1. Introduction

In the last few years, plasmonic nanostructures have been exploited in several ways for plasmon-resonance-based sensing applications [1,2]. Therefore, surface-enhanced spectroscopies (SES), such as surface-enhanced Raman scattering (SERS) [3–6], surface-enhanced resonance Raman scattering (SERRS) [7], surface-enhanced infrared absorption (SEIRA) [8,9] or metal enhanced fluorescence (MEF) [10], are considered as some of the most important optical sensing techniques. They are all based on a significant amplification of the spectroscopic signal by electromagnetic field enhancement induced by localized surface plasmon resonance (LSPR). In particular, SERS was the most widely used SES as an extremely sensitive analytical technique for chemical or bio-analytical applications. Although single molecule detection has already been predicted and observed in SERS [3,5,6,11–14], in most of the cases, giant enhancements of the Raman signal were only reached for coupled nanostructures such as at interstitial sites in nanoclusters or nanostructured surfaces with closely spaced features [6,15–17]. However, SERS behaviors at so-called hot spots are complex and usually show poor repeatability and reproducibility, which limit their use in biosensor applications. Particularly, the LSPR excitations at such interstitials or on nanoparticle sharp edges are highly dependent on the excitation polarization [18]; as a consequence, the SERS signal will show in the same way a strong dependence on the excitation polarization [15,19,20]. For example, using nano-ellipses as SERS-active substrates, intense Raman signals could only be obtained when the excitation polarization is parallel to the major axis of the nano-ellipses [20]. Such polarization dependence also brings limitations to the actual sensing applications because rigorous management of polarization in the measurements set-up will be required. This could be an important drawback especially when the light polarization control cannot be easily completed such as in optical fiber [21] or when the SERS substrate cannot be placed precisely in the right orientation. Further developments for technology transfer of such optically based sensors will depend on their robustness and simplicity for non-physicists users (biologist, medical doctors...).

Moreover, for all the plasmon-resonance-based sensors, but not only for SERS, LSPR excitations of the metallic nanostructures play the key role for the sensor sensitivity. Thus, to achieve the best enhancement factor and to improve the sensor performances, the optimization of the LSPR properties is essential to the sensor characteristics. Actually, the role of the LSPR in the SERS performance has been investigated by several groups with the aim to improve the detection sensitivity in a controllable and reproducible way [20,22–29]. In most of the cases, gold nanoparticles on solid support are widely exploited for the development of lab-on-chip biosensors because of their chemical stability and the potential biocompatibility of gold. Moreover, optical sensing techniques require a transparent substrate but gold nanostructures do not stick well on the glass surface. Usually, a thin adhesion layer (Cr, Ti, or Ni typically) of 5-10 nm is commonly used to guarantee a good bonding between Au nanostructures and glass surface [30,31]. However, the Cr layer degrades the optical properties by shifting and broadening the LSPR of gold nanoparticles [32]. As a consequence, the spectral signal will be decreased due to the damping of the LSPR, which is considered one of the main limitations for the improvement of sensor sensitivity. However, the influence of the adhesion layer on the actual performance of plasmon-resonance-enhanced spectroscopies was rarely studied experimentally [33].

Thus, in this paper, the optimizations of sensing substrates in actual applications are highlighted in two aspects: high sensitivity and ease of manipulation under different conditions. In sec.3, the design and fabrication of apolar plasmonic nanostructures are proposed. This new family of nanostructures defined by symmetry consideration exhibit non polarization dependent LSPR and SERS performance. Such apolar plasmonic nanostructures enable a wide flexibility for the substrate design and specimens detection under different conditions. As high sensitivity and reproducibility are critical for actual sensing, we will demonstrate in sec.4 that the SERS sensitivity could be further increased by using a molecular adhesion layer between gold nanostructures and underlying glass substrates. An enhancement of SERS signal of one order of magnitude was achieved due to the improved LSPR properties that induce higher near-field enhancement.

2. Experimental

The gold nanostructures have been designed by electron beam lithography (EBL). The EBL is achieved by a 30 kV Hitachi S-3500N scanning electron microscope (SEM) equipped with a nanometer pattern generation system (NPGS) [20]. This technique provides precise control of shape, size and arrangement of the nanostructures. To achieve good quality EBL, we use a high resolution resist, Polymethyl methacrilate (PMMA), spun over a glass substrate and covered by 10 nm of aluminium. After exposure, the patterns are developed using methyl isobutil ketone (MIBK): isopropyl alcohol (IPA) 1:3 and the desired mass thickness of gold is evaporated on the sample. After lift-off in acetone, the shape and the lateral sizes of the nanoparticles were checked by using SEM (Hitachi S-3500N) operating in backscattered electron mode.

