1. Introduction

Metamaterial devices have attracted substantial attention due to their marvelous electromagnetic performance in many applications such as antenna systems [1], [2], electromagnetic cloaking [3,4], imaging [5,6], ultra-sensitive sensing [7,8] and refractive index engineering [9,10]. The subwavelength structure of the metamaterials can be tailored flexibly by designing the artificial “meta-atoms”. This characteristic enables the metamaterials a designable permittivity and permeability. Meanwhile, the performance of the energy depletion in metamaterials can be utilized in a positive way to design metamaterial absorbers by introducing the power loss.

During the last decade, trends to achieve light absorption in metamaterials and plasmonic nanostructures have increased tremendously due to the huge interest in the development of solar energy harvesting. Generally, regarding the absorption bandwidth, absorbers can be categorized into two types, namely narrow band absorber and broadband absorber. The former one can easily find applications in linear or nonlinear sensors covering both visible and infrared regions [11–15]. On the contrary, in the applications such as photodetectors [16], thermal emitters [17,18], photovoltaics (PV) [19,20] and ultra-short pulse generation [21,22], broadband absorbers are always required. Up to date, plenty of metamaterial structures have been reported to demonstrate broadband perfect absorbers. For instance, using hole array [23], cylinder array [24], complementary crosses and cylinders [25] and multilayer structures [26]. However, these methods may suffer from the problems such as a relatively narrow absorption bandwidth (i.e. 300 nm - 400 nm), a complicated cell pattern or a requirement of noble metals.

In this paper, we design an ultra-broadband absorber based on a thin metamaterial nanostructure composed of a periodic array of titanium-silica (Ti-SiO₂) cubes and an aluminum (Al) bottom film. The proposed structure can achieve a nearly perfect absorption spanning a broad range from 354 nm to 1066 nm, indicating a wavelength band over 712 nm (>90%) from visible to near-infrared. In this band, the absorber exhibits an average absorbance of 97% and the peak absorbance of 99.8%. The broadband perfect absorption benefits from the excitation of surface plasmon resonance in addition with the resonance induced by the metal-insulator-metal Fabry–Pérot (FP) cavity. The impact of the thicknesses of the top metal layer and the silica layer on the absorption performance is investigated in detail. Additionally, we thoroughly analyze the absorption properties of the metamaterials using various top and bottom metals, and the results demonstrate the effectiveness of the proposed structure. Oblique incidences of both TE and TM-polarized waves are also investigated to demonstrate the angle and polarization insensitivity of the absorber. The proposed metamaterial absorber not only has a simple geometry and ordinary metal based material composition, but also shows a high performance in both bandwidth and absorptivity. Moreover, this nanostructure is straightly compatible with high throughput manufacture technology using soft nano-imprinting lithography, making the low-cost mass-production of the material highly feasible. It is expected that such absorber structure will hold great potential in solar cell and photodetector applications.

2. Geometry, parameters and methods

Figure 1 shows the proposed ultra-broadband metamaterial absorber [Fig. 1(a)] with a magnified cubic unit cell [Fig. 1(b)]. The metamaterial structure consists of a periodic array of Ti-SiO₂ cubes placed directly on the surface of a uniform aluminum (Al) substrate. The Ti-SiO₂ cubes have a width (w) of 190 nm and the unit cell has a period constant (p) of 250 nm. Thicknesses of the Ti-SiO₂-Al layers from top to bottom are 20 nm (t₃), 80 nm (t₂) and 200 nm (t₁), respectively. The surrounding material is assumed to be air. The complex dielectric constants of Ti and Al are modeled by a Drude-Lorentz fitting to tabulate experimental data [27], and the refractive index of the SiO₂ is 1.45. The fabrication process of the proposed metamaterial is compatible with the popular nano-fabrication technologies such as electron beam lithography and nano-imprinting lithography.

 

Fig. 1 (a) Schematic diagram of the proposed broadband metamaterial absorber. (b) Magnified unit cell of the absorber. Here, a periodic array of subwavelength Ti-SiO₂ cubes is placed directly on the surface of a uniform aluminum substrate. Parameters of the unit cell are set as p = 250 nm, w = 190 nm, t₁ = 200 nm, t₂ = 80 nm, t₃ = 20 nm. The aluminum layer is thick enough to prevent the light transmission and the surrounding material is assumed to be air.

Download Full Size | PDF

Numerical calculations are performed by the method of three-dimensional finite-difference time-domain (FDTD). Periodic boundary conditions are employed for the lateral boundaries and perfectly matching layers (PML) are applied along the z direction to eliminate the boundary scattering. The reflection spectrum (R) is recorded by a 2D frequency-domain power monitor, which is perpendicular to the x-y plane. As the aluminum film is thicker than the penetration depth of the incident light, the incident light will be blocked and the transmission (T) of the structure is nearly zero (T = 0), resulting in the absorption A = 1 – R (reflection).

