1. Introduction

Plasmonic nanostructures such as one-dimensional (1D) [1–8] and two-dimensional (2D) periodic structures [9–12], nano-antennas [13–15] and metasurfaces [16–19] are drawing significant interest for various applications such as biological sensors [8,14,20], optical filters [7,21], and thermal infrared (IR) detectors [9–12] and emitters [2–5,17]. Simple and low-cost nanofabrication technologies are thus required to satisfy this emerging demand. Conventional nanofabrication techniques are based on top-down processes using electron beam (EB) lithography. EB lithography has sufficient resolution for nanofabrication; however, it is expensive and requires much longer exposure times than photolithography, which prevents practical mass production. Moreover, noble metals such as Au and Ag, which are common materials for plasmonic devices, are more difficult to shape by dry or wet etching than semiconductors such as Si and GaAs.

1D plasmonic nanogratings with high aspect ratios (larger than 10) and narrow grooves (~150 nm) are highly promising [6] for a wide range of applications, such as polarization-selective absorbers [6,11,19] for thermal IR polarimetric imaging [22,23] and IR emitters for gas sensing [24]. However, the fabrication of these structures remains challenging. Conventional dry etching processes are effective as long as the aspect ratio of the grating is low because the etched metals re-adhere to the side wall of the grooves and fill in the grooves, which prevents a high aspect ratio from being achieved. Metal deposition on semiconductor or oxide molds, which are initially formed as high-aspect-ratio nanogratings, results in filling of the nanogrooves. Therefore, stamp-like mold techniques have been developed [2,3,6] in which reversed nanograting structures are initially formed on other materials, such as semiconductors, and used as a mold. Metals such as gold and chromium are then deposited or electroplated onto the mold, which is removed to obtain the plasmonic nanograting. This method is appropriate if the developed structures do not need to be integrated on other devices or bonded onto other wafers. However, direct fabrication of nanogratings on a device wafer is critically important for practical device integration, such as for sensors and other electronic devices. In addition, incomplete etching of mold material at the bottom of deep grooves results in a shift of the absorption wavelength [6], but complete etching is difficult. Therefore, the direct fabrication of high-aspect-ratio plasmonic nanogratings using conventional top-down photolithography is in high demand.

In this study, a top-down fabrication technique was developed that uses conventional photolithography to achieve high-aspect-ratio plasmonic nanogratings as polarization-selective absorbers. The optical properties of these absorbers were investigated for thermal IR imaging applications. It should be noted that the absorbance is equal to the emissivity due to Kirchhoff’s law of thermal radiation. Thus, the absorbers studied here can also be used as IR emitters.

2. Design and theoretical calculations

Figures 1(a) and 1(b) respectively show oblique and cross-sectional illustrations of a 1D plasmonic nanograting with high-aspect-ratio and narrow grooves. The structural parameters of the groove period, depth and width are denoted p, d and w, respectively. The incident angle (θ) and the transverse-electric/transverse-magnetic (TE/TM) polarization are also defined.

 

Fig. 1 (a) Schematic illustration and (b) cross-sectional view of the 1D plasmonic-nanograting.

Download Full Size | PDF

The absorbance was calculated using rigorous coupled-wavelength analysis (RCWA). The optical constant of Au was taken from Ref [25]. The 1D plasmonic nanogratings investigated here were confirmed to absorb only TM polarized light, as demonstrated in previous research [2]. Therefore, in this section, only the results for TM polarization are presented.

First, the θ dependence of the absorbance was calculated, as shown in Fig. 2, where p, d and w were fixed at 2.0 μm, 1500 nm and 100 nm, respectively, to demonstrate the validity of the proposed fabrication method and for comparison with the highest previously reported aspect ratio of 13.3 and w of 150 nm [6].

 

Fig. 2 Calculated dependence of absorbance on θ. The color scale represents the absorbance.

