Volta Potential of Oxidized Aluminum Studied by Scanning Kelvin

Apr 14, 2010 - Model for Anodic Film Growth on Aluminum with Coupled Bulk Transport and Interfacial Reactions. Stephen DeWitt and Katsuyo Thornton...
0 downloads 0 Views 3MB Size
8474

J. Phys. Chem. C 2010, 114, 8474–8484

Volta Potential of Oxidized Aluminum Studied by Scanning Kelvin Probe Force Microscopy Kiryl A. Yasakau, Andrei N. Salak, Mikhail L. Zheludkevich,* and Ma´rio G. S. Ferreira Department of Ceramics and Glass Engineering/CICECO, UniVersity of AVeiro, 3810-193, AVeiro, Portugal ReceiVed: February 4, 2010; ReVised Manuscript ReceiVed: April 5, 2010

The Volta potential difference (VPD) of oxide layers formed on an aluminum surface as a result of anodization in a neutral solution of boric acid, thermal oxidation in air, and immersion in boiling water was studied by scanning Kelvin probe force microscopy. The measured VPD value was correlated with the layer thickness. It has been suggested that the measured VPD value is contributed by induced potential from the charges trapped by defects. Provided that the defects created during oxidation and thereby the embedded charges are mainly concentrated at the film borders, the induced potential depends linearly on film thickness. The observed difference in the linear dependence of VPD on the thickness of anodic films has been explained in terms of their charge distribution profiles. The effect of annealing in air on the VPD of the anodic films has been associated with charge redistribution and partial disappearance of the defects. It was shown that, on annealing, the VPD of the anodic films tends to the equilibrium level which is the VPD value for a nonanodized film thermally oxidized at the given annealing temperature. 1. Introduction Aluminum alloys have been extensively used in the aeronautical industry and in naval, construction, and automotive sectors owing to successive combination of valuable mechanical properties, low weight, and a relatively low cost.1,2 However, in a thermodynamical respect, oxidation of aluminum is a strongly favorable process. It results in a high probability of corrosion attack especially in the presence of corrosive agents. The main factor why such a reactive metal can be industrially used is a dense passive oxide film on its surface. Different approaches are used to increase the thickness and improve the protective properties of the covering oxide films. However, due to a possibility of local breakdown of the passive film, aluminum and aluminum-based alloys are susceptible to the localized corrosion attack which often leads to growth of deep defects affecting the mechanical integrity of the structures. The pitting potential of aluminum depends on the aggressiveness of anions in solution and concentration3 and can also be affected by the electrical properties of the oxide. Knowledge on the local electrical characteristics of aluminum and alloys can give answers to important questions on the mechanisms of the pitting attack. However, the electrochemical and photoelectrochemical techniques normally used to characterize the electrical properties of passive oxide films are integral. These methods do not provide information on the local distribution of the passive film. With the appearance of the atomic force microscopy (AFM) technique, it became possible to study the surface of different materials with a high resolution down to nanometers. The development of the complementary scanning Kelvin probe force microscopy (SKPFM) method allowed probing the surface electrical properties of various metals, semiconductors, and thin coatings at micro level with a typical resolution of 100 nm or less.4,5 The working principle of SKPFM is similar to that of the conventional SKP (scanning Kelvin probe)6–8 and based on measurement of the Volta potential difference (VPD) between the surface and a reference metal electrode. In the case of SKP * Corresponding author. Phone: +351-234-370-255. Fax: +351-234-425300. E-mail: [email protected].

measurements, the VPD value is equal to the voltage applied to null the induced current between the oscillating probe and the analyzed surface.9 In the SKPFM technique, VPD corresponds to the voltage that nulls the oscillations of the metal coated cantilever. It comes from the theory that measured VPD is actually the difference of work functions of the reference probe and analyzed surface.10 Knowing the work function of the probe, a work function map of the surface can be obtained. The VPD values were shown to have a certain correlation to the electrode potential map10,11 from which certain information on the possible corrosion behavior of the surface can be drawn. This makes SKPFM a valuable experimental tool. SKPFM was used by Schmutz and Frankel12 to study the correlation between VPD measured in air and electrochemical properties of metals in aqueous solutions. A linear relationship between the VPD value of some metals and their corrosion potential in solution was found. A number of works have been devoted to studies of the localized corrosion activity of metals and alloys using both the SKPFM12–21 and conventional SKP techniques.22–24 It should be noted that, although the relation between VPD and work functions is straightforward, the SKPFM/SPM measurement is not. There are several factors which influence the measured local VPD. Changes of capacitance between the probe and the analyzed surface due to the topography effect10,21 can result in inconsistent feedback of the potential difference. The effects related to the geometry of the scanning probe25 in principle reduce both the resolution and sensitivity of SKPFM, since a large area of cantilever can interact electrostatically with a larger area of the sample surface, thus contributing to noise. Contamination of the surface and tip20 as well as some features that occurred as a result of the surface preparation procedure14,26 also create parasitic electric fields affecting the measured potential difference. When the surface under study is not a noble metal, one should also take into account a natural oxide film present on the surface.12,15,26 In this case, the measured value is actually the difference of the work functions of the probe (which is normally covered with a noble metal) and oxide film on the surface.10,11 Thus, the measured VPD is influenced by the parameters of the oxide

10.1021/jp1011044  2010 American Chemical Society Published on Web 04/14/2010

Volta Potential of Oxidized Aluminum film (composition, morphology, thickness, etc.) which determine its electrical conduction properties. Hausbrand et al.11 have discussed in detail how the potential difference measured using the Kelvin probe technique depends on the properties of the conductive oxides (on iron, zinc, and magnesium surfaces). However, the case of dielectric oxide (particularly, Al2O3) has not been considered there. Our recent works have demonstrated that sufficient change of the Volta potential difference in aluminum alloys AA2024 and AA5083 occurs near the active cathodic intermetallics.13,27,28 One of the given explanations was possible local change of the electrical properties of the surrounding oxide film. Controversial results were reported on the VPD measurements of aluminum alloys in NaCl solutions. It was observed for the AA2024 alloy that with increase of the immersion time a difference in VPD between the intermetallics and the matrix almost disappears. This was attributed to either precipitation of corrosion products15,29 or modification of the electrical properties of oxide film on aluminum matrix.27,30 At the same time, an increase of VPD was reported for the AA5083 alloy under similar conditions. After immersion in a NaCl solution, VPD of the matrix in the vicinity of intermetallics became more positive than that of the matrix away from the intermetallics.27 Such an influence of the active intermetallics on the VPD of the surrounding oxide can be explained by modification of oxide film properties due to its polarization in the course of localized corrosion. Both an increase of the thickness of the oxide film near the intermetallics and a change of the electrical properties of oxide film (e.g., due to incorporation of chloride anions)31,32 can be potentially responsible for this effect. An increase of the VPD measured on the AA2024 alloy surface after immersion in boiling deionized water was reported.33 The increase was correlated with the growth of oxide film on the aluminum surface; however, no clear explanation of the observed effect has been given. As already mentioned, the VPD of the oxide-covered metallic surface is determined by the properties of the oxide film (when other conditions are equal).11 Hence, systematic mapping and analysis of the VPD distribution over oxide films formed on aluminum under different conditions could provide a better understanding of the protective (anticorrosion) effect of those films. In principle, the results obtained for pure aluminum substrate can then be generalized to the aluminum alloys. The aim of the present work is to correlate the Volta potential difference measured in oxidized aluminum surfaces with the electrical properties of the formed oxide layers. The layers were obtained using (i) anodization in a neutral solution of boric acid, (ii) thermal oxidation, and (iii) boiling in deionized water. VPD was measured with the SKPFM technique. In order to characterize the thickness and structure of the oxide films, electrochemical impedance spectroscopy (EIS) and transmission electron microscopy (TEM) were applied. 2. Experimental Methods 2.1. Preparation of Samples and Electrolytes. Aluminum rods (99.999% Al, Alfa Aesar) were embedded in epoxy-based resin (Struers) and abraded using SiC paper followed by polishing with nonaqueous diamond pastes (9-0.5 um). After cleaning in isopropanol, the samples were mechanically polished by colloidal silica (OP-S, Struers) with a mesh size of 40 nm and a pH of 9-10 to eliminate residual roughness. Then, the samples were thoroughly rinsed in a flow of deionized water, dried in air, and left in a desiccator for 1 day. The final surface area of the aluminum samples for the VPD measurements was about 1 cm2.