LSPR and SERS performances are investigated by a Horiba Jobin-Yvon micro-Raman spectrophotometer (Labram) (Fig. 1(e) ). The extinction spectra were recorded in transmission configuration with a × 10 objective (N.A. = 0.25) by removing the edge filters. For SERS measurements, the substrates were immersed in a 10⁻³ M solution of trans-1,2-bis(4-pyridyl)ethylene (BPE) during 1h and dried with nitrogen. Raman measurements were carried out with the 632.8 nm line of a He-Ne laser, and Raman spectroscopy is recorded with a × 100 objective (N.A. = 0.90) in back scattering configuration. The SERS intensity was estimated by calculating the area of a Lorentzian fitted BPE band located at 1200 cm⁻¹ (this mode has significant ethylenic C = C stretch character).

 

Fig. 1 SEM images of nano-star (a) nano-triangle (b) nano-cylinder (c) and nano-ellipse (d); (e) schematic of the experimental setup for extinction and SERS measurements

Download Full Size | PDF

3. Apolar plasmonic nanostructures for non polarization dependent sensing activities

In this part, we propose the design and fabrication of a new family of nanostructures with optical response insensitive to the light polarization. Such apolar plasmonic nanostructures are not limited to cylindrical or spherical shapes but could have a wide variety of shapes with sharp tips or coupled architectures that could give rise to “local hot spots” with giant near-field enhancement.

2.1 Design and fabrication of apolar plasmonic nanostructures

With group theory, one can demonstrate that any complex nanoparticle that belongs to C_nv, with n≥3, symmetry point group for at least one scale has an optical response insensitive on the light polarization [34]. It has been proved that nanoparticle of such symmetry appears with the optical behavior of a cylinder. Such nanostructure acts as an apolar system for any optical excitation with a wave vector parallel to the C_n axis.

In order to show that the response of those nanostructures is apolar, it is convenient to use the irreducible tensor formalism [35,36]. Thus, the nanoparticle polarization P depends on the incoming electric field E such as $P=\underset{¯}{\underset{¯}{\alpha }}•E$ with $\underset{¯}{\underset{¯}{\alpha }}$ the second-rank polarizability tensor. We assume that all far field properties depends on this tensor. As the repartition of molecules on the nanoparticles is rather homgeneous, some phenomena including near-field like SERS can also be described by this tensor. $\underset{¯}{\underset{¯}{\alpha }}$ tensor be expanded into its irreducible parts such as: $\alpha ={\alpha }^0\oplus {\alpha }^1\oplus {\alpha }^2$.

${\alpha }^0$ corresponds to the fully symmetric part (i.e. the tensor trace), ${\alpha }^1$ to the antisymmetric term and ${\alpha }^2$ to the traceless antisymmetric term. Since those components are irreducible, they can be rotated by using the Wigner D matrices. In a $C_{nv}$invariant system, we then find the following constrain: $D^J(\theta ,0,0){\alpha }^{JM}=e^{2\pi i\frac{M}{n}}{\alpha }^{JM}={\alpha }^{JM}$. This condition is only verified for M = 0 or 3. Then, only the components ${\alpha }^{00}$, ${\alpha }^{10}$ and ${\alpha }^{20}$ do not vanish for a $C_{nv}$ nanoparticle with $n\ge 3$. Those components are those which have a cylindrical symmetry about the axis orthogonal to the substrate. Such nanoparticles have thus an apolar response to the incoming electric field.

This model was verified on gold nanostructures designed by electron beam lithography (EBL). Four types of nanostructures that belong to the C_nv point group with the lowest number of symmetries were designed and fabricated on the glass substrate, as shown in Fig. 1(a)-1(d); each one has at least a C_n symmetry axis: nano-stars and nano-triangles (C_3v), nanocylinder (C_∞v symmetry group) and nanoellipses (C_2v).

The gap between adjacent particles has been kept constant at 200 nm because with such a distance any near field coupling could be neglected. The apolar behaviors of such nanostructures are experimentally investigated by extinction and SERS measurements.

2.2 Apolar behaviors confirmed by LSPR and SERS measurements

To characterize the apolar behavior of such nanostructures, extinction and Raman spectra were measured with incident polarization varied continuously in the sample plane. Figure 2 gives the comparison of LSPR properties on nano-cylinders and nano-stars.