Download Full Size | PDF

This calculation demonstrates that the absorption wavelength has no θ dependence and sufficient wavelength-selective absorption of over 90% is achieved at around 8 μm. These characteristics are strongly advantageous for IR thermal detector applications due to the high responsivity and wide viewing angle.

Next, the absorbance was calculated as a function of wavelength and d, with w fixed at 100 nm. Figures 3(a)–3(d) show the calculated results for p = 1.5, 2.0, 3.0 and 4.0 μm, respectively. These results clearly indicate that the absorption wavelength is almost proportional to d with a slight dependence on p. Therefore, the plasmonic resonance is produced mainly in the nanogroove depth due to the narrow w. Absorption occurs at wavelengths longer than p because diffraction is dominant at wavelengths smaller than p.

 

Fig. 3 Calculated absorbance as a function of wavelength and d for p = (a) 1.5, (b) 2.0, (c) 3.0, and (d) 4.0 μm. In all cases, w was fixed at 100 nm. The color scale represents the absorbance.

Download Full Size | PDF

There are two resonance modes in 1D plasmonic nanogratings. One is Fabry-Perót, or organ-pipe, resonance in the groove depth direction and the other is the horizontal resonance induced by the surface period. These modes are mixed, and their hybridization is controlled by the relation between d and p [3]. For smaller p, Fabry-Perót resonance is dominant, which causes resonance mode dispersion.

The full width at half maximum (FWHM) of the absorbance as a function of the wavelength decreased with increasing p. The FWHM is an important parameter for thermal detectors; therefore, the dependence of the FWHM on p was more thoroughly investigated. Figure 4 shows the calculated absorbance as a function of p with fixed d and w of 1500 nm and 100 nm, respectively. The FWHM increases with a decrease in p and the absorption wavelength is slightly dependent on p. When p is increased, the horizontal plasmonic resonance caused by the nanogroove period weakens the higher modes of Fabry-Perót resonance in the nanogroove depth direction [3]. Therefore, the FWHM increases with a decrease in p.

 

Fig. 4 Calculated absorption as a function of the wavelength and p.

Download Full Size | PDF

Figures 3 and 4 show that the wavelength and the FWHM can be controlled according to d and p, respectively, which sufficiently fulfills the requirement for polarization-selective IR detectors.

3. Fabrication

Figure 5 shows the procedure developed for the direct fabrication of high-aspect-ratio plasmonic nanogratings. Nanogratings with p = 1.5, 2.0, 3.0 and 4.0 μm, labeled as samples A, B, C and D, were fabricated on the same 3-inch (1 inch = 2.54 cm) Si wafer. Each pattern size was 1 cm².

 

Fig. 5 Direct fabrication procedure using tapered-side-wall molds. The thin Ti layer is omitted for clarity.

Download Full Size | PDF

i) 5-nm-thick Ti (adhesion layer) and 300-nm-thick Au layers were sputtered on a 3-inch silicon substrate. These layers were used to form the Au bottom of the grooves, considering that subsequent metal sputtering is insufficient to cover the bottom of the nanogratings. ii) A 880-nm-thick photoresist layer was coated and exposed to form a line-and-space pattern. The space width was 1.0 μm for all samples. For example, a 3.0 μm line width and 1.0 μm space width were used for sample D. The exposure light was defocused to form the tapered side wall of the photoresist. The taper angle can be controlled by changing the focus of the exposure light. iii) 10-nm-thick Ti and 980-nm-thick Au layers were then deposited by ion beam sputtering. The deposition rate increases from the bottom to the top of the photoresist due to the tapered side wall. iv) Finally, the side wall becomes vertical. The surface Au thickness is at least twice as thick as the skin-depth in the IR region, so that the influence of the photoresist under the Au layer is negligible.

