The proposed programmable metasurface realizes self-stealth by scattering cancellation. To achieve this goal, the meta-atom needs to possess multiple 1-bit phase modulation bandwidths, both in-band and out-of-band. Additionally, independent control of x- and y-polarizations within the desired frequency range is necessary. To accomplish these objectives, we propose a novel meta-atom structure as depicted in Fig. 2A. The meta-atom consists of 4 metal layers and 3 dielectric substrates. The dielectric substrates are made of Shengyi S7136H with a relative permittivity of 3.55 and a loss tangent of 0.003. The thickness of substrates 1 and 2 is 1.524 mm, while substrate 3 has a thickness of 0.5 mm. The 4 metal layers are arranged in the following order from top to bottom: top layer, middle layer, ground layer, and direct current (DC) feedline layer. The top layer consists of 3 different components. Among them, the 4 branch structures belong to the first out-of-band low-frequency band (fl₁), while the 4 cross-shaped dumbbell structures belong to the first out-of-band high-frequency band (fh₁). Additionally, the cross structure corresponds to the in-band. A PIN diode is integrated with the x-polarization branch. The model of this PIN diode is SMP1340, and its equivalent circuit is shown in Fig. 2A. By controlling the ON and OFF states of the PIN diode, a phase difference of 180° can be achieved in the x-polarization, enabling 1-bit phase control. However, there is no PIN diode integrated on the y-polarization branch. Moreover, the branches in the x- and y-polarizations are not directly connected at the center position of the top layer. Instead, they are isolated from each other and their characteristics are unaffected by employing a bridge design, as shown in Fig. 2A. The middle layer also contains 4 metal branches and 4 cross-shaped dumbbell elements rotated by 45°. They correspond to the second out-of-band low-frequency band (fl₂) and high-frequency band (fh₂). The DC feedline originates from the top layer, passes through the middle layer, and finally enters the DC feedline layer. It is located on the side of the cross-shaped patch, with a 56-nH choke inductor. A quarter-wavelength fan-shaped branch can also effectively suppress the influence of the DC feedline on radio frequency (RF). However, the large size of the fan-shaped branch restricts the scalability of the array. The detailed parameters of meta-atom are provided in Table S1. The relationship between the frequencies mentioned above is fl₂ < fl₁ < f₀ < fh₂ < fh₁. Therefore, the proposed meta-atom exhibits multiple resonant modes in the in-band and out-of-band. By adjusting the state of the PIN diode and the sizes of the structure, 1-bit phase differences can be achieved simultaneously both in-band and out-of-band.

The center frequencies of these 5 bands are fl_2c = 3.6 GHz, fl_1c = 4.16 GHz, f₀ = 5.2 GHz, fh_2c = 6.55 GHz, and fh_1c = 7.43 GHz, respectively. The structure corresponding to each frequency band is shown in Fig. 2A. The choice of these 4 out-of-band frequencies is determined by the in-band frequency. Our design principle is to achieve out-of-band RCS reduction without increasing the profile height or adding extra RF components, based on the programmable metasurface. Thus, we introduced parasitic structures on the original metasurface. We realized phase gradient through parasitic structures. This allows the utilization of scattering cancellation for out-of-band RCS reduction. After parameter optimization, we added parasitic structures around the in-band structure to introduce low- and high-frequency resonances. The low-frequency to high-frequency ratio, which does not affect the in-band resonance frequency, is approximately 0.8. Therefore, when fl_1c/f₀ = 0.8 and f₀/fh_2c = 0.8, changes in the parameters of the introduced high- and low-frequency structures do not affect the in-band resonance characteristics. Similarly, changes in in-band parameters do not affect the out-of-band resonance characteristics. Decoupled in-band and out-of-band design is necessary because dynamically regulating the in-band resonance of the programmable metasurface would otherwise affect the out-of-band characteristics. Moreover, to increase the out-of-band bandwidth, new out-of-band high- and low-frequency parasitic structures were introduced, based on the existing ones. The newly introduced parasitic structures couple with the existing ones, creating a new resonance frequency. Specifically, fl_2c/fl_1c = 0.86 and fh_2c/fh_1c = 0.88. The coupled-induced resonance points are not adjacent to the original ones, maintaining a certain frequency interval. Thus, the programmable metasurface achieved RCS reduction in 4 out-of-band frequency bands without increasing the profile or adding extra RF components, because of the above design approach. The meta-atom design underwent the processes of Cell-A, Cell-B, and Cell-C, as shown in Fig. 2B. These are associated with the in-band structure, the out-of-band structure, and the final structural design process. Figure 2C to E depicts the working states of Cell-A, Cell-B, and Cell-C. The characteristics of Cell-A, Cell-B, and Cell-C are described below.

