Thin film electrodes of nominal compositions Ir_0.3Ti_0.4Sn_0.3O₂, Ir_0.2Ti_0.3Sn_0.5O₂ and Ru_0.3Ti_0.4Sn_0.3O₂ were prepared by the thermal decomposition of a polymeric precursor solution (DPP) [11] . This method consists in synthesizing resins of metallic precursors by mixing citric acid (CA) in ethylene glycol (EG). The Ru, Ir, Sn, and Ti resins were prepared separately. First, 8 g of citric acid (Merk) were dissolved in 9 mL ethylene glycol (Merk) at 60˚C - 65˚C. After the dissolution of the acid, a solution of the precursor metal in isopropanol with 0.1 mol∙L⁻¹ concentration (RuCl₃∙xH₂O, IrCl₃∙xH₂O, TiCl₂∙6H₂O all purchased from Aldrich and C₆H₅O₇Sn₂ synthesized from SnCl₂ (Aldrich), as described in [17] ), was added to the CA/EG solution. The temperature was then raised up to 85˚C - 90˚C and the solution under was kept under rigorous stirring (300 rpm) for 1 - 2 hours for esterification and total isopropanol evaporation.

The precursor solutions were deposited on both sides of the pretreated metallic titanium (2.5 × 2.5 cm) by brushing, as described in the literature [12] . After the application of the coating, the electrodes were dried at 130˚C for 5 minutes and then calcined at 450˚C for 5 minutes. This procedure was repeated until the desired mass (125 mg∙cm⁻²) had been achieved. The layers were finally annealed at 450˚C for 1 hour under air flow.

This measurement and others are deliberate, using specifications that anticipate your paper as one part of the entire journals, and not as an independent document. Please do not revise any of the current designations. The crystalline structures were physically characterized by X-ray diffraction (XRD) using an XRD-6000 diffractometer (Shimadzu) with a CuKα radiation source (λ = 1.5406 Å) operating in the continuous scan mode (4˚ min⁻¹) from 10˚ to 90˚.

The surface morphology and microstructure of the deposited oxide films were analyzed through optical microscopy and scanning electron microscopy (SEM). Photomicrographs were obtained by a Zeiss LEO model 440 SEM coupled to an OXFORD operating with electron beam of 15 kV. The average composition was analyzed by PGT PRISM energy dispersive X-ray spectrometer (EDX) coupled to the SEM instrument.

Electrochemical experiments were conducted with AUTOLAB model PGSTAT30 instrumentation. Voltammetric curves were recorded at scan rate of 50 mV∙s⁻¹ using 0.5 mol∙dm⁻³ of H₂SO₄ as the supporting electrolyte. A platinum foil served as the auxiliary electrode and the KCl saturated calomel electrode (SCE) was used as the reference. The cell was thermostated at 25˚C.

Impedance spectra were recorded at constant potential between 0.3 and 1.4 V vs Ag/ AgCl. Electrochemical impedance spectroscopy (EIS) measurements were obtained in the 5 mHz - 10 kHz frequency interval using the “single sine” method and a sine wave amplitude of 5 mV (p/p). An AUTOLAB software program (FRA analyzer) was used for the analysis of the impedance data.

The stability of the electrodes was assessed based on their lifetime (LT) under galvanostatic conditions at a high current density (400 mA∙cm⁻²) in 0.5 mol∙dm⁻³ of H₂SO₄. The electrode lifetime was considered the time necessary for the electrode potential to achieve a value of 8.0 V.

