Engineering of Nanostructured Plasmonic Substrates for Use as SERS Sensors

Plasmonic nanostructures strongly localize electric fields on their surfaces via the collective oscillations of conducting electrons under stimulation by incident light at certain wavelength. Molecules adsorbed onto such surfaces experiences a strongly enhanced electric field due to the localized surface plasmon resonance (LSPR), which amplifies the Raman scattering signal from these molecules. This phenomenon is referred to as surface enhanced Raman scattering (SERS). Further enhancements in the Raman intensity have been achieved by designing plasmonic nanostructures with a controlled size, shape, composition, and arrangement. This review paper focuses on the theory and analyses the influence of protective coating with oxide materials an isolated plasmonic metal nanostructures. Starting with a brief description of the basic principles underlying LSPR and SERS, we compare two plasmonic metals, two dielectric materials and the effect of changing individual parameters of the nanostructure on output enhanced Raman signal.


Introduction
Surface-enhanced Raman scattering (SERS) For a direct, label free and non destructive characterization of molecules, vibrational spectroscopes are efficient interesting techniques. Based on the determination of molecular or crystalline vibrations, Raman scattering provides many information about the studied object whatever is its nature (liquid, solid or gas). It enables the description of the chemical composition (atomic composition and chemical group) and structure (geometry, conformation...) of almost any kind of molecules or crystal [1]. However, Raman scattering is known to present a weak cross section meaning that only a weak part of the incident energy will be effectively scattered inelastically by the observed object. The Raman signal is thus very low( ≈ 10 −30 2 ) compared to others spectroscopies such as fluorescence or infrared spectroscopies. Thus, it is necessary to use a large quantity of matter to collect exploitable Raman signals. Typically, to study proteins in liquid phase for example, a concentration higher than 10 −4 mol −1 is required to obtain sufficient Raman signal with acceptable signal to noise ratio. For lower concentrations, it is necessary to increase the electromagnetic field around the target molecule. One way is to add nanostructured metallic film under irradiation and localized surface plasmon resonance (LSPR) at wavelength will appear, meaning the appearance of surface waves corresponding to a collective oscillation of the excited conductive electrons or plasmons and local electromagnetic field will be induced. If molecules are deposited on or in the vicinity of the surface, the Raman signal will be enhanced. So detection of low concentrated molecules requires metallic surface nanostructure that promote enhanced Raman scattering (SERS). Raman spectra are typically measured from monochromatic light source (laser) to excite the molecules. Scattered light then detected by a CCD camera. Scattered photons that lose energy are called Stokes scattering, and anti-Stokes scattering corresponds to photons that gain energy relative to the energy of the incident photon. The magnitude of the energy shift is called the Raman shift (∆ ) and can be expressed by [2]: Where 0 and are the wavelengths of the incident and Raman scattered beams, respectively.
The origin of SERS phenomenon is still not well understood but community agreed on the existence of two principal contributions: electromagnetic or plasmonic effect (ensemble of mechanisms leading to an enhancement of the electromagnetic field [3]) and chemical effect which encompasses phenomenons that generate the variation of polarisability of a probe molecule. It is admitted that contribution of the chemical effect would not exceed a factor 100 and thus can be neglected in many casess compared to the electromagnetic effect that gives exaltation factor of 10 8 or even 10 10 [1].
Electromagnetic effect include two distinct processes [1]. Creation of local electromagnetic field ( ) by plasmons. The enhancement factor 0 can be expressed as: Where is the enhance local field by the nanoparticle at the incident wavelength 0 .
The second process results in the enhancement of the Raman scattered signal from probe molecule (at ) at the vicinity of the nanoparticle surface and it's called re-radiation process. The enhancement factor at the wavelength in this process is: So the final enhanced intensity of the Raman signal scattered by the molecule can be written as: Where 0 the Raman signal intensity scattered by the molecule.