The in-band component features a cross structure with 2 separate branches, shown in Cell-A of Fig. 2B. The in-band design follows 2 key principles. First, adjusting the PIN state enables 1-bit phase control for x-polarization while maintaining y-polarization resonance. Second, variations in the y-polarized structure’s size do not influence x-polarization. The design principle of Cell-A is described in Note S2. To verify the cross structure’s characteristics, full-wave simulations with periodic boundary conditions were performed. Cell-A parameters are provided in Table S2. The PIN diode was modeled using the equivalent circuit in Fig. 2A (R = 0.85 Ω, L = 0.45 nH, C = 0.21 pF). For x-polarization, switching the PIN diode’s ON/OFF states shifts the meta-atoms’ resonance frequencies (Fig. 3A and B), achieving a 1-bit phase difference (180 ± 20°) at 5.2 to 5.4 GHz. Notably, varying p2y (6 and 6.3 mm) minimally affects reflection coefficients and phases in both PIN states, confirming good x-/y-polarization isolation. Under y-polarization, the meta-atom achieves a 1-bit phase response at 5.25 to 5.4 GHz (Fig. 3B and C), with reflection characteristics closely matching x-polarization except for slightly broader x-polarized bandwidth. PIN switching does not alter y-polarization performance. The cross-polarization isolation is below −40 dB across all 4 states (Fig. 3D). Figure 3E illustrates the electric field distributions: In states 1 and 2, y-polarized structures exhibit weak fields under x-polarized excitation, while states 1 and 3 show minimal x-polarized fields under y-polarized excitation.

Traditional out-of-band stealth approaches typically use absorbing frequency-selective surfaces, which may reduce gain and increase profile height. To address these issues, we implement scattering cancellation. To achieve out-of-band stealth separately for high and low frequencies, we introduce high- and low-frequency resonant structures based on Cell-A. These resonant structures, referred to as Cell-B, are designed in a way that they do not affect the in-band characteristics. Cell-B’s structure, shown in Fig. 2B, integrates the original cross with 4 cross-shaped dumbbells and 4 metal branches. The dumbbells enable high-frequency resonance, while the branches facilitate low-frequency resonance. Adjusting R1a (dumbbell) and pl (branches) independently tunes the high- and low-frequency ranges. Design principles and parameters for Cell-B are detailed in Note S3 and Table S3, respectively. Scattering cancellation requires Cell-B to exhibit 1-bit phase responses across sizes and states. Figure 3F to I shows reflection coefficients and phases of Cell-B under different states. Figure 3J displays in-band electric field distributions for states 1 to 3, revealing minimal field intensity on out-of-band structures during in-band excitation. This weak coupling, further validated in Note S3, enables independent phase control for in-band and out-of-band operations.

Cell-B enables independent reflection phase control in in-band and out-of-band regions. However, introducing out-of-band high-/low-frequency structures increases in-band reflection loss (Fig. 3F and G) and narrows out-of-band bandwidth. To address this, Cell-C (Fig. 2B) enhances out-of-band 1-bit phase bandwidth while reducing in-band loss. Building on Cell-B, Cell-C integrates middle-layer resonant structures: high-frequency cross-shaped dumbbells rotated 45° relative to the top layer, and low-frequency metal branches mirroring the top-layer dimensions. Design principles for Cell-C are detailed in Note S4. Figure 3K to N shows reflection coefficients and phases of Cell-C under different states. Cell-C features dual out-of-band resonances (low/high-frequency) and a single in-band resonance. Adjusting Cell-C’s high-/low-frequency dimensions achieve dual-polarized 1-bit phases across 4 out-of-band bands. Tuning the PIN diode and p2y parameters enables in-band 1-bit operation. The detailed performances are provided in Note S5. Five frequency bands maintain x-/y-polarization phase differences of 180° ± 37° (Table 1). To further enhance the bandwidth of Cell-C, the design methodology is detailed in Notes S6 and S7. This approach involves integrating a dielectric coating onto Cell-C, which substantially improves the 1-bit bandwidth in both the out-of-band and in-band regions, as shown in Figs. S12 and S13. The angular stability in the out-of-band high frequency region is also greatly enhanced, showing minimal variation in bandwidth across different incidence angles. Ultimately, an ultra-wideband characteristic is realized, covering approximately 3.6 to 7.47 GHz (70%), with only a few frequency points slightly deviating from the required 1-bit phase difference. Moreover, despite achieving a nearly continuous ultra-wideband that spans both in-band and out-of-band frequencies, the characteristic frequency ratios fl_c/f₀ ≈ 0.8 and f₀/fh_c ≈ 0.8 are still maintained. In practical applications, a protective cover is also necessary for the metasurface to prevent performance degradation of the PIN diodes and oxidation of the printed circuit board (PCB). Consequently, this protective cover can be implemented as a dielectric coating. This configuration not only safeguards the device but also greatly enhances the bandwidth. This method represents an optimal, concise, and cost-effective strategy for bandwidth extension. Thereby, the meta-atom design becomes more suited to the broad bandwidth requirements.

