Study of the Optoelectronic Properties of Titanium Nitride Thin Films Deposited on Glass by Reactive Sputtering in the Cathodic Cage

We investigate the structural, optical, thermal-optical, and electronic properties of TiNx thin films utilizing a variety of experimental techniques, including spectroscopic ellipsometry, Raman spectroscopy, scanning electron microscopy, atomic force microscopy, thermal lens spectroscopy, and UV-VIS spectroscopy. Our experimental results indicate a remarkable metallic character in the TiNx thin films deposited under lower N2 flow during treatment, as well as an increase in reflectance in the infrared region and thermal diffusivity as the partial N2 flow is reduced, which is consistent with previous experimental studies. The microscopic origin of these trends is explained in terms of the atomic structure of the system, where Ti atoms contribute free carriers that interact with IR radiation and N atoms create imperfections in the lattice, causing greater scattering of phonons, respectively. The experimental results also indicate that the roughness of the films produced with a lower N2 partial flow was lower than that of the films with a higher N2 partial flow. To explain the deviation of optical properties from the ideal case, surface oxidation is also investigated. This facilitates the coating of smarting windows with TiNx thin films.


Introduction
The continuous use of fossil fuels is one of the main factors that intensify global warming and environmental pollution.In addition, the depletion of non-renewable resources essential for the continuous development and production of technologies that help solve problems and improve humanity's living conditions stimulate the search and development of new materials and alternative technologies that effectively replace their precedents [1][2][3][4][5] .Innovating to minimize energy use while achieving the same or better results is significant to tackle the core of the problem: the massive use of fossil fuels and the growing need for electricity.To slow the destructive effects of global warming, energy-efficient buildings, vehicles, and windows are gaining increasing attention, as an indoor environment kept warm or cold with little or no electricity affects energy consumption [1][2][3][4][5][6] .
Used as a coating for windows in homes, buildings, and vehicles with materials in the visible spectrum and with reflectance in the infrared range, the thermoregulatory material is a viable strategy as a starting point to reduce the consumption of electricity spent on heating and heating lighting [1][2][3][4][5][6][7] .Titanium nitride (TiN) is included in the transition metal nitride set that presents remarkable resistance to several factors, such as extreme temperature and pressure conditions, and durability capable of being used in multiple technological areas, from cutting tools to rocket parts 8,9 .TiN thin films have been widely studied in recent decades, given their importance and applicability in several commercial areas due to the combination of their mechanical properties -high melting point, hardness, corrosion resistance -and optoelectronic properties, which derive from the metallic behavior of free electrons existing in the d-band of Ti.TiN can also be applied as a heat mirror, as it is very stable when subjected to high temperatures and highly reflective in the near-infrared region 7,[10][11][12] .
The production of thin films by sputtering is a recent technique derived from the sputtering deposition method.A cylindrical cage, composed of the material to be deposited, houses the substrate inside, and both are positioned in the cathode reactor.The new technique, developed and patented by the Plasma Materials Processing Laboratory of the Federal University of Rio Grande do Norte (UFRN) (PI0603213-3) in 2007, has advantages that make it viable and effective, including the elimination of the edge effect and overheating that commonly occur in plasma nitriding processes, use in some of the existing deposition equipment, high film deposition rate, etc. Bilokur et al. 13 investigated the optical properties of TiN, AlON, and TiAlN thin films deposited on glass and Si.The authors report that stoichiometric TiN has high reflectance in the infrared and low in the visible light spectrum.However, the best results for the spectral selectivity property occurred in films where the combination of dielectric, semiconductor, and conductor materials in multiple layers was performed.
In this work, cathodic cage plasma deposition (CCPD) will be performed varying nitrogen flow to produce thin TiN films on glass substrates.The results should contribute to discussing the relevance and applicability of TiN as a functional solar selective coating for energy-saving technology applications.

