1. Introduction

Nowadays, the Surface Plasmon Resonance (SPR) phenomenon [1] is well known and has been used since the 1980s for sensor applications such as the detection of chemical [2] or biological [3] species. Surface Plasmon Polariton (SPP) phenomenon has also been investigated to realize very small photodetectors (quantum dots) [4] as well as to transport information on short distance in computer chips [5]. These phenomena are also the object of research in the biodetection field [6,7].

Unlike the transmission signal by optical fibers widely used on long distance communications, plasmonic waves have the advantage of being the most promising candidates for subwavelength optical confinement considerintg both evanescent and propagating waves, allowing nanometric dimension achievements. From this point of view, plasmonics waves can be used to perform compact interconnect and offer very high speed [8].

Silver and gold are the traditional metals of choice for plasmonic material; unfortunately they suffer from many problems in addition to be expensive. Silver is not stable in the air, it oxidizes quickly in presence of water and/or oxygen; as regards, gold is very difficult to micro-structure because of its high stability. Moreover, silver and gold suffer from high resistive losses [9] that restrict the propagation of the plasmonic waves in the considered wavelength ranges.

In order to bypass these limits, in the past few decades, new metamaterials (MM) have been the subject of a lot of researches for plasmonic materials [9,10]. Among them, transition-metal nitrides, in particular titanium nitride (TiN), offer interesting properties for plasmonic applications in the Near-InfraRed (NIR) range [11,12].

Usually the TiN layer is deposited by means of expensive technologies such as PVD under nitrogen/argon environment. Its hardness and its chemical stability make it attractive in many domains [13]. Unfortunately, these same properties make it difficult to micro-structure in the shape of periodic microstructures such as gratings. So, all the technological steps from the lithography to the lift-off and/or physical or chemical etching process are necessary to micro-structure this MM [14].

Another way to elaborate TiN layer is possible, using the ammonolysis of TiO₂ [15–17] that leads to non-stoichiometric TiN_x or TiO_xN_y. Although this technique is often used in a way to implement a low level of nitrogen in the TiO₂ layer so as to increase photocatalytic activity [18,19], ammonolysis can potentially bring a high level of nitridation. So now, this technique deserves a new interest because TiO₂ can support a high degree of nitridation and also can be deposited by sol-gel method offering new interests and new fields of applications.

In this paper the authors present for the first time an original approach and an experimental demonstration of a TiO_xN_y plasmonic grating using ammonolysis of titanium oxide, from a sol-gel layer, which acts as a negative photoresist [20]. The idea is to first print the grating in the photo-patternable TiO₂ sol-gel layer and then transform the TiO₂ grating in TiO_xN_y based grating using the process of ammonolysis. The demonstration of structuration of the TiO₂ sol-gel photoresist has been previously reported by the authors using interferential lithography or phase mask process [21] and colloidal-photolithography [22].

The study presented here goes much further showing in the first part, the simple experimental process leading to the nitridation of the TiO₂ sol-gel layer. The second part is dedicated to the measurements of the structural, electrical and optical characteristics of the fabricated TiO_xN_y layer in order to prove its ability to be used as a plasmonic material, demonstrating the existence of surface plasmon polariton at the air-metal interface. Then the whole structuration process is explained followed by the experimental demonstration of a plasmonic effect with this structured TiO_xN_y material. These last results are confirmed by using an electromagnetic model based on coupled waves analysis.

2. Experimental process

2.1 Sol-gel layer preparation

Specific sol-gel formulation and procedure were used to deposit the TiO₂-based photoresist. A sol was prepared from titanium isopropoxyde orthotitanate (TIPT) complexed by benzoyl acetone (BzAc) in methanol and butanol. The whole procedure is fully explained in previous papers [20–22]. Manipulation and functioning are briefly summarized hereafter. The sol was deposited on a silicon substrate by spin-coating at the speed of 7000 rpm. The obtained sol-gel layer was first dried at room temperature and then heated at 110°C during 90 min, leading to a so-called xerogel film, i.e. an inorganic polymer film constituted of Ti-O-Ti chains with organic chain-end groups arising from the sol formulation, mainly TIPT-BzAc complexed species. This xerogel film is soluble in different solvents (alcohol, chloroform, acetone, etc.) as far as BzAc stays complexed with TIPT.

