[0001 ] The present invention is related to a method for spatially selecting an enhanced sum frequency generation signal.

[0002] The present invention is also related to a sum frequency generation spectrometer implementing such a method.

State of the Art

[0003] The plasmonic effects taking place in metallic nanostructures and giving rise to major local electric field enhancements have attracted much interest from the scientific community for their outstanding possible applications.

[0004] A well-known example is surface enhanced Raman spectroscopy

(SERS), fo r wh i ch th e u se of n a n ostru ctu red su bstrates h as tu rn ed Ra m a n spectroscopy into a highly sensitive and efficient surface spectroscopy.

[0005] It is also known from document US2010/0148049 A1 that nanostructures may be used to enhance a sum frequency generation signal. In that document, no hint is given to a spatial selectivity of the signal enhancement.

[0006] Document "Surface enhanced sum frequency generation of carbon monoxide adsorbed on platinum nanoparticles arrays" in J. of Chem. Phys. vol.1 1 3, N°13, p. 4532-4538 describes the enhancement of SFG signal obtained by using platinum nanoparticles on a substrate. No spatial information is determined by the disclosed method.

Summary of the Invention

[0007] The present invention is related to a method for spatially selecting an enhanced sum frequency generation signal comprising the steps of:

- provid ing an adsorption su bstrate com prising a flat su rface and metal l ic nanorods fixed onto the flat surface, said nanorods having an aspect ratio higher than 1 , preferably higher than 1 .2;

- adsorbing molecules of interest at the surface of the adsorption substrate; - illuminating said adsorption substrate with a first and a second coherent light source;

- measuring the sum frequency light generated from the illumination,

wherein the frequency of the first light source or the sum of the frequency of the first light source with the frequency of the second light source corresponds to the resonance frequency of either a transverse or a longitudinal surface plasmon mode, thereby selecting the signal arising from either the sidewalls of the nanorods or the end surfaces of the nanorods.

[0008] A particularly preferred embodiment of the invention is related to a method for spatially selecting an enhanced sum frequency generation signal comprising the steps of:

- providing an adsorption substrate comprising a flat surface and metallic nanorods fixed onto the flat surface, said nanorods having an aspect ratio higher than 1 , preferably higher than 1 ,2 and possessing a longitudinal or a transverse plasmon mode which can be excited at two distinct frequencies with s- or p- polarizations, wherein the material forming the nanorods is different from the material forming the flat surface;

- adsorbing molecules of interest at the surface of the adsorption substrate;

- illuminating said adsorption substrate with a first and a second coherent light source which overlap on said adsorption substrate;

- measuring the sum frequency light generated from the illumination, using the first light frequency and the sum frequency generation signal which does not correspond to said longitudinal or transverse plasmon mode of the nanorods to determine the signal arising from said adsorbed molecules on said flat surface, or

- using the first light source or the sum frequency generation signal s-polarized with a frequency matching the transverse plasmon mode of the nanorods to probe said adsorbed molecules on the nanorod lateral sidewall with enhanced intensity regarding said flat surface, or

- using the first light source or the sum frequency generation signal p-polarized with a frequency matching the longitudinal plasmon mode of the nanorods to probe said adsorbed molecules on the nanorod top surface with enhanced intensity regarding said flat surface.

[0009] According to particular preferred embodiment, the method of the invention further comprises one or a suitable combination of at least two of the following features: - the first light source has a wavelength comprised between 150 and 2500nm preferably comprised between 300 and 1 100nm;

- the signal arising from the molecules adsorbed on the substrate is determined by using a first light frequency and a sum frequency generation signal that do not correspond to said surface plasmon modes;

- the nanorods are chemically bonded to the flat surface;

- the nanorods have a height comprised between 15 and 1000nm;

- the nanorods are parallel to a common direction and the angle between the normal to the substrate and said direction is lower than 15° and preferably, the nanorods are perpendicular to the substrate; in this case, preferably, the first light source is s-polarised, respectively p-polarised for enhancing the signal arising from the sidewall, respectively from the end surfaces of the nanorods;

- the nanorods are parallel to a common direction and the angle between the normal to the substrate and said direction is comprised between 75 and 15°, preferably comprised between 40 and 50°, ideally about 45°.

