The traditional reflective metasurface consists of a metal pattern, a dielectric substrate, and a metal back plate. Different patterns of metal structures represent different properties (Modi et al., 2017; Zheng et al., 2015; Zhang et al., 2016). But no matter what kind of structure, the performance and response of the metasurface cannot be separated from the dielectric substrate. In other words, the thickness of the substrate is an important factor affecting the metasurface properties. A common PC meta-atom (Fan et al., 2018; Fu et al., 2021) shown in Figure 1A is employed to specify the importance of the substrate thickness. The biarc meta-atom has great PC performance. The blue part in Figure 1 is the dielectric substrate. Figure 1B depicts the variation of the reflection with the substrate thickness d _f1 . When d _f1 = 6 mm, the meta-atom has a great PC performance at high frequencies (9–13 GHz) and then increases d _f1 to 10 mm. At this point, the PC efficiency of high frequencies is significantly reduced. When d = 14 mm, it can be seen that the PC efficiency greatly improves at low frequencies. After many simulations, we find that there are some spikes in the co-polarized reflectance, which is mainly caused by the resonance of the metasurface structure. However, this does not affect our analysis of the influence of thickness on the PC performance of the metasurface.

In general, the periodicity of a meta-atom determines its operating frequency band. However, as can be seen from Figure 1, when the periodicity is fixed, different substrate thicknesses will also change the working frequency band. Therefore, in addition to the element periodicity, element geometric pattern and parameters, dielectric constant of the substrate, and the thickness of several shifts are also one of the important factors affecting the metasurface performance. However, as a 2D array of sub-wavelength metallic patterns, the research on the metasurface has always focused on the design of the metal pattern structure and neglected the longitudinal thickness of the substrate. Therefore, the design freedom of the metasurface can be improved by engineering the thickness of the medium.

Inspired by the influence of substrate thickness on the operation frequency band, we can freely design the double PC band metasurface by manipulating the efficient thickness meeting the requirement of different operation bands. For such a design, the most critical step is to realize that the effective thickness varies with the frequency.

The most straightforward idea is to make the unit medium have double “catopter”, which low-frequency electromagnetic waves through the first catopter and reflected by the second, while high-frequency electromagnetic waves are directly reflected by the first reflective plate. The introduction of the antenna design has injected new life into the metasurface. The patch is a typical microstrip antenna and can radiate EMW. From another perspective, the circular patch can be regarded as an FSS structure which realizes the pass band in a low frequency while the stop band in a high frequency.

The functional schematic diagram is shown in Figure 2. The patch antenna radiator is inserted into the substrate of the biarc meta-atom and acts as a second high frequency catopter in addition to the metal back plate. Through such design, the thickness required of the PC is satisfied at both the low frequency and high frequency, and the PC dispersion of the meta-atom is customized. Outside the two PC bands, a co-polarized window is deliberately left as the radiation window. The EMWs radiated by the patch can pass directly through the element without changing the polarization state and radiation direction while ensuring the out-of-band scattering by utilizing the difference of the PB phase response to polarization. Through the above study, a bridge can be built in the design of the metasurface and antenna and makes the integrated design of radiation and scattering possible.

As a typical microstrip antenna, the circular patch has excellent radiation properties. Its structural schematic diagram is shown in Figure 3; the main composition includes the following: the circular patch, dielectric substrate (F4B, ε _r = 2.65, the loss tangent of 0.001), metal, and coaxial port. The geometric parameters of the antenna include the following: the period of the antenna element (18 mm, 0.4λ, where the is the wavelength at 7.3 GHz), the thickness of the substrate (2.7 mm), and the radius of the patch (7 mm). The S11 and radiation patterns of the patch antenna are given in Figures 3B,C by simulation. The main characteristic is that the radiation beam angle is larger, but the gain is lower. When a large gain is needed on a particular occasion, a plurality of antennas are usually required to form an antenna array to meet the requirement. Thus, the proposed antennas are arrayed to form an 8 × 8 array with the size of 144 × 144 mm. Inspired by the Wilkinson power splitter, a sixty-four port microstrip feeding network (MFN) is employed to excite the antenna array. The S11 and radiation diagrams after the array are shown in Figures 2E,F, and the gain increased to 19.7 dB at 7.3 GHz.

The circular patch is not only a radiator but also an FSS with frequency selection characteristics (low frequency transmission and high frequency reflection). The traditional polarized conversion metasurface also has three layers, namely, the polarized rotating structure, dielectric substrate, and metal reflector plate, and the structure form is basically the same as the radiator. According to the theoretical analysis in Section 2, the performance of the polarized conversion surface is not only related to the structure pattern and period but is also closely related to the thickness of the dielectric substrate. Based on this, we can insert the circular patch into the dielectric substrate. The patch has band resistance characteristics, showing band pass characteristics at low frequencies and band resistance characteristics at high frequencies. Therefore, by inserting the patch into the substrate, it can simultaneously satisfy the condition that the thickness is small at high frequency and large at low frequency.

The electrical substrate thickness of the PC surface at different frequencies is adjusted via the band stop characteristic of the circular patch to meet the optimal thickness at different frequencies.

