1. Introduction

Effective mid-infrared (MIR) absorption is essential in various applications, e.g., energy harvesting, radiative cooling, gas sensing [1–3], etc. In recent years, many different MIR absorbers based on plasmonic nanostructures and metamaterials have been reported [4–6]. In terms of materials, perfect MIR absorbers can be realized mainly by metals, semiconductors, two-dimensional (2D) materials and polar dielectrics. The most common approach to enhance absorption is to pattern metals into nanoscale shapes. For example, metal-dielectric-metal three-layer configurations are commonly employed with different metallic nanostructures on top, such as metallic disks [7], strips [8], crosses [9], split-ring resonators [10], etc. By changing the shape and size of the top nanostructures, electrical and magnetic resonances can be simultaneously excited in a three-layer configuration, where the incident light is sufficiently absorbed. Highly doped semiconductors have plasma frequencies in the MIR regime, where surface plasmon polaritons can be excited to enhance the absorption. It can be tuned in a wide MIR range with different doped carrier densities [11,12]. 2D materials, especially graphene, also have tunable optical properties [13]. It can be nanopatterned and greatly enhanced absorption can be realized through excitation of surface plasmon polaritons with tightly confined optical field at the interfaces, despite the weak intrinsic absorptivity [14]. In the MIR wavelength range, polar dielectrics can also give perfect absorption [15]. They exhibit negative permittivities within the Reststrahlen band (i.e., the range between the longitudinal optical and transverse optical phonon frequencies). Similar to surface plasmon polaritons, surface phonon polaritons can be excited in a nanostructured polar dielectric with tightly confined optical field at the dielectric surface, allowing for high absorption in the MIR range [16,17].

All the absorbers mentioned above have utilized micro-/nanopatterns to excite different optical effects for perfect absorption, requiring complex and thus expensive fabrication technologies, e.g., lithography [7–10,12,14,16,17]. To simplify the fabrication process, a lithography-free planar absorber is desirable. A multilayer geometry with alternating layers of BaF₂/NiCr has been designed to achieve broadband, polarization-insensitive and omnidirectional MIR absorption [18]. Impedance is also matched in the Si₃N₄/Ti/Si₃N₄/Cu four-layer planar structure, allowing for high absorption [19]. Earlier reports focused on metal-dielectric-metal three-layer geometries and the middle dielectric layer is quarter-wavelength thick for destructive interferences to achieve zero reflection [20,21]. Double-layer configuration is the simplest among planar MIR absorbers. Kats et al. have reported an MIR absorber consisting of a sapphire substrate coated with an ultrathin VO₂ film [22], where unconventional interferences persist due to the non-trivial interface phase shifts [23,24]. To produce a sufficient phase shift at the interface, special lossy material combinations are always required.

Here in this work, a distinct double-layer dual-band MIR absorber is demonstrated experimentally, which consists of a polar dielectric SiO₂ layer deposited on top of a metallic tungsten (W) layer. In this absorber, both interference resonance and Berreman mode are observed for the greatly enhanced dual-band absorption in the MIR regime. Unlike the unconventional interferences enhanced absorption [22–24], the optical performance obtained in our absorber maintains even when the bottom W layer is replaced by other metals. This provides great flexibility for the absorber design. It is known that the Berreman mode is similar to the epsilon-near-zero (ENZ) mode and exists when the real part of the dielectric constant of the material is close to zero. However, in contrast to the ENZ mode, which lies outside the light cone, the Berreman mode, a leaky mode, lies within the light cone and can be very easily excited by the incidence from free space [25,26]. In our double-layer planar absorber, there is no additional transverse wave vector. Thus the Berreman mode, rather than the ENZ mode, is observed, well verifying the theoretical predictions reported in Refs. [25,26]. Recently, Shaykhutdinov et al. have detected the Berreman mode in a 100-nm thick thermally-grown SiO₂ film on a Si substrate through atomic force microscope infrared spectroscopy [27]. Here in our experiments, room-temperature plasma enhanced chemical vapor deposition (PECVD) is applied to deposit the SiO₂ layer, in order to protect the bottom metallic film from being peeled off in high-temperature environment. The spectral detection is straightforwardly conducted by Fourier-transform infrared spectroscopy (FTIR). The whole fabrication process does not involve lithography and thus is extremely simple.