- the nanorods are parallel or perpendicular to the incoming light or the reflected light of the first light source or of the SFG light;

- the wavelength of the second light source is comprised between 2,5 and Ι ΟΟμηι, preferably between 2,5 and 20μη-ι;

- the nanorods are randomly distributed on the flat surface;

- the flat surface is a conducting flat surface;

- the metallic nanorods comprises a metal selected from the group consisting of Gold, Platinum, Silver, Copper, Indium, Cadmium, Titanium, Nickel, I ron , Cobalt, Zinc, Tin their mixtures and their oxides;

- the material forming the nanorods is different from the material forming the substrate;

- the density of nanorods is higher than 5.10 ⁶ units/cm ² , preferably comprised between 5.10 ⁶ and 5 10 ¹⁰ units/cm ² .

[0010] The present invention is also related to a sum frequency generation spectrometer for implementing the method of the invention comprising:

- an adsorption substrate comprising a flat surface and metallic nanorods fixed onto the flat surface, said nanorods having a height comprised between 15 and 1000nm and an aspect ratio higher than 1 ;

- a first coherent light source;

- a second coherent light source; - a detector able to detect the su m freq uency light generated in use by the interaction of the first and second coherent light on the substrate.

[001 1 ] Advantageously, the first coherent light source has a frequency able to be tuned at least at two frequencies, so that the frequency of the first light source or the sum of the frequency of the first light source with the frequency of the second light source may be tuned to correspond to the resonance frequency of either a transverse or a longitudinal surface plasmon mode for selecting the sum frequency signal arising from either the sidewalls of the nanorods or the end surfaces of the nanorods.

[0012] Preferably, the sum frequency generation spectrometer of the invention further comprises polarisation means for polarising the first light source, s- polarisation of the first light source enhancing the signal arising from side wall of the nanorods, and p-polarisation of the first light source enhancing the signal arising from the ends of the nanorods.

Brief Description of the Drawings

[0013] Fig. 1 represents a Scanning Electron Microscopy (SEM) image of gold nanorods standing vertically on a gold surface. The rod distribution although not organized appears rather homogeneous at large scale. The rod aspect ratio in this case is 2 (140 nm long and 70 nm diameter).

[0014] Figure 2 represents Uv-Vis absorption spectra performed on gold nanorods vertically positioned on a gold substrate (top) and on a platinum substrate (bottom). The nanorods exhibit two localised surface plasmon resonances (LSPR): a transverse mode (TM, s-pola ri sed l ig ht) at 51 0 n m on both su bstrates , a n d a longitudinal mode (LM, p-polarised light) at 71 0 nm on the gold and 660n m on the platinum surface.

[0015] Figure 3 schematically represents an SFG process: a visible and an infrared laser beams are focussed on the surface and generate a third beam at their sum-frequency.

[0016] Figure 4 represents SFG spectra of dodecanethiol molecules chemisorbed on metall ic surfaces. Black dots are raw data , plain cu rve are data averages. In each graph, the top lines are the SFG spectra performed on the nanorods, and the bottom curves are the SFG spectra performed on flat metallic area. The spectra show ssp polarisation measurements of Au nanorods on platinum surface (a) and on gold surface (b), and ppp polarisation measurements of Au nanorods on platinum surface (c). [0017] Figure 5 represents a plot of the CH3 r- mode SFG intensity in ssp polarisation for gold nanorods on gold surface as a function of the visible wavelength.