To elucidate the selection principle of the 4 frequency bands (out-of-band low-frequency and out-of-band high-frequency), a parameter analysis is subsequently conducted. The coupling characteristics between the parasitic structure and the in-band structure are elaborated. For the proposed meta-atom Cell-C, state 1 and state 3 structures are utilized to achieve out-of-band 1-bit phase modulation. Simultaneously, by modulating the PIN diodes in state 1 and state 3 structures, in-band 1-bit phase modulation can be achieved, corresponding to states 2 and 4, respectively. Consequently, when modulating the PIN diodes on the in-band structure, minimal coupling with the out-of-band parasitic structure is desirable. This avoids the deterioration of the out-of-band 1-bit feature. The key parameters influencing the resonant frequencies of out-of-band low-frequency and out-of-band high-frequency are pl and R1a/R2a, respectively. First, the coupling characteristics between out-of-band low-frequency and in-band are analyzed. Taking state 3 as an example, the coupling characteristics between in-band and out-of-band are illustrated, with other states exhibiting similar properties. Figure 4A and B presents the reflection coefficient and phase characteristics of the meta-atom when pl varies, while other parameters remain constant in state 3. It can be observed that when pl varies between 15 and 17 mm, the low-frequency resonant point and the in-band resonant point merge into 2 adjacent resonant points, although the in-band resonant point only experiences a change in reflection amplitude. However, when pl varies between 15 and 17 mm, the in-band reflection phase undergoes a significant change. If the pl value of the out-of-band low-frequency parasitic structure is set to less than 15 and 17 mm to achieve 1-bit phase modulation, it will inevitably cause the out-of-band low-frequency 1-bit to affect the in-band 1-bit characteristic. Conversely, the significant coupling between the out-of-band low-frequency and the in-band structure prevents independent modulation. Notably, when pl varies between 19 and 21 mm, the in-band reflection coefficient and phase remain almost unaffected, with the in-band reflection phases nearly overlapping. Thus, at this point, the coupling between the in-band structure and the out-of-band low-frequency structure is very weak, enabling independent modulation. Consequently, for Cell-C, the pl values corresponding to states 1 and 3 are 19.8 and 19 mm, respectively, to achieve out-of-band low-frequency 1-bit phase modulation. With fl_1c/f₀ = 0.8, a decoupled design between out-of-band low-frequency and in-band is achieved. However, due to the small distance between out-of-band low-frequency and out-of-band high frequency, variations in the out-of-band low-frequency can affect the out-of-band high frequency. Nevertheless, through an integrated out-of-band low- and high-frequency design, considering the coupling characteristics, simultaneous 1-bit phase modulation of out-of-band low and high frequency is achieved. Furthermore, regarding the design principle of the 2 resonant frequencies for out-of-band low frequency, Fig. 4C illustrates the relationship between the 2 low-frequency resonant points with different ratios of the key parameters plu and plm of the top and middle layers’ low-frequency structures, where plm/plu = k. Here, other parameters of state 3 and plu = 19 mm are kept constant. It can be seen that when k = 0.7, there is only one resonant point. When k exceeds 0.7, 2 resonant points emerge in the out-of-band low frequency. For k values of 0.9, 1, and 1.1, fl₂/fl₁ are approximately 0.88, 0.86, and 0.79, respectively. Thus, by selecting different k values, the number of resonant frequency points for out-of-band low frequency can be determined.

