Você está na página 1de 8

Journal of The Electrochemical Society, 160 (6) C277-C284 (2013)

0013-4651/2013/160(6)/C277/8/$31.00 The Electrochemical Society

C277

Ti Anodization in Alkaline Electrolyte: The Relationship between Transport of Defects, Film Hydration and Composition
a Jorge Vazquez-Arenas,a Rom Pr ospero Acevedo-Pena, an Cabrera-Sierra,b, c a, ,z Luis Lartundo-Rojas, and Ignacio Gonz alez
a Universidad Aut onoma Metropolitana, Departamento de Qu mica, Laboratorio de Electroqu mica, CP 09340 DF, M exico b Instituto Polit ecnico Nacional, Escuela Superior de Ingenier a Qu mica e Industrias Extractivas, Departamento de Ingenier a Qu mica Industrial, CP 07738 DF, M exico c Instituto Polit ecnico Nacional, Centro de Nanociencias y Micro y Nanotecnolog as, UPALM, Zacatenco, M exico-D.F. CP 07738, M exico

The hydration state and structural transformations of Ti lms potentiostatically grown in 0.1 M NaOH are characterized utilizing EIS, Mott-Schottky and XPS techniques. Variations presented in the density of donors, the atband potential and the fraction of species measured by XPS, in the region of characterization of these lms are associated with changes in the ratio of different Ti sub-oxides within the passive lm. Oxygen and hydroxyl vacancies and their respective formation rate constants are estimated from EIS spectra ts with a modied Point Defect Model. A consistent behavior is observed for the hydration state and structural transformations of the lms as a function of potential. Films anodized at less positive potentials than 1.17 V vs SCE present higher hydroxyl vacancy diffusivities (e.g higher lm hydration) and a heterogeneous mixture of titanium sub-oxides, whereas the opposite behavior is observed for lms formed at more positive potentials than 1.17 V vs SCE, where the hydroxyl vacancy diffusivities decrease by one order of magnitude and the oxygen vacancy diffusivities increase three times. Thus, lms anodized at more positive potentials present a higher dehydration, are more homogeneous and mainly composed of TiO2 . 2013 The Electrochemical Society. [DOI: 10.1149/2.063306jes] All rights reserved. Manuscript submitted January 15, 2013; revised manuscript received March 12, 2013. Published April 5, 2013. This was Paper 2252 presented at the Honolulu, Hawaii, Meeting of the Society, October 712, 2012.

Titanium anodization consists of the formation of an oxide layer on a Ti substrate by electrolytic polarization. The oxides formed through this methodology have become attractive due to the paramount performance exhibited by semiconductor TiO2 lms in photocatalytic and photoelectrochemical applications, biomaterials and corrosion resistance.14 Indeed, the features and performance of these lms associated with the aforementioned applications can be directly related to the electronic properties and crystal structure of these materials. For instance, the surface coloration, electrocatalytic activity, corrosion resistance, light refraction and absorption of the lms rely upon the properties and crystal structure of the material.58 This anodization could entail the transition through multiple intermediary oxidation states (TiO, Ti2 O3 , Ti3 O5 and/or Ti4 O7 ).931 The proportion of these oxides within the anodic lms varies depending on the conditions utilized to anodize, e.g. applied potential, time, electrolyte composition. This transition also affects the crystallinity and photoelectrochemical performance of TiO2 photoanodes, as well as their corrosion resistance.32,33 Molar fraction of chemical species vs E diagrams for the Ti-H2 O system reveals that the formation of Ti sub-oxides could be connected to the hydration state of the oxide lm, since the stability of the TiO2 is displaced to more positive potentials when hydrated oxides are involved in the thermodynamic calculations.34 Different characterization techniques have been used to study the formation of these intermediary oxides in different electrolytes.1012,15,18,19,21,2325,30,32,34 However, to date few studies have correlated the formation of these sub-oxides with their electrochemical response.9,20,27 Our research group has developed a model to account for the reaction mechanism of semiconductive oxides subjected to the transport of point defects across the lm.35 The model is based on the Point Deffect Model (PDM) formalism proposed by Macdonald et al.36,37 and it has been modied to consider the diffusion of OH ions across the lm (e.g. hydroxyl vacancies). The importance of this input arises from the partial hydration of the lm, which inuence the properties of the oxide, producing oxyhydroxide and/or hydroxide phases.19,21,23,3840 However, the mechanistic analysis of the passivity of titanium oxides has not been used to comprehend the variation in the semiconducting properties of these lms or the formation of different
z

