Self-Organized In-Depth Gradients in Highly Ti-Doped ZnO Films: Thermal Versus MW Plasma Annealing

: Highly Ti-doped ZnO ﬁlms have been produced by a spin-casting sol-gel process. The spin-casted ﬁlms show high in plane homogeneity and optical quality. However, when inspected in depth, the surface composition is Ti rich. We show that two possible annealing processes can be considered depending on the properties to exploit. To promote in-depth homogenization, thermal annealing processes have been applied. Meanwhile, the gradients can be exacerbated, thanks to a non-negligible surface sputtering, by applying microwave (MW) plasma treatments with Ar discharges at di ﬀ erent pressures. The microstructural properties of the di ﬀ erently processed ﬁlms have been obtained prior to a study by grazing incidence X-ray ﬂuorescence (GI-XRF) spectroscopy, which reveals the in-depth composition trends induced by the two alternative annealing procedures. The ﬁnal wetting, electrical and optical properties of the ﬁlms are described in accordance with the Ti distribution pattern revealed by GI-XRF. The study underlines for the ﬁrst time how MW plasma annealing processes can be used to exacerbate self-induced atomic gradients in sol-gel ﬁlms with potential implications in catalytic and biomedical applications.


Introduction
The applications of advanced materials have increasingly implied their deposition in the form of thin films in conjunction with their micro-nanostructuring. ZnO has been exposed to this trend in a wide range of applications, such as in photovoltaics [1], light-emitting devices [2], biosensors [3] or photocatalysis [4], which exploit the tunable light absorption and charge transfer properties of ZnO. From the point of view of processing, ZnO is an extremely flexible material with adapted protocols described for physical and chemical deposition techniques. The criticality is often determined by dopant elements that should tailor the desired properties; i.e., the charge transfer, light absorption, surface state, etc. As examples of doped ZnO deposition by physical techniques, Al-doped ZnO film synthesis has been achieved by using magnetron sputtering [5], while blue-emitting Er-doped ZnO films have been deposited by electron beam evaporation [6] or magnetic Co-rich ZnO was obtained by pulsed laser deposition [7]. In parallel, for the chemical techniques, Li- [8] and Ga- [9] doped ZnO deposition was described by the sol-gel process leading to highly oriented structures. Additional examples can be found of deposition of B-doped (n-type) ZnO by metal organic chemical vapor deposition [10] or Zr-doped ZnO by atomic layer deposition (ALD) [11]. of monoethanolamine diluted with ethanol was mixed with the first solution for stabilization [19]. The mixture was stirred under reflux for 10 h at 60 • C. TiO 2 sol was prepared by dissolving 0.2 M of titanium isopropoxide (Ti[OCH(CH 3 ) 2 ] 4 ≥98%, Sigma-Aldrich) in ethanol. We added 0.35 µl of HCl to the solution to induce an acid condensation catalysis [19]. Our study was performed with 5% Ti-doped ZnO solution (5% TZO) prepared by mixing aliquots of the previously described sols. The spin-casting technique was used to deposit the TZO sol on Si <100> and glass substrates previously cleaned by sequential ultrasonic baths for 5 min in trichloroethylene, acetone and ethanol. Deposition was carried out by casting 50 µl of the precursor and spinning at 2000 rpm spin speed for 30 s under N 2 flow.

Annealing of Ti-Doped ZnO Thin Films
Spin-casted films were exposed to thermal annealing (TA) or microwave plasma annealing (MwPA). TA was performed at 200 • C or 300 • C for 30 min and MwPA in Ar at two different gas pressures (0.3 mbar and 0.7 mbar) was applied for 2 min after each spin casting step. The process of casting and subsequent annealing was repeated 5 consecutive times, except for the films prepared for XRF analysis, which were prepared with a single layer. For simplicity in the description of the samples, they are labelled as (TZt200 or TZt300) for TA films at 200 • C and 300 • C, respectively, and (TZp0.3 or TZp0. 7) for MwPA films at 0.3 mbar and 0.7 mbar, respectively. Our home-made MwPA system was previously described [26].