If these two wavelengths 0 and are close to each other it is considered that this two factors are approximately equal. So SERS enhancement is at fourth power of the local enhancement of the electromagnetic field produced by the nanoparticle ≈ [ ] 4 . Hovewer, these two wavelengths ( 0 and ) can be separated by several tens on nanometers and, in this case, the wavelength shift between 0 and cannot be neglected. From the literature should be located at some intermediate energy between the excitation 0 and Raman electromagnetic fields to provide the maximum enhancement of the two factors [1,4].
The maximal enhancement factor (EF) can reach the value 10 11 or even 10 12 [5,6] and can be calculated as [4]: Where is the number of molecules bound to the enhancing metallic substrate, is normal Raman intensity and is number of molecules in the excitation volume.

Plasmonic Metals
Choice of the metal for use in SERS applications strongly depend on its optical properties.
If, for example, spherical metallic nanoparticle is illuminated with incident electromagnetic field , this leads to the appearance of an induced electric field around nanoparticle and this field can be expressed as: Where is dielectric function of metal and is relative permitivity of the environment. This function is resonant while the real part of the dielectric constant of the metal ( ) is equal to −2 . In this case the enhancement of the surface plasmons increase the local field. For gold and silver this condition can be respected in visible part of the electromagnetic spectrum.
In its turn gold widely preferred to silver for biological applications for its relativey easier surface chemistry, the possibility of attaching molecules via thiol groups, good biocompatibility, and chemical stability [7].
The surface roughness of metallic films [8] or areal density of nanoparticles (NPs) [9,10], shape of nanostructures [11] and environment around NPs (Equation 6) are a keys points to enhance the electromagnetic field for SERS measurements. About influence of the plasmonic metals shape on SERS activity we will talk in next paragraph.

Topology of gold nanostructures
Surface-enhanced Raman spectroscopy (SERS) depends on the highly intense localized electromagnetic fields (also known as "hot spots") produced by the process of surface plasmon resonance (SPR). In its turn, factors of SPR property are very sensitive to the parameters of plasmonic metallic nanoparticles including size, shape, morphology, distribution, and a surrounding environment.
Chemical synthesis is the most widely used method for preparing various kinds of plasmonic metallic structures, such as nanospheres, nanorods, nanocubes, nanostars [12][13][14]. However, using this approach it is difficult to control the distance and ordering of nanostructures in solution.
By far, the lithographic method is gaining popularity because it can give desired high-resolution, and its can be used to fabricate precisely controllable dimensions, interparticle spacing, and ordering of nanoparticles on substrates. There are different kinds of this technique such as electron beam lithography [15,16], on-wire lithography [17] ,nanoimprint lithography, photolithography, laser interference lithography [2], etc. Table 1 Influence of changing each design parameter (independently) on the system spectral response, while keeping the other parameters fixed, including the pitch (p x q) which is maintained to 300 x 300 2 [18]. Where -resonant wavelength, T -transmittance, R -reflectance, A -absorbance.

Monopoles
Dipoles Geometry of a unit cell of the system under study: an Au rectangular dipole antenna on a silicon substrate covered by water. An x-polarised plane wave source illuminates the antenna in the z-direction from the substrate [18].
The high degree of freedom available for such technique has enabled studies of the basic principles underlying electric field enhancements by welldesigned metallic nanostructures. To find out the optimal topology of the nanostructure is the central question in SERS applications. Saba Siadat Mousavi et al [18] have been done a great work in determining the influence of length, width, thickness and gap of periodic rectangular gold nanomonopoles and nanodipoles on the transmittance, reflectance and absorptance response. In this work, infinite arrays of rectangular gold nanomonopoles and nanodipoles on a silicon substrate covered by water ( Fig. 1) have been investigated extensively via 3D finite-difference-timedomain simulations (FDTD). Results of this paper gathered in Table 1.
Looking at Table 1 we can say that, for example, with increasing the length of the dipole from 190 to 280 nm, the resonant wavelength red-shifts, decreases the amount of power transmitted at resonance, and increases the level of reflectance of the system, but the absorptance remains almost unchanged.