sub-oxides, which are commonly detected by other characterization techniques.1012,15,18,19,21,2325,30,32,34 Thus, the purpose of this study is to extend our comprehension of the reaction mechanism operating during Ti anodization in alkaline electrolytes, as well as the existing relationship between point defects, lm hydration and electronic properties. Experimental The experiments were carried out in a typical three-electrode cell. The working electrode was a Ti bar (AlfaAesar, 99.95% purity) with a diameter of 6 mm embedded in Teon. The working electrode was systematically polished to a mirror nish before each experiment.27,41 A saturated calomel electrode (SCE) was used as reference, all potentials along the text are reported versus this scale. The counter electrode was a graphite bar (99.999% AlfaAesar). The electrolyte was prepared using NaOH (JT Baker, 97% purity) and ultrapure water from a Millipore (18.2 M cm) system. The oxide lms were formed on the Ti electrode by imposing different formation potentials, EF , (0.74, 0.8, 0.92, 1.17, 1.67 and 2.17 V vs SCE) for 3.0 hours. These potential values were selected according with the voltammetric characterization. Electrochemical Impedance Spectroscopy was performed at 30-minute intervals after the rst half hour of formation utilizing the following parameters: i) the impedance spectrum was measured at the same potential imposed during lm growth, ii) the amplitude of the perturbation peak to peak was 10 mV and iii) the frequency was scanned from 10 kHz to 10 mHz. The semiconductor properties of the lms were estimated employing Mott-Schottky (M-S) plots at scan rates of 200 mV s1 . This measurement was carried out after lm growth, and consisted of scanning the potential from EF to the open circuit potential (0.77 V). The amplitude of the AC signal used in this technique was also 10 mV and the frequency was 1 kHz.42,43 Angle Resolved X-ray Photoelectron Spectroscopy (AR-XPS) was performed using Thermo Scientic K-Alpha X-ray photoelectron spectrometer with a monochromatized AlK X-ray source (1487 eV). The different angular analyzes were performed over a 060 angular range, which corresponds to photoelectrons take-off angles in the 90o 30 range, relative to the detector. Therefore, the photoelectrons take-off angle remains constant with respect to the hemispherical an-

Electrochemical Society Active Member. E-mail: igm@xanum.uam.mx

Downloaded on 2013-05-02 to IP 148.206.55.254 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

C278

Journal of The Electrochemical Society, 160 (6) C277-C284 (2013) which undergo different dehydration stages since they are immersed in aqueous solutions.19,21,23,3840 Therefore, it would be more appropriate to build a Pourbaix-type diagram for this system considering hydrated Ti oxides. This idea is supported by XPS measurements where the formation of hydrated oxides was found during the early stages of anodization.19,21,23 Figure 1b shows a Pourbaix-type diagram calculated for the system Ti-H2 O, using a Ti2+ concentration equals to 10 M, and considering the equilibrium constants reported for hydrated Ti oxide species (Ti2 O3 and TiO2 ).34 As observed in this gure, the stability region of hydrated TiO2 is mainly delimited by the presence of the hydrated Ti2 O3 phase (negative region) and TiO3 2H2 O (positive region) species. On the other hand, the zones where TiO and hydrated Ti2 O3 are thermodynamically stable increase, whereby their stability zones are displaced to less negative potentials. Additionally, Figures 1c and 1d shows the Pourbaix-type diagrams obtained where species Ti3 O5 (Ti2 O3 TiO2 ) or Ti4 O7 (Ti2 O3 2TiO2 ), are respectively

alyzer. The position of the O 1s peak at 531.0 eV was monitored on each sample to ensure that no binding energy shift due to charging had occurred. Narrow scans were collected at 60 eV analyzer pass energy and a 400 m X-ray spot size. Thermodynamic Analysis of the System Ti-H2 O The Pourbaix-type diagram is a powerful tool to evaluate thermodynamic information concerning the stability regions of the predominant species of aqueous systems. They are generated through the graphic representation of all possible chemical and electrochemical reactions among these predominant species. A typical Pourbaix-type diagram calculated for the system Ti-H2 O (Figure 1a) without considering hydrated species shows that only TiO2 predominates in the predominance region of water.13,25,33 More negative potentials are characterized for the appearance of Ti2 O3 and TiO. However, it is known that anodic lms entail the formation of hydrated oxide lms

Figure 1. Pourbaix-type diagrams of Ti-H2 O system constructed considering a concentration of dissolved species equals to 10 M, at 25 C. (a) Typical Ti-H2 O Pourbaix-type diagram, (b) considering hydrated species (TiO2 and Ti2 O3 ), and overlapping the metastable Pourbaix-type diagrams obtained when (c) Ti3 O5 or (d) Ti4 O7 are considered instead of hydrated Ti2 O3 . The stability region of water is showed in dashed lines.

Downloaded on 2013-05-02 to IP 148.206.55.254 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

Journal of The Electrochemical Society, 160 (6) C277-C284 (2013) considered instead of the hydrated Ti2 O3 phase. These species show a large zone of predominance comprising part of the region dominated by the hydrated TiO2 . A consistent behavior is observed in this gure at constant pH value, a progressive increase in the oxygen contained in the oxide (e.g. stoichiometric) is obtained as the potential becomes more positive. Also note that most of the predominance zones for the hydrated Ti oxides are situated under neutral or alkaline conditions unless the potential in the system becomes more positive. This observation draws attention to the formation of more stable Ti oxides under alkaline conditions. Thus, the chemical species described in the Pourbaix-type diagrams shown in Figure 1 could provide insights to analyze the mechanism controlling the Ti anodization in aqueous solution. Results and Discussion Voltammetric characterization and lm growth. Figure 2a shows a typical voltammogram obtained with a Ti substrate in solutions containing 0.1 M NaOH. The potential was scanned at 10 mVs1 from the Open Circuit Potential (OCP) to more positive values (i.e. forward scan), and then switched back in the negative direction (backward scan). As observed in Figure 2a, a steep rise in current is registered during the early stages of the scan, which could be attributed to the nucleation of the Ti oxides on the surface of the electrode. From potential values 0.30 to 0.90 V, a plateau is observed in current density, followed by the formation of a shoulder which forms a new plateau at the end of the voltammogram. Along the forward scan, the current density becomes slightly higher as the potential scan proceeds, likely due to the water oxidation on the electrode surface.20,22,27 As reported for oxide growth on valve metals, the current density dropped abruptly during the backward scan.4245