 

Fig. 2 Extinction spectra of gold nano-cylinders (diameter 260 nm, height 50 nm) (red line) and gold nano-stars (side length 150 nm, height 50 nm) (green line) (a): each has a resonance peak at 790 nm, the values of the Full Width at Half Maximum (FWHM) are 110 nm and 83 nm respectively for nano-cylinders (red arrow) and nano-stars (green arrow); Extinction spectra of gold nano-stars (b) and nano-cylinders (c) (the same particles in (a)) with perpendicular polarization directions.

Download Full Size | PDF

Nano-cylinder is intrinsically apolar structure which shows optical response insensitive to the excitation polarization. However, in such structure the tuning of LSPR is not so “efficient”: as demonstrated in Fig. 2(a), to get a plasmon resonance at about 800 nm requires the nano-cylinders and nano-stars have a size of 260 nm and 150 nm (both have the same height), respectively. However, the nano-stars show a much narrower LSPR line width than the nano-cylinders one. Since the near-field enhancement factor is inversely proportional to the LSPR line width, as a consequence, a narrower line width indicates a stronger local electromagnetic field [37,38]. Here we have to mention that to get the most efficient near-field enhancement one phenomenon cannot be neglected: the lightning rod effect, in which the confinement of the charges at a surface with a small radius of curvature makes an additional enhancement factor. Such effect takes place in very small nanostructures or in elongated nanostructures such as nanorods, nanowires or tips. Therefore, in the case of spherical or cylindrical Au nanoparticles of large diameters (above 150 nm), the lightning rod effect is ignored because the curvature is too large. Accordingly, in such structures one could not get the maximal near-field enhancement at any plasmon resonance wavelength. However, as indicated in Fig. 2(a), at the same LSPR wavelength, much higher near-field enhancement could be achieved in nano-stars due to their smaller size and sharp tips. To show the polarization dependence, Fig. 2(b) and 2(c) give the extinction spectra of nano-cylinders and nano-stars (the same particles as Fig. 2(a) with incident polarizations in perpendicular directions. Both the nano-cylinders and nano-stars present few changes in the position and intensity of the LSPR for the two perpendicular polarization directions. This means the nano-star of C_3v symmetry shows the same apolar behavior as nano-cylinder. Such property of nano-stars is demonstrated more clearly in Fig. 3 that gives the LSPR position (a) and intensity (b) as a function of the polarization angle for the nano-stars with side length of 150 nm and out-of-plane height of 50 nm.

 

Fig. 3 LSPR position (a) and intensity (b) versus incidence polarization angle for nano-stars (length 150 nm, height 50 nm) and nano-ellipses. Nano-ellipses (length 80 nm, width 40 nm, height 50 nm) have two LSPR position (one for short axis, another one for the long axis). Only their strength depends on polarization. Nanostars have only one major LSPR.

Download Full Size | PDF

Both the position and intensity of the LSPR are nearly independent from the polarization angle, which reveals the nano-stars of C_3v symmetry act as apolar system with respect to their LSPR properties. More details about the apolar behaviors of different nanostructures are shown in Table 1 : the independence of LSPR characteristics (position and intensity) on the polarization is observed in all nanostructures except the ellipses in which the LSPR intensity decreases almost down to zero for a polarization perpendicular to its major axis.

Table 1. LSPR property of nanostructures under polarization rotation for 4 different shapes: LSPR position with its maximum variation between square brackets; standard deviation of the LSPR intensity with the ratio of the maximum to the minimum LSPR intensity between square brackets

View Table

More interestingly, the positions of LSPR under polarization rotations stay in a short wavelength range for all the nano-triangles and stars, regardless of the shape and size, with a LSPR shift comparable to or even smaller than the one observed for the nano-cylinders whereas such structures are intrinsically apolar. For example, the LSPR shift for the nano-triangles (L = 105 nm, H = 80 nm) is about only 20 nm. Such a shift corresponds to a size drift around 15 nm which is almost the resolution of EBL fabrication. Thus, the observed LSPR shift is not due to specific physical processes or polarization dependence but only to size drift or shape deviation appearing during the EBL process because of technical uncertainties. For the apolar structures, the main interest is the nearly constant LSPR intensity which indicates a constant LSPR excitation, and as a consequence, a highly efficient near-field enhancement by the polarized light. To estimate the variation of LSPR intensity under the polarization rotation, the standard deviation around the average value was calculated. As shown in Table 1, the intensity deviation is under 30% for all the nano-triangles and nano-stars as well as the cylinders, demonstrating a highly constant near-field enhancement with respect to polarized excitation. We assume that the LSPR intensity deviation induce a similar deviation on the near-field enhancement. As shown by the following, we observe also a deviation on the SERS intensity and as a consequence on the detection sensitivity.