J. Phys. Chem. C, Vol. 114, No. 18, 2010 8475 A 3 wt % solution of boric acid in deionized water with pH 7 was used for the anodization process and impedance measurements. A solution of ammonium hydroxide (25% in water) was used to adjust the pH of the boric acid solution. The same composition of boric acid solution was used for impedance measurements. 2.2. Anodization Procedure. Anodization of the polished samples was performed both in one and in two steps. To prevent current overload, a galvanostatic polarization at a current density of 1 mA/cm2 was used. After the desired terminal voltage was achieved, the process was either terminated or switched to a potentiostatic regime. In this step, a constant voltage was applied. The procedure was carried out for a set of six samples with terminal voltages of 5, 10, 15, 20, and 30 V. In the case of formation of oxygen bubbles on the surface, the experiment was terminated and the sample was repolished to repeat the procedure. Immediately after anodization, the samples were rinsed with deionized water and dried in air flow. Prior to SKPFM and EIS studies, the anodized samples were held in an oven at 50 °C for 1 h and then left in a desiccator for 1 day. Untreated polished sample was used as a reference. 2.3. Heat Treatment and Immersion in Boiling Water. The polished samples were put in a furnace at room temperature. The initial temperature increase to 300 °C was performed at a rate of 5 °C/min. The samples were held at this temperature for a required time and then immediately removed from the furnace. In order to study the effect of the heat treatment on the VPD of the anodized samples, they were also annealed in a furnace under the same conditions for 3 h. The samples of polished aluminum were also immersed in boiling water for 5 min to 6 h in a Teflon beaker which was previously cleaned by boiled deionized water several times. SKPFM and EIS measurements were performed on all samples both before and after each oxidation precedure: anodization, heat treatment, and immersion in boiling water. 2.4. Experimental Techniques. A commercial AFM Digital Instruments NanoScope III system with Extender Electronic Module was used to study the evolution of the VPD on the aluminum surface. For the SKPFM measurements, the AFM operated in the interleave mode with two pass scans. The first pass acquired the topography of the surface. During the second scan, the tip was lifted up from the surface by 100 nm and an ac voltage of 1 V was applied between the tip and the sample to induce oscillations of the cantilever. Using a nulling technique, the Volta potential difference between the sample and the tip was measured over the whole surface to obtain the map of the Volta potential difference. The measured VPD values were presented versus the values measured on nickel, which was used as a reference due to the stable properties of native oxide.12 For all SKPFM measurements, silicon probes covered with Cr/Pt layers were applied. The EIS measurements were carried out in a three-electrode cell consisting of a mercury-mercurous sulfate reference electrode, platinum counter electrode, and the aluminum sample as the working electrode. The measurements were performed at room temperature in a Faraday cage to avoid any interference with external electromagnetic fields. A Gamry FAS2 Femtostat with the PCI4 Controller in a frequency range from 5 × 104 to 3 × 10-3 Hz was used, taking seven points per frequency decade. All of the spectra were recorded at the open circuit potential with an applied sinusoidal perturbation amplitude of 10 mV. The impedance plots were fitted using appropriate equivalent circuits by means of EchemAnalyst (Gamry Inc.) software.

8476

J. Phys. Chem. C, Vol. 114, No. 18, 2010

Figure 1. Maps of the surface topography (a) and Volta potential difference (b) of the aluminum sample after the final polishing with OP-S suspension.

Figure 2. Characteristics of the anodization process: voltage (left axis) and current (right axis) as functions of time during the galvanostatic polarization at a current density of 1 mA/cm2 and the potentiostatic polarization at different terminal voltages, respectively.

TEM was applied to estimate the morphology and thickness of the oxide films formed on the aluminum surface. Thin sections were obtained using an ultra microtome technique from the samples embedded in an epoxy resin. The samples were examined using a Hitachi H-9000 TEM operating at 300 kV. 3. Results 3.1. Characterization of Aluminum Surface after Polishing. A typical surface topography and the respective VPD map of aluminum after polishing are presented in Figure 1. The borders between individual grains are visible in the topographic image (Figure 1a) due to the chemical etching effect of the polishing suspension, which has a pH of about 9.5. The map of the Volta potential difference (Figure 1b) shows the grain boundaries to have a lower VPD (darker lines) compared with the grain surface. The average VPD of this sample was found to be -1.18 V (hereafter versus a Ni reference sample). 3.2. Anodization of Aluminum Samples. Dependences of voltage and current (for the first and second step, respectively) on anodization time are shown in Figure 2. During the galvanostatic polarization step (at 1 mA/cm2), the voltage increased linearly with a rate of 31 V/min until the desired terminal voltage was achieved. The observed rate of voltage increase during anodization is slightly higher than that reported for aluminum anodized in a boric acid electrolyte (28 V/min).34