C279

The shoulder presented in the voltammogram shown in Figure 2a has been previously reported for Ti and Ti alloys.13,15,16,18,21,24,25,27,29,46,47 However, its formation is still a matter of discussion since it has been associated with different processes: oxygen evolution, rupture and formation of the lm and the formation of different intermediate oxides in the TiO2 matrix, e.g. Tix Oy (where 1 y/x < 2). So far, the formation of intermediate oxides is the most accepted proposal since it has been supported by XRD and XPS studies.15,18,21,24,25 Thus, the presence of different plateaus in Figure 2a might be related to changes in the lm properties as a result of the stoichiometry of the oxide lms.9,20 In order to investigate the processes occurring in this shoulder, Ti anodic oxide lms were potentiostatically grown for 3 h at different formation potentials, EF (0.74, 0.8, 0.92, 1.17, 1.67 and 2.17 V vs SCE, indicated with arrows in Figure 2a). Figure 2b shows current transients, where the abrupt decrease in current as a function of time has been related to the formation of a passive titanium oxide on the substrate. A constant current was observed at later stages due to the pseudo stationary growth attained for the oxide lm. In the frame work of the PDM model, this indicates a constant ux of vacancies controlling the phenomena involved in the formation of the passive lms. Thus, the current of this process is made up of the generation/annihilation of vacancies of Ti4+ , O2 and OH , and their respective uxes across the Ti oxide lms (e.g. O2 and OH vacancies dominate the transport for n-type oxides such as TiO2 ). Figure 2b shows the effect of EF in the steady state current (iss ) associated with the growth of anodic Ti lms. At less positive EF than 1.17 V, the iss values increase as the potential is made more positive, suggesting that the transport of either O2 or OH vacancies is increased across the lm. On the other hand, the stabilization of the iss values could be related to a constant diffusion of these vacancies at more positive potentials. In addition to the transport of these defects, there are faradaic processes occurring within the oxide, which could lead to the transformation of different sub-oxides (Tix Oy with 1 y/x < 2), mainly favored at less positive potentials. The faradaic current due to the oxygen evolution was discarded since most likely this input arises at potentials more positive than 3.5 V (Figure 2b). This is corroborated by observing the steady state currents obtained from the chronoamperometric measurements, which attain a current plateau at formation potentials from 1.17 V to 2.17 V (Figure 2b), while the current continues increasing for the water oxidation region (non shown). Semiconducting properties. The semiconducting properties of anodic lms grown potentiostatically on Ti are very sensitive to changes in lm composition, oxide doping and the combination of different oxides.46,47,49 M-S plots are a powerful tool to assess the semiconductive properties of Ti anodic lms, estimating their atband potential (EFB ) and density of donors (Nd ). Differential capacitance measurements were performed on Ti oxide lms previously grown for 3 h. The space charge capacitance, CSC , was calculated using the following equation: 1 1 1 = + C CSC CH [1]

where C is the measured capacitance, comprising the space charge capacitance (CSC ) in series with the Helmholtz capacitance (CH ). CH is assumed constant, with value of 20 F cm2 .50 M-S plots presented in Figure 3a show that the CSC 2 values increase as EF becomes more positive. A linear behavior with a positive slope is obtained for these plots in the potential region from 0.77 V to 0.40 V, which indicates an n-type semiconductor behavior.42,43,46,47,49 Thus, Nd and EFB can be calculated as follows:42,43,46,47,49 1 2NA RT = Em E f b Csc 2 Nd F r 0 F [2]

Figure 2. in 0.1 M NaOH. (b) Effect of the formation potential (EF ) of anodic Ti lms in the current transients recorded during three hours in 0.1 M NaOH. The variation of the steady state current (iss ) is indicated in this gure.

(a) Typical voltammetric behavior (v = 20 mV s1 ) on a Ti electrode

where NA is the Avogrados number (6.02 1023 mol1 ), Nd is the donors density (cm3 ), F is the Faradays constant (9.65 104 C mol1 ), r is the relative permittivity, assumed to be 38,27 0 is the

Downloaded on 2013-05-02 to IP 148.206.55.254 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

C280

Journal of The Electrochemical Society, 160 (6) C277-C284 (2013) the density of donors is observed as the oxide lms are formed at more positive potentials, indicating modications in the oxide structure. The presence of these donors could be associated with different Ti species (Ti+2 , Ti+3 and Ti+4 ) or oxygen defects according to the PDM prediction criteria for n-type semiconductors. In both cases, the formation of metallic oxides relies upon the mobility of these donors, whose effects are more important for EF < 1.2 V vs SCE. For other valve metals, the atband potential (EFB ) is practically independent of EF 42,43 The dependence of EFB on potential (Figure 3b) could be related to the presence and variation of oxygen vacancies and hydroxyl ions that involve the formation of Ti suboxides (i.e. TiO, Ti2 O3 , TiO2 ) within the lm.11,48,51 This is suggested since the Ti3+ species formed from oxygen vacancies or insertion of hydroxyl anions in the TiO2 lattice5256 induce different energetic states inside the TiO2 bad-gap.55,56 Di Valentin et al.55 have shown using density functional theory calculations that OH insertion in the TiO2 lattice generates energetic states 0.4 eV under the conduction band of the TiO2 and oxygen vacancies generate energetic states 0.7 eV under the conduction band. Then, the variation of the atband potential in a less negative direction agrees with the hypothesis that lm dehydration could take place in this potential range, which varies the proportion of Ti-suboxides present in the anodic lm. AR-XPS measurements were conducted to analyze the proposal concerning the formation of different titanium sub-oxides in the anodic lm. Angle resolved X-Ray photoelectron spectroscopy. Three different anodic lms were potentiostatically grown at 0.74, 1.17 and 2.17 V during 3 h, in order to evaluate the presence of different Ti species by AR-XPS. This technique allows obtaining information from the inner and external layers of the lm, whence it is an accurate manner to identify variations associated with lm composition. To this concern, Figure 4a shows Ti 2p doublet (Ti 2p1/2 and Ti 2p3/2 ) high resolution AR-XPS spectra collected at different photoelectron takeoff angles (formed between the sample normal and the detector) from the lm growth at 0.74 V. For all spectra, titanium chemical species contained in the anodic lm were qualitatively analyzed using deconvolution or decomposition modeling of spectral signals, through the comparison with characteristic spectra reported in the literature.19,21,23 Figure 4b shows a typical deconvolution of an XPS spectrum

