total RCSR, compared with that of a perfect electrical conductor (PEC), has the following formation:

where E^s and Eⁱ are the incident and diffused electric fields of the EM waves, respectively. Specifically, if the diffusion metasurface is composed of only two types of elements, the RCSR in the specular direction can be expressed as follows [19]:

where p ₁ , p ₂ are the filling ratios. r ₁ , r ₂ and φ ₁ , φ ₂ are the amplitudes and phases for the reflectivity of each element, respectively. According to Eq. (3), if the phase difference between φ ₁ and φ ₂ among 180° ± 37° over a broad frequency range, the metasurface can achieve a broadband −10 dB RCSR in the backward direction even for lossless elements, r ₁ = r ₂ = 1. The reflection phase of a specific metasurface element can be adjusted by modifying its structure [29]. Therefore, broadband RCSR can be achieved through phase cancellation by designing different unit structures in a metasurface composed of good conductors. Moreover, Eq. (1) indicates that the diffused scattering of EM waves can be realized in backward space by randomly arranging cells with different reflection phases, thus achieving microwave camouflage [43].

With regard to the IR camouflage performance, according to the efficient medium theory, the IR emissivity (E) is related to the area-filling fractions X₁, X₂ and IR emissivity E₁, E₂ of different materials, as follows [29]:

When the composition of a metasurface is determined, altering the filling fractions of the different components can yield different IR emissivity values. If these elements with different IR emissivity have a special arrangement, IR camouflage can be achieved rather than only low IR emissivity. Therefore, if we can design metasurface units whose reflection phases depend directly on the filling ratio of the low IR emissivity conductive materials, IR camouflage and broadband microwave camouflage can be achieved on a single metasurface.

Chameleons can display different colors by contracting or stretching patch structures on their skin. These structures modulate EM waves in the visible band. Inspired by chameleons, we used a patch pattern to achieve microwave diffusion. The structural parameters of the patch pattern, including its period and gap, affect its EM response [29]. Specifically, the capacitance effect between each patch varies with the structural parameters. Therefore, we used a patch pattern to achieve microwave diffusion.

Figure 2a shows two units: the n₀ × n ₀ and n₁ × n ₁ patches. p is the period of each unit, and g is the gap between each patch. The simulated reflection phases of different unit cells are shown in Figure 2b. A phase difference of 180° ± 37° existed from 9 GHz to 13.5 GHz. Therefore, the reflectivity of the metasurface composed of these two types of units can be calculated according to Eq. (3), which is shown in Figure 2c. The results indicate that the corresponding reflectivity is below −10 dB from 9 GHz to 13.5 GHz. Then, we randomly arranged the different reflective phase unit cells with the same quantity into a two-dimensional array to construct the MSCM. Figure 2d and e shows the 3D far-field scattering pattern of the PEC plate and MSCM under normal incidence at 10 GHz. As shown, the MSCM generates more scattered lobes compared to the PEC plate. The backward lobe intensity of MSCM is lower than that of the PEC plate, indicating that the MSCM achieves a backward RCSR. Figure 2f and g further present the 1D far-field scattering patterns of the MSCM and the PEC plate at 10 GHz for different azimuth angles. The results show that the intensity of main lobe, i.e., the backward lobe of the MSCM, is 13.1 dB lower than that of the PEC plate, while the lobe intensities in other directions are slightly higher than those of the PEC plate. This indicates that the MSCM achieves an RCSR of 13.1 dB at 10 GHz, which is consistent with the reflectivity of the MSCM at 10 GHz shown in Figure 2c. Meanwhile, the total energy scattered by the MSCM in the backward hemisphere is not reduced compared to the PEC plate since lossless patches are deployed. However, due to the interference between unit cells with different reflection phases, the MSCM forms more lobes, thereby achieving microwave camouflage.

Figure 2:

Related information about the proposed MSCM in microwave band (a) top view of units 0 and 1, where p = 30 mm, n0 = 40, g0 = 0.15 mm, n1 = 10, g1 = 0.15 mm, (b) simulated reflection phase by CST of units 0 and 1 and a phase difference of 180°± 37°, (c) simulated reflectivity of the metasurface. Simulated 3D far-field scattering pattern of (d) PEC and (e) the metasurface at 10GHz. Simulated 1D far-field scattering pattern of PEC and the metasurface at 10GHz for the azimuth angles of (f) 30° and (g) 60°.

For a common low IR emissivity material copper and polymethyl methacrylate (PMMA) substrate, these have IR emissivity of 0.05 and 0.95 [44], respectively. Then, Eq. (4) is modified as follows:

Figure 3a shows the relationship between n, g, and E. Moreover, the color bar represents the image captured by the thermal camera corresponding to the IR emissivity. As marked by the black star in Figure 3a, the IR emissivity of unit cells 0 and 1 are 0.19 and 0.35, respectively. A gap of 0.16 existed between the IR emissivity values. When the ambient temperature is determined, the colors of materials with different emissivity are different under the infrared thermal imager, as shown in Figure 3a. Thus, the proposed metasurface under a thermal imager will display the image shown in Figure 3b, which is a typical digital camouflage pattern.