Materials and Methods
In this study, we used as substrate microscopic glass plates cut in the dimensions of 25.4 x 12.3 mm and 1 mm thick, where we deposited TiN x thin films through the method of plasma deposition in a cathodic cage.Table 1 14 and Table 2 show the chemical composition of the substrates and the cathode cage, respectively.
The cathode cage was milled from a titanium sheet, and its dimensions are 80 mm (height) × 100 mm (∅), with evenly spaced holes of 10 mm (∅) and center-to-center spacing of 20 mm, and sanded with 600 mesh sandpaper.The cathodic cage and the samples were cleaned by immersion in nitric acid, washed in running water, and immersed in acetone under ultrasonic agitation, followed by a conventional dryer.
After this process, the substrates and the cage were placed in the treatment chamber.
The equipment used for plasma thin film deposition consists of a direct voltage source with maximum voltage and current of 1200 V and 2 A, respectively, and a stainless steel chamber 30 cm in diameter and 40 cm in height.Plasma is formed in the cathode cage, which acts as the cathode (the chamber lid is the anode).The samples remained at floating potential because they were positioned on an insulating surface, alumina.
The samples underwent a pre-sputtering phase for 1 hour at 350 °C in a gaseous mixture of Ar/H 2 in the ratio of 2:1 (Ar: 20 sccm; H 2 : 10 sccm) to remove oxides and impurities from the surface of the cage.The sputtering phase lasted for 4 hours.The samples were subjected to treatment temperatures of 400 °C, where the proportion of the gas mixture was varied according to Tables 3 and 4 to observe how the characteristics of the films would change according to the different treatment parameters.
Raman spectra were obtained with a Bruker Senterra mono-array spectrometer equipped with a charge-coupled device (CCD) as a detection system, and the samples were excited using a solid-state laser (785 nm) with a power of 50 mW.The scanning electron microscope (SEM) from EMPRESA FEI, model QUANTA FEG 250, was used to evaluate the thickness and morphology of the deposited layers.Ten thickness measurements were taken along the length of the layers to obtain mean values and their standard deviations.The roughness and morphology of the deposited films were evaluated by measurements performed in atomic force microscopy (AFM) NTEGRA Spectra in semi-contact mode.In each sample, five measurements were performed in different regions.The area was 100, 25, and 9 μm 2 digitized with a sweep frequency of 0.78 Hz.Each image was acquired with 512x512 pixels.
A Shimadzu UV 3600 spectrophotometer with a deuterium-tungsten excitation source together with the UV-Probe-233 software was used to measure the reflectance (R) and transmittance (T) of samples prepared in glass in the spectral range of 200 to 1600 nm.The optoelectronic properties of TiN x thin films were verified through spectroscopic ellipsometry in a Semilab GES-5E ellipsometer.A polarizer was used before the light source to generate a spectrum of polarized white light.A Drude Lorentz Tauc-Lorentz dielectric analysis based on the free electron and oscillator model was performed to investigate the relationship between nitrogen flux during thin film deposition and the behavior of its optoelectronic properties.
TLS measurements were performed using a dual beam mismatched beam setup, as detailed by Shen et al. 16 .The excitation laser employed a diode laser operating at 532 nm, while the probe beam utilized a He-Ne laser at 633 nm.TLS measurements were recorded in the transient regime, where the θ (phase difference) and tc (characteristic time) parameters were obtained from transient curves using a theoretical fit based on the TL equation (Equation 1), where m = 15.089 and V = 2.092 were used as fixed parameters, obtained from the calibration sample.Thermal diffusivity was calculated using Equation 2, using the fitted values of the curves from Equation 1, which are (t c ) and ω e = 3.72 × 10 -3 cm.All TLS ratios were performed at the same power, and the probe beam power was to adjust effects and not induce thermal effects.

