1. Introduction

Absorbers have found significant interest in research as it has many applications in energy harvesting, sensing and photonics [1–4,14,20,27,34]. Noble metals such as silver, gold and copper are excellent reflectors at infrared and lower frequencies, but exhibit plasma resonances at visible frequencies [5–7]. Absorbers having excellent absorption in a narrow wavelength range can be easily obtained, but they reflect a large amount of incident light over a wide wavelength range. However, in some specific fields, such as solar energy collection and photon detection, it is desirable to have perfect absorption in a wide spectral band. This wavelength-sensitive characteristic severely limits the application of absorbers in the solar energy collection area. To overcome this limitation, different approaches have been proposed, such as mixing multiple resonators [8–12] within the same unit cell, exciting phase resonances to divide an absorption band into multiple sub-bands [13–15]. Heterogeneous metamaterial waveguides block light [16–18], using high-loss materials as absorbing components [19,20], and others [21–24]. For example, Wu et al. designed a new periodic Al groove structure coated by SiO₂ thin film and investigated numerically for wide-angle and polarization-insensitive broadband absorption [24]. He et al. proposed a dual-band metamaterial absorber made of two stacked metallic cross resonators and a lower metallic ground plane, separated by an isolation material spacer [25]. Cheng et al. presented wide-angle polarization independent infrared dual band absorbers based on metallic multi-sized disk arrays [26]. Recently, Yang et al. presented a broadband infrared absorber consisting of stacked double Cr ring resonators on a Cr mirror [12]. One important application of absorbers is thermal radiation [10,27]. Tunable thermal emission using metal-insulator-metal (MIM) structures, where a continuous Ge2Sb2Te5 (GST) film was selected as the middle layer, had been proposed [27]. Evidently, it is generally more difficult to fabricate the structures with complex resonant elements with good abilities in optical broadband absorption.

Glancing angle deposition (GLAD) has been demonstrated as an excellent technique for fabricating plasmonic nanostructures due to several advantages, such as scalability, capability of tuning geometries of nanostructures and compatible with microfabrication [28–31]. This GLAD method is simple and the nanostructures fabricated generally have large areas greater than 5 cm × 5 cm. The main mechanism of the GLAD is geometric shadowing effects, i.e. materials will not deposit in the shadows created by existing structures [32]. Prepatterned substrates can be used as shadowing templates to produce ordered nanostructure arrays via GLAD [33]. Especially recently, self-assembled hexagonal close-packed (HCPSM) microsphere/nanosphere monolayers have been used as these shadowing templates to create a variety of plasmonic nanostructures. Our previous studies showed that when depositing a thin Ag layer onto the nanospheres of HCPSM at a large incident angle, different shaped and sized patchy would form due to different orientations of the small domains in the HCPSM. Those Ag patches would have localized surface plasmon resonance (LSPR), resulting in absorbance peaks as we demonstrated in [34]. If these different shaped and sized patches can be stacked vertically, and isolated by an insulation layer, it is expected that the LSPR absorbance at different wavelengths can overlap together to form a broadband absorber, which is a basic design method to realize broadband absorption. Glancing angle vapor deposition on hexagonal close-packed (HCP) monolayers can produce patterned patchy films, where various patch shapes depend on the orientation of monolayers with respect to the vapor directions. The patches are mainly distributed on one side of nanospheres, which face the incident vapor. In this work, we employed the GLAD method to fabricate Ag/SiO₂ stacked plasmonic layers (SPLs) with broadband optical absorption property. It was demonstrated that the nanostructure for 750nm spheres is capable of absorbing more than 90% of the incident light over the spectral region ranging from 350 nm to 850 nm.

2. Experiment

2.1 Materials

Silver (99.999%) pellets were purchased from Kurt J. Lesker Co., Ltd. (USA). Polystyrene (PS) nanospheres (Huge Biotechnology) were used to fabricate monolayer templates. Sulfuric acid (≥ 98%) and hydrogen peroxide (≥ 30%) were acquired from Sinopharm Chemical Reagent Co., Ltd. (China). Ultrapure water (≥ 18.2 MΩ) was used in all experiments.

2.2 Fabrication of PS sphere monolayers

Colloid monolayers were firstly fabricated using an air−water interface method reported previously [35]. Briefly, glass slides (CITOTEST), cut into 2 cm × 2cm, were cleaned in hot piranhas solution at a 4:1 ratio volume of sulfuric acid and hydrogen peroxide for at least 20 minutes. After chemical cleaning, all substrates were ultrasonically washed in ultrapure water and then blown dry with nitrogen. Then, the diluted polystyrene sphere suspensions with different diameter (d = 300,500,750 and 1000 nm) were injected at a rate of 0.015 ml/min via a syringe pump (KD Scientific) onto a surface of a tilted clean glass Petri dish (10 cm in diameter) containing about 24 ml of ultrapure water. The process continued until the monolayer formed and covered the entire liquid surface. A Teflon ring was then gently placed on the water surface to protect the monolayer from sticking to the sidewalls of the glass culture dish. The glass substrates were carefully slid into the area under the monolayer film. The films were deposited on the surfaces of the glass substrates when a peristaltic pump (Buchler Instruments) slowly removed water from the dish.