2. Methods

2.1 Fabrication and characterization

Our MIR absorber consisted of a polar dielectric SiO₂ layer on top of a metallic W layer, which was easily fabricated. First, the W layer was 206 nm thick (unless otherwise specified) and deposited on a piece of clean silicon (Si) substrate (2 cm × 2 cm × 0.5 mm) by magnetron sputtering (DC power: 100 W; argon pressure: 3 mTorr). Then, the SiO₂ layer was deposited by room-temperature PECVD (N₂O flow: 200 sccm; SiH₄ flow: 1.7 sccm; pressure: 300 mTorr). The SiO₂ layer thickness was controlled accurately through the deposition time. Cross-sections of the sample were inspected using scanning electron microscopy (SEM; Carl Zeiss Ultra 55). To avoid charging, an ultrathin gold film was sputtered onto the facet before the SEM observation.

Due to the negligible transmission through the 206-nm thick W layer and the Si substrate, the absorption spectra (A) of the sample could be obtained by measuring the reflection spectra (R) according to A = 1 - R. The reflection spectra under normal incidence were measured by a liquid nitrogen refrigerated MCT (mercury cadmium telluride) detector equipped in an infrared microscope (Hyperion 1000) which was connected to a FTIR spectrometer (Bruker Vertex 70). In the microscope setup, the incident light was focused onto the sample. Oblique light beams were unavoidable, leading to the excitation of Berreman mode and thereby small absorption peaks at ∼ 8 µm in all the measured normal-incidence absorption spectra shown below. For the measurement of the reflection spectra under oblique incidence, an A513 attachment with an angle adjustment system was installed in the FTIR spectrometer. Unpolarized light was used in all the measurements since there is no polarizer inside our FTIR system. The incident angle of the infrared light could be changed from 15° to 30° and the reflected light was measured with an inner deuterated triglycine sulfate (DTGS) detector. A gold mirror with reflectivity over 98% in the MIR range was used as a reference.

2.2 Numerical simulation and analysis

To illustrate the optical behaviors of our planar samples, two-dimensional full-field optical simulations were conducted based on a finite-difference time domain (FDTD) method (Lumerical FDTD Solutions). A plane wave was used as the source with varied incident angles and wavelength (λ) ranging from 2 to 14 µm. Perfectly matched layers (PMLs) were set along the light propagation direction, while along the horizontal direction, periodic and Bloch boundary conditions were set for normal and oblique incidences, respectively. The refractive indices of SiO₂ and W were obtained from experimental results [28,29]. The reflected light was detected by a power monitor placed over the source. Since there was almost no light transmitted through the W layer, the absorption was simply calculated by subtracting the reflection from unity (i.e., A = 1 - R). For comparison, both reflection and absorption spectra were calculated by a transfer matrix method (TMM) [30]. In order to analyze the light trapping mechanism, the normalized absorption density profile as a function of the wavelength of the incident light, i.e., A_p(λ), was calculated according to the following equation [27]:

