1. Introduction

The coupling of light to surface plasmons (SP)s in nano-hole arrays (NHA)s depends highly on the dielectric properties of the metal film and the supporting material in contact with the metal film [1]. A change in the refractive index of the dielectric material at the surface of the metal film gives rise to a shift in the resonance wavelength and is the basis for SPR sensing [2]. In multiplexed SPR sensing devices based on NHAs, a grid of NHAs is fabricated on to a single substrate, where each NHA has unique geometric properties and hence distinct optical properties. The miniaturization of multiplexed NHA devices inevitably requires that NHAs be fabricated with smaller size and closer inter-array spacing. However, as a grid of NHA elements is scaled down, cross-talk between elements can affect the spectral transmission of each element [3,4]. Cross-talk between two nano-hole arrays arises due to the constructive or destructive effect between two scattered SP waves between closely-packed nano-hole arrays. In order to minimize the SP interactions between NHAs, several groups have taken advantage of the design freedoms to include various SP optical isolators such as Bragg mirrors to reduce neighbor to neighbor cross-talk and enhance light transmission [5,6]. However, the crosstalk effect and its consequence on spectral transmission and resonance wavelength for large systems of closely packed NHAs has yet to be systematically analyzed.

Hence, the objective of this work was to experimentally investigate the effect of inter-array spacing on the EOT properties of a grid of NHAs that was representative of a multiplexed SPR sensor. The approach was to fabricate different sets of NHAs with various hole diameter (D), hole periodicity (P), and inter-array spacing (S). The NHAs were arranged on a grid of 4 × 4 blocks where each block contained 4 NHAs (distributed as a 2 × 2 grid) with four specific periodicities and hole diameters. The transmission spectrum of each nano-hole array on the grid was measured and the effect of inter-array spacing on the transmission spectra was analyzed.

2. Materials and methods

2.1 Surface plasmon propagation

We estimated the propagation length of the SP wave along the surface of the metal and dielectric material using the relation

 

Fig. 1 (a) SP propagation length [Eq. (1)] along the surface for Pyrex-gold, air-gold, and Pyrex-chromium interfaces. (b) SEM image of a series nano-hole arrays in a gold film. In this design, all nano-hole arrays had an inter-array spacing of 2 µm. (c) Transmission spectra of (NHA1-NHA4) for boundary versus center blocks with 100 µm inter-array spacing. (d) Transmission spectra of (NHA1-NHA4) for boundary versus center blocks with 20 µm inter-array spacing. The solid lines are the NHAs at the boundary region while the dashed lines are for NHAs at the central region. The transmission spectra measured using microscope setup (described in 10])

Download Full Size | PDF

2.2 Nano-hole array fabrication

Samples were fabricated using an Electron Beam Lithography (EBL) lift off technique to produce a 100-nm-thick gold film perforated with circular nano-scale apertures. We started with a Pyrex wafer coated with a 3-nm Cr conductive layer that enabled electron beam writing. Afterward, 500 nm photo-resist (Negative tone photo-resist ma-N 2403) was spin-coated on to the sample. The nanostructure patterns were written using EBL (LEO, 1530 e-beam lithography) leaving behind nano-scale photo-resist pillars after development. In order to improve adhesion between the gold layer and the substrate, a 4-nm Cr adhesion layer was deposited on to the sample before 100-nm thick gold deposition. Finally, the sacrificial layer of nano-pillars was lifted off resulting in a nano-hole array in the gold layer. More detail on the fabrication methodology has been presented elsewhere [10,11]. Each sample was patterned with 16 blocks on a 4 × 4 grid. Each block contained four NHAs in a 2 × 2 grid with each NHA 30 µm × 30 µm in size and with circular nano-holes on a square lattice. Each NHA in the block had a unique hole diameter and spacing, which gave rise to a distinct resonance wavelength. In each sample, all NHAs were spaced by 2 µm, 5 µm, 10 µm, or 20 µm respectively to study the effect of inter-array spacing on the transmission spectra. Figure 1b shows the scanning electron micrographs of a block of 4 NHAs with 2 µm spacing. The four nano-hole array designs named NHA1 to NHA4 had hole diameters of 215 nm, 247 nm, 273 nm, and 303 nm with periodicities of 382 nm, 433 nm, 483 nm, and 536 nm, respectively. Typical variation in the hole diameter and spacing was 4–6 nm (standard deviation).

