1. Introduction

Terahertz (THz) waves have attracted considerable interest because of their potential value in fields including security-based detection [1], chemistry [2], and biology [3]. However, because of the lack of naturally occurring materials with strong electronic and magnetic responses at the relevant frequencies, the manipulation of THz waves has proved to be extremely challenging, making their application more difficult [4]. Metamaterials, which have artificial periodicity, represent one potential solution to overcome this “THz gap” [5].

In recent years, electromagnetic absorption has been employed a variety of optoelectronic applications in fields such as stealth technology [6,7], thermal emitters [8,9], spatial light modulation [10,11], sensors [12], and thermal imaging [13]. Through regulated interactions with THz waves, metamaterial absorbers can produce unity absorption coefficients. Since the first experimental presentation and demonstration of such an absorber in 2008 by Landy and colleagues [14], a diverse range of patterned metal array shapes, including ribbons [15], discs [16], rings [17,18], and cross-shaped structures [19–21], have been proposed, which have an absorption efficiency of nearly 100%. In addition, various theories have been proposed to interpret the absorption mechanism of such devices, including the impedance matching theory [22], interference theory [23], and transmission line theory [24]. Further, development studies have led to gradual improvement in the properties of these metamaterial absorbers.

However, certain shortcomings still exist. In the existing studies, in the processing of the absorbers, their characteristics were fixed, thus making these absorbers inapplicable in certain domains, although some improvements allowed the variation of the operating frequencies and absorption amplitudes of THz absorbers. Conventionally, metamaterial responses have been controlled by introducing active materials into either the absorber geometry or the surrounding medium, wherein the properties of these materials can be varied by using an external stimulus such as an optical [25,26], thermal [27], or electromagnetic field [28]. One of the most widely used media is graphene, which has attracted considerable attention for its tunable absorption characteristics, which are a result of the tunability of the carrier mobility and conductivity of graphite [29,30]. Another efficient approach to achieve a tunable metamaterial absorber is by realizing structural reconfiguration based on microelectromechanical systems (MEMS) [31–33]. These systems permit the design of patterns that result in dynamic control of the absorption characteristics, which is a proven and effective technique to achieve the required tunable characteristics. However, compared with regular metamaterial absorbers, the fabrication process of reconfigured absorbers is inevitably more complex, and a more convenient method is desired.

With the ultimate goal of achieving tunable properties, one potential approach involves overcoming the existing limitations through the use of anisotropic metamaterials. Previous studies have found that anisotropic absorbers are sensitive to the polarization direction of the incident waves, and they also exhibit different absorption characteristics for transverse electric (TE) and transverse magnetic (TM) polarized incident light [34,35].

This paper proposes an anisotropic metamaterial that consists of three layers of gold and dielectric plates. The metamaterial provides a polarization-sensitive response to incident THz waves by breaking the C4 symmetry. An H-shaped resonator—a typical resonator shape—was employed. When configured using a suitable set of resonance units of different sizes and various operating frequencies, the H-shaped resonance units were integrated into a single absorption unit. The absorption frequency was then varied by rotating the device. The objective of this work was to analyze the effect of the polarization direction on the absorption characteristics of the device, and to provide a more concise approach to realize tunable absorption characteristics.

2. Design and fabrication

In this work, for the polarization sensitive tunable absorber, an asymmetric absorber cell design that integrates four pairs of H-shaped patterns with different lengths (l₁–l₄) was selected. To clearly illustrate the absorber structure, the plan and side elevation are shown schematically in Fig. 1(b). The dimensions of the H-shaped pattern were set as follows: l₁ = 8 μm, l₂ = 12 μm, l₃ = 16 μm, and l₄ = 20 μm. The width of each H-shaped pattern was the same (S₂ = 22 μm); the thickness of each pattern and that of the ground plane were the same (T_m = 200 nm); the distance between each pattern of the same pair, S₁, was 22 μm; and the line width was 4 μm. All the patterns were arranged neatly on a polyimide layer with a thickness of T_s = 8 μm, and the lattice constant was P = 150 μm. The size of the structure has been optimized to avoid undesired coupling [36,37].

 

Fig. 1 (a) Schematic of structure of polarization-sensitive tunable metamaterial absorber. (b) Geometric parameters of the unit cell. (c) Photograph of the metamaterial absorber. (d) Optical microscopy image of the absorber.

Download Full Size | PDF

To realize polarization control of the absorption characteristics, the polarization-sensitive tunable absorber was fabricated on a 500-μm-thick silicon substrate. First, an unpatterned ground plane was deposited on the silicon substrate by using the electron beam evaporation technique. The ground plane was composed of a 30-nm-thick Cr adhesion layer and a 200-nm-thick Au layer. Subsequently, a polyimide dielectric layer (~8 μm thick) was added on top of the Au film to form a spacer layer by using the spin-coating method, and the unit was soft cured in a vacuum oven with a nitrogen protective atmosphere at 200°C. The complex permittivity of polyimide was 2.7 + 0.007i. Finally, the metallization patterns (comprising a 30-nm-thick Cr adhesion layer and a 200-nm-thick Au layer) were fabricated using standard positive lithography, electron beam evaporation method, and ion beam etching techniques.

