Disordered array of Au covered Silicon nanowires for SERS biosensing combined with electrochemical detection

We report on highly disordered array of Au coated silicon nanowires (Au/SiNWs) as surface enhanced Raman scattering (SERS) probe combined with electrochemical detection for biosensing applications. SiNWs, few microns long, were grown by plasma enhanced chemical vapor deposition on common microscope slides and covered by Au evaporated film, 150 nm thick. The capability of the resulting composite structure to act as SERS biosensor was studied via the biotin-avidin interaction: the Raman signal obtained from this structure allowed to follow each surface modification step as well as to detect efficiently avidin molecules over a broad range of concentrations from micromolar down to the nanomolar values. The metallic coverage wrapping SiNWs was exploited also to obtain a dual detection of the same bioanalyte by electrochemical impedance spectroscopy (EIS). Indeed, the SERS signal and impedance modifications induced by the biomolecule perturbations on the metalized surface of the NWs were monitored on the very same three-electrode device with the Au/SiNWs acting as both working electrode and SERS probe.

Scientific RepoRts | 6:25099 | DOI: 10.1038/srep25099 accommodation of SiNW based biosensors on plastic and glass provides an effective and cheap method to integrate these sensing nanoelements into macroscopic and/or commercial readout devices including assay plates and microscope slides implementable in benchtop microscopies. The idea of this work is directed to investigate the very same composite structure as Raman probe, allowing ultimately the combination of SERS detection with EIS control. This combination of an optical and electrochemical signal transduction in the same Au/SiNWs can offer an improved method to identify and quantify the bio-analytes over either technique independently. Indeed, EIS is widely used in diagnostic biosensing, because it allows detecting a wide range of analytes with high sensitivity by using chip and miniaturized devices 30 . On the other hand, the impedance biosensors do not provide chemical identification of the analyte and can reveal also non-target biomolecules stuck to the probe causing a false positive signal. This is a critical issue when biological fluids containing a large background of nonspecific interferents (blood, saliva etc.) are analyzed. The SERS offers the complementary chemical information due to the ability to identify analytes from unique Raman signatures reducing the issue of false positive signals.
To show the potential of this approach SiNWs, few microns long, were grown by plasma enhanced chemical vapor deposition (PECVD) and covered by an evaporated thin Au layer, 150 nm thick. Common microscope glass slides were used as substrates to demonstrate these nanostructures can be built via scalable and low-cost fabrication methodology on supports that can be easily integrated into Raman spectrophotometers. The properties of Au/SiNWs as Raman probe and their sensing performance were studied via the ligand-receptor binding of Effects of the NW morphology on the optical properties. Before performing Raman experiments, we investigated the optical properties of the Au/SiNW composite system. The effects of morphology on the optical properties of semiconductor NW arrays are well known: the light passing through a NW forest undergoes where R Au and R Au/SiNW are the reflectivity of the planar Au and the Au/SiNWs, respectively (b). In the inset the enlarged ∆R R at smaller wavelengths. multiple scattering events from each NW. The scattering events fold the light path many times in a random walk inside the NW forest causing an enhancement of the light absorption with consequent drastic reduction of the reflectivity at the frequencies absorbed by the NW material, i.e. light trapping [31][32][33] .
To understand the effect of the NW light trapping to the overlying Au film in our composite system, we measured its total integrated reflectivity, R, in the spectral range between 200-2000 nm and compared it with that of a reference sample consisting of a planar Au film with the same nominal thickness of the NW coverage. In Fig. 2a, spectra obtained from the Au/SiNWs (red line) and reference sample (blue line) are plotted and compared. The reference sample shows the typical Au bulk reflectance 34 of around 95% in the infra-red (IR) range with a steep decrease below λ = 650 nm.
Although the profile of R for the Au on NWs remains quite similar to that of the planar Au, the intensity of R is drastically diminished in the nanostructered sample over the whole spectral range. R is in fact reduced down to 20% in the IR and few percent in the visible range. The NW sizes and their broad size distribution, as evidenced by the SEM images, assure indeed a large light scattering at all wavelengths in the IR-visible range by allowing the Au on the NW to absorb significantly more photons than an equivalent volume of bulk material [31][32][33] . Hence the reduction of the R intensity, defined as = where R Au and R Au/SiNW are respectively R of the planar and nanostructured Au, corresponds to the amount of light trapped inside the NW forest and then absorbed by the overlying Au film. If we plot ∆R R as a function of the wavelength (Fig. 2b) we can observe a quite constant value of about 80% over the whole IR range and a peak of 92% at 532 nm with a shoulder at around 610 nm, inset of Fig. 2b. The shape of the curve and peak position in Fig. 2b suggest the mechanism by which such an absorption occurs could be transverse and longitudinal modes of a plasmonic resonance 35 probably related to the smaller NWs or NW section, which after Au coverage form Au nanopillars. Anyway, a full understanding of the origin of the plasmonic resonances observed in our composite NWs requires deeper investigations that are beyond the scope of this work and will be performed in more details elsewhere.