The apolar behavior of near-field enhancement in such nanostructures is verified by the SERS measurements with a probe molecule BPE as mentioned above. The evolution of SERS intensity (BPE band at 1200 cm⁻¹) versus the polarization angle given in Fig. 4(a) and 4(b) shows a nearly constant value with small deviations for nano-stars and nano-triangles, respectively. As for the LSPR intensity, we have calculated the standard deviation for the SERS intensity around the average value. As shown in Fig. 4, for both nano-stars and nano-triangles, only a few points are actually outside the [μ−σ, μ + σ] range.

 

Fig. 4 SERS intensity (normalized by average value) versus incidence polarization angle for nano-stars (a) and nano-triangles (b) (length: 100 nm, height, 80 nm), standard deviations around the average value are indicated in the figure (σ = 21% for nano-star and σ = 15.7% for nano-triangle). The SERS intensity was estimated by calculating the area of the 1200 cm⁻¹ BPE band fitted by a lorentzian curves.

Download Full Size | PDF

The deviation under 20% is perfectly acceptable for sensor applications and such deviation can be explained by the LSPR shift and then is due to the imperfections of the nanostructure fabrication appearing from lithography process and not from apolar concept. The larger deviation of SERS intensity for nano-stars could be explained by the wider shape distribution appearing during EBL fabrication. The result of SERS measurements confirms the constant near-field enhancement in such nanostructures whatever the polarization direction.

The results of LSPR and SERS measurements confirm that any complex nanostructures with a C_n symmetry, with n≥3, have apolar behaviors both in far-field and near-field. Such apolar structures are not limited to spherical or cylindrical particles but can have sharp tips until they include a proper symmetry axis with an order n≥3. As discussed above, the nano-stars and nano-triangles of C_3v symmetry show apolar optical behaviors as well as efficient near field enhancement. This model could also be generalized to coupled architectures that could provide “local hot spots” with giant near-field enhancement as long as the symmetry identity is satisfied in the whole architecture. Such coupled architectures could be trimers or any higher multimers of nanoparticles including equally distributed nanoparticles around a central axis. Those apolar plasmonic structures used as SERS substrates or other plasmon-resonance based biosensors make the detection more efficient under different environments.

4. Improvements of SERS sensitivity by using the molecule adhesion layer

In this section, we demonstrate that the adhesion layer between gold and glass plays an important role in the SERS process. To avoid the optical damping caused by the metallic Cr adhesion layer, the organic silane, (3-mercaptopropyl)trimethoxysilane (MPTMS), was introduced into EBL process and used as molecular adhesion between Au nanostructures and glass substrates. We will focus on the improvements of the optical properties of the gold nanoparticles using MPTMS adhesion layer and thus on the SERS sensitivity.

3.1 Introducing MPTMS to EBL process as adhesion layer between glass and Au nanostructures

MPTMS is expected to work as an adhesion layer because its silane moieties could covalently bind to the glass surface through siloxane bonds, while the thiol groups attach strongly to the evaporated gold coating through the well-known Au-S binding. Enlightened by the method of immobilizing colloidal particles on glass surface [39], we introduce MPTMS to the EBL process as a molecular adhesion layer between glass surface and gold nanostructures. The scheme of the EBL process is shown in Fig. 5 .

 

Fig. 5 Schematic presentation of EBL fabrication process with MPTMS: the MPTMS is deposited on glass surface just after the glass treatment; then, parameters for a common lift-off process of EBL are slightly adjusted.

Download Full Size | PDF

The glass functionalization with MPTMS was processed according to the method proposed by Charles A. Goss with slight modifications [40]. Briefly, glass slides were immersed in freshly prepared “piranha” solution (1:3 H₂O₂ 30%:H₂SO₄ 98%) for 30 min and then rinsed with distilled water, dried under a nitrogen stream, and then placed on a hotplate at 100°C for about 10 min. After this procedure the glass surfaces were terminated by hydroxyl groups –OH. Subsequently, the silanization solution was prepared by adding 2 ml MPTMS, 2 ml H₂O to 80 ml of 2-propanol. After the solution was heated to be boiling, the pretreated glass slides were immersed for about 10min and then carefully rinsed with enough 2-propanol, blown dry under a nitrogen stream, and then cured at 110°C for 8 min. This procedure was repeated three times. With the MPTMS molecules covalently bound to the surface, the functionalized glass was then used for EBL process.