Microstructural Characterization
Qualitative and quantitative composition analyses were performed by total-reflection X-ray fluorescence (TXRF) and GI-XRF. TXRF and GI-XRF data were acquired in a TXRF 8030C spectrometer from Atomika (Oberschleißheim, Germany), equipped with a 3 kW X-ray tube (Mo/W alloy anode with a double-W/C multilayer monochromator and a periodicity of 2d = 5nm) with an excitation energy of 17.4 keV (Mo Kα). A silicon drift detector (SDD) detector with an active area of 100 mm 2 and a resolution better that 150 eV for 5.9 KeV (Ketek GmbH, Munich, Germany) was used. Additionally, the 8030c spectrometer is able to perform GI-XRF experiments around the critical angle. An offset of 3.5 mrad (0.2 • ) was measured in the spectrometer by comparing the theoretical critical angle of Si(100). The measurements were performed at 50 kV and the intensity was adjusted automatically so that a maximum count rate of about 8500 cps was achieved. Qualitative spectra were acquired during 500 s of acquisition time.
The effect of MwPA on the films' morphology was evaluated by field-emission scanning electron microscopy (FESEM) using a Philips XL-40FEG operated at voltages below 10 kV. X-ray diffraction (XRD) studies were performed using a PANalytical X'Pert PRO THETA/THETA diffractometer in grazing-incidence configuration using Cu-Kα radiation (λ = 1.54 Å), a fixed incidence angle of 0.5 • and 2 • θ range from 10 • to 60 • with 0.04 • increments and 10 s accumulation time. The equipment counts with an X-ray mirror for copper radiation, a parallel plate collimator, a flat pyrolytic graphite secondary monochromator and a Xe gas detector. The system was verified prior to the measurement of the batch of TZO films by measuring a sample of compacted Si microparticles (PANalytical) and verification of diffraction peak positions' deviations within ∆2 θ ≤ ± 0.03 •

Optical, Electrical and Wetting Characterization
The optical properties of TZO films were measured on films deposited on glass substrates using a V-560 double-beam spectrophotometer (Jasco, Easton, MD, USA), equipped with an integrating sphere to avoid scattering losses. The optical absorption was recorded in the ultraviolet-visible (UV-vis) range (250-900 nm). Tauc plots were obtained to compare the optical bandgap of TA and MwPA films.
The electrical performance of the TZO films was compared by electrochemical impedance spectroscopy (EIS) using by a Bio-Logic SP-150 system (Seyssinet-Pariset, France). The measurements were done after deposition of Au point contacts on the films by direct current (DC) sputtering. The alternating current (AC) impedance spectra were acquired in a Faraday cage between two surface electrodes up to 500 kHz frequency and applying a 0.5 V excitation potential. Fitting of data to electrical equivalent circuits was performed using the EC-Lab software from Bio-Logic Science Instruments (Seyssinet-Pariset, France).
The measurement of the water contact angle (WCA) on the different TZO films was performed in static mode using a KSW 100 with droplet volumes of 3 µL. The results are presented as the mean value of five WCA measurements on the each type of TZO film.

Thin Film Morphology
The analysis of Ti-doped ZnO thin films was performed after sequential deposition of 5 layers and application of the annealing process after each spin-casting cycle. Figure 1 shows the comparison between the films' morphological structure subjected to TA at characteristic temperatures of 200 • C Figure 1a and 300 • C Figure 1b and films treated by MwPA at plasma pressures 0.3 Figure 1c and 0.7mbar Figure 1d. All films show good homogeneity with relatively flat surfaces and only soft wavy features with sizes below the wavelength of the visible range, which guarantee a good optical quality. The inspection of the total film thickness denotes a specific degree of condensation of the films related to their annealing conditions. Relevantly, the MwPA annealing treatments appear as the most drastic and less influential annealing treatments, respectively. However, the film annealed at 300 • C shows some granular structure, which is indicative of a more intense condensation and development of a nano-crystalline state [2]. It is convenient to remember that the MwPA process is influenced by both, a thermal component and an ion and electron surface bombardment that can induce surface sputtering. In this sense, in view of the granular microstructure, TZt300 appear as the most influential in the condensation of the TZO films. As regards the comparison of the two MwPA conditions, it is patent that the pressure reduction to 0.3 mbar is critical in providing a higher Ar ionization efficiency and energy per ion [27], which can contribute, besides a facilitation of desorption processes, to an enhanced thickness reduction. The measurement of the water contact angle (WCA) on the different TZO films was performed in static mode using a KSW 100 with droplet volumes of 3 μL. The results are presented as the mean value of five WCA measurements on the each type of TZO film.