The main interest of this protocol relies on the properties of BzAc, which makes the film soluble in a solvent while being sensitive to UVA light. Indeed under UVA illumination, the TIPT-BzAc complex is partially degraded in insoluble species such as carbonates and/or carboxylates. Therefore, it will create a contrast of solubility between illuminated and non-illuminated areas when it is selectively exposed to UVA light.

2.2 Nitridation of TiO₂ sol-gel

Firstly, the nitridation of TiO₂ was made on full xerogel film with no micro-structuration in order to determine the properties of the material for plasmonic effect. The xerogel films had a thickness of 110 – 120 nm before a heat-treatment at 300°C, 400°C or 500°C during 15 min, which induced a shrinkage of 55%, 60% and 65% leading to a film of 50 nm, 45 nm and 40 nm thickness, respectively [20–22]. Then, they were heated under Ar gas (100 sccm) in a SiO₂ glass tube furnace up to 800°C, 900°C or 1000°C. Then, the temperature was kept constant and NH₃ flow (100 sccm) was applied during 30 min to achieve nitridation. According to Kamiha et al [15,16] the nitridation is performed via the following reactions of NH₃ with the reduced titanium dioxide:

After nitridation the tube was cooled down at room temperature under Ar gas (100 sccm). The ammonolysis of the TiO₂ thin layer did not induce a supplementary shrinkage and all films had a constant thickness of around 40 nm whatever the temperature of ammonolysis. This ammonolysis is expected to proceed through nitrogen in-diffusion and oxygen out-diffusion mechanisms, which can in turn be limited by the film thickness, thus leading to a limited nitridation degree and to an eventual O/N composition gradient through the thickness. However, the thickness of samples studied here is very small and we infer that such gradient should be minor. Further investigations, for instance using XPS measurements, should be necessary to go further on this feature.

After characterization of all studied samples, the parameters leading to optimal nitridation of the TiO₂ layer will be kept constant for the grating realization.

3. TiO_xN_y layer characterization

The TiO_xN_y layers were characterized using three different tools. First, the X-ray diffraction (XRD) is necessary to know the atomic, molecular and crystallographic structure of the thin film. This will give us an estimate of the degree of nitridation. Secondly, four-point probe method allows determining the electrical resistance of the layers by a simple measurement, and the comparison between the electrical resistances of TiO_xN_y films elaborated in different conditions will provide information about the conduction performances. Thirdly, the spectroscopic ellipsometry is necessary to give insight in the optical properties of the TiO_xN_y films. This last characterization is very important here because it brings information about the real and imaginary parts of the permittivity and the ability of the TiN thin layer to propagate plasmonic waves.

3.1 Structural properties

XRD patterns of TiO_xN_y films elaborated in various conditions are illustrated in the Fig. 1. These patterns show peaks corresponding to the face-centered cubic (fcc) structure of TiN with no particular preferential orientation. The intensity and full width at half maximum (FWHM) of the peaks increase slightly when the ammonolysis temperature increases, in particular between 800°C and 900°C. These features depict an increase of the crystallization degree and crystallite size with temperature of ammonolysis. Conversely, there was no significant influence of pre-treatment temperature (not represented here). Using Debye-Scherrer law for the most intense peaks, i.e. (111), (200) and (220), leads to an estimate of the crystallite size that increases from 20 nm to 30 nm when the ammonolysis temperature increases from 800°C to 1000°C.