Figure 4D and E illustrates the effect of varying the in-band structural parameter p1x on the out-of-band high and low frequencies, while other parameters are held constant in state 3. Here, p1x is the key parameter influencing the in-band structure. As shown in the previous analysis, in state 3, the out-of-band structural parameter pl is 19 mm, and the in-band and out-of-band low-frequency structures are decoupled. From the reflection coefficient and phase, it can also be seen that when p1x changes, the reflection coefficient and phase of the out-of-band low frequency are almost unaffected. It should be noted that when p1x varies from 6.1 to 6.7 mm, it affects the first resonant point of the out-of-band high frequency, with minimal impact on the second resonant point. Moreover, when p1x changes from 6.3 to 6.7 mm, the effect on the second high-frequency resonant point is also negligible. Therefore, by adjusting p1x, the in-band structural changes can be made without affecting the high-frequency out-of-band structure. Figure 4F also shows the effect of p1x variation on the out-of-band reflection phase under state 1. Similar conclusions can be drawn. Thus, when p1x is 6.65 mm in states 1 and 2, the in-band PIN diode modulation has almost no effect on the out-of-band high-frequency characteristics, and /f₀/fh_2c = 0.8. Furthermore, Fig. 4G and H presents the impact of varying the out-of-band high-frequency structural parameter R1a (R2a) on the out-of-band low frequency and in-band, while other parameters are kept constant in state 3. It can be observed that when R1a (R2a) varies from 4.2 to 5 mm, it has a slight effect on both the out-of-band low frequency and the in-band. That is, as R1a (R2a) increases, the out-of-band high-frequency resonant point shifts toward lower frequencies. Similarly, the in-band and out-of-band low-frequency resonant points also shift toward lower frequencies, but at a much slower rate than the high frequency. Therefore, the structural changes required for the out-of-band high frequency to meet the 1-bit demand result in frequency shifts of the in-band and out-of-band resonant points that are almost negligible. Thus, for state 1, R1a and R2a are 4.85 and 4.7 mm, respectively. For state 3, R1a and R2a are both 4.5 mm. At this point, the structural changes required for the high frequency to achieve 1-bit have minimal impact on the in-band resonant characteristics. Additionally, regarding the design principle of the 2 resonant frequencies for the out-of-band high frequency, Fig. 4I illustrates the relationship between the 2 low-frequency resonant points with different ratios of the key parameters R1a and R2a of the top and middle layers’ high-frequency structures, where R2a/R1a = ka. Here, other parameters of state 3 and R1a = 4.5 mm are kept constant. It can be seen that when ka = 0.7, there is only one resonant point. When ka exceeds 0.7, 2 resonant points emerge in the out-of-band high frequency. For ka values of 0.8, 0.9, and 1, fh₂/fh₁ are approximately 0.89, 0.88, and 0.88, respectively. Moreover, although only the parameter analysis for state 3 is provided, similar results are observed for other states. Therefore, through the above parameter analysis, it can be seen that once the in-band structure is determined, the out-of-band high- and low-frequency phase modulation can be achieved via the parasitic structure. Additionally, the condition for independent modulation between in-band and out-of-band structures is approximately fl_1c/f₀ ≈ 0.8 and f₀/fh_2c ≈ 0.8.

To further illustrate the coupling characteristics among out-of-band low-frequency, in-band, and out-of-band high-frequency, the meta-atom electric field distribution is given. Figure 5 presents the electric field distributions corresponding to 5 resonant frequencies across 4 states. Notably, when the low-frequency out-of-band structure is excited, the coupling with the in-band structure is minimal. Similarly, when the in-band structure is excited, the coupling with both out-of-band low and high frequencies is also weak. The electric field distributions align with the conclusions drawn from the parameter analysis. Furthermore, when the out-of-band high-frequency structure is excited, there is some coupling at the f_h2 frequency point with both the out-of-band low frequency and in-band. However, the coupling at the f_h1 frequency point with the in-band structure is relatively small, although the coupling between the f_h2 frequency point and the in-band structure may lead to the out-of-band high frequency affecting the in-band characteristics when implementing 1-bit modulation. However, as established in the above parameter analysis, the structural changes required for the out-of-band high frequency to meet the 1-bit demand result in frequency shifts of the in-band and out-of-band resonant points that are almost negligible. Additionally, the out-of-band low and high frequencies can be designed integrally. Thus, following the above design guidelines, it is possible to achieve a low-profile self-stealth programmable metasurface with in-band and out-of-band RCS reduction characteristics. Furthermore, under this design principle, higher-frequency and lower-frequency parasitic structures can be added to further enhance the out-of-band RCS reduction bandwidth. For instance, higher-frequency resonant structures can be incorporated into the vacant regions surrounding the high-frequency parasitic units. Similarly, lower-frequency resonant structures can be introduced into the vacant areas around the low-frequency parasitic units.