Figure 3. (a) Mott-Schottky plots of Ti oxides growth at different potentials in 0.1 M NaOH. (b) Effect of the formation potential over the semiconductive properties (density of donors Nd , and atband potential EFB ) of Ti oxides formed during three hours in 0.1 M NaOH.

vacuum permittivity (8.8542 1012 F m1 ), Em is the measuring potential, EFB is the atband potential, R is the gas constant (8.314 J K1 mol1 ) and T is the temperature (298 K). In equation 2, the last term is assumed to be negligible at room temperature. Figure 3b shows the variation of EFB and Nd values with formation potential, calculated with equation 2 and the M-S plots. A decrease in

Figure 4. (a) Ti2p doublet high resolution AR-XPS spectra collected at different photoelectron take-off angles for the anodic oxide lm growth at 0.74 V during 3 h. (b) Typical XPS spectrum deconvolution for the anodic oxide lm grown at 0.74 V during 3 h, and measured using a take-off angle of 0 . Dependence of the fraction of Ti4+ (TiO2 ), Ti3+ (Ti2 O3 ) and Ti2+ (TiO) with the measuring angle, depth concentration prole, for the anodic lms growth at (c) 0.74 V, (d) 1.17 V and (e) 2.17 V.

Downloaded on 2013-05-02 to IP 148.206.55.254 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

Journal of The Electrochemical Society, 160 (6) C277-C284 (2013)

C281

Figure 5. Nyquist diagrams obtained for Ti oxide lms at different EF (indicated in the gure) in 0.1 M NaOH. The continuous lines show the quality of the spectra constructed with the modied PDM model and the best tting parameters to experimental data.

calculated using a Gaussian -Lorentzian mix function and Shirley background subtraction. The Ti 2p doublet spectra were tted with three double contributions attributed to: Ti4+ located at 458.0 and 464.4 eV, Ti 2p3/2 and Ti 2p1/2 peaks respectively. For Ti3+ , the Ti 2p3/2 peak is located at 457.7 eV and Ti 2p1/2 peak at 463.5 eV. The Ti 2p3/2 and Ti 2p1/2 contributions related to Ti2+ , are located at 456.4 and 462.2 eV, correspondingly. Relative Sensitivity Factors were used to scale the raw peak areas in order to calculate the fraction of each species, and thus, extract the depth concentration prole, Figure 4c4e. Figures 4c4e show the variation of the fraction of Ti4+ (TiO2 ), Ti3+ (Ti2 O3 ) and Ti2+ (TiO) species inside the lms grown at 0.74, 1.17 and 2.17 V, respectively. The oxide lm grown at 0.74 V (Figure 4c) is constituted by a heterogeneous mixture of different oxides (TiO2 , Ti2 O3 and TiO), showing the presence of two regions dominated by TiO2 at the anodic lm surface and Ti2 O3 at the titanium/anodic lm interface. In contrast, when the formation potential is 1.17 V (Figure 4d) and 2.17 V (Figure 4e), TiO2 dominates across the entire lm and a more homogeneous composition is observed. This variation could be related to the changes observed for the iss values (Figure 2b), since it represents the ux of vacancies through the anodic oxide. Consequently, at less positive potentials where the iss varies more signicantly with the formation potential, the lm composition is heterogeneous and a larger composition of other titanium sub-oxides is

obtained. However, when the formation potential becomes more positive and iss rises to attain a constant value, the ux of vacancies guarantees the homogeneity across the oxide lm. Note that the presence of titanium sub-oxides is still detected at more positive potentials as a result of the high density of defects (e.g. primarily oxygen vacancies). The aforementioned proposal supports the interpretation given to the anodic shoulder formed in the voltammogram in Figure 1a. However, further evidence is needed in order to explain the changes observed for the electrochemical behavior of the lms analyzed in this study. EIS Characterization. EIS measurements were carried out after growing the oxide lm for three hours. The Nyquist diagrams obtained from these measurements are shown in Figure 5. These diagrams show two different features regardless of the EF value used to grow the lm. At 0.74 V (Figure 5a), a attened semicircle is formed and higher values are obtained for the real component of the impedance in comparison with the imaginary component. As the potential becomes more positive (refer to Figure 5b and 5c), the protective character of the Ti oxide lms decreases, and is reected in the attenuation of both components of the impedance. At EF 1.17 V (Figures 5d5f), the resistance of the Ti lms increases and the lm newly becomes more protective. The components of the impedance are even higher to those

Downloaded on 2013-05-02 to IP 148.206.55.254 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

