1. Introduction

Advanced optics research in recent years have begun to focus on metasurfaces, or two-dimensional equivalents of metamaterials, which have great potential applications from microwave to optical frequency especially in terahertz (THz) spectral region due to the paucity of suitable materials in nature. With the distinct resonance response of the subwavelength constituent elements, metasurfaces can alter the amplitude, phase and polarization of the electromagnetic wave within an optically thin layer [1]. So far, diverse applications have been demonstrated at THz frequency, including beam deflectors [2,3], lenses [4–6], wave plates [7–9], beam splitters [10–12], and absorbers [13–15]. Since the first experimental demonstration by Landy et al. [16], metamaterial or metasurface absorber has undergone a rapid development, presenting a wide variety practical application in selective thermal emitters [17], micro-bolometer detectors [18,19], uncooled THz imaging [20,21] and refractive index sensing [22,23].

Traditionally, the metamaterial absorbers employ a sandwich structure, consisting of arrays of subwavelength metallic pattern on a dielectric spacer, backed by a metallic ground plane. The top metallic pattern acts as electric resonators and the ground plane blocks any transmission. Owing to the thin dielectric spacer, there exists a strong coupling between the two metallic layers, which produces a magnetic response in the dielectric layer [24]. By changing the geometry of the metallic pattern and the thickness of the dielectric layer, the effective permittivity and permeability can be tuned independently such that the effective impedance can be designed to closely match to the free space, resulting in a high absorption of incident electromagnetic waves at specific frequency. Since the absorption mechanism is based on electromagnetic resonance, the bandwidth of the absorption is limited by nature [25]. Nevertheless, a wider bandwidth with high absorption is more favorable in most of practical applications.

To address this issue, two similar approaches were adopted by merging several close absorption peaks into a wider absorption. The first one was to horizontally arrange multiplexed resonators with different geometries into one planar unit cell [26,27], but the bandwidth was relatively narrow. Another alternative way was vertically stacking multiple metallic patterns [28–31], but the fabrication cost and complexity were highly increased. Although the previously reported absorptions were relative high, the bandwidths were still finite and usually no more than 2 THz. A patterned silicon based metamaterial absorber reported by Sheng et al. achieved a bandwidth of 2 THz with nearly 100% absorption [32]. By using a double-layer doped silicon grating array, Yan et al. showed an absorption more than 95% with a bandwidth over 2 THz [33]. However, both silicon based metamaterial absorbers were quite thick and difficult to be integrated with other THz components. Recently, by integrating fractal resonators into one metasurface supercell, Kenney et al. demonstrated a bandwidth of 2.82 THz with absorption exceeding 45% [34], but there still exists a room for improvement of the absorption bandwidth.

In this work, we demonstrate a bi-metasurfaces based ultra-broadband THz absorber numerically and experimentally. The bi-metasurfaces are constructed by horizontally arranging multiplexed cross resonators arrays into two sub-metasurfaces and vertically stacking the two sub-metasurfaces together. The measured absorption spectrum of the bi-metasurfaces absorber shows an ultra-wide bandwidth of 4.46 THz with the absorption more than 50%.

2. Design and simulation

As schematically illustrated in Figs. 1(a) and 1(b), the unit cell of the proposed bi-metasurfaces absorber consists of two stacking gold (Au) resonators layers sandwiched by parylene film. Both resonance layers use the cross as basic element, whose symmetry also assures it insensitive to the polarization of electromagnetic waves. The top and bottom resonance layers are arranged into the four quadrants of the unit cell, which can reduce the coupling between themselves. A 100 nm thick continuous aluminum (Al) film supports the top and bottom resonators arrays with paylene as the spacers, following the typical metal/dielectric/metal absorber configurations. The bottom metasurface has a 40 μm periodicity of resonators array with a 50% fill factor and the period of the top metasurface are 80 μm, thus making the bi-metasurfaces has a periodicity of 80 μm.

 

Fig. 1 Schematic diagram of (a) perspective view and (b) side view of the unit cell of the bi-metasurfaces.

Download Full Size | PDF

The dimensions of the whole unit cell were optimized with a commercially available finite-difference time-domain (FDTD) software. The conductivity of Al and Au were set to be 3.72 × 10⁷ and 4.09 × 10⁷ Sm⁻¹, respectively. A permittivity ε = 2.60 and loss tanδ = 0.04 were used to model the parylene film [35]. To get a broadband absorption targeting at high frequency, the bottom metasurface contains two identical groups in which four multiplexed cross resonators with the same width of 2 μm and different bar lengths on a thin parylene film. The lengths of the four cross are a₁ = 9μm, a₂ = 10μm, a₃ = 11μm, a₄ = 12μm and the thickness of parylene is t₁ = 4 μm, respectively. The top metasurface employs two multiplexed cross resonators with lengths of a₅ = 22 μm and a₆ = 25 μm on a thicker parylene film of t₂ = 12 μm, aiming to obtain two close absorption peaks at low frequency. The thickness of the parylene film is elaborately designed to be approximately equal to the quarter of the effective wavelength, and the objective is to get a Salisbury screen absorption peak located between the absorption bands of the two metasurface absorbers and obtain an ultra-broadband absorption.

