High-efficiency hybrid plasmonic metasurfaces

There has been a growing amount of attention to 2D metasurfaces since their introduction for manipulating the propagation of electromagnetic waves. Indeed, many researchers have explored various designs for the deflection of an impinging beam into anomalous refraction channels (according to the generalized Snell's law) by imparting a controlled gradient of phase discontinuities along the metasurface.^1–3 The thickness of these structures is far smaller than the operational wavelength, which allows the miniaturization and integration of various optical components and systems. In general, all these promising metasurfaces can be categorized as operating either in reflection mode or transmission mode. Although metasurfaces that operate in reflection mode exhibit relatively higher efficiencies than those in transmission mode,^{4, 5} the reflective metasurfaces also introduce inconveniences in many applications. In addition, there has been particular interest in ultrathin metasurfaces operating in transmission mode, but these metasurfaces are still in their infancy because of their low manipulation efficiency and extremely complex fabrication methods (especially for visible light manipulation).^6–8

The theoretical limit of cross-polarized transmitted light through 2D metasurfaces is only 25% of the total impinging energy.⁹ Realistic implementations of these surfaces have exhibited even lower efficiencies, on the order of a few percent. Meanwhile, it has also been reported that high efficiency is possible for transmission mode metasurfaces in the microwave or IR regions.^{10, 11} These designs, however, generally require sophisticated fabrication processes and unit cells. Such complicated designs result in great challenges for their application at visible wavelengths. Improved approaches to fill the wide gap between laboratory and practical applications are therefore still required and are being strongly pursued.

In our work we have proposed and experimentally demonstrated an efficient bilayer plasmonic metasurface that works in transmission mode at visible frequencies (see Figure 1).¹² Our results actually reveal an accidental parallel in the metasurface domain and support the argument that ‘two 2D materials are better than one.’¹³ The top layer of our configuration is a V-shaped 2D array of gold nanoantennas, and the bottom layer is a 2D array of its Babinet-inverted apertures. We marry these two metasurfaces together to form a bilayer metasurface. The entire structure is ultrathin, with a thickness of 130nm (about a sixth of the wavelength).

Figure 1. Efficient manipulation of visible light is achieved with a V-shaped bilayer metasurface. The polarization state of the impinging beam (E_y) can be converted into the cross-polarized (E_x) component and deflected into the anomalous direction, although part of the impinging beam also travels unaffected through the structure (normal beam, E_y). The light spots on the screen in the foreground are combined photoimages of the transmitted light beams for different wavelengths. The subunit cell of the bilayer metasurface is shown in the bottom right. It consists of a top layer of gold nanoantennas and a bottom layer of its Babinet-inverted pattern. These layers are separated by conformal hydrogen silsesquioxane (HSQ) pillars. L: Length of the nanoantenna arms. W: Width of the nanoantennas. θ: Angle between two arms of the V-shape. T: Thickness of the gold film. H: Height of the HSQ pillars. D: Nominal space between top and bottom layers.

A prerequisite in all previous metasurface designs has been the assumption that mutual coupling between adjacent subunits is negligible. In contrast, our compact design gives rise to strong intra-plane coupling among the unit cells and to strong inter-plane coupling between the layers. We exploit both this intra-layer and inter-layer coupling to enhance the overall manipulation efficiency. To design the geometry for a given wavelength, we therefore tailor the parameters of all subunit cells as a whole by considering the coupling of all neighboring elements. Our findings show that the generalized Snell's law still applies in the presence of such strong mutual coupling. The coupling yields a remarkably high conversion efficiency (36.5%)—see Figure 2—from a linear polarization to the orthogonal polarization, i.e., beyond that of a traditional single-layer metasurface. The 25% theoretical limit for cross-polarized conversion efficiency does not apply to our metasurface design because of its finite, yet small, thickness, which allows the radiation symmetries of ultrathin metasurfaces to be broken. Moreover, we have achieved a strikingly high extinction ratio of 12.7dB for the first time (which implies stronger anomalous light than the normal component).

Figure 2. High conversion efficiency performance of the optimized bilayer plasmonic metasurface.

A key feature of our bilayer metasurface design is its ease of fabrication (see Figure 3). We use electron beam lithography to pattern the structure on a quartz substrate with a layer (100nm thick) of hydrogen silsesquioxane. We subsequently use an electron-beam evaporator for gold deposition. In this way we manage to avoid the lift-off process. Although our design consists of two separated layers, this fabrication process is arguably simpler than for single-layer metasurfaces. Our one-step fabrication for the bilayer metasurface is more viable and cost-effective, and therefore largely beneficial for practical applications.

Figure 3. Schematic representation of the fabrication procedure for the bilayer plasmonic metasurface. (i) A 100nm-thick layer of HSQ is spin-coated onto a quartz substrate. (ii) Electron beam (e-beam) lithography is used for patterning. (iii) An electron-beam evaporator is used to deposit a 30nm-thick gold film on the sample. (iv) A scanning electron microscope image of the top view of the final bilayer metasurface. SU: Subunit.

With our ultrathin hybrid structure we also counter-intuitively overturn the widely accepted idea that dielectric metasurfaces outperform plasmonic ones (because of lower ohmic loss).¹⁴ Indeed, in our work we have shown that plasmonic metasurfaces are almost equivalent to dielectric metasurfaces, and that they perform even better in simulations. We obtain a high conversion efficiency (36.5% in simulation and 17% in experiment) and a high extinction ratio simultaneously for visible light.

In summary, we have designed and experimentally demonstrated a new metasurface configuration. In addition to providing all the benefits of plasmonic metasurfaces in terms of light manipulation with our hybrid bilayer metasurface, we also overthrow the theoretical efficiency upper limit of ultrathin plasmonic metasurfaces. We have thus shown that the common perception of the superiority of silicon metasurfaces is uncertain. Our hybrid metasurface potentially exceeds the performance of all previously demonstrated metasurfaces that operate in transmission mode at visible wavelengths (both dielectric and plasmonic). It should be possible to extend our mechanism to other frequencies by using different materials and structure parameters. We are now planning to use our bilayer configuration to design and demonstrate various metasurface devices (e.g., for waveplates, hologram imaging, and planar metalenses). Our one-step fabrication process eliminates the challenges previously associated with sample preparation. In conjunction with the high efficiency, our bilayer hybrid configuration may thus facilitate and enable the practical application of the metasurface concept.

We gratefully acknowledge financial support from the National Research Foundation of the Prime Minister's Office, Singapore, under the Competitive Research Program award NRF-CRP10-2012-04.