Figure 1(c) shows a photograph of the metamaterial absorber. The total size of the sample was 10 mm $\times$ 10 mm. Figure 1(d) shows a microscopy image of the fabricated anisotropic metamaterial structure. The reflection spectra were measured using reflection-mode THz time-domain spectroscopy with a pair of emitting/detecting low-temperature-GaAs photoconductive antennas. The absorption spectrum was obtained using fast Fourier transform reflection time-domain signals. For optimization and comparison, computer simulations were carried out based on the finite integration method by using the commercial program CST Microwave Studio 2018.

3. Results and discussion

The absorption spectra and cell structure of the polarization-sensitive tunable metamaterials are shown in Fig. 2. The simulation results were obtained using frequency domain solver (in CST Microwave Studio). The polyimide is model as lossy polymer with a permittivity of 2.7 + 0.007i, and gold structure as lossy metal with an electrical conductivity of 4.561 × 10⁷ S/m is chosen. Floquet boundary condition is employed, and the normal incidence is used. The absorption A was obtained using A = 1 − |T| − |R|, where T and R are the transmittance and reflectance, respectively. T was close to zero because the Au film thickness on the bottom is considerably larger than the skin depth. The angle between the E-field direction and the positive direction of the y-axis was θ. To illustrate the absorption properties of the designed absorber, the absorption spectra for four typical values of θ (0°, 45°, 90°, 135°) are shown in Figs. 2(a)–2(d). The major absorption peak is noted to be frequency shifted as the E-field rotates counterclockwise. As indicated by the dotted arrow shown in the inset of Fig. 2, the resonance pairs that are oriented parallel to the E-field are the main contributors to the major absorption peaks. Because of their different sizes, the four resonator pairs produce absorption modes with different absorption frequencies. For ease of description, the resonance pair that contributes most strongly to an absorption peak is called the major pair. Further, Figs. 2(a)–2(d) show that the designed absorber has three discrete absorption peaks. However, because of the asymmetric distribution of the resonators and the anisotropic absorption characteristics, the operating frequencies, especially that of the major peak, can be tuned by rotating the absorber. Moreover, it should be noted that the proposed structure can only excite three of the four modes simultaneously. The absorption mode for which the resonator is oriented perpendicular to the electric field is inhibited because no effective excitation is provided by the electric field. The associated absorption peaks result from the interactions between the E-field component and the resonance pairs on both sides of the major pair. The amplitude of the associated absorption is noted to decrease slightly. Additionally, Fig. 2 also shows that with increase in the length of the H-type resonator (l₁–l₄), the absorption frequency decreases in accordance. The absorption frequency is solely controlled by the length of the H-shaped resonator, whereas the amplitude is related to the value of θ. When θ changes continuously, the frequencies of the absorption peaks are modulated by these four frequencies. The amplitude of the absorption of the H-shaped asymmetric unit cell structure is tunable over the 0–0.95 range.

 

Fig. 2 Simulated absorption spectra at (a) 0°, (b) 45°, (c) 90°, and (d) 135°. Inset: Structure of the unit cell.

Download Full Size | PDF

To gain an improved understanding of the absorption mechanism, we also derived the simulated electric field distributions for the four major absorption modes, as illustrated in Fig. 3. The field distribution for each mode was different after the electric field was rotated. A strong electric field was established around the H-shaped resonator pair with length l₄ = 20 μm at 1.60 THz when the polarization angle θ was 0°. Similarly, the E-field was concentrated around the resonator pairs with lengths l₁, l₂, and l₃ at frequencies of 1.71 THz, 1.86 THz, and 2.04 THz, respectively. With increasing θ, the E-field rotated counterclockwise and was concentrated around the resonance pair oriented parallel to its field direction. Consequently, the resonance frequency of the major mode gradually increased with decrease in the length of the H-shaped resonator. Furthermore, as shown in Figs. 2(a)–2(d), the rotation of the sample affected only the amplitude of the absorption, and it had only a minimal impact on the operating frequency.

 

Fig. 3 Simulated electric field distributions of the major modes at 0°, 45°, 90°, and 135°.