Results and Discussion
The Raman pattern shown in Figure 1 indicates typical titanium nitride films deposited on glass samples under all treatment conditions.The Raman peaks found in the regions between 150 and 350 cm -1 and between 450 and 700 cm -1 confirm the presence of TiN.These regions are related to vibrations of heavy Ti and light N ions, respectively, due to Ti and N vacancies in the crystal lattice.The peaks of 163, 360, 486, and 615 cm -1 observed in the spectra refer to transverse acoustic (TA), longitudinal acoustic (LA), transverse optical (TO), and longitudinal optical (LO) modes, respectively, and refer to first-order Raman vibration [17][18][19] .
In order to quantify and understand the observed changes in the Raman spectrum, these were adjusted through the Lorentzian model, as shown in Figure 1.The adjustment of the Raman spectrum was used to obtain the N/Ti ratio present in the films through the peak areas, calculation of the ratio (TO+LO) / (TA+LA) 18,19 .The influence of partial N 2 flow on the N/Ti ratio is shown in Figure 2. It is observed that the addition of N 2 during film deposition follows a decrease in its N/Ti ratio, which is the opposite of what is reported in the literature [20][21][22] .   5. They show that the layer thickness increases with increasing N2/Air ratio during Cathodic Cage Plasma Deposition (CCPD).Thus, the A4 sample had the most negligible thickness, while the A8 was the largest.The increase in the mean thickness that accompanies the growth of the N 2 flow is contrary to that found in studies that do not use a cathodic cage as a target, and this shows that the proportional increase in the N 2 flow decreases the deposition rate because, due to the "poisoning" of the target resulting from the formation of TiN on its surface, the performance of the reactive spraying decreases [23][24][25][26][27] .
The uniformity of the deposited layer, indicated by the standard deviation of the average thickness of the films, also grew along with the N/Ti ratio, showing medium to high values.This indicates that nitrogen in the films contributes to forming more uniform films, which is corroborated by the literature [28][29][30] .However, the increase in the flow of N 2 during deposition contributes to the non-uniformity of the films 31 .
From the observation of 2D and 3D AFM images with a scan size of 9 × 9 μm 2 of samples A4 and A8 in Figure 4, it is evident that the films grew on the surface of the substrate by nucleation and that the roughness increased considerably as the stoichiometry decreased.Samples A4 and A8 showed roughness values of 26 nm and 45 nm, respectively.It is observed that the increase in roughness is accompanied by a decrease in the N/Ti ratio, thus being a result compatible with the literature and with previous results obtained from SEM micrographs.Surface energy, deposition time, and relative gas concentration influence grain size 25,32,33 .
The data obtained through UV-VIS-NIR spectroscopy measurements show that the film that came closest to stoichiometry (sample A4 -deposited with the lowest N 2 /Ar flow) was the one that exhibited the highest reflectance, mainly in the near-infrared region with a value of 30% at 1600 nm.Stoichiometric TiN reflectance is reported in a range of 70-95% depending on the deposition method and parameters and the substrate 34,35 .The reflectance results are illustrated and represented in Figure 5.The literature shows that a high N/Ti ratio in TiN films is accompanied by low reflectance.This is due to nitrogen capturing free electrons from titanium resulting in a decrease in light reflected by the film as the N/Ti ratio increases 34,[36][37][38] .
It has also been reported that the reflectance of stoichiometric TiN films is much higher than that of sub-stoichiometric films 39 .We believe this phenomenon occurs from the point at which the films migrate from stoichiometric to super stoichiometric.In contrast, in sub-stoichiometric films, such as those presented in this work, the reflectance increases as they approach stoichiometry since the literature also reports that stoichiometric TiN films present much higher reflectance than non-stoichiometric ones 39 .Because the thickness of the films also influences this property, where it is lower in thicker films (sample A8) and higher in thinner ones (sample A4) 39 .
UV-VIS-NIR spectroscopy also provided Figure 6, from which it can be concluded that the transmittance decreases as the N 2 flow decreases, which leads us to believe that the increase in N 2 during deposition, or the decrease in the N/Ti ratio, contributed to the increase in transmittance, mainly in the visible and near-infrared spectral region.The flow of N 2 during deposition leads to a departure from the stoichiometry of the films, given the direct relationship and marked difference between the high infrared transmittance and the non-stoichiometric growth of the TiN x films.At the same time, the UV and IR region is wholly blocked in stoichiometric films 36,40 .
However, it is known that the thickness and roughness of the films also influence the reflectance and transmittance properties in such a way that the increase in thickness (sample A8 has greater thickness and roughness) results in a decrease in reflectance and transmittance (sample A8 has lower reflectance) 39,41,42 .
The optoelectronic parameters TiON and TiO 2 were used to guide the phase 2 adjustment of the A4 and A8 samples, thus improving the consistency of the models applied to these films 43,44 .The average thickness and roughness of the deposited films, obtained by SEM and AFM, respectively, were used as a fixed parameter during the adjustments of the ellipsometric measurements to build a more consistent model.The quality of the fits can be seen in Figures 7 and 8 as well as in Tables 6, 7, and 8. Optically, the films are divided into layers and substrate, as shown in Figure 7.Each layer was assigned two or more dispersion laws (Drude, Lorentz, and Tauc-Lorentz models).
The ellipsometric measurements showed Lorentz bands in the ultraviolet and visible spectral regions.These bands explain that we do not see significant reflectance or transmittance in this spectral region.Thus, these bands show absorption in this region, specifically the Lorentz bands at 2.14 eV (phase 1 -Lorentz 1), 2.6 eV (phase 2), and 5.64 eV (phase 1 -Lorentz 2) eV for sample A4, 2.07 eV (phase 1 -Lorentz 2), 4.08 eV (phase 1 -Lorentz 1) and 4.2 eV(phase 2) for sample A8, being related to the absorption of light in this range, as shown in Tables 6, 7 and 8. Through the Drude model, the plasma energy of the samples was verified.The value of this parameter for sample A4 was 4.43 eV, followed by sample A8, which presented a value of 3.53 eV, as shown in Table 6.These values are within the range reported in the literature ranging from 2.6 eV for stoichiometric TiN -and above 2.6 eV for super stoichiometric TiN x (x > 1) 45,46 and 5.45 eV 47,48 .Table 9 shows the parameters derived from the ellipsometric adjustments made where sample A4 presented higher electrical conductivity and electrical resistivity lower than sample A8.Therefore, it was verified that films closer to stoichiometry are better electrical conductors when compared to those more distant from stoichiometry.
Figure 9 illustrates the thermal lens signal observed for films A4, A5, and A8 which were deposited onto glass substrates in atmospheres with different percentages of nitrogen.The thermal lens curves were fitted using the thermal lens model described by Almeida et al 49 .The pumped beam induces excitation, resulting in a phase shift in the probe beam.This phase shift reflects the changes in the optical path length across the entire sample in relation to the trajectory of the probe beam.Variations in the phase shifts were noticed as the nitrogen content was altered in the samples.The thermo-optical parameters are listed in Table 10.The glass substrate did not show a thermal lens signal.Therefore, only the thermo-optical properties of the films were investigated.The variation in the proportion of the N:Ar gas mixture of 10, 22, and 50% affected the heat diffusion of the material, with values of 1.55x10 -3 , 1.89x10 -3, and 3.39x10 -3 cm 2 .s - for the thermal diffusivity for samples A4, A5 and A8, respectively.
Thermal conductivity is proportional to thermal diffusivity.It can be calculated by the following expression, k=Dρc, where ρ is the density and c is the specific heat of the sample.The values of ρ = 5450 kg.m -3 and c = 636 J.kg -1 .K -1 are already present in the literature 50,51 .The values for the thermal conductivity of the TiN thin films were lower than the same material in bulk, 30 W.K -1 .m -1 5 .Figure 10 shows the dependence between thermal diffusivity and the N/Ti ratio, showing that when approaching stoichiometry, there is a decrease in heat diffusion in the material 51,53 .
Therefore, the properties and characteristics observed in the films deposited through the different types of analyzes performed indicate that TiN x thin films deposited through CCPD have the potential to be applied in the development of coatings that act as thermal regulators in smart windows.