3. Results and discussion

Figure 1(a) shows our fabricated SiO₂/W double-layer absorber on a Si substrate. The SiO₂ and W layers are ∼ 947 and ∼ 206 nm thick (i.e., t_SiO2 ∼ 947 nm, t_W ∼ 206 nm), respectively. Figure 1(b) shows real (n) and imaginary (κ) parts of the refractive indexes of SiO₂ [28] and W [29] over the wavelength range from 2 µm to 14 µm. For SiO₂, between ∼8 µm and 9.3 µm, n_SiO2 is smaller than κ_SiO2, resulting in negative real part of the relative permittivity. This range is referred to as the Reststrahlen band [15]. The high κ is mainly due to the vibration of Si-O-Si bridges [28]. Beyond the Reststrahlen band, n_SiO2 is higher than κ_SiO2. Specifically, below ∼ 8 µm, n decreases gradually as λ increases, and becomes lower than 1 when λ > 7.4 µm, while κ_SiO2 increases but remains extremely low. When λ increases in the range beyond 9.3 µm, n_SiO2 first becomes maximal and then decreases gradually, while κ_SiO2 first drops quickly and then becomes almost zero. There is a small κ_SiO2 peak (corresponding to a small change in the n curve) at λ ∼ 12.4 µm also due to the vibration of Si-O-Si bridges [28]. For W, both n_W and κ_W increase almost linearly with λ, and κ_W is always much larger than n_W (a typical feature of metal). Due to the unique dielectric properties, the SiO₂/W double-layer configuration demonstrates distinct optical absorption behaviors to be shown below.

 

Fig. 1. (a) SEM image of the cross section of our fabricated SiO₂/W double-layer absorber; (b) real (n; red curve) and imaginary (κ; black curve) parts of the refractive indices of SiO₂ and W. (c) The measured (red solid curve) and simulated (black solid curve) absorption spectra of our absorber under normal incidence; and (d) the simulated distribution of the normalized absorbed energy density as a function of the incident wavelength (t_SiO2 = 947 nm and t_W = 206 nm). In (b)-(d), the Reststrahlen band of SiO₂ (8.0 - 9.3 µm) is defined between the two vertical dashed lines.

Download Full Size | PDF

From Fig. 1(c), it is seen that under normal incidence, a very high absorption peak exists on the right side of the Reststrahlen band of SiO₂ (i.e., λ ∼ 10 µm) surrounded by two small sidelobes. The absorptivity of the main peak is ∼ 0.98 and the full width at half maximum (FWHM) is ∼ 1.29 µm. In the short wavelength range, there is a very broad small absorption peak around λ ∼ 5 µm. The simulated and measured absorption spectra match very well (Fig. 1(c)), verifying the feasibility of our numerical simulation model. Based on this numerical model, the normalized absorption density profile was calculated with Eq. (1) and plotted in Fig. 1(d). In the Reststrahlen band, n_SiO2 < κ_SiO2 holds (Fig. 1(b)). Therefore, the incident light is mainly reflected by the top SiO₂ layer and almost no light is transmitted into or absorbed by the bottom W layer (Fig. 1(d)). At the left band edge of λ ∼ 8 µm, there is a small absorption peak in the measured curve. This is mainly due to the excitation of Berreman mode by the unavoidable oblique incident light (to be explained in detail below). Beyond the Reststrahlen band, n_SiO2 is higher than κ_SiO2 (Fig. 1(b)). The incident light can be coupled into the SiO₂ layer and reflected by the bottom W layers, being absorbed by both of them (Fig. 1(d)). When they interfere constructively, light can be strongly trapped in the top SiO₂ layer. At λ ∼ 5 and 10 µm, t_SiO2 ∼ λ/(4•n_SiO2) holds, meaning constructive interference appears. Thus, two absorption peaks are observed at these two wavelengths for the sample of this thickness (Fig. 1(c)). Due to the extremely low κ_SiO2 in the short wavelength range, the absorption peak at λ ∼ 5 µm is not as high or sharp as the one at λ ∼ 10 µm, where the incident light is strongly and deeply absorbed in the top SiO₂ layer (Fig. 1(d)).