Thin Film Morphology
The analysis of Ti-doped ZnO thin films was performed after sequential deposition of 5 layers and application of the annealing process after each spin-casting cycle. Figure 1 shows the comparison between the films' morphological structure subjected to TA at characteristic temperatures of 200 °C Figure 1a and 300 °C Figure 1b and films treated by MwPA at plasma pressures 0.3 Figure 1c and 0.7mbar Figure 1d. All films show good homogeneity with relatively flat surfaces and only soft wavy features with sizes below the wavelength of the visible range, which guarantee a good optical quality. The inspection of the total film thickness denotes a specific degree of condensation of the films related to their annealing conditions. Relevantly, the MwPA annealing treatments appear as the most drastic and less influential annealing treatments, respectively. However, the film annealed at 300 °C shows some granular structure, which is indicative of a more intense condensation and development of a nano-crystalline state [2]. It is convenient to remember that the MwPA process is influenced by both, a thermal component and an ion and electron surface bombardment that can induce surface sputtering. In this sense, in view of the granular microstructure, TZt300 appear as the most influential in the condensation of the TZO films. As regards the comparison of the two MwPA conditions, it is patent that the pressure reduction to 0.3 mbar is critical in providing a higher Ar ionization efficiency and energy per ion [27], which can contribute, besides a facilitation of desorption processes, to an enhanced thickness reduction.

Total-Reflection X-Ray Fluorescence Spectrometry
The qualitative evaluation of the composition of the TZO thin films was made by acquiring TXRF spectra at incidence angles lower than the critical angle of the material on films prepared with only one spin casting/annealing cycle. Figure 2 shows the TXRF spectra of TZO thin films (5 at. % Ti to Zn) with different annealing conditions and normalized to Zn Kα signal: Figure

Total-Reflection X-Ray Fluorescence Spectrometry
The qualitative evaluation of the composition of the TZO thin films was made by acquiring TXRF spectra at incidence angles lower than the critical angle of the material on films prepared with only one spin casting/annealing cycle. Figure 2 shows the TXRF spectra of TZO thin films (5 at. % Ti to Zn) with different annealing conditions and normalized to Zn Kα signal: Figure   TXRF spectra show the presence of Zn and Ti, as major elements and Ca, Fe and Ni, as traces with a maximum concentration of Fe and Ca of 1.2 wt% for the TZp0.3 film, which fall within the maximum specified by the supplier. There is also an Ar signal associated to air composition and a Si peak related to the Si(100) substrate. Spectra show strong variations of the Si signal, mainly in Figure 2d. The high intensity of the Si peak and Compton scattering and MoKα from the source can only be understood in terms of a higher incidence angle than the critical angle for total reflection in the TZp0.7 film. Relevantly, this film demonstrated a condensation rate well below the observed for the other samples (see SEM results). Since the critical angle for total reflection is proportional to the square root of the density of the material, this sample does not appear to comply with the rational critical angle during the experiment.
The calculated penetration curve (see a Figure A1) shows that the depth evaluated in the qualitative TXRF analysis was always lower than the first 5 nm of material except for the TZp0.7 film (Figure 2d), where the probing of the coating may have extended deeper due to a bad estimation of the film critical angle. In fact, the density of ZnO and other sol-gel thin films can be determined from the critical angle of incidence [28], which may require a specific calculus of the critical angle for this Ti-rich film. In fact, the theoretical TZO mass proportion, considering 5% of Ti on ZnO, was 6.22:100 for Ti:Zn, which implies an estimated mass proportion of Ti/Zn = 0.062. However, the experimental mass proportions obtained for   TXRF spectra show the presence of Zn and Ti, as major elements and Ca, Fe and Ni, as traces with a maximum concentration of Fe and Ca of 1.2 wt% for the TZp0.3 film, which fall within the maximum specified by the supplier. There is also an Ar signal associated to air composition and a Si peak related to the Si(100) substrate. Spectra show strong variations of the Si signal, mainly in Figure 2d. The high intensity of the Si peak and Compton scattering and MoKα from the source can only be understood in terms of a higher incidence angle than the critical angle for total reflection in the TZp0.7 film. Relevantly, this film demonstrated a condensation rate well below the observed for the other samples (see SEM results). Since the critical angle for total reflection is proportional to the square root of the density of the material, this sample does not appear to comply with the rational critical angle during the experiment.
The calculated penetration curve (see a Figure A1) shows that the depth evaluated in the qualitative TXRF analysis was always lower than the first 5 nm of material except for the TZp0.7 film (Figure 2d), where the probing of the coating may have extended deeper due to a bad estimation of the film critical angle. In fact, the density of ZnO and other sol-gel thin films can be determined from the critical angle of incidence [28], which may require a specific calculus of the critical angle for this Ti-rich film. In fact, the theoretical TZO mass proportion, considering 5% of Ti on ZnO, was 6.22:100 for Ti:Zn, which implies an estimated mass proportion of Ti/Zn = 0.062. However, the experimental mass proportions obtained for the TXRF evaluation of the spectra of the investigated films (shown in Table 1) show a remarkable Ti super-stoichiometry. Surface compositions are far from the theoretical composition and in all cases were always Ti rich. For TZt200 and TZt300 films the Ti enrichment was higher than 2 times the expected and for TZp0.3 and TZp0.7 films the enrichment was around 3 times higher. The results showed that the coatings, irrespective of the post-deposition treatment, were rich in Ti and indicated that the mechanism of formation of the gradient must have an origin in the deposition process. Additionally, the surface enrichment was higher in MwPA thin films, which can be explained in terms of the different sputtering yield of the two atoms in the final material. As illustrated in Appendix B, the sputtering of Zn is higher than for Ti in a wide range of energies for Ar ions impinging a simulated TZO film with 5 at.% Ti with respect to Zn. This is in agreement with the expected sputtering observed for Zn and Ti as induced by Ar ions [29] and explains the observed increase of relative Ti composition on the MwPA TZO samples

