High-Performance Metamaterial Light Absorption from Visible to Near-Infrared Assisted by Anti-Reflection Coating

Abstract

This study experimentally demonstrates two types of ultra-broadband metamaterial absorbers with high performance in the visible-to-near-infrared range by using different anti-reflection coatings (i.e., SiO₂ and Si₃N₄) and a multi-subcell Ti-SiO₂-Ti metasurface. Compared to the bare metamaterial nanostructure, the absorption bandwidth of the coated metasurfaces exhibit increases of 594 nm and 1093 nm, respectively. Such improvements benefit from nearly perfect impedance matching to the free space enhanced by the anti-reflection coating, thin film interference, and excitations of different surface plasmon resonances. As a result, the absorber with SiO₂ coating exhibits a measured bandwidth with an absorption of 0.9 ranging from 502 nm to 1892 nm, while the absorber with Si₃N₄ coating further broadens the bandwidth from 561 nm to 2450 nm. The measured average absorptions for both cases remain above 95% and 87%, respectively. Moreover, both nanostructures are robust to large incident angles of up to 60° for both TE and TM modes. Our findings highlight the promising potential of these absorbers for various applications, including solar energy harvesting, thermal emitters, and photodetectors.

1. Introduction

A metamaterial comprises a series of artificially designed and manufactured nanostructures with unique and outstanding optical properties not found in natural materials. With the development of nanophotonics and nanofabrication, these unique optical properties are applied to practical devices in many ways. By changing the structural morphology and the dimensional parameters of artificial atoms, independent tuning of the dielectric constant and magnetic permeability can be achieved [1]. This offers excellent application potential, such as in imaging [2], filtering [3], and sensing [4,5]. Ultra-broadband light absorption in metamaterials or plasmonic nanostructures has attracted substantial attention due to the considerable interest in developing solar energy harvesting. In the operational bandwidth, nearly perfect absorption efficiency can be achieved on the subwavelength scale for the incident light [6]. This gives metamaterials or plasmonic nanostructures an enormous scope of development in the fields of thermal emission [7,8,9], detectors [10], and photovoltaics [11].

Conventional triple-layer metal–insulator–metal (MIM) configuration with uniform top cells has been widely applied to plasmonic absorbers [12,13,14], where a thin dielectric spacer is used to enable strong plasmonic coupling between the top resonators and the metal film in the bottom. Such an absorber design can be intuitively treated as a resonator coupled to a single-input transmission line. The dielectric spacer thickness influences the radiative damping rates and resonant wavelength. Taking advantage of the strongly radiative damping and inherent metal loss, the resonant absorption bandwidths of these plasmonic absorbers are relatively broader than several hundreds of nanometers. However, due to the inherent narrow bandwidth nature of the plasma, achieving an ultra-broadband metamaterial absorber is still a challenge. So far, many efforts have been made to expand the absorption bandwidth further. One method is to create multiple resonances in each unit cell via several subcells with different sizes on the top layer [15,16]. In this case, each subcell works as an individual resonator that produces multiple resonances. An ultra-broadband absorption can be achieved if these resonances are close enough. Another method is to construct the unit cell in the vertical direction by stacking multiple alternating metal–dielectric layer patterns [17,18]. This method improves the spatial area constraint of the device very well, but the precise pattern alignment during the fabrication is challenging. The third method is based on metallic or metal–dielectric nanocomposites [19,20]. In this method, nanocomposite-induced localized surface plasmon resonance and impedance matching of the metamaterial to the free space contributes to intensive light absorption. Mie resonance also plays an essential role in the latter configuration. Correspondingly, the absorption performance is sensitive to the density and compound ratio of nanocomposites. All of the studies mentioned above have extensively promoted the development and applications of broadband absorbers. However, metamaterial absorbers with higher performance, as well as simple fabrication processes, are still highly anticipated.