Yasakau et al. This difference can be associated with slight deviations in temperature and concentration of the electrolyte. During the potentiostatic polarization regime at different terminal voltages, the current decreased as shown in Figure 2. Figure 3 shows the maps of the topography and VPD for the aluminum sample before and after anodization at a terminal voltage of 20 V. No visible changes resulting from the anodic film growth are seen in the surface topography. At the same time, the average VPD increased from -1.25 to 0.20 V (cf. Figure 3b and d). The VPD map of the anodized sample (Figure 3d) demonstrates slight variations of potential, which could be due to either a presence of defects in the initial oxide film or some artifacts influencing the electric field distribution during the anodization. In order to ensure reproducibility, different areas of each sample were scanned followed by statistical treatment. The VPD values measured on surfaces anodized at different terminal voltages are summarized in Figure 4. It is seen that the dependence of VPD on the terminal voltage is close to linear. Since the magnitude of the voltage (anodizing voltage for the two-step procedure) certainly determines the thickness of the anodic oxide films, it might be expected that VPD correlates with thickness. Therefore, EIS and TEM were used to provide data on the thicknesses of the formed films. 3.3. Estimation of the Oxide Film Thickness. The capacitance of dielectric film on a metal surface can be evaluated from the respective impedance spectra. Then, the thickness is easily calculated using a plane capacitor model if the dielectric constant and capacitance of the film are known. Figure 5a presents Bode plots of aluminum samples anodized at different voltages. Only one time constant associated with the capacitance of the oxide film is seen in the spectra. The impedance modulus increases, indicating that capacitance decreases with the increase of the voltage applied for anodization. The dependence of the phase angle on frequency shown in Figure 5a reflects the practically dielectric behavior of anodic layers. In order to calculate the capacitance values, the impedance spectra were fitted using an equivalent circuit (Figure 5b). The model comprises one time constant element associated with the dielectric anodic film (Ox) connected in parallel with the resistance of the oxide layer (R). A dielectric constant of the aluminum oxide equal to 9 was used for calculation of the thickness of the anodic layers.35 The slope of the plot “thickness versus anodizing voltage” was estimated to be about 1.4 nm/V, which is consistent with the literature data.36 TEM was applied as an independent method to verify the calculations of thickness with the plane capacitor model. A cross section image of the oxide film formed at 20 V is shown in Figure 6. One can see that the film is dense and uniform. The thickness of the oxide layer formed at this voltage is about 30 nm, which is in good agreement with the value estimated from a fitting of the respective impedance spectrum. 3.4. Thermal Oxidation of Aluminum. From the results presented above, one can conclude that the VPD of the anodized aluminum increases linearly with an increase of the oxide thickness. In order to attribute unambiguously the observed increase of the potential to growth of the anodic oxide layer and eliminate other effects related to the anodization process, additional experiments were performed. Thermal oxidation (annealing) in air was used as another approach for the oxide film formation. Figure 7 demonstrates changes in topography and VPD resulting from the heat treatment at 300 °C for 3 h. It is seen that the thermal treatment modifies the sample surface (cf. Figure 7a and c). This can be an effect of annealing which releases mechanical stresses and pushes defects out from the

Volta Potential of Oxidized Aluminum

J. Phys. Chem. C, Vol. 114, No. 18, 2010 8477

Figure 3. Maps of the surface topography (left) and Volta potential difference (right) of the aluminum sample before (a, b) and after (c, d) the two-step anodization at a terminal voltage of 20 V.

Figure 4. VPD of the two-step anodized samples as a function of the terminal voltage.

Figure 6. TEM cross section image of the aluminum sample after the two-step anodization at a terminal voltage of 20 V.

Figure 5. (a) Modulus of impedance |Z| and phase angle of the aluminum samples anodized at different voltages as functions of frequency; solid lines represent the fitting results. (b) Equivalent circuit comprising an R-C element associated with the anodic oxide film (Ox, R).

bulk metal. The heat treatment also causes certain changes in the VPD map: the average VPD level of the samples increases, indicating the same tendency as in the case of anodization. The VPD measured in the samples treated at 300 °C is presented in Figure 8 as a function of the annealing time. One can see that VPD increases by 0.65 V during the first hour of heat treatment. Further annealing has no significant effect on the VPD magnitude.

The thermally treated samples were also tested by EIS in order to estimate the thickness of the respective oxide films. The Bode plots obtained for the thermally oxidized aluminum (Figure 9) are similar to those recorded from the anodized samples (cf. Figure 5). Therefore, the spectra were fitted using the same equivalent circuit. It has been found that after the first hour of heat treatment the oxide film thickness is ∼2 nm. Further increase of the treatment time has little effect on thickness: the increment is about 10% by the end of the fourth hour. Comparison of the above data has revealed that at thermal oxidation at 300 °C the VPD of the samples increases with the film thickness. The same trend was observed in samples thermally oxidized at 200, 400, and 500 °C (not presented here). 3.5. Oxidation of Aluminum in Boiling Water. The polished aluminum samples were also oxidized in boiling deionized water. Immersion of the samples in boiling water results in a continuous growth of the oxide layer. Figure 10 shows maps of the topography and VPD for the sample immersed in boiling water for 4 h. It is seen that, unlike the films grown by means of anodization and thermal oxidation, the immersed film has a porous flaked structure (Figure 10c). The Volta potential difference was measured in the same place of the sample after immersion in boiling water for different dwell times. The summarized data on the VPD as a function of immersion time are shown in Figure 11. Surprisingly, this

8478

J. Phys. Chem. C, Vol. 114, No. 18, 2010

Yasakau et al.

Figure 7. Maps of the surface topography (left) and Volta potential difference (right) of the aluminum sample before (a, b) and after (c, d) heat treatment at 300 °C for 3 h.

Figure 8. Evolution of VPD during the thermal oxidation of aluminum at 300 °C.

Figure 9. Frequency dependence of the modulus of impedance |Z| and phase angle of the aluminum samples thermally oxidized at 300 °C; solid lines represent the fitting results.

treatment brought in no meaningful changes in VPD. During the first minutes of immersion, its value decreases by 0.1 V; however, 1 h after, it comes back to a level close to the VPD value of the reference aluminum surface (about -1.2 V). Thus, changes in Volta potential difference turned out to be negligible, although for such long immersion time an increase of the total film thickness to some micrometers could be expected.37 Results of the EIS measurements of the aluminum samples immersed in boiling water are presented in Figure 12a. The respective impedance spectra suggest two time constants. Indeed, it was previously reported that the frequency behavior of the phase angle of immersed aluminum samples requires a second constant phase element in the fitting process.38 Thus, a twolayer structure of the immersed film is suggested and hence contributions from inner and outer layers can be obtained. Indeed, a combination of two R-C elements in series (Figure

12b) allowed us to fit the obtained spectra with a high goodness. The thickness of the bottom (inner) layer of the oxide film after 1 h of immersion was estimated to be ∼1 nm. The same value of dielectric constant (ε ) 9, part 3.3) was used for the calculations. The thickness of the outer part was impossible to estimate due to a short circuit caused by its porous structure. No regular significant changes in the thickness of the inner layer have been revealed as a result of an increase of the dwell time. It suggests that the inner layer does not thicken when the immersion time exceeds about 1 h. At the same time, a rise of the resistance of the outer part of the film was revealed which can indicate an increase in thickness as well as the partial blocking of the pores. The results presented in parts 3.2-3.4 suggest a certain “film thickness-VPD” correlation for anodized and thermally oxidized aluminum surfaces. At the same time, some spread in both values of VPD and their dependences on thickness have been revealed. It was concluded that the obtained results need further analysis. Therefore, characteristic features of the oxidation processes used in this study and their possible impacts on the properties of the obtained oxide layers were considered in more detail. Moreover, the VPD measurements were also performed in the samples anodized using the galvanostatic procedure only (one-step anodization process). 4. Discussion 4.1. Volta Potential Difference of Anodized Aluminum. It is generally accepted36 that electrolytes without dissolving action (e.g., a neutral solution of boric acid) assist in the formation of a barrier type anodic oxide film on aluminum at anodic polarization. The process of anodic film formation occurs via a mutual migration of aluminum cations (Al3+) from the side of metal toward solution and oxygen ions (O2-) in opposite direction. Cations combine with anions and form Al2O3. The formed oxide layer is compact and has very low electronic conductivity. Ionic conductivity is predominant; however, it requires a high electric field applied across the film to assist ion movement. It follows from the relation between ionic current density and electric field energy that the resulting thickness of anodic film depends on applied voltage36 and the process of ionic transport starts when the electric field strength exceeds the barrier value (about 7 × 106 V/cm).

