1. Introduction

Gradient metasurfaces leverage advances in nanophotonics to modulate the phase of an electromagnetic wave that can result in optical components with a compact flat form factor [1–4]. To engineer the wavefront, the wave interacts with nano-tokens that are on the scale of tens of nm [5]. To date, a variety of optical functions have been demonstrated including dual polarity lensing [6], generation of optical vortex beams [7], efficient coupling of propagating waves to surface waves [8], quarter [9] and half [10] waveplate phase retardation, and axicon lensing [11].

However, metasurfaces still face some roadblocks before they can be widely implemented in consumer devices. In the current work, we focus on the issue of durability and preservation of performance over time for metal-based metasurfaces. This is a particularly important challenge for gap plasmon resonators [12–14] due to the risk of metal oxidation over time [15]. The oxidation can cause rapid decrease in optical performance and sample degradation. Organic polymers such as Poly(methyl methacrylate) (PMMA) have been used to preserve metallic surfaces against oxidation and contamination [16,17]. However, the introduction of a polymer layer may shift the plasmonic resonances which rely on the surrounding refractive index of the metallic nanostructures [18,19]. Here we study the effects of a dielectric coating used to protect metasurfaces from mechanical damage and oxidation as illustrated in Fig. 1.

 

Fig. 1 Unprotected samples degrade with time due to oxidation of the metal nano-tokens causing a drop in diffractive efficiency. This loss in efficiency can be prevented via the use of a dielectric coating that protects the structures.

Download Full Size | PDF

In particular, we are investigating the benefit of this coating on metal-insulator-metal metasurface gratings used in the visible regime [20,21]. We discuss the challenges of designing and fabricating the coated structures while preserving high efficiency in the diffraction order of interest over a broad range of angles of incidence [22,23]. Finally, we evaluate the effectiveness of the coating by comparing the change in efficiency over time for a coated and an uncoated sample.

2. Simulation and modeling

For the current work, our goal is to design a metasurface with no geometric features less than 50 nm, coated with a dielectric, that provides a linear phase response mimicking a 1200 lp/mm reflective grating. The grating is constrained to have high efficiency (~50%) in the first diffractive order at 650 nm. Furthermore, it is important that the grating efficiency does not drastically vary (flat efficiency response) as a function of the angle of incidence (AOI) θ_i for AOI ranging from 0 to 50 degrees. This feature is particularly important if the grating is to be used in an optical system with a wide field-of-view.

The target grating period spacing of 1200 lp/mm corresponds to a periodicity L_x of 834 nm. The linear phase response is simulated stepwise by a sequence of rectangular nano-tokens within a single unit cell (one period) each of which has its own amplitude and phase response [21]. The -π to π phase range is achieved based on the metal-insulator-metal geometry that consists of nano-tokens above a metallic back plane. Here, we use SiO₂ for the insulator and Ag for the back plane and the nano-tokens. PMMA was used for the coating material in order to protect the sample from oxidation and contamination [18,19].

The number of tokens within the unit cell determines the number of steps in the stepwise linear phase response. For instance, for a period with three tokens there are three steps in the phase response across a single unit cell with the distances between the steps being 2π/3. Splitting each unit cell in three square regions (one for each token) we get L_y = 278 nm (one third of L_x = 834 nm) as shown in Fig. 2(a). The main degrees-of-freedom to achieve the desired stepwise linear phase response are the x and y dimensions of each token within the unit cell as well as the thicknesses of the metal and dielectric layers. A commercial-grade simulator based on the finite-difference-time-domain method [24] was used to perform the parameter sweep calculating the amplitude and phase response of tokens with varying x and y dimensions and normal AOI illumination. A set of three tokens were chosen to approximate a linear phase ramp from -π to π across each unit cell of the structure (see next paragraph for more details about the selection process). During the design process, we observed that we can remove the third token (which had very small dimensions) while maintaining the same efficiency response resulting in even simpler two token design (the “empty region” still serves its purpose as part of the linear phase ramp giving a relative phase response of zero). This was particularly beneficial for fabrication purposes. The final token dimensions are shown in Fig. 2(a). In Fig. 2(b) we compare the predicted phase response of the structure (red dashed curve) to the phase ramp of a perfect 1200 lp/mm blazed grating (blue solid line).