Figure 1 shows the XRD patterns for different compositions of electrodes prepared at 450˚C. In the electrodes containing iridium, characteristic diffraction peaks were

observed and attributed to IrO₂, according to JCPDS PDF #15-0870. Ti/Ru_0.3Ti_0.4Sn_0.3O₂ showed two peaks, one at 2θ = 43.7˚ and another at 2θ = 53.6˚, which correspond to RuO₂ JCPDS PDF #40-1290. However, by comparing the positions of the peaks in the XRD obtained with the respective pure oxide, it is possible to observe that Sn-rich electrode composition displays peaks shifted toward the pure SnO₂ pattern JCPDS PDF #46-1088, indicating that Ir and/or Ti atoms may be incorporated into the SnO₂ crystalline reticule. The opposite trend is observed for Ru-electrode composition, which have their peaks shifted toward the pure RuO₂ pattern due to the incorporation of Sn and/or Ti atoms into the RuO₂ crystalline reticule. All samples displayed typical crystalline characteristics of tetragonal, with space group P42/nm. All those evidences suggesting the formation of a saturated solid solution for all the compositions investigated, one for Ir/Ti/Sn and other for Ru/Ti/Sn compositions. The XRD patterns show that all materials synthesized contained the Ti phase attributed to the Ti metallic subtract.

The average crystallite sizes of the oxide particles were estimated by the Debye- Scherrer equation [18] at crystalline planes ((210): 2θ = 43.7˚) for Ti/Ru_0.3Ti_0.4Sn_0.3O₂, and ((101): 2θ = 34.5˚) for Ti/Ir_0.2Ti_0.3Sn_0.5O₂ and Ti/Ir_0.3Ti_0.4Sn_0.3O₂. The values obtained were approximately 4 and 5 nm, respectively.

The roughness of the oxides Table 1 was estimated from the optical microscopy analyses Figure 2 performed in random areas of the film (average of five analyses). The electrode with nominal composition of Ti/Ru_0.3Ti_0.4Sn_0.3O₂ showed the lowest roughness, which suggests that the morphology of the oxide layers is highly dependent on the physicochemical properties of the oxides and the nature of the precursors.

Figure 3 shows some representative SEM images of the oxide films. Films containing

IrO₂ (a, b) show uniform and continuous structures with cracks, i.e., mud-cracked-type morphology which are typical of thermally prepared oxide layers [10] [12] . Moreover, one observe that due to the increase in the amount of SnO₂ in the electrode composition, cracks become larger (b), however the surface becomes less rough (see Table 1). However, the SEM image of the films containing RuO₂ (c) indicate a distinct morphology, and in this case, the morphology change severally where the amount of fissures and cracks increase. The oxide surface morphology shows a clear relationship with the coating compositions investigated.

Table 2 shows the EDX analyses of the micrographs Figure 3. The EDX analysis of the electrodes indicated a good correlation between experimental and nominal compositions. The control of the composition of the films can be explained by the method used, since this polymer is formed before the calcination and the metal atoms are trapped in the matrix, which hinders its evaporation and consequent loss. All electrodes exhibited a homogenous distribution of particles on the electrode surface.

Figure 4 shows the j/E curve obtained in the cyclic voltammetric experiments. This profile is typical of thermally prepared oxide layer electrodes [19] [20] and characteristic of DSA® electrodes [21] . The figure also shows a blurred peak at around 0.5 V associated with the Ru (III)/Ru(IV) redox transition [22] for the Ti/Ru_0.3Ti_0.4Sn_0.3O₂ electrode. The voltammograms of the electrodes containing IrO₂ showed a peak typical of the Ir(III)/Ir (IV) transition in the region between 0.4 and 0.8 V [6] .

The oxygen evolution reaction occurs at a more positive potential for the electrode containing the largest amount of SnO₂. According to Fukunaga et al. [23] , the doping of the electrode surface with SnO₂ is an effective strategy to improve performance even in

the treatment of degradation of organic compounds thus our results are in agreement with previous report.

The comparison between the electrode containing IrO₂ and those containing RuO₂ shows the IrO₂ exhibits lower activity for the oxygen evolution reaction.