The thickness effect of the top SiO₂ layer (t_SiO2) on the absorption spectra of our fabricated SiO₂/W double-layer absorbers is also studied with t_W = 206 nm and demonstrated in Fig. 2(a). When t_SiO2 = 106 nm, the SiO₂ layer is too thin to trap light in it. Therefore, the absorption is very low and there are nearly no apparent absorption peaks in the considered wavelength range. As t_SiO2 increases to 353 nm, minor absorption peaks appear and are enhanced. Further increasing t_SiO2 enables constructive interferences in the top SiO₂ layer, leading to more absorption peaks. In the short wavelength range below λ ∼ 8 µm, the first constructive interference induced absorption peak occurs at λ ∼ 2.7 µm when t_SiO2 = 470 nm and redshifts to λ ∼ 6.2 µm when t_SiO2 = 1552 nm. Meanwhile, more short-wavelength absorption peaks appear when t_SiO2 increases to 1787 nm. In the long wavelength range beyond λ ∼ 9.3 µm, as t_SiO2 increases to 947 nm, the absorption peak redshifts and increases to nearly 1 at λ ∼ 10 µm. As mentioned above, the constructive interference condition, i.e., t_SiO2 ∼ λ/(4•n_SiO2), is satisfied when t_SiO2 = 947 nm. Here, good mode confinement appears in the top absorptive SiO₂ layer (Fig. 1(d)) and contributes to the high absorption (Figs. 1(c) and 2(a)). Further increasing t_SiO2 to 1787 nm leads to another high absorption peak close to 1 at λ ∼ 12.4 µm, which is attributed to the local high κ_SiO2 value (Fig. 1(b)) and the re-satisfied constructive interference resonance.

 

Fig. 2. Measured absorption spectra of our fabricated SiO₂/W double-layer absorbers (a) with various SiO₂ thicknesses and t_W = 206 nm, and (b) with different W layer thicknesses and t_SiO2 = 947 nm under normal incidence. (c) Simulated absorption spectra of the double-layer absorbers with the top SiO₂ layer (t_SiO2 = 947 nm) but different metallic layers with a fixed thickness of 200 nm under normal incidence. In (a), the Reststrahlen band of SiO₂ (8.0 - 9.3 µm) is defined between the two vertical dashed lines.

Download Full Size | PDF

Figure 2(b) shows the measured absorption spectra of the SiO₂/W double-layer absorbers with fixed SiO₂ layer thickness (i.e., t_SiO2 = 947 nm) but different W layer thicknesses (t_W). It is seen that when t_W is thick enough, e.g., t_W = 129 and 206 nm, no light could be transmitted through the W layer and the absorption spectra keeps almost unvaried as t_W varies. When t_W becomes thinner, e.g., t_W = 70 nm, or removed from the structure, the ∼ 10 µm peak becomes lower and even disappears with more light transmission, while the ∼ 5 µm peak is enhanced because of the reduced reflection loss from the W layer through the low-loss SiO₂ top layer. Such phenomena indicates the important roles of the sufficiently thick W layer in both blocking light transmission and providing reflection for interference in the SiO₂ top layer. The numerically simulated absorption spectra shown in Fig. 2(c) illustrate that the bottom metal layer W can be replaced by other metals (e.g., Ta, Ag and Au) without changing the overall light absorption behaviors. This provides great flexibility for the design.