To validate the MPTMS relevance as adhesion layer, scratching tests were performed on gold nanostripes with a diamond tip (data not shown here). The results proved that the MPTMS layer has similar mechanical properties than the Cr one and that the MPTMS could actually act as a molecular adhesion layer sufficient for gold nanostructures on glass for sensing applications.

3.2 Improved LSPR properties and SERS sensitivity using MPTMS as molecule adhesive

As confirmed by the scratching tests the MPTMS has similar mechanical robustness than the Cr adhesion layer, it is even worthy to mention that only MPTMS can improve the optical properties of gold nanostructures on glass. To determine the actual effect of the improved optical properties, i.e. LSPR characteristics, on the SERS signal, nano-cylinders were fabricated with the LSPR tuned in the whole visible range by varying the diameters. To make a comparison, Au nano-cylinders with Cr of 3 nm as adhesion layer were also fabricated with the same geometrical parameters. The gap between particles has been kept constant around 200 nm since this distance is large enough to avoid any near field coupling.

Here let us focus on the optimized LSPR properties achieved by using MPTMS as adhesion layer. The extinction spectra shown in Fig. 6 revealed quite different resonance behaviors of Au nano-cylinders with MPTMS or Cr as adhesion layer: comparing to LSPR for nano-cylinders with Cr layer, a much narrower LSPR line width and significant increase of peak intensity were observed in the same sized Au nano-cylinders with MPTMS as adhesion layer.

 

Fig. 6 Extinction spectra of gold nanocylinders (diameter of 100 nm) with chromium (black) and MPTMS (blue) as adhesion layers: the values of the Full Width at Half Maximum (FWHM) are 116 nm and 81 nm respectively for the nano-cylinders with chromium and MPTMS as adhesion layer. The inset was the evolution of FWHM of the extinction spectra for different nano-cylinder diameters measured with chromium (black squares) and MPTMS (circles) as adhesion layers, the fits are represented to guide the eyes.

Download Full Size | PDF

In fact, a decrease of LSPR line width of about 30% was observed for all diameters in the case of MPTMS layer (Fig. 6 inset). The decrease of line width is a competitive advantage which presents a larger quality factor and then better resonance efficiency. Thus, we observed with MPTMS a more efficient light matter interaction or in another sense a larger energy transfer between the excitation light and the Plasmon. In fact, the influence of Cr layer on the Au nano-cylinders lies in two factors: the high refractive index at the glass/particle interface caused a red shift of LSPR peak; high absorption in the adhesion layer quenched the resonance intensity [41]. The increased imaginary part of the dielectric constant of the metal by Cr deposition caused an obvious broadening of the line width of the extinction spectrum. As is well known, the homogenous line width Γ of the Plasmon resonance is a very important parameter which influenced the SPR decay and local field enhancement by its dephasing time T2 = 2ћ/Γ [37,38]. The effective FWHM of the LSPR spectrum measured from an ensemble of nanoparticles (few tens of thousands on the collection spot) could be considered to approach the homogenous line width Γ of the individual nanoparticle LSPR because the inhomogenous broadening due to size and shape dispersion of individual particles could be neglected since such particle features are highly homogenous thanks to the EBL fabrication. Since the near-field enhancement factor ƒ is inversely proportional to the LSPR width, a narrower LSPR indicates a higher local field enhancement. The enhancement factor of SERS is equal to ƒ²(λ₀).ƒ²(λ_R), with λ₀, the excitation wavelength, and λ_R, the Raman wavelength corresponding to the enhanced Raman mode. If we assume that the two wavelengths are close, the SERS enhancement factor can be estimated, in first approximation, as ƒ⁴. As a consequence, the signal enhancement for SERS should be proportional to ƒ^4$\propto$1/Γ⁴, which means the reduced line width could lead to dramatic increase of the SERS signal of any molecules deposited at the nanoparticle surface. As shown in Fig. 7 , the normalized value of 1/Γ⁴ ($\propto$ƒ⁴) is almost about one order of magnitude larger for MPTMS/Au nano-cylinders than for Cr/Au nano-cylinders. The solid lines are Lorentz fittings for scattered values and each showed a maximum around nearly the same cylinder diameter. The curve being sharper for MPTMS/Au compared to Cr/Au nano-cylinders also demonstrates a larger LSPR Q factor with MPTMS layer. Data shown in Fig. 7 predicts much more pronounced local field enhancement for SERS effect in MPTMS/Au nano-cylinders, as observed in Fig. 8 , giving the BPE SERS spectrum obtained with nano-cylinders of 130 nm in diameters. The Raman intensity was about one order of magnitude stronger for the nano-cylinders with MPTMS as adhesion layer compared to Cr layer.