[0018] Figure 6 represents a plot of the CH3 r- mode SFG intensity in ppp polarisation for gold nanorods on platinum surface as a function of the visible wavelength. As shown in the top curve, the SFG intensity reaches a maximum at the plasmon resonance frequency.

Detailed Description of the Invention

[0019] Sum frequency generation (SFG) spectroscopy is an optical technique that analyzes a material at an interface between two materials.

[0020] In sum frequency spectroscopy, two light beams, a first light beam and a second light beam, are directed to the interface. The first light beam and the second light beam are directed so that the two light beams overlap spatially and temporally with each other at the interface. One of the light beams has a frequency generally in the infrared region of the light spectrum. The other light beam has a frequency generally in the visible region of the light spectrum.

[0021] Because of the overlap, an interaction between the first light beam and the second light beam occurs at the interface between the substrate and the material to be analysed (adsorbed material). From this optically coherent interaction, a third light beam is generated at a frequency that is the sum of the frequencies of the first light beam and the second light beam. The angle that the third light beam makes with the interface is the angle required to conserve momentum parallel to the interface.

[0022] The third light beam is generated because of a nonlinear optical phenomenon known as "sum frequency generation" or "three wave mixing." The third light beam is referred to as the "sum frequency light beam" or "sum frequency light."

[0023] The first light beam and the second light beam are generally provided by lasers, as high intensity and preferably coherent light sources are needed. In order to maximise the intensity of the incoming beams while reducing thermal effect, synchronous pulsed laser are usually used.

[0024] The intensity of the third light beam is modulated by the interaction of the light with the material to be analysed. When the frequency of the incoming light corresponds to an electronic or vibrational excitation mode (resonant frequency), the intensity of the sum frequency beam is enhanced. This property is generally used to probe molecular vibrational modes corresponding to Infrared light wavelength.

[0025] In order to get the vibrational fingerprint of an adsorbed molecule, one of the light beam has a wavelength in the infrared (IR) region of the light spectrum, and this wavelength can be tuned over the IR spectrum corresponding to usual vibrational molecular modes (typically between 2,5 and 20 μηη, but may be extended up to Ι ΟΟμηη by using a free electron laser).

[0026] The sum frequency beam is characterized by an intensity and a wavelength (or wavenumber). As the frequency of the infrared laser light is varied, the intensity of the sum frequency beam varies according to the vibrational modes characteristic of the material to be analysed. Plotting the intensity of the third beam versus the wavelength of the infrared light beam (or the wavelength of the sum frequency light beam) provides a "vibrational sum frequency (VSF) spectrum."

[0027] The present invention takes advantage of the particular sum frequency signal enhancement properties of metallic nanorods standing (fixed) on a flat surface. In that context, "metal" or "metallic" should be understood according to electronic band structure (i.e. a solid having its Fermi level within the conduction band). Such nanorods exhibit very specific plasmon resonance modes which are spatially localised: a transverse mode excitable on the side of the rods and a longitudinal mode excitable on their ends. As will be shown in the example, it was discovered that when the visible light or the sum frequency light is corresponding to the resonant frequency of such plasmon, the sum frequency signal arising from molecular species adsorbed on the excited region is strongly enhanced.

[0028] Excitation of longitudinal mode of both ends of the nanorods are possible, but as the bottom end of the nanorods is normally not accessible to adsorbent molecules the top end is of particular interest for use in SFG spectrometry.

[0029] Furthermore, when using visible light and sum frequency light beams out of the resonant plasmon frequency, the signal arising from the flat surface is dominating the signal, said flat surface representing a much larger area than the side surface or the end surfaces of the rods.

[0030] When the nanorods stand perpendicular to the substrate, the selectivity of the method of the invention may be further improved by using polarised light beams, as p-polarised visible light further improves the intensity of p-polarized sum frequency light from the end surfaces of the rods or from the flat surface, and s- polarised visible light further improves the intensity of s-polarized sum frequency light from the side surface of the rods.