In this work, we present and experimentally demonstrate an ultra-broadband metamaterial absorber (UBMA) via a multiple subcell Ti-SiO₂-Ti metasurface assisted with an anti-reflection coating (ARC). Here, the bare multi-subcell metamaterial initially realizes a comparatively wideband absorption, and the ARC further expands the absorption bandwidth due to the nearly perfect impedance matching to the free space. Two different types of coatings are investigated: SiO₂ and Si₃N_4, Correspondingly, the respective 90% absorption bandwidth (BW_0.9) of the coated metasurfaces exhibit increases of 594 nm and 1093 nm compared to the bare metamaterial. The physical mechanism of the absorption performance enhancement is discussed in detail in the paper. For both transverse–electric (TE) and transverse–magnetic (TM) modes, the plasmonic metasurface maintains a high absorption over a wide range of incident angles up to 60°. Compared to the reported works, our nanostructure has the significant advantage of a simple structure while greatly improving the absorption bandwidth. In addition, the ARC material can be easily fabricated without complicated synthesis, and the fabrication is compatible with nano-imprinting lithography. All these advantages make low-cost and large-scale manufacturing in metamaterial energy harvesting highly feasible for many applications.

2. Design and Simulations

The side view figure and schematic diagram of the ultra-broadband metamaterial absorber are shown in Figure 1a,b, respectively. Based on a basic MIM absorber, whose cell is composed of Ti-SiO₂-Ti with four subcells of multiple sizes, a dielectric anti-reflection layer is coated on the top. Here, two anti-reflection coatings, SiO₂ and Si₃N₄, were analyzed to improve the absorption performance. The selection of ARCs can be guided by three considerations. First, to achieve broadband absorption from the visible to infrared range, the ARC material should be transparent or translucent to reduce reflection in the visible band. Second, the ARC should meet the plasma excitation conditions in terms of its optical properties. We will discuss this factor in more detail in Figure 2. Third, the ARC material is expected to be easily fabricated in both laboratory and industrial settings for future large-scale and cost-effective manufacturing. The whole cell has a period P of 750 nm in both the x and y directions. The thicknesses of titanium (Ti) substrate (t₁), silica dielectric spacer (t₂), patterned Ti layer (t₃), and top anti-reflection layer (t₄) are 300 nm, 70 nm, 40 nm, and 140 nm, respectively. In order to combine multiple resonant wavelengths, for patterned metals, the unit cell contains four subcells. Specifically, two cube-shaped nano-antennas are located on the main diagonal with respective side lengths of 170 nm (w₁) and 300 nm (w₂), which are the width and length of the rest of the two cuboids correspondingly. This arrangement reduces the effect of the direction of polarization on the performance. The gap (g) between the four subcells is 140 nm. Here, we used a 2D grating model with a single subcell to pre-design the proper widths of each subcell while taking further optimization of their combination via 3D finite-difference time-domain (FDTD) analysis. In the simulation, the refractive index data of both SiO₂ and Ti were obtained from Palik [21]. Since the 300 nm underlying Ti is much thicker than the penetration depth of electromagnetic wave in this band, the light transmission (T) can be neglected, indicating the absorption A = 1-R, where R denotes the reflectance.

Figure 1c shows the simulated absorption spectra of the absorber without ARC, with SiO₂ and Si₃N₄ ARCs for normal incidence. It can be seen that the effect of the anti-reflection layer was very significant. After adding SiO₂ ARC, the 90% absorption bandwidth of the absorber increased from 747 nm (836–1583 nm) to 1684 nm (440–2124 nm), and the average absorption increased from 95.9% to 96.2%. The spectrum shows multiple absorption peaks, indicating that multiple resonant modes were excited, which is beneficial when constructing an ultra-broadband absorber. When the material of ARC is changed to Si₃N₄, the performance of the absorber in the longer wavelength band is more significantly improved. In this case, the absorber achieved an average absorption rate of 95.0% in the 833–2994 nm range. The above simulation results demonstrate the advantages of effective bandwidth enhancement and flexible operating band by using ARC.

