1. Introduction

The total absorption of light by a structured metal surface had been predicted and demonstrated as early as 1976, by Hutley and Maystre, on a single, gold coated, sinusoidal interference grating [1]. Efficient absorption of light on nanoscale metal-semiconductor gratings fabricated on a Bragg mirror has also been demonstrated by Collin et al. [2]. More recent work by Bonod et al. on lamellar metallic gratings has demonstrated that total absorption of light for any polarization is possible if there is only the zeroth propagating order [3]. However, the absorption of light on grated surfaces still remains heavily dependent on the incident angle.

In 2005, Teperik and Popov theoretically calculated that a nanoporous metallic film could exhibit total light absorption at plasma resonance when there is optimal coupling of the plasmons in the voids to external light, and this can occur when the voids are buried at a specific distance under the metal surface [4]. In porous metals, the lattice of voids beneath the metal surface has two roles [5]. First, it forms a coupling element which diffracts incident light into surface plasmon-polaritons. Secondly, localized Mie plasmons are excited in the spherical voids. Since void plasmons are radiative excitations, they can couple to light directly without any special coupling device. These localized void plasmons can also be brought into resonant interaction with the propagating surface plasmon polaritons (SPPs) by tuning the structural parameters such as the diameter of the voids, and by filling them with dielectric materials.

It was further suggested that total omnidirectional absorption of light can be achieved in nanostructured metal surfaces that sustain localized optical excitations [6]. These nanoporous metal surfaces can simultaneously support delocalized SPPs and localized void plasmons, where the localized void plasmons are responsible for the omnidirectional absorption of light. It was demonstrated that almost total absorption of light can indeed be achieved with a lattice of voids just beneath the surface, and theoretically predicted that omnidirectional absorption is possible for a wide angle range (up to 40°). This range of angles can be further increased if the voids were filled with a dielectric medium (e.g. silica) because a dielectric medium can bring down the plasmon void energy to a region that it does not interact with the delocalized SPPs.

A separate work by Bonod and Popov mentioned that total light absorption by buried metal cavities can occur without the aid of surface plasmons under TM polarization due to the role of cavity resonance [7]. It stated that while an opening of the cavity is required to couple incident light into the cavity resonance, the opening has to be closed enough to sustain cavity resonance.

In our work, we have made fabricated a series of nanostructured gold surfaces with a layer of buried closed packed voids/spheres to experimentally investigate the effect of tuning the cavity size, the cavity medium as well as the polarization of incident light on the phenomenon of omnidirectional total light absorption on such surfaces. Almost total light absorption has been observed at multiple wavelengths when a monolayer of dielectric spheres is buried in the gold films at distances smaller than the skin depth. The strongly enhanced absorption is observed for angles of incidence up to 65° for p-polarized incident light, which makes our systems good candidates for high-efficiency absorber materials.

2. Results and discussion

It has been reported that a prerequisite of total light absorption by nanostructured gold surfaces is that the metal thickness must be adjusted in such a way that the radiative decay rate of the localized plasmons equals the rate associated with its dissipation [6]. We have previously reported that it is particularly easy to accurately control the thickness of a metallic deposit in a colloidal template thanks to the current density variations [8,9]. Indeed, the local minima/maxima of the current density can be attributed to the periodic variation of the electroactive surface area of the growth front during the gold deposition. Figure 1 shows SEM side views of gold films of well-controlled thickness, t, electrodeposited through a monolayer of polystyrene (PS) beads with a diameter, D, of 260 nm, 430 nm or 595 nm. It can be seen that the deposits are uniform over large distances, illustrating the homogeneity of the electrodeposition process. Another observed feature is that the film growth was slower near the edges in contact with the PS spheres, possibly due to diffusion constraints. This results in ‘billowing’ of the metal film between the spheres, as marked by arrows in Fig. 1 and illustrated in Fig. 2.

 

Fig. 1 SEM side views of the gold deposits with various thicknesses grown through a monolayer of 260 nm (left), 430 nm (middle) or 595 nm (right) polystyrene (PS) beads. Scale bar: 300 nm.

Download Full Size | PDF

 

Fig. 2 Sketch of the nanostructured gold surfaces, consisting of a monolayer of close-packed PS beads or voids of diameter D infiltrated with gold to a thickness t.

Download Full Size | PDF

The absorption (A = 1-R) spectra of the nanostructured gold surfaces for normal incidence of light are shown in Fig. 3. Interestingly, the spectra exhibit multiple main peaks (up to 4 for the 595 nm PS template) with an intensity value higher than 90%. These peaks correspond to the high-order plasma resonances, which can be observed here because of the reduction of their frequencies -which normally fall within the interband absorption spectra of gold- due to the presence of the PS beads, as demonstrated theoretically by Teperik et al. [4]. For D = 260 nm, almost total light absorption (ca. 98%) is observed at the wavelength of the fundamental Mie plasmon mode, i.e. 850 nm for an optimum value of t, found to be equal to 1.02 D. For the films prepared using the 430 nm (595 nm) PS spheres as template, 93% (95%) of light is absorbed at 1020 nm (1305nm) when t = 1.05 D. When the film thickness exceeds the optimum value, the resonant absorption peak shifts towards lower wavelength and decreases in intensity.

 

Fig. 3 Measured (solid lines) and calculated (dashed lines) absorption spectra of gold films of different thicknesses grown through a monolayer of 260 nm (top), 430 nm (middle) or 595 nm (bottom) PS beads under normal incidence.