[0031] To properly define S- and P- polarizations, one first needs to define the plane of incidence. The plane of incidence of a light beam is the plane comprising the incident beam and the surface normal. In the present setup, this plane is perpendicular to the flat surface and comprises the incident and exiting beams. P- polarization refers to light that is polarized into the plane of incidence. S-polarization refers to light that is polarized perpendicularly to the plane of incidence.

[0032] Using nanorods not perpendicular to the substrate can also advantageously be used. More particularly, when the nanorods exhibit an angle of 45° with respect to the substrate normal, three favourable conditions can be obtained:

- when the nanorod axis is aligned with the visible and SFG reflected beams, the SFG signal becomes independent of the nature of the substrate and of the length of the nanorods for the longitudinal mode (polarisation p and 45° incidence);

- when the nanorod axis is aligned with the visible and (virtual) SFG incident beams, the SFG signal becomes independent of the nature of the substrate and of the length of the nanorods for the transversal mode (polarisation p and 45° incidence);

- when the nanorod axis is tilted in the vertical plane perpendicular to the incident plane, an additional increase of the longitudinal mode is observed.

[0033] Contrary to typical prior art setup, in which the wavelength of the visible light beam is fixed, in the present invention, the wavelength of the visible light is varied in order to select the different excitation modes, thereby selecting the localisation of the signal.

[0034] For example, the visible light beam may be produced by three different lasers, two operating at the wavelengths corresponding to the longitudinal and transverse plasmon resonance mode, and one operating out of resonance.

[0035] Alternatively, the visible light beam is generated by a tunable laser, able to be operated at the different wavelengths involved.

[0036] The wavelength/frequency of both transverse and longitudinal modes may be adjusted by selecting the nature of the metal constituting the rods. On the other hand, the longitudinal mode may also be adjusted by changing the aspect ratio (height/diameter) of the nanorods.

Examples

[0037] Several experiments have been carried out and are summarised hereafter.

[0038] In those examples, gold nanorods standing vertically on a gold or platinum metallic surface were synthesized as shown in Figure 1 . The rods were electrochemically grown through ion track etched membranes as described by V. Callegari et Al. in Chemistry of Materials, Vol. 21 , No. 18, 2009, p. 4241 . [0039] The rod synthesis parameters are adjusted to obtain a rod density in-between 5.10 ⁸ to 5.10 ⁹ units per cm ² , a rod length of 90 n m to 400 n m and a diameter between 60 nm to more than 100 nm. Note that for rod lengths longer than the SFG wavelength, optical diffusion becomes detectable. Such diffusivity shades off the coherent propagation of the nonlinear light emission.

[0040] As shown in Figure 2 for gold nanorods on gold, with a surface density of 3.10 ⁹ rods/cm ² and an AR of 1 .6, the TM mode is localised at 510 nm (s polarisation of the light) while the LM is a broad band centred around 710 nm (p polarisation). For gold nanorods on platinum, with a density of 9.10 ⁸ rods/cm ² and an AR of 1 .25, the TM mode has been measured at 510 nm and the LM at 660 nm.

[0041] The weaker intensities of both TM and LM modes when nanorods stand on the platinum surface by comparison to gold surface can reasonably be attributed to the lower rod density on the platinum sample than on gold as well as to additional effects due to the differing optical properties of the two substrates.

[0042] The lower frequency of the LM mode when nanorods stand on platinum (660 nm) instead of gold (710 nm) is ascribed to the lower aspect ratio of the nanorods on the platinum surface.

[0043] More specifically (see Figure 3), the TM mode can be excited by both the visible and the SFG photons, while the LM mode can only be excited by the visible photons due to the limited tunability of the laser used in the present experiment.