 

Fig. 7 The reverse dependence of the 4th power of line width 1/Γ4 (FWHM) on the nano-cylinder diameters. The continuous and dotted lines are just guide to the eyes.

Download Full Size | PDF

 

Fig. 8 SERS measurements of BPE on Au nano-cylinder of 130 nm with Cr (black) and MPTMS (blue) as adhesive layer. For both spectra, the baseline has been substracted to compare their relative intensity.

Download Full Size | PDF

Thus, the decrease of LSPR line width, observed in the far field extinction measurements, has a clear influence on the near field enhancement. Moreover, SERS intensity is strongly dependent on the LSPR position and needs to be optimized in respect to this latter parameter [20,25,29]. Thus, we compared SERS measurements performed with LSPR continuously tuned in a wide visible range for Cr and MPTMS layers, respectively. Figure 9 presents the Raman intensity plotted against the LSPR position of Au nano-cylinders with diameters varying from 80 to 300 nm. As shown in Fig. 9, much stronger SERS signal was observed for MPTMS/Au nano-cylinders (about one order of magnitude larger compared to the SERS signal obtained with the Cr/Au nano-cylinders) whatever the LSPR position is.

 

Fig. 9 Evolution of SERS intensity versus LSPR position of Au nano-cylinders (square, with Cr; circle, with MPTMS), the dashed lines are Lorentz fitting of the measured data.

Download Full Size | PDF

This means that the improvement of SERS is then intrinsic to the change of adhesion layer but not an effect of the LSPR position. Also observed in Fig. 9, the evolution of Raman intensity versus the LSPR position shows the same trend: both gave a maximum of Raman signal around 645 nm, exactly between the excitation wavelength (632.8 nm) and the Raman wavelength (685 nm). Such results are in good agreements with Wokaun’s model and consistent with our previous work [20,25]. As is well accepted that SERS process could be optimized by tuning the LSPR at proper position, results shown in Fig. 9 demonstrated that SERS signal could be further amplified by using MPTMS as adhesion layer. We attribute such improvements of SERS sensitivity mainly to the optimization of LSPR by using MPTMS as adhesion layer and more especially to a narrower line width of the LSPR inducing a higher near-field enhancement.

Besides the electromagnetic mechanism, other factors may also have influences on SERS efficiency such as the chemical interactions between Raman molecules and metal surface. Especially, when Cr was used as adhesion layer, it could diffuse along the grain boundaries to the metal surface. Such inter-diffusion or alloy formation between Cr and Au may change the surface properties of metallic particles, which could also restrain the SERS signal [42]. Such negative effects induced by Cr could be avoided by using MPTMS as adhesion layer, thus further optimization of SERS could be achieved.

Anyway, using this molecular adhesion layer, optical properties of gold nanostructures was greatly improved by showing a much reduced LSPR line width. As expected, an enhancement of SERS intensity of one order of magnitude was achieved due to the higher near-field enhancement. The improved LSPR properties pave the way of highly sensitive plasmon-resonance-based biosensors.

5. Conclusions

In the present work, we have presented the optimization of plasmonic nanostructures to improve their SERS activities in order to induce high sensitivity and flexible manipulation. All these results are mainly based on the improvements of LSPR properties of metallic nanostructures essential to the sensing performance. First, we presented the design and fabrication of a new family of complex nanostructures with apolar behavior. We have then be able to demonstrate that their far field (LSPR) and near-field (SERS) interaction with light are independent of the excitation light polarisation. This new family of nanostructures is only defined from symmetry consideration. Indeed, we demonstrate that any complex nanostructure with a C_n symmetry (with n≥3) has optical responses insensitive to the light polarization. Such apolar plasmonic structures enable a wide flexibility the design and sensing activity of plasmon based sensors. Second, a new process for the adhesion layer was proposed allowing an adhesion improved of EBL fabricated gold nanostructures on glass surfaces. We demonstrated that the SERS sensitivity could be further increased by using MPTMS as adhesion layer instead of Cr. An improvement of the SERS signal of one order of magnitude was achieved due to a higher quality factor of the LSPR and as a consequence due to a higher near-field enhancement.