Download Full Size | PDF

We have simulated the optical spectra of the gold structures by the FDTD method, taking into account the billowing of the metal film between the PS beads (see Materials and Methods). Figure 3 shows a nice matching between the positions of the predicted (dashed lines) and measured (solid lines) fundamental plasma resonance peaks, while the band intensities differ slightly, probably due to a non-optimised design of the ellipsoidal caps which represent the billowing of the metal between the PS spheres.

After dissolution of the PS template, porous gold surfaces containing a lattice of interconnected voids were obtained. Figure 4 shows SEM top views of the porous gold films, evidencing their uniformity over large areas and the existence of an aperture on the top of the films even when t is larger than D, which is a consequence of the ‘billowing’ of the metal film between the spheres.

 

Fig. 4 SEM top views of the porous gold films with various thicknesses after the dissolution of the 260 nm (left), 430 nm (middle) or 595 nm (right) PS beads. Scale bar: 300 nm.

Download Full Size | PDF

The optical spectra of the nanoporous gold films also exhibit a quasi total light absorption (Fig. 5). For instance, when D = 260 nm, 97% of the incident light is absorbed at 600 nm for t = 1.05 D, while 96% of light is absorbed at 670 nm (840 nm) for t = 1.08 D when D = 460 nm (595 nm). As previously, the absorption peak decreases in intensity and shifts towards lower wavelength when t exceeds the optimum value. These results are in good agreement with those obtained by Teperik et al. [6]. As theoretically predicted by these authors [6], the near total light absorption of light by the nanoporous gold films persists over a large range of angles of incidence (AOI) for both s- and p-polarized incident light, as shown in Fig. 5. For instance, more than 90% of the p-polarized incident light is absorbed for AOI values up to 45° when D = 260 nm.

 

Fig. 5 Experimental absorption spectra under normal incidence (left column) and incidence-angle dependence of absorption for p-polarized (center column) and s-polarized (right column) incident light of porous gold films with various thicknesses after the dissolution of the 260 nm (top), 430 nm (middle) or 595 nm (bottom) PS beads.

Download Full Size | PDF

Teperik et al. also suggested that this range of AOI can be further extended by infiltrating the voids with a dielectric medium so that the void plasmon is brought down in energy to a region where it does not interact with the delocalized SPPs. Figure 6 shows that this statement is experimentally verified for the gold films containing the PS beads, which absorb more than 90% of the p-polarized incident light for an AOI up to 65°, whatever the PS bead size.

 

Fig. 6 Incidence-angle dependence of absorption of gold films of different thicknesses containing a monolayer PS beads for p- (top) and s-polarized (bottom) incident light. D = 260 nm (left), 460 nm (middle) and 595 nm (right).

Download Full Size | PDF

3. Conclusion

In summary, we have fabricated nanostructured gold films containing an embedded layer of spherical PS beads or voids by electrodeposition. Near total absorption of light at normal incidence by gold films with an optimum thickness was experimentally and numerically demonstrated. We have shown that the gold films containing a monolayer of PS beads show light absorption larger than 90% for AOI values up to 65°, which makes the nanostructured gold films good candidates for low cost, high-efficiency absorber materials.

4. Materials and methods

Polystyrene (PS) spheres with a diameter, D, equal to 260 nm, 430 nm or 595 nm were synthesized by emulsion polymerization according to a reported procedure [10]. Close-packed monolayers of the PS spheres on gold-coated glass slides were fabricated by direct assembly of the particles at the air–water interface followed by their transfer onto the substrates [11]. After the formation of the template, the slides were used as working electrodes in a typical three-electrode cell with a platinum foil as counter electrode and an Ag/AgCl electrode as reference. A cyanide-free gold plating bath purchased from Metalor (ECF-60; gold concentration 10g.L⁻¹) was used as received for the metal deposition. The potentiostatic electrodeposition experiments were made in a water bath set at 25 ± 1°C. The intensity of the faradaic current generated from the Au-ion reduction was measured using an Autolab PGSTAT 20 potentiostat (EcoChemie) system monitored by a PC running the GPES 4.9 software. The dissolution of PS spheres was done by immersing the samples into tetrahydrofuran at room temperature for 2 hours.

The UV–vis–NIR spectra were recorded under normal incidence using a CRAIG 2020 microspectrophotometer. The incidence-angle dependence of the UV-vis-NIR reflection spectra was performed with a Lambda 950 Perkin-Elmer spectrophotometer equipped with the Universal Reflectance Accessory. The SEM experiments were carried out using a JEOL 6700F microscope operating at 5 kV.

The reflection spectra of the various experimental situations have been simulated by solving Maxwell equations using the three-dimensional finite-difference time-domain (FDTD) method, as implemented in the freely available MEEP software package [12]. The dielectric permittivity of gold was specified by using the Drude-Lorentz model with parameters reported in [13]. By Fourier transforming the response to a short, broadband, spatially extended gaussian pulse in the far-field of the structures and normalizing with the response of the reference for the same excitation conditions, a single simulation yielded the reflection spectra over a wide spectrum of frequencies. The reference reflective sample was taken as a 150 nm thick gold flat layer, mimicking the commercially available gold substrates used in this study. The structured sample was modelled as a close-packed monolayer of PS beads deposited on the 150 nm thick gold flat substrate layer and infiltrated with gold in the voids between the PS beads, up to a given thickness, t. To take into account the billowing of the metal film between the PS beads, ellipsoidal caps with different aspect ratios were set at the voids located between the beads.