Introduction

During the last decades, plasmonics has attracted ever-increasing attention from researchers and engineers due to the high tunability, enhancement and confinement of the electric field and novel properties provided by plasmonic nanostructures [1–3]. This interest is catalyzed by expanding experimental capabilities for accurate fabrication, characterization and fine-tuning of nanostructures as well as deeper understanding supported by theoretical descriptions and the development of numerical simulation methods. One of the most promising application of plasmonics is for refractive index based label-free biosensing, where the sensitivity of plasmon resonance frequency to the dielectric constant of the surrounding media is utilized for the detection of biomolecular binding events at the surface of nanostructure [4].

Substrate-supported plasmonic nanostructures are advantageous for biosensing compared to nanoparticle dispersions as it allows controlled illumination/spectral readout geometry, compatibility with microfluidic lab-on-a-chip approaches and the potential for sensor regeneration. At the same time, the refractive index sensitivity of such structures is lower than that of unsupported nanostructures since a significant part of enhanced electric field is directed into the substrate becoming inaccessible to the analyte molecules and hence not involved in sensing. Such a‘substrate effect’ has been used to rationalize the higher sensitivity of supported nanoring structures compared to disks, since the larger part of the surface in the latter geometry (≈50% for disks versus ≈17% for rings) is not available due to the contact with the substrate [5]. Moreover, the supported sensing nanostructures are typically made on a material with the refractive index higher than that of the operating media (for example, n≈1.5 and ≈1.3 for glass and aqueous solutions, respectively). This difference in dielectric constants affects the local electric field near the surface of plasmonic nanostructures, often making it asymmetric and localized close to or inside the substrate. It has been demonstrated that lifting nanodisks above the substrate using dielectric nanopillars can decrease this effect and enhance the refractive index sensitivity [6, 7].

One of the most promising biosensing platforms is based on plasmon resonances in perforated optically thin metal films, since they are easy to fabricate via colloidal lithography (CL) over a large (cm²) area, provide for simple functionalization in the high sensitivity regions and can be combined insitu with other sensor approaches e.g. QCM-D [8–11].However, the reported values of the refractive index sensitivity of such glass-supported short-range ordered (SRO) hole arrays [9, 12] are lower than those of other structures such as disks [13, 14]and rings [5, 15]. It is reasonable to expect that the sensitivity of such films is to a large extent suppressed by the substrate. As it has been shown recently, the refractometric sensitivity of these films can be enhanced by using low refractive index substrates [16]. In addition, electrodynamics simulations show that the local sensitivity reaches a maximum near the bottom rim of the holes in SRO hole arrays, also suggesting that the local electric field is reshaped by the substrate [12].

Unlike metallic nanoparticles and metal films, SRO hole arrays combine surface plasmon polariton and localized surface plasmon modes making them promising extensions of the traditional SPR approach [17]. Much effort has been focused on understanding the optical spectra of these systems. It has been suggested that extinction maximum is related to coupled SPP Bloch wave (SPP-BW) modes, with energy defined by the average nearest neighbor hole distance, whereas the longer wavelength extinction minimum is connected to localized surface plasmon resonance (LSPR) inside the holes [12].

Two approaches can be exploited to fabricate the metal structure with holes elevated above the substrate: etching through the holes or deposition of a dielectric layer prior to metal using the same mask. The former approach has been recently used to greatly enhance the extraordinary optical transmission-based sensitivity of 80 nm thick Ag films with periodic hole arrays by etching the glass substrate using wet etching [18] and form perforated Au films lifted by Nb₂O₅ layer using the reactive ion etching [19]. The latter way has been used to create the perforated multilayered AlN/Al/AlN films [20], but no comparison of refractive index sensitivity of lifted and non-lifted SRO hole arrays has been reported. The aim of the present work is to study the substrate effect on refractometric sensitivity of SRO hole arrays in optically thin Au films.

Fabrication and characterization

Colloidal lithography has been previously used to fabricate a variety of structures including triangular pyramids [21], disks [6, 7, 22], rings with different asymmetry [5, 23], as well as SRO hole arrays [9, 24].A common CL technique for the fabrication of SRO hole arrays involves three major steps: (i) formation of a colloidal mask, (ii) vacuum deposition of material(s) and (iii) removal of mask. The protocol used here for the formation of the colloidal mask has been described in detail elsewhere [8]. Briefly, negatively charged monodispersed polystyrene particles have been deposited from aqueous dispersion (0.2% wt.) on positively charged substrates by electrostatic self-assembly. The substrates (glass slides, 25 mm diameter, 0.3 mm thick, Menzel-Gläser) have been cleaned previously by ultrasonication in acetone (10 min), ethanol (10 min) and deionised water (MilliQ10 min) followed by 30 min oxidation in UV/Ozone. Then they have been coated by a triple polyelectrolyte layer, absorbed sequentially from aqueous solutions: poly(diallyldimethylammoniumchloride (PDDA, MW 200,000–350,000, Sigma–Aldrich, 2% wt.),poly(sodium 4-styrenesulfonate) (PSS, MW 70,000, Sigma–Aldrich, 2% wt.), and polyalluminium chloride (PAX, Kemira, 5% wt.). Each solution has been applied for 30 sec, followed by sample rinsing with MilliQ and drying with nitrogen. Particle adsorption time was in the range 2-5 minutes, sufficient to reach saturation. An excess of the particles has been rinsed off with MilliQ followed by rapid drying under the nitrogen flow to avoid particle aggregation caused by capillary forces.