Volta Potential of Oxidized Aluminum

J. Phys. Chem. C, Vol. 114, No. 18, 2010 8479

Figure 10. Maps of the surface topography (left) and Volta potential difference (right) of the aluminum sample before (a, b) and after (c, d) immersion in boiling water for 4 h.

Figure 11. Volta potential difference of the aluminum sample versus time of immersion in boiling deionized water.

Figure 12. (a) Frequency dependences of the modulus of impedance |Z| and phase angle of the aluminum samples after immersion in boiling water; solid lines represent the fitting results. (b) Equivalent scheme comprising two R-C elements connected in series: one associated with the porous outer part (Ox1, R1) and the second one with the dense inner layer (Ox2, R2) of the oxide film.

Supplementary measurements (SKPFM and EIS) performed to compare the parameters of the aluminum samples oxidized using either the one-step anodization procedure (only galvanostatic growth until the terminal voltage) or the two-step procedure (galvanostatic growth followed by potentiostatic step at terminal voltage) have revealed certain differences in their properties. One can see from Figure 13 that a total thickness of the oxide layer formed at a given terminal voltage in the case

Figure 13. Volta potential difference of the aluminum surface as a function of the thickness of the oxide layer formed using a two-step anodization procedure (squares) and a one-step procedure (triangles).

of the one-step anodization process is regularly lower by only about 10% compared with the value measured after the twostep procedure. A similar dependence was also reported by other authors.39 Thus, the observed difference in thickness is regular and small. At the same time, the difference in the VPD value (∆VPD) measured in the samples anodized at the same terminal voltage using one-step and two-step procedures is significant (Figure 13). The higher the applied voltage, the larger the observed difference. As an example, for the sample anodized at 30 V, ∆VPD is about 1.25 V. Thus, the difference in the Volta potential is high and cannot be explained only by the difference in the film thickness. It seems that the description of this effect should proceed from the mechanism of the oxide layer growth and the resulting electrical properties of the obtained films. In the course of the anodization, a gradient of charge density naturally occurs in the growing oxide film. Although cations Al3+ and anions O2- move throughout the film, the concentration of these ions is the highest at the metal/oxide and electrolyte/ oxide interfaces, respectively. The processes of aluminum and oxygen ionization occur mainly near the film borders. Moreover, these borders are not well-defined. In this respect, the processes in the growing oxide film during the first (galvanostatic) stage take place in an open system. Due to the specificity of the oxidation of a metal surface, the layers bordering the interfaces are the most defective regions of the oxide film. Positively charged defects, (VO2+) and (Ali3+) are supposed to build up near the metal/oxide boundary, whereas a negative charge is

8480

J. Phys. Chem. C, Vol. 114, No. 18, 2010

Yasakau et al.

Figure 14. Schematic representation of charge distribution in oxide film on an aluminum surface polarized as a result of anodization: (a) one-step procedure; (b) two-step procedure.

increased at the oxide/electrolyte interface due to the presence of (VAl3-), (Oi2-), and other species including those coming from solution. Switching of the anodizing voltage (when the onestep oxidation only is applied) results in a global slowing down of the charge movements. As a result, the charge distribution being qualitatively similar to that existing in the external field is kept as shown schematically in Figure 14a. When the second (potentiostatic) regime is switched on (immediately after the first step), the system comes to a new dynamic equilibrium. The second step, although longer than the first one, leads to no appreciable growth of the oxide layer. During this step, the current continuously decreases and the thickness of the oxide increases insignificantly (see Figure 13). Although the ionization processes at the film borders persist, growth slows down, since the electric field strength in the film decreases. Thus, the characteristic processes occurring in the potentiostatic regime are accumulation and redistribution of charge carriers and charged defects. These species still increase the density of positive and negative charges at the outer and inner boundaries of the film, respectively (Figure 14b). It should be noticed that an increase of duration of the second anodization step does not influence the measured VPD, which suggests a fast charge redistribution process. The measured VPD can be represented as a sum of the potential induced by embedded charges (induced potential) and the contact potential, which is the VPD value provided that the layer is not charged (polarized). The latter is determined by natural properties of the oxide and is believed to be independent of film thickness. Thus, one can assume that changes in the measured VPD are caused by changes in the induced potential. Lambert et al.39 have reported on a permanent polarization of alumina films (100-400 nm thick) due to the charges embedded during anodization. It was pointed out that those charges are located in relatively thin layers compared to the film thickness, at the metal/oxide and electrolyte/oxide interfaces. A capacitor representation is applicable in this case and hence the effective surface charge density (σ) can be calculated using the following equation:39,40

V)

σ·d ε · ε0

(1)

where V is the potential induced by the charges, d is the thickness of the film, ε is the dielectric constant of the oxide, and ε0 is the permittivity of the free space. A linear fitting of the experimental dependence of the Volta potential difference on thickness yields a slope (σ/ε · ε0) of 0.019 V/nm for the onestep anodization process (Figure 13). The surface charge density

calculated using this slope value is 9.46 × 1011 charges/cm2. For the two-step anodization process, the slope is 0.055 V/nm and σ is 2.76 × 1012 charges/cm2. The obtained value of charge density is higher than that for the one-step process by almost a factor of 3, and both are in a good correlation with the data reported in ref 39. However, a recent study of the polarization and conductivity of anodic oxide layers on an aluminum surface41 has revealed that the embedded charge is rather distributed throughout the film than located near the borders. It should be noted that the oxide layers mentioned in ref 41 were 12-54 nm thick, i.e., about 1 order of magnitude thinner as compared to those studied by Lambert and co-workers.39 Nevertheless, the contradictory conclusions on charge distribution in anodic alumina films suggest that the applicability of the capacitor model should be proved. Assuming that the distribution of positive and negative charges in anodic oxide films is symmetrical with respect to the middle plane, the induced potential is

V)

2 ε · ε0

∫0d/2 ( 2d - x)F(x) dx

(2)

where F(x) is the (volume) density of the embedded charges located in a thin layer at a distance x from the film surface. (The dielectric permittivity ε is supposed to be constant over the film.) Calculations performed using eq 2 have shown that a linear dependence of potential V on thickness d is met without a requirement of location of the embedded charge in a thin layer near the film border, provided that the profile of the charge distribution F(x) (see Figure 14) is the same for each d. In other words, the increment of potential is proportional to the increment of film thickness if the total embedded charge (Q+ + |Q-|) and its distribution remain invariable. The maximal thickness of a layer in which the charge is distributed is allowed to be comparable with d/2. Thus, the observed linear dependence “V Versus d” does not necessarily mean that the separated charges are located in very thin layers near the film surfaces. On the basis of the above model, a higher slope of the linear VPD(d) dependence for the two-step anodized sample than that for the one-step anodized Al (Figure 13) can result from a sharper profile of the charge distribution, as shown in Figure 14. A wider F(x) provides a smaller induced potential even if the total embedded charge is the same. The presence of the embedded charges in the case of anodic film growth is directly associated with the transfer processes taking place both at the oxide/metal and oxide/electrolyte interfaces.42,43 On the assumption of the embedded charges resulting from defects,39 the possible origins of the charges, namely, from water and/or electrolyte anions, should be considered in more detail. Hickmott44,45 has reported a difference of about 1 order of magnitude in the measured polarization charge between the samples prepared in an aqueous borate as compared with borate electrolytes based on ethylene glycol. Hydrogenation of aluminum oxide is known to result in an increase of density of defects and is responsible for the decrease of breakdown potential and conductivity of the oxide.35 However, incorporation of protons in the anodic layer seems to be insignificant owing to their positive charge that would prevent them from going inside the oxide at anodic potential applied to aluminum.35 At the same time, hydroxyl species being negatively charged can be incorporated in the oxide at positive potential. These species are certainly not free in the oxide and entrapped in the structure.