This study achieves in-band and out-of-band stealth while preserving the programmable metasurface’s in-band radiation performance. Cell-C is arranged in a 12 × 12 array to exhibit the radiation characteristics for x-polarization when excited by a horn, as shown in Fig. 1. The out-of-band structures are arranged in a checkerboard topology, as depicted in Fig. 6A. Codes 1 and 2 represent 2 element states with a 1-bit phase difference. The phase distribution of the in-band x-polarized structure switches to different encoding patterns based on different beam steering states. For example, when the beam points to 0°, the in-band encoding pattern is depicted in Fig. 6A. The phase encoding pattern for in-band y-polarization is consistent with the array beam pointing to 0° as well. By controlling the PIN diodes of the 12 × 12 programmable metasurface, the array can achieve 2-dimensional beam scanning in the E-plane and H-plane. For an array beam deflection angle of (θ, φ), each element requires continuous compensatory phase:

where θ and φ represent the elevation and azimuth angles, respectively. k is the propagation constant, d is the period, and φ_f denotes the phase delay from the feed source to the (m, n) element. The phase delay caused by the feed source can be expressed as:where (x, y, z) represents the position of each element. (x₀, y₀, z₀) is the position of the feed source, where x₀ = 0, y₀ = 0, z₀ = F. The focal diameter ratio (FDR) is defined as F/D, where D is the size of the array. After 1-bit quantization, the required compensatory phase for each unit is given by:

The discrete phases are encoded as codes 0 and 1, respectively. FDR is 0.75. Figure 6B and C presents the beam scanning results in the E-plane and H-plane, respectively, at 5.2 GHz. It is evident that the programmable metasurface can achieve good 2-dimensional ±45° beam scanning with a scanning step size of 15°. The coding pattern for beam pointing at 15°, 30°, and 45° is shown in Fig. S4. The average sidelobe is below −10 dB, and the cross-polarization is below −30 dB. Figure 6D illustrates the gain and efficiency at different frequencies. It is clearly observed that the 1-dB gain bandwidth is 5.15 to 5.35 GHz. The maximum gain is 17.96 dBi, and the aperture efficiency is 21.25%. Therefore, the programmable metasurface exhibits excellent beam scanning characteristics and high efficiency.

As previously noted, the programmable metasurface demonstrates excellent x-polarization radiation characteristics within the in-band. This section examines its in-band stealth performance during beam scanning. For a 2-dimensional metasurface, the monostatic RCSR can be approximated analytically as follows [52]:where {\mathrm{\Gamma }}_{\mathit{mn}} and {\phi }_{\mathit{mn}} are the reflection coefficient and phase of the mnth element, respectively. Equation (1) neglects the coupling between the elements and edge effects, but it provides a good guideline to predict the RCS reduction characteristics of the metasurface.