Grazing Incidence X-Ray Fluorescence Spectrometry
The K-line intensities of the elements Zn, Ti and Si were evaluated around the critical angle of the TZO material and Si(100) wafer. Lower angles imply information of the first depth nanometers and the depth of material inspected increases when the angle increases as shown in Figure 3. A depth profile of the relative composition of Ti/Zn along the first hundreds of nanometers was obtained. Figure 3 shows, for each annealing condition, the Ti/Zn mass ratios obtained from the GI-XRF measurements from incidence angles between 0 to 6.4 with a step of 0.1 mrad (i.e., 64 spectra for each thin film). For all the films, a surface enrichment in Ti was observed around the first tens of nanometers. After this depth, the Ti/Zn mass ratio reached a plateau in all the cases. The horizontal dotted line displays the theoretical Ti/Zn mass ratio. For sample TZt200 (Figure 3a) the Ti/Zn mass ratio varied from 0.15 to 0.05 and for TZt300 ( Figure 3b) the ratio varies from 0.18 to 0.12 from the surface to bulk, respectively. The gap was lower for the films annealed at 300 • C probably due to the highest activation of diffusion that tends to equilibrate the Ti/Zn ratio all over the surface. In any case, the Ti/Zn mass ratio measured for all concentrations was higher than expected, which confirms that the condensation process during spin coating is not homogeneous and more Ti than Zn was retained during xerogel film formation. There were also evident differences between TZp0.3 and TZp0.7 films as shown in Figure 3c,d, respectively. In both cases the Ti/Zn mass ratio was always higher than the expected value of 0.062. We can also observe a concentration gradient, with a Ti/Zn ratio higher at the surface and lower in the bulk. However, for the TZp0.3 film it varied from 0.19 to 0.11 while for the TZp0.7 film varies from 0.7 to 0.11. These results confirm on the one hand that the condensation process led to a preferred overall retention of Ti with respect to Zn. On the other hand, it becomes also clear that the plasma annealing induced an opposite trend to the TA, with an enhancement of the Ti stoichiometry on the surface with respect to the bulk. This can be again understood in terms of the difference in the  Figure 4 shows the normalized intensity for the different film/substrate elements at different Xray incidence angles. The results allow us to compare how the composition of Ti and Zn follow each other in terms of depth, and how their compositions evolve with respect to the Si substrate. The TZt200 film (Figure 4a) presents a remarkable mismatch between the increase of the surface Ti and Zn, making it possible to visualize the film as a real stack a TiO2 film doped with Zn (on the surface) and a ZnO film doped with Ti. This observation again emphasises the importance of the condensation process of the xerogel film during spin casting. After annealing at 300 °C (Figure 4b) such stack does no longer exist and although Ti is still over-stoichiometric, the evolution of the stoichiometry is much more homogeneous, as can be derived from a process that has notably activated diffusion (note the decreasing steepness of the normalized intensity curves). For the TZp0.3 film (Figure 4c) the steepness is high, although there is no shift between the curves suggesting a stacking. For the TZp0.7 film (Figure 4d), the steepness of the curves for both Ti and Zn is radically reduced, which suggests that the thermal activation effect at this pressure is higher than the one induced in the TZp0.3 film.