C282

Journal of The Electrochemical Society, 160 (6) C277-C284 (2013) the oxide lm, Figure 6a shows a linear relationship of 1/Cp as a function of EF . This behavior indicates that the growth ratio a is not modied, as reported for oxide lms grown on valve metals.18,42,43 Additionally, this behavior indicates that the changes observed in the electrochemical behavior of the lms (Figures 2, 3 and 5) are not due to the variation in the growth ratio of the anodic lm, but structural changes occurring within the oxide lm when the potential is varied. The estimation of a can be obtained by considering a typical Helmholtz capacitance of CH = 20 F/cm2 ,50 the relative permittivity of the oxide lm r = 38,27 and the vacuum permittivity 0 = 8.8542 1014 Fcm1 in the following eq.: 1 a b 1 EF + = Cb CH r 0 a [3]

recorded at the less positive potential. Presumably, this change in the resistive behavior of the Ti oxides is associated with the modication of the semiconductive properties, which could rely on the hydration and structure of the Ti lms.9,11,20,51 Therefore, it is suggested that more protective coatings could be obtained when the proportion of TiO2 is signicantly higher within the lm. This behavior ideally agrees with the decrease in the density of donors observed for the EF in Figure 3b. To this concern, Shvets et al. reported that passive lms composed of TiO2 and titanium sub-oxides (TiO and Ti2 O3 ) are less protective against corrosion than layers mostly composed of TiO2 in the SiC-TiB2 -B4 C system.33 As described in Figure 5, the changes observed in the EIS spectra associated with the increase of the iss values (refer to Figure 2b) and the shoulder obtained in the passive region of the voltammogram (Figure 2a) are consistent with each other. In order to determine the kinetic parameters of the phenomena controlling the formation of the Ti oxides, a modied version of the PDM model was tted to the experimental EIS spectra.35,44,45 This modication incorporates the transport of hydroxyl vacancies which has not been described by any model yet. Analysis of EIS spectra using the modied PDM model. Figure 5 shows the quality of the ts (continuous lines) of the modied PDM model to the experimental Nyquist diagrams (see Appendix for details of the model implementation). Note that the ts describe adequately the experimental data over the entire range of frequencies. The values of the lm capacitance Cp , the resistances R1,O 2 and R2,OH involving forward and backward rate constants for vacancies generation, the backward reaction constant for oxygen vacancies generation k-3a and hydroxyl vacancies generation k-3b , the diffusion coefcient for oxygen ( DV O ) and hydroxyl ( DV O H ) vacancies are plotted in Figure 6. The value of Cp decays as EF is made more positive (Figure 6a) as a result of the increase in the lm thickness. Although the voltammetric and M-S characterization indicates structural transformations across

where b represents the lm thickness at EF = 0 V. A value of 3.4 nm V1 was calculated for a, which is slightly higher than the values commonly reported for Ti anodic lms.25,5759 The resistances R1,O 2 and R2,OH increase as the formation potential becomes more positive. While R2,OH raises progressively, the changes in R1,O 2 are steeper, but both resistances attain the same value at the most positive potential (see Figure 6b). Presumably, the structural modications undergone by the Ti lms at less positive potentials are mainly due to resistances associated with the formation of the oxygen vacancies. These modications also involve variations in lm hydration. The resistances R1,O 2 and R2,OH are independent of frequency and inversely proportional to the forward and backward reaction constant for the oxygen vacancies generation and hydroxyl vacancies generation, respectively.35,44,45 The variation of the backward reaction constants for the oxygen (k3a ) and hydroxyl (k3b ) vacancies generation with EF is shown in Figure 6c. At 0.74 V, both reaction constants show a value approximately of 104 cm s1 , however, they behave distinctively when the formation potential becomes more positive; meanwhile, k3a remains constant, and k3b gradually decreases (Figure 6c).

Figure 6. Kinetic parameters obtained from the tting of the modied PDM to the experimental EIS spectra experimentally obtained for Ti oxide lms at different EF (indicated in the gure) in 0.1 M NaOH. (a) Film capacitance Cp , (b) the resistances R1,O 2 and R2,OH , (c) backward reaction constant for oxygen vacancies generation k3a , and hydroxyl vacancies generation k3b and (d) diffusion coefcient for oxygen (DVO ) and hydroxyl (DOH ) vacancies. The arrows indicate the position of the axis.

Downloaded on 2013-05-02 to IP 148.206.55.254 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