Volta Potential of Oxidized Aluminum Nevertheless, existing data of the distribution of protons inside the oxide film testify that the oxide contains hydrogen at a level less than 0.17% wt.46 Such a negligible amount of protons cannot explain the effect of charge incorporation. On the other hand, there is much evidence that electrolyte species can be incorporated in the anodic alumina during anodization.34,47 This process can cause mechanical stresses in the oxide. During the potentiostatic step, the incorporation of electrolyte anions is enhanced, since the movement of charges becomes slower at lower electric field strength. Some authors associate such incorporation with the appearance of the embedded charges.46,48 The incorporation of anions in the oxide film leads to the formation of structural defects which are able to carry charge. Defects and impurities determine the existence of so-called “traps”, donor or acceptor levels in the band gap of alumina. The appearance of negative charge in anodic aluminum oxide is ascribed to injection of electrons in the trap states.39,44,49,50 This approach assumes the electron injection to come from water decomposition. In order to clarify the influence of entrapped species from water electrolyte on the shift in VPD, additional experiments were performed. Aluminum samples were anodized using the two-step method in borate glycol electrolyte. The same trend of increase of the Volta potential difference was revealed (not shown) as on the samples anodized in water-based borate electrolyte (Figure 13). It indicates that the role of water in the charge incorporation is likely overestimated. At the same time, it should be noticed that some amount of water can be present in borate glycol electrolyte and affect the anodic oxidation process. These investigations are in progress and will be published elsewhere. It is known that annealing of anodic oxide films results in recombination of charge carriers and in a general decrease of number of defects.36,51 This effect is evidenced by a decrease of conductivity and an increase of the breakdown potential of the alumina film.39,51 One can expect that a heat treatment will reduce the amount of embedded charges and thereby influence the measured Volta potential difference. Therefore, additionally, the anodized samples were heat treated and studied using SKPFM. 4.2. Effect of Heat Treatment. Heat treatment of aluminum results in an increase of the oxidation rate; however, such an increase depends on temperature and partial pressure of oxygen. The kinetics of thermal oxidation of aluminum in ultra high vacuum with controlled partial pressure of oxygen were studied by X-ray photoelectron spectroscopy (XPS) and high resolution transmission electron microscopy (HR-TEM).52–54 It was established that until 300 °C an amorphous oxide film is formed, whereas at higher temperature the amorphous oxide formed at the initial step gradually crystallizes into a γ-Al2O3 phase. The mechanism of the thermal oxidation was described55,56 to consist of several stages. The first stage of the process includes adsorption of oxygen molecules on the metal surface and transfer of electrons either by tunneling or thermionic emission to acceptor levels in oxygen species. The potential formed due to a negatively charged layer of oxygen results in a high electric field across the oxide that favors outward movement of Al cations from metal. Oxide film with a limited thickness is formed at relatively low temperature, whereas at higher temperatures a continuous formation of oxide is observed. A charged layer near the surface of the oxide can be created as a result of the chemisorption process during oxidation of aluminum at high temperature. The measurements performed at elevated temperatures using a Kelvin probe method revealed

J. Phys. Chem. C, Vol. 114, No. 18, 2010 8481

Figure 15. Volta potential difference versus terminal voltage for anodized aluminum samples before (solid squares) and after (open circles) annealing in air at 300 °C for 1 h: (a) two-step procedure; (b) one-step procedure. The reference levels corresponding to the measured VPD values for the untreated matrix and nonanodized matrix oxidized at 300 °C are also indicated in both graphs.

an increase of the Volta potential difference of NiO when oxygen was introduced in a chamber.57 This increase of potential unambiguously indicates the formation of a charged layer near the oxide surface due to the chemisorbed oxygen ions. As seen from Figure 8, there is an increase of the measured Volta potential difference after the heat treatment at 300 °C. A stable VPD offset (around -0.6 V) measured after 1 h of heat treatment suggests the presence of a charged layer. Natural processes of ionization and movement of participating charged species during thermal oxidation are accompanied by the appearance of inhomogeneities resulting in the formation of defects. In fact, the same occurs during anodization of aluminum in borate electrolyte, with an anodic potential being the driving force. In the case of heat treatment, the final thickness of the oxide film and the respective embedded charge are entirely thermodynamically determined by temperature and oxygen pressure. Indeed, an almost constant value of Volta potential difference of the annealed samples measured by SKPFM (Figure 8) reflects some equilibrium conditions that correspond to the certain charge embedded after the first hour of annealing. An increase in dwell time was found to result in no regular change of VPD. In order to clarify the effect of annealing on the Volta potential difference of anodized aluminum, supplementary experiments were done. Prior to the treatments and measurements, a part of the polished surface of each sample was isolated from the rest. The samples were oxidized using either the onestep or two-step anodization process. The samples were annealed at 300 °C for 1 h in the way described in the experimental section, and then, their VPD was measured. The reference measurements were also performed on the nonanodized parts of the surfaces both before and after the heat treatments. The values of VPD measured on the anodized and annealed surfaces as well as the reference values are presented in Figure 15. It should be noticed that similar trends in the Volta potential dependences (black squares) were obtained (cf. Figure 13) for both the one-step and two-step anodized samples, verifying the

8482

J. Phys. Chem. C, Vol. 114, No. 18, 2010

reproducibility of our experiments. The reference VPD values for the aluminum matrix were also found to be the same (within an experimental error) as those measured previously. It is seen from Figure 15 that the thermal treatment of the anodized surfaces results in a considerable change in their Volta potential difference. In spite of some spread in values, a general trend for the Volta potential difference to achieve a certain level independent of the film thickness is observed. The annealing leads to an alignment of the VPDs to the levels -0.85 ( 0.06 and -0.75 ( 0.05 V for the one-step and two-step anodized samples, respectively. One can also see in Figure 15a that the Volta potential difference of the sample anodized using the two-step procedure at a terminal voltage of 15 V is close to the VPD level of the annealed aluminum matrix. At this voltage, an anodic oxide film of about 20 nm thick was formed. At the same time, the thickness of the oxide film on annealed aluminum was calculated to be approximately 3 nm. This suggests again that the measured VPD depends on the distribution of the embedded charges in the oxide layer rather than on the thickness of the film. The following observations related to the thermal treatment effect can be drawn from Figure 15. (i) The Volta potential difference measured in the annealed anodized samples is always lower than the respective one found in the nonanodized matrix oxidized in air under the same conditions. (ii) The average VPD levels for the one-step and two-step anodized samples are similar in magnitude; nevertheless, the level for the former is slightly lower than that for the latter (cf. the values above). (iii) The samples oxidized at the lowest nonzero anodizing potential (5 V) demonstrate the smallest VPD value after annealing. In order to clarify these features, two aspects of thermal treatment, namely, oxidation (of the nonanodized samples) and annealing (of the anodized samples), should be considered. Since thermal oxidation of the samples under study was performed at relatively low temperature, an oxide film of certain thickness was formed during the first hour of treatment (see part 3.4) and the film thickness increased negligibly with an increase of the treatment duration. This suggests that the Volta potential difference of an aluminum surface oxidized at relatively low temperature (e300 °C) in air for 1 h or longer possesses the value which can be considered as the equilibrium one corresponding to given conditions of the thermal treatment. It seems reasonable to suppose that this VPD tends this level regardless of the treatment prehistory of the sample. As already mentioned, the interface layers of as-anodized oxide films are believed to be under mechanical stress and contain many defects. When annealing, most of these defects disappear and stresses are released. These actions as well as some increase of mobility of charge species at the annealing temperature promote redistribution and recombination of charges in the film. One can suggest that some of the defects do not disappear after annealing at 300 °C. Residual defects are expected to impede establishment of the polarization state corresponding to the respective equilibrium VPD value at a given temperature. As a result, this level is not reached (see Figure 15), since the annealed anodic films are less polarized. It has been suggested in ref 39 that a considerable amount of defects is created at the second (potentiostatic) stage of anodization. In contrast to them, on the basis of the above considered pattern of polarization of the oxide layers, one can suppose that the relative concentration of defects and their distribution are rather comparable in both the one-step and twostep anodized films. At the same time, in this assumption, a sample anodized using the one-step process is characterized by a smaller gradient of the embedded charge density across the