Figure 1c and Fig. 1d show the transmission spectra of NHA1-NHA4 for samples with 100 µm and 20 µm inter-array spacing collected using a microscope setup [10]. Although The NHA sample with 20 µm inter-array spacing showed a slightly greater variation in resonance peaks compared to the 100 µm sample, the NHA sample with 20 µm inter-array spacing had a small crosstalk effect compared to 2 µm, 5 µm, 10 µm cases and was used for comparison.

2.3 Device characterization

We employed a multispectral transillumination imaging system with a spectral scanning procedure using a broadband halogen lamp source and a random access monochromator (DeltaRAM V, PTI, Birmingham, NJ), which enabled parallel characterization of all NHAs in each device. Unpolarized white light from a 100 W halogen lamp was focused on the input to the random access monochromator. The input and output slits were set to provide light with a bandwidth of 4 nm (FWHM) at each scanning step. The output from the random access monochromator was delivered to the NHA sample using a broadband liquid light guide (Series 2000 Lumatec, Germany) and a NIR achromatic pair 1:2.86 with 35 mm and 100 mm EFL achromats (NT47-297, Thorlabs, Newton, NJ, USA). The output light intensity from the NHA sample was collected and delivered to a CMOS camera with a NIR achromatic pair (1:1) with 75 mm and 75 mm EFL achromats (NT47-300, Thorlabs, Newton, NJ, USA). A NIR-enhanced monochromatic CMOS camera with 1312 × 1082 pixels was used to capture the images (MV1-D1312I, Photonfocus AG, Switzerland). The monochromator was scanned by computer control from 652 nm to 985 nm in 86 steps with a step size of ~4 nm. Twenty (20) images were captured at each monochromator step. Each image was 3 × 3 median filtered to remove salt and pepper noise due to the CMOS camera. Each set of 20 filtered images was then averaged on a pixel-by-pixel basis to produce a single image. Repetition of the process at each scan step resulted in an image stack (data cube), which was interrogated in the spectral direction on a pixel by pixel basis. Each spectrum was normalized to the transmission spectrum for the Pyrex substrate. The resonance wavelength was estimated for each NHA by first computing the first derivative of the transmission spectrum at each pixel in the image stack. Next, a search algorithm was used to locate the wavelength of the zero-crossing representative of the largest peak within each first-order spectrum, which was interpreted as the resonance wavelength of the EOT. Utilizing a region of interest analysis, the mean ± standard deviation of the resonance wavelength for each NHA was computed. The original sample

3. Results and discussion

3.1 Spectral response of NHAs within a block

Figure 2 shows the optical transmission spectra experimentally acquired from a central block (third row and the third column of the 4 × 4 block grid) for four different devices fabricated with inter-array spacings of 2 µm, 5 µm, 10 µm, and 20 µm. The spectra show a number of distinct spectral features. The larger peaks at longer wavelengths corresponded to the (1,0) resonance mode for the Pyrex-gold side (λ_(1,0)). Resonance peaks representative of the air-gold side were not observed. The minima in transmission spectra for NHA2, NHA3, and NHA4 corresponded to Wood’s anomaly [12]. The smaller peaks in the spectra for NHA3 and NHA4 corresponded to the Pyrex-gold (1,1) resonance mode (λ_(1,1)). The Pyrex-gold λ_(1,1) for NHA2 and NHA1 as well as the Wood’s anomaly for NHA1 were not observed as they were expected to lie outside the spectral range of the instrumentation. The experimental optical transmission spectra were qualitatively similar to the transmission spectra observed in previous work for isolated NHAs [10,11,13,14].

 

Fig. 2 Experimental transmission spectra for nano-hole arrays within a block with specific hole diameter (D), hole periodicity (P) and inter-array spacing (S). All NHAs were selected from the same central block position within the 4 × 4 grid of each device.