X-Ray Diffraction
The TZO films produced by 5 cycles of spin casting and annealed by the different processes were analyzed by XRD. The diffraction diagrams are shown in Figure 5 and denote, in general, the absence of a consolidated microstructure. In the case of TA films, the diagrams show some wide peaks for the TZt200 film (Figure 5a) but some additional emerging peaks for the TZt300 film (Figure 5b), which suggests a short range order organization.  Figure 4 shows the normalized intensity for the different film/substrate elements at different X-ray incidence angles. The results allow us to compare how the composition of Ti and Zn follow each other in terms of depth, and how their compositions evolve with respect to the Si substrate. The TZt200 film (Figure 4a) presents a remarkable mismatch between the increase of the surface Ti and Zn, making it possible to visualize the film as a real stack a TiO 2 film doped with Zn (on the surface) and a ZnO film doped with Ti. This observation again emphasises the importance of the condensation process of the xerogel film during spin casting. After annealing at 300 • C (Figure 4b) such stack does no longer exist and although Ti is still over-stoichiometric, the evolution of the stoichiometry is much more homogeneous, as can be derived from a process that has notably activated diffusion (note the decreasing steepness of the normalized intensity curves). For the TZp0.3 film (Figure 4c) the steepness is high, although there is no shift between the curves suggesting a stacking. For the TZp0.7 film (Figure 4d), the steepness of the curves for both Ti and Zn is radically reduced, which suggests that the thermal activation effect at this pressure is higher than the one induced in the TZp0.3 film.

X-Ray Diffraction
The TZO films produced by 5 cycles of spin casting and annealed by the different processes were analyzed by XRD. The diffraction diagrams are shown in Figure 5 and denote, in general, the absence of a consolidated microstructure. In the case of TA films, the diagrams show some wide peaks for the TZt200 film (Figure 5a) but some additional emerging peaks for the TZt300 film (Figure 5b), which suggests a short range order organization.
The two peaks in the diagram of the TZt300 film are in agreement with the (100) and (002) peak assignments for ZnO crystals in a wurzite structure [2]. In any case, the lattice constriction effects induced by the high concentration of Ti dopant are responsible for an inhibition of the crystallization of ZnO for a given thermal activation, as previously observed [19]. This decrease in the crystal structure by TiO 2 attributed to the higher activation energy required for the crystallization of TiO 2 [30]. In fact, sol-gel ZnO films annealed at 200 • C have been observed to develop an incipient polycrystalline microstructure on Si substrates, which is consolidated after annealing at 300 • C [26,31]. Figure 5 shows that all the samples induced only an incipient crystallization with wide diffraction peaks denoting nanocrystalline phases. The diffraction diagrams from TZp0.3 (Figure 5c) and TZp0.7 (Figure 5d) present even wider peaks than the TA ones. Even though the peaks appear at positions which agree with the ZnO pattern (30-35 and around 56 • ), the wideness rather suggest a fully amorphous state for these TZO films.