Figure 6 shows top-view and cross-sectional scanning electron microscopy (SEM) images of samples A and B. The lines for groove width measurements and the measured groove shape profile in the depth direction are also shown. w and d were approximately 100 and 1500 nm, respectively. The parameter w = 100 nm was maintained over the 3-inch Si wafer. Plasmonic nanogratings with a high aspect ratio of 15 and a narrow width of 100 nm were thus successfully achieved.

 

Fig. 6 Top-view and cross-sectional SEM images for samples (a,b) A, and (c,d) C.

Download Full Size | PDF

4. Measurements

The polarization dependence of the absorbance was measured in air using Fourier transform infrared spectroscopy (FTIR) with a polarizer. During these measurements, the incidence angle and reflection angle were held constant at 15° due to the mechanical restrictions of the FTIR system used. The FTIR used in this study can be operated at wavelengths over 3 μm. The measurement results for the wavelength region from 3 to 15 μm are presented, which is the relevant wavelength region for IR sensor applications. The measurement results for samples A–D with TM and TE polarization are shown in Figs. 7(a)–7(d), respectively. The dip in reflectance at 4.3 μm is attributed to absorption by CO₂ in the air. The reflectance is lower at shorter wavelengths for the TE polarization because the measurement is a mirror reflection, so only the zero-order reflection was detected.

 

Fig. 7 Measured reflectance with TE and TM polarization for samples (a) A, (b) B, (c) C, and (d) D. (e) Calculated and measured FWHM as a function of p.

Download Full Size | PDF

Samples A–D have a dip in reflectance at almost the same wavelength (8 μm). The surface Au thickness of the samples is twice as thick as the skin depth, so that the transmission is zero. The absorbance can be obtained by 1 – reflectance. Therefore, the main reflectance dip at 8 μm directly corresponds to absorption due to plasmonic resonance in the groove depth direction. An absorbance of over approximately 90% was achieved. Additional dips at 3 and at 4 μm were observed in Figs. 7(c) and 7(d), respectively. The diffraction emerges at a smaller wavelength than the surface period [2]. These dips were thus attributed to diffraction. These experimental results are consistent with those obtained by calculation shown in Fig. 3. Therefore, the developed fabrication method is demonstrated to be effective to achieve high-aspect-ratio plasmonic nanogratings.

Figure 7(e) shows the calculated and measured FWHM as a function of p. The measured FWHM was larger than the calculated FWHM for all p. Figure 8 shows the calculated absorbance as a function of wavelength and w with fixed d and p of 1500 nm and 2.0 μm, respectively. The calculated results indicate that the FWHM is not proportional to w, but the absorption wavelength changed slightly with w. The developed samples exhibit structural fluctuations in w, as shown in Fig. 6. Therefore, the FWHM of the experimental results was larger than that of the calculated results. However, both the calculation and the measurement demonstrated that the FWHM can be controlled simply according to p. This tunability is a strong advantage for thermal IR detectors, particularly for analytical devices such as gas sensors [24].

 

Fig. 8 Calculated absorption as a function of wavelength and w.

Download Full Size | PDF

5. Conclusions

A direct fabrication method was developed for high-aspect-ratio plasmonic nanogratings that employs conventional photolithography and tapered sidewall structures. A high aspect ratio of over 15 and narrow grooves of 100 nm were realized for various periods. This method allows the integration of plasmonic nanogratings with other devices. Optical measurements indicate that a strong polarization-sensitive wavelength-selective absorption of more than 90% can be obtained, and the absorption wavelength is less dependent on the period, which corresponds to RCWA calculation results. Moreover, the FWHM can be tuned according to the period. Such controllability is highly advantageous for thermal IR absorber/emitter applications.

Although photoresist was used to produce the tapered molds in this study, other materials such as semiconductors or oxides could be used to improve the durability and heat tolerance of the molds. Therefore, the proposed tapered-mold method should be generally applicable to the fabrication of a wide range of devices. We expect that the results from this study will facilitate the mass production of a wide range of plasmonic nanodevices.