1. Introduction

The extraordinary transmission of light through nanostructured arrays of subwavelength holes [1] in thin gold films and its relation to surface plasmon resonance (SPR) [2] has recently inspired researchers to use these devices for label-free biosensing [3]. The diffraction properties of the periodic structure allows for light momentum enhancement, thus allowing excitation of surface plasmon polaritons. Such excitation of surface waves increases the amount of light scattered inside the holes, giving rise to precise peaks in the transmission spectrum at the resonance wavelength. Since the conditions of plasmon excitation are extremely sensitive to the refractive index (RI), the shift of the resonance wavelength is used for real-time detection and studies of biological binding events. The surface plasmon’s propagation constant can be retrieved by solving Maxwell’s equation at the interface of a metal and a dielectric, which yields bounded modes for TM polarization. In the approximation that the holes do not change the dispersion relation of the plasmon, and that there is no strong coupling between the two interfaces of the film, the wavelength that satisfies the condition of resonant excitation of these bounded modes for a diffraction array at normal incidence is given by [2]:

Such plasmonic systems based on nanostructured hole arrays are known to provide new properties and advantages over the conventional Kretschmann SPR sensor configuration: the collinear setup allows for easy integration, hence giving great miniaturization possibilities and permitting incorporation into a microfluidic cell with high-throughput and multiplexed sensing [4–7]. The applications of modern nanofabrication technologies allow for low-cost and large-scale fabrication of regular plasmonic metallic nanostructures on different dielectric or semiconductor substrates [8,9].

Despite a number of advantages, nanohole arrays show sensitivity and resolution orders of magnitude lower compared to conventional SPR devices [3,7,10–14]. So far, resolution of 6.6 x10⁻⁵ RI units (RIU) has been demonstrated [10]. In general, it has been suggested that grating coupler-based surface plasmon sensor exhibit lower sensitivities than prism-coupler ones [15], and devices based on extraordinary transmission are usually affected by a low signal-to-noise ratio since the transmitted intensity is usually of about 5% [2]. In this article, we improve the resolution of the nanohole array configuration by using polarization modulation, a biperiodic array and balanced detection. This configuration retains the collinear geometry that is favoured for integration into a portable SPR device, and also uses only a single laser source and intensity detection.

2. Basic idea and approach

The performance of portable nanoplasmonic sensing and imaging systems depend on the complexity and cost of the excitation/detection system. An intensity interrogation approach with one or several fixed wavelength sources is well suited for such applications because it can provide an efficient and simple system, using a low-cost and stable monochromatic laser diode sources in combination with 2D CCD camera detector. However, the resolution of such reported intensity systems is low, typically 10⁻⁴ RIU [11]. Intensity based sensors must also be corrected for any changes in intensity that are not correlated with the displacement of the resonance peak, such as changes in light absorption through the addition of tested analytes, or source intensity noise. To reduce such spurious intensity changes, differential systems with a reference or calibration channel can be realized by introducing structure with identical sensing characteristics [11,12]. Here, we exploit polarization diversity within a biperiodic nanohole array to achieve self-referencing and thereby reduce the spurious contributions to the noise.

We report a resolution-enhanced nanohole array sensor based on the intensity measurements of a single wavelength source. In this setup, we use the transmission signal of two orthogonal polarizations impinging on a rectangular nanohole array (inset of Fig. 1 ), as was introduced previously [16]. The periods in x and y of the array are chosen in order to have one peak below the source’s wavelength, and one above, as shown in Fig. 1. A modification of the refractive index close to the metal surface thus gives rise to an increase in transmission for one polarization and a decrease for the other one, and the difference between the two corresponds to the detection signal. This method allows up to double the sensitivity of the sensor since it uses the variation of two slopes instead of only one. Furthermore, with a careful choice of the position of the peaks and of the working wavelength the initial calibration point can be set to zero amplitude, allowing for better contrast in the detection signal and better noise reduction by having similar amplitude on the two polarizations.

 

Fig. 1 Principle of the proposed technique using a nanohole rectangular array with different periodicities a_x and a_y along the two main axes. The transmission spectrum of such a structure is shown with resonant peaks at different wavelength for the two orthogonal polarizations. The point of intersection (Point A) corresponds to the initial calibration point. The change in refractive index (red curves) shows an increase of intensity for one polarization (point B) and a decrease for the other polarization (point C).

