1. Introduction

The control of exotic electromagnetic properties has drawn significant interest to cloaks and has been extensively studied in electromagnetics. A cloak is a device that can manipulate the incident electromagnetic waves to achieve the goal of hiding objects [1]. Recently, transformation optics took advantage of the fundamental connection between the spatial material properties of the surrounding medium and a suitable coordinate space transformation to realize interesting optical phenomena and devices [2–4], for example, illusion optics [5], light steering [6–8], light source control [9], meta-lenses [10], and so on [11,12]. In addition to transformation optics, scattering cancellation [13–16], which suppresses the dominant multipolar scattering orders, has proved to be another powerful method to build cloaking devices. Hence, cloaks have been experimentally investigated from the microwave to the optical frequencies. However, transformation optics requires specific material permittivity and permeability profiles of inhomogeneity and anisotropy, which in many cases are hard to realize in practice, whereas scattering cancellation cloaks perform poorly if the detector is sensitive enough.

Recently, attention has been paid to metasurfaces, an optically thin layer consisting of subwavelength elements that can engineer the amplitude, phase, and polarization of the incident radiation in a broad electromagnetic spectrum [17–19]. A novel cloak design approach based on metasurfaces, called local phase compensation, has been proposed and experimentally verified [20,21]. The discrete phase strategy adapted in local phase compensation greatly simplifies the fabrication of the cloaking devices, and the ultra-thin thickness of the metasurface extends the potential range of applications. A remarkable ultrathin skin cloak for visible light was first realized by this method [22], followed by similar theoretical and experimental cloaking investigations at microwave frequencies [23–25]. However, terahertz carpet cloaks based on local phase compensation remain unexplored, which does not match the urgent needs for terahertz devices, especially, the cloaking devices.

Here, we experimentally demonstrate an ultrathin metasurface cloak to eliminate the wavefront disturbances coming from a triangular bump with a bottom length of 19 mm (47.5 λ) and a center height of 3.64 mm (9.1 λ) at 0.75 THz. A carpet cloak working at large incident angles is achieved by periodically arranging 12 different unit cells in the terahertz region. Good cloaking performance is observed from numerical results in both the near and far fields. The scanning results of the far field agree well with the simulation, which further verifies our design. Asymmetric and symmetric unit cells are also used to establish polarization-sensitive and polarization-insensitive cloaks, which may be of use to improve terahertz cloak designs, as well as the related radar imaging and RCS detection [26] performances against the cloaking devices.

2. Design and Numerical Results

Most carpet cloaks to date create a triangle cloaking area with dimensions of several wavelengths with metamaterials [27,28]. Due to limited cloaking area, these designs are hard to generalize in real-world applications. However, another feasible implementation is reconstructing the wavefront of the reflected wave by a metasurface with a gradient phase, which makes large-dimension cloaks possible. Figure 1 shows a schematic of the metasurface-based carpet cloak, where the incident wave with an angle of incidence θ illuminates a metal bump with a base angle of φ. The restoration of the reflected wavefront is achieved through eliminating the distortions caused by any object on the bare bump. The inset in the figure gives a view of the unit cell. All metasurfaces are designed with a sandwiched structure with two aluminum layers of 200 nm thickness separated by a 30 μm-thick polyimide (PI) spacer. The upper aluminum layer consists of a patch array with varied width a and length b. The compensating phase distribution along the object surface should meet the following phase $\varphi$ relationship:

 

Fig. 1 Schematic of the metasurface-based carpet cloak and unit cell (inset).

Download Full Size | PDF

When the period of the unit cell p and the thickness d are varied, the phase distribution changes as:

The initial phase at the edge is ${\varphi }_0=\pi \text{-}2k_0[(p/2)\sin \phi +d\cos \phi ]\cos \theta$and the integer n becomes larger as the height is increased. However, the phase variation between the adjacent unit cells is π/3. Noticeably, the largest obstacle of local phase compensation is how to maintain the continuation of phase variation at the edges, which will effectively avoid the split of the reconstructed wavefront. Keeping this in mind, we will apply the discrete phase method to design the cloak.

The initial phase (closest to the ground) is set to be 2π/3 and the phase variation between the unit cells is set to π/3, which is consistent with the phase variation between the ground and the first cell. Then the phase varies smoothly from the ground to the cloaking area, which meets the requirement of rebuilding the wavefront. When the period p and thickness d are set to 138 µm and 30 µm, respectively, the initial phase will be 2π/3 at the working frequency f = 0.75 THz and the phase variation is simplified to π/3. In our work, the 2π phase variation of the metasurface layer is realized by a six-order discretization, whose periodic feature may allow to enlarge the cloak size at will. The metasurfaces covered on the two sides of the bare bump have incident angles of $\theta -\phi$ and $\theta +\phi$, respectively. Here, the incident angles are, respectively, 25° and 65° at $\theta =45°$ and $\phi =20°$. To obtain the required amplitude and phase distributions, we have simulated the responses of the unit cell under various widths and lengths using a commercially available CST MICROWAVE STUDIO software. Figures 2(a) and 2(b) show, respectively, the amplitude and phase responses of the unit cell as a function of width a and length b under an incident angle of 25°, while Figs. 2(c) and 2(d) represent the corresponding results at an incident angle of 65°.