Note that almost no redshift is observed for the absorption peak at λ ∼ 8 µm as t_SiO2 increases (Fig. 2(a)). In order to investigate the optical mechanism behind it, absorption spectra of the SiO₂/W double-layer absorbers with t_SiO2 = 229 and 947 nm (t_W = 206 nm) are measured for unpolarized light under oblique incidence and plotted in Figs. 3(a) and 3(e), respectively. Here, the wavelength range is reduced to 7-12 µm, including only one interference-induced absorption peaks at λ ∼ 10 µm when t_SiO2 = 947 nm for comparison. As mentioned before, t_SiO2 = 229 nm is too thin to keep constructive interference within the SiO₂ layer and thus there is only a very small absorption peak appearing at λ ∼ 9.5 µm. For both absorbers, as the incident angle (θ) increases, the long-wavelength absorption peak remains almost unchanged, while the short-wavelength peak at λ ∼ 8 µm is greatly enhanced, especially when θ increases from 15° to 30°. Due to the limitation of our facility, the absorption spectra with larger θ values could not be measured accurately. In order to compensate this deficiency, numerical simulations based on both TMM and FDTD were used to calculate the absorption spectra with θ ranging from 0° to 60° for both p- and s-polarizations. For the absorbers with t_SiO2 = 229 and 947 nm, the averaged absorption spectra were calculated and plotted in Figs. 3(b) and 3(f), while the p- and s-polarized spectra were shown respectively in Figs. 3(c) and 3(g), as well as in Figs. 3(d) and 3(h). The TMM calculated spectra match well with the FDTD numerically simulated spectra, except small ripples appearing in FDTD under θ = 60° (see Figs. 3(b) and 3(c)). Such deviations are probably due to the insufficient absorption of PMLs for the incident light with larger oblique angles. The consistence between the experimentally measured spectra in Figs. 3(a) and 3(e) and the calculated spectra (averaged for s- and p-polarizations) under θ = 15° and 30° in Figs. 3(b) and 3(f) verifies again the feasibility and reliability of our numerical modeling. Under normal incidence, theoretically there are no absorption peaks at λ ∼ 8 µm for both absorbers (Figs. 3(b) and 3(f)). However, such absorption peaks were observed experimentally (Figs. 3(a) and 3(e)). This is probably induced by some unavoidable oblique light illuminating on the samples in the microscope measurement setup (see Methods). The calculations predict that when θ becomes larger, the absorption peak at λ ∼ 8 µm keeps increasing, while the long-wavelength peak remains almost unchanged (Figs. 3(b) and 3(f)). When we see closer the absorption spectra for p- and s-polarizations (respectively shown in Figs. 3(c) and 3(g), as well as in Figs. 3(d) and 3(h)), it is found that for both absorbers with t_SiO2 = 229 and 947 nm, the peak at λ ∼ 8 µm only occurs for p-polarization under oblique incidences, while the long-wavelength peak is almost independent of the light polarization and the incident angle.

 

Fig. 3. Measured absorption spectra of our fabricated SiO₂/W double-layer absorbers with (a) t_SiO2 = 229 nm and (e) 947 nm under different incident angles (θ = 0°, 15° and 30°). (b, f) The simulated average absorption spectra, (c, g) the simulated absorption spectra for p-polarization only and (d, h) s-polarization only of the SiO₂/W double-layer absorbers with (b, c, d) t_SiO2 = 229 nm and (f, g, h) 947 nm under different incident angles (θ = 0°, 15°, 30°, 45° and 60°). The p- and s-polarized light is schematically shown in the inset of (c) and (d), respectively.

Download Full Size | PDF

In order to analyze the optical behaviors of the long-wavelength absorption peak, the phase shifts at the SiO₂-air and SiO₂-W interfaces (i.e., φ₁₀ and φ₁₂, respectively) and the phase shift accumulated along the propagation path (φ_t) of the absorbers with t_SiO2 = 229 nm at λ = 9.5 µm and t_SiO2 = 947 nm at λ = 10 µm are calculated and plotted in Fig. 4 for both s- and p-polarizations. The incident angle ranges from θ = 0° to 60°. The total phase shift, i.e., φ_T = φ₁₀ + φ₁₂ + φ_t, is also plotted. φ₁₀, φ₁₂, φ_t, φ_T as well as θ are defined schematically in the inset of Fig. 4(c). For light propagating in the SiO₂ film, φ_t = 2•2π•n_SiO2•t_SiO2/λ holds for both polarizations. The phase shift at the SiO₂-air interface φ₁₀ are actually reflection phase, which are obtained from the Fresnel formula (r_10s and r_10p) below for both s- and p-polarizations, i.e.,

 

Fig. 4. Calculated phase shifts at the SiO₂-air interface (φ₁₀) and at the SiO₂-W interface (φ₁₂), the phase delay in the SiO₂ film (φ_t), and the total phase shift (φ_T) of the SiO₂/W double-layer absorber with (a,b) t_SiO2 = 229 nm at λ = 9.5 µm and (c,d) t_SiO2 = 947 nm at λ = 10 µm for (a,c) s-polarized and (b,d) p-polarized light with different incident angles from θ = 0° to 60°. Light propagation in the absorber is schematically shown in the inset of (c).