Download Full Size | PDF

3. Fabrication, design and optimization of the nanohole rectangular array

The two-dimensional nature of nanohole array offers the possibility of having distinct transmission features for orthogonal polarizations. These resonant properties are dependent on the many parameters of the array, such as the periodicity, the size of the holes and the thickness of the metal layer. Optimization of the sensing performances relies on the proper choice of these parameters for operation at 632.8 nm wavelength and in an aqueous environment.

Figure 2(a) shows an example of a nanohole array with periods 380nm × 420nm. The gold layer samples were obtained from Platypus Technologies and nanostructured by Focused Ion Beam (FIB) milling with a FEI DB-234, using a 30 kV acceleration voltage and 30 pA ion current. The thickness of the film was 100 nm and was chosen to be a few times the skin depth of gold in order to be optically opaque while reducing the exponential decrease of the guided modes inside the holes. The diameter of the holes is chosen to be of 200 nm in order to be before the cut-off frequency of any guided modes in the holes [17]. This allows optimization of the transmitted intensity while reducing spurious noise from, for example, the intensity fluctuations of the source or absorption of the laser in the analyte.

 

Fig. 2 (a) Scanning electron microscopy image of an example of a nanohole rectangular array 380 nm x 420 nm (b) Transmission spectra of nanohole rectangular arrays with period 380 x 415nm (solid line), 390 x 425 nm (dashed line) and 395 x 435 nm (dotted-dashed line). The positions of the peaks are red-shifted of (from left to right 28 nm, 27 nm, 29 nm, 28 nm, 29 nm, 26 nm in comparison to the dispersion relation (square dots).

Download Full Size | PDF

The choice of the periodicity determines the position of the resonant peaks. Structures can easily and precisely be tuned in order to meet the requirements of a specific application. In the case of the proposed sensor, the wavelength of a He-Ne source was chosen since highly stable laser sources and high performance optical equipment are widely available. The periods of the array were chosen with the help of Eq. (1) in order to have one peak below 632.8 nm and one above. It has to be noted that the transmission peak wavelength, when measured experimentally, is red-shifted in comparison to what is calculated. The simple dispersion relation is therefore incomplete to fully take into account the phenomenon of extraordinary transmission. Interference between the resonant state (the surface plasmon) and a continuum of state (light scattered into the hole) gives rise to a red-shifted asymmetric peak typical of a Fano-type resonance [18]. Also, the holes themselves can give rise to resonances [19] which will affect the position of the transmission maxima. Therefore, when a precise position of the transmission peak is critical, a slight correction in the period must be taken into account when fabricating the nanohole array. In Fig. 2(b), the transmission spectra of several nanohole rectangular arrays are presented, as well as the wavelength of resonance calculated with Eq. (1).

Since the position of the curves translates to longer wavelengths with an increase in refractive index, the dynamic range of the sensor will correspond to the range between the working wavelength and the position where the slopes vanish. Therefore, having the maxima of the right peak as close as possible to the right of the source’s wavelength will increase the dynamic range of the sensor. Making the curves intersect at the source’s wavelength can still be managed by adjusting the optical setup in order to suppress or increase the intensity of a particular polarization (see the instrumental methodology section below). The choice of period was made at 380 × 420 nm to increase dynamic range while having the steepest parts of the curves at the source’s wavelength.

4. Instrumental methodology

To test the nanoplasmonic structures in liquid, the experimental set-up illustrated in Fig. 4 was developed. A stabilized He-Ne laser was used as the source. Spatially filtered and polarized light passes through a Photo-Elastic Modulator (Hinds Instruments) and a quarter-wave plate in order to periodically modulate the state of polarization of light from linear to circular at a frequency of 50 kHz. Fine rotation of the second polarizer also serves for initial equalization of the two orthogonal polarizations intensities to perfectly match the intersection point at the source’s wavelength. Indeed, changing the orientation of this polarizer modulates the transmitted intensities by factors:

 

Fig. 4 (a) Schematics of the optical setup used for the sensing experiment. (b) Adjustment of the initial calibration point by rotating the second polarizer from 45° to 35°. The intensities for both polarizations intersect at the source’s wavelength.