3. Numerical analysis and discussions

3.1 Performances and principles

The absorption spectrum of the proposed metamaterial absorber is illustrated in Fig. 2. It is observed from the spectrum that the absorber exhibits a high absorption over 90% in a wide wavelength range from 354 nm to 1066 nm, indicating a 90% bandwidth of 712 nm from visible to near-infrared. The average absorbance over this band is calculated as >97% and a nearly perfect absorbance (over 99%) is achieved from 804 nm to 921 nm. The peak absorption is obtained up to 99.8% around 886 nm with a bandwidth of 26 nm.

 

Fig. 2 Simulated absorbance spectrum. BW: Bandwidth.

Download Full Size | PDF

To reveal the physical mechanism of the proposed metamaterial absorber, current density (J), electric and magnetic field distributions (|E| and |H|) of the TE-polarized light with a normal incidence at various resonant wavelengths (i.e. 380 nm, 470 nm, 603 nm and 890 nm) are simulated in the x-z plane and illustrated in Fig. 3. Overall, it is seen from Figs. 3(a)-3(d) that, electric current distributes in the thin titanium layer and on top of the aluminum surface, implying the origins of the energy loss which result in the broadband absorption. In Figs. 3(e)-3(h), one can clearly see from the electrical distributions that surface plasmon polaritons (SPPs) are excited in the metematerial structure that light is coupled into the air-slot and localizes around the metal corners between the adjacent unit cells, creating a SPP-induced light absorption. However, the distributions of the magnetic field are intrinsically different. In specific, at the short wavelength of 380 nm shown in Fig. 3(i), the resonance is considered to be the propagating surface plasmon (PSP) resonance between the continuous Al film and SiO₂ spacer, where the magnetic field is not only strongly confined in the gap region underneath the nano-cubes, but also intensively enhanced between the adjacent cells. While for Fig. 3(l), at the long wavelength of 890 nm, localized surface plasmon (LSP) resonance dominates the absorption. In this case, the magnetic field is mainly localized within the gap between the top Ti cap and the Al substrate. Since Ti is a highly lossy metal, the Q-factor of this LSP resonance is rather low, which further broadens the absorption bandwidth. On the other hand, the incident light penetrates through the thin Ti cap and bounces back at the Al bottom layer. These two reflectors together with the silica spacer form a lossy Fabry–Pérot (FP) cavity with a low-Q factor, causing the FP resonance-induced wideband light absorption. Thus, the magnetic distributions of the resonances at 470 nm and 603 nm, respectively illustrated in Figs. 3(j) and 3(k), indicate a combination of the PSP, LSP and FP resonances.

 

Fig. 3 (a)-(d) Current density J (top row, color bar in the x-z plane), (e)-(h) distributions of the electric field |E| (color bar in the x-z plane) and (i)-(l) magnetic field |H| (color bar in the x-z plane). The observed range in x-direction is from −125 nm to 125 nm.

Download Full Size | PDF

To additionally verify the advantages of introducing the SPP modes, we compare the absorption spectra of the proposed metamaterial structure with the planar Ti-SiO₂-Al continuous films, where the layer thicknesses are identical to those of the proposed one. The absorption spectra of both cases are shown in Fig. 4. We can find that within the analyzed wavelength range, the absorption of the metamaterial absorber is much stronger than that of the continuous films because of the SPP enhancement. This result further shows that the combined SPP and FP resonances together lead to the strong absorption in such a broadband region.

 

Fig. 4 Spectrum of the proposed metamaterial absorber (red solid line) compared to that of the Ti-SiO₂-Al planar structure (blue dash-dot line).

Download Full Size | PDF

3.2 Absorbance with oblique incidences

Absorptions with oblique incidences for both TE-polarized (x-polarized) and TM-polarized (y-polarized) waves are compared and shown in Figs. 5(a) and 5(b), respectively. The incident angle is increased by a step of 20^o from 0^o to 60^o. One can see that, at the relatively small incident angles (e.g. 20^o and 40^o), the structure shows almost angle independent absorptions for both polarizations. At an incident angle of 40^o, the absorption bandwidth is slightly broadened beyond 1200 nm for TM-mode due to the increased coupling at the longer wavelengths. Specifically, as depicted in Fig. 5(b), the absorption bandwidth (≥90%) extends from 1066 nm to 1200 nm. Here, the average absorbance of TM-mode is estimated higher than 96% from 354 nm to 1200 nm. For the incident angle of 60^o (cyan dashed line), the spectra exhibit small ripples caused by the increased resonance modes within the analyzed bandwidth, while the average absorptions are still adequate as 90% and 92% for TE and TM modes, respectively.

 

Fig. 5 Absorbance spectra with different incident angles for (a) TE-mode (x-polarized) and (b) TM-mode (y-polarized). The incident angle steps in 20^o from 0^o to 60^o.