 

Fig. 2 Responses of the unit cell under different incidence angles. (a) and (b) show, respectively, the amplitude and phase of the unit cell when the incoming wave is at 25° for 0.75 THz. (c) and (d) illustrate the corresponding results at 65°. The white crosses represent the amplitude and phase response of selected unit cells for incident angles of 25° and 65°, respectively.

Download Full Size | PDF

To achieve the phase distribution calculated from Eq. (2), we choose 12 discrete unit cells with a large amplitude response and a π/3 phase increment to cover the 0-to-2π phase to provide full phase control of the wavefront. In Figs. 2(a) and 2(b), the white crossings represent the responses of the selected unit cells for the incident angle of 25°. Similarly, the corresponding amplitude and phase responses for the 65° angle are indicated in Figs. 2(c) and 2(d). It can be inferred that the chosen unit cells meet the requirement of a 2π phase variation in addition to a uniform amplitude. The uniform response of the elements may influence the performance of the cloak. Another design based on a symmetric patch will be polarization insensitive.

To validate the design, a numerical simulation of the metasurface cloak is first carried out. Two metasurface layers designed at the two incident angles are covered at the corresponding sides of the triangular bump. Here, a perfect conducting bump with a triangular shape in the x-z plane and infinite long in the y direction is assumed, and we numerically check the effectiveness of the metasurface design by comparing the wavefronts reflected from the ground plane, bare bump and bump covered with the metasurface cloak. With the time consumption taken into account, the size of the triangular bump has a base angle φ = 20° and an edge length of 2.484 mm (6.21 λ). To be consistent with the practical situation, the amplitude distribution of the incident wave has a Gaussian distribution. Meanwhile, the periodic boundary condition is applied in the y direction while the x direction is set to be open in the simulation.

Figure 3 shows the simulation results for different cases. It can be seen that the far-field radiation patterns of the ground plane and bare dump are obviously different, as shown in Figs. 3(a) and 3(d), respectively. The related electric field distributions in Figs. 3(b) and 3(e) indicate the change in the reflected wavefronts at 0.75 THz. To explore the influence of the undistorted and distorted wavefronts in detail, the far-field intensity distributions are plotted in Figs. 3(c) and 3(f). These results reveal that the splitting of the near field leads to the difference in the far-field patterns. Meanwhile, there are three lobes consisting of two side lobes and a center lobe in the results of the bare dump shown in Figs. 3(e) and 3(f), which are due to the reflections from the two edges of the triangular bump and the aluminum ground, respectively. It can be found that the split in the center lobe is reasoned to be due to the triangle apex. After covered with the metasurface layers, the situation changes greatly, as illustrated in Figs. 3(g) and 3(h). The wavefronts are rebuilt well in Figs. 3(h) and 3(k) using both a six-step polarization-sensitive layer and a six-step symmetric metasurface layer at 0.75 THz. Compared with Fig. 3(c), for an undistorted far-field intensity distribution, Figs. 3(i) and 3(l) verify the design. However, there is still a small deviation of the reflected wave which mainly comes from two aspects: (i) the large unit cell size (138 µm ≈λ/3) in the cloaking layers will introduce the higher order diffraction, which may affect the restoring of the reflected wavefront; (ii) the contribution from the apex of the triangle. Moreover, as shown in Figs. 2(c) and 2(d), the two edges of the bump have different reflectivity, which may be an obstacle for rebuilding the reflected beam. Despite this fact, the field distributions after the metasurface cloak and the far-field radiation of the ground verify that the design based on a six-step gradient phase can compensate for the reflection phase and rebuild the reflected wave well.

 

Fig. 3 Simulation results. (a), (d), (g) and (f) Far-field radiation pattern of ground plane, bare bump, bump covered with six-step polarization-sensitive and bump covered with six-step polarization-insensitive layer from 0.65 THz to 0.85 THz, respectively, under the y-polarization incident wave. (b), (e), (h) and (k) Corresponding electric field distributions at 0.75 THz. The red lines mark the place of 0.75 THz. (c), (f), (i) and (l) Far-field intensity distributions corresponding to (b), (e), (h) and (k), respectively.

Download Full Size | PDF

3. Experiment and Analysis

First, a triangular surface bump with a bottom length of 19 mm (47.5 λ), a center height of 3.64 mm (9.1 λ) and a base angle about 20.96° was made. Then, the cloaking layers were fabricated with standard photolithography process. Low loss flexible polyimide (PI) with a permittivity of ε = 2.898 + 0.14i (measured by experiment) was used as the spacer medium. The fabrication process was as follows: (1) A 10 µm PI layer was spin coated on a silicon substrate, followed by a four-step baking process to make it stable: 80 °C for 10 minutes, 120 °C for 10 minutes, 180 °C for 15 minutes and 250 °C for 15 minutes. This process was repeated three times to meet the desired thickness; (2) On the PI layer, the 200-nm-thick Al patches were patterned were by standard lithography and thermal evaporation; (3) The PI layers with the aluminum patterns, as well as the reference, were sliced into designed sizes and stripped off from the substrate.