To confirm that these simulations are right we can look at the work of Bryan F. Mangelson et al. [17]. Using on-wire lithography (OWL) they synthesized a composite plasmonic-semiconductor material composed of Au nanorod dimers embedded within anatase 2 sheets. Optical properties of these structures can be tailored via the geometric control afforded by the OWL process to produce structures with various gap sizes exhibiting different electric field intensities near the surface of the metal particles. Spectra on (Fig. 2 a) were obtained by subtracting a solution containing only 2 sheets as a background. It can be seen that as the gaps get larger the plasmon resonance of the structures blue-shift. The plots on (Fig. 2 b) show the intensity of the field at the resonances for each dimer, which are 1100, 1050, and 1030 nm, respectively. It is observed that as the gap becomes larger the electric field intensity decreases, showing the tunability available using this system. Additionally it can be seen that the electric field penetrates well into the semiconductor material. Electric field intensities are plotted on a log scale where zero is equal to the intensity of the incident field. So experimentally has been confirmed that the position of the dipole plasmon resonance shifts with increased coupling between rod segments as a result of the changing gap size.
In work Zhibing Zhan et al [11] have been fabricated, via a cost-effective nonlithographic route, perfectly ordered nanoparticle arrays with identical shape, uniform size and distance between these plasmonic gold nanoparticles on large-area substrates (> 2 ). Experimentally has been shown that with increasing the height (or in other words thickness) of nanopillar array we get a blue shift in resonant wavelength (Fig. 3). Also here excitation of multipole SPR modes will not necessarily result a peak broadening effect. In fact, the width of the SPR peak mainly involved in the uniformity of NPs, including size, shape and distribution in addition to the multipole excitation. Increasing heights strengthened the restoring force to the displaced electron gas and electromagnetic coupling among neighboring nanoparticles, which leads to a shorter resonance wavelength.
In another work of Hyung Ju Kim et al [19] sizecontrollable Au nanodot arrays (height of nanodot 12-15 nm) on Ti adhesion layer (ca. 1 nm) are directly printed onto indium tin oxide (ITO) glasses and then coated with 2 layer (110 nm in thickness) by dip-coating followed by annealing process at 300 . Experimentally has been shown that the plasmon resonance wavelength ( ) experiences a redshift as the Au nanodot size (or length) increases from 50 nm to 83 nm (Fig. 4 a). Increasing the gap (at the same time decreasing the size or length   ( Fig. 4 c-e)) between nanodots we can see blue shift ( Fig. 4 a, b) and decrease in absorbance (for bare gold).
But not all possible structures of the gold topography could be interpreted as arrays of monopoles and dimers. Another way of designing gold topography was proposed by Alyssa J. Pasquale et al. [16] where thanks to the E-beam lithography were formed concentric necklace nanolenses which consist of metal nanoparticle dimers placed in the center of one or more concentric rings of plasmonic necklaces. The introduction of concentric rings around metal nanodisk dimers significantly concentrates the electric field at the gap within the dimer region. The intensity of the electric field increases as the number of necklaces increases, as shown in the FDTD simulations presented in (Fig. 5).
In work of Bouvree et al. [15] by E-beam nanolithography, two kinds of patterns, lines and circles were formed from Au nanoparticles with fixed size of 22 nm (Fig. 6 b,c). The interline and inter-circle distances were set from 20 to 2000 nm. Two line thickness values, 30 and 60 nm, had been achieved, corresponding to bands of two or three AuNP widths. It should be pointed here that "line"patterns were sensitive to the light polarization, whereas "circle"patterns were nonsensitive. Signals recorded on concentric rings were lower than observed for lines. (Fig. 6 a) displays experimental Raman scattering results obtained on various patterns. The intensity (counts/s) of the band at 1, 620 −1 (crystal violet) is reported as a function of the distance between lines.  5. Calculated near-field intensity distributions at 785 nm of (a) isolated dimer, (b) dimer/enneadecagon, and (c) dimer/ enneadecagon/tetracontagon. Scale bars correspond to 500 nm. (d) Near-field intensity spectra of an isolated dimer (red) embedded into optimized single (green), double (blue), triple (blue), through seven (navy) necklace(s). Inset shows the evolution of the maximum near-field intensity with the number of surrounding necklaces, N, fitted to a logarithmic function [16].