Vacuum deposition has been carried out after removal of triple layer by 5 min UV/Ozone treatment. All materials have been deposited by thermal evaporation using home-built electron beam-stimulated physical vapor deposition set-up (EB-PVD, 2kV e-gun, base pressure 1x10⁻⁷Torr, deposition rate of 0.1-0.4 nm/sec). To create elevated hole arrays we have modified the standard CL fabrication procedure by adding a silica layer deposition step. Thus, 40 nm of silica has been deposited through the colloidal mask followed by deposition of 1 nm Ti adhesion layer and 20 nm of Au. Colloidal masks have been removed after vacuum deposition by tape stripping with an adhesive tape (90µm thick blue PVC Tape, Semiconductor Production Systems).The non-lifted structures have been formed using the same procedure by omitting the silica deposition step. The series of non-lifted and lifted samples have been produced using polystyrene particles with average diameter of 110, 140 and 160 nm. The fabrication scheme of elevated hole array is depicted in Fig. 1(A).

 

Fig. 1 Perforated Au film lifted by 40 nm silica layer: A) fabrication scheme; B) Cross-sectional SEM view. Scale bar is 100 nm.

Download Full Size | PDF

Samples with perforated gold films have been studied by scanning electron microscopy (FEI Magellan FEG XHR-SEM). An example of the resulting elevated film structure, formed using 200 nm colloidal particles on Si substrate for better image quality, is presented in Fig. 1(b). Due to shadowing effect caused by the growing lateral size of the silica caps on the particles during the deposition, the SiO₂ holes possess a conical shape with wall inclination of about 30°.Thus,lifting is accompanied by the increase of the hole diameter.

Results and discussion

Optical spectra

Optical properties of the hole arrays have been studied by UV-vis-NIR spectroscopy (Shimadzu 3600 UV/Vis-NIR) in transmission mode on glass substrates using a clean glass slide as a reference. Extinction spectra for the holes fabricated using 110,140 and 160 nm particle mask with and without a silica spacer layer, are presented in Fig. 2(a) and Fig. 2(b), respectively. The maxima in the spectra of the lifted holes are blue-shifted in comparison to the non-lifted ones formed in the Au film of the same thickness and using the same particle size. At the same time, the extinction minima are slightly shifted to longer wavelength. The background extinction is lower for the lifted films due to larger hole sizes at the same hole density resulting in effectively lower amounts of gold.

 

Fig. 2 Extinction spectra of perforated thin Au films fabricated by CL using 110 nm (black), 140 nm (red) and 160 nm (green) particles: A) without silica spacer, B) with 40 nm silica layer. Spectra taken in air. Insets show the example of hole arrays fabricated using 160 nm colloid particles, scale bar is 1μm.

Download Full Size | PDF

The observed spectral changes can be rationalized by taking into account that an addition of a silica layer results in lower effective refractive index of the substrate and larger hole sizes. The former is expected to blue-shift both extinction peak and dip, whereas the latter should red-shift the dip only. The average center-to-center hole spacing, which defines the surface plasmon polariton Bloch wave condition (BW-SPP), and thus the extinction maximum wavelength [12], is not changed.

Refractometric sensitivity

Bulk refractive index sensitivity measurements have been performed using a home-built flow cell, compatible with the UV-vis-NIR spectrometer, and aqueous solutions of glucose (Sigma-Aldrich Denmark) with different concentrations having the refractive index in the range 1.333−1.355. Series of spectra were taken at each concentration measuring the spectral profile. The peak position has been determined by numerical differentiation of the spectrum using the Savitzky-Golay algorithm [25] and finding the roots of its first derivative. The sensitivity has been then determined as a slope of the linear fit to the thus obtained extinction peak positions at different values of refractive index.

The experimentally determined bulk refractive index sensitivity of perforated Au films with and without silica layer is presented in Fig. 3(a). Unlike linear dependence of the sensitivity versus the peak position, observed in case of LSPR [26], the SPP sensitivity exhibit second-order behavior, that is in agreement with analytical model developed for SPR [14], assuming linear approximation of dielectric function for Gold in this wavelength range. Estimated bulk sensitivity of the smallest holes gives 116 ± 4 and 158 ± 9 nm·RIU⁻¹ for non-lifted and lifted holes, respectively, that corresponds to more than 37% sensitivity enhancement. This result indicates that despite the fact that the investigated extinction peak is described as a SPP-BW resonance localized between the holes, there still is a large portion of the high field concentrated near the rims of the holes which can be revealed to improve sensitivity. At the same time, it should be noted that lifting is accompanied by broadening of the extinction peak, which can negatively affect the overall sensor performance expressed in terms of figure of merit (FOM).