The simulated absorption spectrum of top metasurface absorber is plotted in Fig. 2(a), where three absorption peaks are clearly observed at f₁ = 3.50 THz, f₂ = 3.92 THz and f₃ = 5.20 THz with the absorption of 95.4%, 78.2% and 85.4%, respectively. The two peaks at low frequency f₁ and f₂ are close and merge into a broader absorption band with a 1.19 THz bandwidth for the absorption exceeding 50%. To gain into the physical insight of the two low frequency absorption peaks, we simulated the 2D current density distributions at f₁ and f₂. As shown in Figs. 2(b) and 2(c), the strong current flows on the cross resonators and the ground plane resulting from the electric dipole response at both f₁ and f₂. It is also obvious that the transient current directions of the cross resonator and the ground plane are just opposite and form a current loop, indicating a magnetic dipole response at f₁ and f₂. These facts are in agreement with the standard metamaterial absorber explanation [36]. However, no evident opposing currents occur between the cross resonators and the ground plane at f₃, which is not shown here.

 

Fig. 2 (a)Simulated absorption spectrum of the top metasurface absorber. The inset is the schematic diagram of the unit cell of the top metasurface absorber. (b, c) Simulated surface current density at the two resonance frequencies of f₁ and f₂. (d) The real part, imaginary part and magnitude of the normalized effective impedance spectra.

Download Full Size | PDF

We then analyzed the effective impedance of the top metasurface absorber, which can be calculated using the following equation [37]:

Additionally, the thickness of the parylene dielectric spacer is approximately equal to the quarter of the effective wavelength (λ/4), which is the third absorption wavelength divided by the refractive index of the parylene film. Therefore, we have reasons to believe that the third absorption peak at f₃ = 5.20 THz should be attributed to the Salisbury screen absorption, a unique absorption mechanism different from the metamaterial absorber theory, which requires a top resistive layer match to the free space impedance on a dielectric layer with a thickness of λ/4. The Salisbury screen type operation, relying on the matched impedance and phase modification just caused by the presence of the top metasurface, effectively expands the bandwidth of the original metasurface absorber [34].

As for the bottom metasurface absorber, the simulated absorption spectrum is shown in Fig. 3(a), where four resonance peaks are obtained at f₄ = 7.12 THz, f₅ = 7.56 THz, f₆ = 7.87 THz, and f₇ = 8.72 THz with the absorption of 92.4%, 89.5%, 99.7% and 97.6%, respectively. Since the four resonance peaks are designed so close that they also merge into a much wider absorption band from 6.78 THz to 9.05 THz with absorption higher than 50%. The simulated current density distribution on the surface of the absorber are illustrated in Figs. 3(b)-3(e) at the four frequencies, indicating each resonance peak is mainly dominant by one of the cross resonators, and higher frequency resonance peaks correspond to the cross resonators with shorter bar length.

 

Fig. 3 (a)Simulated absorption spectrum of the bottom metasurface absorber. The inset is the schematic diagram of the unit cell of the bottom metasurface absorber. (b-e) Simulated surface current density at the four resonance frequencies of f₄ = 7.12 THz, f₅ = 7.56 THz, f₆ = 7.87 THz, and f₇ = 8.72 THz.

Download Full Size | PDF

The simulated absorption spectrum of the bi-metasurfaces absorber is shown in Fig. 4(a). Obviously, the Salisbury screen absorption peak at middle frequency band makes the two absorption bands connected, resulting in an ultra-broadband absorption. The bandwidth with absorption higher than 50% is 3.31 THz and the full width at half maxima (FWHM) is calculated to be 87.3%. Compared with the recently reported THz broadband metasurface absorber with a bandwidth of 2.82 THz while absorption exceeding 45% [34], the bandwidth of our metasurface absorber is as high as 4.89 THz with absorption exceeding 45%. Noticeably, the resonance peaks in bi-metasurfaces absorber show apparent red shifts respect to the single bottom metasurface absorber. It is mainly caused by the additional capacitance from the top parylene film on the bottom cross pattern, leading to an obvious decrease in the resonant frequencies [38].

 

Fig. 4 (a)Simulated absorption spectrum of the bi-metasurfaces absorber. (b, c) Microscope photos of the fabricated bi-metasurfaces absorber sample with the focuses located at bottom and top resonator layers, respectively. (d, e) SEM pictures of the bottom and top metasurface absorbers.