Download Full Size | PDF

Due to the effects of cross-talk, the position of the resonance peak was hypothesized to depend on the inter-array spacing. Analysis of Fig. 2 confirmed that the Pyrex-gold λ_(1,0) for all 4 NHAs within a block were blue shifted (dotted lines) when the spacing between the arrays was decreased. The average Pyrex-gold λ_(1,0) blue shift resulting from a decrease in the inter-array spacing from 20 µm to 2 µm was roughly equivalent for all four NHAs. For example, blue shifts (mean ± s.d.) of 32 ± 15 nm, 31 ± 18 nm, 35 ± 19 nm, and 36 ± 13 nm, were observed for NHA1 through NHA4, respectively. Similar blue shifts in the wavelength of each Wood’s anomaly were observed as the inter-array spacing decreased. One of the reasons for resonance blue shifting might be related to a SP reflection effect. Specifically, neighboring NHAs can reflect back the scattered SP waves from the NHA edge with a similar mechanism shown for Bragg mirrors described in [5]. The blue shift effect has been shown in for NHAs with Bragg mirrors as well. Another experimental study with super-lattice NHAs with the same periodicities but with the different pitches shows the same effect of resonance blue shift [15]. Although a strong dependence of transmission on inter-array spacing was not observed, NHAs spaced 2 µm apart had the highest transmission at the Pyrex-gold λ_(1,0) compared to the others.

3.2 Variation in spectral response of NHAs across each device

Since each sample contained 64 NHAs in an 8×8 grid pattern, some NHAs were situated in closer proximity to the edge of the device than others. We hypothesized that the NHAs on the perimeter of the device would be less affected by cross-talk effects compared to NHAs near the center of the device. For example, NHAs at each corner were adjacent to only three neighboring NHAs, while a NHA at least one position in from an edge had 8 NHAs immediately adjacent.

The resonance wavelength measurements supported the hypothesis since for NHAs near the boundary (see Figs. 3a–c ), the resonance wavelength tended to be blue shifted to a smaller degree than NHAs near the center of the device. The spatially dependent blue shift of the resonance wavelength was observed in all devices, but became progressively larger as the inter-array spacing decreased (see difference maps in Fig. 3d and Fig. 3e).

 

Fig. 3 Resonance wavelength maps of NHAs with (a) 20 µm inter-array spacing, (b) 10 µm inter-array spacing, and (c) 2 µm inter-array spacing. Resonance wavelength difference map of NHAs with (d) 10 µm inter-array spacing compared to 20 µm inter-array spacing, and (e) 2 µm inter-array spacing compared to 20 µm inter-array spacing.

Download Full Size | PDF

Analysis of Fig. 3a suggested that even at an inter-array spacing of 20 µm there was cross-talk between NHAs. For example, the resonance wavelength for NHA1 was blue-shifted for blocks from the left side of the device to the right side. However, the differences in resonance wavelength were typically ≤ 20 nm and much smaller for NHA2 through NHA4. These observations lead us to conclude that larger inter-array spacings or intervening structures are required to completely prevent cross-talk effects between closely packed NHAs. In the case of 2 µm inter-array spacing (Figs. 3c and 3e), cross-talk introduced blue-shifts in Pyrex-gold λ_(1,0) of approximately 20 to 50 nm depending on the location and design parameters of the specific NHA being considered. For multiplexed SPR sensing applications, this systematic effect is competitive with the expected changes in resonance wavelength due to local index of refraction changes. However, if small systematic shifts in resonance wavelength are tolerable, then close packing of NHAs by reduction of the inter-array spacing may be a worthwhile approach to miniaturization of multiplexed SPR sensors.

4. Conclusion

The goal of this study was to evaluate the effect of SP cross-talk on the transmission spectra of closely packed NHAs. We fabricated four NHA devices, each with unique inter-array spacing. Each device contained a set of sixteen identical blocks of NHAs and each block contained 4 individual NHAs each 30 µm × 30 µm in size with unique hole diameter and periodicity. Each NHA had SP resonance peaks in the near infrared, which were related primarily to the Pyrex-gold interface. The transmission spectrum of each NHA was measured and the resonance wavelength was extracted. Experimental results revealed progressively larger changes in the resonance wavelength as the inter-array spacing decreased from 20 µm down to 2 µm. The resonance wavelength and the Wood’s anomaly minima blue shifted as the inter-array spacing was decreased. No systematic behavior was observed for the optical transmission intensity of NHAs with different inter-array spacings.