 

Fig. 1 X-ray diffraction patterns for pre-treatment at 300°C and ammonolysis at a = 800°C, b = 900°C and c = 1000°C: a) entire patterns and b) zoom around the most intense (200) peak

Download Full Size | PDF

XRD patterns indicate the presence of a single fcc crystallographic phase that may correspond to the TiN phase. However, from this analysis one cannot conclude that a total nitridation happens [23–25]. Indeed, this crystallographic structure may also describe a partially oxidized composition, which could be the result of a TiO_xN_y solid solution or a mixture of two phases: TiO_x (x<<2) and TiN_y (y<1). However The lattice cell determined from the three most intense peaks was observed to increase from 0.4190 nm after ammonolysis at 800°C to 0.4210 nm after ammonolysis at 1000°C. Compared to the lattice cell of a fcc stoichiometric TiN structure (a = 0.4240 nm) [26], a partial oxidation is expected to reduce the lattice cell. This oxidation must be then reflected by a XRD peak shift toward greater 2θ values, which is actually observed in Fig. 1(b) for ammonolysis performed at the lowest temperature studied here (800°C). Let us note that XRD cannot provide a direct insight of the nitridation degree since the peak position can also depict stress features. However, this tool is sufficient to compare samples elaborated in different conditions and to conclude to a similarly good degree of nitridation after ammonolysis at 900°C or 1000°C, which will be confirmed by electrical and optical analysis.

3.2 Electrical properties

The square resistance (Rs) values measured using the four-point probe method are illustrated in Fig. 2 for the different experimental conditions.

 

Fig. 2 Resistance measurements according to the pre-treatment temperature for ammonolysis at 800°C, 900°C and 1000°C

Download Full Size | PDF

This figure shows that Rs is influenced both by the pre-treatment and subsequent ammonolysis. For ammonolysis at 800°C, the resistance is very high when the pre-treatment is carried out at 110°C. Resistance drops by a factor 5 for pre-treatment at 300°C, and then rises again continuously for pre-treatment at higher temperatures. The values measured after a pre-treatment at 110°C and 500°C are nearly the same. Similar trends are observed after ammonolysis at 900°C, but this temperature leads to much lower values than those measured after ammonolysis at 800°C. Ammonolysis at 1000°C leads to a Rs reduction only in the case of a pre-treatment at 110°C. For other pre-treatments, Rs values measured after ammonolysis at 900°C and 1000°C are very similar and exhibit a continuous increase when the pre-treatment temperature increases from 300°C to 500°C. Overall, the best conduction performances are obtained for a pre-treatment at 300°C. For this temperature, the resistivity (ρ) and conductivity (σ) can be deduced from the film thickness (d) and Rs values using Eq. (4) and Eq. (5):

Table 1. Resistivity and conductivity of the TiOxNy films after a pre-treatment at 300°C and ammonolysis at 800°C, 900°C and 1000°C

View Table

Values illustrated in this table depict good conduction performances after ammonolysis at 900°C or 1000°C. Indeed, the resistivity (conductivity) is about one order of magnitude larger (weaker) compared to pure TiN, but these values are very similar to those of ammonolyzed layers elaborated by other authors [24]. Beside, compared to pure metals such as aluminium, gold or copper, the resistivity of ammonolyzed layers is two orders of magnitude larger. The discrepancy between pure TiN and ammonolyzed TiO_xN_y layers can probably be attributed to a partial oxidation of the latter. Nevertheless, TiO_xN_y layers reveal an extremely low resistivity, by more than six orders of magnitude, compared to a TiO₂ thin layer, which exhibits a resistivity around several kΩ.cm. This difference proves the high degree of nitridation reached during the TiO₂ ammonolysis. Moreover, in agreement with the conclusions drawn from XRD patterns (Fig. 1), the best conduction performances measured after pre-treatment at 300°C and ammonolysis at 900°C or 1000°C are likely the result of a high degree of nitridation.

3.3 Optical properties

Spectroscopic ellipsometry provides a third characterization of the TiO_xN_y layer. It enables measurements of the real (n’) and imaginary (n”) parts of the refractive index and to deduce corresponding ε' and ε” values of permittivity. These latter parameters are very important to conclude to the ability of the layer to be a good material for propagation of plamonic waves. SPR relies on an evanescent wave stimulated by an electromagnetic field and characterized by the resonant oscillation of conduction electrons at the interface between two materials of negative and positive permittivity, such as a metallic layer and a dielectric, respectively. Today, the conditions of existence of a surface plasmon are well known and among them the most restrictive is:

Equation (6) implies that the real part of the TiO_xN_y permittivity has to be negative and its absolute value should be greater than the permittivity of the surrounding dielectric (in our case air, ε' = 1).