Journal of The Electrochemical Society, 160 (6) C277-C284 (2013) The effect of the formation potential on the diffusion coefcient for the oxygen vacancies ( DV O ) and hydroxyl vacancies ( DV O H ) is shown in Figure 6d. These coefcients are higher than those determined for other valve metals using a modied PDM model.35,44,45 DV O H decreases more than one order of magnitude as the EF becomes more positive, whereas DV O increases three times in the same region of potential. Note the important variation of the reaction constants (Figure 6c) as well as those related to transport defects (Figure 6d) at more positive potentials. Indeed, the variation of these variables could account for the formation of different titanium sub-oxides and conrm the results shown in previous sections. The results presented above offer signicant insights into the formation mechanism of titanium lms grown potentiostatically in 0.1 M NaOH aqueous solutions. Thus, the following summary can be proposed to describe the phenomena occurring during the different stages of formation. The variation of the kinetic parameters related to hydroxyl vacancies (R2,OH , k3b and DV O H ) results from the changes in lm hydration with the formation potential. The dehydration of Ti lms with time and potential has been previously conrmed by other researchers.19,21,23 Of course, this situation is accompanied with the transport of defects within the oxide lm, which leads to the formation of different sub-oxides, since the insertion of each OH ion in the oxide lattice stabilizes a Ti3+ species.55,56 The ratio existing between their formation/annihilation and diffusion rate results in different stochiometries for the titanium oxide lms. In this study, this ratio is controlled through the application of different formation potentials. Thus, the formation of these sub-oxides is enhanced at less positive potentials, as a result of the high rate to produce hydroxyl vacancies and high DV O H (Figure 6c and 6d). At more positive potentials, where the anodic shoulder was observed in the voltammetric characterization and a rise in the iss was obtained (Figure 2), a decrease in the hydroxyl ion insertion into the oxide was observed accompanied by an increase in the coefcient for the oxygen vacancies ( DV O ), Figure 6d. This suggests that more oxygen anions (O2 ) are being inserted in the oxide lattice instead of hydroxyl anions (OH ), thus driving the oxidation of the Ti3+ species to Ti4+ (Figure 4c and 4d). Additionally, the lm dehydration when the formation potential becomes more positive agrees with the variation of the atband potential (EFB ) to less negative values (Figure 3b), since OH ions inserted in the oxide lattice induce energetic states inside the bandgap of the grown oxide, which are closer to the conduction band than those induced by oxygen vacancies.55 Conclusions EIS, Mott-Schottky and AR-XPS techniques were used to evaluate and correlate the hydration state with the structural transformations of Ti anodic lms formed in 0.1 M NaOH solutions. Variations in the density of donors, the at band potential and the fraction of species measured by AR-XPS were related to changes in the ratio of different Ti sub-oxides. r A modication of the PDM model was utilized to t experimental EIS spectra and calculate the kinetic parameters describing the transport of point defects and the reaction mechanism of the lms. The oxide lms formed at less positive potentials than 1.17 V presented higher hydroxyl vacancies diffusivities reecting a higher hydration state. These lms were composed of a heterogeneous mixture of titanium sub-oxides. The opposite behavior was observed at more positive potentials than 1.17 V, where the hydroxyl vacancies diffusivities decreased by one order of magnitude and the oxygen vacancies diffusivities increased by three times, indicating a higher dehydration of the oxide lms. Conversely, these lms were more homogeneous and mainly composed by TiO2 . r The changes observed with the PDM analysis conrmed the changes in lm hydration, which agrees with the formation of the sub-oxides found with the semiconducting properties evaluation and the AR-XPS. This study demonstrates that the faradaic and nonfaradaic processes (e.g. hydration, dehydration) occurring during the

C283

Ti anodization in alkaline media can be accounted for the formation of point defects (oxygen and hydroxyl vacancies) at the metal-lm interface. Acknowledgments The authors are indebted to the CONACyT (Mexico) for their nancial support to carry out this work (Project CB-2008/105655). Pr ospero Acevedo-Pe na is grateful to CONACyT for the PhD fellowship granted through the program doctorados nacionales. Appendix
Previous publications of our research group have extensively dealt with a modication of the point defect model intended to account for hydration mechanisms of passive lms, involving the diffusion of OH ions through hydroxyl vacancies to form oxyhydroxide or hydroxide phases.35,44,45 A schematic description of the interface metal/lm/solution used to describe the Ti anodization in the present study is sketched in Figure A1. This gure represents the formation and annihilation processes of different types of defects subjected to diffusion across the passive lms utilizing a Kroger-Vink notation. These lms are modeled in this study considering that are comprised by an inner layer (oxide) poorly hydrated. The metal/lm interface does not involve an uniform metal/oxide interface because of the incorporation of some hydroxyl ions, and may entail the formation of sub-oxides. On the same hand, the inner layer is a highly defective semiconductor (either p-type or n-type), where the electronic conduction across it, is described by the transport of vacancies acting as electronic dopants. To this concern, on p-type oxide lms, the majority of carriers are cation vacancies (VM ) and for n-type (e.g. Ti oxides), the majority + ) and/or interstitial metallic ions ( Mi ).3537,44,45 of carriers are oxygen vacancies (VO For both cases (p-type and n-type), the model considers the formation and/or annihilation of hydroxyl vacancies (VO H , reaction 3b and 6b), at the metal/lm and lm/solution interfaces, respectively. Molecular hydrogen can be formed by a recombination process (i.e. Tafel mechanism) owing to diffusion of hydroxyl ions across the oxide lms, where mono-atomic hydrogen is trapped due to reduction reactions at the metal/lm interface. Fluxes of cation, oxygen and hydroxyl vacancies are described by the arrows direction indicated in Figure A1. Protons can migrate through the oxide lm via interstitial sites, but only under cathodic polarization. The probable occurrence of water oxidation on the surface of the lm has not been considered in the present model because TiO2 lms were grown at less positive potentials than the onset value for oxygen evolution. Moreover, the oxidation of water can be considered negligible because this passive oxide covers most of the titanium surface. The metal/lm interface does not involve a uniform metal/oxide interface because of the incorporation of some hydroxyl ions (e.g. reaction 6b). In the model, it is assumed that the total impedance is dominated by the barrier layer (e.g. poorly hydrated) which is a highly defective semiconductor (either p- or n-type), where the electronic conduction across it, is described by the transport of vacancies acting as electronic dopants.