Avidin detection by Raman spectroscopy.
To test the capability of Au/SiNWs to act as SERS biosensor we progressively modified their surface by adding cysteamine to achieve a self-assembled monolayer (SAM) coating, NHS-biotin to build the sensing layer, finally avidin as the detected material.
Visual inspection of the sample after each functionalization step shows a change of the surface color as it results in Fig. 3 where optical images of the pristine Au/SiNWs (Fig. 3a), modified with cysteamine ( Fig. 3b) and NHS-biotin (Fig. 3d), and after immersing in an avidin solution with a protein content of 1 μM (Fig. 3e) are shown. As control experiment we immersed in the 1 μ M avidin solution the Au/SiNWs coated by sole SAM layer, and Fig. 3c shows the related optical image. Any change in the surface aspect is visible after cysteamine SAM formation ( Fig. 3b), maybe due to the very small size of the cysteamine monolayer which does not produce any interference effects visible with an optical microscope.
After NHS-biotin treatment a change of the surface color from green to blue can be noticed (Fig. 3d), indicating a quite uniform adsorption of the NHS-biotin molecules onto the Au/SiNW surface. Then the blue color becomes attenuated in a very large areas of the samples after avidin exposure (Fig. 3e), whereas any color change can be observed when avidin is added on the sample modified with sole cysteamine (control experiment in Fig. 3c). These observations suggest the avidin immobilization onto the surface of the Au/SiNWs through its capture from the biotin ligand. This is best confirmed by the Raman analysis in Fig. 4a where the Raman spectra corresponding to each step involved in the biosensing experiment are plotted.
We observe a progressive increase of the signal in the spectral range comprised between 1280 and 1450 cm −1 . In particular, a quite large band at about 1380 cm −1 appears in the spectra obtained after the last passage with avidin. This contribution is ascribable to the specific interaction of avidin with biotin 36 thus demonstrating the stable immobilization of the target molecule onto the treated substrate. In addition, two control experiments were carried out. As first, to confirm that the observed Raman spectra changes in Fig. 4a are due to the specific binding of avidin to the biotin ligand we immersed in the 1 μ M avidin solution the Au/SiNWs coated by the sole SAM layer, i.e. without the NHS-biotin ligand. As we can see in Fig. 4b by comparing the Raman spectra of the Au/SiNW coated with cysteamine before (Au/SiNW+ Cyst) and after immersion in avidin solution (Au/ SiNW+ Cyst+ Avd), the avidin did not produce any significant change in the signal suggesting a little non-specific binding of the protein to the bare cysteamine modified Au/SiNWs. The second control experiment was realized to evidence the essential role played by the SiNWs in the enhancement of Raman signal. For this purpose the sensing capability to detect avidin was also tested on planar Au film (Fig. 4c), which was coated with cysteam-ine+ NHS-biotin and successively immersed in the 1 μ M avidin solution. Any relevant signal variation can be observed between the pristine NWs and after each modification step, suggesting that the nanostructuring of the Au improves the Raman signal. The enhancement of the Raman signal in the Au/SiNWs is most likely due to the occurring of plasmonic resonances observed in the spectrophotometric investigation.
On the basis of the control test, we explored the sensing ability of the biotin modified Au/SiNW to detect avidin in a wide range of concentration by exploiting the Raman signal without using blocking molecules. Despite the non-perfect uniformity of target molecule adsorption over the whole sensor surface, a definite trend of Raman signal enhancement with protein concentration can be identified. This behavior can be clearly recognized in Fig. 5 showing Raman spectra at different avidin concentrations extending from 1 μ M down to 1 nM, as reference we have also plotted the spectrum before protein adsorption (black line). When avidin is added, the intensity of the band at 1380 cm −1 increases with the specific concentration of the target solution. The band intensity evolution can be easily quantified by applying a fitting procedure to the collected spectra and used as transduction signal for the biosensor. The figure in the inset exhibits, as an example, the result of the fitting procedure applied to the spectrum collected at 100 nM, demonstrating the presence of the quite large band ascribed to avidin binding (blue line).

Avidin detection by combined Raman and electrochemical probes.
We tested the biosensing capability of the Au/SiNWs to detect avidin through two different and independent transduction signals: Raman and electrochemical via impedance measurements. For this purpose, we exploited the Au coverage wrapping the SiNWs as working electrode (WE) of a three-electrode device, like as shown in Fig. 6a and discussed in more details in ref. 29. This device is simply obtained by adding to the Au/SiNWs, the Au counter electrode (CE) and the Ag reference electrode (RE). We note that the Au/SiNW WE was fabricated in the very same conditions of the above discussed Raman substrates. This configuration enables the collection of both the Raman and electrochemical signal coming from nanostructured WE.
The impedance measurements were performed, after Raman detection, in KCl solution by connecting the electrodes to the potentiostat. The avidin content range was varied between 10 pM and 1 μ M, and Fig. 6b shows the related impedance modulus, |Z| measured in the frequency range extending from 100 KHz down to 1 KHz. When avidin is added, a significant and continuous increase of the |Z| values in the whole frequency range analyzed is observed. In Concerning the sensitivity limits achieved through the two detection methods, the impedance analysis is characterized by a value as low as 10 pM whereas the Raman signal shows, at the chosen laser power of 3 mW, a higher value of 1 nM. This is due to the fact that for lower concentrations the Raman signal would require a higher excitation power, associated to a different calibration curve, but such as not to damage and alter the biological sample. Moreover, we expect an improvement of performance for both the sensing responses by optimizing some key material features, including the NW size, Au coverage thickness and uniformity. Nevertheless the low detection  Table 1, where some works reported for the (strept)avidin detection based on nanosized materials via optical or electrochemical approach are summarized. In addition our biosensor benefits the

