1. Introduction

Textured metallic structure with a subwavelength scale is known to strongly modify the electromagnetic response. This remarkable feature has potential in applications in both passive and active optical devices. In passive optical devices, a metallic film patterned with periodic subwavelength hole arrays can exhibit extraordinary optical transmission [1] which can be used in photolithography and near-field microscopy [2–4]. In active optical devices, the surface plasmon polaritons (SPPs), which play a crucial role in light-matter interactions, were proposed to modify the emission properties of a nearby active emission layer [5] and lead to more efficient light emitting devices [6]. Moreover, a dielectric cavity sandwiched between a perforated metal film and a flat metallic film has strong coupling effects via evanescent fields [7, 8]. The coupling interaction strongly depends on the thickness of the metal film [9]. When the thickness of the metal film is small, the coupling interaction is strong, and new SPP modes can be formed. Through tuning the coupling of the SPP modes, versatile plasmonic infrared (IR) emitters were made [10, 11]. This kind of the IR emitter with narrow bandwidth emission property is very useful in the studies of the reaction of biological systems, environmental protection, and industrial environment [12–18].

2. Device descriptions and fabrications

In this Letter we study the reflection properties of a plasmonic multilayer structure which consists of a relief metallic grating, a waveguide layer, and a metallic substrate as depicted in Fig. 1. This plasmonic multilayer structure can also be used as an IR emitter. The relations of the reflection and emission properties were investigated in details. The grating periodicity is denoted by Λ, the filling factor is f, the depth is denoted by t_g, and its complex dielectric constant is denoted by ε_m where ε_m=ε_mr+iε_mi. Between and above the grating grooves the dielectric constant is denoted by ε_Air. The thickness of the SiO₂ waveguide layer is assumed to be t_w and its dielectric constant is denoted by ε_w. The material of the metallic substrate is identical to the grating, ε_m. The substrate thickness is assumed to be infinite. The structure is illuminated with an incident plane wave at an angle θ_i. The input light is TM-polarized in which the magnetic vector is parallel to the grating grooves (i.e., parallel to the y-axis). The wavelength-dependent optical constants of the investigated materials are taken from ref. [19].

 

Fig. 1. Basic geometry of an IR emitter considered. It is a plasmonic multilayer structure consisting of a relief metallic grating, a waveguide layer, and a metallic substrate.

Download Full Size | PDF

The fabrication processes of the plasmonic multilayer structure are described as follows. An optically thick Ag metal film was deposited on the front side of a Si substrate followed by a SiO₂ layer deposited with the electron beam evaporator. Then, a grating structure of 100nm thick silver film was patterned onto the SiO₂ layer by deposition and lifted-off techniques. The geometric parameters of the fabricated system are as follows: Λ=3000nm, fΛ=1800nm, t_g=100nm and t_w=23nm. The reflectance spectrum of the device was measured by a Perkin Elmer 2000 Fourier transform infrared spectrometer (FTIR) system with a wave number resolution of 8 cm⁻¹.

3. Angular-dependent reflectance spectra of the plasmonic multilayer filter

Figures 2(a) and 2(b) shows the measured and simulated angle-dependent reflectance spectra. Darker shading represents lower reflectivity. Because of the limitation of the FTIR system, we can only measure the reflectance spectrum with unpolarized light with an incident angle from 12° to 65°. The TM-polarized spectrum was simulated by the Rigorous Coupled Wave Analysis (RCWA) method [20]. The simulation parameters of the plasmonic multilayer structure used to give a best fit are: Λ=3000nm, fΛ=1900nm, t_g=100nm and t_w=25nm. For normal incidence, there were three resonance dips occurring at 0.17eV, 0.38eV and 0.41eV. As the incident angle increases, the resonant dip at 0.41eV splits into two dips. At the same time, the resonant dips at 0.17eV and 0.38eV are independent of the incident angle. The simulated pattern agrees well with the experiment.

 

Fig. 2. Angle-dependent reflectance spectra of the IR emitter. (a) measured. (b) RCWA simulation. The geometric parameters used in the simulation were: Λ=3000nm, fΛ=1900nm, t_g=100nm and t_w=25nm.

Download Full Size | PDF

First, we discuss the resonant dip at 0.41eV for normal incidence. This dip corresponds to the grating-coupled SPPs. Typically, the dispersion curves reveal the resonances due to the excitation of SPPs at either the SiO₂/Ag (SA) or Air/Ag (AA) interfaces of the metal film. The resonant dip of the dispersion is denoted according to the conservation of momentum for SPPs: k_SPP=k_x+G, where k_SPP is the wave vector of SPPs; G=2πm/Λ denotes the reciprocal lattice vectors of the grating; m is an integer. The simulated dispersion reveals the grating-coupled SPP AA (AA[+1], AA[-1] and AA[-2]) modes with sharp resonant features. In our measurements, the contrast of the grating coupled modes are low owing to that the probe beam is a converging beam with a numerical aperture (NA) smaller than 0.2.

When the structure is an Ag grating on a SiO₂ substrate with infinite thickness, the field strength of the SPP decays exponentially away from the SA interface, as shown by the dotted line in Fig. 1. On the other hands, when the lower part of the SiO₂ substrate is replaced by a flat Ag layer, leaving an SiO₂ slab between the Ag grating and the Ag substrate, the SPP modes localized at the top and bottom of the SiO₂ slab become coupled. At this time, the field distribution at the sandwiched SiO₂ layer can be seen as a linear combination of two exponentially decaying SPPs of the two SA interfaces. Therefore, new SPP modes can be formed within the Ag/SiO₂/Ag (ASA) interface. Except the grating-coupled modes, there were some additional resonant reflection dips shown in Fig. 2. These resonant dips show very little angular dispersion. Especially for the dip at 0.17eV, which is angular-independent for almost the entire range of incident angles. The dips at 0.27eV and 0.5eV exist for a larger incident angle while the dip at 0.38eV exists for a small incident angle. This angular-independent behavior implies that the group velocity in x-direction, ∂ω/∂k_x is zero. This is because that the SPP modes, which are localized SPP (LSPP) modes, also have a standing wave character in the x-direction as we shall illustrate below.