The n’ and n” values measured on TiO_xN_y layers elaborated in different conditions are illustrated in Fig. 3 as a function of the wavelength. For a pre-treatment at 300°C, this figure shows that, whatever the ammonolysis temperature (900°C or 1000°C) is, the respective n’ and n” values in the visible and infrared regions are very close. This feature is in good agreement with the conduction properties observed previously for both ammonolysis temperatures. Moreover, compared to a pre-treatment at 300°C, a sample pre-treated at 500°C with ammonolysis at 900°C exhibits a similar real part of the refractive index but a much lower imaginary part.

 

Fig. 3 Real part (n’ in blue) and imaginary part (n” in red) of the refractive index according to the wavelength (Vis-NIR), for pre-treatment at 300°C and 500°C and ammonolysis at 900°C and 1000°C

Download Full Size | PDF

From the Eq. (7) we can now deduce the numerical values of the real part ε’ and imaginary part ε” of the relative permittivity ε_r:

Derived values of permittivity are plotted in Fig. 4, giving us direct idea of the metallic behavior of the TiO_xN_y layer and its ability to exhibit SPR.

 

Fig. 4 Permittivity according to the wavelength for pre-treatment at 300°C and 500°C and for ammonolysis at 900°C and 1000°C: a) real part (ε’) and b) imaginary part (ε”)

Download Full Size | PDF

Figure 4(a) shows that for a pre-treatment at 500°C and ammonolysis at 900°C, the criterion of Eq. (6) is fulfilled in a very small wavelength range around 1200 nm and the ε’ maximal absolute value (1.70) is weakly superior to that of air. In contrast, for a pre-treatment at 300°C and ammonolysis at 900°C or 1000°C, the criterion of Eq. (6) is fulfilled in a very large NIR range, from 800 nm to more than 2000 nm, which depicts the metallic behavior of the TiO_xN_y layer. Ammonolysis at 1000°C leads to a metallic behavior extending in the largest spectral range, but ammonolysis at 900°C leads to the lowest ε’ value (−5.70 at 1450 nm). Figure 4 also shows that, while the ammonolysis temperature has a strong influence on the ε’ value, it does not significantly influence ε”, which is observed to continuously increase with wavelength in the NIR range.

Thus, thanks to the systematic characterization of ammonolyzed TiO₂ xerogel layers, the structural, electrical and optical properties of derived TiO_xN_y layers are known and we have identified a new process giving access to a plasmonic material. Of course, permittivity measured on the TiO_xN_y layers are far from that of traditionally used gold or silver metamaterials, which exhibit an ε’ value below −80 at 1500 nm, and also far from that of stoichiometric TiN, which exhibits an ε’ value close to −20 for the same wavelength [11]. However, it has been inferred that the best permittivity obtained in this work, i.e. for a pre-treatment at 300°C and subsequent ammonolysis at 900°C, is sufficient to create a plasmonic wave. Thus, this process has been extrapolated to develop a TiO_xN_y grating using a two-step approach which consists i/ in the development of a TiO₂ grating from a photopatternable TiO₂-based sol-gel layer, and ii/ in the subsequent nitridation of the TiO₂ grating in previously optimized ammonolysis conditions.

4. Fabrication of the TiO_xN_y grating

4.1 Illumination process

As previously mentioned, the sol-gel film deposited by spin-coating on a silicon substrate forms a xerogel layer after baking at 110°C during 90 min. The pattern to be printed on the samples is obtained by illumination through a chromium photomask of 1 µm period during 15 min under a 365 nm wavelength collimated light with an irradiance of 26 mW.cm-2. Then a short heating step of 8 min at 110°C is performed in order to increase solubility contrast between insolated and non-insolated areas. The photo-chemical or thermo-chemical effects of this three step process have been described in [20]. The development of samples was optimized by two successive washing operations in ethanol and in deionized water during 35 s and 1 min, respectively. Ethanol dissolves easily the non-insolated xerogel layer and water stabilizes it, leading to a grating whose period is the same as that of the photomask, i.e. a TiO₂ xerogel grating of 1µm period.