The theory of optical impedance matching can explain the physics behind the proposed enhanced ultra-broadband absorbers with anti-reflection coating. The equivalent impedance Z of the metamaterial absorber is calculated by the following equation [22]:

$$\mathrm{Z}=\sqrt{\frac{{\left(1+{\mathrm{S}}_{11}\right)}^2-{\mathrm{S}}_{21}{}^2}{ {\left(1-{\mathrm{S}}_{11}\right)}^2-{\mathrm{S}}_{21}{}^2}}$$

(1)

S₁₁ and S₂₁ denote the scattering matrix coefficients of reflection and transmission under normal incidence, respectively. Since the Ti substrate is thick enough, S₂₁ can be considered zero. It is worth noting that the equivalent impedance Z is a complex number. Figure 2a shows the simulated real and imaginary parts of the optical impedance compared to the measured absorption spectrum for the SiO₂-coated absorber. At the resonance-induced absorption peaks (vertical dashed line), both real and imaginary parts of the optical impedance of the absorber are almost perfectly matched to that of the free space, whose respective values are 1 and 0 as a reference in dashed lines. Apart from the resonances, the wavelength band between the upper and lower cut-off resonances also enables the optical impedance of the absorber to be highly close to that of free space, resulting in an ultra-broadband absorption. We can also achieve a similar conclusion for the Si₃N₄-coated absorber, as displayed in Figure 2b.

Generally, the coupling efficiency of light to absorber is comparatively low for an absorber without an anti-reflection layer. Surface plasmon resonance (SPR) excitation is relatively weak in this scenario. SPR usually includes two kinds of resonances [23,24,25]: localized surface plasmon (LSP) resonance and propagating surface plasmon (PSP) resonance, which dominate the absorption of the metamaterial nanostructure. The weak SPR modes would undoubtedly result in low absorption efficiency and narrow absorption bandwidth. The anti-reflection coating then changes the impedance of the metamaterial absorber to match it with free space impedance, leading to higher light coupling efficiency to metasurface and more intensive SPR modes. Hence, the absorption efficiency increases, and the absorption bandwidth is widened.

To further understand the absorption mechanism of the absorber, electromagnetic field distributions are additionally analyzed at each resonance with normal incidence under the TE mode. For simplicity, only the SiO₂-coated nanostructure is considered here. As illustrated in Figure 3a, we can observe that the electric field at 480 nm is mainly concentrated at the interface between the air and SiO₂ anti-reflection layer. According to the thin film interference theory [26], it is associated with the interference between two kinds of light. One is the reflected light from the incidence at the interface of air and SiO₂ coating. The other originates from the reflection at the interface of silica and Ti, which then transmits to the air. Figure 3b,c clearly show the excitation of a dipole; the electric field is intensively confined at the edge and corners of the Ti pattern, indicating strong LSP resonance. The magnetic field distribution allows us to gain more insight into the coupling mechanism of the incident light. The magnetic field shown in Figure 3d is concentrated at the interface between the Ti pattern and SiO₂ anti-reflection coating, which is attributed to the PSP resonance between them. In Figure 3e, LSP resonance dominates the absorption since the magnetic field is strong at the surface of the Ti pattern. This LSP-resonance-induced absorption enhancement can also be found in Figure 3f. Additionally, we can still observe magnetic field confinement in the silica spacer. In this case, the incident light penetrates through the thin Ti pattern and bounces back to the Ti substrate. These two metallic mirrors, together with the silica spacer, form a Fabry–Pérot (FP) cavity. Since Ti is a high-loss metal, the quality factor of this FP cavity is relatively low, which further broadens the absorption bandwidth. Thus, the thin film interference and the PSP, LSP, and FP resonances make ultra-broadband absorption possible.

Table 1. Performance comparison of the proposed and reported metamaterial solar absorbers.

Refs. and Published Year	Number of Patterned Layers	Materials	BW0.9 (nm)	Region (nm)	Average Absorption within BW0.9
[27] 2020	2	Al, MgF2, Ti, SiO2	1100	405–1505	95.14%
[28] 2020	1	Ti, SiO2, TiN	1182	415–1597	-
[29] 2020	2	Ti, SiO2	1376	456–1832	94.60%
[30] 2021	2	Ti, a-Si	865	250–1115	97.11%
[31] 2021	18	Ni, Al2O3	2240	300–2540	99.17%
[32] 2022	3	Ti, a-Si, SiO2	1576	329–1905	96.09%
SiO2-coated UBMA (This work)	1	Ti, SiO2	1684	440–2124	96.2%
Si3N4-coated UBMA (This work)	1	Ti, SiO2, Si3N4	2161	833–2994	95.0%