Dielectric oxide materials
From practical application viewpoint, not only strong enhancement but also stability, recyclability as well as cost-effective preparation methods are necessary for SERS sensors productivity. To this end, the fabrication of SERS-active substrates with optimized properties is still faced with numerous challenges. Because of the mixture among metal nanoparticles, solutions and target molecules, the nanoparticle metal surface influenced by the adhesion of solutions and target molecules, thereby influencing the accuracy of the Raman spectra of the target molecule.
To reduce these errors and incorporate new functionalities, these plasmonic metal nanoparticles have been coupled with various dielectric oxide materials (e.g., 2 [12,20], 2 [10], 2 3 [21] etc.) to form metal-oxide hybrid nanostructures and prevent contact between the metal surfaces and target molecules. For example in work of Lingwei Ma et al [22] slanted silver nanorodes were prepared on oblique angle vapor deposition (OAD) technique and then covered with different 2 layers thicknesses using atomic layer deposition (ALD) (each ALD cycles added 0.55 nm of 2 thickness). (Fig 7 a) shows influence of the covering layer thicknesses on the output intensity of the Raman signal from 10 −5 M crystal violet molecules. Also temporal stability of bare Ag and @ 2 NRs was supervised via counting the SERS intensities of 10 −5 M CV from the stored substrates as a function of time. (Fig 7 b) displays the normalized Raman intensities at 1171 −1 band obtained from different substrates over a period of 48 days. As can been seen, CV signals on bare Ag NRs emerged a substantial decline even only after 8 days ( 50% drop), and were more than one order smaller after 48 days. However, when the NRs were uniformly wrapped with ultrathin  nanostructures. (c) Optical-extinction spectra of bare gold NPs (50 nm in diameter), as well as Janus and core-shell @ 2 nanostructures in isopropyl alcohol, which were taken from experimental measurements [10].
stability was dramatically enhanced.
Adding new material also change combined optical, electronic and magnetic properties. Via depositing high optical index of refraction film on metal surface we can tune the LSPR peak position [23]. Increase in the refractive index of the dielectric environment that surrounds the plasmonic metal nanoparticle causes red shift in LSPR peak position [10,12,21,23]. Even the covering deposition technique, the way of covering [10] (Fig 8 a,b), sequence in which this oxide material was added [24,25], annealing temperature with which material was treated [26] (Fig 9), thickness of this additional Fig. 9. Evolution of the LSPR peak position as a function of the annealing temperature of 2 host matrix with embedded gold NPs [26].
In this study we compare two materials for protective covering shell: 2 and 2 . The task is to decide which material to use and what parameters it has to have to produce the best SERS results. Looking at the previous results [21,27] we can assume that thickness changing of 2 shell with the Au core even on 1 nm can dramatically damp the scattered signal (Fig 11) (minus 1 order of electric field magnitude from 1 nm to 2 nm thickness of covering layer). On the figure below we can see that 2 as covering layer material is very sensitive to the thickness variation. So it would be difficult to produce the same substrates with the same properties without any changes in a layer thickness, even on 1 nm.
In the work of Sergey Scherbak et al. [23] were tested different thicknesses of 2 layer on a gold films (Fig. 10). Was determined optimal thickness of  layer (about 20-30 nm for a truncated gold nanospheres of 8.8 nm in radius on the glass substrate), where the induced electric field is almost fully localized in the covering layer and the particle barely "feels"any further increase in the coating thickness. It was demonstrated that the can be precisely shifted up to 100 nm towards longer wavelengths by depositing titania layer on the island film. So we can conclude that the thickness changing of 2 material as covering layer is not so crutial for the final enhancement.
Another way to study the enhancement mechanism in such bilayer thin films is submit these substrates as photoelectrodes. With this approach we can characterize the enhancing mechanism throught photoconversion in metal-semiconductor heterojunction-based photovoltaics and photocatalysts.