The phase required for beam scanning of the 1-bit programmable metasurface with horn excitation is a combination of code 0 and code 1 elements. For example, the coding pattern of the broadside is shown in Fig. 6E. It is evident that the phase distribution is composed of multiple code 0 and code 1 elements. This indicates that the phase distribution of the array exhibits scattering cancellation for the incident plane wave during beam scanning [53,54]. Based on Eq. (4), the coding patterns of the 1-bit programmable metasurface are calculated for different deflection angles, and their RCS reduction characteristics under these coding patterns are analyzed. The theoretical calculations yield the RCS reduction results under x-polarized and y-polarized plane wave incidence, as shown in Fig. 6F and G. It is evident that the −10-dB RCS reduction bandwidth is 5 to 5.27 GHz for x-polarization and 5.05 to 5.3 GHz for y-polarization under different deflection angles. This indicates that the 1-bit programmable metasurface with horn excitation achieves scattering cancellation characteristics for plane wave incidence and thus achieves stealth functionality. Figure 6F and G also demonstrates the RCS reduction characteristics of the array under the chessboard topology. It can be observed that the RCS reduction characteristics of the array under the chessboard topology are nearly identical to the RCS reduction bandwidth and magnitude achieved by the phase topology with horn excitation for beam deflection. Therefore, while maintaining the 1-bit programmable metasurface’s in-band radiation functionality, it also possesses in-band stealth capabilities. Figure 6H presents simulation results of RCS reduction using coding at 0° and 30°. For transverse magnetic (TM)-polarized plane wave incidence, the −10-dB RCS reduction bandwidth is 5.05 to 5.33 GHz. The RCS reduction bandwidths under the 0° and 30° coding patterns are nearly the same, with only slight differences in reduction magnitude. For transverse electric (TE)-polarized plane wave incidence, the −10-dB RCS reduction bandwidth is 5.12 to 5.32 GHz, and the RCS reduction bandwidths and magnitudes under the 0° and 30° coding patterns are almost identical. Figure 6H also illustrates the RCS characteristics under chessboard coding. It can be observed that the bandwidth of y-polarized RCSR is greatly improved under chessboard coding, while the effect of x-polarization is weaker. Specifically, under chessboard coding, the bandwidth of y-polarization RCSR at −10 dB is 5.105 to 5.35 GHz, while that of x-polarization is 5.017 to 5.327 GHz. It can be observed that the −10-dB RCSR bandwidth is almost the same in the aforementioned coding schemes. Furthermore, considering that the subsequent y-polarization at 0° is also used as a communication transmission mode, the y-polarization is selected for 0° coding. Thus, the coding scheme for the y-polarization component is unique post-fabrication. Therefore, while maintaining the in-band beam scanning characteristics, the programmable metasurface also exhibits RCS reduction characteristics similar to the array’s chessboard topology under different radiation modes.

In the previous section, we discussed the stealth characteristics of the programmable metasurface within the in-band. In this section, we will explore its stealth capabilities outside the operating band. To achieve monostatic RCS reduction beyond the operating band, we arrange the proposed Cell-C in a chessboard pattern to achieve scattering field cancellation. We employ 2 chessboard arrangements as shown in Fig. 6I. In these arrangements, code 1 corresponds to state 1 (or state 2) in Fig. 2G, while code 2 corresponds to state 3 (or state 4) in Fig. 2G. Due to the good isolation between the in-band and out-of-band frequencies, using states 1 and 2 as code 1 yields the same effect, while using states 3 and 4 as code 2 also produces the same effect.

Such an arrangement proves to be highly effective in achieving RCS reduction in the out-of-band. Figure 6J and K illustrates the RCS reduction characteristics under the 2 array topologies. It is evident that by arranging the meta-atoms in a chessboard pattern, 4 RCS reduction bands can be achieved in the out-of-band. Moreover, the RCS reduction characteristics are nearly identical for both TE- and TM-polarized plane wave incidence, although there is a slight difference in RCS reduction bandwidth and magnitude between array A and array B. Figure 6L presents the out-of-band −10-dB RCS reduction bands for array B. Therefore, the proposed programmable metasurface demonstrates 4-frequency band RCS reduction characteristics in the out-of-band, in addition to its RCS reduction characteristics in the in-band. Next, we will fabricate the programmable metasurface using the topology of array B to validate its full-band stealth capabilities both in- and out-of-band.

To validate the proposed self-stealth programmable metasurface, we fabricated a 12 × 12 programmable metasurface. The layout of the actual processed array is shown in Fig. S15. The top and bottom layers of the processed 12 × 12 array after soldering are shown in Fig. S16. In a microwave anechoic chamber, we conducted tests on the metasurface’s far-field radiation characteristics and monostatic RCS, as illustrated in Fig. 7A. For the far-field radiation characteristics test, we used a transmitter horn to stimulate the metasurface at a position with an F/D ratio of 0.75. The metasurface’s coding sequence was controlled by an FPGA. In the far-field, we employed a receiver horn to capture the radiated field of the metasurface, thus testing its radiation characteristics. In the RCS test, a transmitter and receiver horn were positioned in the far field to measure the monostatic RCS.