Yasakau et al. oxide film (cf. Figure 14). Therefore, in the one-step anodized films, a charge distribution profile is wider and the charge species have to pass a longer mean path to reach the equilibrium state. That is why the average VPD level measured in the onestep anodized surface after annealing is farther from the equilibrium VPD value (corresponding to the nonanodized reference sample thermally treated under the same conditions) than that measured in the two-step anodized sample. For the relatively thin anodic film (e.g., that formed at a terminal voltage of 5 V is about 7 nm thick), this effect is expected to be more pronounced, since double thickness of a defect layer with localized charges is comparable with the film thickness. 4.3. VPD of Aluminum Oxidized in Boiling Water. According to Vedder and Vermilyea,37 the main stages of the reaction between hot water and aluminum are the following: formation of amorphous oxide, dissolution of amorphous oxide, and formation of aluminum hydroxide. The nature of the aluminum hydroxide was studied by infrared spectroscopy and described to be a pseudoboehmite which is aluminum oxyhydroxide containing an excess of water with the approximate composition AlOOH · H2O. Previously, Bernard and Randall58 assumed that the reaction product is boehmite (AlOOH). The surface hydrolysis and nucleation of aluminum hydroxide go faster at higher temperature. It was found59 that an induction time of the hydrous oxide formation in boiling water (∼100 °C) is less than 1 min. Our AFM measurements confirmed the high reaction rate. It has been found that after 5 min of immersion the rms roughness increases from 2.6 nm for the as-polished substrate to 28 nm for that treated in boiling water. TEM study of aluminum foil after 20 min of immersion in boiling water has revealed the formation of a double layered structure of pseudoboehmite on a metal surface.60 Features of an inner layer about 130 nm thick and a fibrous outer layer of ∼350 nm made of platelets were observed. It was also reported60 that the outer part is porous, while the inner part was considered to be dense. It was also found that during anodization over a preformed pseudoboehmite layer the anodic alumina grows below the inner layer. This seems to be possible when the layer is porous, allowing electrolyte solution to penetrate through. Our EIS measurements of the samples treated in boiling water showed the presence of two relaxation processes which can be associated with the double layered structure of the product of oxidation/hydration. Fitting of the EIS spectra using the model presented in Figure 12b yielded the capacitance (Ox2) of the inner part to be 9.12 × 10-6 F/cm2 and the parameter of the constant phase element, n ) 0.948. Such a high value of n indicates the presence of a homogeneous and dense bottom dielectric layer. Even if the dielectric constant of this layer is assumed to be 20 (as reported for pseudoboehmite35), its thickness estimated from the EIS spectra is only about 2 nm. As regards the outer (porous) part of the double layered structure, using the EIS technique, it is impossible to evaluate the thickness of the layer short-circuited by penetrating electrolyte. Thus, our EIS measurements actually determined a dense bottom layer of the oxidation product (Figure 12b). As mentioned in part 3.5, the calculated thickness of this layer remains virtually constant (about 1 nm) when the dwell time is 1 h and more. The observed small irregular changes in the thickness value suggest that the dense layer is continuously formed and dissolved during the immersion. These competing processes are also reflected by some variations of the measured VPD value (Figure 11). In certain aspects, the thin dense layer detected in the double layered structure in boiling water is similar to a natural oxide

Volta Potential of Oxidized Aluminum film formed on the aluminum surface under normal conditions. Their measured VPDs (cf. Figures 11 and 15) were estimated to be very close to the value of the difference of work functions of aluminum (sample) and platinum (the AFM tip was coated with Pt). Both the natural oxide film and the dense layer of the double structure formed in boiling water are not polarized. The charge transport in such thin films is controlled by tunnel effects,55 and any charge-separated state cannot be implemented. Regarding the porous part of the double structure, its composition and morphology suggest that the mechanism of formation is drastically different from that acting in the cases of anodization and thermal oxidation in air. Under conditions of oxidation/hydration in boiling water, a macroscopic separation of the charge in the respective film does not occur. 5. Conclusions Formation of oxide films on an aluminum surface as a result of application of different oxidation methods was studied using the SKPFM technique and EIS and TEM as supplementary tools. The combination of these techniques allows one to correlate the measured VPD, morphology, and thickness of the films with the oxidation conditions (applied voltage, temperature, time of treatment). It has been found that the oxide films formed by anodization in a neutral solution of boric acid and those grown by thermal oxidation in air are rather similar with respect to their structure and properties. An increase of VPD with film thickness (d) is observed for both the anodized and thermally oxidized films; however, the character of the dependence is characteristic of the oxidation method. The observed variation VPD(d) in those films is caused by contribution of the potential induced by embedded charges which are distributed throughout a film with the maximal charge density near the metal/oxide and oxide/air interfaces. It was shown that the slope of the dependence is determined by the profile of the charge distribution across the film. The application of potentiostatic step after galvanostatic anodization leads to contraction of the profile toward the interfaces. It results in a regularly lower induced potential (and thereby VPD) in the only galvanostatically anodized surface. Annealing of the anodized films favors disappearance of defects which are natural traps for the embedded charges. This induces redistribution/recombination of the charges in the oxide films, resulting in a change of the Volta potential difference measured on the Al surface. VPD trends to achieve the certain equilibrium level independent of the film thickness. This level is equal to the VPD value measured in a nonanodized Al surface thermally oxidized at the given annealing temperature. Since some defects remain in the anodized films after annealing, this level is not reached. Unlike the anodized and thermally oxidized films, the oxidation product obtained by means of immersion of the aluminum surface in boiling water was found to have a double layered structure: a thin dense inner layer and a thick porous outer one. SKPFM revealed no meaningful changes in both the thickness of the dense layer and the measured VPD as a function of immersion time. It is suggested that the conduction properties of this layer are similar to those of a natural oxide film formed on an aluminum surface under normal conditions. SKPFM has been shown to be a powerful technique for nondestructive investigation of oxidized surfaces, providing information on both the topography and electric properties of oxide films. It seems promising to continue works in this direction, particularly in order to study the possible impact of the polarized state on protective (anticorrosion) properties of

