1. Introduction

Since Pendry et al. first proposed the thin metal wires and split-ring resonators (SRRs) for obtaining negative refraction [1–3], a variety of metamaterials and plasmonic structures have been demonstrated, which exhibit interesting electromagnetic properties that are otherwise unavailable in natural materials. These include ultrahigh or zero refractive index [4–6], spoof-surface plasmon propagation [7,8], unidirectional optical transmission [9,10], and frequency-selective perfect absorption [11,12], etc. Broadband absorption is one of these engineered properties, which is of importance for a number of applications such as bolometric sensing, thermophotovoltaics, and camouflaging, etc. One reported design to realize broadband absorption is the multilayered thin film structure [13]. By matching the effective impedance of the multilayered thin films to that of free-space, perfect absorption can be achieved over a wide bandwidth. The multilayered thin film absorbers have the advantages of simple lithography-free fabrication and sharpness in spectral filtering [14–16]. However, their thicknesses have to be at least on the order of a quarter wavelength, presenting a limitation in term of device sizes and complexity especially for the long wavelengths in the infrared region. Plasmonic metamaterials and antennas offer a pathway to developing compact absorbers with ultrathin sub-wavelength thickness at the cost of complex fabrication involving sophisticated techniques, such as e-beam lithographic and etching processes [11]. Broadband absorption in metamaterials and antennas is usually achieved by superimposing multiple localized resonances to form a wide absorption band. Previous studies have demonstrated broadband absorbers using the structures of multiplexed square patches [17–19], cross resonators [20,21], and the vertically stacked patch resonators [22–25], etc. Such multiplexed absorbers were designed by merging several subunits to form a macro unit-cell, which involved complex fabrication techniques. Recently, Aydin et al. reported a non-multiplexed structure made of cross-trapezoid units, which exhibited strong broadband absorption in the entire visible frequencies [26,27]. The relevant resonances, however, were shown to involve both propagating modes and localized modes, whose overlapping and hybridization behaviors are quite complex and thus present a difficulty in design and control. Moreover, the angular dependence of the propagating modes made the absorption feature sensitive to the incident angle [26].

In this paper, we report on broadband antenna absorbers based on multiple localized magnetic resonances, which exhibit angle-insensitive broadband absorption. Our absorbers are made of tapered metal strips atop a dielectric film of Al₂O₃ backed with a metal ground plane, extending the classical strip array antenna for broadband operation [28,29]. By using well-defined localized magnetic resonances as correlated to the varied width of the tapered strips, near-perfect absorption is designed and achieved over a bandwidth from 29.2 THz to 38 THz. The broadband absorption feature is maintained over a wide range of incident angles up to 75°. Our demonstrated tapered patch antenna design offers a simple and promising approach for obtaining omnidirectional broadband absorption.

2. Design and simulations

The unit-cell of our designed antenna absorber is sketched in Figs. 1(a) and 1(b), made of a bottom copper film, an Al₂O₃ dielectric layer, and a top tapered Au strip. The width of the strip is linearly tapered down from 4 µm at the center to 2.6 µm at the two ends. The lengths of period are 8 µm and 32 µm along the x and y directions, respectively. The thicknesses of the Al₂O₃ layer and the Au strip are designed as 600 nm and 100 nm, respectively. The incident light is within the x-z plane. Strip array antenna typically exhibits magnetic resonances with excited anti-parallel surface currents in the strips and the metal ground plane. Frequencies of these resonant modes are determined by the half-wavelength resonance condition, and thus are inversely proportional to the strip width [30]. In our designed strips with varied widths, we can expect closely spaced multiple resonances to form an absorption band.

 

Fig. 1 (a) Schematic of one unit-cell of the designed patch antenna absorber. (b) Cross-section of the unit-cell in x-z plane.