We used an angular scanning terahertz time-domain spectroscopy system to characterize the metasurface cloak. The experimental setup is illustrated in Fig. 4(a). The terahertz transmitter and detector were both photoconductive antennas (Menlo Systems) with a default horizontal polarization and relatively high polarization degree. Before illuminating the samples, the generated terahertz beam was collimated by lens L1. To enable an angular scan, the detector and terahertz lens L2 were placed on a rotator to collect the reflected terahertz wave from 0° to 90°. In order to improve the angular resolution, a 6 mm aperture was employed in front of the detector. The distance between the aperture and the center of the holder was 17.7 cm. Figure 4(b) shows the bump covered with six-step asymmetric cloaking layers (right) as well as the bare bump (left).

 

Fig. 4 Measurement system and carpet cloak models. (a) Experimental system based on angular scanning terahertz time-domain spectroscopy is used for terahertz cloak demonstration. (b) The bare bump (left) and the bump covered with six-step asymmetric cloaking layers (right).

Download Full Size | PDF

Figure 5 illustrates the scanned results of the ground plane, the bare bump, the bump covered with asymmetric and symmetric cloaking layers, respectively. Here, the far-field patterns for incident terahertz wave along the x and y directions were both measured. As can be seen, the experimental results agree well with the simulation. For convenience of comparison, we used the electric field intensity of the ground plane to normalize the experimental results under the same polarization. Figures 5(a) to 5(h) are scanned to check the effectiveness of the system under two linear polarizations, which perfectly match the simulation. It should be noted that in Figs. 5(f) and 5(h), the scanning accuracy was not high enough to distinguish between the split around 45°. Nevertheless, a comparison of the experimental results of the bare dump and the dump covered with the asymmetric cloaking layers shows that this cloak did not work when illuminated by the x-polarized incidence and the radiation pattern was in disorder, as exhibited in Figs. 5(i) and 5(j). When the y-polarized incident wave was applied, the bump covered with asymmetric cloaking layers was concealed well as the two side lobes were suppressed and the maximum of the reflected terahertz wave was distributed at 45°, see Fig. 5(l). The field intensity near 45° was at least 10 dB larger than that for the bare bump in Fig. 5(h). It indicates that the metasurface layers effectively transformed the energy to the center, as illustrated in Fig. 5(l). The efficiency compared with the incident light was about 44.2%, which is due to absorption caused by the polyimide. More importantly, it can be inferred from Fig. 5(k) that the cloak worked well from 0.73 to 0.8 THz. As a result of the polarization-sensitive feature, we can easily turn the cloak on and off by switching the polarization. From another aspect, it is also meaningful to realize a cloak for all polarizations. As shown in Figs. 5(m) to 5(p), the symmetric cloaking layers worked well under two polarizations, which suggests that the symmetric cloaking layers are polarization insensitive over the same working bandwidth as the asymmetric cloaking layers. Although the side lobes at the angles of 5° and 85° were reduced a lot by the metasurface, still observable lobes remained due to the fact that the measured efficiency of the symmetric metasurface layer was about 43.36% (compared with the incident energy). The small difference between Figs. 5(e) and 5(g) might be caused by misalignment. In short, the experimental results well proved the effectiveness of our design in concealing the object.

 

Fig. 5 Experimental results. (a), (c), (e), (g), (i), (k), (m) and (o) Far field radiation patterns of the ground plane, bare bump, bump covered with six-step asymmetric cloaking layers, and bump covered with six-step symmetric cloaking layers, under the x and y polarized incidences, respectively. The red lines mark the place of 0.75 THz. (b), (d), (f), (h), (j), (l), (n) and (p) Corresponding results of (a), (c), (e), (g), (i), (k), (m) and (o) at working frequency, respectively.

Download Full Size | PDF

4. Conclusion

A large-dimension carpet cloak (216.125 λ²) has been numerically and experimentally demonstrated by adopting local phase compensation and ultrathin metasurface layers at 0.75 THz. The cloak is constructed by periodically arranging 12 different elements to build two metasurface layers, which are both 10.18 mm (25.45 λ). The principles can be extended to build much larger cloaks without extra design. Moreover, polarization-selective and polarization-insensitive cloaks are also investigated, showing a narrow working bandwidth. Additionally, the measured efficiency of the metasurface layers is about 44.2%. Our metasurface layers based on polyimide are promising in building freestanding and flexible cloaking devices, paving the way for practical applications. These proposed cloak designs have promising applications in radar, illusion optics and antenna systems as well as in information encryption.