4.2 Ammonolysis of the TiO₂ grating

According to previous studies performed on TiO₂ layers, the best conditions of ammonolysis offering appropriate optical properties for plasmonic material were adopted to the nitridation of the TiO₂ grating, i.e. a pre-treatment at 300°C and subsequent ammonolysis at 900°C.

Figure 5 shows the grating profile measured by Atomic Force Microscopy (AFM) before and after pre-treatment and ammonolysis. As illustrated in Fig. 5(a), the grating depth measured after development of the xerogel film is 200-250 nm. This value has to be compared to the xerogel film thickness of 350 nm. It means that there is a continuous under-layer of 100-150 nm thickness below the patterned grating. After ammonolysis, Fig. 5(b) shows that the grating depth falls to 80 nm, thus revealing shrinkage of 60-70% occurring during pre-treatment and ammonolysis. The continuous under-layer of 100-150 nm thickness is subjected to the same shrinkage leading to a final thickness of about 40 nm after ammonolysis. The AFM profiles of Fig. 5 also show that the grating shape evolves with ammonolysis. Before ammonolysis, the grating presents lines and grooves close to 500 nm each, i.e. a line-space ratio close to 1. Moreover, lines reveal a flat surface, exhibiting almost a square shape, while the grooves reveal curved bottoms. After ammonolysis, this structure is no longer observed. The grating presents lines and grooves close to 200 and 800 nm, respectively, which leads to peak-shaped lines. This grating structure is closer to a sinusoidal shape rather than a square shape. Finally, the AFM profiles illustrated in Fig. 5 depict a certain roughness of the TiO_xN_y grating that is not observed before ammonolysis. In any case, and whatever eventual structural, geometrical and dimensional defects, the TiO_xN_y grating yields strong diffraction effects clearly depicted by the green color illustrated in the micrograph of Fig. 6(a).

 

Fig. 5 AFM profile before a) after pre-treatment and ammonolysis and b) performed on a TiO2 xerogel grating.

Download Full Size | PDF

 

Fig. 6 a) Picture of a TiO_xN_y grating on silicon and b) SEM image of the TiO_xN_y grating in cross section view.

Download Full Size | PDF

AFM features are confirmed with the cross section Scanning Electron Microscopy (SEM) image of Fig. 6(b), which has been obtained after cleaving the microstructured substrate. This image depicts a grating that can be described as a thin film whose thickness exhibits quasi-sinusoidal modulation of 80 nm amplitude and 1 µm period with a maximal thickness of about 120 nm. The SEM image also confirms the grating roughness, which is attributed to structural changes occurring over ammonolysis [23]. As previously illustrated in the case of a TiO₂ layer, ammonolysis at 900°C leads to the crystallization of TiO_xN_y grains of around 30 nm in diameter. Even if so small grains cannot precisely be observed in the SEM image of Fig. 6(b) owing to resolution limitations, this image clearly depicts the granular morphology of the ammonolyzed grating. Thermo-mechanical stress occurring at the film-substrate interface during the pre-heating and post-cooling ammonolysis steps may also cause the formation of localized micro-cracks. On the one hand, grain boundary and localized micro-crack features are likely to induce the roughness depicted by AFM profiles after ammonolysis. On the other hand, these features can also reduce the conduction and optical performances of the fabricated grating, leading to losses for plasmonic applications. In other words, a same TiO_xN_y layer free of grain boundary and micro-crack features should exhibit better conduction and optical performances closer to that of stoichiometric TiN.