In metal-semiconductor heterojunctions, the plasmon is able to induce charge separation in the semiconductor via three mechanisms: (i) photonic enhancement, (ii) direct electron transfer (DET) also known as hot electron injection, and (iii) plasmoninduced resonance energy transfer (RET) [3,24,28].
For the photonic enhancement, the plasmon can interact with the semiconductor by far-field radiation, improving light absorption in the semiconductor via light trapping. It can happen only when the incident photon's energy is larger than the band gap of the semiconductor [3].
For DET, the nonthermalized electrons excited by a localized surface plasmon resonance (LSPR) have energy above that of the Fermi level in the metal, allowing the Schottky barrier created by charge equilibrium between the metal and the semiconductor to be overcome in the excited state [29]. Thus the electron transfer process does not require the energetically favorable equilibrium band alignment. On (Fig. 12) illustrated the scheme of these electron transfers. The gold core converts the energy of incident photons into localized surface plasmon resonance oscillations and transfers the plasmonic energy to the 2 semiconductor shell via resonant energy transfer. RET generates electron-hole pairs in the semiconductor by the dipole-dipole interaction between the plasmonic metal (donor) and semiconductor (acceptor), which greatly enhances the visible-light photocatalytic activity as compared to the semiconductor alone [28].
In the work of Scott K. Cushing et al. [3] nanospheres of gold (50 nm in diameter) were covered with 10 nm of 2 and then with 20 nm of 2 . Transient absorption spectroscopy was chosen to experimentally determine the plasmonic enhancement mechanism, since it allows the transfer and creation of carriers to be measured on a picosecond time scale, before recombination can reduce the carrier population. It has been found that 2 serves as an insulating barrier, whose 10 nm thickness is sufficient to block hot electron transfer but still allowed coupling through the plasmon's near field.
As for 2 layer, it was found some interesting properties. In the work of Hongjun Chen et al [24] only by switching the deposition sequence of the 2 layer on a simple bilayer structure of 5 nm thick Au and 10 nm 2 on fluorine-doped tin oxide (FTO) as a substrate (Fig. 13), it is possible to change the photocurrent direction under visible-light irradiation. It is found that when Au layer is deposited as the outer layer on 2 coated FTO substrate, an anodic photocurrent is obtained due to the band bending formed at the electrode-electrolyte interface. But when Au is under 2 layer the cathodic photocurrent appears.
Here photogenerated electrons can be easily trapped in the energy well formed between the band bending and the Schottky contact, which allows electronic tunnelling through the 1.6 nm thick space charge layer, resulting in a unique anodic to cathodic photocurrent conversion.
For our task cathodic photocurrent is more desirable since the generated photovoltage under visible illumination are totally changed into positive values, suggesting positive charges are accumulated at the surface. If the 2 layer of / / 2 photoelectrode increases to 100 nm, the thus-obtained photoelectrode still keeps cathodic photocurrent [24], suggesting the tunnelling effect still exists, not being dependence of the thickness of 2 layer. It is also worth to mention work of Hyung Ju Kim et al [19] where has been found that the activity of each catalyst (Au nanodot array with different sizes on Ti adhesion layer on ITO substrate) increased by a factor of 6 times when 110 nm of 2 was added to the Au nanodots for all samples.

Conclusion
Over the past four decades, since the discovery of SERS, a series of theoretical and technological innovations have stimulated broad interest in plasmonic nanomaterials. This review was devoted to the analysis of the of the latest research on the elucidation of occurrence of Raman signal enhancement mechanisms. At the beginning the reader was introduced to the SERS phenomenon. Then it was said why it is better choose gold for a role as a plasmonic metal over silver. The function of sensor's recyclability and stability with two oxide materials 2 and 2 were analysed. It has been found that change of upper protective layer thickness in case of 2 at 1 or 2 nm is not so crutial as for 2 . This allows the production of sensors with less precision, and hence cost. The main trends in the signal change were considered depending on changes in individual parameters of the nanostructure in order to determine the optimal topography of the metal surface.