The impedance behavior of the electrodes was investigated to further characterize the different Ti/Ir_0.3Ti_0.4Sn_0.3O₂; Ti/Ir_0.2Ti_0.3Sn_0.5O₂ and Ti/Ru_0.3Ti_0.4Sn_0.3O₂ systems. The Nyquist diagram (Z' vs Z") of the electrodes obtained between 0.3 and 1.4 V vs Ag/AgCl is shown in Figure 5. In the low frequency domain, electrodes Ti/Ir_0.3Ti_0.4Sn_0.3O₂ and Ti/Ir_0.2Ti_0.3Sn_0.5O₂ formed a straight line parallel to Z", characteristic of an ideally polarizable electrode, and a slight deviation from the straight line along Z", suggesting a non-ideally polarizable electrode [24] [25] . The shift from the ideal capacitor behavior is a consequence of the material’s porosity [24] [26] . In the low-frequency domain region a decrease in impedance was found when the applied potentials were positively shifted. This result suggests that the EIS responses in this domain region indicate a faradaic process of the bulk redox transitions of the oxide material. The difference observed between the Ir-based electrodes and Ru-based electrodes maybe could be explained due to the higher electronic conductivity in the Ru-Ti-Sn/solution than in the Ir-Ti-Sn/solution interfaces.

The Bode plot (θ vs. log f) obtained at 0.3 V vs Ag/AgCl for electrodes is shown in Figure 6. An analysis of this figure indicates that the main feature of these electrodes is the appearance of a well-defined time constant (τ) for the Ti/Ru_0.3Ti_0.4Sn_0.3O₂ electrode, which is characterized by a maximum phase angle ranging from 1 to 100 Hz. This behavior can be ascribed to the large number of RuO₂ transition states contributing to the charging system [27] . This results corroborated with the Nyquist plot interpretation above because the Ru-based electrode shows more pseudo capacitive behavior than the Ir-based electrode [10] [24] .

The stability of the electrodes was assessed based on their lifetime, considering the time necessary for the electrode potential to reach 8.0 V. The electrode containing larger amounts of SnO₂ (Ti/Ir_0.2Ti_0.3Sn_0.5O₂) has a shorter lifetime. The 20% decrease in the stoichiometric amount of this oxide as well as the high amount (10%) of IrO₂ (Ti/ Ir_0.3Ti_0.4Sn_0.3O₂) increase the time to a value above 70 hours. Comparing the electrode Ti/Ir_0.3Ti_0.4Sn_0.3O₂ with Ti/Ru_0.3Ti_0.4Sn_0.3O₂, IrO₂ has higher stability under drastic conditions of electrolysis than RuO₂ (Table 3).

The lifetime of oxide electrodes is directly correlated with two factors: passivation and dissolution of the coating [21] . The first factor is due to the penetration of the electrolyte through the pores or cracks towards the substrate, resulting in the oxidation of the metallic support and forming a non-conductive layer between the substrate and the oxide coating [28] - [30] . The second factor involves the loss of electroactive material (erosion or dissolution), resulting in a gradual reduction of the voltammetric charge. This may occur due to the pores in the layer and the rapid evolution of gas on the surface, inducing the separation of weakly bound parts of the active layer [28] [31] [32] .

Morphological changes of the electrode surface after the lifetime test can be observed

through microstructural analysis (Figure 7), which shows worn structures with erosion of the active layer. The EDX analysis revealed a decrease in the quantity of Ir and Ru, confirming the loss of the electroactive material, well as a decrease of Sn (Table 4).

The curves obtained for the lifetime showed a slow increase in the potential followed by an abrupt increase at the end of the experiment for all compositions investigated. This behavior indicates a rise in the electrode structure resistance. Such an increase may have resulted from the loss of Ir or Ru in the top layers of the electrode and/or the formation and growth of a non-conductive oxide film between the metallic substrate and the conductive oxide [9] [31] .

EDX analysis after lifetime revealed a considerable increase in the titanium signal. These results suggest that besides the process of erosion, there is also a process of anodic passivation of the metallic base due to the formation of an insulating film composed primarily of TiO_x.