A set of 11 plates of Ti (Johnson Mattey, 99.7%) of approximately 100 mm² were employed in all experimental conditions. The plates were polished till the surface becomes mirror-like using a 1 µm and 0.3 µm particle size alumina suspension (Büehler). These plates were heated in an oven in atmosphere of air at a rate of 20˚C∙min⁻¹. As soon the desired temperature was reached, the plates were taken out of the oven. Four of these plates were heated till 450˚C, 600˚C, 750˚C and 900˚C and taken out of the oven when the temperature reached the desired value. These temperatures will be referred as quenching temperatures. Another set of three plates were allowed to be heated 30 minutes at 450˚C, 600˚C and 750˚C respectively. Finally, a third set was 3 plates were heated 60 minutes at 450˚C, 600˚C and 750˚C. A reference plate remains without thermal treatment.

The reference and the 10 plates, once cooled, were covered by an epoxy resin leaving a section of 64 mm² for photo-electrochemical experiments.

The cell consisting of three electrodes, the photoelectrode as a working electrode (WE) with Pt electrodes symmetrically arranged as counter electrodes and a reference of Hg/HgO were employed in all measurements.

All experiments were carried out at 25˚C. A 6 M KOH (p.a.) aqueous solution was the electrochemical media. Measurements under light irradiation were carried out employing a dichroic lamp of 250 W (Zurich MR16) settled at 16 cm of the photoelectrode surface.

The potential of the photoelectrodes in darkness and under irradiation as well as the output photocurrent is listed in Table 1, where the results in darkness and under irradiation are shown. Experimental conditions can be read in Table 1 legend.

As seen in this Table, excluding the reference, the response of the photoelectrodes depend on the quenching temperature and the time of heating of the material at a given temperature. Even though the behavior is strongly dependent how the material was handled (temperature quenching and time of heating at a given temperature during 30 and 60 minutes), the best photochemical answer is observed in the material prepared at 750˚C. This behavior is consistent with previous results obtained in our laboratory [13] .

An interesting behavior can be observed in Table 1. In fact, the potential in darkness depends dramatically on the temperature program applied to the material whereas the potential measured under light irradiation is almost independent on the quenching temperature applied on the same material. The thermal treatment induces morphological changes on the surface [4] -[6] , being these changes responsible of the voltage measured in darkness, but the similar values observed under irradiation can be attributed to the TiO₂, which is the photo active material.

Moreover, the largest output photocurrent is obtained from the material prepared at 750˚C. Except for the reference and the photoelectrode prepared at 900˚C; these results are reproducible, even though the measurements were carried out several days after their preparation.

The behavior of these magnitudes show that the photoelectrode quenched at 750˚C presents the optimal photocurrent. These results can be correlated with the optimal thickness and the ratio of the anatase/rutile crystalline forms on the films as seen by X-ray diffraction and SEM studies reported in earlier studies [5] .

A small current intensity (1 mA) is recorded in the blank and the photoelectrode prepared at 900˚C. However, these values decays to negligible values (<1 µA) in a few minutes. Some instability are observed which can be associated to morphological changes on the electrodes, i.e., if the mirror-like Ti electrode is dipped alone into the solution, there are visual changes in the reflectance of the surface, but when is connected to the Pt counter electrode, the visual appearance of the surface changes, and a dark-grayish tonality is developed. This behavior is not observed on the other photoelectrodes, which show stability and reproducibility as reported in the literature [5] .

The thickness or amount of the TiO₂ on the surfaces should increase regularly with the time of heating leading to a better performance of the photoelectrochemical behavior. At 750˚C, the photocurrent is the highest observed in our experiments, but it decreases by a factor 3 to 4 with the heating time of this photoelectrode.

Nyquist diagrams obtained on the photoelectrodes prepared at 750˚C are shown in Figure 2. As seen in these plots, at higher frequencies the typical loops are observed. At lower frequencies a difusional behavior can be observed but not depicted in these Figures, because we are only interested in the parameters associated with the charge transfer processes. The difusional contribution is only observed in the photoelectrode prepared at 750˚C which presents a lower charge transfer resistance in comparison with the other electrodes listed in Table 2.

Nyquist diagrams shown dramatic differences when the measurements are carried out in darkness and under illumination. A significant decreasing the charge transfer resistance in the illuminated photoelectrode is recorded, as consequence of the catalytic effect of the TiO₂ under irradiation.

In Table 2 are resumed the experimental results measured in these set of experiments. All the R_ct listed this Table show an important decrease when the photoelectrodes are irradiated. These parameters are temperature dependent as can be observed for all photoelectrodes analyzed in this study. Even though the photocatalytic effect of light is observed in all samples, the effect is more important in the photoelectrode prepared at 750˚C.

As seen in Table 1, the photocurrent output increases with the temperature, being higher on the photoelectrodes prepared at 750˚C, meanwhile, as observed in Table 2, the R_ct value is the most small. As can be seen the Table 1, the parameter n » 1 suggests that the model depicted in Figure 1 fits reasonably well the experimental results.

Plots of versus the applied potential V are shown in Figure 3. In all cases V was measured against a Hg/HgO electrode. The slopes in all cases are positive as expected for a semiconductor type n [11] .

The slopes and the intercepts of MS plots are also temperature dependent as seen in these Figures. New sets of photoelectrodes were prepared to perform these experiments as described in section 2.1.

As usual, experiments were performed in a wide range of frequencies, but those showing a reasonable electrochemical behavior were selected as relevant in the construction of MS diagrams.

Experimental results are resumed in Table 3. It can be observed that the donor density (N_d) decrease as the temperature increases. The comparison between those experimental results listed in Table 1 are not parallel as expected in a simplified analysis, i.e. larger N_d values should imply an increase in the current output (I_light).

The flatband potentials do not show a significant difference below 750˚C, but at this temperature a much higher value is measured.

The information presented in Table 3 can be correlated to the film thickness (w) of the space charge layer (SCL) as well as to structural changes on the TiO₂. The film thickens of the SLC can be estimated by Equation (3), as proposed in the literature [12] , assuming that the SLC is fully depleted at high anodic potentials, where the layer behaves as a capacitor. This assumption allows us to correlate its capacity with the e_r of the TiO₂.

V_int is the potential in the erfase between the support and the TiO₂. A rough estimation can be obtained is this value is assumed smaller than V_int, as observed on other systems [14] .

The simplified expression leads to an approximate linear dependence between w and. Assuming that the geometric area of the photoelectrode is constant, the increase of the TiO₂ films should be proportional to the relative increase of the TiO₂ mass with the quenching temperature (see Table 3).

This proportionality between these variables allows us to estimate the dielectric constant of the film, assuming that the natural thickness of the film is of the order 10 - 20 nm. Therefore, taking a value of 15 nm, an applying the Equation (3) assuming that, a value of e_r = 230 is obtained. Weber et al. [13] reported values of 300 - 1000 for ε_r in defect-doped TiO₂. In Table 3 are shown the estimated values of the film thickness. It can

be seen that the donor density decreases with the thickness as observed in by R. van de Krol et al., [15] . The donor density behaves in a similar way as the value reported by Spagnol et al. working on TiO₂ nanotubes (on stainless steel) i.e., at 145 nm, a value of N_d = 1.4 × 10¹⁸ cm⁻³ is reported.

Studies performed on films of anatase on indium-tin oxide showed that the donor density decreases with the thickness. Values of 2.9 × 10¹⁹, 7.9 × 10¹⁸ and 4.1 × 10¹⁶ cm⁻³ have been evaluated for thickness of 40, 80 and 120 nm respectively (See reference [15] ), employing a dielectric constant of 55 for anatase.