J. Phys. Chem. C, Vol. 114, No. 18, 2010 8483 covering aluminum oxide films as well as on the adhesion of organic molecules to the metal surfaces covered with oxides. Acknowledgment. The authors thank Dr. M. D. Cunha Belo for useful discussion. The Portuguese Foundation for Science and Technology (FCT-Portugal, grant SFRH/BD/25469/2005 and FCT-project PTDC/CTM/72223/2006) is gratefully acknowledged for financial support. References and Notes (1) Starke, E. A.; Staley, J. T. Application of Modern Aluminum Alloys to Aircraft. Prog. Aerospace Sci. 1996, 32, 131–172. (2) Hecht, R. L.; Kannan, K. In Superplasticity and Superplastic Forming; Ghosh, A. K., Bieler, T. R., Eds.; The Metallurgical Society: Warrendale, PA, 1995. (3) McCafferty, E. The Electrode Kinetics of Pit Initiation on Aluminum. Corros. Sci. 1995, 37, 481–492. (4) Jacobs, H. O.; Knapp, H. F.; Muller, S.; Stemmer, A. Surface potential mapping: A qualitative material contrast in SPM. Ultramicroscopy 1997, 69, 39–49. (5) Yasutake, M.; Aoki, D.; Fujihira, M. Surface Potential Measurements Using the Kelvin Probe Force Microscope. Thin Solid Films 1996, 273, 279–283. (6) Grundmeier, G.; Reinartz, C.; Rohwerder, M.; Stratmann, M. Corrosion Properties of Chemically Modified Metal Surfaces. Electrochim. Acta 1998, 43, 165–174. (7) Grundmeier, G.; Schmidt, W.; Stratmann, M. Corrosion Protection by Organic Coatings: Electrochemical Mechanism and Novel Methods of Investigation. Electrochim. Acta 2000, 45, 2515–2533. (8) Nazarov, A.; Thierry, D. Rate-determining Reactions of Atmospheric Corrosion. Electrochim. Acta 2004, 49, 2717–2724. (9) Yee, S.; Oriani, R. A.; Stratmann, M. Application of a Kelvin Microprobe to the Corrosion of Metals in Humid Atmospheres. J. Electrochem. Soc. 1991, 138, 55–61. (10) Rohwerder, M.; Turcu, F. High-resolution Kelvin Probe Microscopy in Corrosion Science: Scanning Kelvin Probe Force Microscopy (SKPFM) Versus Classical Scanning Kelvin Probe (SKP). Electrochim. Acta 2007, 53, 290–299. (11) Hausbrand, R.; Stratmann, M.; Rohwerder, M. The Physical Meaning of Electrode Potentials at Metal Surfaces and Polymer/Metal Interfaces: Consequences for Delamination. J. Electrochem. Soc. 2008, 155, C369–C379. (12) Schmutz, P.; Frankel, G. S. Characterization of AA2024-T3 by Scanning Kelvin Probe Force Microscopy. J. Electrochem. Soc. 1998, 145, 2285–2295. (13) Yasakau, K. A.; Zheludkevich, M. L.; Lamaka, S. V.; Ferreira, M. G. S. Mechanism of Corrosion Inhibition of AA2024 by Rare-Earth Compounds. J. Phys. Chem. B 2006, 110, 5515–5528. (14) Tanem, B. S.; Svenningsen, G.; Mardalen, J. Relations between Sample Preparation and SKPFM Volta Potential Maps on an EN AW-6005 Aluminium Alloy. Corros. Sci. 2005, 47, 1506–1519. (15) Schmutz, P.; Frankel, G. S. Corrosion Study of AA2024-T3 by Scanning Kelvin Probe Force Microscopy and In Situ Atomic Force Microscopy Scratching. J. Electrochem. Soc. 1998, 145, 2295–2306. (16) Leblanc, P.; Frankel, G. S. A Study of Corrosion and Pitting Initiation of AA2024-T3 Using Atomic Force Microscopy. J. Electrochem. Soc. 2002, 149, B239–B247. (17) Zheludkevich, M. L.; Yasakau, K. A.; Poznyak, S. K.; Ferreira, M. G. S. Triazole and Thiazole Derivatives as Corrosion Inhibitors for AA2024 Aluminium Alloy. Corros. Sci. 2005, 47, 3368–3383. (18) Jonsson, M.; Thierry, D.; LeBozec, N. The Influence of Microstructure on the Corrosion Behaviour of AZ91D Studied by Scanning Kelvin Probe Force Microscopy and Scanning Kelvin Probe. Corros. Sci. 2006, 48, 1193–1208. (19) Andreatta, F.; Apachitei, I.; Kodentsov, A. A.; Dzwonczyk, J.; Duszczyk, J. Volta Potential of Second Phase Particles in Extruded AZ80 Magnesium Alloy. Electrochim. Acta 2006, 51, 3551–3557. (20) Bengtsson-Blucher, D.; Svensson, J.-E.; Johansson, L.-G.; Rohwerder, M.; Stratmann, M. Scanning Kelvin Probe Force Microscopy A Useful Tool for Studying Atmospheric Corrosion of MgAl Alloys In Situ. J. Electrochem. Soc. 2004, 151, B621–B626. (21) Shiraishi, S.; Kanamura, K.; Takehara, Z. Imaging for Uniformity of Lithium Metal Surface Using Tapping Mode-Atomic Force and Surface Potential Microscopy. J. Phys. Chem. B 2001, 105, 123–134. (22) Williams, G.; McMurray, H. N. The Kinetics of Chloride-Induced Filiform Corrosion on Aluminum Alloy AA2024-T3. J. Electrochem. Soc. 2003, 150, B380–B388.