Electrical, Optical and Wetting Properties
A comparative study of the electrical properties of the TZO thin films with different annealing was performed by impedance spectroscopy. Figure 6a shows, as an illustrative example, the Nyquist plot for the TZp0.3 film with an indication of a resistance (R) and capacitor (C) in parallel equivalent circuit model. Such equivalent circuit was valid for all the studied films, which allowed obtaining by fitting an estimation of R and C for each film. An R vs. C map is presented in Figure 6b and shows the improvement in the electrical conduction for MwPA with respect to TA films as demonstrated by the decreasing resistance. We had previously shown that inclusion of Ti implies an increase in surface resistivity with respect to pure (TA) ZnO [19], so that a MwPA process can partially compensate for this drawback. This confirms the trend of reduced surface resistivity of MwPA samples for MwPA with respect to TA [26]. However, the MwPA-treated samples lead also to an increasing capacitive behavior.
In any case, while the resistance of MwPA films with respect to the best TA film (TZt300) decreases by a factor of 10, the capacitance increases only by a factor of 2-3. Since the general applications of TZO in optoelectronics imply operation in DC (photovoltaic cells), the results point to a favorable integration of MwPA TZO films conduction/extraction layer. In addition, the value of the optical bandgap for the transparent films was extracted by measurement of the UV-vis optical characteristics in the 250-900 nm wavelength range: transmission, from TZO films deposited on glass, and reflectance from TZO analogues deposited on Si (100). The data were transformed to obtain Tauc plots [32] and estimate the absorption coefficient (α). Moreover, assuming that TZO has a direct bandgap, the relationship between (α.d) 2 and the photon energy (where d is the film thickness) was plotted to extract the optical bandgap (Eg). This was achieved by extrapolating the linear behavior at high energies to the horizontal line at (α.d) 2 = 0, as indicated in Figure 6c (as an illustrative example for the TZp0.3 film). The curves presented in all cases a dual linear regime, which could be utilized to the determination of a double optical bandgap that can be observed in compounded semiconductors. For the TZp0.3 film, the first linear regime can be identified from 3.0 to 3.7 eV, while the second extends from 3.85 to 4.1 eV. Figure 6d shows the variation of the value of the highest Eg obtained for the four TZO films with the different annealing treatments. The results confirm the added value of the MwPA to induce higher Eg structures with respect to TA films [26], even with respect to temperatures of 400 and 600 • C [19]. The results obtained are in agreement with the physicochemical properties and suggest that the high doping level of Ti is able to maintain an amorphous/nanocrystalline structure of ZnO, which gives rise to high values of the optical bandgap. Meanwhile, the values obtained for TA films remain close to the standard for ZnO (3.37 eV), the values for the MwPA annealed samples increase higher than the values obtained for ZnO nanowires (3.65 eV) [33].

Electrical, Optical and Wetting Properties
A comparative study of the electrical properties of the TZO thin films with different annealing was performed by impedance spectroscopy. Figure 6a shows, as an illustrative example, the Nyquist plot for the TZp0.3 film with an indication of a resistance (R) and capacitor (C) in parallel equivalent circuit model. Such equivalent circuit was valid for all the studied films, which allowed obtaining by fitting an estimation of R and C for each film. An R vs. C map is presented in Figure 6b and shows the improvement in the electrical conduction for MwPA with respect to TA films as demonstrated by the decreasing resistance. We had previously shown that inclusion of Ti implies an increase in surface resistivity with respect to pure (TA) ZnO [19], so that a MwPA process can partially compensate for this drawback. This confirms the trend of reduced surface resistivity of MwPA samples for MwPA with respect to TA [26]. However, the MwPA-treated samples lead also to an increasing capacitive behavior.  We further studied the wetting surface characteristics of the TZO films after the different annealing. Sessile droplet water contact angle measurements on the surfaces of the different TZO films are presented in Figure 7. As expected from the previously described physicochemical differences from TA and MwPA samples, the WCA of the surfaces were determined by the post-processing technique. Even though all the surfaces could be described as hydrophilic, TA induced higher contact angles close to 40 • , with no clear differences between annealing at 200 and 300 • C. On the other hand, the MwPA surfaces presented lower contact angles with a reinforced hydrophilic character for the sample annealed at 0.3 mbar. Relevantly, such WCAs are similar to values obtained for Ar plasma processed anatase thin films produced by magnetron sputtering at similar pressures [34]. This observation supports the trend observed in MwPA TZO films, which consist of a Ti rich surface finishing.
processing technique. Even though all the surfaces could be described as hydrophilic, TA induced higher contact angles close to 40°, with no clear differences between annealing at 200 and 300 °C. On the other hand, the MwPA surfaces presented lower contact angles with a reinforced hydrophilic character for the sample annealed at 0.3 mbar. Relevantly, such WCAs are similar to values obtained for Ar plasma processed anatase thin films produced by magnetron sputtering at similar pressures [34]. This observation supports the trend observed in MwPA TZO films, which consist of a Ti rich surface finishing.