Download Full Size | PDF

Figure 4 shows that φ₁₀, φ₁₂, φ_t, and thus φ_T of the SiO₂/W double-layer absorbers with t_SiO2 = 229 nm at λ = 9.5 µm and with t_SiO2 = 947 nm at λ = 10 µm all remain almost unchanged at different incident angles from θ = 0° to 60° for both s- and p-polarized light. This is likely due to the relatively high refractive index of SiO₂ at λ = 9.5 µm and λ = 10 µm (Fig. 1(b)). Such phase shift behaviors well explain the almost insensitive absorption peak to the light incident angle. For the thick absorber at λ ∼ 10 µm, the phase shifts at the top and bottom SiO₂ film boundaries as well as the phase shift of a roundtrip propagation compensate each other, leading to approximately zero total phase shift (i.e., φ_T ∼ 0) under oblique incidences for both polarizations (Figs. 4(c) and 4(d)). This confirms that this absorption peak is induced by interference resonance and is insensitive to the light polarization. For the thin absorber at λ ∼ 9.5 µm, the total phase shift is also almost independent of light polarization but is far from zero (see Figs. 4(a) and 4(b)) under different incident angles. Therefore, resonance is not excited and the intrinsic material absorption is likely to induce that very small absorption peak.

For p-polarization, the peak at λ ∼ 8 µm becomes as high as 1 when θ = 60° for the absorber with t_SiO2 = 229 nm (FWHM = 0.69 µm; Fig. 3(c)) and when θ = 45° for the absorber with t_SiO2 = 947 nm (FWHM = 0.97 µm; Fig. 3(g)). At this wavelength, the real part of the relative permittivity of SiO₂ approaches zero (Fig. 1(b)), allowing us to infer that the Berreman mode is likely to be excited for the high absorption at λ ∼ 8 µm under oblique incidences. In order to further confirm our inference, we calculated the angle of refraction (θ_r) for p-polarization as a function of the angle of incidence (θ) when light with λ = 8, 9.5 and 10 µm illuminating on a semi-infinite SiO₂ substrate (Fig. 5), according to the following equations [31]:

 

Fig. 5. The correlations between the angle of refraction (θ_r) and the angle of incidence (θ) for p-polarized light at λ = 8, 9.5 and 10 µm incident on a semi-infinite SiO₂ substrate.

Download Full Size | PDF

As shown in Fig. 5, as θ increases, θ_r rises first quickly and then slowly for all light wavelengths. Different from the light with longer wavelengths, the light with λ = 8 µm from the air sees an optically thinner medium (n_SiO2 < 1; Fig. 1(b)), and thus θ_r is always larger than θ until θ increases to 81°. Especially when θ is larger than 45°, the refracted light propagates almost horizontally within SiO₂. This is another characteristic feature for the Berreman mode. When a SiO₂ film with a specific thickness is stacked on a W layer, the coupled light can be reflected by the metal with high reflectivity and bounced back and forth within the SiO₂ layer. At λ = 8 µm, only p-polarized light is coupled and possesses a longer propagation path as θ (and thus θ_r) becomes larger. This induces enhanced absorption though κ_SiO2 is not as large as those at λ = 9.5 and 10 µm. For the absorber with t_SiO2 = 229 nm, the absorption peak at λ ∼ 8 µm increases to nearly 1 when θ ranges from 0 to 60° (Fig. 3(c)). When t_SiO2 increases to 947 nm, θ does not necessarily need to be very large to enable a sufficiently large θ_r. Thus, at a smaller incident angle θ = 45°, the highest absorption peak occurs (Fig. 3(g)). Such polarization- and incident-angle-dependent optical behaviors are quite different from the absorption at λ ∼ 9.5 and 10 µm for the absorber with t_SiO2 = 229 and 947 nm, respectively (Fig. 3). The angular dependence is desirable in applications where omnidirectional performances are not necessary, e.g., thermal sensing and laser machining. According to Kirchhoff’s law [31], our absorber can serve as a thermal emitter. The angle-sensitive performance is beneficial for mitigating the influence of the surrounding environment.