5. Plasmon resonance demonstration

5.1 Condition to couple plasmon mode by a grating

The diffraction grating is necessary to couple the incident wave into the plasmon mode. The grating diffracts the light according to different m orders with different directions β_m relative to the perpendicular to the grating [29]. If λ is the wavelength of the incident wave, Λ is the period of the grating and α the incident angle relative to the perpendicular to the grating, the direction β_m of each diffracted m orders is given by Eq. (8):

To excite the SPR, it is necessary to couple one of the m orders diffracted by the grating into the plasmon mode, according to β_m = −90°. As shown in Fig. 7, for a given wavelength λ, a particular angle of incidence α_c enables the diffracted order to match the natural frequency of electron oscillation at the dielectric-metal interface. With the condition β_m = −90°, and starting from Eq. (8), the α_c angle is given by Eq. (9):

 

Fig. 7 Coupling of the −1st order diffractive order into the plasmon mode.

Download Full Size | PDF

From this last equation and for a grating period of Λ = 1µm corresponding to the studied structure, we deduce that the wavelength λ must range between 1000 nm and 2000 nm, which is actually the spectral range where the real part of the permittivity ε’ of the TiO_xN_y film is optimal for a plasmon effect (Fig. 4).

Due to the continuity of the electromagnetic field at the grating interface, the resonant oscillation is excited only if the electric field of the incident wave is perpendicular to the lines of the grating, i.e. in TM polarization. Then, the plasmon in TM polarization is depicted by a minimum of power in the 0th reflected order for a particular wavelength and a specific incident angle, according to the excitation and phase matching between the incident wave and the plasmon mode. The phase matching is so performed with the −1st order of the grating and the energy injected into the plasmon mode is taken from the 0th reflected order, thus conducting to a deep in the 0th reflected order

5.2 Plasmon resonance modeling

In order to predict the SPR effect on the 0th reflected order, simulations are realized. Modeling was made from the measured optical properties of the TiO_xN_y layer and using Rigorous Continuous Wave Analysis (RCWA) method with MC grating software [30]. The modeled structure (Fig. 8(a)) is described as a thin film whose thickness exhibits sinusoidal modulation of 80 nm amplitude and 1 µm period with a maximal thickness of 120 nm, which matches the actual TiO_xN_y grating structure as illustrated in the SEM image of Fig. 6.

 

Fig. 8 a) Sinusoidal structure of the grating modeled with MC grating and b) power in false colors (from blue to red) of the TM 0th reflected order versus wavelength and angle of incidence.

Download Full Size | PDF

Figure 8(b) shows in a qualitative way the plasmon mode coupled by the diffraction grating. The power of the 0th reflected order in TM polarization is plotted in false colors as a function of the wavelength on the abscissa and of the angle of incidence on the ordinate. For each wavelength, a dark blue area clearly depicts a minimum power at a particular angle of incidence according to the phase matching between incident TM wave and plasmonic mode propagation. So a suitable couple of wavelength - incident angle values, fulfilling the condition of Eq. (9) and leading to excitation of the plasmon resonance, can be predicted provided that the structure (substrate, layers, cover, period and depth of the grating) is well known. For instance, for a 25° angle of incidence, the plasmon resonance occurs at a 1440 nm wavelength for the structure proposed in our model. The amplitude of the resonance is linked to the grating depth and to the material used for the plasmonic effect.

5.3 Plasmon resonance measurement

In order to prove the ability of the fabricated TiO_xN_y grating to be used as a plasmonic component, measurements were performed with the experimental set-up shown in Fig. 9. A collimated and linear polarized (TE or TM) polychromatic wave is incident onto the grating. This source is a halogen lamp, which emits on the desired bandwidth in the very near IR. A spectrometer measures the wavelength spectrum of the 0th reflected order. The spectrometer is adapted to the NIR range and is able to detect the signal in the 900-2000 nm bandwidth. The angle of incidence α is fixed at three values: 25°, 30° or 35°.

 

Fig. 9 Optical set-up for the measurement of the 0th reflected order of the TiO_xN_y grating.