2.3 Fabrication of Ag/SiO₂ stacked layer

Figure 1(a) shows that GLAD deposition configuration with respect to HCPSM orientation. Here the vapor incident angle was fixed at θ = 75°, but the relative azimuthal angle φ was changing. For a fixed incident angle θ, if φ is different, the formed patches on PS spheres will be different. Since the formed HCPSM has domain distributions, for the deposition with fixed θ and φ, the patches on different domains will vary. The morphology of nanopatterns can be predicted by numerical calculation using an in-house Matlab code [35]. The purpose of this morphology calculation is to provide an estimation of the geometric shapes of the deposited nanostructures, which serves to guide the design of desired structures. We can conclude the following conclusions: (1) An oblique angle deposition (OAD) at large θ₀ creates isolated nanostructures coated on nanospheres. (2) The shape of these isolated nanostructures fabricated at a fixed θ₀ strongly depends on the domain orientation φ₀. (3) The depositions at φ₀ and 60°-φ₀ produce the same shape of nanostructures with reflection symmetry. For the design of broadband absorbers, we wanted to fabricate silver patches of different size and shape on the nanospheres.

 

Fig. 1 (a) Illustration of the HCP lattice of a colloidal domain, showing the definitions of the polar and azimuthal angles, θ and φ. (b) Illustration of seven deposition steps:① Ag, SiO₂, Δ φ = 0°, ② Ag, SiO₂, Δ φ = 65°, ③ Ag, SiO₂, Δ φ = 130°, ④ Ag, SiO₂, Δ φ = 195°, ⑤ Ag, SiO₂, Δ φ = 260°, ⑥ Ag, SiO₂, Δ φ = 325°, and ⑦ Ag, SiO₂, Δ φ = 390°. (c) Top-view figures at each step of the expected structures.

Download Full Size | PDF

In order to construct similar shaped patches stacked on each sphere, according to the HCP lattice symmetry as well as the reflection symmetry, we rotated the substrate azimuthally every Δ φ = 65° to deposit Ag and SiO₂ layers till an accumulated Δ φ = 390° is rotated, i.e., a relative Δ φ = 30° is rotated. More specially, silver and SiO₂ layers were deposited on the HCPSM by using a double-source electron beam evaporation coating system (DE500, DE Technology Inc). The deposition procedures followed those shown in Fig. 1(b). The Ag and SiO₂ vapors were then deposited alternately on the monolayers with the polar angle θ = 75° with respect to the substrate normal. Specifically, we defined Δ φ as a relative change in the orientation of the substrate relative to the original position. The sequence of depositions could be described by: ① Ag, SiO₂, Δ φ = 0°, ② Ag, SiO₂, Δ φ = 65°, ③ Ag, SiO₂, Δ φ = 130°, ④ Ag, SiO₂, Δ φ = 195°, ⑤ Ag, SiO₂, Δ φ = 260°, ⑥ Ag, SiO₂, Δ φ = 325°, and ⑦ Ag, SiO₂, Δ φ = 390°, where each Ag or SiO₂ layer had a nominal thickness of 30 nm. During the deposition, the chamber pressure was maintained as 4 × 10⁻⁷ Torr and the deposition rate was kept as 0.1 nm/s. The shapes and coverages of the Ag coating on nanospheres are determined by both the θ and the φ. For a monolayer domain with initial φ = 0°, after each step’s deposition, the expected patches are shown in Fig. 1(c), i.e., with increasing Δ φ, the Ag patch shapes change from a near rectangle to a triangle. Following the procedures, samples with area sizes of 2 cm × 2 cm were obtained.

2.4 Characterization

The optical absorption spectra of HCPSM were characterized by a UV-vis-NIR spectrophotometer (Lambda950, PerkinElmer). Transmission was calibrated using a transparent glass substrate. For calibration of reflection, in our case, we measured the reflectance of a standard silver mirror (THORLABS Inc. Model: PF10-03-P01) whose reflectance was provided by the company. Measured reflection from nanostructures was then calibrated using the reflection spectra of the standard silver mirror. The morphology of the HCPSM substrate was examined using a scanning electron microscopy (SEM) (SU8010, Hitachi). The cross-sections of HCPSM were revealed by a transmission electron microscopy (TEM) (HT7700, Hitachi).

It was observed that the SPLs obtained by the above method had a HCP domain of PS spheres as determined by SEM. Due to the 60° rotation symmetry of HCP lattices, the azimuth angle φ can be expressed as φ = φ₀ + n × 60°, where φ₀ is an initial azimuth with 0° ≤ φ₀ < 60° and n is an integer. For θ = 75°, however, the shape and coverage of the Ag/SiO₂ coating at different φ are very different. As shown in Figs. 2(a)-2(g), the representative evolution of the patch shapes after each step deposition on d = 750 nm monolayer. A general trend can be found from Figs. 2(a)-2(g), as the number of deposition increases, the coverage of the Ag/SiO₂ coatings become more symmetrical. With the increase of the number of depositions, Ag/SiO₂ stacked above the spheres step by step, as φ has been changed. The coverage areas were also gradually increased, which eventually led to nearly entire covering on the spheres.