Figure 3:

Related information about the proposed MSCM in infrared band (a) relationship between n, g, and E. The black stars represent the IR emissivity corresponding to the structural parameters of two units. (b) Calculated IR thermal imager of the metasurface.

2.2 Design of high transparency conductive film

Optical transparency is a crucial property for various scientific and technological applications including light transmission and optoelectronics [44]. Metals such as gold, silver, and copper are candidate IR camouflage materials. However, these exhibit low visible transmittance, which limits their applications. Currently, multilayer films based on oxide and metal overlays are widely used in visible-light antireflection [45], low IR emission [46], and microwave shielding [47]. These can achieve low IR emissivity and high transparency. As common and widely used materials, Ag and AZO play important roles in many optoelectronic devices owing to their modulating properties on EM waves [48], [49], [50].

For a k layer nonmagnetic multilayer film, the transmission matrix is calculated according to the transmission line theory [51]. The schematic model and ECM model of the oxide and metal overlapping multilayer film are shown in Figure S1 (Supporting Information).

where h is the thickness of the layer, λ is the wavelength of the working frequency, β = 2 π ε / λ is the corresponding propagation constant, Y t = ε / Z 0 represents the characteristic admittance of the corresponding material, Z₀ = 377 Ω represents the wave impedance of vacuum, and each layer has its dielectric constant ɛ = ɛ′ − 1j ×ɛ″. The optical parameters and corresponding EM parameters of Ag and AZO in IR and visible bands are shown in Figure S2 (a) to (d) (Supporting Information). R and T represent the reflection power and transmission power of the material, respectively. They can be calculated as follows:

According to Kirchhoff’s laws of thermal radiation and energy conservation, the emissivity (E) of a flat material can be expressed as follows [52], [53]:

where α represents the absorption power. Usually, multilayered films are fabricated on a PMMA substrate, and the influence of the substrate on the IR emissivity and visible transmittance can be omitted (see Section S1.2 in the Supporting Information for details). Using a typical total oxide thickness of 100 nm and total metal thickness of 10 nm, we calculated the visible transmittance and IR emissivity under different configurations such as oxide (O)–metal (M)–oxide (O) according to Eq. (9). The aim was to initially obtain an ideal configuration.

As shown in Figure 4a, Model I exhibits the best visible light transmission performance, followed by Model II. Reference [54] explains the mechanism behind this result. This is due to the fact that Ag and AZO have negative and positive dielectric constants, respectively, in the visible light range (see Supplementary Information Figure S2). Therefore, Model I forms a symmetric structure with positive–negative–positive dielectric constants. The EM tunneling effect in such symmetric structure allows the visible light to pass through almost perfectly, thereby achieving higher transmittance. This mechanism is different from the Fabry–Pérot effect since it is independent of the thickness of the negative dielectric constant layer and is also broadband. Model II can be considered as the superposition of two sets of symmetric structures, but the additional interfaces somewhat affect the transmittance, making it suboptimal. The other four models lack this symmetric structure, so their transmittance is inferior to that of Models I and II. Reference [54] also points out that when EM tunneling occurs, the magnetic field localizes near the negative dielectric constant layer. Figure 5a and b shows the normalized magnetic field distribution of the Model I at wavelengths of 450 nm and 700 nm. As shown, the magnetic field indeed localizes near the Ag film.

Figure 4:

Design of the multilayer films: (a) visible transmittance and (b) IR emissivity of different configurations.

Figure 5:

Normalized magnetic field distributions at (a) 450 nm, (b) 700 nm, (c) 4.5 μm, and (d) 11.5 μm of Model I.

Figure 4b presents the infrared emissivity performance of each model calculated based on Eq. (6)–(9). In addition, Figure 5c and d displays the normalized magnetic field distribution of Model I at wavelengths of 4.5 μm and 11.5 μm, respectively. It can be observed that the incident electromagnetic waves are completely reflected by the Ag film and do not penetrate into the bottom AZO layer. The magnetic field distribution in the top AZO layer is relatively uniform, indicating a tiny absorption rate of electromagnetic wave energy within it. Therefore, according to Eq. (9), Model I exhibits low infrared emissivity, as shown in Figure 4b. The detailed field distributions of Model II to VI are shown in Section S1.3 (Supporting Information) for further analysis.