Download Full Size | PDF

Absorption property of the structure is simulated with Ansys’ HFSS solver in driven mode, which is based on finite element method in the frequency domain. The Cu and Au layers were described using the Drude model with optical parameters as given in [31]. The complex refractive index of the Al₂O₃ layer was taken from [32]. Floquet excitation port and periodic boundary condition were applied in one unit-cell, and the S parameter was used to calculate the reflection coefficient. The absorption was then obtained by one minus the reflectance since the copper ground plane is thick enough to prevent light transmission. The calculated absorption spectrum for TM polarization and normal incidence is shown in Fig. 2(a). An absorption band is obtained from 29.5 THz to 37.5 THz with absorption efficiency exceeding 80%. This broadband absorption consists of three overlapped resonances at 31.4 THz, 33.5 THz, and 36.7 THz, respectively. Figures 2(b)–2(d) illustrate the electrical field and surface current distributions on the top metal strip, and the ground metal surface for these three resonant modes, respectively. One can see that the surface currents on the top strip and the bottom metal layer are anti-parallel to each other, forming current loops. Therefore, the resonances at 31.4 THz, 33.5 THz, and 36.7 THz are considered as magnetic resonances. The field for the 31.4 THz mode is mainly located at the center of the strip, while fields of the other higher frequency modes are located at the tapered region and near the ends with narrower strip widths. These features indicate an anti-correlation between the resonant frequency and the strip geometric width, consistent with the half-wavelength resonance condition.

 

Fig. 2 (a) Simulated absorption spectrum of the absorber for TM polarization at normal incidence. (b), (c), and (d) show the electrical field amplitude and surface current distributions for the resonant mode at 31.4 THz, 33.5 THz, and 36.7 THz, respectively. The top row is the electrical field amplitude distribution. The middle and bottom rows represent surface current distribution on the top metal strip and the bottom metal layer, respectively.

Download Full Size | PDF

To further examine the nature of resonances involved in the broadband absorption, we also calculated the absorption spectrum as a function of period as shown in Fig. 3(a). It is seen that the resonant frequencies of the three modes exhibit a negligible dependence on the period, which is indicative of localized modes instead of propagating modes such as the surface plasmon polaritons or waveguide modes. In addition, these mode frequencies are blue-shifted as the thickness of the Al₂O₃ increases as shown in Fig. 3(b), which rules out the contribution of Fabry-Perot resonances from the Al₂O₃ layer. This frequency blue-shift can be explained by the intuitive inductance and capacitance (LC) resonator model [33,34]. The larger Al₂O₃ thickness results in a smaller capacitance, which leads to higher resonant frequency. These features support that the modes contributing to the broadband absorption are localized magnetic resonances.

 

Fig. 3 Simulated absorption spectrum as a function of (a) period and (b) Al₂O₃ thickness at normal incidence. The dotted lines illustrate traces of the three relevant resonant modes.

Download Full Size | PDF

3. Experimental results and analysis

To fabricate the designed antenna absorber, we first evaporated a 300 nm thick Cu film on a silicon substrate to serve as the ground plane, and then grew a 600 nm thick Al₂O₃ film on top of the Cu using e-beam evaporation technique. Finally, a tapered gold strip array was fabricated on top of the Al₂O₃ with standard stepper photolithography, metal deposition and lift-off processes. Figure 4 shows the scanning electron microscope (SEM) images of the fabricated sample. The area size of the strip array is 8 mm × 8 mm. The fabricated strip dimensions are very close to those designed values.

 

Fig. 4 Top-view SEM images of the fabricated absorber sample.

Download Full Size | PDF

The sample was characterized with Fourier transform infrared reflection (FTIR) spectroscopy in ambient. A broadband globar IR light from the FTIR spectrometer was focused by an 8 inch focal length off-axis parabolic mirror onto the sample with a spot size of about 1.5 mm. The reflection from a copper mirror was used as the reference spectrum. A wire-grid polarizer was used in the optical path to control the incident light to be TM or TE polarization. The measurement setup consisted of two coaxial sample and detector rotational stages. The blocking of incident beam by the detector limits the minimum incident angle to be 15° in our measurements. The solid red curve in Fig. 5(a) shows the measured absorption of the sample at 15° angle for TM polarization. Strong absorption over 80% is observed over the frequency range from 29.2 THz to 38 THz, which consists of three overlapped resonances at 31.5 THz, 35.2 THz and 37.1 THz. Frequencies of these resonances are slightly shifted as compared to the calculated values, which is likely originated from fabrication uncertainties. However, the overall absorption band is in fairly good agreement with the simulated result shown in black. As the incident light changes to TE polarization, the absorption feature disappears as shown by the dashed curves. This feature is consistent with the polarization dependence of the magnetic resonance in strip array. It is noticeable that there is still about 10–20% absorption at frequencies outside the absorption band for TM polarization but not for TE polarization in both simulation and experiment, which likely originates from the Al₂O₃ and metal losses in the structure enhanced by tails of the multiple antenna resonances. Figure 5(b) shows the measured absorption spectra at different incident angles. The sample exhibits a strong absorption band with amplitude of around 90% for angles less than 35°. As the angleincreases, the amplitude of the absorption band slightly decreases to 80% even when the angle is as large as 75°. The strong absorption band is maintained for incident angles up to 75° without much degradation, as seen from the absorption contour plot shown in Fig. 5(c). This angle-insensitive absorption is attributed to the unique mechanism of coupling between localized magnetic resonances and free-space incident light [35]. As illustrated in Fig. 5(d), the magnetic resonance couples to free-space light via the magnetic field component. For incident TM light, its magnetic field component is always perpendicular to the circulating current loop for various incident angles, which makes the coupling between magnetic resonance and incident light independent of the incident angle. Therefore, our absorber design using magnetic resonances offers the advantage of omnidirectional absorption as desired in many practical applications.