[0044] To perform SFG measurements, the experimental SFG spectrometer was designed as follows. A mode-locked picosecond Nd:YAG oscillator clocked at 100 MHz and delivering one 1064nm laser train containing one hundred pulses of 15ps is used as pump at a 25Hz repetition rate. This fundamental pump beam is then sent in two optical parametric oscillators (OPO), one in the visible (25mW

- 420 to 700nm), and one in the infrared (25mw - 2,8 to 10 μηη). The two beams overlap together on the sample (spot size ~ 1 mm) and the generated SFG is collected by a double grating monochromator. Further description of the SFG spectrometer can be found in "Surface Science" 2002, vol. 502, p.261 -267 which is incorporated herein by reference.

[0045] In order to measure a vibrational SFG signal from the nanostructures as well as from the flat surface, a monolayer of dodecanethiol (DDT) has been chemisorbed onto the surfaces. The DDT signal intensity is used to give information about the local electric field enhancement at the nanorod surroundings when excited at the plasmon resonance. [0046] To be able to quantify such amplification factor of the SFG signal emitted from the rod area by comparison to flat regular surface, SFG measurements have been performed on both types of surfaces. Practically, the two measurements have been carried out on the same sample, the rod zone covering only a part of the flat metallic surface.

[0047] To excite the two plasmon modes, s- and p-polarisations have been used, as done for the UV-Vis spectra, with the difference th at i n a S FG measurement this may actually be done with both the visible and the SFG beams. Note that because of the strong screening of the metallic substrate in the I R frequency region, this latter has been kept p-polarised.

[0048] Therefore, to excite selectively the TM and LM modes, we have used ppp and ssp polarisation configurations, respectively (with the conventional order SFG, vis, IR). Figure 4 shows the SFG spectra recorded in the infrared spectral range between 2800 and 3000cm ^"1 . There is no scale normalisation in order to be able to compare the different spectrum intensities. Black dots are raw data and plain curves are data averages (5 nearest neighbours). The spectra show three main vibrational fingerprints, the first at 2883 cm ^"1 corresponding to the methyl symmetric stretching mode (CH3 r+), one at 2975 cm ^"1 matching the degenerate methyl asymmetric stretching mode (CH3 r-), and one around 2945 cm ^"1 being the methyl Fermi resonance (CH3 FR). Two weaker modes are also observed, accounting for the methylene symmetric (CH2 r+ at 2858 cm ^"1 ) and asymmetric (CH2 r- at 2920 cm ^"1 ) stretching vibrations.

[0049] The experiments show that metallic nanorods support well organized dodecanethiol self-assembled monolayers (SAM ) . I n d eed , S F G spectroscopy is sensitive to the layer conformation and order through centrosymmetry.

It is able to detect gauche defects into the methylene chain structure from their vibrational activity. In the different spectra shown in Figure 4, the CH2 activity is in general weak. This testifies that the amount of gauche defects in the DDT SAM is small, whether the DDT is adsorbed on the rods or on the flat surface.

[0050] Let us first discuss the LSPR of the TM. The spectra in the two top panels of Figure 4 depict the SFG intensity in ssp polarisation on platinum surface (a) and gold surface (b). In each panel, the lower curve is measurement from flat metallic area while the higher one is originating from the gold nanorods zone. The right drawings depict the experimental conditions for each curve. The SFG intensity increase by comparison to flat surface is close to 150 times for gold rods on platinum (maximal CH3 r- mode intensity of 3.8 vs 0.025) and reaches nearly 200 times for gold rods on gold substrate (maximal CH3 r- mode intensity of 18.6 vs 0.100).

[0051] Captions on the right picture the measurements and the wavelengths used in the different cases.

[0052] Those amplification factors are obtained when the frequency of the visible or the SFG matches the transverse LSPR. Indeed, as shown in Figure 5, on the gold substrate, the SFG intensity of the CH3 r- mode described a maximum when the wavelength of the incident visible laser is fixed around 515 nm and at 605 nm. At the first wavelength, the TM mode is excited directly by the visible beam while at the second value, the excitation is ensured through the SFG photons themselves, close to 515 nm in wavelength. Those measurements are in good agreement with the Uv-Vis absorption spectroscopy. However, it is interesting to note that the coupling efficiency between the plasmonic field and the nonlinear SFG emission appears to be higher when the LSPR is directly excited by the SFG electric field (Figure 5, peak b) instead of the visible one (Figure 5, peak a).