Table 1 provides a comprehensive comparison of the performance of these multilayer film models in both the infrared and visible light ranges. Model I and Model III exhibit more balanced advantages. However, Model III is more complex in terms of processing technology compared to Model I, due to its greater number of layers and thinner individual layer thicknesses. Therefore, we selected Model I in this work. The calculated average IR emissivity s and average visible transmittance are 0.048 and 0.86, respectively. The parameter sweeping results are shown in Figure S6 (Supporting Information) to analyze the influence of thickness variation. Furthermore, we calculated the square resistance R_S of the proposed film at ambient temperature and pressure, as follows [15]:

where σ_i and h_i are the conductivity and thickness, respectively, of each layer. At ambient temperature and pressure, the conductivities of AZO and Ag are 1 × 10⁶ S/m and 6.3 × 10⁷ S/m [54], respectively. The calculated is approximately 1.1 Ω/sq. Such a low R_S means that it exhibits the same EM properties as metals in the microwave range [55]. Therefore, the copper in the original MSCM can be replaced by the AZO–Ag–AZO multilayered film, which has a high visible transparency and low IR emissivity.

2.3 Multitarget optimization for multispectral camouflage via genetic algorithm

We have constructed an MSCM with a hierarchical microstructure, featuring a minimum characteristic scale of 10 nm. The target wavelength (band) of such metasurface is from 300 nm (1,000 THz) to 0.15 m (2 GHz), thus extending over six orders of magnitude. This implies that even for full-wave simulation of its microwave performance, a mesh grid at the nm scale is required. Such simulations involve an astronomical computational scale, making it impossible to jointly optimize the multispectral performance of the proposed metasurface in the microwave range using full-wave simulation [31].

As mentioned in Section 2.2, we calculated the responses in the visible and IR bands using ECM. Similarly, we also can establish an ECM of the proposed MSCM [29]. We first calculated the ABCD transmission matrix as follows:

where ω is the operating angular frequency, Y t = ε PMMA / Z 0 , and β = 2 π ε PMMA / λ . k is equal to zero and one and represents each unit cell. The ECM model and calculation process of equivalent parameters R _k and C _k are shown in Section S1.5 (Supporting Information). Then, the reflectivity Γ under normal incident can be calculated as follows:

Now that we have the analytical ECMs for the proposed MSCM in the visible, infrared, and microwave bands, we can integrate them with some typical intelligent algorithms to perform inverse design of the metasurface on demand of multispectral performance. The genetic algorithm (GA) has been demonstrated as an efficient algorithm to search for an optimal solution for physical problems. It has been utilized successfully in the design of metamaterials [56], [57], [58], [59], [60], [61]. Therefore, we built a GA-based design program for the MSCM. In this work, the design task consisted of the following five parts:

1. visible transmittance T > 0.85,

2. IR emissivity of the film E < 0.05,

3. difference between the emissivity of units ΔE > 0.25,

4. microwave reflectivity Γ < −10 dB,

5. thickness h < 3 mm.

The setting, principle, and progress of the GA program are shown in Section S1.6 (Supporting Information).

After completing the optimization, a group of structural parameters extending from millimeters to nanometers was obtained in addition to the multispectral camouflage performance. The optimized result is as follows: h₁=50 nm, h₂=12 nm, h₁=50 nm, h=2.7 mm, g₀=0.14 mm, g₁=0.21 mm, n₀=46, n₁=7, x=0.5. Figure 6a and b shows the optimized visible transmittance and IR emissivity of the film, with an average transmittance of 0.89 and average emissivity of 0.04. The difference in IR emissivity between units is larger than 0.25. Figure 6c plots the calculated reflection phase of units 0 and 1. The purple region represents the frequency range of microwave reflectivity smaller than −10 dB. The optimized result is shown in Figure 6d, i.e., a −10dB microwave camouflage at 7 GHz–16 GHz. Evidently, the GA program has accomplished the design task. Compared to the original design shown in Figure 2, the optimized version exhibits superior camouflage performance across the visible, infrared, and microwave spectrums.

Figure 6:

Optimization results including (a) calculated visible transmittance; (b) calculated IR emissivity of the IRCM and each unit; (c) calculated reflection phase of units 0 and 1 (a phase difference of 180° ± 37° is obtained from 7 GHz to 16 GHz); (d) calculated reflectivity of the MSCM; (e) random order and the IR thermal imaginer of the MSCM; (f) simulated 1D far-field scattering pattern of the metasurface at 10 GHz; calculated oblique microwave reflectivity under (g) TE, (h) TM polarization modes.

Moreover, Figure 6e shows the IR thermal imaginer of the MSCM. Herein, the irregular IR camouflage performance would deceive the detectors. Figure 6f plots the 1D far-field scattering pattern of the MSCM at 10 GHz. As shown, the incident EM wave energy is also diffused in different directions. More information about the microwave camouflage mechanism of the optimized MSCM is presented in Section S1.7 (Supporting Information). Moreover, we calculated the microwave reflectivity under oblique incidence according to the method presented in Ref [29]. The calculated results are plotted in Figure 6g and h. As shown, for both transverse electric (TE) and transverse magnetic (TM) polarization modes, the MSCM is still capable of performing microwave camouflage at certain oblique angles of incidence. More details about the far-field scattering patterns under oblique incidence are shown in Section S1.8 (Supporting Information).