Metamaterial perfect absorbers [1] have attracted the attention of the electromagnetics community in recent years. The ability to obtain any desired effective permittivity and permeability in a metamaterial by carefully designing its structure has resulted in the discovery of exotic phenomena such as negative refraction, cloaking, perfect absorption etc., which are not possible with ordinary materials [2]. Electromagnetic absorbers [1, 3] made of metamaterials can have significant applications such as controlled reflectors [4], sensors [5], bolometers [6], imaging devices [7], thermal emitters [8] and Q-switching of infrared lasers [9]. The key to a highly absorbing medium is to design resonant structures that can simultaneously be driven by both the electric and magnetic fields of radiation. This results in resonant absorption of radiation along with a perfect or optimized impedance matching of the metamaterial with vacuum ( $\text{Z}=\sqrt{\mu (\omega )/\epsilon (\omega )}$). One of the most common designs for the metamaterial absorber structure consists of metal-dielectric-metal (MDM) tri-layered structures with a patterned metallic top surface. Proper choice of the size of the structure and layer thickness can result in resonances for the electric and the magnetic excitations at a common frequency [10]. The electric field of the incident radiation induces a time varying charge distribution in the top structured pattern and the corresponding mirror image charges are formed in the bottom metallic film (ground plane). This creates opposite current densities at the top and bottom metal-dielectric interfaces. These anti-parallel currents along with the displacement currents at the edges form a circulating current loop that gives rise to a magnetic dipole resonance.

Usually, the top structured surface is designed to be highly sub-wavelength in size and resonate via a LC resonance [3], although the dipole resonance of a sub-wavelength sized patch can also be used [10]. The MDM tri-layer structure resonance wavelengths are given by ${\lambda }_{\text{nm}}=2l\sqrt{{\epsilon }_{\text{r}}}/\sqrt{n^2+m^2}$, where l is the length of the top metallic patch, n, m are positive integers with m² + n² ≠0 and ε_r is the permittivity of the dielectric spacer [11]. As the polarization in the patch is mostly along the direction of electric field, one can assume m = 0. The dipolar mode (n=1, m=0) is most often dominating. Moderate sized resonators can also exhibit in-plane, multi-polar resonances at shorter wavelengths compared to the dipolar resonance [12]. Thus, one can utilize higher-order resonances with moderate patch sizes comparable to a wavelength. The larger sizes could ease the fabrication demands and well established micro-fabrication techniques can be applied to make infra-red (IR) metamaterial absorbers.

The bandwidth of a metamaterial absorber is of great importance for many applications such as bolometers, solar cells and imaging devices. Metamaterial absorbers usually have a few hundred nanometer (∼ 200 nm) bandwidth at IR wavelengths, which mainly arises due to the resonant nature. Broadening the absorption bandwidth by combining non-degenerate multiple resonant structures within a unit cell that give rise to multiple or broadband absorption bands has been demonstrated [13, 14]. Another strategy is to create multi-layer resonators [15]. Recently, the use of metals with low conductivity to control the dispersive permittivity of the metal, as the constituents of the metamaterial absorber, has been shown to enable specific device performance over required bands [16, 17].

So far, only highly conducting metals like Ag, Au and Al have been used as constituents in metamaterial absorbers. These metals have plasma frequencies at ultra-violet frequencies and highly dispersive plasma-like permittivities at IR frequencies. Recently, doped semiconductors such as indium tin oxide (ITO) and aluminum doped zinc oxide have been used for NIR plasmonic applications [18]. ITO thin films are good infrared reflectors while transparent at visible wavelengths, and well known for transparent electrodes in photo-voltaic applications. The optical properties of doped semiconductors are well described by the Drude free electron model below the plasma frequency. Due to the lower charge carrier concentrations compared to plasmonic metals, the plasma frequency of ITO is much smaller and the dispersion of the permittivity is comparatively very small at IR frequencies. The plasmon frequency of ITO corresponds to λ ∼ 1.2μm for carrier densities of 10²¹ cm⁻³ [19]. The Drude dielectric permittivity of Au, Al and ITO are plotted in Fig. 1. Here we present the design, fabrication and characterization of a metamaterial that efficiently absorbs radiation over the 4 μm to 7 μm band and allows visible radiation to transmit through. The design consists of an array of circular conducting (aluminum or ITO) micro-disks separated by a thin film of ZnS from a continuous ITO film that acts as a ground plane as shown in Fig. 2. A tri-layer structure with strong electric and magnetic resonances is formed, while the ITO ground plane renders the structure transparent to visible light. This metamaterial shows broadband absorption as compared to the usual design where the top and bottom layers are both metallic.