The combined structure is designed and shown in Figure 4. Figures 4A,B are the top and perspective view of the unit. The circular patch structure, a typical FSS structure, is inserted into the biarc unit. The metal patterns are etched on the F4B (ε _r = 2.65, the loss tangent of 0.001). The patterned slabs are pasted on polymethacrylimide (PMI) foams with ε _r = 1.1. Other geometrical parameters are as follows: a = 18.0 mm, w ₁ = 0.6 mm, w ₂ = 0.5 mm, d ₁ = 1.5 mm, d ₂ = 6 mm, d ₃ = 2.7 mm, α = 66, and r = 7 mm. The reflection (r _yx , r _xx , r _xy and r _yy ) is shown in Figure 4C, in which r _yx (r _xx ) represents the cross- (co-) polarization reflectivity under the x polarized wave while r _xy (r _yy ) is the cross- (co-) polarization reflectivity under y polarized waves. Encouragingly, we can see the design can realize high PCR in two bands. One band is 2.9–6.1 GHz, and the other is 8.2–13.7 GHz. The co-polarized reflection is even lower than −10 dB. Besides, between the two cross-polarization reflection frequency bands, there is a co-polarization reflection window. In the co-polarization band, the cross-polarization reflection is even lower than −10 dB, too.

The polarization characteristics required by the theory are realized by inserting the antenna. In 2014, the researchers developed the PB phase concept and studied its relationship to the angle of rotation meta-atom. It is found that the phase of PB is only related to the orientation of the element and is independent of the frequency, that is, when the element is rotated by a, the phase changes by 2a. In addition, the PB phase only affects cross-polarized waves. According to the characteristics that the PB phase only responds to the cross-polarized wave, the phase profile of the out-of-band wave can be designed by rotating the biarc unit. Based on such an idea, the biarc structure was rotated 90 around the center. The cross-polarization reflection amplitude and the phase before and after the rotation are simulated and displayed in Figure 5A. Through rotation, the reflection phase required for the checkerboard arrangement is obtained while the amplitude is kept unchanged.

However, there is another issue need to be confirmed, that is, whether the radiation performance would be affected while increasing and arranging the PC surface into a chessboard structure. In this regard, the transmission and the reflection of the PC surface and the dielectric sheet are also simulated and depicted in Figure 5B. It is attested that the co-polarized transmission mainly occurs when the wave through the PC surface without the metal board and the phase of the co-polarized transmission wave will not change. Thus, it will not worsen the antenna array radiation performance.

The previous section demonstrated that the desired polarization characteristic can be obtained by inserting the antenna radiator into the substrate and forming a new meta-atom. The metasurface can be seen as an antenna aperture with low scatter and a great radiation performance.

First, we consider arranging the PC surface to manipulate the scattering wave. One of the most significant research methods to reduce the RCS is to propose the checkerboard structure. Its operation mechanism is based on the cancellation effect of both contributions. The RCS reduction can be approximately calculated by the following equation (Pang et al., 2018): R R C S = 10 lg | ( 1 − f ) A 1 + f A 0 e j △ φ | 2 , (1) where A ₀ and A ₁ are the amplitudes of the different areas of the checkerboard, is the phase difference of the different area; f is the area fraction of the area 0. In other words, if the two meta-atoms have the same amplitude and the 180 ± 37 phase difference, the total scattering field will be cancelled in the normal direction and redirected toward the four principal quadrants.

In this case, the RCS and the far-field scattering pattern of the metasurface after arranging the chessboard are simulated and shown in Figure 6. Compared to the same size metal sheet, the bandwidth of RCS reduction covers 2.5–5.5 GHz and 8–12 GHz corresponding to a fractional bandwidth of 75 and 40%, respectively. From the far-field scattering pattern shown in the illustration, it is clearly seen that the EMW is scattered away from the normal incident direction.

Then, we consider the radiation performance comparison, it is shown in Figures 6B,C. When inserting the antenna array into the PC surface substrate, we can see that the radiation gain is improved. This is mainly due to the cavity formed between the upper PC surface and the antenna surface. As can be seen in Figure 6B, when an electromagnetic wave passes through a PC surface, some is reflected rather than completely passed through the PC surface. This energy is reflected again by the antenna and re-reaches the polarized rotating surface. After multiple reflectance and transmission superposition, the gain is improved.

Compared with the metasurface of non-rotation, the gain loss is lower than 1 dB. The reason causing the gain loss is that the upper PC surface will inevitably produce a transmitted PC, even though the energy is small. This causes a small part of the radiated EMW to be scattered as cross-polarized waves when it passes through the PC surface, and the 3-D far-field scattering pattern depicted in Figure 6B also demonstrates it. Nevertheless, overall, the simulation result indicates that the proposed metasurface can realize the out-of-band RCS reduction while radiating the EMW. The radiation efficiency is simulated and shown in Figure 6C. The in-band high efficiency is realized, although the efficiency has a little loss. While for the out-of-band, the efficiency decreases rapidly because of the impedance mismatch. Additionally, the corresponding aperture efficiency of the proposed metasurface from 6.5 to 8.5 GHz is also given in Figure 5C, which is calculated by (Zhang et al., 2020): ε a p = G λ 2 4 π A , (2) where G and A denote the simulated gain and the physical area of the reflect array aperture. It is seen that the aperture efficiency has a maximum value of 74.1% at 7.3°GHz, which indicates a good radiation performance.

To verify the designed metasurface’s radiation property, its radiation patterns are simulated and shown in Figure 7. At the central frequency 7.3°GHz, the gain is 20.3°dB, the 3 dB angular width is 14° degrees, and the side lobe is −12.4 dB. By comparing the cross-polarized and co-polarized components, the cross-polarized level is lower than -10dB, which demonstrated that the metasurface’s radiation still has good polarization purity despite the existence of the PC metasurface in the upper layer.