provides a comprehensive comparison of the proposed absorber’s performance with previously reported works operating in the visible to near-infrared bands. In many cases, employing two or more patterned layers [27,28,29,30,31,32] in the configuration has proven to be an efficient way to improve the absorption bandwidth associated with the enhanced excitation of plasmon resonances. However, such configurations often come with increased fabrication complexity and costs. In contrast, our proposed absorbers, featuring a single patterned layer and an anti-reflection coating, offer ease of fabrication and are well suited for large-scale manufacturing thanks to the nanoimprinting technology. Moreover, the materials selected in our device, such as Ti and Si₃N₄, exhibit excellent chemical stability and high melting points, ensuring the robustness of the nanostructure. By capitalizing on the benefits of the anti-reflection coating and multicell configuration, our absorber greatly broadens the operational bandwidth while maintaining high absorptance.

3. Fabrication Process

The fabrication process of the proposed UBMA is illustrated in Figure 4a. Based on a 500 μm silicon wafer, electron beam evaporation with the rate of 1.2 A/s was first utilized to form a 300 nm continuous film of Ti as the substrate. After that, the SiO₂ spacer with a thickness of 70 nm was deposited on the substrate by plasma-enhanced chemical vapor deposition (PECVD, PlasmaPro 80), whose deposition rate was 54.5 nm/min. A standard metallic lift-off process was then used to build up the patterned layer. Specifically, a 150 nm thick positive resist (PMMA 950k 679.03) was coated onto the SiO₂ spacer and exposed via electron-beam lithography (EBL, Raith EBPG5150) as a pattern mask. The EBL works at an acceleration voltage of 100 kV, with a beam current of 0.5 nA and an exposure dose of 800 μC/cm². It is worth noting that the thickness of the resist should be well-optimized. Ideally, the resist should be thicker than the desired Ti cubes to separate the pattern area and the metal to be lifted off. However, if it is too thick, the resolution of the e-beam lithography will reduce due to the electron scattering and proximity effect. Here, dimethyl sulfoxide was used to dissolve the photoresist and remove the extra metal. The sample was obtained by depositing 140 nm SiO₂ or 170 nm Si₃N₄ anti-reflection layer onto the bare metamaterial by PEVCD. Figure 4b,c display the fabricated 3 mm × 3 mm metamaterial sample and its SEM image after lithography. The fabricated nanostructure shows good agreement with the proposed design.

4. Experimental Results

The reflection spectra were observed with an angle-resolved spectrum system (R1, ideaoptics, Shanghai, China) equipped with a highly sensitive spectrometer (NOVA, ideaoptics, China). A halogen lamp was used as the light source to provide a broadband light in the wavelength range of 360–2500 nm. The spot size of the incident light irradiated onto the sample was 1 mm, and the divergence angle was less than 3°, which can be regarded as a parallel light incidence. Since the reflectance of the silver sheet is almost 100% in this spectral range, the reflectance spectrum of the silver sheet was used as a normalized spectrum to eliminate systematic measurement errors. According to the measured results shown in Figure 5a,d, the absorption spectra exhibited significant broadening with SiO₂ or Si₃N₄ anti-reflection layers. Specifically, with the anti-reflection layer of SiO₂, the absorption bandwidth was increased from 796 nm to 1390 nm, indicating a broadening of 594 nm. Moreover, average absorption remained higher than 95%. The nanostructure with Si₃N₄ anti-reflection layer shows an even wider absorption bandwidth of 1889 nm with an average absorption of up to 87.1%. The measured spectra generally matched well with the simulation results, while there was still some discrepancy between the measured and simulated results of Si₃N₄-coated UBMA. This discrepancy is mainly attributed to the refractive index mismatch of Si₃N₄ between the simulation data and deposited Si₃N₄ film. Specifically, Si₃N₄ has different crystal phases, which are largely dependent on the deposition temperature during the fabrication. These different crystalline Si₃N₄ compounds in turn induce instability of the refractive index, resulting in the absorption drop or resonant wavelength shifting. Additionally, Si₃N₄ is more liable to agglomerate in the PECVD, and the comparatively rough surface of Si₃N₄ film may also affect the absorption performance.