Conclusions
TiNx substoichiometric thin films were deposited on glass using the CCPD deposition technique.Through Raman measurements, it was possible to estimate the stoichiometry of the films produced.The film deposited with less N 2 flow showed higher stoichiometry than the film deposited with higher N 2 flow.Images obtained through SEM and AFM show the formation of uniform thin films in samples treated with a lower N 2 flow, where the stoichiometry achieved was higher.
This uniformity decreases as the N 2 flow increases and the stoichiometry decreases.The films produced became thicker with increasing partial N 2 flow during deposition.The UV-VIS-NIR spectra indicated that the TiN x thin films produced with the lowest N 2 /Ar ratio had higher stoichiometry (N/Ti ratio of 38%) and reflectivity (30% at 1600 nm), mainly in the near-infrared than those produced with the highest N 2 /Ar ratio (N/Ti ratio of 33%), which is promising for the production of heat mirrors or thermally regulated coatings since the deposition method can be improved for such.
The Drude, Lorentz, and Tauc-Lorentz models were applied and refined to better fit the ellipsometric data.The information collected from these adjustments indicated why the transmittance is lower in the UV and visible spectral region and higher in the infrared region since they presented Lorentz bands positioned from 2 to 5.74 eV, indicating light absorption in this spectral range.The higher reflectance in the near-infrared is due to the number of free carriers described by the Drude model, being closely linked to the plasma energy of the material.The more free carriers, the greater the reflectance in that spectral region.Thus, as plasma energy is also closely linked to the stoichiometry of thin films, future studies should address the problem of obtaining stoichiometric and super stoichiometric films.
The analysis using thermal lens spectroscopy showed a strong dependence on the thermal diffusivity of the deposited films and their stoichiometry, where the closer to the stoichiometry the film is, the lower its thermal diffusivity.Thus, the A4 sample showed the highest electrical conductivity and the lowest thermal diffusivity, thus uniting metallic and ceramic properties.This shows that, with a refinement of the method and more profound knowledge of the relationship between the deposition parameters and the general properties of the films produced, it will be possible to develop coatings for thermal regulation to be applied in smart windows.The perspective is that, in the future, coatings produced by this method can add optoelectronic and thermal properties to the treated material, thus expanding the spectrum and quality of its applications.