Figure 7B confirms that the proposed metasurface exhibits a high isolation between the co- and cross-polarization. Figure 7C and D displays the beam scanning results in the E-plane and H-plane at 5.2 GHz. It is evident that the metasurface achieves 2-dimensional ±45° beam scanning. The average sidelobe level is below −10 dB. Figure 7E illustrates the gain and efficiency at different frequencies. It can be observed that the 1-dB gain bandwidth is 5.11 to 5.36 GHz, with a maximum gain of 17.23 dBi and an aperture efficiency of 18%. The simulation and measured results are consistent. Therefore, the proposed metasurface demonstrates excellent beam scanning characteristics and efficiency.

During the monostatic RCS test, the programmable metasurface coding is depicted in Fig. 6A, which includes in-band x- and y-polarizations, as well as an out-of-band coding topology. The measured results demonstrate the RCS reduction characteristics of the metasurface in both the in-band and out-of-band, as shown in Fig. 7F to H. It can be observed that the metasurface exhibits RCS reduction in 4 frequency bands in the out-of-band. The first low-frequency band with a −10-dB RCS reduction extends from 3.57 to 3.82 GHz, and the second low-frequency band from 4.07 to 4.3 GHz. The first high-frequency band is from 6.56 to 7 GHz, and the second high-frequency band is from 7.3 to 7.87 GHz. In the in-band, the −6-dB RCS reduction bandwidth for x-polarization ranges from 4.9 to 5.3 GHz, while for y-polarization, it spans from 5 to 5.38 GHz. The detailed RCSR measured bandwidth is shown in Table 2. The measured in-band RCS reduction was only −6 dB, with a more severe degradation observed for x-polarization. The discrepancies described above are primarily attributable to the following factors. First, the implementation for x-polarization required the soldering of PIN diodes and choke inductors. Consequently, this introduced greater uncontrolled errors compared to other polarizations and frequency bands. Thus, soldering inaccuracies for the PIN diodes and choke inductors resulted in a frequency shift of the 1-bit phase bandwidth for x-polarization. This led to an in-band RCS reduction for x-polarization of less than −6 dB. Second, a deviation in the dielectric constant of the fabricated PCB substrate from the design value caused the RCS reduction across all frequency bands to be slightly inferior to the simulation results. This is because the dielectric constant error altered the 1-bit phase bandwidth. Third, errors were introduced by the measurement environment. This is a common source of uncertainty inherent to all experimental disciplines. Figure S17 presents the measured RCSR performance of the metasurface under oblique plane-wave incidence. Simulated in-band radiation and RCSR performance of metasurface with the coating for oblique incidence are shown in Figs. S18 and S19. Therefore, the proposed programmable metasurface exhibits monostatic RCS reduction characteristics in both the in-band and out-of-band. Moreover, the transient processes and nonlinear effects of PIN diodes, as well as the spectral leakage under rapid beam switching, are crucial for evaluating the performance of metasurfaces in stealth applications. We have conducted corresponding analyses and incorporated relevant additions and clarifications in Note S13.

We have added Table 3, which provides a comparison across various aspects including reconfigurability, stealth methodology, requirement for an additional coating, number of RF components, in-band stealth, out-of-band stealth, stealth bandwidth, profile height, and in-band radiation characteristics. It can be observed that the proposed programmable metasurface achieves dual-polarized stealth for both in-band and out-of-band frequencies while exhibiting the lowest profile height. The number of required RF components is also relatively small. Furthermore, it maintains high radiation efficiency within the operating band. Additionally, existing programmable metasurfaces [18,26,48] primarily focus on in-band RCS reduction, whereas only Ref. [50] and the present work achieve out-of-band RCS reduction. However, compared to designs with higher profile heights, such as those in Refs. [50,53], the achievable RCS reduction bandwidth still requires further improvement. Therefore, Table 3 also includes the performance characteristics of the proposed metasurface when integrated with a dielectric coating. It is evident that incorporating the dielectric coating greatly enhances both the in-band and out-of-band RCS reduction bandwidths. Moreover, the in-band and out-of-band frequencies merge to form an ultra-wideband, ultimately resulting in ultra-wideband RCS reduction performance. The overall profile height remains low at merely 0.14λ.