 

Fig. 1 Left panel: Drude dispersion of the real and imaginary parts of the dielectric permittivity of Au (ω_p/2π = 2176 THz, ε_b = 5.7 and γ/2π = 6.5 THz), Al (ω_p/2π = 3464 THz, ε_b = 5.1 and γ/2π = 19.41 THz) and ITO (ω_p/2π = 461 THz, ε_b = 3.9, γ/2π = 28.7 THz). The right panel shows dispersion for ITO in an expanded view..

Download Full Size | PDF

 

Fig. 2 Left panel: Schematic of unit cell of an absorbing metamaterial. The structure absorb the IR radiation and diffract through the visible radiation. Middle: The atomic force microscope scan shows a disk height of about 100 nm. Right: SEM image of the fabricated structure with h = 200 nm, d = 380 nm, t = 100 nm, disk diameter = 3 μm and periodicity in X–Z directions are 8μm. The bar indicator is 8 μm long. Inset: Diffraction of a He-Ne (632.8 nm) laser transmitted through the structure with a zeroth order transmittance of 45%.

Download Full Size | PDF

We first optimized the geometrical parameters for the structure to resonant at the first higher order mode (m=3) with a resonant wavelength of ${\lambda }_{\text{nm}}=2l\sqrt{{\epsilon }_{\text{r}}}/3$. For a disk of 3 μm diameter, the resonance wavelength is 4.6 μm. Numerical simulations of the electromagnetic fields were performed using the commercial finite element method based COMSOL Multiphysics software [20]. In the simulations, the Drude free-electron model was used for the dielectric permittivities of Al and ITO as $\epsilon (\omega )={\epsilon }_{\text{b}}-\left[{\omega }_p^2/\omega (\omega +i\gamma )\right]$ with ω_p/2π = 3464 THz, ε_b = 5.1 and γ/2π = 19.41 THz for Al [22], and ω_p/2π = 461 THz, ε_b = 3.9, γ/2π = 28.7 THz for ITO [18]. The dielectric permittivity of ZnS was taken from experimental data in Ref. [23]. More details of the computer simulations can be found in Ref [10]. Wavelength dependent reflectance [R(λ)] and transmittance [T(λ)] were obtained by integrating the power flow (Poynting vector) over two surfaces placed above and below the metamaterial structure respectively. The absorbance, calculated as A(λ) = 1 − R(λ) − T(λ), is plotted in Fig. 3 and shows peaks with absorbances exceeding 98% at 4.5 μm, 6.5 μm and 7 μm. We also observe a lower peak absorbance of about 80% at 16 μm corresponding to the fundamental dipole mode of the disk. The resonance peaks at 4–7 μm almost merge and have a much wider net bandwidth of about 2.5 μm wavelengths compared to Al/ZnS/Al perfect absorber structures in the mid-wave IR region [21]. This significant increase of bandwidth is due to the small dispersion of the complex permittivity of ITO (compared to aluminum or gold) that enables optimal impedance matching of the structure for the same dielectric layer thickness at multiple peaks within the given band. The overall dispersion of the metamaterial response is small and the field enhancements within the structure occurs over a wider wavelength band.

 

Fig. 3 Left panel: Simulated absorption versus the wavelength for the metamaterials absorber structures designed. Right panel: Measured absorption from the fabricated metamaterials absorber structures and the reflectance of plane ITO film.

Download Full Size | PDF

We also evaluated the absorbance for the tri-layer design with the same geometric parameters as shown in Fig. 2, but with ITO disks and a ITO ground plane. The simulated absorbance shows two distinct peaks at 4.25 μm and 7 μm with absorption exceeding 95% corresponding to the higher order modes with m=3 and m=5 respectively. The absorption band corresponding to the fundamental dipole mode at long wavelengths, cannot be effectively excited as the skin depth in ITO at MWIR and LWIR wavelengths is very different for semi-conducting films as compared to the skin depth of metals. Hence, the optimum layer thicknesses for effective excitation of the far spaced modes becomes different. The calculated electric and magnetic field distributions at the peak absorption wavelength of 4.5 μm give insights on the nature of the resonances and are plotted in Fig. 4. The field distributions confirm that the absorption properties of the metamaterial are due to the electric and magnetic resonances excited in the tri-layer structure. The electric field distribution shows half-wavelength-like charge distributions excited in the top disk and the corresponding image charge formation in the ITO film. This results in strong localization of the electric field within the dielectric layer. The magnetic field distribution clearly shows a magnetic resonance with a first higher order mode (n = 3) of the tri-layer. There are three magnetic loops corresponding to anti-parallel currents induced in the metallic disk and the ground plane. Simultaneous excitation of the electric and magnetic resonances leads to a strong confinement of the electromagnetic fields inside the tri-layer. Significant electric and magnetic fields penetrate into the lower ITO layer as compared to the Al disk, and can be clearly seen in Fig. 4.