Download Full Size | PDF

Figure 10 shows the 0th reflected order of the TiO_xN_y grating in the 1300 - 1700 nm bandwidth and for the three chosen angles of incidence. This figure reveals a nice correlation between the modeling (Fig. 10(a)) and the experimental values (Fig. 10(b)). In both cases, SPR depicted by a minimum of the 0th reflected signal in TM polarization appears at 1440 nm, 1515 nm and 1590 nm for incident angles of 25°, 30° and 35°, respectively. The TE mode exhibits a constant reflection close to 60% with no particular change whatever the incidence angle is. Outside the SPR areas (reflection minima of the TM signal), the difference in the mean level of reflection between TM and TE signals is around 15% both for modeling and experimental values.

 

Fig. 10 0th reflected order of the TiOxNy grating for incidence angles of 25°, 30° and 35° and for TE or TM incidence: a) modeling curves and b) experimental measurement.

Download Full Size | PDF

Astonishingly, the experimental curves is very close to the theoretical one, experimental SPR reaches a deep variationclose to 6-7% for the three corresponding angles while modeling depicts a variation close to 12%. This difference can be explained, in part, by the approximation of the real grating profile for modeling. The grating is not perfectly sinusoidal and the line-space ratio is not equal to 1. In other words, the line has a width of 200 nm and forms an angle with the groove (800 nm width) that disturbs propagation of the plasmon wave. Moreover, the modeling considers a grating without any imperfection like the surface roughness or internal cracks illustrated in the SEM image of Fig. 6(b). But, whatever those differences between modeling and experimental data, Fig. 10 proves that detection of the resonance has effectively been achieved and demonstrates ability of the TiO_xN_y grating elaborated by ammonolysis to be used as a plasmonic component. It is worthwhile noticing that several measurements (very sensitive to the material permittivity) have been realized in the first months after ammonolysis and no significant changes have been observed, which provides first insight in the stability of our ammonolyzed samples.

6. Conclusion

In this article, a process for the ammonolysis of a TiO₂ grating involving a directly photo-patternable sol-gel thin film layer has been demonstrated for the first time, with the aim to use the derived gating as a plasmonic component. The characteristics of the nitrided grating led to the experimental demonstration of the plasmonic resonance at the metal-air interface in the NIR range. The grating elaborated from the TiO₂ nitridation does not lead to a pure TiN and induces roughness due to the ammonolysis treatment but the derived plasmonic component is easy to realize and excitation of the plasmon is directly achieved by coupling the incident wave by a quasi-sinusoidal grating. From these good and promising results and despite the chemical structure and the roughness and the defects of the TiO_xN_y layer, the authors are currently working on further understanding of the plasmon mode excitation on such TiO_xN_y layer.

The main advantage of our method is the easy way to fabricate the component. Thus, it can be applied to non-standard substrates (for example non-planar, flexible and large substrates), since the complete process is summarized by a lithography and heating process, without any expensive and heavy alternative processes such as etching and/or lift off. On one part, this approach can thus open a new route for metallic grating fabrication and, on the other part, the sol-gel technology is no longer restricted to ceramic and dielectric material and opens the route toward structured metallic materials.

In this paper, our primary goal was to demonstrate the feasibility of a plasmonic material from a sol-gel. This has been proved by the negative real part of the permittivity and by the strong difference in the orders of magnitude of the electrical property between the TiO₂ and our TiO_xN_y. To go further, even if a stoichiometric TiN leads to plasmonic compound with less loss than the TiO_xN_y [11], the process of ammonolysis can again be improved in order to reach a plasmonic material with an optimized grating profile and to reduce roughness of the grating surface. For instance, ammonolysis can be achieved by Rapid Thermal Annealing (RTA), enabling a better control of the temperature and of the gas pressure to acces. On the basis of previous studies [21,22], authors are also currently improving the heating process as well as the lithography one using interference lithography set-up and colloidal lithography set-up for large and non-planar substrates. Finally, it will also be fruitful to assess the TiO_xN_y stoichiometry, for instance by XPS measurements, in order to correlate the y/x ratio to electrical and plasmon properties.