To understand the nature of these resonances, it is instructive to investigate their field distributions. As shown in Fig. 2, there were four resonant reflection dips, respectively located at 0.17eV, 0.27eV, 0.38eV and 0.5eV, which were almost independent of the incident angle. The resonant dips display different resonant behaviors versus the incident angle. Fig. 3(a) shows the H_y ² distribution along the x-axis at the grating/SiO₂ interface for the four LSPP modes at 0.17eV, 0.27eV, 0.38eV and 0.5eV, respectively. From the nodal structures, we can identify them as the n=1, n=2, n=3 and n=4 Fabry-Perot cavity modes, respectively. Fig. 3(b) shows the H_y ² distribution (averaged over the period in x) along the z-axis under the Ag ridge of the grating for the four LSPP modes. The dips at 0.17eV (square) and 0.38eV (triangle) are calculated for an incident angle of 0° while the dips at 0.27eV (circular) and 0.5eV (star) are calculated for an incident angle of 89°. From Fig. 3(b), it can be seen that the LSPP modes were resonant along lateral direction under the Ag ridge with an effective cavity length of β(λ)Λf/k ₀, where β(λ) is a wavelength dependent propagation constant. For a deceasing t_w/λ, β(λ) is increasing. The field strength decays exponentially first and then grows exponentially for 0nm< z <100nm. In the waveguide region (100nm< z <125nm), the field strength is enhanced significantly for the 0.17eV mode. In the Ag substrate region, the field strength exponentially decays again, as it should.

 

Fig. 3. Magnetic field strength, H_y ², (a) along the x-axis at the center of the Ag ridge (b) along the z-axis at the grating/SiO₂ interface. The dips at 0.17eV (square) and 0.38eV (triangle) are calculated for an incident angle of 0o while the dips at 0.27eV (circular) and 0.5eV (star) are calculated for an incident angle of 89°. The geometric parameters were: Λ=3000nm, fΛ=1900nm, t_g=100nm and t_w=25nm.

Download Full Size | PDF

4. Thermal radiation of the IR emitter

Finally, the thermal radiation of the plasmonic multilayer structure was measured by heating the sample with a DC current in a chamber with a pressure of 3 mTorr. The device area is 1×1cm². The structure is a SiO₂ cavity sandwiched between a one-dimensional relief metallic grating and a flat metallic film. The thermal radiation was collected by a 45° off-axis concave mirror with a NA of 0.1 and then reflected into the FTIR system. The solid lines shown in Fig. 4 were the measured thermal emission spectrum of the IR emitter which was heated up to 220°C (red line) and 260°C (black line), respectively. The spectra show a sharp emission peak at 7.2μm which was very different from a typical blackbody emission spectrum. Nevertheless, the emission spectrum can be predicted by investigating the angular-dependent reflectance spectrum. For the case that λ>>Λ, the absorptivity of the device is (1-reflectance). According to the Kirchhoff’s law [21], the emissivity is equal to absorptivity. Therefore, the emission spectrum of the plasmonic structure can be predicted by multiplying the inherent thermal radiation profile and the angular-dependent absorptivity[22]:

where R(λ,θ) is the reflectance of the proposed structure at an incident angle, θ. In this paper, TR(λ) is taken to be the thermal radiation curve without the grating structure. Previously, Kreiter et al. also performed emission measurements related to surface plasmons on a grating.[23] Other related works can be found in Refs. [24–26]. Using the simulated R(λ,θ), the emission spectrum can be calculated from Eq. (1). Here, R(λ,θ) has been shown in Fig. 2(b). The dashed lines shown in Fig. 4 are the simulated thermal radiation spectrum of the IR emitter which was heated up to 220°C (red line) and 260°C (black line), respectively. The simulated results show a fair agreement with the measurement. However, the sideband of the measured radiation spectrum was higher than the simulated one. This sideband might come from the heating plate. Furthermore, the dielectric function of the materials involved could also change at high temperature, which would lead to discrepancies between theory and experiment. Comparing to Fig. 2(b), the emission peak coincides with the angular-independent reflection dip which corresponds to an LSPP mode mentioned above. It is worth mentioning that the thermal emission light inside the SiO₂ cavity suffers from the total internal reflection (TIR) for a grating-less case. However, for our proposed structure, the emitted light is coupled into the LSPP mode and then re-radiates into the free-space. At this time, the emission efficiency is not limited by the TIR. This implies that an angular independent localized SPP mode can lead to an enhanced emission peak.

 

Fig. 4. (a) Simulated (dashed line) and measured (solid line) emission spectrum of the IR emitter for 220°C (red line) and 260°C (black line), respectively. The geometric parameters were: Λ=3000nm, fΛ=1900nm, t_g=100nm and t_w=25nm.

Download Full Size | PDF

5. Conclusion

In summery, the reflection and emission properties of an IR emitter, which is a plasmonic multilayer structure consisting of a relief metallic grating, a waveguide layer, and a metallic substrate, are investigated both experimentally and theoretically. The resonant grating-coupled and localized SPP modes are identified. An LSPP mode, which is angular-independent in almost the full range of incident angles is observed. The relations of the reflection and emission properties are investigated in details. The emission peak coincides with the angular-independent reflection dip simulated by the RCWA method. The thermal radiation spectrum of the structure can be predicted by investigating the reflectance properties. We demonstrate that this method is useful for the design of an effective LED light extraction structure.