8484

J. Phys. Chem. C, Vol. 114, No. 18, 2010

(23) McMurray, H. N.; Coleman, A. J.; Williams, G.; Afseth, A.; Scamans, G. M. Scanning Kelvin Probe Studies of Filiform Corrosion on Automotive Aluminum Alloy AA6016. J. Electrochem. Soc. 2007, 154, C339–C348. (24) Williams, G.; McMurray, H. N. Polyaniline Inhibition of Filiform Corrosion on Organic Coated AA2024-T3. Electrochim. Acta 2009, 54, 4245–4252. (25) Jacobs, H. O.; Leuchtmann, P.; Homan, O. J.; Stemmer, A. Resolution and Contrast in Kelvin Probe Force Microscopy. J. Appl. Phys. 1998, 84, 1168–1173. (26) Guillaumin, V.; Schmutz, P.; Frankel, G. S. Characterization of Corrosion Interfaces by the Scanning Kelvin Probe Force Microscopy Technique. J. Electrochem. Soc. 2001, 148, B163–B173. (27) Yasakau, K. A.; Zheludkevich, M. L.; Lamaka, S. V.; Ferreira, M. G. S. Role of Intermetallic Phases in Localized Corrosion of AA5083. Electrochim. Acta 2007, 52, 7651–7659. (28) Yasakau, K. A.; Zheludkevich, M. L.; Ferreira, M. G. S. Lanthanide Salts as Corrosion Inhibitors for AA5083. Mechanism and Efficiency of Corrosion Inhibition. J. Electrochem. Soc. 2008, 155, C169–C177. (29) Leblanc, P.; Frankel, G. S. A Study of Corrosion and Pitting Initiation of AA2024-T3 Using Atomic Force Microscopy. J. Electrochem. Soc. 2002, 149, B239–B247. (30) Robin, F.; Jacobs, H.; Homan, O.; Stemmer, A.; Bachtold, W. Investigation of the Cleaved Surface of a p-i-n Laser Using Kelvin Probe Force Microscopy and Two-dimensional Physical Simulations. Appl. Phys. Lett. 2000, 76, 2907–2909. (31) Martin, F. J.; Cheek, G. T.; O’Grady, W. E.; Natishan, P. M. Impedance Studies of the Passive Film on Aluminium. Corros. Sci. 2005, 47, 3187–3201. (32) Pyun, S.; Moon, S. M.; Ahn, S. H.; Kim, S. S. Effects of Cl-, NO3- and SO42- Ions on Anodic Dissolution of Pure Aluminum in Alkaline Solution. Corros. Sci. 1999, 41, 653–667. (33) Muster, T. H.; Hughes, A. E. Applications and Limitations of Scanning Kelvin Probe Force Microscopy for the Surface Analysis of Aluminum Alloys. J. Electrochem. Soc. 2006, 153, B474–B485. (34) Ozawa, K.; Majima, T. Anodization Behavior of Al, and Physical and Electrical Characterization of Its Oxide Films. J. Appl. Phys. 1996, 80, 5828–5836. (35) Sullivan, J. P.; Barbour, J. C.; Dunn, R. G.; Son, K. A.; Montes, L. P.; Missert, N.; Copeland, R. G. In Proceedings of the Symposium on Critical Factors in Localized Corrosion III; Kelly, R. G., Natishan, P. M., Frankel, G. S., Newman, R. C., Eds.; The Electrochemical Society Proceedings: Boston, MA, 1998. (36) Tajima, S. In AdVances in Corrosion Science and Technology; Fontana, M. G., Staehle, R. W., Eds.; Plenum Press: New York, London, 1970. (37) Vedder, W.; Vermilyea, D. A. Aluminum + Water Reaction. J. Chem. Soc., Faraday Trans. 1969, 65, 561–584. (38) Hsu, C. H.; Mansfeld, F. Technical Note: Concerning the Conversion of the Constant Phase Element Parameter Y0 into a Capacitance. Corrosion 2001, 57, 747–748. (39) Lambert, J.; Guthmann, C.; Ortega, C.; Saint-Jean, M. Permanent Polarization and Charge Injection in Thin Anodic Alumina Layers Studied by Electrostatic Force Microscopy. J. Appl. Phys. 2002, 91, 9161–9169. (40) Szunerits, S.; Pust, S.; Wittstock, G. Multidimensional electrochemical imaging in materials science. Anal. Bioanal. Chem. 2007, 389, 1103–1120.

Yasakau et al. (41) Hickmott, T. W. Electrolyte Effects on Charge, Polarization, and Conduction in Thin Anodic Al2O3 Films. I. Initial Charge and TemperatureDependent Polarization. J. Appl. Phys. 2007, 102, Art. No. 093706. (42) Davies, L. W.; Collins, R. E. Anodic Oxide Electrets. Electron. Lett. 1969, 19, 462–463. (43) Bernstein, J. J.; White, R. M. Surface Potential Difference of Anodized Al2O3 Electrets. J. Electrochem. Soc. 1985, 132, 1140–1144. (44) Hickmott, T. W. Polarization and Fowler-Nordheim Tunneling in Anodized Al-Al2O3-Au diodes. J. Appl. Phys. 2000, 87, 7903–7912. (45) Hickmott, T. W. Interface States at the Anodized Al2O3-Metal Interface. J. Appl. Phys. 2001, 89, 5502–5508. (46) Despic, A.; Parkhutik, V. P. In Modern Aspects of Electrochemistry; Bockris, J. O. M., White, R. E., Conway, B. E. S., Eds.; Plenum Press: New York, 1989; Vol. 20, pp 401-504. (47) Thompson, G. E.; Xu, Y.; Skeldon, P.; Shimizu, K.; Han, S. H.; Wood, G. C. Anodic Oxidation of Aluminum. Philos. Mag. 1987, B55, 651–667. (48) Parkhutik, V. P.; Shershulskii, V. I. The Modeling of DC Conductivity of Thin Disorder Dielectrics. J. Phys. D: Appl. Phys. 1986, 19, 623–641. (49) Zudova, L. A.; Agapova, S. I.; Zudov, A. I.; Sterkhov, V. A. Effect of Condidtions of Anodic Oxide Film Formation on Polarization State Established During Film Growth. SoViet Electrochemistry 1975, 11, 1153– 1156. (50) Zudov, A. I.; Zudova, L. A.; Sadakova, G. P.; Najmushina, S. I. Influence of Oxidation Conditions on Electret Characteristics and Space Charge of Aluminum-Aluminum Oxide Systems. SoV. Electrochem. 1983, 19, 164–168. (51) Dignam, M. J. Oxide Films on Aluminum. J. Electrochem. Soc. 1962, 109, 184–191. (52) Jeurgens, L. P. H.; Sloof, W. G.; Tichelaar, F. D.; Mittemeijer, E. J. Composition and Chemical State of the Ions of Aluminium-Oxide Films Formed by Thermal Oxidation of Aluminium. Surf. Sci. 2002, 506, 313–332. (53) Jeurgens, L. P. H.; Sloof, W. G.; Tichelaar, F. D.; Mittemeijer, E. J. Structure and Morphology of Aluminium-Oxide Films Formed by Thermal Oxidation of Aluminium. Thin Solid Films 2002, 418, 89–101. (54) Jeurgens, L. P. H.; Sloof, W. G.; Tichelaar, F. D.; Mittemeijer, E. J. Growth Kinetics and Mechanisms of Aluminum-Oxide Films Formed by Thermal Oxidation of Aluminum. J. Appl. Phys. 2002, 92, 1649–1656. (55) Fromhold, A. T.; Cook, E. L. Kinetics of Oxide Film Growth on Metal Crystals: Electron Tunneling and Ionic Diffusion. Phys. ReV. 1967, 158, 600–612. (56) Fromhold, A. T.; Cook, E. L. Kinetics of Oxide Film Growth on Metal Crystals: Thermal Electron Emission and Ionic Diffusion. Phys. ReV. 1967, 163, 650–664. (57) Bak, T.; Nowotny, J.; Sorrell, C. C. In Electrical properties of oxide materials; Nowotny, J., Sorrell, C. C., Eds.; Trans Tech Publications: Zurich, Switzerland, 1997. (58) Bernard, W. J.; Randall, J. J., Jr. An Investigation of the Reaction between Aluminum and Water. J. Electrochem. Soc. 1960, 107, 483–487. (59) Alwitt, R. S. The Growth of Hydrous Oxide Films on Aluminum. J. Electrochem. Soc. 1974, 121, 1322–1328. (60) Uchi, H.; Kanno, T.; Alwitt, R. S. Structural Features of Crystalline Anodic Alumina Films. J. Electrochem. Soc. 2001, 148, B17–B23.

JP1011044