Ref. Sensing Material Signal Transduction Target Low detection limit Working range
Cui 2001  unique advantage to couple the SERS chemical information with the electrochemical quantification that increases the analysis efficiency and enlarges the application areas over the common approach based on detection via one transduction mechanism.

Conclusion
We explored forest of SiNWs covered by a Au layer as effective substrate for enhancing Raman signal of the adsorbed molecules. We showed several unique features of these systems. First, the NW morphology imparts an enhancement of light absorption in the overlaying Au with the occurrence of plasmonic peaks resonant to the wavelength of Raman excitation source. Second, the disordered arrangement of the NWs offers the advantage to use high yield techniques, such as PECVD, that involve relative low temperature procedures compatible with non-conventional substrate such as microscope glass slides, rapidly implementable in Raman spectrophotometers. Third, the design involving a conductive coverage of the SiNWs allows developing innovative devices able to detect one analyte through both Raman and impedance transduction signals by increasing the analysis efficiency and reducing the detection issues related to nonspecific interaction. In conclusion, the proposed composite system not only is efficient Raman substrate, fully compatible with the industrial integrated silicon based technology, but also can be designed aiming at multi transduction methods, thus putting in perspective the application of SERS in innovative analytical contexts. Surface functionalization of SiNWs. The bio-sensing tests were performed using the biotin-avidin scheme. The immobilization of the NHS-biotin group onto the Au/SiNWs was performed via the modification of the Au/SiNWs surface with SAM by immersing the samples in cysteamine solution 20 mM and incubating them overnight at 70 °C. They were rinsed off the residual cysteamine molecules with deionized water, Raman (and also impedance in the case of the three-electrode device) spectra were measured to ensure that the cysteamine molecules had been linked to the NW surface. The cysteamine functionalized samples were immersed in a fresh solution of 10 mM NHS-biotin in pH 7.4 phosphate-buffered saline (PBS) solution for 60 min at room temperature, and then cleaned with PBS. Also in this case, the immobilization of NHS-biotin onto cysteamine was controlled by Raman (impedance) investigation. Afterwards, the biotin-functionalized Au/SiNWs were immersed in 1.5 mL of aqueous avidin solution with different concentrations ranging from 1 μ M down to 10 pM for 30 min at room temperature, rinsed again in PBS and dried before being tested. For each avidin concentration fresh samples were used and the experiments were sequentially conducted after cysteamine coverage, NHS-biotin modification and avidin detection. In order to treat with the biomolecules only the area having the NWs, we delimited it with removable Teflon wells fixed on the NW area under mild pressure.

Methods
Morphological and optical characterization. The morphology of the SiNWs was verified by field emission scanning electron microscopy (FE-SEM), equipped by an energy dispersive X-ray spectrometer for composition analysis. The SEM images were acquired at an accelerating voltage of 5 kV. The EDX spectra were measured at 10 keV. The optical properties were studied by measuring the angle-integrated total reflectivity in the spectral range between 200 and 2500 nm with UV-Vis-IR spectrophotometer equipped with an integrating sphere.
Raman spectroscopy. Raman spectra were acquired with a DXR Thermo Fisher Scientific Raman Microscope, by exciting the samples at 532 nm with a 5 mW power and a 50× objective. The measured spectral range was 50-3000 cm −1 , and each spectrum resulted from 1 s acquisition time and 200 accumulations over an area of about 800 nm (laser spot). The spectra were analysed performing a lorentzian fit to obtain the values of the band intensities calculated as the area of the selected spectral features.
Impedance measurements. The impedance measurements were performed on the three-electrode device at room temperature in a Faraday cage at DC 0 V and with an AC signal of 10 mV in a stimulus range between 30 mHz-100 KHz. The tests were carried out in a solution 100 mM of KCl in deionized and distilled water. A test circular area of 1.5 cm 2 , with the Au/SiNW WE at the center of that area and part of RE and CE immersed in the solution, was delimited by using a Teflon tube. In order to avoid solution leakage, an O-ring was placed between the test sample and the bottom of the Teflon tube. The potentiostat used was a versaSTAT 4 by PAR. Each measurement was performed with a minimum of four repetitions on the same electrode over a time period of 1 h.