 

Fig. 2 (a) The unit cell for the final designed coated metasurface – top x-y and side x-z view (diagram) including dimensions in nm (table); (b) the modeled phase response (red dashed curve) across the x-axis over four periods compared to the phase response of a perfect 1200 lp/mm blazed grating (blue solid line); the red squares identify the position of the tokens; (c) SEM image of the fabricated sample.

Download Full Size | PDF

It should be noted that there are various sets of three tokens that can provide a linear phase across the structure with a high efficiency for normal AOI. However, it is challenging to achieve a high efficiency for a large range of AOIs. In our workflow we first chose a set of three tokens approximating a linear phase based on the parameter sweep of a single token at normal AOI. We then modeled the complete unit cell using those three tokens for various AOIs. The previous process was iterated until a variation of less than 5% was achieved across 0-50 degrees.

Furthermore, it was observed that the addition of a thin dielectric coating results in a peak of the efficiency around the Littrow AOI (~23°) and a decreased efficiency away from that region. To explain the change in efficiency when the PMMA coating is added, we observe that the surface plasmon resonance which give rise to the enhanced 1st diffractive order efficiency, are dependent on the refractive indices of both materials at a metal-dielectric interface as shown by the following equation for the surface plasmon propagation constant [25]:

We found out that the severity of the peak in efficiency around the Littrow angle can be tuned by changing the thickness of the coating. Thus, by varying the thickness of the coating and the dimensions of the tokens we achieved a design that provides a relatively high efficiency, uniform over a large range of AOIs. A coating thickness of 80 nm was chosen for the final design resulting in a relatively flat efficiency response of 45 ± 3% over a 0-50° AOI.

3. Fabrication and testing

The metasurface described above was fabricated using standard electron beam lithography (EBL) and lift-off pattern-transfer techniques. First, a 130 nm-thick Ag film was deposited on a 4-inch Si wafer using e-beam evaporation (PVD-75 Lesker, base pressure is ~1 × 10⁻⁶), followed by an 80 nm of SiO2 and a 30 nm of Ag film on top. Secondly, a bi-layer positive resist (495 A2 and 950 M2) was spin coated on the substrate and then exposed to the pattern designed using an EBL (JEOL9500) tool with an acceleration voltage 100 kV and area dose of 1000 μC/cm². After EBL, the bi-layer resist [26] was developed for 45 seconds in a mixture of isopropanol (IPA): methyl isobutyl ketone (MIBK) of 3:1. The Ag tokens were formed by e-beam evaporation of 30 nm of Ag followed by a lift-off process (2 hours soak in acetone and 10 seconds sonication). For adhesion purposes, a thin layer (3 nm) of Cr was buried below each metal layer. The fabricated metasurface pattern was imaged using scanning electron microscopy (SEM) as shown in Fig. 2(c). As evident by the SEM image, the fabrication quality of the resulting metasurface is in good accordance with the initial design and the rectangular shapes of nano-tokens are well formed. A thin layer of PMMA was applied by spin coating on the sample. The thickness of protection layer is controlled by the selection of the appropriate PMMA molecular weight and spin speed.

The optical setup used to measure the efficiency of the manufactured grating is presented in Fig. 3(a). The sample was illuminated with a pseudo-collimated beam from a broadband laser source (Koheras SuperK Versa). The laser beam was filtered through a 10-nm bandpass filter centered at 650 nm. A linear polarizer was used to achieve the desired TE polarization. The sample was overfilled to minimize spatial intensity variations across the structure representing well the plane wave illumination used in the simulations. Due to the overfill it was required to evaluate the exact amount of power that arrives on the diffraction grating at normal AOI. We used a knife edge measurement to experimentally measure the Gaussian beam width and then derived the corresponding power based on the size of the sample. This derivation was only valid at normal AOIs, as for larger AOIs the beam footprint on the sample changes shape (becomes more elliptical), resulting in different power arriving on the grating. We used an analytic model to account for this change and to derive the accurate power input for each AOI of interest.