 

Fig. 5 (a) Measured and simulated absorption spectra of the sample for TM and TE polarizations at 15° incident angle. (b) Measured absorption spectra and (c) absorption contour plot of the sample at different incident angles for TM polarization. (d) Illustration of coupling between a localized magnetic resonance and the incident TM light.

Download Full Size | PDF

As the resonances contributing to the wideband absorption are primarily correlated to the width of the tapered strips, there is flexibility in design to tune and engineer the absorption band in our proposed absorber structure. To shift the absorption band in frequency, we also fabricated a tapered strip array with a smaller width, which decreases from 3 µm to 1.6 µm when moving from center toward the ends as shown in Fig. 6(a). The measured absorption band, as shown in Fig. 6(b), is blue-shifted by 2.5 THz as compared to the structure presented in Fig. 4. In addition, by multiplexing multiple tapered units, it is also possible to further increase the absorption bandwidth. Figure 6(c) shows one representative of our fabricated absorbers with two multiplexed subunits. The strip in the left subunit is tapered from 3.1 µm to 1.6 µm, while the one in the right subunit is tapered from 4.1 µm to 1.6 µm. For this multiplexed structure, there are four localized magnetic resonances involved, which result in a larger bandwidth as evident from the measured and simulated curves in Fig. 6(d). The measured result shown in red curve is in fairly good agreement with the simulation. The oscillations from individual contributing resonances are smeared out in the measured spectrum, which is likely caused by fabrication uncertainties, especially the variation in the optical constants of Al₂O₃ film. The refractive index of aluminum oxide is known to depend on the evaporation parameters such as vapor pressure, deposition rate, and substrate temperature [36] etc. A variation in the optical constants of Al₂O₃ could contribute to the deviation between model and experiment curves. Figure 6(e) shows the measured absorption spectra of the multiplexed sample at different incident angles. Again here, the strong wideband absorption feature is maintained up to 75° without much degradation.

 

Fig. 6 (a) Top-view SEM image of one fabricated tapered strip array with smaller width. (b) Measured and simulated absorption spectra of the sample with smaller width at 15° incident angle. (c) Top-view SEM image of one fabricated multiplexed structure with two subunits. (d) Measured and simulated absorption spectra of the multiplexed sample at 15° incident angle. (e) Measured absorption spectra of the multiplexed structure at different incident angles for TM polarization. The dotted curve in (b) and (d) is the measured absorption of the sample presented in Fig. 4 for comparison.

Download Full Size | PDF

4. Conclusions

In conclusion, we have designed and realized broadband antenna absorbers using tapered-strip array structures. The magnetic resonances as correlated with the strip width were engineered to create near-perfect broadband absorption, which was insensitive to incident angle. Strong absorption exceeding 80% was measured over the bandwidth from 29.2 THz to 38 THz for a wide range of incident angles up to 75°. The angle-independent absorption was attributed to the unique mechanism of coupling between relevant localized magnetic resonances and free-space incident light. Our demonstrated absorber design offers a simple and promising way in engineering natural materials for strong omnidirectional absorption with a variable and wide bandwidth. Such broadband antenna absorbers could be of interest for applications ranging from electromagnetic filtering, shielding, thermal energy harvesting to infrared detection etc.