To further demonstrate the different optical behaviors of the Berreman mode and the interference resonance, electric field distributions and time-averaged Poynting vectors at λ = (8, 9.5) µm and (8, 10) µm were simulated with FDTD and plotted in Fig. 6 for various values of incident angle θ. It is seen that for the absorber with t_SiO2 = 229 nm, under oblique incidence, the electric field at λ = 8 µm is tightly confined within the SiO₂ layer and the energy flows and dissipates slowly almost horizontally due to the excitation of the Berreman mode (Fig. 6(a)). The electric field is a little bit weaker in the 947-nm thick SiO₂ layer but the Poynting vector still keeps almost horizontally under oblique light at λ = 8 µm (Fig. 6(b)). With the greatly extended propagation length, the thin absorber (t_SiO2 = 229 nm; over three times thinner) can still have nearly 100% absorption as the thick one (Figs. 3(c) and 3(g); due to the intrinsic absorption of the SiO₂ layer at λ = 8 µm, the Berreman mode will be absorbed quickly while propagating horizontally). In contrast, the SiO₂ layer should be thick enough to get the interference resonance. Quite differently, at λ ∼ 9.5 µm, t_SiO2 = 229 nm is too thin and nearly no light in the MIR range is coupled or resonant (Fig. 6(c)), and consequently only leading to an extremely weak absorption peak (Fig. 3(c)). Figure 6(d) shows that the absorber with t_SiO2 = 947 nm is sufficiently thick and allows enhanced electric field confined in the SiO₂ layer in comparison with the thinner absorber at λ ∼ 9.5 µm. However, in this layer, the energy flows almost vertically from the top to the bottom, even under a large incident angle of θ = 60°. The much smaller Poynting vector arrows (in comparison with those in Figs. 6(a) and 6(b)) indicate that the energy dissipates much more quickly, leading to the resonant absorption peak at λ ∼ 10 µm (Fig. 3(g)).

 

Fig. 6. Electric field distributions of the SiO₂/W double-layer absorber with t_SiO2 = 229 and 947 nm under obliquely incident p-polarized light with λ = (8, 9.5) µm and (8, 10) µm, respectively. The considered incident angles are θ = 0°, 15°, 30°, 45° and 60°. The arrows over the electric field distributions represent time-averaged Poynting vectors.

Download Full Size | PDF

4. Conclusion

In conclusion, we have experimentally demonstrated a perfect MIR dual-band absorber based on a very simple lithography-free SiO₂/W double-layer nanostructure. A nearly 100% absorption has been obtained at λ ∼ 10 µm for a relatively thicker absorber with t_SiO2 = 947 nm, due to the constructive interference resonance. Such a resonant absorption peak is insensitive to both the light polarization and incident angle. Another enhanced absorption has been identified at λ ∼ 8 µm under oblique incidence. Both the FDTD and TMM modeling results have shown that such absorption enhancement is induced by the excitation of Berreman mode, where the refracted light propagates almost horizontally within the SiO₂ layer. Different from the interference-induced absorption, the Berreman mode induced absorption exists even for a thin absorber with t_SiO2 = 229 nm. This absorption peak is sensitive to the incident angle and the light polarization. The angular dependence of our absorber is beneficial for thermal sensing, thermal emitting with mitigated influence of the surrounding environment, laser machining, etc.