Download Full Size | PDF

3.3 Effects of material and geometry on absorption performance

For a thorough investigation, we analyze the effects of the material and geometry on the absorption performances of the proposed structure. Firstly, as shown in Fig. 6, we study the impact of the metal material by replacing the top and bottom metals with several commonly used metals while keeping the geometry parameters constant. Here, the proposed structure with a top Ti cap and bottom Al substrate is illustrated in the figure for comparison. One can see from Fig. 6(a) that the top metal largely affects the performance of the metamaterial absorber. When using Ag or Al as the top cap, the absorption spectra exhibit narrow resonances around 415 nm or 363 nm/861nm, corresponding to the respective FP and SPP resonances. The narrow bandwidth implies a relatively large quality factor (Q factor) due to a small metal loss. The structure with a Ni top cap shows an even broader absorption band up to ~1294 nm, while the absorbance is relatively lower within the visible range, which is decided by the metal’s intrinsic dispersion property.

 

Fig. 6 Absorbance spectra with different top and bottom metals. (a) Top metal cap replaced by silver (Ag), aluminum (Al) and nickel (Ni). (b) Bottom metal substrate replaced by gold (Au), silver (Ag) and copper (Cu). The spectra of the proposed structure with the top cap of Ti and bottom layer of Al are depicted in a red solid line in Figs. 6 (a) and 6(b) for comparison.

Download Full Size | PDF

The behind physics of the absorption sensitivity to the thin metal cap can be explained as follows. The top metal cap has the role to bind the incident wave to the dielectric spacer by the evanescent wave propagating through the top metal. When the perfect absorption condition (i.e., critical coupling condition) related to the metal cap's permittivity and thickness is satisfied, the incident wave tunnels through the top metal cap, excites the cavity mode and is perfectly absorbed there. The top cap influences not only the Q factor of the FP cavity, but also the coupling strength between the cavity mode and the external light incidence. Correspondingly, the top metal cap largely affects the absorption bandwidth and absorption strength, which in turn determines the characteristics of the designed absorber.

On the other hand, as shown in Fig. 6(b), when replacing the bottom Al film with different metals, the absorption performance is slightly changed since the bottom layer serves mainly as a back reflector. These conclusions can be proved by the absorption ability of the respective top and bottom metals as illustrated in Fig. 7. Based on the proposed model, the absorbance is significantly dominated by the top cap, while slightly affected by the bottom metal. The absorption in Al film shows small peaks around 380 nm and 890 nm, which indicate the PSP and LSP resonances, respectively.

 

Fig. 7 Absorption ability in different parts of the proposed absorber.

Download Full Size | PDF

The influence of the cell cube width (w), SiO₂ thickness (t₂) and Ti thickness (t₃) on the absorption spectra is shown in Fig. 8. As depicted in Fig. 8(a), with a relatively larger filling factor (i.e., w/p), the structure behaves more like a continuous sandwiched film, which leads to a wider absorption bandwidth but a decreased absorbance. On the contrary, the structure with a relatively smaller filling factor shows an intensive SPP resonance and the absorbance increases at the expense of a decreased bandwidth. Thus, the tradeoff of w to achieve an ideal absorption within the analyzed wavelength range is around 190 nm. As to the impact of the SiO₂ thickness, it is seen from Fig. 8(b) that the absorption band gradually red shifts as t₂ increases. This is mainly because the effective cavity length correspondingly increases, and the ideal spectral property is achieved around t₂ ~80 nm. The influence of the Ti thickness is shown in Fig. 8(c). A thinner Ti cap promises a less lossy cavity while a thicker one indicates a weaker coupling. In both cases, the absorber suffers from the limitations of the absorbance and the bandwidth. As a result, the optimal performance can be obtained with the Ti thickness t₃ of 20 nm.

 

Fig. 8 Demonstration of the geometric effects on the absorption performances with normally incident TE-polarized light: (a) the width of the Ti-SiO₂ cell cube (w), (b) the thickness of SiO₂ (t₂) and (c) the thickness of Ti (t₃).

Download Full Size | PDF

4. Conclusion

In summary, we have proposed an ultra-broadband metamaterial absorber with a nanostructure configured by a periodic array of titanium-silica (Ti-SiO₂) cubes and an aluminum (Al) bottom film. The proposed structure exhibits a nearly perfect absorption with an average absorbance of 97% and 90% absorption bandwidth over 712 nm from visible to near-infrared. The peak absorption is as high as 99.8% during this band. Superior surface plasmon resonance combined with the Fabry–Pérot resonance between the metals enables this broadband absorption. The absorption average level remains higher than 90% for both TE and TM-polarized light with an incident angle up to 60°. Effects of different metal materials and structure geometry on the absorption performance are compared in detail to investigate the characteristics of the proposed absorber. The structure shows the advantages of simple configuration and simultaneous high performances in both bandwidth and absorptivity. Moreover, it requires no noble metal and is compatible with the manufacturing technology of nano-imprinting lithography, making the cost-effective manufacturing process feasible. Such a broadband absorber will easily find applications in solar cell, thermal emitters and nonlinear plasmonics.