Download Full Size | PDF

3. Experimental results and discussion

The bi-metasurfaces absorber was fabricated using the standard micro-fabrication process. First, a 100 nm thick Al film, serving as the ground plane, was evaporated on a 4-inch silicon substrate. The two-layer dielectric spacer parylene of the thickness of 4 μm and 8 μm were deposited by chemical vapor deposition (CVD) with vapor phase parylene monomers at room temperature and a pressure of 22 mTorr. Parylene film are chemically inert to most acids, alkalines and organic solvents, which makes the fabrication of the bi-metasurfaces a good process compatibility. Additionally, parylene is a widely used polymer with great flexibility and biocompatibility, and its thickness can be precisely controlled in the course of CVD, while the thickness of other polymers like polyimide is difficult to be accurately controlled as they are usually made by spin-coating method. Two layers of 20/100 nm Cr/Au films were sputtered on the parylene film followed by the standard optical lithographs and wet etching processes. The total size of final sample is 1 × 1 cm. Figures 4(b) and 4(c) display the two microscope photos of the fabricated bi-metasurfaces absorber with the focuses located at the bottom and top gold layers, respectively. For comparison, the bottom and top metasurface absorbers were also fabricated separately by the similar processes, and their scanning electron microscope (SEM) pictures are shown in Figs. 4(d) and 4(e), respectively. The final thicknesses of parylene films of the bottom and top (hybird) metasurface absorbers were measured to be 4.4 μm and 11.6 μm by a thin film reflectometry system (K-MAC ST2000-DLXn).

A Fourier-transform infrared (FTIR) spectrometer (Bruker Vertex 80v) was used to characterize the absorption performance of the fabricated absorbers under the vacuum condition. A reflection accessory with a fixed incident angle of 11° was equipped to measure reflection spectra R(ω), which were normalized with respect to a gold mirror using a 6 mm aperture. The absorption is calculated by A(ω) = 1 − R(ω) since the transmission through the samples is nearly zero owing to the continuous Al ground plane.

The measured absorption spectra of the top and the bottom metasurface absorber are plotted in Figs. 5 (a) and 5(b). As expected, the top metasurface absorber displays three absorption peak at 3.56 THz, 4.19 THz and 4.89 THz with the absorption of 72.5%, 64.6% and 56.6%, respectively, as shown in Fig. 5(a). We notice that the measured frequency of the Salisbury screen absorption peak is a little bit lower than the simulation, which is caused by the thinner parylene film than the design value. Figure 5(b) shows the absorption spectrum of the bottom metasurface absorber, where four peaks merge into a wide absorption band with two less obvious absorption peaks at 7.96 THz of 76.1% and 9.20 THz of 79.6%. The FWHM is 51.3% and the bandwidth with the absorption more than 50% is 3.80 THz. Compared with simulations, the measured resonance frequencies of both top and bottom metasurface absorbers show blue shifts, which may result from the fabricated errors of the cross resonators.

 

Fig. 5 Measured absorption spectra of the fabricated (a) top metasurface, (b) bottom metasurface and (c) bi-metasurfaces absorbers.

Download Full Size | PDF

The measured absorption spectrum of the bi-metasurface absorber is given in Fig. 5(c), which presents an exceptionally broad absorbing response. Two absorption peaks result from the top metasurface at low frequency band, as expected. At high frequency band, a wide absorption band is also obtained due to the existence of the bottom metasurface. The Salisbury screen absorption peak located between the high and low frequency absorption bands, which plays a significant role in realizing the ultra-broadband absorption. Within the overall absorption band, the maximum absorption is 81.9% at 7.26 THz and the minimum absorption is 43.6% at 4.43 THz. The FWHM of the broadband absorption is calculated to be 97.7%, which is almost two times of the bottom metasurface absorber. The maximum bandwidth of the bi-metasurfaces absorber with absorption higher than 50% is 4.46 THz. This bandwidth with the absorption exceeding 50% simultaneously demonstrates a big advance toward ultra-broadband planar absorber in THz regime.

Although the experimental broadband response agrees with the simulation, it is noticeable that the measured absorption is lower than the predication. We attribute the discrepancy to several reasons as follows. Firstly, the various non-idealities of the fabrication, including geometrical errors of the resonator patterns, thickness errors and surface roughness of the parylene film. Secondly, a constant permittivity parameter was used to model the parylene film during the simulation, while the parylene film is more possibly dispersive with frequency in reality. The last, the measured spectra were taken with THz waves incident at 11°respect to the surface of metasurface absorber, while the simulations were carried out at normal incidence.

4. Conclusions

In summary, an ultra-broadband THz absorber using bi-metasurfaces has been investigated. The bi-metasurfaces is composed of two stacking metasufaces sandwiched by parylene film. The bottom metasurface brings in a broad range of absorption at high frequency, while the top metasurface is response for the absorption at low frequency, plus a Salisbury screen absorption peak, which results from the parylene dielectric spacer with the thickness about λ/4 and the adequate surface impedance caused by the top metasurface. The Salisbury screen absorption peak links the high and low frequency absorption bands, achieving a wider range of absorption. The fabricated bi-metasurfaces absorber shows a bandwidth of 4.46 THz with absorption greater than 50%. Meanwhile, the measured FWHM of the broad absorption band is as high as 97.7%. The bi-metasurfaces absorber also offers a better trade-off between the absorption bandwidth and the fabrication complexity, which brings it great application potentials on uncooled THz bolometric imaging.