 

Fig. 3 (a) The experimental setup used to measure the diffraction efficiency; θ_lim is the angular region where the input and output arm collide and θ_i is the angle of incidence measured from the normal of the surface; (b) comparison between simulated η_s (circle/square) and measured η_m (star/rhombus) efficiency of the first diffractive order for the samples (with/without coating) at 650 nm; the bottom inset shows the absolute difference between the simulation and measurement for the samples with/without coating (upward/downward pointing triangle); (c) the measured efficiency at week 1 (circle/square) and week 25 (upward/downward pointing triangle) for the sample with/without coating; for all results the measurements error bars were smaller than the shown markers; the blue color symbols in (b) and (c) corresponds to the sample with coating where the red data is for the sample without coating. An x-axis break was used in (b) and (c) to emphasize the regions of interest where measurements were performed (the region left out corresponds to the regime where the input and output arm collide in the experimental setup).

Download Full Size | PDF

A custom kinematic mount for the sample was built to allow illumination at various AOIs between −90 and 90 degrees. Furthermore, the mount included a collection arm rotating concentrically around the sample’s center of rotation (the arm containing the detector in Fig. 3(a)). This allowed measurements of the diffracted power at any angle of diffraction. For this work, the sample’s efficiency was measured for AOIs from 0 to 50 degrees. Measurements in a region θ_lim around the Littrow angle are not possible as the input and output arm collide in that regime. AOIs larger than 50 degrees produce unreliable measurements given that a very small amount of the input illumination hits the sample causing low signal to noise ratio (as previously mentioned, the beam footprint on the sample becomes highly elliptical for larger AOIs). The efficiency of the first diffractive order was measured for each applicable AOI. For both the input and the output beams, all power measurements were done using a Si photodiode and a pico-ammeter.

4. Results and discussion

The efficiency as a function of the AOI is presented in Fig. 3(b) for 2 different samples based on the previously discussed design – one with the 80 ± 2 nm PMMA coating and one with the same geometry but without any coating. Both efficiency curves are compared to the predicted modeled results and they are in good agreement. The coating thickness was confirmed using an ellipsometry measurement. The surface roughness of the PMMA layer was measured using a Zygo NewView interferometer. The achieved peak-to-valley roughness was 20 nm (<λ/30) with a RMS of only 1 nm.

It was observed experimentally that the efficiency behavior at various AOIs is sensitive to the thickness of the coating film. Small deviations (~20 nm) cause the efficiency curves to vary significantly (up to 20%). This is consistent with the investigation made during the design of the metasurface grating. An important takeaway is that the coating effects on the diffraction grating efficiency are not negligible. They need to be accounted for during the design process and the thickness of the coating layer during fabrication needs to be closely monitored.

To evaluate the durability of both the coated and uncoated samples we repeated the efficiency measurement over the span of 25 weeks. During that time both samples were kept in standard room temperature and humidity. The efficiency at week 1 and week 25 are shown in Fig. 3(c). As it can be seen from the plots, the efficiency of the samples without coating drastically decreases over time dropping over 80% within 25 weeks. During that time, the sample with coating shows no significant change in the efficiency over all AOIs. These results clearly indicate that metal-dielectric metasurfaces require coating protection to prevent metal oxidation and sample degradation over time. A thin (<100 nm) PMMA film provides the demonstrated protection and sample stability.

5. Conclusion and future work

We demonstrated the design and fabrication of a 1200 lp/mm reflective metal-dielectric metasurface grating with ~45% efficiency in the first diffractive order at 650 nm illumination. The metasurface design was optimized with an 80 nm PMMA coating to protect against metal oxidation and sample degradation. A time laps comparison was performed with one sample with and one without coating. It was shown that over 25 weeks the sample without coating lost over 80% of its efficiency in the first diffractive order, while the coated sample does not show any drop in efficiency or sample degradation. These results demonstrate that durable coated metasurfaces with custom phase response can be designed and fabricated, which lays the path for the integration of such devices in consumer optical systems. In our future studies we will further investigate the interaction of the coating with the surface plasmon resonances and the corresponding effect on the efficiency of the metasurface grating.