The measured absorption of the two different nanostructures with oblique incidences for both TE and TM modes are compared and shown in Figure 5b,c,e,f. It can be seen that the absorber was nearly insensitive to the incident angle. For an absorber with SiO₂ ARC, when the incident angle rose to 30°, the absorption efficiency in the 500–2000 nm band reached 95% under TE polarization. Under TM polarization, the effect of the incidence angle increased slightly, and the absorption efficiency remained above 94% at an incidence angle of 30°. The performance of the Si₃N₄-coated nanostructure behaved similarly. For both configurations, with the increase of the incident angle, there was a slight blue shift of the absorption band. Compared with TE mode incidence, the absorption spectrum of the device was more sensitive to the angle of TM mode incidence light. This polarization-dependent angular dispersion of the proposed nanostructure is associated with the direction of the magnetic field. Specifically, the magnetic field direction of the TE incidence remains unchanged as the incident angle varies, which efficiently drives the circulating currents at all oblique incidences. On the contrary, for the TM polarization, the magnetic field rotates with the variable incident angles, and the circulating currents in turn cannot be driven as efficiently, especially at large-angle incidence. Nevertheless, from 500 nm to 2000 nm, the average absorptances of both nanostructures under the TM incidence at 60° still exceeded 87% and 85%, respectively.

5. Discussion

To further evaluate the solar energy harvesting capability of the fabricated nanostructures, we calculated their solar absorption rates by using the following equation [33]:

$${\mathsf{\eta }}_{\mathrm{A}}=\frac{{\displaystyle {\int }_{{\mathsf{\lambda }}_{\min }}^{{\mathsf{\lambda }}_{\max }}\mathrm{A}\left(\mathsf{\lambda }\right){\mathrm{I}}_{\mathrm{AM}1.5\mathrm{G}}\left(\mathsf{\lambda }\right)\mathrm{d}\mathsf{\lambda }}}{{\displaystyle {\int }_{{\mathsf{\lambda }}_{\min }}^{{\mathsf{\lambda }}_{\max }}{\mathrm{I}}_{\mathrm{AM}1.5\mathrm{G}}\left(\mathsf{\lambda }\right)\mathrm{d}\mathsf{\lambda }}}$$

(2)

where λ_min and λ_max denote the working bandwidth covering the visible (300 nm) to near-infrared (2500 nm) band. I_AM1.5G (λ) is the incident solar power of the air mass 1.5 global (AM 1.5G) spectrum, and A (λ) refers to the absorption of the device. In this case, the numerator represents the total absorbed solar energy by the absorber, which is also illustrated in Figure 6 in red and compared with the standard solar energy spectrum. According to Equation (2) and Figure 6, the absorption rate of SiO₂-coated ultra-broadband absorber reaches 90.9%. For the Si₃N₄-coated structure, more solar energy is missed due to the slightly worse performance of the absorber. However, the solar energy absorption rate is higher than 82% with a large absorption bandwidth. These nearly perfect absorption characteristics make the proposed nanostructures a superior candidate for many applications in the natural environment, such as high-performance solar cells, detectors, and thermal emitters.

6. Conclusions

We propose and experimentally demonstrate an ultra-broadband metamaterial absorber working from visible to near-infrared using anti-reflection coating and a multi-subcell configuration. The multi-subcell configuration fundamentally provides multiple resonant units, promising fundamental broadband absorption. Additionally, with the help of anti-reflection coating, absorption performance is further improved because of better impedance matching. The anti-reflection coatings of SiO₂ and Si₃N₄ are analyzed in detail. Compared to the nanostructure without any coating, the measured BW_0.9 of a nanostructure with SiO₂ coating exhibits a broadening of 594 nm in the range from 502 nm to 1892 nm. Moreover, BW_0.9 increases even higher to 1093 nm, ranging from 561 nm to 2450 nm for the nanostructure with Si₃N₄ coating. The measured average absorptions in both cases remain beyond 95% and 87%. Thin film interference combined with the resonances of PSP, LSP, and FP is the possible reason behind such a high absorption. Angular insensitivity is also numerically demonstrated for both TE and TM modes. The fabrication of such devices is also simple, requiring no etching or pattern alignment. Additionally, this fabrication is compatible with nano-imprinting lithography, making large-scale and cost-effective manufacturing feasible. In addition, the superior solar energy absorption further offers great promise for future versatile utilizations of the proposed absorbers in photovoltaic technology, thermal devices, and broadband detectors.