Figure 1 .
Figure 1.Raman spectrum of TiNx thin films deposited under partial N 2 flow variation and the adjustment performed on these: A4 (a), A5 (b), and A8 (c).

Figure 3
Figure3shows cross-sectional SEM micrographs of samples treated with different proportions of Ar and N 2 , indicating that all treatments resulted in the formation of a surface layer.The average film thicknesses are listed in Table5.They show that the layer thickness increases with increasing N2/Air ratio during Cathodic Cage Plasma Deposition (CCPD).Thus, the A4 sample had the most negligible thickness, while the A8 was the largest.The increase in the mean thickness that accompanies the growth of the N 2 flow is contrary to that found in studies that do not use a cathodic cage as a target, and this shows that the proportional increase in the N 2 flow decreases the deposition rate because, due to the "poisoning" of the target resulting from the formation of TiN on its surface, the performance of the reactive spraying decreases[23][24][25][26][27] .The uniformity of the deposited layer, indicated by the standard deviation of the average thickness of the films, also grew along with the N/Ti ratio, showing medium to high values.This indicates that nitrogen in the films contributes to forming more uniform films, which is corroborated by the literature[28][29][30] .

Figure 2 .
Figure 2. Relationship of dependence between the N/Ti ratio of the deposited TiN x films and the partial flow of N 2 during deposition: A4 (a), A5 (b), and A8 (a).

Figure 3 .
Figure 3. Electron micrographs of the cross-section and the top of the samples A4 (a), A5 (b), and A8 (c).

Figure 7 .
Figure 7. Illustration of the organization of the simulated layers by adjusting the ellipsometric measures of samples A4 (a) and A8 (b).

Table 1 .
Chemical composition of substrates.

Table 2 .
Chemical composition of the cathodic cage (Ti level 2).

Table 4 .
Sample treatment conditions.

Table 5 .
N/Ti ratio was found through Raman spectroscopy, average thickness and estimate of uniformity (S) obtained through SEM measurements, and roughness (  ) obtained through AFM measurements of the deposited films.

Table 7 .
Adjusted parameters of phase 2 (surface layer possibly oxide) resulting from ellipsometric measurements of samples A4 and A8.

Table 8 .
Parameters derived from the adjustments of the ellipsometric measurements of samples A4 and A8 where phase 3 (EMA) simulates the surface roughness of the sample.

Table 9 .
Parameters derived from the adjustments of ellipsometric measures and the total thickness of samples A4 and A8.

Table 6 .
Phase 1 adjusted parameters (thin TiN x film) resulting from the ellipsometric measurements of samples A4 and A8.