 

Fig. 2 (a-g) SEM images showing the various morphologies of the coatings for seven step depositions. (h) TEM images of SPLs on the d = 750 nm spheres.

Download Full Size | PDF

We measured transmission utilizing a spectrometer with a beam spot of 1.8 mm in length and 0.6 mm in width. An un-polarized light ranging from 350 nm to 850 nm was employed to measure the transmitted beams of the glass substrate as reference lights. The absorption data (A) of samples fabricated on each different diameter PS nanospheres were obtained by measuring both the reflectance (R) and transmission (T) spectra of three samples prepared under the same conditions and using the principle of conservation of energy, which dictates that A = 100% – R – T. Therefore, the final absorbance is obtained by averaging the absorptions of the three samples.

3. Results and discussions

We fabricated a series of [Ag/SiO₂] _N/PS samples with N = 1 ~7. Figure 2(h) shows the TEM images of SPLs on the d = 300 nm spheres. The cross-section TEM images demonstrate that the desired SPLs have been achieved. As can be seen from Figs. 3(a)-3(c), the measured absorption spectra at normal incidence for the samples with sphere diameters of d = 750 nm, 500 nm, and 300 nm, respectively. It is found that the [Ag/SiO₂] _N/PS sample with N = 7 has the best broadband absorption performance, with absorption higher than 90% over the whole measured wavelengths. From Figs. 3(a)-3(c), we can also see that although the [Ag/SiO₂] _N/PS samples with N < 4 are relatively inefficient in absorption at the long wavelengths (λ > 650 nm). By increasing the number of N, the absorption at the long wavelength range approaches gradually towards the unity. The broadband absorption is due to broad wavelength plasma resonances resulting from different sizes and shapes of Ag coating on polystyrene spheres.

 

Fig. 3 Absorption as a function of number of deposition N for (a) 750 nm, (b) 500 nm, and (c) 300 nm PS spheres, respectively. (d) Absorption (N = 7) as a function of wavelength for d = 1000 nm, 750 nm, 500 nm, and 300 nm PS spheres.

Download Full Size | PDF

We repeated the experiment under the same conditions for d = 1000 nm PS spheres. The corresponding absorption as a function of wavelength for deposition N = 7 is plotted in Fig. 3(d), where the curves conclude the cases of d = 300 nm, 500 nm, and 750 nm. From this figure, we can find that the absorption for the d = 1000 nm PS spheres becomes lower than the one for the d = 750 nm PS spheres. The reason is that the plasma resonance in the sample for d = 750 nm brings about effective permittivity and permeability, meeting well impedance matching with the surrounding. As a result, the reflective waves for the sample are fewest (Here, the reflectance curves for the four samples are not plotted). In another word, the incident waves almost enter the sample and are absorbed in great degree. Figure 4(a) shows the integrated absorption A_int as a function of N for different diameters of the PS spheres. Here, A_int is obtained over the wavelength range from 350 nm to 850 nm. This figure clearly reveals that the rule of gradual growth in the integrated absorption with increasing N is maintained greatly well for different-sized PS spheres. When N is less than 5, the absorption of the d = 750 nm PS spheres is obviously better than those of the d = 300 nm and 500 nm PS spheres. However, when N is more than 5, the absorption of the three samples is getting closer.

 

Fig. 4 (a) Integrated absorption A_int integrated over the wavelength range from 350 nm to 850 nm as a function of number of deposition for the sphere diameters of d = 300 nm, 500 nm, 750 nm and 1000nm, respectively. (b) Integrated absorption A_int (N = 7) integrated over the wavelength range from 350 nm to 850 nm as a function of size of PS spheres.

Download Full Size | PDF

Figure 4(b) shows their integrated absorption. The figure demonstrates more clearly that the sample with d = 750 nm achieves the best absorption. In addition, it is found that the increasing rate of absorption for the three samples gradually decreased with the increase of deposition times. When N is greater than 6, the absorption has reached saturation, which means that absorption will not increase indefinitely as N increases. Figure 4(b) also shows that the sample for d = 750 nm has the strongest integrated absorption of 93%.

4. Conclusions

In summary, we have developed a new plasmonic absorber with good absorption of light in the range of 350 nm to 850 nm. The absorber consists of Ag/SiO₂ stacked plasmonic layers on a self-assembly polystyrene sphere monolayer. The experimental results show that the absorber for 750 nm polystyrene spheres has an average absorption of more than 90%. Important features of the absorber are simple and large-area for fabrication, which indicates that the production costs greatly low and it is proper to practical applications in light cloaking and conversion.