 

Fig. 3 Sensitivity of the perforated Au films. Left panel (A): as a function of extinction peak position at n = 1.333 for different diameter of the holes without (blue squares)and with spacer layer (red circles). Black lines are to guide the eye. Right panel: extinction peak shift with an increase of the refractive index for theB) non-lifted and C) lifted holes fabricated using 110 nm particles as a mask. The refractive index values are presented in the insets.

Download Full Size | PDF

FDTD Simulations

To model the optical properties of the smallest SRO hole arrays we have used a system with a hexagonally close packed array of holes in a 20 nm Au film on a glass substrate surrounded by water. The lattice parameter a = 273 nm has been chosen to reproduce the experimental hole number density (15,5μm⁻²), estimated from low magnification SEM images (data not shown). The diameter of the holes was set to 110 nm for non-lifted and to 155 nm for lifted ones estimated from SEM images. In the latter case the conical wells with walls inclination of 30° in a glass substrate were added under each hole to represent the structure with the spacer layer. The rims of the holes have been rounded to avoid unrealistic field enhancement at unphysicallysharp edges.

Numerical calculations have been performed using commercial finite-difference time domain simulator (FDTD Solutions) [27]. To reproduce the dielectric function of gold a numerical fit to experimental data has been used [28]. Constant refractive index values have been used for glass (1.52) and water (1.333). A mesh size of 1 × 1 × 1 nm was used. Periodic boundary conditions in the Au film plane and perfectly matching layers in the perpendicular direction have been used to represent the periodic structure. The simulation box included two hexagonal close-packed unit cells, allowing one complete hole in the center.

The simulated extinction spectra presented in Fig. 4(a) are consistent with the experimental data for SRO hole arrays registered in air and water. Thus, the extinction maxima corresponding to the lowest order SPP-BW (1 −1 0) resonance in n = 1.333 (black) are 673 and 645 nm for the holes without (top) and with spacer layer (bottom), respectively. The simulated data match the experimental values (673 and 634 nm, respectively).

 

Fig. 4 FDTD simulation results: A) extinction spectra for non-lifted (top) and lifted (bottom) hole arrays in n = 1(red) and n = 1.333 (black); local electric field (|E|²/|E₀|²) distribution near a hole in hexagonally close packed periodic array of holes at the wavelength of extinction maximum. The refractive index of the media is n = 1.0 (B,C); and n = 1.333 (D,E).

Download Full Size | PDF

To understand the origin of the sensitivity enhancement observed when the holes are lifted above the substrate, we have analyzed the local field distribution near the hole with and without spacer layer. The field plots are presented in Fig. 4 for refractive index n = 1.0 (Figs. 4(b) and 4(c)) and n = 1.333 (Figs. 4(d) and 4(e)) for both geometries. It can be seen that the high field is localized near the bottom rim of the hole. Adding a silica layer reveals a fraction of the field to the sensing media. However, it is not obvious to what extent the effective refractive index of the silica layer reduced by the presence of holes, can change the distribution of the local electric field between the top and bottom surface of the gold film.

To quantify the redistribution of the electric field we have calculated values of E_top and E_bot, which characterize the fraction of the field localized at the top and at the bottom of a hole, respectively. The absolute field values in each point above and below the half thickness of the Au film (z>0 and z<0) have been summed up and related to the total sum of the field in all the points. Similarly, to get information on the amount of field buried in the substrate and available to the media, the values E_substr and E_media have been calculated by finding the part of the field located in the regions with refractive index n>1.4 (inside the glass substrate) and n<1.4 (inside the media).

The results, presented in Table 1, explicitly show that the sensitivity enhancement is achieved by revealing the otherwise unavailable part of the high electric field in samples with the spacer layer. Thus, the fraction of the field that is buried in substrate is changed from 0.54 (0.52) to 0.44(0.43) in air (in water, respectively). The relative increase of the field available for sensing is approximately 17% at n = 1.333.

Table 1. Electric field (|E|/|E₀|) distribution near the hole at extinction maximum wavelength.

View Table

Lifting SRO holes arrays by deposition of a spacer layer is a simple process and can be adopted easily for sensing application of current SRO arrays [10, 12]. We reveal that despite the fact that the extinction peak is related to SPP modes, the major cause for its enhanced sensitivity is the increase in available local high field sites at the hole suggesting that the increase in sensitivity for localized detection at the hole (which is the approach typically performed) would potentially be significantly higher than in the system reported here. Potentially, deeper holes combined with the appropriate biochemical functionalization of the inner surface can be used as a suitable sensing platform for the detection of larger objects, such as vesicles or viruses.

Conclusions

We have developed a CL technique for fabrication of the lifted hole arrays by adding a dielectric spacer layer between plasmonic film and a substrate. It is demonstrated that lifting the perforated Au film results in up to 37% increase of bulk refractive index sensitivity at extinction maximum frequency. According to the numerical simulation, the enhanced sensitivity is due to revealing of otherwise buried local high field near the hole. This approach is simple to incorporate in current fabrication approaches and can be of use in future SRO based nanohole plasmon sensors.