Figure A1. Schematic description of the interface metal/lm/solution used to describe the Ti anodization. This gure represents the formation and annihilation processes of different types of defects subjected to diffusion across the passive lms employing a Kroger-Vink notation: m , vacancy in metal phase, + anion vacancy, V hyVM cation vacancy, Mi interstitial cation, VO O H droxide vacancy, M M metal cation in a cation site, O O oxygen ion in anion + cation in solution.35 site, O HO H hydroxyl ion in hydroxide site, M(ac)

Downloaded on 2013-05-02 to IP 148.206.55.254 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

C284

Journal of The Electrochemical Society, 160 (6) C277-C284 (2013)


18. C. E. B. Marino, E. M. de Oliveira, R. C. Rocha-Filho, and S. R. Biaggio, Corros. Sci., 43, 1465 (2001). 19. M. A. M. Ibrahim, D. Pongkao, and M. Yoshimura, J. Solid State Electrochem., 6, 341 (2002). 20. M. Methiko s-Hukovi c, J. Bo zi cevi c, and S. Brini c, J. Electrochem. Soc., 149, B450 (2002). 21. Y. Z. Huang and D. J. Blackwood, Electrochim. Acta, 51, 1099 (2005). 22. E. Pel aez-Abell an, L. Rocha-sousa, W.-D. M uller, and A. C. Guastaldi, Corros. Sci., 49, 1645 (2007). 23. Z. Xia, H. Nanjo, T. Aizawa, M. Kanakubo, M. Fujimura, and J. Onagawa, Surf. Sci., 601, 5133 (2007). 24. I. Milo sev, T. Kosec, and H.-H. Strehblow, Electrochim. Acta, 53, 3547 (2008). 25. N. T. C. Oliveira and A. C. Guastaldi, Corros. Sci., 50, 938 (2008). n, D. Y. Pe 26. H. Estupi na na, Y. O. Garc a, R. Cabanzo, and E. Mej a-Ospino, Eur. Phys. J. D, 53, 69 (2009). 27. P. Acevedo-Pe na, G. V azquez, D. Laverde, J. E. Pedraza-Rosas, and I. Gonz alez, J. Solid State Electrochem., 14, 757 (2010). 28. Y.-L. Chung, D.-S. Gan, K.-L. Ou, and S.-Y. Chiou, J. Electrochem. Soc., 158, C319 (2011). 29. C. C. Manole and C. Pirvu, Surf. Interface Anal., 43, 1022 (2011). 30. M.-Y. Hsu, W.-C. Yang, H. Teng, and J. Leu, J. Electrochem. Soc., 158, K81 (2011). 31. D. I. Ptukhov, A. A. Eliseev, I. V. Kolesnik, K. S. Napolskii, A. V. Lukashin, A. V. Garshev, Y. D. Tretyakov, D. Chernyshov, W. Bras, S.-F. Chen, and C.-P. Liu, J. Porous Mater., 19, 71 (2012). 32. M.-Y. Hsu, W.-C. Yang, H. Teng, and J. Leu, J. Electrochem. Soc., 158, K81 (2011). 33. V. A. Shvets, V. A. Lavrenko, V. I. Subbotin, V. N. Talash, and L. I. Kuznetsova, Powder Met. Metal. Ceram., 49, 702 (2011). 34. G. H. Kelsall and D. J. J. Robbins, J. Electroanal. Chem., 283, 135 (1990). 35. R. Cabrera-Sierra, M. A. Pech-Canul, and I. Gonz alez, J. Electrochem. Soc., 153, B101 (2006). 36. D. D. Macdonald and M. Urquidi-Macdonald, J. Electrochem. Soc., 137, 2395 (1990). 37. D. D. Macdonald, S. R. Biaggio, and H. Song, J. Electrochem. Soc., 139, 170 (1992). 38. G. Okamoto, Corros. Sci., 13, 471 (1973). 39. S. Pyun, C. Lim, and R. A. Oriani, Corros. Sci., 33, 437 (1992). 40. B. C. Bunker, G. C. Nelson, K. R. Zavadil, J. C. Barbour, F. F. Wall, J. D. Sullivan, C. F. Windisch Jr., M. H. Engelhardt, and D. R. Baer, J. Phys. Chem. B, 106, 4705 (2002). 41. P. Acevedo Pe na and I. Gonz alez, J. Electrochem. Soc., 159, C101 (2012). 42. G. V azquez and I. Gonz alez, Electrochim. Acta, 52, 6771 (2007). 43. G. V azquez and I. Gonz alez, J. Electrochem. Soc., 154, C702 (2007). 44. R. Cabrera-Sierra, J. M. Hallen, J. V azquez-Arenas, G. V azquez, and I. Gonz alez, J. Electroanal. Chem., 638, 51 (2010). 45. R. Cabrera-Sierra, J. Vazquez-Arenas, S. Cardoso, R. M. Luna-S anchez, M. A. Trejo, J. Mar n Cruz, and J. M. Hallen, Electrochim. Acta, 56, 8040 (2011). 46. F. Spadacecchia, G. Cappelletti, S. Ardizzone, M. Ceotto, and L. Falciola, J. Phys. Chem. C, 115, 6381 (2011). 47. H. Uchiyama, M. Yukizawa, and H. Kozuka, J. Phys. Chem. C, 115, 7050 (2011). 48. Z. T. Y. Al-Abdullah, Y. Shin, R. Kler, C. C. Perry, W. Zhou, and Q. Chen, Nanotech., 21, 505601 (2010). 49. P. Acevedo Pe na, G. V azquez, D. Laverde, J. E. Pedraza Avella, J. Manr quez, and I. Gonz alez, J. Electrochem. Soc., 156, C377 (2009). 50. F. Di Quarto, A. Di Paola, and C. Sunseri, J. Electrochem. Soc., 127, 1016 (1980). 51. C. da Fonseca and M. da Cunha Belo, Mater. Sci. Forum, 192194, 177 (1995). 52. G. Lu, A. Linsebigler, and J. T Yater Jr., J. Phys. Chem., 98, 11733 (1994). 53. N. Serpone, J. Phys. Chem. B, 110, 24287 (2006). 54. A. V. Emeline, V. N. Kuznetsov, V. K. Rybchuk, and N. Serpone, Int. J. Photoenergy, 2008, 19 (2008). 55. C. Di Valentin, G. Pacchioni, and A. Selloni, J. Phys. Chem. C, 113, 20543 (2009). 56. G. Mattioli, P. Alippi, F. Filippone, R. Caminiti, and A. A. Bonapasta, J. Phys. Chem. C, 114, 21694 (2010). 57. A. Prusi, L. Arsov, B. Haran, and B. N. Popov, J. Electrochem. Soc., 149, B491 (2002). 58. M. Schneider, S. Schroth, J. Schilm, and A. Michaelis, Electrochim. Acta, 54, 2663 (2009). 59. M. Schneider, U. Langklotz, and A. Michaelis, Surf. Interface Anal., 43, 1471 (2011).