 

Fig. 4 Left: Electric field magnitude, Right: Magnetic field magnitude in tri-layer of Al/ZnS/ITO metamaterial at 4.6 μm wavelengths. The nature of the m = 3 mode is apparent.

Download Full Size | PDF

The designed metamaterial was fabricated by deposition techniques. Thermal evaporation was used for the deposition of aluminum as well as ZnS, which was the dielectric spacer material used. A uniform 380 nm thick ZnS film was deposited on a commercially available ITO coated glass with an ITO film thickness of 200 nm and sheet resistance of 50 Ω/sq. Subsequent to the deposition of the continuous ZnS layer, a shadow mask containing an array of micro-holes was employed to form the Al disks on top by direct deposition of Al vapor through the holes in the membrane. The shadow mask containing 3 μm holes in a square array of period 8 μm over 1.5 mm² area was fabricated using excimer laser micromachining. A scanning electron microscope (SEM) image of the structure is shown in Fig. 2. The Al disks were measured to have average diameters of (3 ± 0.1) μm and disk heights of 100 nm with some side-wall slope by atomic force microscope (XE 70, Park Systems). Reflection and transmission measurements over the wavelength range 2.5 μm to 17 μm range were performed using a Fourier transform infrared spectrometer (Agilent, Model Cary 660) coupled to a IR microscope (Agilent, Model Cary 600) and a cooled HgCdTe detector. The measurements are averaged over an angular range of 14° due to the numerical aperture of the 10× microscope objective. Reflection spectra were taken with polarized light and normalized to that obtained from a smooth gold film. The reflection from the plain ITO film was about 85% to 90% over the 3 μm to 14 μm wavelength band. The reflection from the metamaterial was reasonably polarization independent due to the circular symmetry of the disks. The measured absorption spectrum [A(λ) = 1 − R(λ)] of the metamaterial absorber is shown in Fig. 3. Broadband absorption with a peak absorbance of 97% at 5 μm was measured and ascribed to the merged resonances corresponding to m = 3 and 5. The measured spectra are in reasonable conformity with the predicted spectra with minor differences that presumably arise from the inverted cup-like shapes of the disks (see Fig. 2). Fabrication imperfections cause inhomogeneously broadened peaks that merge and give rise to a broadband absorption >70% in the 4 μm –7 μm band. We have only shown the data up to 17 μm as ZnS is transparent only up to 14 μm and starts absorbing at wavelengths beyond.

Most metamaterial designs utilize an LC resonance or the fundamental dipole (m=1) mode for use as a sub-wavelength sized resonator. However, highly localized higher order modes that need not to be in sub-wavelength limit, can also give rise to near-perfect absorption as shown here for the m = 3 mode. For a particular size of the patch, the resonance wavelength of the higher order modes are at lower wavelengths than for the fundamental mode. For a given band, the size of the resonator utilizing a higher order mode will be larger. This leads to significant easing of the demands on fabrication of the metamaterial at MWIR and NIR wavelengths. Note that using a higher order mode with a larger unit cell comparable to the wavelength renders the system not homogenzable as diffracted modes will also be present. The simulated R(λ) and T(λ) include the power in the diffracted beams while only zeroth order specular reflection is present in the experimental measurements. Very little power flow in the diffracted modes in reflection and negligible overall transmission in noted in simulations [21]. The reasonable agreement between simulated and measured data indicates that diffraction effects do not considerably affect the absorbance/reflectance, presumably due to the strength and localization of the resonances. The lowered plasma frequency and conductivity of ITO allows the broadening of the resonances due to the smaller quality of the resonances.

Use of ITO in the metamaterial has several advantages over previous designs and allows for increased flexibility: e.g., the ITO ground plane can be configured in a CMOS configuration to tune the absorption of the metamaterial via charge injection into the ITO [24]. Another great advantage is the transparency at visible wavelengths that allows such an IR absorbing metamaterial to interface with other applications that require visible light. With rapid micro-patterning techniques, such metamaterials with ITO disks can be used for smart window applications that need high visible transparency and high absorptivity/emissivity at 8 μm to 12 μm wavelengths.

In summary, we have designed, fabricated, and experimentally demonstrated a broadband infrared metamaterial absorber that is transparent at visible wavelengths. A maximum absorption level of about 97% at 5 μm wavelength with 50% bandwidth of over 3.5 μm is achieved in the simulations as well as in the fabricated structures. The large bandwidth is due to the small dispersion in the dielectric permittivity of ITO at IR wavelengths. Fabrication by shadow mask deposition techniques is rapid and allows for direct fabrication of large structured areas. The fabricated absorber is transparent to visible light and allows the flexibility of using visible radiation to control the metamaterial or for other applications.