Conclusions
In the present work, we demonstrate that the surfaces of sol-gel processed Ti-rich ZnO thin films exhibit from their casting an enrichment in Ti. Spin casting from sols containing 6.22 wt.% Ti with respect to Zn led to overall Ti contents higher than 14 wt.%. This phenomenon can be attributed to the fastest condensation catalysis of the Ti precursor with respect to the Zn precursor. Not only were the films enriched, but also there was an in-depth gradient with an acute surface enrichment in Ti that decreased smoothly to reach a stable composition after a few tens of nm. This surface enrichment in Ti can evolve differently depending on the post-processing technique selected to anneal the TZO films. In order to produce more homogeneous films, a TA at 300 • C can be applied. Inversely, one can exacerbate the surface enrichment in Ti by applying a MwPA process. At the particular pressure of 0.7 mbar, the preferential surface sputtering of Zn allowed defining the formation of a ternary compound (Ti/Zn mass ratio reaching 0.7) at the surface. For the compared post-processing conditions, the films exhibited rather amorphous (especially MwPA samples) or nanocrystalline microstructure (especially TA samples).
The post-processing techniques determined also different optical, electrical and wetting properties. The MwPA films exhibited lower surface resistivity (and slightly higher capacitance) in spite of a remarkable optical band gap (comparable to that of ZnO nanostructures) and behaved as highly hydrophilic surfaces. These results are extremely attractive for implementation of the thin films in photovoltaic devices (in view of the promising low resistivity) and for photocatalytic processes (featuring wide optical band gap and hydrophilic surfaces) ensuring high molecular affinity for the degradation of hydrophilic (polar) water pollutants.
This novel approach proposed to exacerbate the interface stoichiometric gradients self-generated during sol-gel processing may be attractive for a wide range of ternary ceramic thin films (i.e., composed of transition metal oxides such as Ti, Zr, Va, Ta, Sn, etc.) where surface properties may be optimized with respect to interface properties.

Acknowledgments:
We greatly acknowledge the technical assistance of L. García Pelayo and I. Poveda during materials processing and characterization.

Conflicts of Interest:
The authors declare no conflict of interest. Figure A1 shows the simulation of the penetration depth curve as a function of incidence angle for a monochromatic MoKα X-ray beam (17.45 keV) irradiating a layer of ZnO (pZnO = 5.61 g cm −3 ) doped with 5% of Ti (pTi = 4.506 g cm −3 ) following the model of Slaboski et al. We considered a final material density (pTZO = 0.95pZnO + 0.05pTi) of 5.554 g.cm −3 . Additionally, the X-ray optical constants of the TZO material were also calculated obtaining a scattering term δTZO = 3.55 × 10 −6 , an absorption term βTZO = 1.328 × 10 −-7 and a critical angle for this material of α c = 0.153 • .

Appendix A
Coatings 2020, 10, x FOR PEER REVIEW 12 of 14 Figure A1. Calculated depth penetration curve for the 5% doped TZO film composition considered in this work. Figure B1 shows the simulation of the sputtering yield for a TZO film with 5 at.% Ti and a density of 5.554 g·cm −3 for different energies up to 200 eV. The simulations were performed using a SRIM-2013 (Berlin, Germany) [35] to calculate the surface sputtering damage with Ar ions impinging in the normal direction to the film surface. The resulting sputtering yields (Y) for Zn, O and Ti were satisfactory fitted to a Bohdansky behaviour [29] in an extended energy range.  Figure A2 shows the simulation of the sputtering yield for a TZO film with 5 at.% Ti and a density of 5.554 g·cm −3 for different energies up to 200 eV. The simulations were performed using a SRIM-2013 (Berlin, Germany) [35] to calculate the surface sputtering damage with Ar ions impinging in the normal direction to the film surface. The resulting sputtering yields (Y) for Zn, O and Ti were satisfactory fitted to a Bohdansky behaviour [29] in an extended energy range.

Appendix B
Coatings 2020, 10, x FOR PEER REVIEW 12 of 14 Figure A1. Calculated depth penetration curve for the 5% doped TZO film composition considered in this work. Figure B1 shows the simulation of the sputtering yield for a TZO film with 5 at.% Ti and a density of 5.554 g·cm −3 for different energies up to 200 eV. The simulations were performed using a SRIM-2013 (Berlin, Germany) [35] to calculate the surface sputtering damage with Ar ions impinging in the normal direction to the film surface. The resulting sputtering yields (Y) for Zn, O and Ti were satisfactory fitted to a Bohdansky behaviour [29] in an extended energy range.