[0053] The transverse LSPR can be excited selectively by both the visible

(visible wavelength at 515nm [a]; bottom left) and SFG (visible wavelength at 605nm

[b] ; bottom right) photons. However, as shown in the top curve, the SFG intensity is greater when the plasmon mode is directly excited by the SFG beam itself [b]. Black dots are raw data.

[0054] When considering the shape of the spectra of Figure 4, both ssp measurements on platinum (a) and gold (b) surface lead to vibrational modes appearing clearly as peaks. This is the usual case on platinum flat surfaces (lower spetra of Figure 4a and 4c). However, it is known that on flat gold, for both ssp and ppp polarisations, the vibrational SFG modes of alcanethiols appear as troughs when the

Vis or SFG beam wavelengths are close to the gold electronic transition, as it is the case here (lower spectra of Figure 4b).

[0055] Besides, in the Vis and SFG wavelength window considered here, the vibrational modes appear always as peaks on platinum.

[0056] The lineshape of SFG bands results from interferences between the substrate and molecular contributions to the total SFG signal. The resulting peaklike or trough-like shape depends on phase and amplitude differences between the electronic oscillations in the metallic substrate and the molecular vibrational modes.

[0057] Consider now the LSPR excitation of the LM. The two bottom spectra of the Figure 4 depict the SFG intensity in ppp polarisation on platinum surface

(c) . The excitation of the longitudinal LSPR is clearly observed when gold nanorods stand on the platinum substrate, as shown on the c spectrum of Figure 4. In this case, the amplification factor is about 1 1 for the CH3 r- mode SFG intensity.

[0058] Figure 6 shows the trend of the SFG intensity (maximum of CH3 r- mode intensity) as a function of the visible wavelength in ppp polarisation configuration for the gold nanorods on platinum. Due to experimental limitation of the visible laser tunability, the longitudinal mode has been excited here only by the incident visible beam, and not by the SFG beam. This leads to only one intensity peak in Figure 6, located around 650 nm, which is in close agreement with what was obtained by UV-Vis spectroscopy.

[0059] The Table 1 summarises the amplification factors and the sample properties. The SFG intensity recorded from the DDT molecules when probing the flat substrates or the nanorods zones are mentioned for both platinum and gold substrates, in ppp and ssp polarisations. The different amplification factors (AF) are determined by calculating the ratio of the SFG intensities of the corresponding flat and nanostructured surfaces. We also evaluate normalized amplification factors (NAF) that take into account the quantity of DDT molecules that generates the increased SFG intensity. For this, we assume that the SFG signal of nanostructured surfaces probed at the plasmon resonances comes from the rods top surface in the case of the LM mode and from the rods lateral surfaces in the case of the TM mode. To evaluate the NAF, we calculate thus the surface coverage (SC) of the SFG-active molecules on the nanorods walls and top surface, relatively to the molecule surface coverage on the flat substrate (SC=1 ). Finally, we take into account the quadratic dependence of the SFG intensity on the amount of probed molecules, so that NAF=AF/SC ^A 2. The NAF give the increase of the SFG signal that would be observed if the nanostructured and flat surfaces had equal SFG-active areas.

I SFG I SFG

AF SC NAF NR Flat

Pt - ppp 8.800 0.800 1 1 0.045 5432

Pt - ssp 3.800 0.025 152 0.226 2976

Au - ssp 18.600 0.100 186 0.659 428

Rod density Rod length Rod diameter AR

Pt

9.10 ⁸ units/cm ² 100 nm 80 nm 1 ,25 substrate

Au

3.10 ⁹ units/cm ² 106 nm 66 nm 1 ,6 substrate Table 1. Summary of the SFG intensities measured on Au and Pt substrates, with and without gold nanorods on them. AF, SC and NAF designate the amplification factor, the surface coverage and normalized amplification factor by the square of the surface coverage, respectively.