Download Full Size | PDF

In the performed experiments, the polarization angle θ was defined as being zero when the incident wave was polarized in the y-direction. The THz wave impinged on the sample at an angle of incidence of 30°. All measurements were performed in a nitrogen atmosphere to avoid the absorption of water vapor. Figure 4(a) shows the experimentally measured absorption spectra of the isotropic metamaterial absorber at four different polarization angles θ. These spectra were noted to be highly sensitive to the polarization angle, thus indicating the notable frequency shift that can occur in the proposed anisotropic metamaterial absorber, which demonstrates the realization of the tunability of the absorber via sample rotation. However, the most remarkable fact was that the four absorption peaks appeared simultaneously, and the absorption amplitudes were slightly different than those in the simulations. In theory, one of the four absorption peaks should be inactive at this polarization angle. To explain this phenomenon, several simulation analyses were carried out. For ease of description, the absorption modes are labeled f₁, f₂, f₃, and f₄ according to the fundamental frequency values of the modes. Figure 4(b) shows the simulated absorption spectra of the anisotropic metamaterial absorber at ten different polarization angles θ, ranging from 0° and 45° in increments of 5°. When the incident polarization was parallel to the y-axis or had a 0° orientation, three absorption peaks occurred at 1.60 THz, 1.84 THz, and 2.04 THz (i.e., f₁, f₂, and f₄). As θ increased, the absorption gradually decreased at both f₁ and f₂, and f₂ disappeared at 45°; however, the resonance at f₃ appeared, and f₄ gradually dominated. In other words, the proposed structure demonstrated four absorption peaks simultaneously when θ was not an integer multiple of 45°. These simulation results indicate the reasons for the unexpected results obtained in the experiments. Because of the uncertainties in the rotation angle θ during the experiments, four absorption peaks appeared, generating a difference in the absorption amplitude. For example, for the measured 90° absorption curve, the absorption at f₄ was expected to be lower than the amplitude that was obtained in the experiment. This aspect indicates that the real rotation angle θ, which lies between 45° and 90° in the experiment, leads to the amplitude variation.

 

Fig. 4 (a) Measured absorption spectra at θ = 0°, 45°, 90°, and 135°. (b) Simulated results at various polarization angles between 0° and 45°.

Download Full Size | PDF

As discussed above, the absorption amplitudes of the resonator pairs are related to the polarization angle θ. For further validation of the correlation between the amplitude and the polarization angle, a structure with only one pair of H-shaped resonators was analyzed (Fig. 5), with the H-shaped resonator pair rotated counterclockwise from 0° to 90°. The black squares represent the normalized absorption rate values obtained from the simulations. As the angle of polarization increased, a monotonic reduction in the absorption occurred because the incident electric field no longer excited the H-shaped resonator pair efficiently owing to the decrease in the E-field component oriented parallel to the resonator pair. The interpolation polynomial was applied for curve fitting to provide a better visualization of the relationship between the absorption and the polarization angle. The results of the simulations indicated that the relationship between the absorption amplitude and cos²θ is quadratic.

 

Fig. 5 Relationship between absorption and polarization angle.

Download Full Size | PDF

The simulated absorption spectra of the polarization tunable metamaterial absorbers with different values of l₄ under excitation at normal incidence are shown in Fig. 6(a), where the incident electric field is along the y-axis. The absorption resonance frequency can be noted to be red-shifted with increase in l₄. Furthermore, when l₄ is approximately the length of the resonance pairs on both sides of the major pair, the interactions of the different modes produced by the different resonance pairs can be observed. Therefore, resonance pairs with adjacent lengths were arranged vertically to avoid this coupling.

 

Fig. 6 (a) Simulated results for various values of l₄. (b) Elementary cell of the extended metamaterial absorber and simulated absorption spectra at 0°, 45°, 90°, and 135°.

Download Full Size | PDF

The proposed concept can be readily adaptad to allow variation in the tunable frequency range of an absorber. As an example, we describe one specific structure. The unit cell of the absorber is shown in Fig. 6(b); the lengths of the H-shaped resonators were set as 20, 24, 28, and 32 μm. In addition, the periodicity of the absorber and the distance between each pattern of the same pair were changed, with P = 170 μm and S₃ = 80 μm. The thickness and the dielectric constant of the medium were set to previous values. The absorption spectra of the proposed absorber shown in Fig. 6(b) at four different polarization angles (0°, 45°, 90°, and 135°) indicate that the absorber exhibits the same absorption characteristics as the previously obtained ones; however, the frequency range is different. The major absorption peak is frequency shifted during the sample rotation. Therefore, if the length of the H-shaped resonator is changed appropriately, an absorber that is tuned at the desired frequencies can be designed.

4. Conclusion

An anisotropic metamaterial absorber with polarization-tunable operating frequencies was demonstrated experimentally by using an asymmetric resonator distribution. In contrast to most previous studies, which focused on the introduction of active materials or performing structural reconfiguration, in this work, several H-shaped resonators were directly integrated to enable operation at different frequencies. When using an appropriate configuration for the resonance units, the absorption peaks could be tuned by simply rotating the sample. In addition, the relationship between the absorption and polarization angle was analyzed, which verified the results of the experiments and provided a better understanding of the absorption process. As an extension, by modifying the lengths of the resonators, the structure could be operated at any frequency of interest. This work provides a more concise approach to realize tunable characteristics that have immense potential in applications including tunable filters, polarization sensitive sensors, polarization detectors, and terahertz switches.