From the reaction mechanism sketched in Figure A1, the corresponding transfer function was derived for n-type (present case) semiconductor lms considering that the total impedance is dominated by this barrier layer and its rate-controlling steps, equations A1 to A5.
1 ( Z T Rs )1 = j C p + Z f

[A1]

ZT and Rs are the overall impedance and the solution resistance, respectively. Cp is the interfacial capacitance, Zf the overall faradaic impedance, which is given by the expression: Z f = Z fo + Z fh [A2]

where, Zfo and Zfh are the faradaic impedances of oxygen and hydroxyl ion vacancies at the metal-lm interface, respectively. Reactions 3a and 3b are related to the formation of oxygen and hydroxyl vacancies at the metal lm interface, whereby the transfer function (equation A2) arising from these reactions can be expressed as: Z f = R1, O 2 1 + k3a 1 + k 3b e O L ( K + O ) e O L ( K O ) + R 2, O H j (e O L e O L ) [A3]

eh L /2 ( K + h ) eh L /2 ( K h ) 2 j (eh L /2 eh L /2 )

1 0 a1 = 2 F b3a k3a + b3a k3a C O R 1, O 2 1 0 = F b3b k3b + b3b k3b C O H a1 R 2, O H

[A4]

[A5]

Where, O = K 2 + j / DV O , h = K 2 + 4 j / DV O H .k3a and k3b are the backward rate constants, corresponding to the formation of both oxygen and hydroxyl vacancies. a1 is the potential drop at the metal/lm interface. The resistive terms R 1, O 2 and R 2, O H involve forward and backward rate constants (reactions 3a and 3b, respectively), which do not depend on frequency. DV O and DV O H are the diffusion coefcients of the oxygen and hydroxyl vacancies. K is a constant parameter, given by K = F / RT , where is the electric eld strength; F,the Faraday constant; R, the universal gas constant, and T, the temperature.

References
1. J. Yu, G. Dai, and B. Chang, J. Phys. Chem. C, 114, 19378 (2010). 2. Z. Jiang, X. Dai, and H. Middlenton, Mater. Chem. Phys., 126, 859 (2011). 3. C. Liu, Y. Wang, M. Wang, W. Huang, and P. K. Chu, Surf. Coat. Technol., 206, 63 (2011). 4. P. Roy, S. Berger, and P. Schmuki, Agnew. Chem. Int. Ed., 50, 2904 (2011). 5. L. Hamadou, A. Kadri, and N. Benbrahim, Appl. Surf. Sci., 252, 1510 (2005). 6. B. Roh and D. D. Macdonald, Russ. J. Electrochem., 43, 125 (2007). 7. L. A. S. Ries, M. Da Cunha Belo, M. G. S. Ferreira, and I. L. Muller, Corros. Sci., 50, 676 (2008). 8. S. Auvinen, M. Alatalo, H. Haario, J.-P. Jalava, and R.-J. J. Lamminm aki, J. Phys. Chem. C, 115, 8484 (2011). 9. M. Methiko s-Hukovi c and M. Ceraj-Ceri c, Surf. Techol., 24, 273 (1985). 10. M. Pankuch, R. Bell, and C. A. Melendres, Electrochim. Acta, 38, 2777 (1993). 11. C. da Fonseca, S. Boudin, and M. J. da Cunha Belo, Electroanal. Chem., 379, 173 (1994). 12. J. Pouilleau, D. Devilliers, F. Garrido, S. Durand-Vidal, and E. Mah e, Mater. Sci. Engin. B, 47, 235 (1997). 13. S. Y. Yu and J. R. Scully, Corrosion, 53, 965 (1997). 14. A. L. Pergament and G. B. Stefanovich, Thin Solid Films, 322, 33 (1998). 15. I. Milo sev, M. Methiko s-Hukovi c, and H.-H. Strehblow, Biomater., 21, 2103 (2000). 16. E. M. Oliveira, C. E. B. Marino, S. R. Biaggio, and R. C. Rocha-Filho, Electrochem. Commun., 2, 254 (2000). 17. E. Vasilescu, P. Drob, M. V. Popa, M. Anghel, A. Santana Lopez, and I. Mirza-Rosca, Mater. Corros., 51, 413 (2000).

Downloaded on 2013-05-02 to IP 148.206.55.254 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use

Você também pode gostar