[0060] As mentioned earlier, the amplification in ppp polarisation due to the LM mode is about 1 1 , which corresponds to a NAF of 5432. I n ssp, the amplification is 152 and 186 on platinum and gold respectively when excited by the TM, which gives 2986 and 428 as normalized amplification factors.

[0061] In addition to the increase of the sensitivity of surface SFG by several orders of magnitude through LSPR amplification , the novelty of this invention lies also in the ability to excite the LM or the TM mode in a controlled way which enables to probe selectively the lateral or the end surfaces of the nanorods.

[0062] Furthermore, by using working on- or off-resonance through an adequate choice of the visible wavelength and polarization, we can probe selectively the molecules adsorbed on the rods or on the substrate surface between the rods.

[0063] Indeed, when working in ppp polarisation on the platinum sample and far from the LM mode (e.g. at 500 nm), the ppp SFG signal is clearly dominated by the molecules adsorbed on the flat surface between the rods (~ 95% of the probed surface).

[0064] Next, by switching the visible light to the LM mode, the majority of the signal comes from the top end of the rods ([8.8-0.8]/8.8 = 91 %).

[0065] With ssp polarisation, on Pt and Au, the SFG intensity remains high between the two visible wavelengths of excitation (515 and 605nm), while the ssp signal from the flat surface is very weak. We may therefore claim that the ssp signal is nearly exclusively originating from the molecules adsorbed on the rod bodies, with insensitivity to molecules adsorbed on the flat substrate between the rods.

[0066] As a conclusion, the present invention discloses a metallic nanostructured substrate combined with SFG spectroscopy to obtain an enhanced nonlinear spectroscopy with high sensitivity and spatial selectivity. The excitation of the two plasmonic modes of the nanorods with both visible and SFG beams allows enhancing significantly the sensitivity of sum-frequency generation. The measured SFG signal is about 200 times more intense in ssp polarisation, and 10 times in ppp polarisation by comparison to a flat surface.

[0067] The SFG spectra of alkanethiols adsorbed on the nanorods indicate that the molecules are self-assembled in an organized configuration on the nanostructures. [0068] Furthermore, the lineshape of the molecule vibrational resonances on gold nanorods is simpler than on flat gold substrate, which should facilitate the SFG spectra analysis on gold.

[0069] Finally, we gain a spatial selectivity. By choosing appropriate wavelengths and polarisations to work on- or off resonance with the LM and TM localised surface plasmons of the nanorods, we can probe preferentially the rods' top, the rods' sidewalls, or the flat substrate area in-between them.

[0070] The use of such nanostructured substrates for SFG enhancement may become a great solution to overcome one major drawback of the technique, which is the usual weakness of the signal intensity. This will pave the way to the development of new SFG-based surface sensor and imaging devices.

[0071] Enhanced SFG imaging on such nanostructured substrates should give remarkable results for many applications.

[0072] Moreover, several parameters can be used to tune the spatial selectivity according to the application needs, by exploiting the conditions of the excitation and the emission rules of the plasmon-enhanced SFG process occurring at the rods on the substrate. In particular, in addition to the beams frequencies and polarizations, we can adjust the angles of incidence, the optical properties of the flat surface (by choosing the adequate substrate), and the nanorods orientation on the surface (characterized by the tilt angle theta with respect to the surface normal, and the azimuthal rotation phi with respect to the plane of incidence). Taking into account how the total electric field of the visible and SFG beams vary above the flat surface along the rod axes, these parameters can be selected adequately to tune the preferential excitation of localised plasmons in order to control finely the resu lti ng signal enhancement and spatial selectivity.