Download Full Size | PDF

After transmission through the nanohole array, the beam is split in its two polarization components using a Wollaston prism towards a balanced photodetector (Nirvana Detector, New Focus), that together with a lock-in amplifier (Stanford Research System) allows for significant noise rejection at all the frequencies that differ from the modulation frequency. An open and flow-injection measurement cell was used for tests in liquid.

5. Results and discussion

The wavelength sensitivity of the nanohole array was evaluated from the shift of the maxima when the solution was changed from water to pure ethanol, as shown in Fig. 5(a) . Shifts of 9.9 nm were recorded for the y polarization and for the x polarization, which corresponds to an estimated sensitivity of 396 nm RIU⁻¹, when considering the difference of 0.025 RIU between the two solutions [20,21]. The sensitivity in terms of change in intensity was calculated by multiplying the slopes of the peaks by the wavelength sensitivity. The slopes at the source’s wavelength were of 4.3% nm⁻¹ for the polarization along the y axis and 3.0% nm⁻¹ along the x axis (the slope on the red side of the resonant peak is lower, as can be expected from the shape of the Fano resonance peak), which corresponds to an intensity sensitivity of 1700% RIU⁻¹ and 1200% RIU⁻¹ respectively, for a total of 2900% RIU⁻¹ for the balanced detection. The sensitivity of this setup corresponds to an almost two-fold improvement compared to the sensitivity of a square array in similar conditions, since it corresponds to the addition of the displacement of two transmission peaks instead of only one.

 

Fig. 5 (a) Transmission spectra of the nanohole array with 380 x 420 nm as periods in water and ethanol. (b) Responses of the sensing system for small increase of ethanol. Insert: Noise levels from the signal of the balanced detection and from the signal of a single polarization.

Download Full Size | PDF

The intensity difference between the two polarizations was recorded as a function of time with different concentrations of ethanol successively added to water in order to measure the maximum resolution of the system (Fig. 5(b)). Every step represents an addition of a small amount of ethanol, in order to obtain a 0.7% concentration difference, which corresponds to 1.7 x 10⁻⁴ RIU change. The signal to noise ratio (SNR) achieved is 27.4, allowing for a detection sensitivity of 6.4 × 10⁻⁶ RIU. The SNR was determined by taking the difference between the mean value of the signal for the bottom and top part of each step, divided by the standard deviation of the signal at those levels.

In order to demonstrate the improvement accomplished by the balanced detection of the rectangular array, the signal coming from one polarization only was compared to the signal from the balanced detection, as shown in the inset of Fig. 5(b). The single polarization signal in this case corresponds to the signal that is obtained from a single periodicity, which would be analog to the signal obtained from a square array on the presented setup. The SNR obtained from the single polarization is 5 times lower than the one obtained from the balanced detection, clearly demonstrating the advantage of such detection method. The dynamic range was calculated by using a 10% sensitivity decreasing as range limits, which corresponds to a spectral shift of 20 nm or to a change in refractive index of 0.05 RIU.

The proposed experimental setup benefits from several features in comparison to the previous intensity based nanohole sensing systems [11–13]. First, two slopes are used instead of only one, allowing for a greater change in intensity to be measured (approximately double) from a modification of refractive index. Also, the problem of uncorrelated (i.e., spurious) intensity fluctuations is solved by the use of the two polarization components and balanced detection, which rejects the common fluctuations of the signal. Noise originating from the light source is reduced this way. Overall, the polarization modulation, bi-periodic array and balanced detection allow for an overall sensing system with significantly improved SNR. Despite the sensitivity being low in comparison to conventional angular-based SPR devices [14], the very low noise of this setup allows for similar sensing performance.

5. Conclusion

We demonstrated the use of a rectangular nanohole array with polarization modulation and balance detection in order to increase the performance of a nanohole array intensity based biosensor. A resolution of 6.4 x 10⁻⁶ RIU is achieved, which is an improvement in comparison to previously reported nanohole array sensors [10]. Since this type of sensor has such a small footprint, many arrays can be integrated into a chip in order to realize a multiplexed sensor, using a simple laser diode as the source and a 2-D CCD camera. Balanced detection would require the splitting of the light emerging from each array (using a Wollaston prism) and using two areas on the camera. Together with microfluidic channels, this structure and detection scheme could contribute to the realization of a fully integrated and highly sensitive sensor which can find many applications in the area of detection of chemical and biological species, or for the study of biomolecular interaction.